Wall St. Journal: Why the Florida recounting gives biased results.

greenspun.com : LUSENET : TB2K spinoff uncensored : One Thread

Recount 'Em All, or None at All

By Ed Glaeser

(from The Wall St. Journal: 11/13/00)

There is a well-known trick among statistical economists for biasing your data while looking honest. First, figure out which data points don't agree with your theory. Then zealously clean up the offending data points while leaving the other data alone. The key to maintaining academic dignity is to ensure that you do nothing to the data other than eliminate errors.

But while this approach may seem to improve accuracy, it actually leads to biased results. If you only clean the offending data points, then you will disproportionately keep erroneous data that agrees with your prior views. This leads many scholars to believe that data that is partially cleaned at the discretion of a researcher is worse than bad data.

This lesson from the ivory tower has a clear implication for the current mess in Florida. Hand counting ballots in only a few, carefully chosen counties is a sure way to bias the results. Even if hand counting is more accurate than machine counting, there is a clear bias introduced because Al Gore chose which counties to hand count. Mr. Gore has selected the state and counties where recounting has the best chance of helping him.

This is exactly the same as cleaning other data selectively. Naturally, if this opportunity for selective recounting becomes the norm, the floodgates will open and any candidate who loses a close election would be foolish not to demand a recount.

The immediate implication of this is clear. If there is to be recounting by hand, it cannot be selective. There needs to be total hand counting, not just within Florida, but across the U.S. in any state that was close. One candidate cannot be allowed just to choose where he wants the data cleaned. If this is prohibitively expensive, or time consuming, then it is better to leave the process unchanged than to introduce the selective recounting bias.

More generally, one of the principal lessons of macroeconomics is that rules generally work better than discretion. This is as true in elections as any place else. Giving candidates influence over how election results are processed does not help democracy to accurately reflect the will of the people. Judicial discretion is not much better, as judges will be responding to cases selectively filed by candidates. Furthermore, judges determining elections will exalt the judiciary to a king-making role it should not have.

While it certainly may be appropriate to ban butterfly ballots for all of eternity, and while reform of balloting procedures seems like a must, it is also clearly wrong to selectively recount certain areas.

(Mr. Glaeser is a professor of economics at Harvard University and a visiting fellow at the Brookings Institution.)

-- eve (eve_rebekah@yahoo.com), November 13, 2000

Answers

Who said this?

(I think it was Twain (Mark -- not Shania), but I'm not sure. I could look it up, but I guess I'll just turn it into a contest. The winner gets a cyber-kiss from my new sock-puppet, "Chad, the Butterfly Ballot.")

"There are three kinds of lies: Lies, damn lies, and statistics."

-- eve (eve_rebekah@yahoo.com), November 13, 2000.


Hope I beat ya to the bump!

So, essentially, figures lie & liars figure?

-- flora (***@__._), November 13, 2000.


>> If there is to be recounting by hand, it cannot be selective. There needs to be total hand counting, not just within Florida, but across the U.S. in any state that was close. <<

Mr. Glaeser doesn't need to convince the readership of the WSJ. He needs to convince George W. Bush and his advisors to request the recounts he is advising, for the reasons he is presenting, within the legally prescribed time limits for making such requests.

So far, Bush has failed to take the legal recourse available to him. Gore is exercising the options he has.

As far as I can see, based on observing the moves and countermoves the candidates have been making, each one understands that a handful of close states fall so narrowly within the limits of random variation, that a recount could go either way.

However, each candidate has conciously decided that the 25 electoral votes in Florida are the only prize worth taking, since this one prize would be sufficient for victory. Therefore, they have both ignored the possibility of recounts in other states - even though Bush is winning New Mexico by a mere 17 votes!

In Florida, the Bush campaign survived the automatic recount by a mere 320-some-odd votes. They have decided that any attempt at a recount, in any county, could easily wipe out that lead and is not worth undertaking. They have clearly discarded the recount path as far too risky - no matter where or how they might pursue it. Now they are concentrating all their efforts on preventing any further recount by any and every means they can think up.

Based on this, I would say that Dr. Glaeser's arguments will fall on deaf ears in the Bush campaign. Recounting every county in every close state would be a monumental gamble. Instead, they want to prevent recounting of any sort at any cost to preserve the all-important fiction of owning the lead.

The Gore campaign, meanwhile, is concentrating their efforts on finding the best strategy to pick up at least 350 votes. They have decided to selectively request hand recounts in those counties where they have the best statistical chance of finding 350 previously uncounted Gore votes in excess of previously uncounted Bush votes.

Based on this, I would say that Dr. Glaeser's arguments will fall on deaf ears in the Gore campaign. Recounting every county in every close state would be a monumental gamble for them as well. Instead, they want to recount only the counties that give them the best statistical chance to win.

So, while Dr. Glaeser may have some points to make, they are totally irrelevant to the practical politics of grabbing Florida and nailing it down in the realm of officially-accepted fiction. Neither candidate has the slightest interest in knowing what the real vote count was. At best, some idle curiosity. They just want the electoral votes and the rest be damned!

-- Brian McLaughlin (brianm@ims.com), November 13, 2000.


Brian,

I hear ya. Any lip service paid towards giving the disenfranchised voters their due is just a sham. Gore (and probably Bush too, if the situation were reversed) just wants to win -- by tweaking the right voters in the right counties in the right state.

-- eve (eve_rebekah@yahoo.com), November 13, 2000.


>There is a well-known trick among statistical economists for biasing your data while looking honest.

... and there are well-known (and perfectly legitimate, _in context_) methods among writers, including Ed Glaeser, for buttressing an idea being advocated in a persuasive (rather than objective) article without being too obvious about it -- draw inappropriate (or, at least, ill-fitting) analogies or parallels, omit relevant considerations or counter-arguments, draw unjustified conclusions, and so on.

A persuasive article is not designed to be balanced, but that doesn't make it illegitimate. Persuasive (maybe this category had some other name in my speech or English classes, but I can't think of it right now) articles, or speeches or other presentations, have their rightful places.

Glaeser is going to draw analogies or parallels between elections and economics or statistics. Well, okay. Some such analogies or parallels could be good, while others might be ill-fitting at best.

>This lesson from the ivory tower has a clear implication for the current mess in Florida.

Glaeser omits some important differences between elections and statistics, but writes as though they were irrelevant. Others might disagree with Glaeser on whether some of those differences are irrelevant or can be safely ignored.

This election was political. While many election provisions and procedures are intended to make the outcome as objectively correct as possible, a la statistics, others are designed to balance or guarantee the rights of competing political interests which are inherently unobjective, or to cope with practical limitations of government.

>Even if hand counting is more accurate than machine counting, there is a clear bias introduced because Al Gore chose which counties to hand count.

So what? George Bush had exactly the same legal rights and opportunity to choose counties in which to request hand recounts.

And recounts cost money. AFAIK, in Florida, after a certain point challengers have to put up the money to pay for recounts. Governments do not have unlimited funds with which to conduct their operations. Candidates have to consider this cost in relation to their resources.

Bush freely and voluntarily _chose_ not to request hand recounts.

>Mr. Gore has selected the state and counties where recounting has the best chance of helping him.

Does Glaeser expect anything else? Did he expect Bush to request recounts where recounts have the best chance o helping Gore? What's his point here?

Glaeser's point is to associate the perfectly legitimate choices of Gore and Bush with the idea of statistical bias in order to persuade his reader that there is something wrong with Gore's and Bush's actions, that's what.

>This is exactly the same as cleaning other data selectively.

No, it isn't. Here Glaeser either lies or shows his ignorance.

>Naturally, if this opportunity for selective recounting becomes the norm, the floodgates will open and any candidate who loses a close election would be foolish not to demand a recount.

Do you see Glaeser pointing out economic or partly-economic considerations such as weighing the cost of a recount versus the likelihood that the recount will change the result? No? Odd ... he's "a professor of economics", isn't he?

Guess he decided to just omit that consideration because it doesn't reinforce the idea he wants his reader to accept. Acceptable tactic in a persuasion piece.

>The immediate implication of this is clear. If there is to be recounting by hand, it cannot be selective.

Invalid conclusion, Professor Glaeser. I hope you'd disallow this sort of reasoning by students taking your exams. Of course, such exams are generally intended to be objective, not persuasive.

The economic expense of hand recounting must be borne by someone. If governments do not, perhaps in response to past voter complaints about overspending, then the complaining candidates become the next logical choice. And if candidates are to pay, then they're going to choose to pay only where it suits their interests.

Professor, you're in danger of failing in regard to objectivity. But your grade in persuasion is okay.

>There needs to be total hand counting, not just within Florida, but across the U.S. in any state that was close.

Who decides "close"?

>One candidate cannot be allowed just to choose where he wants the data cleaned.

So, you propose that a candidate have to pay for cleaning data he's not interested in? Do economists do that?

Or did you mean, while not expressing it clearly, that more than one candidate should be allowed to choose where to clean data? If this was your meaning, then you may not have noticed (or, is it that you don't want your reader to notice?) that each candidate already _did_ have an equal right and opportunity to do so.

>If this is prohibitively expensive, or time consuming,

At last, you mention economic cost! That might yet save your D!

>then it is better to leave the process unchanged than to introduce the selective recounting bias.

"Better" from what point of view? Better to pay for recounting where it doesn't have any reasonable chance of changing the election outcome? Guess your point of view here is from within the ivory tower, all right.

>More generally, one of the principal lessons of macroeconomics is that rules generally work better than discretion. This is as true in elections as any place else.

Okay. The rule is that each candidate has equal right and opportunity to request recounts or otherwise challenge election results.

>Giving candidates influence over how election results are processed does not help democracy to accurately reflect the will of the people.

Repeat: The rule is that each candidate has equal right and opportunity to request recounts or otherwise challenge election results.

That's equal choice.

Democracy does not compel candidates to make equal decisions among their equal choices.

>judges will be responding to cases selectively filed by candidates.

Duh!

>Furthermore, judges determining elections will exalt the judiciary to a king-making role it should not have.

Judges carry out the law. Laws govern elections. Carrying out the law doesn't determine an election result -- the voters did that. Carrying out the law (we're assuming here that the law is well-designed, impartial, and so forth) simply determines how the voters voted.

>it is also clearly wrong to selectively recount certain areas.

Glaeser's article clearly is not an objective, balanced exploration of all points of view about the topic of selective recounts. It is a persuasion piece, legitimate as long as everyone recognizes that and doesn't pretend otherwise.

-- No Spam Please (nos_pam_please@hotmail.com), November 13, 2000.



"In Florida, the Bush campaign survived the automatic recount by a mere 320-some-odd votes. They have decided that any attempt at a recount, in any county, could easily wipe out that lead and is not worth undertaking."

Whoa, now! That's quite a leap there, Brian. Like I said on the other thread, assuming the same recounting methods as PBC were used, the outcome of the recounts should be somewhat predictable: in the four heavily Gore-leaning counties, Gore comes out ahead; in some number of Bush leaning counties with populations and percentages equal to Gore's choices, Bush comes out similarly ahead; statewide... flip a coin.

Referring back to my earlier thread, what I was "missing" watching the news coverage over the weekend was:

1) Any discussion of the fact that hand recounting the four counties chosen by themselves logically skews the election results.

2) Any discussion of why the Bush campaign didn't request a like number of recounts.

Reading the WSJ piece above makes me feel at least a little bit better...

-- RC (randyxpher@aol.com), November 13, 2000.


>> Whoa, now! That's quite a leap there, Brian. Like I said on the other thread, assuming the same recounting methods as PBC were used, the outcome of the recounts should be somewhat predictable <<

Ah! But would you stake your chance at the presidency on "somewhat predictable", if you thought you could drive a stake at a point where you were certain to be ahead by 320-some-odd votes?

The Bush people were faced with a fork in the road: they could go for competing recounts and hope the gains in Bush's counties would offset Gore's, or try to halt the process at a point where they had the lead. Bush could not logically go for both. Halting Gore's recounts would require arguments that could not be supported if Bush were doing the same thing.

Bush apparently chose to try to nail last Friday's numbers to the wall, let Gore file for his selective recounts, then stop them before they could be certified.

On Saturday he tried to stop them using the Federal courts. Today he tried to stop them using the Florida Secretary of State's discretion. A short time ago I predicted that Gore's campaign would go to court today to overturn the Secretary of State's decision. They did.

I now expect both cases to go to appeal, regardless of their outcome, as high as they can go - to the Florida Supreme Court and the US Supreme Court, if those courts can be persuaded to hear them.

So far, I'd speculate that Bush's efforts to stop Gore's recounts won't succeed. However, there is no telling if the selective recounts will turn up enough Gore votes to prevail in the end anyway. If I were a betting man, I'd give a slight edge to Gore as of today.

I think Bush was seduced by the idea that he could stop the recount death march by manuevering. Gore had no choice but to go out and rescue every vote he could and I think in the end that strategy is the one that will pull his chestnuts out of the fire. People want to hear the final numbers. They want to know that it was their votes that elected the President, not lawyers. Judges will sense that and lean over backward to let the votes be counted.

-- Brian McLaughlin (brianm@ims.com), November 13, 2000.


I agree with almost everything you've just said, Brian. I knew deep down inside we weren't on totally different pages.

All I can figure in regards to the Bushman's motivation for not pursing recounts is that anything beyond the current status quo is a coin flip. If he can freeze things where they are, he wins - if not, he gets to whine about the unfairness of the recount and play the martyr.

And if I were a betting man, my money would be on Gore right now, too. There's going to be more and more pressure each day to settle this, esp. if the markets continue to react - I'd also bet we have an answer by this Sunday at the latest.

Having said that, I'm not sure I like a system where there is a CHOICE to be made regarding whether we look at the data in a logically biased way or not. But like the ballot "problems", that's an issue for future discussion and change. Right now, you gotta play it as it lays...

-- RC (randyxpher@aol.com), November 13, 2000.


Sorry Brian, this is not true. "Today he tried to stop them using the Florida Secretary of State's discretion" It's not at the discretion of any one. The law is clear that the recount must be done one week after the election. No discretion to it. The Sec of State is just following the law. Bush has nothing to do with this Floridian non-partisan law. Democraps cry foul where this is none.

-- Maria (anon@ymous.com), November 13, 2000.

>> The law is clear that the recount must be done one week after the election. No discretion to it. <<

I may have been misled by Flint when he said in another thread that "legally it was Harris' decision to make." To me this means Harris had discretion. Now you say there was no discretion. Someone is right about that.

I presume that, if you are right and no discretion is allowed to Harris by Florida law, then Harris's action will be upheld and the court challenge will flop - big time. ;) If we reach that point, then things may begin to look dark for Gore. His chances will hang on the overseas vote and the overseas vote is a slender thread for him.

If the Florida court rules that discretion in this matter is allowed, then I assume it would be allowed because the lawmakers envisioned circumstances under which discretion would be needed to preserve the integrity of the election process.

If we ever do reach that point, Harris would not be able to plead that her hands were tied by the law and she would need to show cause for her action. That's when things could get, uh, very interesting.

-- Brian McLaughlin (brianm@ims.com), November 13, 2000.



The Presidential election snafu has resulted in at least one positive, namely, the insightful analysis that has graced this thread and others on our forum.

-- David L (bumpkin@dnet.net), November 13, 2000.

>>Hello gang!

>>Hey, want to be able to tell a No Spam Please post without looking ahead for the name?

>>Just look for lots of these things>>

>>No Spam cannot post without them>>>

>>

>>

>>

>

> >>> >>>

>>>>

>>>

>> >>> >>

>>> >>

>>

-- Cannot answer without>>> (no@spam.please.think), November 13, 2000.


Brian:

Discretion is always possible. I don't believe there is a single law ever adjudicated that permitted no discretion under any circumstances. Harris does have the authority to override existing Florida law if she feels a powerful enough case can be made that doing so is warranted. Perhaps demonstrable widespread fraud would qualify. Perhaps a hurricane would qualify. *Certainly* being a member of the party that stands to gain by suspending current law would qualify [grin].

But no such circumstances apply here. This election is being contested *solely* on the grounds that it is so close that if a recount is biased enough, it can swing the election. Nobody has yet found any reason for suspicion of foul play on anyone's part. Gore's tactic of recounting ONLY in the very limited cases where he stands to gain by the observed partisan farce of the process, has the advantage of nearly guaranteeing his objective (hey, let ME pick where and how to count, and do you think the results would vary?). It has the disadvantage of taking a very long time as a committee ponders each ambiguous ballot trying to read the voter's intent.

Consider that the entire state was counted twice in 2 days. What's taking all this time is the introduction of howling bias, of both Glaeser's description and the kind noted by the direct observers.

I must say, though, I think No Span is hilarious. Here he is saying out of one side of his mouth that manual counting is more accurate, and out of the other side that Bush should have countered by introducing his OWN biases and inaccuracies! I guess the though that winners don't demand recounts has yet to occur to him. He's too busy saying a Harvard professor doesn't know what he's talking about!

-- Flint (flintc@mindspring.com), November 13, 2000.


I find "No spam please's" essays a joy to read.

Why make fun of them?

His/her reasonable and calming explanation to 'eve' about the possible consequences of recounting votes was outstanding.

-- Pam (Pam@j udge.of election), November 13, 2000.


>> This election is being contested *solely* on the grounds that it is so close that if a recount is biased enough, it can swing the election. <<

The bias, as noted by Glaeser, by myself and by No Spam, is introduced solely by the selection of which counties to recount, not by the actual methodology used when counting. Bush did have the opportunity to select his own counties to recount, but declined.

I am clarifying, because you might have made this clearer than you did.

>> Gore's tactic of recounting ONLY in the very limited cases where he stands to gain by the observed partisan farce of the process...<<

Actually, the process is fairly well-designed to minimize partisanship in the actual determination of how a ballot cast it's vote. Counting by machine introduces machine failures. Counting by people introduces human failings. However, the process you find farcical embodies much of the same "checks and balances" approach favored by the authors of the US constitution.

>> I must say, though, I think No Span is hilarious. <<

I disagree. His analysis has been fairly dispassionate and balanced, although it does have a discernable tinge of bias in favor of Gore's arguments. Your own bias is also detectable.

>> Here he is saying out of one side of his mouth that manual counting is more accurate, and out of the other side that Bush should have countered by introducing his OWN biases and inaccuracies! <<

And he is correct. A perfect count is not possible, either by machine or by human judgement. However, the opportunity to introduce a competing bias DOES tend to cancel out the bias competed with.

>> I guess the though that winners don't demand recounts has yet to occur to him. <<

This is a frankly partisan comment. No one has won. For you to declare the winner is merely a declaration of your personal bias.

>> He's too busy saying a Harvard professor doesn't know what he's talking about! <<

Dr. Timothy Leary, who advised all of us to, "Turn on. Tune in. Drop out." was a Harvard professor. So was Baba Ram Dass, if I recall correctly.

-- Brian McLaughlin (brianm@ims.com), November 13, 2000.



Eve:

Mr. Glaeser is a professor of economics at Harvard University and a visiting fellow at the Brookings Institution.

May well be true, but he isn't a great statistican. His argument only works if the probability of error is the same for all data sources. But then you knew that.

Best wishes,,,,

Z

-- Z1X4Y7 (Z1X4Y7@aol.com), November 13, 2000.


Brian:

[The bias, as noted by Glaeser, by myself and by No Spam, is introduced solely by the selection of which counties to recount, not by the actual methodology used when counting. Bush did have the opportunity to select his own counties to recount, but declined.

I am clarifying, because you might have made this clearer than you did.]

I'm not sure I understand this clarification. I consider that there are two sources of bias. One indeed lies in the narrow selection of ballots most likely to change to Gore votes simply because random machine-nonreadable ballots in those districts have a 2-1 probability of being for Gore. The other is the evident difficulty of "determining the intent of the voter" by methods which change as the day goes on, using shifting definitions, new terminology, and the like.

[the process you find farcical embodies much of the same "checks and balances" approach favored by the authors of the US constitution.]

I guess I'm just suspicious that the checks don't balance in this case.

[I disagree. His analysis has been fairly dispassionate and balanced, although it does have a discernable tinge of bias in favor of Gore's arguments. Your own bias is also detectable.]

What careful discernment! [grin]

[And he is correct. A perfect count is not possible, either by machine or by human judgement. However, the opportunity to introduce a competing bias DOES tend to cancel out the bias competed with.]

The machines are biased? Or do you mean that Bush should demand recounts equally carefully tailored to favor his side? Yes, if the rules of the game have changed from "what did the voters really say?" to "How can my distortions legally exceed HIS distortions?" then of course that's what Bush should have done.

But No Spam was vehemently arguing over the weekend that manual counts by partisan counters would be MORE accurate than the machines, NOT that their biases could cancel out! And here he is admitting that the LESS accurate and more biased human count is preferable. But this is situational...

[This is a frankly partisan comment. No one has won. For you to declare the winner is merely a declaration of your personal bias.]

And this is the situation. I did not declare a winner of the election, I *observed* a winner of both machine counts as they stand so far, recognizing that these are incomplete pending the absentee ballots, which can't be counted until they arrive. However, this status is clearly critical, eager as you and No Spam seem to be to pretend otherwise. It is a status which places the party ahead in these counts in a position of having everything to lose and nothing to gain by any further action.

After all, leading by a wider margin makes no difference. A win by 1 vote or a million is a win, and counts the same. Conversely, the trailing party has absolutely nothing to lose by pulling every stunt, trick, technique, or strategem permitted by law, in an effort to reverse this status. So the situation is very binomial. Any support for further counts can ONLY work in favor of Gore and against Bush, and vice versa.

[Dr. Timothy Leary, who advised all of us to, "Turn on. Tune in. Drop out." was a Harvard professor. So was Baba Ram Dass, if I recall correctly.]

Who were promptly fired, as I recall. But this gets us to:

Z:

Yes, of course Glaeser's statistics assume equal error distribution around all data points. But why is this wrong? We have no indication that this assumption is false, and quite a bit of data suggesting it's true. The variance among counties was quite high (most with little or no change, a few with a LOT of change), and the direction of change wasn't overwhelming. I think you're trying to say that the Big Errors were limited to what Gore is recounting, yet we know otherwise.

In any case, after TWO machine counts, we're looking at a fight not at the opening bell, but entering the last round with one fighter WAY ahead on all cards. The other fighter needs a knockout, and has nothing to lose, the first fighter's best tactic is to run and duck. As I said before, how much someone wins BY is irrelevant.



-- Flint (flintc@mindspring.com), November 13, 2000.


Maria, FYI. This is taken from an article from Reuters news service, posted in the thread: FLORIDA WILL NOT EXTEND!

>> Debby Kearney, general counsel for the secretary of state's office, acknowledged that state law gave Harris discretion to extend the deadline but said she refused to do so because the counties had given no adequate reason why they could not comply. <<

Apparently, general counsel for the Secretary of State disagrees with your position that Florida law allows no discretion to the SOS in this matter.

-- Brian McLaughlin (brianm@ims.com), November 14, 2000.


No Spam Please,

I still don't see how Glaeser's points aren't indeed objective and persuasive in a non-partisan way. He's simply urging us to look at the big picture, pointing out universal statistical principles that would apply no matter who the candidate was. I may have overlooked a point you were trying to make, but I really don't follow how his view is partisan, because the same principles would apply if Bush had adopted Gore's tactics. Perhaps you could elaborate.

I don't take issue with your points about the pragmatic motives by the candidates -- as long as Gore abandons any pretense at a fair result, and that his obviously outrageous "the will of the people" lie is retracted. If he wants to talk about "the will of the people" he should restate it truthfully in terms of his aims and actions, i.e., "the will of certain carefully selected people, only to the extent that I win."

And I'm not necessarily supporting Bush in this, as I believe he'd probably adopt the same tactics if the situation were reversed.

Your emphasis on the the importance of cost considerations in recounts is a very good point; I agree completely.

Z,

I support Flint's response re the assumption of equal error distribution.

-- eve (eve_rebekah@yahoo.com), November 14, 2000.


Brian, "acknowledged that state law gave Harris discretion to extend the deadline but said she refused to do so". The discretion which the law states refers to natural disaster not failure to act quickly on the part of a candidate. That's the only allowances that can be made. So barring a hurricane or snow storm, the count stands.

The is all OBE anyway. The county commissioner has decided by 3 to 1 vote to stop all manual counts. It has been determined to be illegal to do the manual counts. Only if the machines were not working correctly could a recount be done. Again the law is pretty clear.

-- Maria (anon@ymous.com), November 14, 2000.


>> The discretion which the law states refers to natural disaster not failure to act quickly on the part of a candidate. <<

I am not aware of any failure on the part of Gore to act quickly. AFAIK, he acted less than a day after the state-mandated recount was completed and the result announced.

As far as your interpretation of Florida state law goes, I presume your interpretation carries about as much weight as mine (meaning: none). A judge will be making this determination quite soon. I would expect the judge's decision to be appealed, regardless. We'll see.

As I said earlier, I believe that most judges will interpret the law in the way that is most favorable to honoring the rights of voters. I still believe this case favors Gore's position over Harris', in that no truly compelling public interest is served by enforcing a deadline that interferes with ongoing vote counting.

-- Brian McLaughlin (brianm@ims.com), November 14, 2000.


"That's the only allowances that can be made."

Welcome to the World of Maria, where absolutes reign supreme, and where her iron-clad mind frolics behind closed doors.

"The county commissioner has decided by 3 to 1 vote to stop all manual counts."

In the World of Maria, objective news sources don't matter. What is factual is what Maria saw on TV. Even if the New York Times says the vote was 2 to 1, Maria says it's 3 to 1. That's the Logic of Maria.

"It has been determined to be illegal to do the manual counts. Only if the machines were not working correctly could a recount be done. Again the law is pretty clear."

In the World of Maria, law are unambiguous and never subject to varying interpretation. Everything is "clear." In the World of Maria, reality is all SO SIMPLE!

Welcome to the narrow and one-dimensional World of Maria, a cartoon of the Republican party.

-- World of Maria (bush@league.net), November 14, 2000.


Hawk had decided. All should stop discussing now. The NEW YORK TIMES is gospel, the law has spoken. Thanks Hawk for showing us your hairy butt.

-- Maria (anon@ymous.com), November 14, 2000.

"All should stop discussing now."

Maria's motto!

In the World of Maria, Maria is always right, even when it can be proved Maria is wrong. When Maria is shown how she's wrong, she resorts to foul language.

Remember, in the World of Maria, what Maria saw on TV is more accurate than what the New York Times reports.

-- Bright & Sane World of Maria (bush@league.net), November 14, 2000.


Flint,

Now you've done it again -- attributing ideas to me that I did not express.

Didn't what you and I just went through on another thread persuade you to be careful about that? What's wrong with you lately?

>Here he is saying out of one side of his mouth that manual counting is more accurate, and out of the other side that Bush should have countered by introducing his OWN biases and inaccuracies!

Flint, why did you decide to twist my saying that Bush should have requested manual recounts, which I said were more accurate, into somehow meaning that Bush should have introduced his own biases and inaccuracies? By requesting manual recounts, Bush would have been requesting a _more_ accurate count, not a less accurate one.

Now you need to apologize again.

>I guess the though that winners don't demand recounts has yet to occur to him.

Flint, in my first November 12 posting on that other thread, I wrote, "After all, recount procedures are likely to be requested only by losing sides, ..." But maybe it didn't occur to you to read that.

>He's too busy saying a Harvard professor doesn't know what he's talking about!

Were you too busy composing some witty remark to think about checking whether what you were writing about me was true?

>But No Spam was vehemently arguing over the weekend that manual counts by partisan counters would be MORE accurate than the machines, NOT that their biases could cancel out!

I wrote that the manual recounts were designed to be more accurate than the preceding counts, and that they would be inspected by both Democratic and Republican observers.

>And here he is admitting that the LESS accurate and more biased human count is preferable.

I wrote that the human count would be MORE accurate than the preceding machine counts.

-- No Spam Please (nos_pam_please@hotmail.com), November 14, 2000.


eve,

>I still don't see how Glaeser's points aren't indeed objective and persuasive

Re: "objective" -- Because he leaves out important stuff that renders his conclusions invalid.

Re: "persuasive" -- That's what I'm saying, that his article is in the category of persuasion rather than objectivity. Nothing ethically wrong with that, _as long as everyone's clear on what's what_.

>He's simply urging us to look at the big picture, pointing out universal statistical principles that would apply no matter who the candidate was.

He's being _too_ simple. He's urging a look at only _part_ of the "big" picture. He omits mentioning some aspects of elections, aspects to which statistical principles do not apply.

>I really don't follow how his view is partisan,

Well, I never said Glaeser's view was partisan.

Do a word search for "partisan" starting at the top of this thread. It doesn't appear before Flint used it in his post. It doesn't appear in any of my postings before now except where I quoted Flint.

_Now_ that we're discussing that aspect, however, I'll say that Glaeser's argument seems designed to persuade the reader to accept a conclusion that currently happens to favor one party's candidate over another's. Maybe Glaeser's motivation is partisan, maybe it isn't. But I wasn't arguing that earlier.

>because the same principles would apply if Bush had adopted Gore's tactics.

If you mean the same statistical principles, then yes, of course.

But I'm pointing out that Glaeser is not mentioning important non-statistical considerations that make his statistics-based conclusion invalid for our real-world situation. And that would still be true if the Bush/Gore situation were reversed.

>I don't take issue with your points about the pragmatic motives by the candidates

My comments in this thread before today were about Glaeser's article, but now I'll respond to your comments about a candidate.

>-- as long as Gore abandons any pretense at a fair result, and that his obviously outrageous "the will of the people" lie is retracted.

What did Gore say about "the will of the people" that was an obviously outrageous lie? I don't catch all the news stories.

Should Bush abandon any pretense at a fair result, as well as Gore, in your view?

-- No Spam Please (nos_pam_please@hotmail.com), November 14, 2000.


No Spam:

We must agree to disagree in this case. My experience has been that manual procedures tend to show much wider variation than mechanical procedures, and tend to depend distressingly strongly on the orientation of the person performing the procedure. I suggest you might read Gould's book The Mismeasure of Man for some insights into this process.

I should think the observations we have read, about the frequency with which honest ballot interpreters of different orientations disagree about "the intent of the voter" on a particular ballot, should alert you. The machines had a variance of less than 1%, and the human disagreement is on the order of 5-10% by observation. This can be called "more accurate" only with tongue firmly planted in cheek. By observation, this is the Great Kreskin approach to "accuracy".

I roared with laughter at your sputtering response to Glaeser's article. Yes, Glaeser's analysis assumed roughly equal error of measurement across all precincts, which may be a false assumption. But his overall claim that a process selectively applied introduces bias, *even if* it's more accurate, is still correct.

In any case, the basic point you are very carefully evading is that no recount can help Bush. Bush has already come out ahead on two prior counts, and has nothing to gain and everything to lose by further counts of any kind. Therefore, prima facie, ANY support for ANY further count by ANY means of ANY portion of the vote is a pro- Gore position.

So your embarrassing eagerness to agree with what is conceded even by Gore to be *intended* to bias the count in his direction, is itself political fervor. Your claim that this clearly biased sample (in precincts that went 2-1 for Gore, so all ambiguous or non-machine readable ballots are ALSO likely to be 2-1 for Gore) is *more accurate* is simply fatuous. I know better, Bush knows better, Gore knows better, and most of all, I'm sure YOU know better. And you want *me* to apologize again? I respectfully suggest it's your turn this time.

Now, I can admire Gore's tactics here. We know what he's doing, we know why, we know what he stands to gain and we know HOW he stands to gain it. What makes it such a neat trick is, the public at large doesn't understand statistics and probabilities. Even if a manual count of these votes really IS more accurate, the real difference is made through the *selection* of votes to count, not how accurately they are counted.

Do you recall that one of the first pre-election polls picked Alf Landon to defeat FDR handily? The pollsters wrote down the responses with perfect "accuracy" and analyzed their data honestly. Problem was, the poll was done by telephone, which only the rich could afford at the time. And the rich favored Landon. Meanwhile, FDR won the election by the largest margin in history up to that time. And the editors of the Atlantic Monthly who performed the poll were smart people, they just didn't understand statistics.

And here you are, supporting *exactly* this type of bias, on the grounds that it's "more accurate"! And demanding that those who see through this should apologize. Your eagerness to paint Glaeser as an advocate and dismiss his argument is pathetic. If you know better, shame on you. If you don't, your best bet is to admit it.

-- Flint (flintc@mindspring.com), November 14, 2000.


Flint,

Now you're just increasing your distortions. Are your political views so important to you that you feel justified in distorting what I've written in order to make your points?

>My experience has been that manual procedures tend to show much wider variation than mechanical procedures,

People can determine the intent of a voter who has punched partly, but not all the way, through a punch-card ballot better han manchines, generally.

>and tend to depend distressingly strongly on the orientation of the person performing the procedure.

... which is why the recount procedures are designed to be monitored by party observers so that they might detect and protest improper decisions.

Why don't you ever acknowledge that there can be safeguards built into manual counting to make it more accurate, though slower and more expensive, than manchine counting, Flint?

>I should think the observations we have read, about the frequency with which honest ballot interpreters of different orientations disagree about "the intent of the voter" on a particular ballot, should alert you.

Why don't you ever acknowledge that there can be safeguards built into manual counting to make it more accurate, though slower and more expensive, than manchine counting, Flint?

>The machines had a variance of less than 1%,

I presume you mean that successive machine counts varied by less than 1%. So? If the machines failed, on each run, to count, say, 10% of the valid ballots, then their error in vote count is at least that 10%, regardless of how small the variance between runs.

Ever head of systematic error? You'd better, if you're going to consider yourself qualified to recommend books to me.

>and the human disagreement is on the order of 5-10% by observation.

Please state exactly what you mean by that.

>This can be called "more accurate" only with tongue firmly planted in cheek.

If you're using sufficiently fuzzy definitions, the "1%" and "5-10%" cannot reasonably be compared at all.

Please tell us exactly what your "1%" and 5-10%" figures refer to.

>By observation, this is the Great Kreskin approach to "accuracy".

Do you think you have just written some rational refutation of anything?

>I roared with laughter at your sputtering response to Glaeser's article.

Do you care to refute anything I wrote about it?

>Yes, Glaeser's analysis assumed roughly equal error of measurement across all precincts, which may be a false assumption.

I don't recall writing anything about an assumption of equal error of measurement. So, to what is your "yes" agreeing?

>But his overall claim that a process selectively applied introduces bias, *even if* it's more accurate, is still correct.

Well, if it's more accurate, then apparently the bias is toward accuracy. Do you see anything wrong with that?

You _do_ want an accurate count in Florida, don't you?

>In any case, the basic point you are very carefully evading is that no recount can help Bush.

(1) Glaeser didn't assert that. My earliest comments in this thread were about Glaeser's article.

(2) Do you not know that recounts might help Bush in states such as Iowa or Wisconsin? I do. Maybe you should not opine about what I'me evading or not.

Why are you "carefully evading" staying on subject, Flint?

>Bush has already come out ahead on two prior counts, and has nothing to gain and everything to lose by further counts of any kind.

Do you not know that recounts might help Bush in states such as Iowa or Wisconsin? I do.

>Therefore, prima facie, ANY support for ANY further count by ANY means of ANY portion of the vote is a pro- Gore position.

Apparently you are unaware that Bush lost some states by small margins. Does support for a recount in Iowa or Wisconsin or New Hampshire constitute a pro-Gore position?

>So your embarrassing eagerness to agree with what is conceded even by Gore to be *intended* to bias the count in his direction, is itself political fervor. Your claim that this clearly biased sample (in precincts that went 2-1 for Gore, so all ambiguous or non-machine readable ballots are ALSO likely to be 2-1 for Gore) is *more accurate* is simply fatuous. I know better, Bush knows better, Gore knows better, and most of all, I'm sure YOU know better. And you want *me* to apologize again? I respectfully suggest it's your turn this time.

Flint, if you're drunk or high, maybe you ought to sober up before posting again in this thread.

>And here you are, supporting *exactly* this type of bias,

No, I'm not.

Please sober up.

-- No Spam Please (nos_pam_please@hotmail.com), November 15, 2000.


NoSpam to Flint: >> Are your political views so important to you that you feel justified in distorting what I've written in order to make your points? <<

No Spam,

Personally, I have found Flint more than willing to distort my words, to misrepresent my views, and to draw unwarranted inferences about what I said or what I meant, if he thought it would advance his argument, by fair means or foul. You may either choose to respond to these tactics, or not.

I have chosen each course at various times. When he gets too far out of hand, I suggest you whack him on the nose with a rolled-up newspaper as vigorously as you think necessary. He tends to quiet down his, uh, tendencies if you repeat this treatment several times in a short period.

-- Brian McLaughlin (brianm@ims.com), November 15, 2000.


102.112 Deadline for submission of county returns to the Department of State; penalties.--

(1) The county canvassing board or a majority thereof shall file the county returns for the election of a federal or state officer with the Department of State immediately after certification of the election results. Returns must be filed by 5 p.m. on the 7th day following the first primary and general election and by 3 p.m. on the 3rd day following the second primary. If the returns are not received by the department by the time specified, such returns may be ignored and the results on file at that time may be certified by the department. keywords here are "may be"

102.166 Protest of election returns; procedure.--

(d) The manual recount must include at least three precincts and at least 1 percent of the total votes cast for such candidate or issue. In the event there are less than three precincts involved in the election, all precincts shall be counted. The person who requested the recount shall choose three precincts to be recounted, and, if other precincts are recounted, the county canvassing board shall select the additional precincts.

(5) If the manual recount indicates an error in the vote tabulation which could affect the outcome of the election, the county canvassing board shall:

(a) Correct the error and recount the remaining precincts with the vote tabulation system;

(b) Request the Department of State to verify the tabulation software; or

(c) Manually recount all ballots.

102.168 Contest of election.--

(c) Receipt of a number of illegal votes or rejection of a number of legal votes sufficient to change or place in doubt the result of the election.

Just a few observations..This thing is NOT over and will not be for quite sometime. This will drag-on in the courts for who knows how long. Supreme Court anyone?

From my readings of the Florida Statutes and the Texas Election Statutes, it is pretty clear to me the Dems are taking Bush and company on a ride thru their own Rhetoric, their own laws they signed and supported. They(the Dems)understand the Florida law and counted on Harris doing what she did yesterday. Now the real actions can proceed. Repubs are falling right into a trap and can't help themselves but fall into it.

It would not surprise me in the least if the Dems KNEW them butterfly ballots would produce x amount of errors resulting in a tainted election in PB. All of this just seems a bit too "perfect" for my tastes.

-- Doc Paulie (fannybubbles@usa.net), November 15, 2000.


bold off sorry

-- Doc Paulie (fannybubbles@usa.net), November 15, 2000.

off I say!

-- (fannybubbles@usa.net), November 15, 2000.

No Spam:

Let's try again. I'm willing to keep trying until you make even a token effort to address the points I raise. You (and Brian) seem content to call names at those with whom you disagree, and then sit back and think you've made points. You can do better than this.

[People can determine the intent of a voter who has punched partly, but not all the way, through a punch-card ballot better than machines, generally.]

This is a bald assertion. Now, how do we determine the correctness of this assertion? Well, we can either repeat it like a mantra (your approach), or we can observe what has actually happened when this approach is tried. We have the nominal definition that a hole should be punched to constitute a vote. But it isn't, completely. Now, did the voter *intend* to punch a hole?

We have several issues here: If the voter *intended* to punch a hole but failed, should that vote count? (Much like saying if the batter *intended* to hit a home run but instead fouled out, should we count it as a home run based on intention?) And if we decide that an invalid vote should count because this is what someone *thinks* the voter *probably* wanted, then who gets to make the decision as to what the voter probably thought? What we've seen in practice as the manualcounts continue, is growing stacks of ambiguous ballots. These are "ambiguous" because, with humping consistency, the democratic interpreter thinks they are Gore votes and the republican interpreter thinks they are not. And both of these interpreters are completely sincere and doing their best to be "objective." But the underlying goal is politics -- victory, not accuracy.

This *still* hasn't got to the issue of how we go about deciding which ambiguous ballots should be interpreted and which should not. We'll get there later. Meanwhile, permit me to doubt that your passion for "accuracy" is what motivates you to put in the effort you are just to know "for certain" that, say, Bush really won by 3000 rather than 300 votes! How many times do I need to point out that since the margin is irrelevant, recounts can ONLY help whoever is currently behind. We'll get nowhere so long as you can't admit that your goal is to see the current temporary status reversed, no matter HOW many names you can think to call people.

[Why don't you ever acknowledge that there can be safeguards built into manual counting to make it more accurate, though slower and more expensive, than manchine counting, Flint?]

Acknowledging that this is possible is not quite the same as observing that *therefore* it is being done. It is *possible* to add safeguards to any such procedure. Now, are the safeguards actually *working*? Why can't YOU acknowledge that when a process is performed by people strongly committed to a given result, safeguards have a history of failing in one way or another?

Yet when I pointed out what we have actually observed, that rules are changing on the fly, that votes about those rules fall along party lines, that debates about particular ballots fall along party lines, that debates about where and whether to count votes fall along party lines, and that these OBSERVATIONS are a clue to the political nature of this process, you repeat your same accusation that I'm somehow failing to acknowledge a theoretical possibility.

I adknowledge that our metaphorical batter *could* have hit a home run. In practice, he fouled out.

[If the machines failed, on each run, to count, say, 10% of the valid ballots, then their error in vote count is at least that 10%, regardless of how small the variance between runs. Ever head of systematic error? You'd better, if you're going to consider yourself qualified to recommend books to me.]

Well, minus the gratuitous insults, we have another theoretical possibility here, depending again on a definition. What do we mean by a "valid ballot"? In practice, for logistical reasons, a valid ballot is one the tabulating machine can read consistently. So in practice, your hypothetical situation violates this definition -- if the machine can't read it, it's not a valid ballot.

And again in practice, we *observe* that this category of ballots (partially punched, dimpled, pregnant, etc.) fails to meet the strict legal requirements to constitute a "vote". Now, should we let people examine them so as to deduce what they think the voter may have intended? Well, that depends on who you ask. Is it all that surprising that those with everything to gain say yes, and those with everything to lose say no?

[>and the human disagreement is on the order of 5-10% by observation.

Please state exactly what you mean by that.]

From what I've read, human interpreters have disagreed on about 5-10% of non-machine-readable ballots, and these disagreements fall along party lines (surprise).

[>By observation, this is the Great Kreskin approach to "accuracy".

Do you think you have just written some rational refutation of anything?]

I have tried to find an accurate and colorful description of a process we have seen to be steeped in political prejudice, in the sense of defining the process in such a way as to produce the predetermined results. A good part of the battle consists of each side fighting to have their definition of a "vote" accepted, because the difference in counts based on differences in definitions (all of which are perfectly reasonable) determines the ultimate winner. Do you think that simply dismissing this point somehow makes it go away? If different reasonable definitions produce different winners, which one is more "accurate"? Who gets to pick?

[Well, if it's more accurate, then apparently the bias is toward accuracy. Do you see anything wrong with that? You _do_ want an accurate count in Florida, don't you?]

OK, I'll take you at face value. Here we go...

Glaeser's point was that "cleaning up" ONLY the data points you don't like, while leaving alone those you are happy with, distorts the final picture. And that this is NOT a bias toward overall accuracy UNLESS this same procedure is applied to the entire population of raw data. In this case, we are applying the "cleanup" process ONLY where it is likely to do Gore the most good and NOT cleaning anything where it is most likely to help Bush. Maybe I am misunderstanding you, but I cannot believe you are making the claim that the count "in Florida" becomes more accurate when those uncounted votes likely to run 2-1 for Gore are reinterpreted, and those uncounted votes likely to run 2- 1 for Bush are NOT reinterpreted. To me, this selection is highly and obviously biased. As I said, everyone knows this.

So no, I don't want an accurate count in Florida, I want a count that favors Bush. We have one, so I want no further counts. At worst, I'd accept the same procedure being applied statewide. But that's been done twice, and YOU claim that it's inaccurate because Gore lost them. YOU would like a count that favors Gore, and I can respect that. The currently debated procedure is NOT designed to produce an accurate count "in Florida" at all, it is designed to use probabilities to bias the count in Gore's favor. If this were not the case, Gore would not be so desperate to continue with it. Your continued insistence that this bias is "more accurate" does you no honor. You aren't really arguing for accuracy. And I just can't understand how you can claim BOTH that Bush could have insisted on contervailing distortions (which is true), and ALSO that these recounts are making the results more accurate and not more distorted!

George Orwell would have appreciated your notion of "accuracy in Gore's favor" which might be countered by "accuracy in Bush's favor". If it's in anyone's favor, it's a bias. Duh!

[Why are you "carefully evading" staying on subject, Flint?]

Have you stopped beating your wife? The subject here is, should we keep counting (in different places and with changing rules) until Gore wins? The subtext is, *how can we justify* this strategy so as to cover our ass and make the other side look evasive?

Your schtick is to claim "greater accuracy", despite the observed political nature of the counters and their rules, and the statistical principle that selective cleanup produces very different overall results depending on WHO gets to do the selection. And you "support" this claim by calling those who observe otherwise fools or drunks. Proud of yourself?

[I don't recall writing anything about an assumption of equal error of measurement. So, to what is your "yes" agreeing?]

I am agreeing to the only valid statistical objection raised to Glaeser's argument. You did not address the statisics at all, and thus raised NO valid objections. You merely obfuscated. Go back and read your response to Glaeser. Essentially, you deny that statistical methods can be appropriately applied to a political process no matter how numerical. You end up concluding that *because* both parties could legally have distorted the count, therefore there is no distortion when only one party does so! Astounding!

[Apparently you are unaware that Bush lost some states by small margins. Does support for a recount in Iowa or Wisconsin or New Hampshire constitute a pro-Gore position?]

Well, I tried to be clear. I'll try again. The first count in any jurisdiction that didn't produce an exact tie, produced a winner. The winner has NOTHING TO GAIN and EVERYTHING TO LOSE by a recount within that jurisdiction. Support for a recount is support for the loser. Is this really *that* hard to understand?

Brian:

What I try to do is understand what you and everyone else is driving at. If you make a point and I refute it, you have the option of clarifying your point (if the refutation missed the point), answering the refutation (if you can), or accusing me of distortion (if you find yourself backed into a corner). (As a random example, supporting fines for polluting but opposing pollution licenses is a self- contradictory position. Saying the first is functionally equal to the second is NOT a distortion, it's an unpleasant observation. Whoever said the first said the second *whether they wish to admit it or not*. It's not always my problem that you don't like the implications of your own positions).

Now I ask you. Glaeser pointed out that applying one set of rules some places and another set of rules elsewhere biases results, and that if you get to select *where* to apply the new rules and *what* those new rules are, you can often change the outcome significantly. No Spam (as far as I can tell) said Glaeser was wrong. Now, this is odd. You don't need to be a rocket scientist to see that Glaeser's point is self-evident. No Spam is foursquare in favor of selective recounts.

Now, does this position support self-evident bias? Of course it does. How does No Spam defend this support? He flat denies it ("No, I'm not"), and accuses me of being drunk. This is support? Is it my fault that NSP believes that if he denies the clear implications of his positions, those implications will somehow not be there?

When I go back and read his answer to Glaeser, NSP doesn't address selective rule changing at all. He talks about the *cost* of recounts, which Glaeser never mentioned and which has NOTHING to do with statistical principles. He claims repeatedly that Bush could have pulled the same stunt, which while true has nothing to do with statistical principles. He argues that because both Bush and Gore could legally have chosen to bias the results, therefore Glaeser's *principle* is either a lie or ignorance! You will note that No Spam has NOT addressed Glaeser's statistics even once. The *principle* cannot be refuted, only obfuscated.

Finally, NSP is reduced to calling Glaeser an ignorant liar and me a drunk. Are we supposed to admire this "support" for a logical position?



-- Flint (flintc@mindspring.com), November 15, 2000.


Eve and Flint:

If you are correct, then I agree. I just don't have enough information. What I have suggests that the error rate for "Opti-scan" is different than punch card [we got rid of them 2 decades ago]. Anyway without this consideration, the presented agrument is nonsense. If you have information that I don't, I concede the argument.

Bye-the-bye, eve; "There are three kinds of lies: Lies, damn lies, and statistics." This is usually attributed to Diz. You know, the PM of England, in the 1800's.

Best wishes,,,,

Z

-- Z1X4Y7 (Z1X4Y7@aol.com), November 15, 2000.


One more thing, eve:

You aren't considered credible, just because you agree with Flint. Give some data; this is an argument about statistical methods.

My comment was about the original post being acceptible; it is not without the discussion of the appropriate variables. I have rejected better publications than this. It shouldn't be politics when discussing statistical approaches.

Best wishes,,,,,

Z

-- Z1X4Y7 (Z1X4Y7@aol.com), November 15, 2000.


Z:

To quote myself from a post above...

[Glaeser pointed out that applying one set of rules some places and another set of rules elsewhere biases results, and that if you get to select *where* to apply the new rules and *what* those new rules are, you can often change the outcome significantly...Glaeser's point is self-evident.]

To me, one need not know the exact method used in every precinct to understand the statistical basis of Glaeser's argument. Cleaning up the data isn't something that can only be done with a specific type of data. It can be applied to ANY data, simply by adjusting the cleanup procedure to the nature of the data. Even counties that had NO variation between the two machine counts remain subject to cleaning up, which process could make a significant change in the results.

In the case under discussion, we have thousands of ballots that had no vote for President visible to the machines. This was true in counties with 2-1 margins for BOTH candidates. At random, one would expect reasonable interpretations of the voters' intentions to match the ratio within each such jurisdiction. Glaeser argues that applying this manual interpretation ONLY in counties likely to favor ONE of the candidates biases the statewide count. This is, again, self- evident.

By what reasoning do you claim this is nonsense? What additional data do you lack?

-- Flint (flintc@mindspring.com), November 15, 2000.


Warren Mitofsky's Florida exit polls had it right after all.

Florida voters went for Gore. It was the voting machines that were wrong when they threw out tens of thousands of ballots -- more than all of Ralph Nader's 92,000 Florida votes. And that's NOT counting the 19,000 ballots thrown out due to the voting irregularities caused by the Palm Beach County "butterfly" ballot. There were an additional 11,000 Palm Beach ballots thrown out for "under voting," i.e., no presidential vote registered on them because they were not completely punched through -- the "hanging chad problem," and the machines could not read them. With Bush leading Gore by less than 400 votes, the Bush machine is moving heaven and earth to stop the hand count of ballots that would, evidence shows, give the election to Gore.

The news media for the most part have not made it clear to the public that Palm Beach board is NOT proposing to count the "over votes," i.e., the thousands of votes that were punched twice due to the confusing design of the ballot -- despite Bush attempts to imply that they are.

Warren Mitofsky heads the Voter News Service (VNS), the exit poll operation subscribed to by most of the news media, which called Florida for Gore at 8:15 p.m. on election night based on interviews conducted with voters as they left the polls. "I've done over 3,000 elections over thirty years," the professorial Mitofsky told the Lehrer Newshour, "and have called it wrong only five times in all that time." Mitofsky, who has been taking a beating in the media, may be exonerated -- but he may not be if Bush's lawyers succeed in stopping the manual count. And they are playing for keeps.

Now we've learned the astonishing news that George W. Bush's first cousin John Ellis, a producer in the political shop at the Republican- dominated Fox News Channel, was the one who pronounced Bush the president-elect at 2:16 a.m., November 8, based on the erroneous vote count at that time. That gave Bush the advantage of seeming the dethroned but rightful king. Republicans and the media, led by Tim Russert and the partisan Chris Matthews joined Republicans to demand that Gore "concede gracefully." Yet, not until Tuesday, November 14, did the New York Times and the Washington Post spell out to their readers Ellis's questionable involvement in the naming of the next president of the United States, including being in constant communication with his cuzzes Jeb and W., Auntie Bar, and Unca George that night. The Times's story appears deep inside the paper, and in the Post, in the "Style" section. The major media are so frightened of being labeled "liberal media elites," by Republicans that they skew and modulate their reporting to avoid the charge.

James Baker, III ("the third stands for the three coverups he's handled for the Bush family so far, not including this one," writes Internet activist J.M. Prince), has been spinning madly, but even he may not be able to put this one over on the American people, despite his past love affair with the "elite" media. There's a lot of information out there. And, unlike the financial and legal details of Whitewater, this is one the public, and even the media, show evidence that they understand. Support for the process is shrinking a bit, however, mainly due to the successful PR efforts of the Bush cabal who relentlessly pound away with untruthful claims.

Baker says that manual counts are less accurate than machine counts, and hints at corruption among the Palm Beach County voting officials. But Florida election officials, and many others around the country vehemently dispute that, saying that manual recounts are often taken to resolve problems in close elections. It is well established that computer glitches give inaccurate counts for a variety of reasons, in the Florida case because of incompletely punched ballots that didn't register properly. And as Baker knows very well, the recount will take place with Democrats and Republicans and all manner of other officials monitoring it closely, as was the case with Palm Beach's earlier sample recount of under votes that showed significant differences with the machine recount.

Baker and Company hope to paint Gore as a whining poor loser by saying that Gore wants "recount after recount after recount." But the only recount taken so far was the machine recount automatically triggered by Florida election law because of the closeness of the original count. And that recount reduced Bush's lead from about 1800 votes to 327, an 85 percent drop.

Florida law allows the counties and Gore to ask for a manual count of the votes, and a sampling recount of several precincts began in Palm Beach County and two other counties on Friday. However, Bush asked the Federal District Court in Southern Florida to stop the recounts -- despite Bush's own request on Friday for a manual recount that was taken in another Florida county. (It helped Bush and Karen Hughes wouldn't say whether they would accept the votes or not.) The hypocrisy doesn't stop eminence gris Baker from trying to spin the country for the Bushes. The request to bar manual recounts was rejected by Federal Judge Donald Middlebrooks, but Florida's Secretary of State, Katherine Harris, a George W. Bush supporter who campaigned for him, ordered all voting results to be in her office by 5 p.m. Tuesday, November 14, which if upheld, would end any manual vote recount in Florida. And would, by all signs, hand the election to Bush.

Baker has said repeatedly that he "was in the room" when former President Gerald R. Ford decided he would not contest his 1976 defeat by Democrat Jimmy Carter. Carter had 297 electoral votes, however, and won the popular vote by 1,682,970 while Ford had only 240 electoral votes. Does Houston corporate lawyer Baker really think he can revise history before our very eyes. A veteran political reporter remembers, "That wasn't even much of a story at the time."

Today, Baker today is trying to fool the nation into believing that the lawsuits filed by Florida citizens were in fact filed by the Gore camp.

But there is a danger that Gore could lose the PR battle. Baker has been joined in that aspect of this political war by Ted Olson who argued and lost before Middlebrooks. Olson was the chief architect of the Machiavellian and successful conjoining of the Whitewater witchhunt and Paula Jones charges which led to Clinton's impeachment. Olson's efforts in Arkansas and in Washington were funded by the right-wing scion of the Mellon fortune, Richard Mellon Scaife, whose billions have funded every single one of the assaults on the the Clinton administration, including the ongoing effort by Larry Klayman of Judicial Watch to disbar Bill Clinton. (And Olson's wife Barbara was one of the most poisonous anti-Clinton TV commentators.)

This political drama closely resembles the uncertain playing out of Nixon's coverup of his Watergate crimes. No one knew whether the system would work in that crisis, but it did. Now we shall see if the system still works.



-- M.D. (m@d.net), November 15, 2000.


No Spam Please,

[...he leaves out important stuff that renders his conclusions invalid.]

Such as?

[...his article is in the category of persuasion rather than objectivity.]

Why do you feel you need to distinguish this? And do you feel that persuasion and objectivity are mutually exclusive? If so, please give your rationale.

Do you feel that, say, we're persuaded that the sun will rise tomorrow based on various objective information? Is Glaeser's article persuasive in that sense? If not, why not?

[He's being _too_ simple. He's urging a look at only _part_ of the "big" picture. He omits mentioning some aspects of elections, aspects to which statistical principles do not apply.]

Could you give me an example that would stand to refute his points?

[But I'm pointing out that Glaeser is not mentioning important non- statistical considerations that make his statistics-based conclusion invalid for our real-world situation.]

Again...for example?

[What did Gore say about "the will of the people" that was an obviously outrageous lie?]

In the context of the selected Florida hand recounts, he said, "There's something very special about our process that depends totally on the American people having a chance to express their will without any intervening interference." (New York Times, 11/14/00, page A19, column 2).

As I indicated in so many words above, the truth would have been something like, "a certain, selected, limited group of American people" instead of "...the American people..."

[Should Bush abandon any pretense at a fair result, as well as Gore, in your view?]

Well, that depends on what you mean by "fair." Seriously. (no Clinton jokes, please):)

Z,

First off, (if I read you right) I don't see how the probability of error of all data points has to meet some sort of Platonic ideal of exact sameness across-the-board, in order to be applicable in a practical sense -- i.e., in the real world.

Why couldn't there be an allowable reasonable range of possible error probability differentials across all data points that would still allow Glaeser's theory to stand?

-- eve (eve_rebekah@yahoo.com), November 15, 2000.


Oh, by the way, Z, I just saw a blurb on the web that Twain apparently had quoted "The Diz" (Benjamin Disraeli) on that little aphorism -- I believe in his (Twain's) autobiography. So I was kinda sorta right and you're WAY right.

Hey, wasn't "The Diz" a musical, put out in the seventies? (or was it "The Wiz"? Whatever.)

-- eve (eve_rebekah@yahoo.com), November 15, 2000.


Eve and Flint:

I see that you aren't big into statistics. I am not arguing with you. I just think that the person who wrote the originially quoted article is a moron; Harvard or not. [note; just rejected an article from a professor at Harvard].

Best wishes,,,,

Z

-- Z1X4Y7 (Z1X4Y7@aol.com), November 15, 2000.


Z:

Your response of "you are all wrong, but I need not explain myself" probably doesn't impress anyone but yourself. As the only known rejector of the principles Glaeser cites, you might show why the rest of the world is wrong, rather than just call Glaeser a moron.

-- Flint (flintc@mindspring.com), November 15, 2000.


Flint:

As the only scientist discussing this, I think that is ok. I know the man. This article is a political discussion. He wouldn't dare put this before fellow scientists. The fact that I am the only person questioning his article [although not your comments] suggests that is true.

Best wishes,,,,

Z

-- Z1X4Y7 (Z1X4Y7@aol.com), November 15, 2000.


By-the-way:

AOL is having a bad night, so this comes in segments. Don't know when they will fail all together.

I in no way discussed either your or eve's analysis. I just referred to the original article. It is seriously deficient and only can be considered as a biased political piece [sort of common for the WSJ]. Simply put, science is science. If you are going to declare your analysis as scientifically valid, then you have to meet the minimal criteria. Evidently not in the WSJ. Nothing new.

Best wishes,,,,

Z

-- Z1X4Y7 (Z1X4Y7@aol.com), November 15, 2000.


Z:

Except you go so far as to say eve and I are "not big into statistics", a euphemistic way of saying we don't know what we're talking about. Yet I find it hard to believe that, despite the shortcomings of this editorial as a statistical treatise, you would be arguing that applying selective fudge factors to get desired results (which really IS going on in Florida) is a valid tabulation procedure either.

Would you have preferred if Glaeser had written "Gore is unfairly applying a process most likely to swing the count in his favor, which is highly unlikely to happen if this same process were applied fairly."? Or words to that effect? The point he makes is still valid.

-- Flint (flintc@mindspring.com), November 15, 2000.


An excerpt from the article posted at the top. My comments will be on the portion where I've added bold.

This lesson from the ivory tower has a clear implication for the current mess in Florida. Hand counting ballots in only a few, carefully chosen counties is a sure way to bias the results. Even if hand counting is more accurate than machine counting, there is a clear bias introduced because Al Gore chose which counties to hand count. Mr. Gore has selected the state and counties where recounting has the best chance of helping him.

I disagree. If the effect of hand counting a vote were truly unpredictable, then recounting a given county could just as likely decrease a candidate's vote as increase it. A candidate could not then readily discern the counties where he should seek a recount.

That the Gore Campaign can judge where a recount is most likely to profit him, is not because recounts carry an inherent bias. Rather it is because in some places, the method of recounting tends to inflate the vote or is prone to outside influence (or both).

-- David L (bumpkin@dnet.net), November 15, 2000.


David L:

[That the Gore Campaign can judge where a recount is most likely to profit him, is not because recounts carry an inherent bias. Rather it is because in some places, the method of recounting tends to inflate the vote or is prone to outside influence (or both).]

Yes indeedy. We read about thousands of ballots where the counting machinery found no vote for President. Historically, submitting a ballot without a vote for President is vanishingly rare. By machine count (even repeated machine count), such ballots do not fall into the count for either candidate, but we can be pretty damn sure that the voters who submitted these ballots almost all *intended* to vote for *somebody* for President. It just didn't "take" for some reason, as far as the machine is concerned.

If we examine these ballots manually, we can see failed efforts to punch a hole. Quite often, it's very clear which hole the voter was *trying* to punch, even if the puncher didn't quite make it all the way through. Should this "partial punch" ballot be counted for the candidate the voter was clearly trying to select? If not, then we might as well accept the machine count. If so, is any candidate favored? Isn't the voter as likely to have not quite punched one candidate as the other? Certainly the ballot is no more difficult to punch beside the name of either candidate, right?

Where the bias comes in, is that the "not quite punched" ballots in any given jurisdiction are highly likely to reflect the same voting ratio as the fully punched ballots in that jurisdiction. That is, if we have 1,000 "partial punch" ballots in a precinct that favored candidate A over candidate B by a ratio of 2-1, then those ballots will likely be interpretable as roughly 670 for candidate A and 330 for candidate B. This is a swing of 340 votes, which is plenty to swing the entire state if the starting difference was only 300 votes.

So what's most important isn't so much HOW you count the ballots, as WHERE you count them. The overall winner depends entirely on which jurisdictions are recounted and WHICH ARE NOT.

-- Flint (flintc@mindspring.com), November 15, 2000.


Flint,

Please stop misrepresenting my views and apologize for your pas misrepresentations.

- - - - - - - - - - - - - - - - - - - - - - - - -

Folks,

Let me point out one more of Flint's misrepresentations of what I wrote:

I wrote: "People can determine the intent of a voter who has punched partly, but not all the way, through a punch-card ballot better han manchines, generally." ["... better han manchines ..." was intended to be " ... better than machines ..."]

Flint responded with: "Now, did the voter *intend* to punch a hole?"

But I didn't write about the voter's intent to punch a hole. I wrote about the intent of a voter who _did_ punch a hole, but only partially, not completely.

-- No Spam Please (nos_pam_please@hotmail.com), November 16, 2000.


No Spam:

You wrote: "People can determine the intent of a voter..."

I wrote: "did the voter *intend*..."

NOW you say "I didn't write about the voter's intent..."

Now I ask you bluntly. Did you or did you not write about the intent of the voter when you wrote about "the intent of the voter"? If you just happened to use these words but didn't mean them, just what are people determining here anyway? "People can determine the intent of a voter" by asking themselves "What did the voter intend". How ELSE do they do it? This isn't a misrepresentation by any stretch at all. These are YOUR OWN DAMN WORDS! Sheesh.

Now, let's consider a situation that is not at all hypothetical. Here we have two ballots, both with a "partial punch", namely a dimple, in the Gore chad. One of these ballots has no other defomations of any kind, the other has the Bush chad punched completely out. But the dimples in the Gore chad are identical.

Now, how does one count these ballots? A Republican counts the one with the Bush hole for Bush and counts the other as a "no vote" because there is no hole, and discards it. This is perfectly reasonable. A Democrat discards the first as a "double vote" because there are two "holes", and counts the second for Gore. This is also perfectly reasonable. Counting the dimple as a hole in this case works in the favor of Gore.

But we've had just the opposite, where the dimple is in the Bush chad. In this case, counting the dimple as a hole works *against* Gore. Not surprisingly, the Republican and Democrat reverse their definitions in this case. After all, BOTH definitions are equally reasonable.

No Spam, I ask you. How would YOU count these two ballots? Does the Gore dimple count as a "partial hole" or doesn't it? Can you understand how a desire to see a specific outcome might influence the reasonable interpretation of a ballot? Just how partial is partial enough? Do other holes on the card change this partiality? When and how?

If every hole, however partial, were unambiguous we'd have no problem. But they ARE ambiguous in many cases. Some ballots have dimples in BOTH the Bush and Gore chads, so do we count the larger dimple? Who gets to decide which is larger? Who gets to decide who gets to decide?

I am NOT misrepresenting you by recognizing that this ambiguity exists and must be dealt with. I'm not impressed with your clear implication that if we pretend it doesn't exist, it will go away. What manual counts are doing is determining the intent of the voter, plain and simple. This intent is NOT always clear, and that's a fact. Deal with it.

-- Flint (flintc@mindspring.com), November 16, 2000.


Flint,

You continue to misrepresent what I write.

Here's an example from right at the beginning of your latest posting:

"No Spam:

You wrote: "People can determine the intent of a voter..."

I wrote: "did the voter *intend*..."

NOW you say "I didn't write about the voter's intent..."

Tell us, Flint: Why do those last two quotations show only part of the sentences involved instead of the whole sentences?

Could it be that if you had quoted entire sentences, the quotes would not have seemed to be so parallel? And it would have been apparent that I was not contradicting myself?

Your whole sentence was "Now, did the voter *intend* to punch a hole?" My first subsequent sentence in response to that was "But I didn't write about the voter's intent to punch a hole." Then I clarified with my next sentence, "I wrote about the intent of a voter who _did_ punch a hole, but only partially, not completely."

Why did you leave out important parts of your and my sentences that would show that we were referring to different aspects of intent? Why did you want your reader to think we were both referring to just "intent" with no qualification such as "... to punch a hole..."?

In your latest post, you continued with: "Now I ask you bluntly. Did you or did you not write about the intent of the voter when you wrote about "the intent of the voter"?".

I ask _YOU_ bluntly, Flint: Why did you quote parts of sentences out of context in a way that obscured those sentences' differences in meaning?

-- No Spam Please (nos_pam_please@hotmail.com), November 16, 2000.


eve,

>[...he leaves out important stuff that renders his conclusions invalid.]

>Such as?

a) That the rights of candidates to request recounts is not a matter of statistics; it's a matter of providing each candidate with a way to ensure a fair election result in the face of a host of possible types of irregularities, many of which are nonstatistical in nature, and some of which may not be specifically foreseeable.

b) The reasons for requiring candidates who request recounts to pay the costs of those recounts include: (1) discouraging frivolous requests, and (2) that governments are not expected to budget funds to cover potential recount costs.

c) Given that candidates pay the cost of recounts, they have the right to request recounts only where they want them (subject to the limitation that whole, not partial, precincts are specified). If another candidate wants recounts in other areas, that candidate is entitled to request such, provided s/he is likewise willing to pay the cost.

d) That "biased" is not automatically bad. A bias toward accuracy and away from error is usually considered desirable.

I can list more points if you wish.

>[...his article is in the category of persuasion rather than objectivity.]

>Why do you feel you need to distinguish this?

Because (a) the categories have different aims, and (b) many people may not recognize this.

It's been a long time since my speech andd English classes in school where we covered this stuff, and I may not be remembering the correct designations for the categories. I'd welcome help on this.

By "persuasive" I mean the category of presentations designed to persuade the audience to adopt a particular idea or point of view.

By "objective" I mean the category of presentations designed to present balanced information to an audience without aiming them to persuade them to adopt a particular idea or point of view. It aims to present roughly equal information about each point of view if the topic is a subject that involves multiple points of view. I can see that "objective" probably isn't the best name for this category, but I can't think of what the name should be. "Balanced"?

>And do you feel that persuasion and objectivity are mutually exclusive?

No. Maybe I should use "balanced" to denote the category I've been calling "objective, but I'm sure that's not really the correct term either.

>Do you feel that, say, we're persuaded that the sun will rise tomorrow based on various objective information?

We're capable of being so persuaded. :-)

>Is Glaeser's article persuasive in that sense? If not, why not?

Yes and no. It does present objective information, with the aim of persuasion. But, as I wrote earlier, it omits important aspects, and is left unbalanced: it does not attempt to include some pretty obvious objective information (see my list above) that conflicts with the conclusion it draws.

>[He's being _too_ simple. He's urging a look at only _part_ of the "big" picture. He omits mentioning some aspects of elections, aspects to which statistical principles do not apply.]

>Could you give me an example that would stand to refute his points?

See my list above.

Items a), b), and c) collectively refute Glaeser's assertion that hand recounts cannot be selective.

>[But I'm pointing out that Glaeser is not mentioning important non- statistical considerations that make his statistics-based conclusion invalid for our real-world situation.]

>Again...for example?

See above.

>[What did Gore say about "the will of the people" that was an obviously outrageous lie?]

>In the context of the selected Florida hand recounts, he said, "There's something very special about our process that depends totally on the American people having a chance to express their will without any intervening interference." (New York Times, 11/14/00, page A19, column 2).

Oh, please. That is not an obviously outrageous lie.

>As I indicated in so many words above, the truth would have been something like, "a certain, selected, limited group of American people" instead of "...the American people..."

That is your opinion, not a demonstration that Gore's statement was an obviously outrageous lie.

>[Should Bush abandon any pretense at a fair result, as well as Gore, in your view?]

>Well, that depends on what you mean by "fair."

I meant it in the same way that you meant it when you wrote "... as long as Gore abandons any pretense at a fair result, ..."

My point was to inquire whether you would apply the same standard to Bush as to Gore.

So, how did you mean "fair" when you used it in that phrase about Gore? Barring some outrageous definition I don't now foresee, I'll mean "fair" in the same way, because I intended my question to parallel your comment about Gore in that regard.

-- No Spam Please (nos_pam_please@hotmail.com), November 16, 2000.


No Spam:

I give up trying to get you to address the topic of the actual partially punched ballots. But I can't blame you for your extreme reticence to address it, since I can no longer see how you could do so and save face.

I'd like to respond to this one however. As carefully as I can, and without any misrepresentations that I know of.

>>>>[...he leaves out important stuff that renders his conclusions invalid.]

>Such as?

a) That the rights of candidates to request recounts is not a matter of statistics; it's a matter of providing each candidate with a way to ensure a fair election result in the face of a host of possible types of irregularities, many of which are nonstatistical in nature, and some of which may not be specifically foreseeable. <<<<<

Yes, without question a great deal more is involved in an election than the mechanical process of casting and counting votes. The problem we're dealing with here is the intended SCOPE of Glaeser's article. By my reading, he is specifically and narrowly addressing the statistical issue of partial recounts by different methods. It seems unfair to criticize Glaeser for failing to make an unrelated point, much like it's unfair to criticize an article about cabbages because it failed to discuss kings as well (apologies to Lewis Carroll). Glaeser isn't discussing WHAT the candidates' rights are, he is discussing HOW the exercise of those rights can introduce bias into the results if those rights are actually USED to introduce bias.

>>>>b) The reasons for requiring candidates who request recounts to pay the costs of those recounts include: (1) discouraging frivolous requests, and (2) that governments are not expected to budget funds to cover potential recount costs.<<<<

But what does cost have to do with statistical theory? Is the conclusion that 2+2=4 less valid if it costs more to do the addition? This is a peculiar objection. When the payoff for jiggering the count is the Presidency, cost is no object. Believe me.

>>>c) Given that candidates pay the cost of recounts, they have the right to request recounts only where they want them (subject to the limitation that whole, not partial, precincts are specified). If another candidate wants recounts in other areas, that candidate is entitled to request such, provided s/he is likewise willing to pay the cost.<<<<

This is as true as it is irrelevant, namely completely. Partial counts by new procedures bias the total. Period. Whether both candidates have the right to do so or not, whether it costs money to do so or not, whether the candidates are willing to pay or not, all notwithstanding. Partial counts by new procedures bias the total. Period.

>>>>d) That "biased" is not automatically bad. A bias toward accuracy and away from error is usually considered desirable.<<<<

Not true. This seems to be a subtle point, and I believe you sincerely don't understand it. Let me give you a simple example here.

Let's say we want to know whether men or women tend to be taller on average, and what we have to measure with is a rubber ruler because this is fast and cheap. Our results aren't going to be perfectly precise, but we go ahead anyway, and we find that men and women *by our measurements* are very close to the same height, with men just slightly taller.

Now, let's say we don't like these results. We want women to be taller. So we decide to remeasure *just* the men, and this time we won't use the rubber ruler, we'll use a very expensive precision device. This gives us a new and more accurate height for the average man, which turns out to be a good bit *smaller* than the rubber ruler height. NOW women are taller, which is the desired result, so we are happy. We have *INCREASED OUR ACCURACY* so how could our new conclusion be worse?

In fact, of course, men are taller, so our increased accuracy led us to the wrong conclusion. If all women had ALSO been measured using the high precision method, then our final conclusion would again have been correct. The increase in accuracy was, unfortunately, more than offset by our methodological error of partial remeasurement.

>>>>[...his article is in the category of persuasion rather than objectivity.]

>Why do you feel you need to distinguish this?

Because (a) the categories have different aims, and (b) many people may not recognize this.<<<

A good point here. Glaeser makes his point as an advocate. We need to take this into consideration when we evaluate what he writes. And clearly, his preference is for Bush and against Gore. But this in itself doesn't render his point invalid or suspect. His argument must be evaluated on its own statistical merits. I believe you are making invalid objections because you ignore the statistical merit and focus on irrelevancies. You need to entertain the idea that some people might favor Bush and still have valid points.

>>>>And do you feel that persuasion and objectivity are mutually exclusive?

No. Maybe I should use "balanced" to denote the category I've been calling "objective, but I'm sure that's not really the correct term either. <<<

Granted Glaeser is not disinterested. Few are. But you seem to take this to an extreme, leaving the distinct impression that as far as you are concerned, an observation that favors Bush simply cannot be objectively true under any circumstance.

>>>>Do you feel that, say, we're persuaded that the sun will rise tomorrow based on various objective information?

We're capable of being so persuaded. :-) <<<<

Unless Bush says this. In that case, I strongly suspect some of us are NOT capable of being so persuaded. [grin]

>>>>>Is Glaeser's article persuasive in that sense? If not, why not?

Yes and no. It does present objective information, with the aim of persuasion. But, as I wrote earlier, it omits important aspects, and is left unbalanced: it does not attempt to include some pretty obvious objective information (see my list above) that conflicts with the conclusion it draws. <<<<

Here is where things are interesting. You are absolutely correct that Glaeser's article is a persuasion piece. You are correct that it makes too many assumptions to hold water all by itself. But the list you created does NOT consist of valid objections. What the candidates have the right to do, what they must pay to do, what they must be willing to pay to do, what reasons they might have for doing various different things, are beside the point. The point is what they *ACTUALLY DID* do, and whether that introduces a statistical bias into the results. IT DOES! And as my simple example above illustrated, it does NOT necessarily make the final result more accurate. In fact, nobody denies that Gore expects to gain by this stratagem. Not gain some hazy notion of "accuracy" at all. Gain VOTES!

>>>>>[Should Bush abandon any pretense at a fair result, as well as Gore, in your view?]

>Well, that depends on what you mean by "fair." <<<<

Absolutely. What Gore is doing is perfectly fair. He has that right, Bush has that right, both of them have the money, the same stunt will work to gain Bush votes in his counties exactly like it works to gain Gore votes in HIS counties. And very probably, if Bush had done the same thing, then after enough time to do all these manual counts, we'd ALSO end up with a Bush victory.

But why should Bush trade DEFINITE for VERY PROBABLY? Let's say I give you a million bucks. NOW, let's say I offer you a chance to roll 2 dice. You can decline and keep the million, or you can roll the dice and keep the million if you don't roll a 7. Which do you pick? Why in Heaven's name would you roll the dice? All you can do is lose your money!

Recounts are rolls of the dice. The dice may be perfectly fair, but if you've already won it all, why keep playing? To favor a recount is to favor the loser. Period.

-- Flint (flintc@mindspring.com), November 16, 2000.


You need to entertain the idea that some people might favor Bush and still have valid points.

This is pretty hard to do, considering what some folks are saying, Flint. I'm old enough to have worked with punch cards for MANY years. They were mangled, dropped, handled, dropped, read for years, dropped, and dropped again. The first thing programmers did when they received their programs back from keypunch were to check them [by hand], and hand number them in anticipation of them being dropped by the operators. None of this manipulation created new punches.

Cory had to be in the business when punched cards were used, but now he's asserting that the recounters are card sharks and magicians. He's claiming that they're wearing ornate rings and the unaware observers won't notice them poking a hole in the card because they're innocents to the skillset of these "card sharks and magicians." His remarks are on csy2k, repeated on EZBOARD, and come from his "latest", BTW.

-- Anita (Anita_S3@hotmail.com), November 16, 2000.


Anita:

You really ought to examine one of these ballots, or a good picture of one, before you start talking about IBM punchcards. Those old Hollerith cards, before being punched, were fairly thick smooth paper, and the punchers actually cut little rectangles out of them with great force. They were damn durable -- I worked with them too.

These ballots are not like Hollerith cards at all. They are pre- perforated. The holes to be punched out are nearly punched out to begin with, and held only at the 4 corners by very tiny pieces of cardboard, easily ripped or broken. It doesn't take much force at all to push one of these nearly-cut chads hard enough to rip one or more of the perforation hooks. Yes, you can make a dimple if you press very lightly. But the process of making a dimple cannot help but weaken one or more of the 4 tiny pieces holding the chad in place.

Now you take a dimpled card with weak corners, stack it with a bunch of others, rub them back and forth against one another. The dimple catches on holes in adjacent cards and the corners break. A bit of rubbing, unavoidable when handling large numbers of these things, and you have an "inadvertent vote". Even a corner of an adjacent card can hook a near-performation (the part already cut out) and "vote" for someone or something.

In any case, Glaeser isn't talking specifically about the logistics of ballot handling

-- Flint (flintc@mindspring.com), November 16, 2000.


Flint,

I see that you continue to misrepresent things I wrote.

But for a change of pace, let's look at something you wrote near the middle of your latest posting addressed to me that was not really in the category of misrepresentation, but more of a failure of careful thinking about sources of error. This one is especially interesting and educational because it involves a mistake that many, perhaps most, readers might overlook.

In a posting addressed to Eve, I wrote:

d) That "biased" is not automatically bad. A bias toward accuracy and away from error is usually considered desirable.

You responded with:

Let's say we want to know whether men or women tend to be taller on average, and what we have to measure with is a rubber ruler because this is fast and cheap. Our results aren't going to be perfectly precise, but we go ahead anyway, and we find that men and women *by our measurements* are very close to the same height, with men just slightly taller.

Now, let's say we don't like these results. We want women to be taller. So we decide to remeasure *just* the men, and this time we won't use the rubber ruler, we'll use a very expensive precision device. This gives us a new and more accurate height for the average man, which turns out to be a good bit *smaller* than the rubber ruler height. NOW women are taller, which is the desired result, so we are happy. We have *INCREASED OUR ACCURACY* so how could our new conclusion be worse?

Well, I'll explain how your example's new conclusion can be worse. It's stuff I learned in freshman physics lab, and it's not because of the explanation you give.

First let's lay out some reasonable assumptions about the rubber-ruler and precision-device measurements to eliminate sources of error other than the two different instruments. Let's assume that all the _same_ men were measured in the two different ways -- no one was measured one way but not the other. Furthermore, let's assume that everything about the measurement environment was the same for the two different runs of measurements, except only for the difference between the two measuring instruments themselves. Also, let's assume that all measurement instrument readings were made by the same impartial observer who read each instrument's result exactly the same way insofar as possible -- that there was no difference in the parallax angle, reference point, or any other observer error. Also, let's assume that none of the men's _actual_ heights differed between the two sets of measurements -- none of them either grew or shrank in stature between the two measurements. So, we can assume that any difference between the two height measurements of a particular individual, and any difference in the average (the same type of arithmetic average in each case, of course) height measurement, was due solely to some difference between the measuring instruments.

(Please feel free to point out anything I may have overlooked in listing these assumptions that could have any bearing on the differences in measurement or render my last sentence of the preceding paragraph invalid. But note that I intend "the measurement environment" to have a very broad meaning, to cover anything like temperature, wind speed, gravitational constant, relativistic contraction, and so forth.)

So since the only possible reasons for the observed difference in average men's height must be related to the difference in instruments, and the precision-device measurements are more accurate than the rubber-ruler measurements, we can draw valid conclusions by comparing the precision-device measurements of the men with the rubber-ruler measurements of the women, right? That is, according to the logic you used in your example, right?

Wrong.

There could be a systematic error in the rubber-ruler measurements that was not present in the precision-device measurements that accounted for the difference in the averages of the two sets of measurements of men's heights. For instance, the rubber ruler might have indicated that every height was twelve inches more than the actual height, while the precision device always indicated the correct actual height. Then there would be a systematic twelve-inch difference between the men's heights (and their average) measured by rubber ruler and the men's height (and their average) measured by the precision device.

Now, perhaps the reader has been following this so far, nodding "yes, yes, yes". But _I have not yet pinpointed the reason why the example's conclusion that women are taller than men is invalid_.

The reason why your example's conclusion that women are taller than men is invalid is NOT that the precision-device measurement was more accurate. It's that you compared measurements made by one device with measurements made by another device without accounting for possible difference in systematic error between those two measuring instruments! This failure to account for systematic error, or to correct for it by comparing only sets of measurements made with the same instrument, renders _any_ conclusion about the difference in average heights (as measured by the different instruments) of men and women invalid.

Remember my mentioning "systematic error" earlier in this thread?

Let's continue with the current analysis. Your example continued with:

In fact, of course, men are taller, so our increased accuracy led us to the wrong conclusion.
Wrong. It was the comparing of sets of measurements made with two different instruments without correcting for systematic differences between the two instruments' measurements that led to the wrong conclusion, not the increased accuracy. Your example's false conclusion did not arise from the difference in accuracy; it arose from failure to account for systematic error.
If all women had ALSO been measured using the high precision method, then our final conclusion would again have been correct.
So close, and yet so far ... You were right on the verge of the correct error analysis, but failed to discern the actual source of error. It is true that if the women had been measured with the same precision-device instrument, with all the same assumptions listed above, then the correct conclusion about men's and women's average heights would probably might have been reached. Not necessarily because of the increased accuracy of measurement, but because the men's and women's average heights would have been derived from measurements with the same instrument for each gender.

Continuing ...

The increase in accuracy was, unfortunately, more than offset by our methodological error of partial remeasurement.
Again -- so close, but still not quite on-target. It's the unaccounted-for difference in systematic error that made the height averages not validly comparable.

The failure to account for the systematic error was a methodological error, all right. But it still would have been possible to validly compare the average of men's heights taken by precison-device with the average of women's heights taken by rubber ruler, at least for the restricted purpose of determining whether men were taller than women or not, even though we had only that partial remeasurement, IF we had an accurate-enough determination of the systematic error between the two instruments to be able to correct for it.

-- No Spam Please (nos_pam_please@hotmail.com), November 17, 2000.


Flint,

I hope you understand from this that you have been misleading your readers with your arguments that a hand recount of an election (e.g., in Florida) would introduce more error than was present in the preceding machine counts. You need to retract that argument and apologize to all the folks you've mislead with your false reasoning in all the threads where you've presented it.

-- No Spam Please (nos_pam_please@hotmail.com), November 17, 2000.


No Spam Please,

I haven't yet read through your latest response to Flint in detail, but I'd like to interject something for now.

For what does Flint need to apologize? If he's wrong, prove him wrong. But all these insinuations about intentional misrepresentation on his part are nothing more than empty ad hominem attacks. As far as I can recall, his posts to you have been thoughtful, reasoned, objective, and civil. He believes some of your points to be irrelevant, or otherwise incorrect -- and so do I (again, prior to analyzing your recent post). From there, how can you possibly make this leap of faith to assumptions about dark, ulterior motives on his part?

I really think that baseless pejorative insinuations about others' psychological motives should have no part in these discussions. And with all due respect, if there is to be any apology in order, it should come from you.

-- eve (eve_rebekah@yahoo.com), November 17, 2000.


I give up trying to get you to address the topic of the actual partially punched ballots. But I can't blame you for your extreme reticence to address it, since I can no longer see how you could do so and save face.

And yet you never answered his question.

-- (hmm@hmm.hmm), November 17, 2000.


"...if you've already won it all, why keep playing? To favor a recount is to favor the loser. Period."

And why is this a bad thing, Flint? Are Bush and his "minions" interested in the truth (remember, he's bringing "integrity back into the White House"), or are they simply interested in winning? If you were in Gore's position and you knew that the results were quite close and you further knew that a number of ballots were not counted, a total that is well more than the difference between you and your opponent, what would you do?

Again, is Bush interested in "the truth" or in "winning"?

And I have to ask again, if the tables were turned, what do you suppose Bush would do? (Hint: I don't think we'd have SOS Harris upholding any deadlines. Why that woman hasn't excused herself is beyond my scope. I can just hear the Republican screams of "conspiracy" and the like if the situation were reversed.)

BTW, that one statement of yours that I cited, IMO, lends credence to the opinion that the article in question is NOT objective, but is more of a persuasion piece. There is no mention in that article that Bush had exactly the same opportunity for a re-count, but chose not to take advantage of it. That information would have lent an enormous amount of balance to the article.

If the author is trying to convey his opinion that the laws need to be changed, he needs to present the whole picture; not just his opinion on a selected part of the picture.

And, BTW, isn't that the point he's trying to make?

eve, you never indicated what type of article this is -- e.g., is it an op-ed piece? That needs to be clarified.

-- Patricia (PatriciaS@lasvegas.com), November 17, 2000.


Flint, sorry; I hadn't read your response above wherein you addressed much of what I wrote.

But it is a good example of what can happen when one does not take into account "the big picture"; "the whole story".

eve, is this an op-ed piece?

-- Patricia (PatriciaS@lasvegas.com), November 17, 2000.


Patricia, yes it was an op-ed piece from the Wall Street Journal, so I have little doubt the guy's rooting for Bush. But if that's your concern, I'm not sure why that would matter; I haven't been able to detect any bias in the piece itself. For example, if the tables were turned (with Bush asking for the recounts), you could apply the same essay/principles to him.

Back at y'all after a while...

-- eve (eve_rebekah@yahoo.com), November 17, 2000.


eve, no offense, but of course you haven't been able to detect any bias; you have clearly been pro-Bush on this and many other threads. Let me clarify that; it's not so much that you've been pro-Bush as it is you appear to be anti-Gore.

The fact that it's an op-ed piece makes all the difference. It is purely opinion; by its very definition it isn't supposed to be objective.

Had that been laid out at the start of the thread, I'm not sure there would be the extent of the ensuing discussion we're now seeing. I can only speak for myself, but I originally took great issue with the fact that the author only gives half the story. But because I now know that it is an opinion piece and is in no way meant to be objective, I understand why it only gives half the story.

-- Patricia (PatriciaS@lasvegas.com), November 17, 2000.


"There are three kinds of lies: lies, damn lies, and statistics." So reportedly said Benjamin Disraeli, prime minister of Great Britain from 1874 to 1880. ... www.gpsworld.com/columns/9805innov.html

-- (your@answer.here), November 17, 2000.

Patricia,

I'm sorry I hadn't made it clear at the start that it was an op-ed piece. But I think the others were probably generally aware of this, as it came from a conservative paper -- yet that didn't stop anyone from debating its points.

But, regardless of the author's political leanings, the guy presented an argument and a position: If X, then Y, etc. etc. Do you see bias or other flaws in the argument itself? If so, would you address them directly? Doesn't it make sense that to refute him you have to do this?

I will admit to one possible point of bias: That if the tables were turned (Bush pressing for selected recounts) you probably wouldn't have seen a piece like this in the Wall Street Journal. You would have seen it in the New York Times.

-- eve (eve_rebekah@yahoo.com), November 17, 2000.


Eve,

>I haven't yet read through your latest response to Flint in detail,

Then you haven't learned the lesson of Flint's basic mistake.

>But all these insinuations about intentional misrepresentation on his part are nothing more than empty ad hominem attacks.

Eve, please look back at my accusations of misrepresentation.

Do you see where I wrote that Flint was misrepresenting my views _intentionally_?

No.

I didn't say Flint was doing it intentionally.

So why did you write that I had accused him of that?

Because you made a mistake. It's not the same mistake as Flint made in his example about measurement, but you were careless with the truth.

If you still think there were insinuations such as you accuse me of making, please point out exactly where they were so that I can clarify any inadvertant misunderstanding.

>As far as I can recall, his posts to you have been thoughtful, reasoned, objective, and civil.

If you were to carefully compare what Flint has asserted that I wrote or meant with what I actually did write, you would find a large number of misrepresentations by Flint in these post-election discussions.

And if you will carefully and thoughtfully finish reading my just-posted analysis of the mistake Flint made in his measurement example, you would see that Flint's reasoning about the issue of recounts has been faulty lately.

>He believes some of your points to be irrelevant, or otherwise incorrect

Then he could have explained that without misrepresenting what I actually wrote.

>and so do I (again, prior to analyzing your recent post).

Once you have carefully and thoughtfully analyzed that recent post, you may modify your opinion about that.

>From there, how can you possibly make this leap of faith to assumptions about dark, ulterior motives on his part?

Eve, please look at what you just wrote about my making a leap to assumptions, then look in a mirror, then carefully and thoughtfully re-read what I _actually_ wrote, then think about which of us made a leap of faith to assumptions.

>I really think that baseless pejorative insinuations about others' psychological motives should have no part in these discussions.

So do I.

>And with all due respect, if there is to be any apology in order, it should come from you.

Once you have carefully and thoughtfully read or re-read what I actually wrote, including my recent long post addressed to Flint, you may wish to reverse that opinion.

-- No Spam Please (nos_pam_please@hotmail.com), November 17, 2000.


Eve,

Pardon me for intruding on your conversation with Patricia, but I wish to point out something.

>I'm sorry I hadn't made it clear at the start that it was an op-ed piece.

Notice that you made a mistake of omission. You omitted something that was relevant to the subject -- that it was an op-ed piece.

Yes, we guessed it anyway. And it wasn't a big, big, big thing you omitted. But you did omit it, and that is relevant to what I'm going to say next.

>the guy presented an argument and a position: If X, then Y, etc. etc. Do you see bias or other flaws in the argument itself?

One of the things I've explained about the article is that the author omitted important stuff that was relevant to his subject. The main flaw (if the article were to be considered objective, that is) in his argument is not what he included; it's _what he omitted_.

- - - - - - -

Patricia,

Pardon me for responding to comments Eve addressed to you.

-- No Spam Please (nos_pam_please@hotmail.com), November 17, 2000.


Not a problem, No Spam; your reply is exactly what mine would have been, as that was the point I was trying to make.

eve, I really don't understand your response to me. We ALL have certain biases; it's part of what makes us human. As I stated, I meant no offense, but your "short" response (The New York Times versus The Wall Street Journal) tells me that you took offense. I apologize as that was not my intention.

-- Patricia (PatriciaS@lasvegas.com), November 17, 2000.


No Spam Please,

A flaw in an argument can easily encompass an omission. To add "omissions" to what I said is therefore redundant -- that's why I didn't see the need to mention "omissions" separately.

Regarding Flint, you combined an emphasis on his "misrepresentations" with the requirement for an apology. I saw in this an implication that you felt he was intentionally misrepresenting you.

I mean, if he had no ill intent, even if you prove him wrong, of what relevance would an apology be? If I assert 2+2=5 and you prove to me that 2+2=4, I'll say something like, "You know, you're right. I stand corrected." Are you saying I should add, "I apologize" to this?

I do plan to review your recent replies to Flint.

-- eve (eve_rebekah@yahoo.com), November 17, 2000.


Patricia, I took no offense at all from anything you said. And I really hope you weren't offended by my post, either. But I meant every word of it. And not sarcastically at all -- I really believe if the situation was reversed you'd probably see a piece like that in a liberal publication, and not in the WSJ.

Yes, we all have biases -- of course. But you can't automatically extrapolate from that and assume there are biases within Glaeser's particular essay. You could try to show that there are, though, by addressing it on the merits.

-- eve (eve_rebekah@yahoo.com), November 17, 2000.


Again from this ridiculous article:This lesson from the ivory tower has a clear implication for the current mess in Florida. Hand counting ballots in only a few, carefully chosen counties is a sure way to bias the results. Even if hand counting is more accurate than machine counting, there is a clear bias introduced because Al Gore chose which counties to hand count. Mr. Gore has selected the state and counties where recounting has the best chance of helping him.

Yes Gore chose as the LAW provided. Thus in one sentence this article which the Wall Street Journal thinks legitimate by printing it, falls of its own weight as the pile of ridiculous crud it is.

Has the WSJ printed a retraction? Have they since indicated the article is based on false information? Course not cause their readers don't care they want to believe in Fairies.

WSJ? Crud source along with the rest of the "Liberal" media.

-- Doc Paulie (fannybubbles@usa.net), November 17, 2000.


Well, I'll get the easy ones out of the way first...

Hmm:

I did in fact answer his questions, which answer consumed the entire remainder of a fairly long post, which he chose to ignore, perhaps for that very reason. Go back, read the rest of the post, read his question, and you will find my answer. In the remainder of that post, you will find that I detailed sample ballots and asked him to interpret them. He pretended I asked no such question, yet I notice you don't take HIM to task for this. I wonder why?

Patricia:

Are you serious? Neither Bush nor Gore has the slightest interest in the truth whatsoever. They are both interested in winning. Winners get to write the history. However, both are quite interested in pointing to whatever is in their favor and calling it truth. Winning with as much public support as possible is certainly preferable.

I think it's frankly stupid to point out that the winner could legally have demanded a recount also, and to fault him for not doing so. How many different ways can I say that once you have won, a recount can NOT help you. The winner can legally shoot himself in the foot, or have a sex change operation, or join the hare krishnas, all of which would make more sense than a recount. Why not criticize Glaeser for failing to mention THESE options as well?

I believe if the situation were reversed, then what both candidates are doing would also be reversed. The loser might as well demand a recount, why not? If possible, the loser should jigger the recount to his maximum advantage in terms of votes gained, however invalid this might be. And the winner would take steps to prevent this. These are elementary game strategies, so self-evident there ought to be no need to point them out.

And to amplify eve's point. 2+2=4 is just as true on the op-ed page as anywhere else. Dismissing it as "opinion" for its location in the paper alone is, uh, not too bright.

--------

Now to No Spam's rather confused but very detailed statistical critique.

No Spam, you have carefully set things up so that one simply cannot disagree with you without "misrepresenting" you. The idea that you might be wrong is cleverly disallowed by this technique. Now, how about being honest for a change?

You write, [we can draw valid conclusions by comparing the precision- device measurements of the men with the rubber-ruler measurements of the women, right? That is, according to the logic you used in your example, right?]

No, absolutely the contrary. You cannot draw correct conclusions by comparing apples and oranges. I regard a total machine count or a total hand count as being apples and apples. I regard a machine count where it's to your advantage, combined with a hand count where THAT'S to your advantage, as invalid. And in my example, it was the invalid comparison that caused the wrong conclusion.

First, I think we're safe in assuming that there *might* have been some systematic error in the machine count. The precision of that count was fairly good considering the fragility of some of the ballots themselves and the handling they received in a hurry to get the counts out on election night. The almost everywhere small difference between initial and second machine count implies pretty good precision by the machines. But as you point out, if the rubber ruler consistently measures a foot too tall, it's a foot off regardless of how precise it might be from one measurement to the next.

Your argument seems to hinge on having a "rosetta stone" comparison factor between the two methods of measurement. At least, when you say "It's the unaccounted-for difference in systematic error that made the height averages not validly comparable", I assume you mean this difference should be accounted for by that comparison factor. You say we could have factored out the systematic difference "IF we had an accurate-enough determination of the systematic error between the two instruments."

Now, I agree with this entirely. This factor is what tells us that the rubber ruler IN FACT measures a foot too long. IF we could correctly measure this factor, and IF we apply it to all of our data consistently, we end up not changing the RATIO of heights, but we do correct the MAGNITUDE of the heights. Our magnitudes are now MUCH more accurate. Too bad we don't give a wiz about magnitudes, but ONLY about ratios.

Now, back to Florida. First, nobody is looking for a correction factor. Second, nobody is thinking of applying that correction factor to improve overall accuracy. Third, nobody WANTS to maintain the ratio, the goal is to CHANGE the ratio. Nobody could possibly care for a correction factor that adds, say, 1.327% to each candidate's votes as counted by machine, to arrive at a total as counted by hand. This is "more accurate" in the sense of coming closer to the number of votes actually cast, but does not change the winner. We don't care if 6 million votes are cast, or only 60 votes. We don't care if 6 million were actually cast but through monstrous incompetence we were only able to count 60 of them. We care WHO WON. And THAT'S ALL we care about.

But at least now I understand the semantics of our positions. I admit I never gave a thought to the value of "greater accuracy" in the counts themselves, but only in the ratio of counts. I can assure you that Gore isn't trying to recount specific areas manually, and Bush trying to prevent this, because Gore wants to lose by the "more accurate" count of 3,000,600 to 3,000,300 if the previous count had him losing by 3,000,300 to 3,000,000. This is NOT his goal here.

This is a race, and what counts is who wins. Not by how much, not by how fast they went, not by how far they went, not by any other utterly irrelevant factor that may or may not be measured with great accuracy. Only the DELTA matters. The purpose of doing selective hand counts isn't to get more accurate totals, it's to CHANGE THE WINNER. It's to "make the count more accurate in Gore's favor", which is a bias.

And one point you seem to have overlooked in your long post, very strangely considering your prior sensitivity to it. In my example, I wrote "we don't like these results. We want women to be taller." This was absolutely critical to my point, and should not have been overlooked. I tried to make it clear that my example researchers KNEW DAMN WELL that their techniques were incorrect. They did NOT use precision instruments to increase accuracy, they used them selectively to GET THE DESIRED ANSWER!

What Gore is doing in Florida has a long and well documented history. It is called the fudge factor, the finagle factor, the Monk's coefficient, and many other terms. It's a way of getting the desired results despite the data, without obviously cheating.

Now, is this the place where I should pompously demand that you trot around to all your other threads and apologize for missing the point and refusing to understand why? [grin]

eve:

While you notice I called No Spam dishonest, nowhere did I say he was *intentionally* dishonest. I don't want anyone to leap to unwarranted conclusions here. Hehehe.



-- Flint (flintc@mindspring.com), November 17, 2000.


I did in fact answer his questions, which answer consumed the entire remainder of a fairly long post, which he chose to ignore, perhaps for that very reason. Go back, read the rest of the post, read his question, and you will find my answer.

Actually, the question I was referring to was this one:

Why do those last two quotations show only part of the sentences involved instead of the whole sentences?

I did indeed read through your answer, but I was unable to find where you answered it. I could have missed it, though. Perhaps you can point it out for me.

-- (hmm@hmm.hmm), November 17, 2000.


Flint,

When I wrote at the top of my recent long posting addressed to you that I saw that you continue to misrepresent things I wrote, that was too short.

I see that you were more careful about misrepresentation in your latest posting addressed to me than in some earlier ones. I'm sorry I didn't say that earlier.

But still, there were misrepresentations in your posting, despite your care.

- - -

>I give up trying to get you to address the topic of the actual partially punched ballots.

I don't have enough time or energy to address all topics brought up in this thread. I didn't realize you considered my addressing of that particular topic to be of special importance. Earlier in this thread, I concentrated on the article and Glaeser's reasoning.

>The problem we're dealing with here is the intended SCOPE of Glaeser's article.

I have no problem with his intended scope until he broadens it in his conclusions to a blanket disapproval of selective recounts without considering nonstatistical aspects.

Had Glaeser properly and consistently qualified and limited his conclusions by saying that they were valid only in a statistical sense, admitting that there were important nonstatistical considerations that could lead to a different overall conclusion, that would be more acceptable. But he didn't.

>By my reading, he is specifically and narrowly addressing the statistical issue of partial recounts by different methods.

Okay, until he comes to a conclusion that is broader than that specific narrow scope.

>It seems unfair to criticize Glaeser for failing to make an unrelated point

It would be, but I'm criticizing Glaeser (from the points of view of whether the article's balanced and whether his conclusions are justified) for failing to make _related_ points.

>Glaeser isn't discussing WHAT the candidates' rights are,

Correct.

And because he doesn't discuss that, among other things, his conclusion about selective recounts fails to be valid for the real world.

>But what does cost have to do with statistical theory?

Cost has to do with the conclusions Glaeser draws. Omission of cost considerations makes his conclusions unsuitable to the real world election situation.

>>>>c) Given that candidates pay the cost of recounts, < snip > >This is as true as it is irrelevant, namely completely.

It is relevant to the issue of the right to partial recount versus a forced choice of either statewide recount or nothing.

>Partial counts by new procedures bias the total. Period.

No. Not "period". In the specific case we are discussing, the partial recounts bias the total in the direction of correctness.

Each county's total that is recounted by hand is almost surely more accurate, or at least no less accurate, than the same county's total counted previously by machine.

Of course the statewide total would be more accurate if all counties were recounted by hand than if only part of the counties were recounted by hand. So why doesn't Bush agree to a statewide recount by hand? If he's interested in accuracy of result, that is.

>>>>>d) That "biased" is not automatically bad. A bias toward accuracy and away from error is usually considered desirable.<<<< >Not true.

Not true? You mean a bias toward accuracy and away from error is _not_ usually considered desirable? Is that what you really intended to say?

Or just that in the current situation, Bush and his supporters do not consider accuracy desirable?

>You need to entertain the idea that some people might favor Bush and still have valid points.

Do you have any evidence that I don't? If not, why did you make this statement? If you do, what is it?

I haven't objected to everything you've asserted. I think you have valid points. But I don't argue much about the points I consider valid.

>But you seem to take this to an extreme, leaving the distinct impression that as far as you are concerned, an observation that favors Bush simply cannot be objectively true under any circumstance.

Flint, that is clearly an exaggeration (i.e., a form of misrepresentation), even though an opinion.

Can you point to any evidence that genuinely led you to think that I seemed to take the idea that Glaeser is not disinterested to "an extreme" or that I consider that "an observation that favors Bush simply cannot be objectively true under any circumstance"? If so, please specify what that evidence is. If you can't do that, please retract and apologize.

>You are absolutely correct that Glaeser's article is a persuasion piece. You are correct that it makes too many assumptions to hold water all by itself.

Weren't you, just a little earlier, arguing that the article _did_ hold water all by itself?

>But the list you created does NOT consist of valid objections.

It consists of points that are relevant to Glaeser's conclusions, the omission of which, among others, renders his conclusions invalid.

>What the candidates have the right to do, what they must pay to do, what they must be willing to pay to do, what reasons they might have for doing various different things, are beside the point. The point is what they *ACTUALLY DID* do,

... and they *ACTUALLY DID* have equal rights and oppotunity to request hand recounts by county, they *ACTUALLY DO* have to pay for hand recounts, and they *ACTUALLY MUST* be willing to pay for the recounts they request.

>whether that introduces a statistical bias into the results.

... a bias toward more accurate counts.

>And as my simple example above illustrated, it does NOT necessarily make the final result more accurate.

No, your simple example illustrated your failure to understand some basic principles of error analysis.

>In fact, nobody denies that Gore expects to gain by this stratagem. Not gain some hazy notion of "accuracy" at all. Gain VOTES!

... just as Bush would expect to Gain VOTES if there were hand recounts in some other Florida counties.

>What Gore is doing is perfectly fair.

Does that mean you withdraw your objections, if any, to a partial manual recount in Florida?

>But why should Bush trade DEFINITE for VERY PROBABLY?

That's not the choice he faced. He didn't have "definite". He had to decide whether to trade one set of "very probably"s and "probably"s for a different set of "probably"s that might or might not be "very".

We've already seen that one of his tradeoffs was a bummer: he lost in his first attempt to have a federal court override state law. He traded his supposedly-"definite" (judging by what he claimed, anyway) -- well, make that a "very probably" -- preference for states' rights for a blatant attempt to override Florida's right to govern its election.

-- No Spam Please (nos_pam_please@hotmail.com), November 17, 2000.


Flint,

Your most recent posting appeared while I was still composing my most recent one, so mine does not take into account what you said there.

-- No Spam Please (nos_pam_please@hotmail.com), November 17, 2000.


Eve,

>A flaw in an argument can easily encompass an omission. To add "omissions" to what I said is therefore redundant -- that's why I didn't see the need to mention "omissions" separately.

If you already realized that Glaeser's argument had important omissions, why did you ask Patricia whether she saw flaws? >I mean, if he had no ill intent, even if you prove him wrong, of what relevance would an apology be?

It demonstrates comprehension of the mistake.

>I'll say something like, "You know, you're right. I stand corrected."

That's an apology. Flint has not yet said something like that.

>Are you saying I should add, "I apologize" to this?

No, the specific word "apologize" is not essential. But an expression of consciousness of the mistake made, is.

-- No Spam Please (nos_pam_please@hotmail.com), November 17, 2000.


Flint,

Okay. I'll deal with some easy ones, also.

>In the remainder of that post, you will find that I detailed sample ballots and asked him to interpret them. He pretended I asked no such question,

Oh? How did you determine that I pretended you asked no such question?

Are you saying that lack of time to respond to all questions implies a pretension that questions not responded-to were not asked?

>I think it's frankly stupid to point out that the winner could legally have demanded a recount also, and to fault him for not doing so. How many different ways can I say that once you have won, a recount can NOT help you.

So you've determined that Bush is the "winner", not merely "the candidate with the leading vote total so far announced"?

Bush _has not yet won_. (For one thing, remember those absentee ballots?) A recount in counties favorable to him _can still_ help him. Except that he passed up his ordinary chance to request such.

>I believe if the situation were reversed, then what both candidates are doing would also be reversed.

Does that mean that you would then be arguing that partial recounts were okay, when following established legal rocedures?

>These are elementary game strategies, so self-evident there ought to be no need to point them out.

Does that mean that you acknowledge that the arguments you have been presenting in this thread have not been objective on your part, but have been consciously intended as game moves (and I too mean that in the sense of game theory, not as a trivialization)?

>Dismissing it as "opinion" for its location in the paper alone is, uh, not too bright.

But classifying it as "opinion" on the basis of its content is okay.

>Now to No Spam's rather confused but very detailed statistical critique.

I presented an error analysis, not a statistical critique.

>No Spam, you have carefully set things up so that one simply cannot disagree with you without "misrepresenting" you.

No, I haven't. There is a difference between saying, "I disagree with your statement '1 + 1 = 2'" and saying "Your '1 + 1 = 3' statement is absurd" when my actual statement was "1 + 1 = 2".

>The idea that you might be wrong is cleverly disallowed by this technique.

... by the hypothetical technique to which you refer, maybe. But that's not what I've done, so stop claiming I did.

>Now, how about being honest for a change?

I am being honest, and have been all the way through.

How about responding to my honest statements as though you recognized they were honest?

>You write, [we can draw valid conclusions by comparing the precision- device measurements of the men with the rubber-ruler measurements of the women, right? That is, according to the logic you used in your example, right?] >No, absolutely the contrary. You cannot draw correct conclusions by comparing apples and oranges.

But _you wrote in your example_ that "our increased accuracy led us to the wrong conclusion", NOT that your comparison of apples to oranges led to the wrong conclusion!!! It's right up there in plain text for everyone to see, Flint!!!

My point was that your example's error analysis was mistaken!

You attributed the error in conclusion to the wrong source!

-- No Spam Please (nos_pam_please@hotmail.com), November 17, 2000.


No Spam Please,

Regarding Patricia:

[If you already realized that Glaeser's argument had important omissions, why did you ask Patricia whether she saw flaws?]

When did I say Glaeser's argument had important omissions? The point I was trying to get across to Patricia was that if she wants to show bias in this guy's essay and/or flaws in his arguments, she needs to come up with REASONS -- reasons that specifically address the merits of THIS (Glaeser's) ESSAY. These reasons may include, for example -- in theory -- an omission or omissions that she feels should have been part of his argument. My apologies to you both if I miscommunicated.

Regarding Flint:

[the specific word "apologize" is not essential. But an expression of consciousness of the mistake made, is.]

To the extent there's a mistake, and that he doesn't express consciousness of the mistake, why not simply assume he's simply not yet aware of and/or doesn't understand any flaw or flaws in his position, instead of waiting for a concession or "apology"?

At that point, you can attempt to reformulate your argument in many different ways, or take the Socratic method -- simply question him, starting with his premises, his axioms, etc. Then probe, probe, probe -- until absurdities start popping out -- until they become so obvious that no one need continue, or until your opponent gives up. Or until you realize you're in the wrong. If you've got a winning hand, in many cases you then won't even need a formal concession. Both parties and most knowledgeable, interested observers would usually have a pretty good idea who came out on top.

-- eve (eve_rebekah@yahoo.com), November 17, 2000.


Flint, come down off your damn high horse. Did you miss the part where I apologized because I hadn't read your previous response (you know, where I indicated that you had ALREADY ADDRESSED THOSE POINTS)?

And I'm the one who's "not too bright"?

Additionally, as you seem to feel you are the only one with any grasp of reality, why don't YOU define an OP-ED PIECE for us?

-- Patrica (PatriciaS@lasvegas.com), November 17, 2000.


Eve,

>When did I say Glaeser's argument had important omissions?

You didn't.

But you did write, "A flaw in an argument can easily encompass an omission. To add 'omissions' to what I said is therefore redundant -- that's why I didn't see the need to mention 'omissions' separately."

If we back up to your preceding post, to see "what you said", we find that you said, "Do you see bias or other flaws in the argument itself? If so, would you address them directly?"

Now, at the time you wrote that you either had realized that Glaeser's argument had omissions, or you hadn't. If you hadn't realized that, then are you now trying to claim after-the-fact that your question to Patricia was consciously intended to include "omissions" when you wrote "flaws", when in fact you didn't so intend at that time? If the answer to that latter question is "no", then to add "omissions" to what you said was _not_ redundant.

If you _had_ realized, at the time you wrote that question to Patricia about flaws, that Glaeser's argument had omissions, then why did you ask Patricia whether she saw flaws, when you were already aware of flaws?

> The point I was trying to get across to Patricia was that if she wants to show bias in this guy's essay and/or flaws in his arguments, she needs to come up with REASONS -- reasons that specifically address the merits of THIS (Glaeser's) ESSAY. These reasons may include, for example -- in theory -- an omission or omissions that she feels should have been part of his argument. My apologies to you both if I miscommunicated.

Regarding Flint:

[the specific word "apologize" is not essential. But an expression of consciousness of the mistake made, is.]

To the extent there's a mistake, and that he doesn't express consciousness of the mistake, why not simply assume he's simply not yet aware of and/or doesn't understand any flaw or flaws in his position, instead of waiting for a concession or "apology"?

At that point, you can attempt to reformulate your argument in many different ways, or take the Socratic method -- simply question him, starting with his premises, his axioms, etc. Then probe, probe, probe -- until absurdities start popping out -- until they become so obvious that no one need continue, or until your opponent gives up. Or until you realize you're in the wrong. If you've got a winning hand, in many cases you then won't even need a formal concession. Both parties and most knowledgeable, interested observers would usually have a pretty good idea who came out on top.

-- No Spam Please (nos_pam_please@hotmail.com), November 17, 2000.


Oh, rats.

Eve, I clicked on Submit just as I was realizing that I hadn't finished my response. So in my preceding posting, the last intentional line was "If you _had_ realized, ... already aware of flaws?"

-- No Spam Please (nos_pam_please@hotmail.com), November 17, 2000.


Eve,

(continuing from where I left off)

>My apologies to you both if I miscommunicated.

Okay.

And I apologize for not having finished editing my last-but-one posting. The part at the top came out complicated, and I would've simplified.

Shoot.

Okay, multiple miscommunication here.

(*takes deep breath, counts to one hundred the way our choir director taught three decades ago [she could do two hundred], and lets out last remnants of deep breath*)

- - -

>To the extent there's a mistake, and that he doesn't express consciousness of the mistake, why not simply assume he's simply not yet aware of and/or doesn't understand any flaw or flaws in his position

His mistakes are misleading other readers of this thread about a very important matter, the U.S. presidential election.

>you can attempt to reformulate your argument in many different ways

I am/will.

>take the Socratic method -- simply question him, starting with his premises, his axioms, etc. Then probe, probe, probe -- until absurdities start popping out -- until they become so obvious that no one need continue, or until your opponent gives up.

Yeah -- that's my favorite, but it takes soooooooo long to do online when the subject's this complicated.

>Or until you realize you're in the wrong.

If someone shows me, I will.

-- No Spam Please (nos_pam_please@hotmail.com), November 17, 2000.


Eve:

Glaeser The simple fact that the arguments presented in his name are not possible to defend. Even he would be embarassed. Now interpretation of the correct explanation of his arguments are OK.

Your continual defense of a flawed presentation, leaves questions in my mind about the honesty of your mission. But then you consider Ayn Rand a meaningful person. Oh Well.

Best wishes,,,,

Z

-- Z1X4Y7 (Z1X4Y7@aol.com), November 17, 2000.


OK, more responses. We are getting our posts out of order, but I think we can piece together which post responds to which with a little effort.

hmm:

[Why do those last two quotations show only part of the sentences involved instead of the whole sentences?]

Because I extracted the only portions of the sentences I considered critical. You know, the subject and the verb, that sort of thing. No Spam's subsequent posts make it clear that he was trying to distinguish between two different "senses" of voter intent, though he does not deign to explain what these senses might be. After much puzzling, the closest I can come is that he considers the critical difference to be whether the voter intends to "punch", or whether the voter intends to "vote", which of course makes a clear difference between punching holes and voting. Or something like that. Do you have a better guess? (hint: NO misrepresentations allowed in your guesses).

No Spam:

[I didn't realize you considered my addressing of that particular topic to be of special importance.]

But that particular topic is the ONLY possible justification for doing hand counts in the first place. If there were no possible differences between machine and hand counting, the entire issue would be moot. Glaeser's point, with which you appear to agree, is that manual counts can produce different results. That difference might be critical. HOW that difference comes to pass is therefore just as critical. Am I seriously intended to believe you don't realize this?

[I have no problem with his intended scope until he broadens it in his conclusions to a blanket disapproval of selective recounts without considering nonstatistical aspects.]

Well, I keep trying to explain why nonstatistical aspects are irrelevant and your objection is nugatory. I will keep trying, what else can I do?

[In the specific case we are discussing, the partial recounts bias the total in the direction of correctness.]

Yes and no. But I addressed this in detail in the other post, so I won't belabor it here. I'll just repeat that "correctness in Gore's favor" is Orwellian.

[Of course the statewide total would be more accurate if all counties were recounted by hand than if only part of the counties were recounted by hand.]

Now I know to distinguish between "more accurate" in the sense of determining the magnitude ofthe final counts, versus "more accurate" in the sense of picking the winner.

[So why doesn't Bush agree to a statewide recount by hand? If he's interested in accuracy of result, that is.]

Because the current counts place him ahead, which is all that matters. I'm running out of different ways to say that Bush has nothing to gain and everything to lose by agreeing to ANY new counts of any description whatever. NEITHER SIDE IS INTERESTED IN ACCURACY! BOTH SIDES ARE INTERESTED ONLY IN WINNING! Good Grief, man! Wake up!

[You mean a bias toward accuracy and away from error is _not_ usually considered desirable? Is that what you really intended to say? Or just that in the current situation, Bush and his supporters do not consider accuracy desirable?]

No Spam, let's put it this way. Let's say that Candidate A IN FACT got 300 more votes than candidate B, *provided* that EVERY ballot were properly counted. By definition, candidate A should win the election. NOW, let's say that due to counting errors, we have missed exactly 1000 votes for EACH candidate. Candidate A still has 300 more than candidate B, but both of their totals are 1000 votes short of fully accurate.

NOW, let's say that through a carefully aimed partial recount we add in Candidate B's 1000 votes, while adding none to candidate A. Yes, in your sense the totals are now "more accurate", because candidate B's total is now precisely accurate whereas before it was not. However, now the situation is that THE WRONG CANDIDATE WON THE ELECTION! If you favor candidate A, this "correction" was a "misinterpretation" because it only told part of the story and made key omissions. If you favor candidate B, then this was an improvement because the counts were "more accurate" and this is "considered desirable". Right? Faugh!

How very sensitive you are to omissions (real or imagined) that work against you, and how very hard you work to deny they make any difference when it works in your favor. Isn't that amazing?

[Do you have any evidence that I don't?]

Only indirectly. Many people have written many things in support of both candidates. You have agreed with pro-Gore points and disagreed with pro-Bush points without exception. So it's like if all your customers are north of Main Street and all your competitor's customers are south of Main Street, no one needs direct evidence that you and your competitor every communicated. Guilty!

[If so, please specify what that evidence is. If you can't do that, please retract and apologize.]

I continue to point out, and explain in detail, that your objections to Glaeser remain irrelevant. (Except for your misunderstanding about accuracy, explained in my other post and reiterated above). I sincerely feel that I HAVE specified. Should I now demand that you realize this and apologize for missing the point and being so thick- headed? Come on.

[Weren't you, just a little earlier, arguing that the article _did_ hold water all by itself?]

No, I was arguing that your non-statistical objections were irrelevant. If someone proposes a theorem, it is NOT a valid criticism to argue that said person *could* have proposed some other theorem and failed to mention this. You must critique Glaeser on his merits and within the intended scope of his argument, NOT dismiss him on the grounds that he FAILED to make some UNRELATED argument.

[It consists of points that are relevant to Glaeser's conclusions, the omission of which, among others, renders his conclusions invalid.]

Glaeser's conclusion was that "rules generally work better than discretion." Fine. I utterly fail to see how the relative cost of following different rules changes this principle, or even relates to it at all.

[No, your simple example illustrated your failure to understand some basic principles of error analysis.]

When we dug deeper, we found we were saying the same thing, as close as I could follow your sentences. But just to be sure, a few paragraphs up I presented yet ANOTHER illustration of the same basic point. In any case, if I am in error, I'm as unable to see it as you are if you are in error. ONE of us is in error here, and we remain equally convinced it's the other guy. To me, the key difference is I *illustrate* why you're wrong, and you are content to *assert* that I'm wrong.

[... just as Bush would expect to Gain VOTES if there were hand recounts in some other Florida counties.]

Yes, certainly this is true. So you are now admitting that accuracy is irrelevant, and these recounts are to change ratios? If so, congratulations!

[Does that mean you withdraw your objections, if any, to a partial manual recount in Florida?]

Nice try. I agree that the law, in its majestic equality, permits the winner as well as the loser to keep counting in the hopes of changing his status. (Apologies to Anatole France). What the two parties are doing now is partly PR and partly legalistic. The primary goal is to win, whatever it takes. The secondary goal is to look (in the public's mind) like the win was deserved, but this is *very* secondary.

I love what Peggy Noonan wrote:

>>>>...the Gore-Clinton Democratic party is trying to steal the election. Not to resolve it--to steal it. That is, they are not using hand-counting to determine who won, they are using hand-counting to win. They are attempting to do this through chicanery, and by interpreting various ballots any way they choose. As in, "This ballot seems to have a mild indentation next to the word Bush. Well, that's not a vote. Person might have changed his mind. This ballot seems to have a mild indentation for Gore; the person who cast this ballot was probably old, and too weak to puncture the paper card. But you can see right here there's a mark kind of thing. I think that's a vote, don't you Charley?" "Oh yeah, that's a vote all right."<<<<

Of course, this addresses the key issue here, so important that it just never occurred to you that I might want you to face it simply because I asked you to.

--------

OK, now on to No Spam's next post. Maybe we'll get back in sync here. Good thing I can't respond while I'm at work, or we'd never get synched back up.

[How did you determine that I pretended you asked no such question?]

By neither answering it nor referencing it. Maybe I should have said you ignored it? Do you see a substantive difference between ignoring something and pretending it's not there?

[So you've determined that Bush is the "winner", not merely "the candidate with the leading vote total so far announced"? Bush _has not yet won_. (For one thing, remember those absentee ballots?) A recount in counties favorable to him _can still_ help him. Except that he passed up his ordinary chance to request such.]

OK, ordinarily the winner is the one with more votes. You have my permission to go back and replace "winner" with "the candidate with the leading vote total so far announced" wherever it appears, if you wish. This doesn't change the sense of anything I wrote. I will certainly agree that IF the need should arise and IF he had foreseen that need, THEN in hindsight a retaliatory recount in favorable counties would have been a superior strategy. We don't yet know whether Bush has selected the optimum strategy, because we don't yet know whether the actors in this drama will decide that "more votes" means "wins" yet.

[Does that mean that you would then be arguing that partial recounts were okay, when following established legal procedures?]

This is a damn good question, and I don't know the answer. I suspect that our relative positions would indeed be reversed in this discussion. It seems to me that both of us are making as strong a case as we can for what we see as the good and the right, namely our preferred candidates.

[Does that mean that you acknowledge that the arguments you have been presenting in this thread have not been objective on your part...]

Yes, of course. I make no pretense of being objective. I admit I'm somewhat grateful that I am arguing the stronger side, but I'd still make the best case I could either way. And I do try to make a case honestly. I'm making every attempt to address my best understanding of the points you make. I regard you as the loyal opposition, doing your best with inferior material, but I don't keep demanding you apologize for being wrong. I regard being wrong as inherent in the weaker position.

[There is a difference between saying, "I disagree with your statement '1 + 1 = 2'" and saying "Your '1 + 1 = 3' statement is absurd" when my actual statement was "1 + 1 = 2".]

Granted. And there's yet another difference between saying 1+1=3, and then later claiming that the 3 was really a 2 after all when caught out. As I said, I am responding to my best understanding of your meaning. If you don't feel I'm doing this deliberately, then what might be the problem? Maybe I don't WANT to understand your points. Maybe you aren't expressing them very well. Maybe you make hazy arguments that can be interpreted several ways, so that no matter which interpretation I pick, you can claim you meant another.

[... by the hypothetical technique to which you refer, maybe. But that's not what I've done, so stop claiming I did.

>Now, how about being honest for a change?

I am being honest, and have been all the way through.]

I believe you HAVE BEEN using this technique, and your denial is disingenuous. However, I'm willing to accept that we have a failure to communicate despite a good faith effort to understand what is really intended, by both sides. Even though sometimes this is hard to maintain...

[But _you wrote in your example_ that "our increased accuracy led us to the wrong conclusion", NOT that your comparison of apples to oranges led to the wrong conclusion!!! It's right up there in plain text for everyone to see]

I'm not sure whether you are picking nits here, or drawing distinctions without differences, or bantering with semantics, or really making a valid point. I'll assume you are making a point. In which case, you are correct. I intended strong sarcasm. Of course it wasn't accuracy per se that caused error. Perhaps I should have said that our originally correct conclusion was reversed *despite* greater accuracy? I explicitly said "The increase in accuracy was, unfortunately, more than offset by our methodological error of partial remeasurement." This remains true. The methodological error was in comparing apples (the rubber ruler measurements) with oranges (the precision measurements).

In any case, the point (with which you seem to agree) is that increased accuracy, selectively applied, does not necessarily improve the final result, and can make it worse. For another example, see my "1000 votes" example earlier.

[You attributed the error in conclusion to the wrong source!]

Yes I did. I admit I should have made some indication like HEAVY SARCASM--->>>. To make sure nobody misunderstood, I quickly clarified (see above) that it wasn't the increased accuracy that caused the error, but rather its selective application. Apparently this quick clarification wasn't sufficient to make sure *nobody* misunderstood. But as we have both learned, you just can't express yourself clearly enough for someone *determined* to misunderstand you.

However, the point that selective increases in accuracy can bias the overall result remains true. And the fact that Gore is trying to do exactly this to win the election remains true. And the fact that it may be legal to do so, and it may be fair to do so (because Bush could also have done so), does not change the fact that this is a methodological error capable of generating incorrect results.

[>Or until you realize you're in the wrong.

If someone shows me, I will.]

Not quite. What you mean is, if someone demonstrates *to your satisfaction* that you are wrong, you will change your mind. This qualification, IF it's impossible to meet, renders your statement misleading in the extreme. I believe *you* are the one whose "mistakes are misleading other readers of this thread about a very important matter." I'm doing all I can correct what I regard as your errors. The Socratic method has bugs [grin].

-- Flint (flintc@mindspring.com), November 17, 2000.


Because I extracted the only portions of the sentences I considered critical. You know, the subject and the verb, that sort of thing.

But as we saw, the subject and verb alone were insufficient for comparison.

His first full sentence was:

People can determine the intent of a voter who has punched partly, but not all the way, through a punch-card ballot better han manchines, generally.

You responded with:

Now, did the voter *intend* to punch a hole?

And he clarified with:

But I didn't write about the voter's intent to punch a hole. I wrote about the intent of a voter who _did_ punch a hole, but only partially, not completely.

It makes a difference when you add the original text back in, as you illustrate below:

No Spam's subsequent posts make it clear that he was trying to distinguish between two different "senses" of voter intent,

Actually, it wasn't in a subsequent post. He made it clear when he said . . . But I didn't write about the voter's intent to punch a hole. I wrote about the intent of a voter who _did_ punch a hole, but only partially, not completely. . . Your response to this quote indicated that you still did not understand that he was distinguishing between two senses of voter intent, even though he said just that.

though he does not deign to explain what these senses might be.

Of course he did, as you illustrate below:

After much puzzling, the closest I can come is that he considers the critical difference to be whether the voter intends to "punch", or whether the voter intends to "vote", which of course makes a clear difference between punching holes and voting.

Indeed it does, though it's unfortunate that it took "much puzzling" for you to figure it out.

Or something like that. Do you have a better guess?

No, you had it right. However, you should be aware that your inability to realize the distinction he was making between the two types of "intent" was not necessarily shared by all.

-- (hmm@hmm.hmm), November 18, 2000.


hmm:

And do you count angels on pinheads in your spare time? Just what do you think those people were doing with those ballots, punching or voting? This is like arguing whether the victim was killed by the bullet, the gun, or the shooter. The answer is YES to all three. You have the smarts, I think, to be uncomfortable that your position forces you away from the issues under discussion and toward splitting hairs and playing semantic games. At least it should.

Let's summarize rather than trying to redirect, OK? There are two basic problems with the Gore recount.

1) The disposition of numerous ballots depends on WHO is interpreting them. This is unavoidable. I gave several specific examples of this problem (as did Peggy Noonan), and note with interest that you, No Spam, and Patricia are very carefully avoiding this specific question, and focusing instead on the absurd "distinction" of whether one votes by punching or punches to vote! Why?

2) Even if a consistent set of rules could be followed unambiguously, *where* those rules are followed and NOT followed is critical. I have provided several clear illustrations of this principle, and again note with interest that No Spam dodges the principle and focuses on whether my precise wording can be misinterpreted if he tries hard enough. And the issue being illustrated? Let's pretend it doesn't exist and start hollering about apologies and splitting hairs and basically ANYTHING other than the topic at hand.

When this kind of weaseling is the only response I can solicit, I'm pretty damn comfortable that I have the stronger hand here. But YOU should feel a bit distressed at how enthusiastically willing you are to get distracted by the irrelevant. You should be holding yourself to higher standards than this, and you usually do.

-- Flint (flintc@mindspring.com), November 18, 2000.


Just what do you think those people were doing with those ballots, punching or voting?

Both. How is this relevant to my previous post?

You have the smarts, I think, to be uncomfortable that your position forces you away from the issues under discussion and toward splitting hairs and playing semantic games.

That depends on what the issue under discussion is. There happen to be many in this thread. The issue I have focused on is the discrepancy between what No Spam is saying and what you are responding to. You may not feel that this is relevant, and avoid the issue by simply calling this "splitting hairs and playing semantic games" but the fact remains that the statements he made were not quite the same as the ones you quoted. And your subsequent argument about "intent" was based entirely on these misquotes. I have no other issue with your arguments at this time. Just that one.

Let's summarize rather than trying to redirect, OK? There are two basic problems with the Gore recount.

That's nice. Except that these have nothing to do with my previous post. In effect, it is now you who are trying to redirect.

When this kind of weaseling is the only response I can solicit, I'm pretty damn comfortable that I have the stronger hand here.

Then perhaps congratulations are in order on your strong hand.

But YOU should feel a bit distressed at how enthusiastically willing you are to get distracted by the irrelevant.

That depends on what you consider relevant. I believe that the issue of arguing someone's points by accurately refuting what he actually said is relevant in this case. Obviously, you deem it a "distraction" which is why you've changed the subject yet again. So it goes.

-- (hmm@hmm.hmm), November 18, 2000.


No Spam (and others):

Thanks for taking the time and effort to keep Flint occupied in this thread. Do not be discouraged that Flint gives ground grudgingly, if at all. Do not be dismayed that he can tangle up the thread faster than you can successfully disentangle it. These are givens.

There is a term I've seen bandied about on UseNet: squink. It cleverly combines the words "squid ink". It refers to the defense mechanism of the squid (and octopus) of squirting a cloud of ink into the water to confuse a predator. Flint is a master of argumentative squink. What is maddening (if you permit it to affect you) is that Flint's particular brand of squink mimics a sincere effort to get to the bottom of some misunderstanding between you. Meanwhile, he is furiously undermining all attempts to actually clarify the argument, introducing red herrings and generally throwing a cloud of unknowing and confusion over all.

I sympathize with your plight. If you endure, if you patiently unravel each and every knot Flint has tied, you will exhaustedly look up from your task, days later, only to find the thread has run to a length, and reached a degree of tedium that it ensures Flint has been unmasked before the smallest possible audience.

If you ever tangle with Flint again, just be sure to understand that he is not sincere in his protestations of sincerity. When Flint argues, it is not to arrive at an understanding - he believes he has already arrived. He has no interest in an exchange of views. His views are set. He sees your position, to the exact degree it doesn't correspond with his own, as something to assail, to undermine, to destroy.

Flint appears to argue from a conviction that it is futile to try to change your mind - rather he sets out to demean you and your position so that your ideas cannot infect others. He knows that others do have neither the attention nor the attachment to your position that you do, so they will fail to detect the subtle distortions he filters into your ideas, the dose of arsenic he slips into the drink.

Knowing Flint, he will respond to this post by pointing out that I am indulging in name calling. He will assume a stance of moral superiority and dismiss such stuff as not serious argument. And he will NEVER FAIL to draw the conclusion that, having stooped to name calling, it proves that in my own estimation my position has been destroyed, is no longer defensible and he must have prevailed in order to have provoked this Parthian shot.

Flint:

You are very predictable. Now that I have seen you at close hand over along time I see you do not change. Your tactics have been honed to fine edge and you trust them to always carry you through. The trouble is that, by treating argument as the moral equivalent of warfare, you easily abandon morality when, where, and as it suits you. This may win battles. It also defines victory as the only measure of right and wrong.

Because you adopt a win-at-all-costs position, you are Jesuitical to the core. While you unfailingly assume a cloak of moral authority, you put it on and off depending on how the wind blows. Each one of us who has seen underneath that cloak knows there is something ugly in you.

Read Dante's Inferno again, Flint. It was written for your instruction.

-- Brian McLaughlin (brianm@ims.com), November 18, 2000.


Texas Law Allows Voter "Intent"

http://hv.greenspun.com/bboard/q-and-a-fetch-msg.tcl?msg_id=0045yI

-- (also@see.this), November 18, 2000.


Golly. If any of you (hmm, Brian, No Spam) would care to look at the post that started this thread, the issue is whether selective recounts, *even if more accurate*, might produce an incorrect verdict as to who got the most votes. Glaeser (remember him?) said this was possible. I agree, and have provided several examples of how this can be. I continue to support this position. Do you need more examples?

Now it seems I am being blamed for TRYING to keep to this original topic, while those who cannot or will not respond to Glaeser's point have decided that their purpose is best achieved by changing the subject and attacking the person rather than the argument. And *I* am accused of changing the subject when I try to drag it back to Glaeser's original point. Don't any of you see a problem with this?

So now this thread has two themes. The first has to do with the bias introduced by selective recounts, and this theme is being addressed by those who appreciate what it implies. The second theme is to attack Flint, change the subject, blame Flint for NOT changing the subject, and otherwise do everything possible to keep from addressing the idea that started this thread. And this second theme is followed by those with the intelligence to recognize that Glaeser's argument is deadly to their positions. This is not a coincidence. Indeed, it's kind of transparent and a bit pathetic.

Now, back to the actual subject of the thread. Do any of my critics dare to talk about it? Hello?

-- Flint (flintc@mindspring.com), November 18, 2000.


Personally, I support a statewide manual recount of the votes in the interests of fairness. Otherwise, there are always going to be large numbers of people who will not believe in the legitimacy of this election no matter who officially wins.

I believe the Democrats have the right to ask for recounts where they think the vote is close or where there may have been voting irregularities. It's too bad the Republicans have not exercised their same rights.

-- (Recount@em.all), November 18, 2000.


Candidates Raise Cash for Recount

http://hv.greenspun.com/bboard/q-and-a-fetch-msg.tcl?msg_id=0044N7

-- The (election@of.2000), November 18, 2000.


Flint,

You act, once again, as if you have done nothing to merit the complaints made against you. You act, predictably, as if they are a ruse to avoid addressing the main issue of this thread. You pretend, as usual, that such complaints are a tacit admission that your stronger arguments cannot be answered, so that such misdirection is necessary to avoid 'losing' the argument.

Just as there are fouls in basketball and penalties in hockey, there are foul ways to argue. You persist in using as many of them as you can get away with.

A man who can construct an argument as well as you can, can understand his opponent's points quite well. You can't plead incapacity. Yet, you consistantly do injustice to your opponents' arguments - misrepresent them, distort them and add or subtract from them to suit your own purposes.

I am not involved in this argument. I came in to explain you to No Spam, who is probably still laboring under the illusion that you are an honorable man. You are not. I will point this out from time to time, whenever you abuse others in your accustomed fashion. I do it because I feel like it. Call it my revenge for your having done me as you are doing them.

-- Brian McLaughlin (brianm@ims.com), November 18, 2000.


Brian:

I have not intentionally distorted No Spam's words. I have addressed them according to my best understanding of what he has said. I may have missed his point in some cases, and if so this is my shortcoming.

I continue to believe that selective recounts, *even if accurate*, can cause the wrong candidate to be selected as "winner", and I think I have shown this to be true in principle. I firmly believe that this principle remains true whether or not such a recount is legal, or whether both parties could have used the same technique, or whether or not recounts cost money to the party requesting them.

As far as I can tell, No Spam has been arguing that these extraneous factors change the underlying principle. I agree that they are salient to the overall situation, but that Glaeser's scope did not attempt to address the overall situation, but ONLY the strategem of selective recounts.

Perhaps the argument that people bring bias to the interpretation of individual ambiguous ballots should not have been brought up, since it too (while true) is beside the point and muddies the water. If I brought it up, then I should not have done so.

I concede it is entirely possible that I put words into other people's mouths that they didn't intend to say, or extract meanings from their statements that they didn't mean. But despite your claim, this is incapacity on my part. I'm doing my best. And it's interesting that this "tendency" seems to become visible to you only when you disagree with my position. The sides drawn in this thread fall entirely along party lines. As you say, my approach is consistent. Yet last year you did not choose to break into arguments I was having with doomers, to call me ugly and mendacious and otherwise deploy gratuitous insults. Doomers did that. You might reflect on this. I was called all kinds of names last year. Again along party lines. Coincidence?

Recount:

I agree with you entirely. Bush should have agreed to a statewide hand recount, which he is quite likely to win with a far greater overall sense of legitimacy. There might be a somewhat larger probability that his chosen course will net him the presidency, but at great cost to any ability to actually lead. And as things stand, BOTH candidates can properly be viewed as seeking the office via a trick, not helpful no matter who wins. This is really a shame.

-- Flint (flintc@mindspring.com), November 18, 2000.


the argument about a selective recount is bogus because Gore offered to count the whole state.

according to Flint all our great presidents like Jefferson, Lincoln, Teddy Roosevelt, are all illegitimate because they were elected by hand counting. time to take Washington down off Mt. Rushmore.

-- -- (count@em.all), November 19, 2000.


Well, y'all, I just can't keep up with this debate. Partly because I'm just mesmerized with the latest developments. And I just heard that counters were being charged with EATING CHADS! (caution: unsubstantiated rumor) This is getting funnier than watching it on Comedy Central!

Anyway, what I've had the time to get through here does make for an instructive and absorbing read.

I would like to respond to a couple of points directed at me, though.

No Spam,

My previous comments are headed by a carrot. Your responses are in brackets.

>When did I say Glaeser's argument had important omissions?

[You didn't. But you did write, "A flaw in an argument can easily encompass an omission. To add 'omissions' to what I said is therefore redundant -- that's why I didn't see the need to mention 'omissions' separately."

If we back up to your preceding post, to see "what you said", we find that you said, "Do you see bias or other flaws in the argument itself? If so, would you address them directly?"

Now, at the time you wrote that you either had realized that Glaeser's argument had omissions, or you hadn't. If you hadn't realized that, then are you now trying to claim after-the-fact that your question to Patricia was consciously intended to include "omissions" when you wrote "flaws", when in fact you didn't so intend at that time? If the answer to that latter question is "no", then to add "omissions" to what you said was _not_ redundant.

If you _had_ realized, at the time you wrote that question to Patricia about flaws, that Glaeser's argument had omissions, then why did you ask Patricia whether she saw flaws, when you were already aware of flaws?]

With respect to the point that Glaeser was trying to get across, I saw no significant omissions. My comments to Patricia were simply of a hypothetical nature that she should focus on any flaws (the concept of which includes significant omissions, since an omission is a type of flaw) to the extent such flaws existed at all. Again, if you found this difficult to follow, focus on what I'm saying now.

Z,

You had said to me,

[Your continual defense of a flawed presentation, leaves questions in my mind about the honesty of your mission. But then you consider Ayn Rand a meaningful person. Oh Well.]

You know, Z, although you drive me nuts sometimes, I love ya anyway. :)

-- eve (eve_rebekah@yahoo.com), November 19, 2000.


I finally heard from someone regarding the six Republican counties that had been hand-recounted. I still haven't received any information on WHY these six counties were recounted.

"Included in the Tuesday, 11/14 tallies are the results of hand counts from Franklin, Hamilton, Taylor, Washington, Lafayette and Seminole Counties - All Bush-dominated counties - where Bush gained hundreds of votes on the recount."

-- Anita (Anita_S3@hotmail.com), November 19, 2000.


Although I STILL haven't heard exactly why those six counties had a hand recount, I WAS sent this article demonstrating that only 3/4 of Florida counties engaged in a MACHINE recount. Franklin County IS mentioned therein, and I don't see how anyone could know what they know if they didn't look at the ballot manually.

Votes have NOT been counted AGAIN and AGAIN

Sometimes I think that the truth about this election will be a secret that is only revealed after my death. Maybe one of my kids can engage that crossover guy on the Sci-Fi channel and send a message stating, "Hey, mom! Gore[Bush] really DID win in 2000."

-- Anita (Anita_S3@hotmail.com), November 19, 2000.


Did somebody fart?

-- ? (?@?.?), November 19, 2000.

the argument about a selective recount is bogus because Gore offered to count the whole state.

Actually YOUR argument is bogus because the timeframe for requesting a hand recount has passed in Florida. Or do you think that the best course of action is to ignore the laws that you find inconvenient or troublesome?

-- Uncle Deedah (unkeed@yahoo.com), November 19, 2000.


I have to agree with the following statement by Z from earlier in the thread.

"If you are correct, then I agree. I just don't have enough information. What I have suggests that the error rate for "Opti-scan" is different than punch card [we got rid of them 2 decades ago]. Anyway without this consideration, the presented agrument is nonsense. If you have information that I don't, I concede the argument."

I recall seeing a 6% error rate for punch cards in the newspaper (meaning, I think, 6% of the ballots that people can read were unreadable by the machine, but newspapers seldom give enough explanation to be sure of their meaning). If true, it means the higher recount numbers should swamp the differential between the counts at present.

dandelion

-- dandelion (golden@pleurisy.plant), November 20, 2000.


Ha, I'm number 100!

-- Lars (lars@indy.net), November 20, 2000.

"Actually YOUR argument is bogus because the timeframe for requesting a hand recount has passed in Florida. Or do you think that the best course of action is to ignore the laws that you find inconvenient or troublesome?"

Nonsense. The counties that Gore wanted are being counted at this moment -- because a judge said go ahead. A judge could say go ahead and count the whole state. Even Senator Chuck Hagel, repsected republican and military hero, is suggesting a recount of the whole state. At least there's one republican that isn't afraid of the results. At least there's one republican that doesn't want to suppress the truth.

All the other republicans want to keep the truth away from the voters, suppress the real results.

If you don't like the law and what the judges are ruling in Florida, why don't you move to Communist China Unc. They don't have judges or the rule of law there and information is routinely suppressed. It sounds right up your alley.

-- Commies Suppress Truth (just@like.gop), November 20, 2000.


Flint,

In the recent past, I commended you on your analyses. However, I have never had the time to read all threads in this forum, and I was especially disinterested in following threads which contained the sort of disagreement we are having in this one, except in the very small fraction of cases where the thread's topic was of special interest to me. I always knew I hadn't read a substantial portion of your postings, but did not think that the ones I had missed would've changed my overall opinion of you.

Until this thread, I had not noticed that some of your postings consistently had the characteristics pointed out by Brian McLaughlin. So until a couple of days ago I have been surprised to see you use certain rhetorical tactics in this thread even after I and others have specifically pointed them out. The extent of my surprise may be gauged by my off-target speculation that intoxication might be a case of your abnormal (from my point of view) arguments.

(By the way, another of the misrepresentations of yours about which I have previously not commented was that after I wrote, "Flint, if you're drunk or high, maybe you ought to sober up before posting again in this thread", you wrote "He ... accuses me of being drunk." [The ellipsed part is not relevant to the issue of accusation of drunkenness]. No, Flint, I did not accuse you of being drunk, or call you a drunk. I issued a conditional statement that _IF_ you were drunk, then such-and-such. I did not state that you actually were drunk or accuse you of actually being so. You could have replied to my conditional statement by simply stating that the if-condition was false.)

Now that I have a clearer view of what you've written lately, I withdraw my previous commendation of your analyses.

-- No Spam Please (nos_pam_please@hotmail.com), November 20, 2000.


Flint,

Thank you for clarifying one issue.

In response to my question, "Does that mean that you acknowledge that the arguments you have been presenting in this thread have not been objective on your part, but have been consciously intended as game moves (and I too mean that in the sense of game theory, not as a trivialization)?", you answered, "Yes, of course. I make no pretense of being objective."

Okay. Let's try the matter of honesty:

Do you acknowledge that some of your statements in this thread have been dishonest?

-- No Spam Please (nos_pam_please@hotmail.com), November 20, 2000.


Flint,

Part of my interpretation of Glaeser's article at the top of this thread is that he is discussing "bias", but not "fraud". That is, that part of his argument is that a proper, clean, nonfraudulent partial recount can bias the results.

Because of that interpretation, throughout this thread when I have referred to the merits of a recount and have not qualified that by specifically mentioning fraud, I have meant a nonfraudulent recount.

I consider fraud to be a separate issue from whether nonfraudulent recounts are okay. As you can see in some other recent threads on this forum, I condemn election fraud and support prosecution of those who commit it.

Am I correct in now thinking that you have often or mostly meant to include the category of fraudulent recounts when you have written about "recounts" unqualified by an explicit attribution of fraud? If so, that explains some of our disagreements.

In the future, will you please make it clear whether an argument you present about recounts is intended to apply to fraudulent recounts as well as, or instead of, nonfraudulent recounts? I pledge to do so -- unless otherwise qualified, my references to recounts in this thread will refer only to nonfraudulent recounts.

-- No Spam Please (nos_pam_please@hotmail.com), November 20, 2000.


Flint,

(from your last November 17 posting)

>>Why do those last two quotations show only part of the sentences involved instead of the whole sentences?

>Because I extracted the only portions of the sentences I considered critical. You know, the subject and the verb, that sort of thing.

I think you can see very well that portions you left out are important in showing that the three sentences you partially quoted do _not_ demonstrate that I contradicted myself. This was simply a knowing misrepresentation on your part.

>No Spam's subsequent posts make it clear that he was trying to distinguish between two different "senses" of voter intent, though he does not deign to explain what these senses might be.

The basic intent of a voter is to register his/her choice of votes. I thought that was pretty obvious to everyone, and that this is the intent meant by the parts of Florida law which refer to the "intent of the voter".

The mechanics of how to indicate one's voting choices, such as punching a card, pulling a lever, or filling in an arrow or oval or circle, are a separate matter.

- - -

(from your first November 16 posting)

>This isn't a misrepresentation by any stretch at all. These are YOUR OWN DAMN WORDS!

But you wrote that right after selectively quoting me out of context in such a way as to hide the nonparallelism between the statements I actually wrote. That _was_ misrepresentation -- you showed only a part of my OWN DAMN WORDS while omitting other parts of my OWN DAMN WORDS that were relevant to the meaning of what you atrributed to me. Sheesh, yourself.

>Here we have two ballots, both with a "partial punch", namely a dimple, in the Gore chad. One of these ballots has no other defomations of any kind, the other has the Bush chad punched completely out. But the dimples in the Gore chad are identical.

The interpretation should depend on the official legal definition of what constitutes a valid ballot, i.e., on Florida state law. That has been my stated position in earlier postings on other threads, and I maintain it here.

If Florida law does not sufficiently define that, shame on sloppy Florida election law, and any ballot which cannot be definitely categorized should not be counted either way, but set aside as indeterminant until clear definitions are in place.

>A Republican counts the one with the Bush hole for Bush and counts the other as a "no vote" because there is no hole, and discards it. This is perfectly reasonable.

... only if those decisions are in accord with Florida election law.

>A Democrat discards the first as a "double vote" because there are two "holes", and counts the second for Gore. This is also perfectly reasonable.

... only if those decisions are in accord with Florida election law.

In this particular example, one of the two recounters is making wrong decisions. Which one, depends on the specifics of Florida law.

From what I read lately, Florida law is not sufficiently precise or clear in this regard, so there needs to be a clarification made by court or legislature.

>But we've had just the opposite, where the dimple is in the Bush chad. In this case, counting the dimple as a hole works *against* Gore. Not surprisingly, the Republican and Democrat reverse their definitions in this case. After all, BOTH definitions are equally reasonable.

... only if they are in accordance with Florida law.

>No Spam, I ask you. How would YOU count these two ballots? Does the Gore dimple count as a "partial hole" or doesn't it?

I would count them according to Florida law. If the law is not sufficient to provide a clear means of determining which interpretation is correct, then I would have the appropriate combination of Florida courts and legislature produce a clear definitive description of how to treat the various ballot possibilities ASAP.

>Can you understand how a desire to see a specific outcome might influence the reasonable interpretation of a ballot?

Yes, if the legal guidelines were not precise or complete enough. In that case, I would have the Florida Courts and legislature produce a sufficiently clear and complete set of guidelines for ballot interpretation ASAP.

>Just how partial is partial enough? Do other holes on the card change this partiality? When and how?

Depends on the law, as above.

>Some ballots have dimples in BOTH the Bush and Gore chads, so do we count the larger dimple? Who gets to decide which is larger? Who gets to decide who gets to decide?

Depends on the law, as above.

>I am NOT misrepresenting you by recognizing that this ambiguity exists and must be dealt with.

... and I didn't say you were.

>I'm not impressed with your clear implication that if we pretend it doesn't exist, it will go away.

Yet here in your very next sentence you _do_ misrepresent me, Flint. Habit's hard to break once you got it established, isn't it?

I have never implied that if we pretend it doesn't exist, it will go away.

As I've said, I don't have time to respond to all points in all postings. Please refrain from drawing any implication from my failure to respond to a particular point other than that I might not have yet had time to respond.

- - -

(back to your last November 17 posting)

>[I didn't realize you considered my addressing of that particular topic to be of special importance.]

>But that particular topic is the ONLY possible justification for doing hand counts in the first place.

No, it isn't.

In my sentence, "that particular topic" referred to your earlier sentence, "I give up trying to get you to address the topic of the actual partially punched ballots."

As I've explained elsewhere, there are other things besides partially-punched ballots that could cause a machine count to differ from a manual recount. Misalignment of the printing, misalignment of the ballot frames, and so on.

Please don't cite a reference to a particular topic, then criticize me while pretending that the reference was to a broader topic. That isn't fair, and you know it.

>Glaeser's point, with which you appear to agree, is that manual counts can produce different results.

Glaeser makes several points. None of them refer to particular detailed causes, such as partially-punched holes, of the differences between manual and machine counts.

>That difference might be critical. HOW that difference comes to pass is therefore just as critical. Am I seriously intended to believe you don't realize this?

Earlier in this thread I was interested in discussing Glaeser's points, which did not include details of the causes of differences between machine counts and manual recounts. Please stop concluding that I don't realize something just because I haven't commented on it.

>Well, I keep trying to explain why nonstatistical aspects are irrelevant

Yet _you_ insist that I respond to your introduction of the nonstatistical topic of how partially-punched ballots are to be judged!

Do you think that how partially-punched ballots are to be judged is a statistical aspect? If so, how do you justify that categorization? If not, then why do you claim to be trying to explain why nonstatistical aspects are irrelevant while you're plainly arguing that particular nonstatistical aspects are relevant?

>"correctness in Gore's favor" is Orwellian.

Do you consider correctness in Bush's favor to be Orwellian?

>NEITHER SIDE IS INTERESTED IN ACCURACY! BOTH SIDES ARE INTERESTED ONLY IN WINNING! Good Grief, man! Wake up!

I'm interested in accuracy.

>Let's say that Candidate A IN FACT got 300 more votes than candidate B, *provided* that EVERY ballot were properly counted. By definition, candidate A should win the election. NOW, let's say that due to counting errors, we have missed exactly 1000 votes for EACH candidate. Candidate A still has 300 more than candidate B, but both of their totals are 1000 votes short of fully accurate.

Okay so far, as long as we note that this example assumes that the "exact" counts are known in advance of the completion of all accurate counting, whereas in a real election the "exact" counts are _not_ known in advance of the completion of all accurate counting.

>NOW, let's say that through a carefully aimed partial recount we add in Candidate B's 1000 votes, while adding none to candidate A. Yes, in your sense the totals are now "more accurate", because candidate B's total is now precisely accurate whereas before it was not.

... presuming that the 1000 votes picked up for candidate B were the same 1000 votes that were previously specified to have been missed due to counting error.

If so, then at this point, candidate A's total is just as accurate as it was before, candidate B's total is more accurate than it was before, and the total of candidate A's and candidate B's votes is more accurate than it was before.

BUT because of the difference in the basis of measurement between A's votes (not all of which known to have been missed have been counted yet) and B's votes (all votes have been counted; no more are missing), we cannot validly compare A's total to B's total to determine who won the election, in this example.

However, in a real election, we would not know that there were 1000 of A's votes that had not yet been counted.

>However, now the situation is that THE WRONG CANDIDATE WON THE ELECTION!

But we can draw this conclusion only in this example where we have perfect knowledge that's not available in a real world election. In a real election, we would not know that there are 1000 votes for A that have been missed.

>If you favor candidate A, this "correction" was a "misinterpretation" because it only told part of the story and made key omissions.

No, the "correction" was not a "misinterpretation". The correction did properly represent an improvement in accuracy of B's vote total and of the combined vote total of A and B.

>If you favor candidate B, then this was an improvement because the counts were "more accurate" and this is "considered desirable". Right?

Wrong.

The correction was properly represented as an improvement in accuracy of B's vote total and of the combined vote total of A and B. In a real-world election, we would not yet know that it did not represent an improvement in accuracy of A's vote, because we would not yet know that there were still 1000 votes for A that had been missed and not yet counted.

What you've done here is perpetuate your failure to properly apply principles of error analysis to the situation.

>How very sensitive you are to omissions (real or imagined) that work against you, and how very hard you work to deny they make any difference when it works in your favor.

But as we've gone through the details, we've found that that assertion of yours about me is false.

>Isn't that amazing?

No. It's a false statement.

If you disagree, please cite specific examples that support your assertion.

- - -

In response to an earlier statement of yours:

>You need to entertain the idea that some people might favor Bush and still have valid points.

I responded:

>Do you have any evidence that I don't? If not, why did you make this statement? If you do, what is it?

and you responded to that with:

>Only indirectly. Many people have written many things in support of both candidates. You have agreed with pro-Gore points and disagreed with pro-Bush points without exception.

That last sentence is not true. I have agreed with some pro-Bush points and disagreed with some pro-Gore points.

Don't say "without exception" unless you really have examined the whole forum and found no exception. I know you haven't done that in my case.

Furthermore, even if your statement were true, it still would not demonstrate that I had not entertained the idea that some people might favor Bush and still have valid points.

>ONE of us is in error here, and we remain equally convinced it's the other guy. To me, the key difference is I *illustrate* why you're wrong, and you are content to *assert* that I'm wrong.

No. I illustrated why your original example was wrong (failure to consider the systematic error), and earlier in this posting I illustrate why your second example was wrong.

>[... just as Bush would expect to Gain VOTES if there were hand recounts in some other Florida counties.]

>Yes, certainly this is true. So you are now admitting that accuracy is irrelevant, and these recounts are to change ratios?

No, I was pointing out the parallelism between Bush and Gore in a particular respect. My comment did not say accuracy is irrelevant.

>I love what Peggy Noonan wrote:

< snip >

>Of course, this addresses the key issue here,

It address fraud. As I explain above, that is a separate issue from nonfraudulent recounts.

>so important that it just never occurred to you that I might want you to face it simply because I asked you to.

After all the stuff I've asked you to face that you haven't, you have no standing to hurl that accusation.

Are you willing to discuss the merits of recounts without involving fraud?

>[How did you determine that I pretended you asked no such question?]

>By neither answering it nor referencing it. Maybe I should have said you ignored it?

No, all you knew was that I neither answered it nor referenced it. You did not have sufficient grounds to conclude that I was either ignoring it or pretending it wasn't there.

As I have explained, I don't have time to respond to everything. So some points simply are too far down on my priority list to make it into my postings. That involves neither ignoring nor pretending.

>Do you see a substantive difference between ignoring something and pretending it's not there?

Yes. Do you see a substantive difference bwteeen not having time to respond to something and (either ignoring it or pretending it's not there)?

>[Does that mean that you would then be arguing that partial recounts were okay, when following established legal procedures?]

>This is a damn good question, and I don't know the answer. I suspect that our relative positions would indeed be reversed in this discussion.

Wrong. I'd argue for what I think is right either way.

Recounts are a perfectly legal part of election procedures in every state, as far as I know. To argue that recounts should be blocked is wrong.

>It seems to me that both of us are making as strong a case as we can for what we see as the good and the right, namely our preferred candidate

I argue for points favorable to Gore only if it's right to do so. I don't argue for Gore by taking standa that are wrong.

Allowing recounts in accordance with the law is right. Trying to squelch them is wrong.

>And there's yet another difference between saying 1+1=3, and then later claiming that the 3 was really a 2 after all when caught out.

But I haven't done the latter, to the best of my knowledge. You seem to have done so with regard to the matter of attributing a false conclusion to increased accuracy rather than to a failure to account for systematic error.

>Maybe I don't WANT to understand your points.

Well, that explains a lot.

>Maybe you aren't expressing them very well.

Then why don't you request clarification instead of misrepresenting me?

>Maybe you make hazy arguments that can be interpreted several ways,

Then why don't you request clarification instead of misrepresenting me?

>so that no matter which interpretation I pick, you can claim you meant another.

If you asked which interpretation was correct instead of just attributing an unfavorable one to me, you'd look a lot more honest in that regard.

>[ I am being honest, and have been all the way through.]

>I believe you HAVE BEEN using this technique,

If you are referring to a technique of making hazy arguments that can be interpreted several ways, then why haven't you applied the obvious countermeasure of requesting a clarification instead of just attributing an unfavorable interpretation to me? If you had been doing this, your current accusation about my "technique" would look more honest. As it is, it looks very much as though you deliberately choose not to use a countermeasure that would come naturally to an honest man.

>and your denial is disingenuous.

My denial is honest.

>[But _you wrote in your example_ that "our increased accuracy led us to the wrong conclusion", NOT that your comparison of apples to oranges led to the wrong conclusion!!! It's right up there in plain text for everyone to see]

>I'm not sure whether you are picking nits here, or drawing distinctions without differences, or bantering with semantics, or really making a valid point. I'll assume you are making a point.

I was.

>In which case, you are correct. I intended strong sarcasm. Of course it wasn't accuracy per se that caused error. Perhaps I should have said that our originally correct conclusion was reversed *despite* greater accuracy?

That would be better, except that the correct answer is that it was not valid to reverse our original conclusion, because the two sets of measurements were not comparable without taking the systematic error into account.

You seem still to have missed that point.

You have used that erroneous logic to argue against a partial recount.

>In any case, the point (with which you seem to agree) is that increased accuracy, selectively applied, does not necessarily improve the final result, and can make it worse.

No, I do not agree that that is the point.

>For another example, see my "1000 votes" example earlier.

... the one that contained further errors.

>[You attributed the error in conclusion to the wrong source!]

>Yes I did. I admit I should have made some indication like HEAVY SARCASM--->>>. To make sure nobody misunderstood, I quickly clarified (see above) that it wasn't the increased accuracy that caused the error, but rather its selective application.

No, that's not correct. Why do you keep avoiding the term "systematic error"?

>However, the point that selective increases in accuracy can bias the overall result remains true.

No, it doesn't.

>What you mean is, if someone demonstrates *to your satisfaction* that you are wrong, you will change your mind.

Of course, as is true for you and everybody else, too.

>I'm doing all I can correct what I regard as your errors.

You might be more effective if instead of attributing unfavorable interpretations to another's words, you first asked that person for clarification.

-- No Spam Please (nos_pam_please@hotmail.com), November 20, 2000.


Commie, (apt name you chose for yourself, BTW)

I said that the time frame for a hand recount has passed, to which you replied "Nonsense". Sorry, but you are wrong, the time to ask for a hand recount HAS INDEED passed in Florida under Florida LAW.

The counties that Gore wanted are being counted at this moment -- because a judge said go ahead. A judge could say go ahead and count the whole state.

Yes the judge said go ahead, BECAUSE the request for a recount in those counties was filed BEFORE the deadline passed. The judge was following the LAW. Could a judge say hand recount the entire state? Sure, if he decided to IGNORE the law as it now stands he could demand one.

If you don't like the law and what the judges are ruling in Florida, why don't you move to Communist China Unc. They don't have judges or the rule of law there and information is routinely suppressed. It sounds right up your alley.

Really? I should move to China? I am the one saying that the LAW should be followed, why don't YOU move there, since YOU are the one who thinks the LAW should be ignored?

-- Uncle Deedah (unkeed@yahoo.com), November 20, 2000.


No Spam:

First you write "Please sober up." NOW you write "I did not accuse you of being drunk." Now, this is splitting hairs mighty fine. If I ask you to pull your head out of your ass, I'm not actually *accusing* you of having your head up your ass, oh no! I'm using the implied subjunctive mood, right? This distinction would embarrass a crooked lawyer, No Spam. You told me to sober up. That's an accusation in any context.

[Now that I have a clearer view of what you've written lately, I withdraw my previous commendation of your analyses.]

But not without cost, because you commended analyses that agreed with your own, and you only have problems with analyses with which you disagree. You are going to have to change quite a few of your opinions by doing this, all of which are probably correct to begin with [grin].

[Do you acknowledge that some of your statements in this thread have been dishonest?]

Opinionated yes. Dishonest, well, I submit NONE of my statements have been as dishonest as telling you to sober up and then turning around and denying I accused you of being drunk. Think about this.

Now as for the difference between fraudulent recounts and bias, I addressed this specifically a couple of times. There are several issues involved here. Glaeser's involves bias without direct fraud, and this should be what we're focusing on. Glaeser correctly argues that perfectly accurate recounts, if selectively applied, can be used to change the winner of the election. Since ALL we care about is the winner, and we do NOT care about the accuracy of the vote counts but only about the *relative* accuracy of those counts, changing the relative accuracy (by making one side more accurate and not the other) can, and is *intended* to, result in selecting *either* candidate as the winner, depending on which results we choose to make more accurate. Glaeser argues, and I have illustrated several times, that this is a valid principle and is being used to do exactly as specified -- change the winner despite the underlying data through selective improvement (without applying that correction factor to the population of results generally).

Now, all of the above assumes that hand recounts are more accurate than machine counts. This is very problematical, as we have seen. The ballots the machine was unable to count tend to be ambiguous in some way, and the humans who are interpreting these ballots tend to differ in their interpretation *along party lines*. This in itself makes the hand count process suspect. However, this is an entirely separate issue and should not be confused with the principle of apples and oranges we discussed above.

[The basic intent of a voter is to register his/her choice of votes. I thought that was pretty obvious to everyone, and that this is the intent meant by the parts of Florida law which refer to the "intent of the voter". The mechanics of how to indicate one's voting choices, such as punching a card, pulling a lever, or filling in an arrow or oval or circle, are a separate matter.]

I must strongly disagree with this. Indeed, I find your distinction mind-boggling. The person's intent is to vote and he votes by punching a hole. Therefore the intent to punch a hole and the intent to vote are logistically identical. Punching a hole IS voting, as implemented in these counties. This is an identity. I insist that you are sowing confusion deliberately by creating a distinction that IN PRACTICE does not exist, and using that confusion to distract from the principle under examination. And I don't consider that honest. Here you are talking about "nonparallelism" where there is no effective difference.

[From what I read lately, Florida law is not sufficiently precise or clear in this regard, so there needs to be a clarification made by court or legislature.]

This is all very well, but perhaps disingenuous. No law can anticipate all future contingencies; these must be ironed out case by case. Nor can any legal system separately consider every individual chad (there being no two precisely the same, like snowflakes). So we need principles on which decisions should be based, ballot by ballot. And we saw these principles changing as time went on. The sunlight test didn't handle dimples. The dimple test didn't handle cases where another hole was completely punched. And so on.

Now, here you are saying "I would have the appropriate combination of Florida courts and legislature produce a clear definitive description..." This is fine, all I wanted to demonstrate was that there *really are* ambiguities, and these *really are* being resolved differently depending on who is doing the counting. The overall context concerned "accuracy" and what that might be. You have dodged this, dumping it back on the courts and the legislature. Well, which party controls the legislature? Assuming the legislature took it upon itself to arbitrate every ambiguous ballot, the end results (vote counts) is STILL a political hot potato, and NOT a matter of some ideal measurement accuracy.

You just keep repeating "depends on the law" without recognizing that the law is simply incapable of defining these things to the required level of detail. Ultimately, *whatever* the law is, these counts come down to a matter of personal judgment, ballot by ballot. People are partisan.

Now, back to honesty. Do you claim really NOT to understand the situation I was asking about? Saying "depends on the law" over and over is an intellectual cop-out. OK, NOW you are the legislature, damn it. You are MAKING the law. Exactly what should it say, that I cannot find cases the law does not properly address? How can I make the point that we are ultimately obliged to use individual partisan interpretations, without you dodging this point and repeating that these decision should belong to *somebody else*. No Spam, it doesn't matter which *somebody else* ends up making these calls. People are partisan, whether they are you, me, or "somebody else".

[>Glaeser's point, with which you appear to agree, is that manual counts can produce different results.

Glaeser makes several points. None of them refer to particular detailed causes, such as partially-punched holes, of the differences between manual and machine counts.]

Now what am I to make of this? The essence of Glaeser's article was that selective improvements in nominal accuracy could change the winner. Now, this principle applies regardless of the nature of the data (except that it is initially inaccurate in some way), regardless of the details of the methods. Without a whole lot of effort, we could likely come up with a thousand scenarios, all completely different in detail, illustrating the same basic principle. I interpret your objection as saying, since Glaeser didn't talk about holes, his principle doesn't apply. Did I misunderstand?

[Yet _you_ insist that I respond to your introduction of the nonstatistical topic of how partially-punched ballots are to be judged!]

Guilty as charged. I should never have introduced this. It is indeed a separate topic having nothing to do with the principle Glaeser was presenting.

[Do you consider correctness in Bush's favor to be Orwellian?]

Yes, absolutely. While a theoretical correct count probably favors someone (assuming we don't have an exact tie), it is Orwellian to imply that "accuracy" is unbiased, and at the same time try to weild "accuracy" so as to assure a desired winner. The goal here, to all but the most uninformed and naive, is NOT to "increase accuracy", it is to CHANGE THE WINNER. This is a perfectly proper political goal. But claiming that the goal is "really" greater accuracy is not honest.

>>>snip most of an interesting discussion, cutting to the chase here...<<<

[But we can draw this conclusion only in this example where we have perfect knowledge that's not available in a real world election. In a real election, we would not know that there are 1000 votes for A that have been missed.]

This is true. The illustration was intended to show how this principle worked assuming perfect knowledge, i.e. to demonstrate the validity of the principle. In real life, we simply do not have this knowledge. We must work according to the odds, based on what we do know. First, we do know that the large majority of ballots where no presidential candidate was selected according to the machine, probably had an intended selection. We know this from experience -- people tend to vote more for president than any other office on the ballot. Second, we do know from unambiguous ballots the approximate distribution of votes within a given county. Third, we know that within that county, ambiguous ballots *properly* interpreted are likely to favor whichever candidate received the majority of unambiguous votes within that county, and by approximately the same ratio.

Now, armed with all these probabilities, we can make a pretty damn good estimate as to where hand recounts will help one candidate more than the other. True, it is *possible* we're wrong, just as it is possible to flip heads 100 times in a row. So you'd be a fool to bet against it. And this is exactly the reasoning behind the selective Gore recounts. As I've said before, everyone expects Gore to pick up a LOT of votes this way. They aren't talking about the number of votes he picks up, you realize. Nobody else cares about the number. When they say "pick up a vote", they mean change the DELTA by one vote. If BOTH candidates are found to have added 100 votes, then neither side "picked up" ANY votes. Only the winner counts, the accuracy of the totals is irrelevant. The accuracy of the *difference* is all that matters.

[No. It's a false statement.]

No, I was teasing you. Sorry.

[If you disagree, please cite specific examples that support your assertion.]

OK, I think I have done that. Your objection to my theoretical example was that in the real world, we lack this knowledge. To me, there are two issues here. First, is the principle valid? In an example where we had full knowledge, I think I demonstrated that yes, the principle is valid. Second, can the principle be used to bias the outcome in a desired direction in a situation where we lack full knowledge but we have some damn fine, reliable probabilities to work with? And I think I showed that yes, indeed we can. And are.

And then there's the issue of whether the *process* of hand counting can't just help these probabilities along a little. This is a different issue, I admit, but observation more and more shows that there is a LOT that partisan counters can do to, uh, goose the odds a bit.

[Don't say "without exception" unless you really have examined the whole forum and found no exception. I know you haven't done that in my case. Furthermore, even if your statement were true, it still would not demonstrate that I had not entertained the idea that some people might favor Bush and still have valid points.]

OK, granted. I personally haven't found any exceptions, but I haven't read every thread either. And I agree that I can't read your mind. I can only deduce what might be going on inside it by what I see coming out of it. Maybe I should qualify everything I say with "As far as I know" and "to the best of my knowledge" and "I can't recall". Gawd, I sound like a politician!

[>This is a damn good question, and I don't know the answer. I suspect that our relative positions would indeed be reversed in this discussion.

Wrong. I'd argue for what I think is right either way.]

But that's exactly what I said!

[Recounts are a perfectly legal part of election procedures in every state, as far as I know. To argue that recounts should be blocked is wrong.]

No, to the best of my knowledge I've never argued that they should be blocked. I've argued that they should NOT be carefully tailored to help one side. I think Bush made a mistake turning down a hand recount of the entire state, if only because the winner of that process would have more legitimacy, whoever it might end up being. But I can understand Bush gambling that he could preserve his narrow lead. I just think he underestimated Gore's "win at any cost" scorched earth determination.

[You seem to have done so with regard to the matter of attributing a false conclusion to increased accuracy rather than to a failure to account for systematic error.]

I admitted poor phrasing. I admit it again. It was NOT increased accuracy that led to the wrong conclusion. I was trying to illustrate that increasing the accuracy of measurement did NOT guarantee better results, and could make them worse. Results being a winner, a binary choice. Accuracy must be increased *evenly* across the data. Applying increased accuracy unevenly so as to create a desired predetermined result is an error. I say so, Glaeser says so, and I think you say so as well. Exactly how we express this differs, but the principle holds.

[>In any case, the point (with which you seem to agree) is that increased accuracy, selectively applied, does not necessarily improve the final result, and can make it worse.

No, I do not agree that that is the point.]

Sigh. Well, that's my point, it's eve's point, it's Glaeser's point, and (most critical of all) it's *Gore's* point.

[>However, the point that selective increases in accuracy can bias the overall result remains true.

No, it doesn't.]

Groan. Apparently your professor of freshman physics was a real stickler for terminology. Selective increases in accuracy CAN bias the total. This is because such selective increases introduce systematic error. OK? These increases in accuracy are not being properly uniformly applied. They permit us to select the wrong candidate as the winner. That is exactly *why* they are being incorrectly applied. And it works, if we don't do anything to correct this procedural error. "Works" in the sense of producing the desired results despite the underlying data.

Unk:

I think you are being unreasonably strict about this. The law means what the courts say it means. This is a long and time-honored principle. The courts might decide that the deadline for demanding a recount was set up to prevent nuisance demands. But in this case, such a demand isn't being submitted just to cause trouble, it is part of a single, ongoing statewide controversy that got started within the proper time frame, and is simply expanding in scope. So the courts try to interpret the *intent* of the law, not merely the narrowest interpretation the words permit.



-- Flint (flintc@mindspring.com), November 20, 2000.


No Spam,

So, how do you like our man Flint, so far?

He will admit to sloppiness. He will admit to incapacity. He will admit to imprecision. He will admit to not wanting to understand your points. The one thing he will never admit is that any of these affect the quality or accuracy of his conclusions by one iota.

Flint is an interesting material. It can produce a cutting edge finer than steel. But try as you might, flint cannot bend even slightly.

-- Brian McLaughlin (brianm@ims.com), November 20, 2000.


Hi Brian,

Still avoiding the subject under discussion? Not surprising.

I like your selective reading here. No Spam says "IF you are drunk, sober up", claims he never said I was not sober, and uses that IF as the justification. And this is fine, No Spam isn't distorting his meaning, right? The notion that I might not be sober would never have crossed anyone's mind, right?

I wrote "Maybe I don't want to understand" and you pop right up and claim I am *admitting* I don't want to understand. But you are NOT distorting my meaning, oh no! Let's omit the context, let's ignore the "maybe", let's interpret an intent that wasn't there. Uh, who was it you said did these things again?

I suggest that rather than keep applying the same double standard, you come up for air and examine the topic we're discussing. It's ever so much more interesting than "Why Brian is having a hissy" to everyone else. Trust me (giggle).

-- Flint (flintc@mindspring.com), November 20, 2000.


Flint,

Although I can't support your position in all its detail, as I haven't been able to keep up with this debate, a pretty good sign that you've scored some points is when they draw their beads on YOU, drawing attention away from your ideas. So my guess is that you're probably doing a very good job so far.

Just hang tough and stay the course, good man.

Oh yes...and watch your back.:)

-- eve (eve_rebekah@yahoo.com), November 20, 2000.


Flint,

ME? Unreasonably strict? Good heavens!

I understand that courts interpret the law and make rulings on the law. These decisions are then referred to as "case law". BUT, until a court rules otherwise, the law that defines the deadline for requesting a hand recount stands as the applicable law. So I cannot agree that I am being unreasonably strict, I am merely stating what can or cannot occur under the laws as they stand now.

-- Uncle Deedah (unkeed@yahoo.com), November 20, 2000.


eve:

No Spam has made some good points, I think. His analysis of the cause of error is excellent and quite rigorous, and worth thinking hard about. What I've been having trouble with is, the error itself remains just as real whether or not I have pinned the precise cause. My interpretation is that No Spam has been taking the position that because I have misanalyzed the cause of the error, therefore there is no error. Otherwise, why doesn't he say something like "Gore's selective recount is indeed likely to introduce systematic error in Gore's favor, but you haven't understood exactly what that error really is" or words to that effect?

I feel as though the patient is just as dead no matter how poorly I may have explained or understood the cause of death. No Spam seems to me to be confusing HOW Gore is cheating with WHETHER Gore is cheating according to Glaeser's principle (NOT according to the law as written, which clearly permits such cheating by either side).

Brian has started to remind me of Our Pal Ray last year, cheerleading from the sidelines by saying "Yeah, what he said. Hit him again!" He has yet to address the subject under discussion, preferring to wander up to the front of the stage and exhort the audience to boo his designated villain. An amusing tactic, but without intellectual pretensions.

Unk:

I think your point is weak, that's all. If only recounts that favor Gore are permitted because of this deadline, this seems quite opposed to the intent of the deadline itself. Selective recounts can indeed help the trailing party. In effect, you are recommending the legal technicality of waiting until the last minute to file for a partial recount, forcing it to be one-sided in your favor from sheer timing. I don't think this ruse will pass muster in any court.

I think an intelligent court would rule that recounts cannot be "structured" so as to preferentially benefit one side. It's possible that the law will be tossed out as it currently reads. But nobody is going to allow the US Presidency to hinge on such a technicality.

-- Flint (flintc@mindspring.com), November 20, 2000.


http://dailynews.yahoo.com/h/wcvb/20001120/lo/253821_1.html

LINK

Monday November 20 06:00 PM EST

Mass. Company Tests Punch-Card Ballots

Tests Shows Hand Counts More Reliable

A Needham company on Monday tested the accuracy of punch-card ballots, similar to those used in Florida.

NewsCenter 5's Jim Boyd reports that AmeriCounsel.com staged a mock election in Copley Square on Sunday, and the results were tabulated Monday. The election asked voters for their favorite ice cream flavor and season of the year.

The ballots were then counted by machine and then recounted by hand. The machines rejected 10 percent of the ballots, but hand counters were able to count a third of those, after determining that they were valid votes.

"It's clear that by doing a hand recount, people were able to pick up votes that were missed by the machine," Jonathan Slater of AmeriCounsel.com said.

Just to be sure, the counters recounted their recount and came up with the same results. Based on this experiment, representatives from AmeriCounsel said that a similar recount in Florida could put back as many as one third of the votes that were thrown out, enough to affect the results of the election.

-- (something@to.consider), November 20, 2000.


Flint,

It is not the fault of the law that Bush did not file for hand recounts, he had the same time to do so that the Gore team did. The law does not favor either side, UNLESS, as you said, recounts always favor the losing side, since they have nothing to lose by recounting.

If it IS true that recounts favor the losing side, then your statement that "an intelligent court would rule that recounts cannot be 'structured' so as to preferentially benefit one side" would mean that NO recounts should be conducted, since you have yourself said that recounts can only end up favoring the losing side.

-- Uncle Deedah (unkeed@yahoo.com), November 20, 2000.


Or are you are saying that a recount must ALWAYS include the entire state?

-- Uncle Deedah (unkeed@yahoo.com), November 20, 2000.

WOW! I feel like I've been through the looking glass after reading this thread! Check this out:

"The correction was properly represented as an improvement in accuracy of B's vote total and of the combined vote total of A and B. In a real-world election, we would not yet know that it did not represent an improvement in accuracy of A's vote, because we would not yet know that there were still 1000 votes for A that had been missed and not yet counted.

What you've done here is perpetuate your failure to properly apply principles of error analysis to the situation."

If the principles of error analysis say that, then the principles of error analysis is a ass.

Are you saying that, in the absence of perfect knowledge of the election result, any partial recount, no matter how selectively the part is chosen, increases the "accuracy" of the total result? This is so obviously, logically wrong that I'm at a loss as to how to respond to it.

I'll just say this - we may not have perfect knowledge of the Florida results, BUT we do know that:

1) Palm Beach County voted roughly 2 to 1 for Gore.

2) The problems with punch card voting should occur randomly - anything else would indicate the possibility of fraud.

3) Therefore, the new votes added by the hand recount process should also break roughly 2 to 1 for Gore.

4) Double therefore, the addition of hand count results from PBC (as well as the other Democratic counties) biases the total result in Gore's favor.

This is in essence what Glaeser's saying, and I fail to see how it's debatable.

-- RC (randyxpher@aol.com), November 20, 2000.


Flint,

>First you write "Please sober up." NOW you write "I did not accuse you of being drunk." Now, this is splitting hairs mighty fine. If I ask you to pull your head out of your ass, I'm not actually *accusing* you of having your head up your ass, oh no! I'm using the implied subjunctive mood, right? This distinction would embarrass a crooked lawyer, No Spam. You told me to sober up. That's an accusation in any context.

Two sentences before my "Please sober up" sentence, I stated "if you're drunk or high, maybe you ought to sober up before posting again in this thread." (as we've already discussed). My "Please sober up" sentence was intended as a continuation of the same theme: that _if_ you were intoxicated, _then_ sober up.

In informal conversation, people often abbreviate the second and subsequent statements of a set of parallel statements, because they expect the parallelism to be understood. They omit some parts that would be repetitious, for efficiency in communication.

I sincerely apologize for not realizing that I needed to repeat the same if-then structure to make the parallelism with my preceding sentence clear. Were I to do that part over again, instead of the short "Please sober up" sentence I would have written a copy of my "if you're drunk or high, maybe you ought to sober up before posting again in this thread."

Elsewhere in this thread, I have made such exact copies of statements in response to two different statements of yours that were related in topic. In fact, I did so earlier in the very same posting where I ended with "Please sober up". If you'll look at the first part of that posting, you'll find that after I first wrote, "Why don't you ever acknowledge ... manchine [sic] counting, Flint?", I copied that sentence as my response to the next item I quoted, deliberately to make a point (though my misspelling of "machine" was unintentional). Had I done the same at the end, we probably wouldn't be having this sub-discussion.

Furthermore, I think now that even if I had written the abbreviated copy "Maybe you ought to sober up" instead of "Please sober up" as my second sentence of the intended parallel set, the parallelism probably would have been clearer. My non-parallel wording obscured my intent of parallelism.

I never intended "Please sober up" to be an accusation of drunkenness. Aside from the consideration that there are non-drunkenness-related meanings of "sober", I intended it to mean a parallel expression of my if-then previous statement, and I regret that I did not make that intent clear at that time.

NSP>[Now that I have a clearer view of what you've written lately, I withdraw my previous commendation of your analyses.]

F>you commended analyses that agreed with your own, and you only have problems with analyses with which you disagree.

Not true. I commended your analyses in general, not particular cases. I have read analyses of yours with which I at least partially disagreed but thought were sound nevertheless.

>Dishonest, well, I submit NONE of my statements have been as dishonest as telling you to sober up and then turning around and denying I accused you of being drunk. Think about this.

1) See my apology above.

2) Please give us a direct answer to my question, "Do you acknowledge that some of your statements in this thread have been dishonest?"

>Since ALL we care about is the winner, and we do NOT care about the accuracy of the vote counts but only about the *relative* accuracy of those counts,

You are speaking for yourself, but not for me.

I care about the accuracy of the vote counts.

It is important that the next President be widely thought to have a legitimate basis for winning the election. Because of the closeness of the vote nationwide and in Florida, it is probable that some people will never consider whoever wins to have a legitimate power base, but I think we can try to do all we can to minimize the fraction who so consider.

But I also care about the law.

There is a conflict between the desire to have an ideal vote count and the practical considerations hindering such, not to mention legislative oversights not realized until circumstances bring them to prominence.

E.g., recently I learned from a Florida resident, whom I consider reliable about this matter, that the part of Florida law that sets a 10-day time limit on certification of election results was designed by the legislature to provide adequate time between declaration of results of September primary elections and October runoff elections for preparing and campaigning for the October runoffs, and similarly for the period between the October runoff election and the November general election. _However_, the law was written simply so as to require the 10-day time limit for the September, October, _and_ November elections, with no distinction made for a different schedule following the November election. Yes, there's a need to determine presidential electors by December 18, but that's longer away from the November election than the time between the September and October elections or between the October and November elections, and there's no corresponding need to allow time for campaigning and preparation for a following election, so there was no deliberate legislative intent, according to my source, to restrict the time for November election certification as strictly as for September and October elections. But that's the way the law's written anyway. My source claims that Florida has never before failed to complete a manual recount (which they _have_ conducted before) within the 10-day time limit, but did indicate that there may have been some sloppiness in previous manual recounts that current intense scrutiny may have changed.

So I want the most accurate vote count, in total, possible within the state's legal framework (which includes not only existing law, but potential court and legislative actions to clarify unclear matters).

It is important to work within the existing legal framework because that is what the citizens of Florida have set up to handle their partisan disputes.

NSP>[The basic intent of a voter is to register his/her choice of votes. I thought that was pretty obvious to everyone, and that this is the intent meant by the parts of Florida law which refer to the "intent of the voter". The mechanics of how to indicate one's voting choices, such as punching a card, pulling a lever, or filling in an arrow or oval or circle, are a separate matter.]

F>I must strongly disagree with this. Indeed, I find your distinction mind-boggling. The person's intent is to vote and he votes by punching a hole. Therefore the intent to punch a hole and the intent to vote are logistically identical. Punching a hole IS voting, as implemented in these counties. This is an identity. I insist that you are sowing confusion deliberately by creating a distinction that IN PRACTICE does not exist, and using that confusion to distract from the principle under examination. And I don't consider that honest.

Flint, I am being honest, regardless of whether you like it or agree.

NSP>[From what I read lately, Florida law is not sufficiently precise or clear in this regard, so there needs to be a clarification made by court or legislature.]

F>This is all very well, but perhaps disingenuous.

No, I'm not being disingenuous. I mean what I've said. I'll be glad to clarify what's unclear, when pointed out.

>So we need principles on which decisions should be based, ballot by ballot. And we saw these principles changing as time went on.

My view is that Florida needs as soon as possible to establish standards for those decisions, and that recounts not conducted according to the eventual standards should be redone if there is any possibility that the noncompliance to the eventual standards could have affected the results.

> You have dodged this, dumping it back on the courts and the legislature.

I didn't dodge anything. I stated my position straightforwardly and honestly.

If you don't like my opinion about having the courts and legislature resolve some things, that's one thing. But you didn't have to misrepresent me in order to register your disagreement.

>You just keep repeating "depends on the law" without recognizing that the law is simply incapable of defining these things to the required level of detail.

How did you determine that I didn't recognize that?

My honest opinion is that certain aspects of the election, the aspects under discussion when I wrote that they should be decided by the law, or, more generally, by the legal process, should indeed be so decided. To have that opinion is not to maintain that the law can define every possible detail. It doesn't need to. It just needs to define enough to allow the election result to have a sound legal basis.

>Now, back to honesty. Do you claim really NOT to understand the situation I was asking about?

I've lost the thread of exactly what you're referring to by "the situation I was asking about". If you will delineate it, I will answer your question.

>OK, NOW you are the legislature, damn it. You are MAKING the law. Exactly what should it say, that I cannot find cases the law does not properly address?

Since I am neither a resident of Florida nor do I claim to be aware of many details of Florida state law, I cannot say what I would do if I were the Florida legislature, except as general principles appplicable to all state legislatures to the best of my knowledge.

My point was that I trusted the Florida state legal structure to make an appropriate resolution. I know I haven't written that explicitly, so here it is: I trust the Florida state legal structure to make an appropriate resolution.

>How can I make the point that we are ultimately obliged to use individual partisan interpretations,

You've made it. I disagree with certain accompanying ideas.

>without you dodging this point

Where have I dodged that point? Please cite an example.

>and repeating that these decision should belong to *somebody else*. No Spam, it doesn't matter which *somebody else* ends up making these calls. People are partisan, whether they are you, me, or "somebody else".

A word search finds that I have not used the phrase "somebody else", or even the word "somebody", in this thread.

Going by your context, my response is that the decision belongs to the people of Florida, as expressed in their courts, legislature, and other governmental bodies according to the state legal structure. I think that could've been reasonably expected to be implied by my previous statements.

Do you think that non-Florida residents/citizens/inhabitants should have a particular say in Florida state affairs, or leave those to "somebody else" in Florida?

BTW, do you care to express an opinion here of Bush's initial move to have a federal court override Florida state law? (Perhaps you've previously done so -- I haven't checked.)

NSP>Glaeser makes several points. None of them refer to particular detailed causes, such as partially-punched holes, of the differences between manual and machine counts.]

F>Now what am I to make of this? The essence of Glaeser's article was that selective improvements in nominal accuracy could change the winner.

But he did not mention detailed causes on the level of partially-punched holes for instance.

>Now, this principle applies regardless of the nature of the data (except that it is initially inaccurate in some way), regardless of the details of the methods. Without a whole lot of effort, we could likely come up with a thousand scenarios, all completely different in detail, illustrating the same basic principle. I interpret your objection as saying, since Glaeser didn't talk about holes, his principle doesn't apply. Did I misunderstand?

Yes.

But I'll drop my objection, because it does not seem important enough to continue discussing.

F> >>>snip most of an interesting discussion, cutting to the chase here...<<<

You're conceding my points in the section so snipped?

>Second, we do know from unambiguous ballots the approximate distribution of votes within a given county.

... in the absence of some systematic error, that is. E.g., perhaps shaky-handed citizens overwhelmingly intended to vote for Bush in a county otherwise favoring Gore, but had systematically more difficulty in punching their ballots properly.

>Third, we know that within that county, ambiguous ballots *properly* interpreted are likely to favor whichever candidate received the majority of unambiguous votes within that county, and by approximately the same ratio.

(*sigh*) Once again, no consideration of systematic error, eh?

Note that this third point depends on the second point, which I've shown to be potentially subject to systematic error invalidating the conclusion.

>Now, armed with all these probabilities, we can make a pretty damn good estimate as to where hand recounts will help one candidate more than the other.

... in the absence of systematic error invalidating the preceding conclusions about probabilities. (*sigh*)

NSP>[Don't say "without exception" unless you really have examined the whole forum and found no exception. < snip >

F> < snip > Maybe I should qualify everything I say with "As far as I know" and "to the best of my knowledge" and "I can't recall".

Why not adopt the simpler approach of just not writing "without exception" unless it's justified? I don't ask that you add qualifiers, as long as you omit the unjustified generalization.

F>[>This is a damn good question, and I don't know the answer. I suspect that our relative positions would indeed be reversed in this discussion.

NSP>Wrong. I'd argue for what I think is right either way.]

F>But that's exactly what I said!

No, it's not. You said that you suspected I would change my relative position. But I'd argue in favor of following Florida legal procedures, either way.

>No, to the best of my knowledge I've never argued that they should be blocked. I've argued that they should NOT be carefully tailored to help one side.

So am I mistaken in thinking you want to block the manual recounts currently in progress in Florida?

Do you instead want them to proceed to completion?

>I think Bush made a mistake turning down a hand recount of the entire state

If Bush loses, history will almost surely point to that as one of the all-time greatest political blunders.

>It was NOT increased accuracy that led to the wrong conclusion.

Okay. Thank you for making this clear on the record.

>Selective increases in accuracy CAN bias the total. This is because such selective increases introduce systematic error. OK?

Close enough for now ... as long as it's understood that the primary source of trouble is systematic error in general, not just the part of that attributable to selective increases in accuracy.

F, to eve>What I've been having trouble with is, the error itself remains just as real whether or not I have pinned the precise cause.

But if you misanalyze the error you know, you could more easily overlook the errors you don't yet know.

F, to eve>My interpretation is that No Spam has been taking the position that because I have misanalyzed the cause of the error, therefore there is no error.

Oh, come on!

Just a short while earlier in your post to me, you claimed to understand that it was because the selective increase in accuracy introduced systematic error that the conclusion was wrong, which is what I've been consistently saying.

Now you're claiming something different when writing to eve?

You've just shown yourself, again, to be dishonest.

F, to eve>Otherwise, why doesn't he say something like "Gore's selective recount is indeed likely to introduce systematic error in Gore's favor, but you haven't understood exactly what that error really is" or words to that effect?

1) Because my reasoning doesn't depend on whether it was Gore or Bush who requested the recount.

2) Because the introduced systematic error does not necessarily favor the candidate who requested the recount.

There. Two reasons why I didn't say what you'd like for me to have said.

F, to eve>No Spam seems to me to be confusing HOW Gore is cheating with WHETHER Gore is cheating

Do you continue to be so blinded by partisan fervor that you cannot recognize a nonpartisan argument? Or is it that you just couldn't bring yourself to make a nonpartisan statement about what I've written?

F, to Unk>If only recounts that favor Gore are permitted because of this deadline, this seems quite opposed to the intent of the deadline itself.

Your if-clause is false. Bush had equal rights and opportunity to request recounts. Will you plainly acknowledge that?

-- No Spam Please (nos_pam_please@hotmail.com), November 20, 2000.


>> [Brian] has yet to address the subject under discussion, preferring to wander up to the front of the stage and exhort the audience to boo his designated villain. An amusing tactic, but without intellectual pretensions. <<

I could care less about the "topic under discussion". If you recall, my man in this election was Nader and not Gore. Gore's "victory" in Florida would be of only minor interest to me. My interest in his assuming the Presidency is, at best, lukewarm.

I made up my mind about the validity of Glaeser's critique days ago, and your argument with No Spam and others over the validity of Glaeser's (or your) points chops no wood and carries no water with me. Further, I could care less whether anyone else is influenced by your arguments.

Yup. No "intellectual pretensions" is good characterization of my attitude in this thread. I just showed up to dog you a bit, to reassure those who have little experience with your debating techniques that they are not imagining your willingness to cram meanings into their words that do not belong there, or your perfect indifference to having this waved under your nose.

That eve still believes in you and your abilities should be gratification enough, Flint. Revel in it.

Your problem, if I diagnose you correctly, is that you have no interest in conquering an unworthy foe, but, when faced with one, you refuse to be act the part of a worthy foe yourself. Anyone such as myself or No Spam, with whom you might clash swords on a somewhat equal basis, motivated out of a similar sense of challenge, you delight to engage in argument. Others, who cannot present a worthy opposition, you scorn. And yet, when you find a worthy foe, you simultaneously cannot prevent yourself from violating all the rules and trashing all the trust we would need for us to want to engage you.

You have ability. But you are an Alkibiades. When Alkibiades was caught biting his opponent in the palestra, he was taunted for roaming outside the rules when threatened with honest defeat, and biting "like a girl". He replied, "No. Like a lion." You are equally afflicted with a need to win, to always win, to always be right, to be invincible, to be more rational than any other person in the room, to come to the "best" conclusion, to defend it to the bitter end, and to be admired for it.

But, if I encourage your "worthy" opponents to stop wrestling with you because you are known to bite, then I will have succeeded in my sole aim. And, yes, my purpose has no "intellectual pretensions", and I hope you are damn well 'amused' with me when you find you have no one left to play with except cpr and his ilk. And you can't blame me. Once somebody has your teeth marks in his leg, my job is easy.

-- Brian McLaughlin (brianm@ims.com), November 20, 2000.


RC,

>WOW! I feel like I've been through the looking glass after reading this thread!

Perhaps after more careful perusal it won't seem so bad.

>Are you saying that, in the absence of perfect knowledge of the election result, any partial recount, no matter how selectively the part is chosen, increases the "accuracy" of the total result?

No, I'm not. That is, if by "the total result", you meant the determination of the winner. If I'm mistaken about that meaning, please post a clarification of what you meant by "the total result".

If you'll carefully reread what I wrote, you'll find that I said that the partial recount (assumed in this context to be more accurate than preceding counts and to be nonfraudulent) would increase the accuracy of the totals of votes for the particular candidates, but I did not say, because it's not necessarily true, that it would increase the accuracy of determining the winner of the election, until we reach the point of having a total recount instead of a partial one.

The winner is determined by comparing the two (assuming a two-candidate race) candidates' individual vote totals. When the difference between those totals is small, a small change in one or both of the candidates' individual totals can reverse the result of the comparison.

>This is so obviously, logically wrong that I'm at a loss as to how to respond to it.

Perhaps you're not familiar with many examples of comparing nearly-equal quantities.

Let's consider an example with specific numbers:

In a state with three counties, RST, UVW, and XYZ, there is a two-candidate race between candidate A and candidate B. After the initial round of counting, by machine, A has 2,100,000 votes and B has 2,099,800 votes. At this point, A leads by 200 votes, which is less than 0.005% of the total of all votes counted so far.

Then there is a manual recount of county XYZ. It doesn't matter for my example which candidate requested that recount. Also, let's assume, for this example, which will not affect my conclusions about how selective recounts affect accuracy, that manual recounts always find the same totals for the portion of all ballots which were counted by machine, but in addition correctly determine the voter's choice on some additional ballots which are valid but were not counted by the machines (because of alignment errors and whatnot).

In the machine count, XYZ's results were: A 600,000; B 701,000. The manual recount comes up with new totals: A 603,000; B 704,500, because of ballots that were valid according to election law, not counted by the machines, but determinable by manual inspection. A gained 3,000 votes, while B gained 3,500 votes.

Substituting the new county XYZ totals for its previous count, we find that the new state totals are: A 2,103,000; B 2,103,300. Now B leads by 300 votes, which is less than 0.008% of the total of all votes counted so far.

A's new statewide total is nore accurate than A's old statewide total, because it now includes at least some ballots that were not properly counted in the previous round of machine counting in county XYZ, in addition to all the ballots that were properly counted by machine in county XYZ. The statewide fraction of voters who intended to vote for A, but whose votes have not yet been properly counted, is now smaller than it was before the recount in county XYZ. Similarly, B's new statewide total is more accurate than B's old statewide total.

Now we recount county UVW. In the machine count, UVW's results were: A 801,000; B 700,000. The manual recount comes up with new totals: A 804,000; B 702,800. A gained 3,000 votes in the UVW recount, while B gained 2,800 votes.

Substituting the new county UVW totals for its previous count, we find that the new state totals are: A 2,106,000; B 2,106,100. Now B leads by 100 votes, which is less than 0.003% of the total of all votes counted so far.

A's new statewide total after the UVW recount is more accurate than A's statewide total after the XYZ recount but before the UVW recount, because it now includes at least some ballots that were not properly counted in the previous round of machine counting in county UVW, in addition to all the ballots that were properly counted by machine and the extra ones counted by manual recount in XYZ. The statewide fraction of voters who intended to vote for A, but whose votes have not yet been properly counted, is now smaller after recounts in both UVW and XYZ than it was before the UVW recount but after the XYZ recount. Similarly, B's new statewide total is more accurate than B's old statewide total.

Finally, we recount county RST. In the machine count, RST's results were: A 699,000; B 698,800. The manual recount comes up with new totals: A 704,000; B 703,550. A gained 5,000 votes in the RST recount, while B gained 4,750 votes.

Substituting the new county RST totals for its previous count, we find that the new state totals, after manual recount of all counties, are: A 2,111,000; B 2,110,850. The lead has switched back to A, who leads by 150 votes, which is less than 0.004% of the total of all votes after manual recount in every county.

A's new statewide total after the RST recount is nore accurate than A's statewide total after the UVW and XYZ recounts but before the RST recount, because it now includes at least some ballots that were not properly counted in the previous round of machine counting in county RST, in addition to all the ballots that were properly counted by machine and by the extra ones counted by manual recounts in UVW and XYZ.

The statewide fraction of voters who intended to vote for A, but whose votes have not yet been properly counted, is now smaller after recounts in all counties than it was before the RST recount but after the UVW and XYZ recounts. In fact, only those voters who intended to vote for A, but whose votes could be determined by neither machine nor manual inspection, do not have their votes included in A's statewide total. If all ballots could be determined by manual inspection, we'd now have a 100% accurate vote total for A.

Similarly, B's new statewide total is more accurate than B's old statewide total.

Notice that as partial recounts were done, the statewide winner could change back and forth. That's because the winner's margin was very sensitive to the difference between nearly-equal quantities. Increasing the accuracy of one or both of those quantities does _not necessarily_ increase the accuracy of the difference between them. The result of subtracting or comparing nearly-equal quantities is sometimes many orders of magnitude less accurate than either of the quantities themselves.

In the _ideal_ case of absolutely accurate manual recounts, our final margin of 150 in A's favor would be exactly correct, after margins in earlier stages of counting of 200 in A's favor, 300 in B's favor, and 100 in B's favor.

-- No Spam Please (nos_pam_please@hotmail.com), November 21, 2000.


RC (continued),

>I'll just say this - we may not have perfect knowledge of the Florida results, BUT we do know that:

>1) Palm Beach County voted roughly 2 to 1 for Gore.

Okay. (Actually. I don't know whether that's true, but I'll accept it for purposes of discussion.)

>2) The problems with punch card voting should occur randomly - anything else would indicate the possibility of fraud.

Not necessarily true.

In the middle of my latest long posting addressed to Flint, I give an example reason for some problems with punch card voting _not_ to be randomly distributed.

Consider the plight of shaky-handed citizens who have trouble pressing the punch properly to register their votes. Such shaky-hands (I can refer to them thusly because I am one -- a side effect of one of my prescription medicines) might plausibly be concentrated in groups of voters who preferentially vote for a particular candidate on the basis of his stand on health care -- e.g., senior citizens or the handicapped.

There you are - an example of a non-random error in punch-card voting.

>3) Therefore, the new votes added by the hand recount process should also break roughly 2 to 1 for Gore.

... and this third point is not necessarily true because it depends on the second point, which is not necessarily true.

>4) Double therefore, the addition of hand count results from PBC (as well as the other Democratic counties) biases the total result in Gore's favor.

Maybe, or maybe not, depending on how the non-random, systematic, errors are distributed.

>This is in essence what Glaeser's saying, and I fail to see how it's debatable.

Do you see now?

-- No Spam Please (nos_pam_please@hotmail.com), November 21, 2000.


Spam, that was a truly outstanding attempt at spin, but no matter how thin you slice it "if you're drunk or high, maybe you ought to sober up before posting again in this thread." implies drunkeness. "Maybe you ought to sober up" implies drunkeness, and "Please sober up" more than implies drunkeness.

Do you still beat your wife?

;-)

BTW, I think we both agree that Bush screwed up by not following Gore's lead. He should have requested recounts in Republican strongholds, or across the entire state. This tactic may have also made Sec of State Harris look less partisan when she tried to enforce the fixed deadline for certified vote totals, since Bush counties may have also not been finished with their hand recounting.

-- Uncle Deedah (unkeed@yahoo.com), November 21, 2000.


Glaeser's opening paragraphs warn us about discounting observations solely because they don't jibe with one's theory. Then he proceeds to do the same thing with his own theory, which is, "Hand counting ballots in only a few, carefully chosen counties is a sure way to bias the results."

To support his theory, he cites the vote recount in selected Florida counties, which has been shown to have the tendency to capture votes that had gone unnoticed when machines had originally counted the vote. Therefore, the tendency of the selected recount in Florida is to increase the margin of the leading vote getter in those counties, a point which has been effectively addressed above.

But consider a vote counting technology that is about as likely to double count votes as to ignore them. In this case, one cannot reasonably anticipate the effect that a recount would have, except that it would be random. A county's original vote count, taken by itself, would not provide any candidate an incentive to request a recount. Rational recount requests would be limited to where a candidate suspects some irregularity other than the normal error in vote counting.

Glaeser fails to consider voting counting technologies that don't support his theory. Hence, he erroneously concludes that a partial recount in any closely contested state, will distort the intent of the people.

-- David L (bumpkin@dnet.net), November 21, 2000.


">1) Palm Beach County voted roughly 2 to 1 for Gore.

Okay. (Actually. I don't know whether that's true, but I'll accept it for purposes of discussion.)"

No, No (or may I call you Mr. Please?), this is THE essential point, and the point that's entirely missing from from your example above.

We know what the original count and the machine recount said about PBC - in fact, the sample precincts used for the first sample hand recount were even more slanted. One was more than 10 to 1 in favor of Gore. We KNOW that even the fairest hand recount of this kind of slanted sample of votes distorts the outcome. This is elementary logic.

"Consider the plight of shaky-handed citizens who have trouble pressing the punch properly to register their votes. Such shaky-hands (I can refer to them thusly because I am one -- a side effect of one of my prescription medicines) might plausibly be concentrated in groups of voters who preferentially vote for a particular candidate on the basis of his stand on health care -- e.g., senior citizens or the handicapped.

There you are - an example of a non-random error in punch-card voting. "

But this is an example of the classic "pregnant chad", which, since I last heard, were not being counted anyway (has that changed now? - I know this was being contested). The other chad problems (hanging, swinging, gay, etc...), the ones which ARE being examined in the ongoing hand recounts, ARE random in nature.

So I still maintain that errors found in the hand recounts as they have been done up to this point should be randomly distributed. And that Glaeser's point is valid.

It also brings me back to a point I made earlier - the PBC ballot confusion problems, which were the original justification for requesting hand recounts, are in no way being addressed by the hand recounts.

-- RC (randyxpher@aol.com), November 21, 2000.


"But consider a vote counting technology that is about as likely to double count votes as to ignore them. In this case, one cannot reasonably anticipate the effect that a recount would have, except that it would be random. A county's original vote count, taken by itself, would not provide any candidate an incentive to request a recount."

I'm not sure I'm following this, but the effect of the current hand recount methods being applied should be random - the chance that upon punching a stylus through a perforated punch card would not totally detach the perforations.

Distribute a random variable across a 2 to 1 distribtion of data, and the changed data should also break 2 to 1.

"Rational recount requests would be limited to where a candidate suspects some irregularity other than the normal error in vote counting."

I believe that's what the stae law says.

"Glaeser fails to consider voting counting technologies that don't support his theory. Hence, he erroneously concludes that a partial recount in any closely contested state, will distort the intent of the people."

Glaeser is only considering the voting technology that is in question here, the punched card. I imagine the error rate is much higher with punch cards than any other technology. BUT, assuming a random distribution of errors, a partial recount of data which is slanted a certain way DOES distort the outcome. Absolutely.

-- RC (randyxpher@aol.com), November 21, 2000.


RC, My problem with Glaeser's argument is that its scope is inconsistent. When he begins discussing the Florida recount, he seems to be restricting his scope to the punched card method of voting. If he stuck to that throughout, I'd have no problem.

But he ends up here: "If there is to be recounting by hand, it cannot be selective. There needs to be total hand counting, not just within Florida, but across the U.S. in any state that was close." He is attempting to extrapolate from the recounting of punched card votes, to recounts of votes collected using any and all other voting technologies. In trying to make an overly broad conclusion, his argument is rendered unsound.

-- David L (bumpkin@dnet.net), November 21, 2000.


David,

A point I meant to make in response to your earlier post: I'm also assuming a random distribution of errors even with different technologies - because if the errors are non-random, it's obviously a lousy way to run a railroad - and Glaeser's tacitly making that assumption, too.

You are correct that he didn't spell that out, but I don't see that this invalidates his position.

One can certainly argue that the astronomical error rate that goes with punch card ballots distorts the outcome in and of itself. Anyone want to do the research to find out how many Floridians voted on punch cards (regardless of county) and how they broke out for the candidates?

Gore's recounts distort the outcome. Had Bush requested hand recounts, they would have distorted the outcome.

Seems like we left "the will of the people" behind several distortions ago.

-- RC (randyxpher@aol.com), November 21, 2000.


Uncle Deedah,

>that was a truly outstanding attempt at spin,

It was the truth, whether you like it or not.

>but no matter how thin you slice it "if you're drunk or high, maybe you ought to sober up before posting again in this thread." implies drunkeness.

No, it implies the _possibility_ of drunkenness.

>"Maybe you ought to sober up" implies drunkeness, and "Please sober up" more than implies drunkeness.

In _Webster's Third New International Dictionary_, the first definition given for "sober" as a verb is: "1 : to make (a person) serious, grave, or thoughtful".

I apologized for the mistake I made.

>BTW, I think we both agree that Bush screwed up by not following Gore's lead. He should have requested recounts in Republican strongholds, or across the entire state.

If Bush loses, it will be considered a great blunder.

If Bush wins, it will be a footnote.

>This tactic may have also made Sec of State Harris look less partisan when she tried to enforce the fixed deadline for certified vote totals, since Bush counties may have also not been finished with their hand recounting.

Another reason to wonder about Bush's political acumen in making the decision.

-- No Spam Please (nos_pam_please@hotmail.com), November 22, 2000.


David L,

>But consider a vote counting technology that is about as likely to double count votes as to ignore them.

Intriguing, for theoretical discussion, but highly impractical, of course.

>In this case, one cannot reasonably anticipate the effect that a recount would have, except that it would be random.

... depending on the details of "about as likely", that is. :)

>A county's original vote count, taken by itself, would not provide any candidate an incentive to request a recount.

Maybe not the same incentives as we have now, but other incentives instead.

>Rational recount requests would be limited to where a candidate suspects some irregularity other than the normal error in vote counting.

... depending on how "normal" "about as likely" is! :)

-- No Spam Please (nos_pam_please@hotmail.com), November 22, 2000.


RC,

NSP>Okay. (Actually. I don't know whether that's true, but I'll accept it for purposes of discussion.)"

RC>No, No (or may I call you Mr. Please?),

Yes, either.

Acceptable forms include No, Mr. Please, No Spam (a la "Joe Bob" or "Debbie Sue"), No Spam Please, and NSP, with varying degrees of formality. Use of the middle name Spam without the preceding first name No is impolite, as it verges on misrepresentation. :)

>this is THE essential point,

Wait a minute. My throw-in comment of "Actually, I don't know ..." was a self-deprecation with reference to my perfectionistic tendencies. I meant that I didn't know whether the ratio was 2.2:1, 2.035:1, 1.9997:1, or whatever.

I _did_ say "Okay."

I _did_ say "I'll accept it for purposes of discussion."

>and the point that's entirely missing from from your example above.

... because my examples were and are intended to be independent of the actual vote ratios in the current election.

>We know what the original count and the machine recount said about PBC

What did you think they said? Spit it out.

If you mean fraud, say so, and see my previous comments on election fraud.

>in fact, the sample precincts used for the first sample hand recount were even more slanted. One was more than 10 to 1 in favor of Gore.

Do you have evidence that that particular precinct did not actually vote for Gore in a ratio of 10:1? I happen to live close to a highly-partisan area which would neither surprise me nor cause me to think of fraud if it went 10:1 or 20:1 for Bush.

>We KNOW that even the fairest hand recount of this kind of slanted sample of votes distorts the outcome. This is elementary logic.

Please define what you mean by "the outcome", in accordance with distinctions I drew in recent examples I have posted.

NSP>"Consider the plight of shaky-handed citizens < snip >

RC>But this is an example of the classic "pregnant chad",

... or the "hanging chad", or the "swinging chad", or any other partially-punched-but-not-completely-separated chad. I meant that shaky-handed voters might not press the punch tool completely and totally all the way down to absolutely, completely, unequivocally separate the chad from the ballot in larger proportion than steady-handed voters did.

>The other chad problems (hanging, swinging, gay, etc...), the ones which ARE being examined in the ongoing hand recounts,

... and which also might be produced in larger proportion by shaky-handed voters than by steady-handed voters ...

>ARE random in nature.

Not when shaky-handed voters produce those partially-punched cases in greater proportion than other voters.

>So I still maintain that errors found in the hand recounts as they have been done up to this point should be randomly distributed.

... perhaps because you yourself are steady-handed and did not fully realize the potential consequences of a shaky hand on the voting punch.

>And that Glaeser's point is valid.

Nope. Try again.

>It also brings me back to a point I made earlier - the PBC ballot confusion problems, which were the original justification for requesting hand recounts,

... but not the only justification ...

>are in no way being addressed by the hand recounts.

But other justifications, such as the various categories of votes represented by partially-separated chads, _are_ being addressed by the hand recounts.

-- No Spam Please (nos_pam_please@hotmail.com), November 22, 2000.


Uncle Deedah,

Let me amend my previous reference to dictionary definition of "sober".

When I wrote:

NSP>In _Webster's Third New International Dictionary_, the first definition given for "sober" as a verb is: "1 : to make (a person) serious, grave, or thoughtful".

... I was quoting the definition for the transitive use of "sober".

For the intransitive use, the first definition given by that same dictionary is: "1 : to become sober: as a : to become serious or thoughtful".

-- No Spam Please (nos_pam_please@hotmail.com), November 22, 2000.


Does it depend on what the meaning of is is?

-- Uncle Deedah (unkeed@yahoo.com), November 22, 2000.

I'm going to post these both here and on the continuation thread, to which, if someone would provide a link for the HTML impaired, I'd be indebted, if one could get past the tortured syntax of this sentence.

NSP said: "Wait a minute. My throw-in comment of "Actually, I don't know ..." was a self-deprecation with reference to my perfectionistic tendencies. I meant that I didn't know whether the ratio was 2.2:1, 2.035:1, 1.9997:1, or whatever.

I _did_ say "Okay."

I _did_ say "I'll accept it for purposes of discussion." "

RC sez: I _did_ say "roughly". I have no perfectionistic tendencies. Just ask my boss.

RC (that's me) blurted out: >and the point that's entirely missing from from your example above.

NSP responded: "... because my examples were and are intended to be independent of the actual vote ratios in the current election."

I say: That's why your examples were not particularly useful - the actual vote ratios for the current election are the crux of the biscuit, as it were, related to the precincts/counties chosen for hand recounting.

I sed: >We know what the original count and the machine recount said about PBC

NSP sed: "What did you think they said? Spit it out."

I say: I think they said what I said they said earlier - they said that PBC voted "roughly" 2 to 1 for Gore. Gack.

NSP... "If you mean fraud, say so, and see my previous comments on election fraud."

Me... I do not mean fraud.

I said earlier: >in fact, the sample precincts used for the first sample hand recount were even more slanted. One was more than 10 to 1 in favor of Gore.

NSP said: "Do you have evidence that that particular precinct did not actually vote for Gore in a ratio of 10:1? I happen to live close to a highly-partisan area which would neither surprise me nor cause me to think of fraud if it went 10:1 or 20:1 for Bush."

I say: I have no problem believing a particular precinct voted 10:1 or 20:1. That's not the issue.

The issue is the choice of those precincts for a SAMPLE recount - the issue is that folks look at the results of that recount and say "Gore gained a bunch of votes --> there was a problem --> this justifies recounting the whole county" - as opposed to saying "the ratio of votes gained by the two candidates in the sample recount is not out of line with the ratio in the original machine count --> this indicates a physical problem with punch cards which can't be fixed on a local basis without skewing the overall outcome".

I said: >We KNOW that even the fairest hand recount of this kind of slanted sample of votes distorts the outcome. This is elementary logic.

NSP responded: "Please define what you mean by "the outcome", in accordance with distinctions I drew in recent examples I have posted."

I say: "The outcome", as I meant it there, is the number of votes cast and counted for each candidate. As in: changing the "outcome" is different from changing the "winner".

-- RC (randyxpher@aol.com), November 22, 2000.


NSP>"Consider the plight of shaky-handed citizens < snip >

RC>But this is an example of the classic "pregnant chad",

NSP>... or the "hanging chad", or the "swinging chad", or any other partially-punched-but-not-completely-separated chad.

NoNoNoNoNo, No.

Yer "pregnant chad" is fundamentally different from the other chad species. Hanging/swinging chads can be produced by any complete punch of the card. You'll have to explain to me how shaky-handedness plays a role in this.

The knocked-up chad is, by definition, not a complete punch.

RC>The other chad problems (hanging, swinging, gay, etc...), the ones which ARE being examined in the ongoing hand recounts,

NSP>... and which also might be produced in larger proportion by shaky-handed voters than by steady-handed voters ...

Like I said - I need some explanation on that. I don't buy it. One could argue that maybe the strident thrust of the hulking bodybuilder's stylus is more likely to leave a dangling chad than the deliberate soft pressure of a little old lady...

RC>So I still maintain that errors found in the hand recounts as they have been done up to this point should be randomly distributed.

NSP>... perhaps because you yourself are steady-handed and did not fully realize the potential consequences of a shaky hand on the voting punch.

Speculate not on my "handedness".

RC>And that Glaeser's point is valid.

NSP>Nope. Try again.

Back atcha.

-- RC (randyxpher@aol.com), November 22, 2000.


No Spam:

[I never intended "Please sober up" to be an accusation of drunkenness]

This is silly. You are seriously telling me that should I write "IF you are going to insist on being such a jerk, why should I respond", that I never suggested you were a jerk? That by writing "IF". I absolved myself of any insinuation? I can only laugh, and suggest that IF you think you are making an honest argument here, THEN you need to re-consult your copy of the dishonest debater's handbook, beginner's edition.

[Because of the closeness of the vote nationwide and in Florida, it is probable that some people will never consider whoever wins to have a legitimate power base, but I think we can try to do all we can to minimize the fraction who so consider.]

I agree. I think it's important that the public believes that the winner actually got the most votesin the entire state. I think Bush should have agreed to a hand recount of the whole state for this reason. I don't think people will consider Gore's customized recount legitimate, and already there is a suspicion that this is a trick to steal the election. Not a good sign.

[Flint, I am being honest, regardless of whether you like it or agree.]

OK. I'm just trying to say that since punching IS voting, there being no practical difference between the two, it simply didn't occur to me that you would be emphasizing a difference that does not exist in this implementation. Since I wasn't able to see a nonexistent difference, I was obliged to look elsewhere for the source of my "misrepresentation" and I took my best guess as to what that might be.

[How did you determine that I didn't recognize that?]

By recommending that it be done. In practice, your recommendation *requires* the legislature and/or the courts to either individually ponder every ballot, or predict unknowable details about procedures. Neither of these is possible. If I recommend that you flap your arms and fly to the moon, and I mean this *seriously*, I think you're safe in saying that I failed to recognize the impossibility of my recommendation.

[A word search finds that I have not used the phrase "somebody else", or even the word "somebody", in this thread.]

You have bucked the responsibility for ballot interpretation from the counters to the courts or the legislature. When I asked YOU how you would decide, you kept saying you'd decide according to properly written and interpreted law designed and written by someone else other than you. I admit I assumed that someone MUST have the ultimate responsibility for determining the disposition of ambiguous ballots, and I tried to ask you how you would phrase any such rules if it were your responsibility to do so. Instead, you just said you would follow the rules laid down by *somebody else", like the courts or the legislature. Don't dodge this, please.

My point was that WHOEVER has such ultimate responsibility cannot avoid ambiguity such that partisan interpreters cannot disagree, no matter WHAT they write into the law. Saying you will follow what someone else writes into law fails to address the intended issue, which is the core ambiguity that cannot be avoided no matter who does what.

[... in the absence of systematic error invalidating the preceding conclusions about probabilities. (*sigh*)]

Yes, I agree with you. We lack complete information. There *might* be systematic sources of underlying error we are not aware of. Gore's recount *might* backfire, producing a net gain for Bush because of exactly such underlying systematic trends. I agree we are playing what we believe to be the odds *without* full knowledge of those odds.

Going back to Glaeser, he knows (and so do I) that sometimes, improving the accuracy of ONLY those data points you don't like, makes them EVEN WORSE in terns of your desired preselected outcome. In that case, typically you choose not to publish [grin].

[F>But that's exactly what I said!

No, it's not. You said that you suspected I would change my relative position. But I'd argue in favor of following Florida legal procedures, either way.]

Chuckle. You said you would argue for "what you think is right". Yes, so would I. My whole point was that depending on our partisan positions, what we think is right would flip sides. The Florida law is clearly ambiguous, and the courts are now disagreeing *with one another* about what it means. They disagree along party lines, isn't that amazing? So you'd favor following *your interpretation* of Florida legal procedures. But what are these procedures when even the courts can't agree? Depends on whom you want to win, of course. Just as I said.

[Do you continue to be so blinded by partisan fervor that you cannot recognize a nonpartisan argument? Or is it that you just couldn't bring yourself to make a nonpartisan statement about what I've written?]

This is a misinterpretation on your part, not intentional. Your argument is nonpartisan, since the error in procedure doesn't depend *in general* on which candidate makes the error. The issue becomes partisan when this source of error is being employed for the explicit purpose of introducing deliberate, directed bias. This is what is indeed happening. I do not believe selective recounts should be permitted by *anyone*, but I recognize that the law permits candidates to do so, which in turn permits selective advantage (as far as we can tell, anyway, without complete knowledge). Bad law, in my opinion.

[Your if-clause is false. Bush had equal rights and opportunity to request recounts. Will you plainly acknowledge that?]

How can an if-clause be false? It's a statement contrary to fact, hence the "if". Yes, either candidate could have demanded a recount. Nominally, the deadline for demanding recounts has now expired, with only Gore having met that deadline. I suspect the courts would in fact decide to permit a statewide hand recount if it were requested, nonetheless.



-- Flint (flintc@mindspring.com), November 22, 2000.


RC,

>I say: That's why your examples were not particularly useful - the actual vote ratios for the current election are the crux of the biscuit, as it were, related to the precincts/counties chosen for hand recounting.

But my examples were not addressing the choice of precincts/counties for hand recounting.

If you want to criticize my examples further, please specify just which examples you are talking about.

>The issue is the choice of those precincts for a SAMPLE recount

So? The choice was legal. Bush had the same rights and opportunity to request recounts as Gore in counties favoring him, and I think county officials could easily have found precincts favoring Bush by 10:1.

>the issue is that folks look at the results of that recount and say "Gore gained a bunch of votes --> there was a problem --> this justifies recounting the whole county" - as opposed to saying "the ratio of votes gained by the two candidates in the sample recount is not out of line with the ratio in the original machine count --> this indicates a physical problem with punch cards which can't be fixed on a local basis without skewing the overall outcome".

The sloppy thinking of folks following that particular train of thought does not constitute a valid objection to a legal procedure.

>We KNOW that even the fairest hand recount of this kind of slanted sample of votes distorts the outcome. This is elementary logic.

By "slanted", do you mean a dishonest count? If so, please say that. By "distorts the outcome", do you mean "portrays other than a correct outcome"? If so, please say just what kind of distortion you mean. Otherwise, I've already posted a numerical example of how the lead in a close race can switch back and forth according to various partial recounts, without "slanting" or "distortion" in a perjorative sense.

>"The outcome", as I meant it there, is the number of votes cast and counted for each candidate. As in: changing the "outcome" is different from changing the "winner".

Well, my previous example shows what happens to numbers of votes cast and counted for each candidate. NSP>... or the "hanging chad", or the "swinging chad", or any other partially-punched-but-not-completely-separated chad.

RC>NoNoNoNoNo, No.

>Hanging/swinging chads can be produced by any complete punch of the card.

... especially if the voter is not holding the punch straight enough for a clean complete separation of chad from ballot, which could happen because the voter's hand was shaky.

>You'll have to explain to me how shaky-handedness plays a role in this.

Shaky-handedness could contribute to not holding the punch tool within the tolerance of verticality that produces a clean complete separation of chad from ballot. Also, the pressure exerted downward by a shaky voting hand might let up just at the moment when the punch tool has separated some, but not all, of the perforations around a chad.

NSP>... and which also might be produced in larger proportion by shaky-handed voters than by steady-handed voters ...

>Like I said - I need some explanation on that. I don't buy it.

Keeping in mind that the shakiness (i.e., involutary muscle tremors) can produce split-second variations in angle, position, and pressure of the punch tool, imagine an irregular descent of the tool upon the ballot.

>One could argue that maybe the strident thrust of the hulking bodybuilder's stylus is more likely to leave a dangling chad than the deliberate soft pressure of a little old lady...

Well, there you have provided your own example of how different people can get different results. But don't confuse shakiness, which is the result of involuntary muscle tremors, with weakness. I didn't specify weakness.

>So I still maintain that errors found in the hand recounts as they have been done up to this point should be randomly distributed.

When you finally understand what shakiness in using the stylus implies, you'll see where nonrandomness can arise.

RC>And that Glaeser's point is valid.

NSP>Nope. Try again.

RC>Back atcha.

Double backs atcha.

-- No Spam Please (nos_pam_please@hotmail.com), November 22, 2000.


Brian:

[I hope you are damn well 'amused' with me when you find you have no one left to play with except cpr and his ilk. And you can't blame me.]

Yes, I understand. Armed with righteousness, you pursue a crusade to educate those intelligent people with whom I might enjoy conversing, in the hope of altruistically saving them from blundering into the error of holding opinions dishonestly presented. As soon as all the smart people here avoid me, you will have succeeded in exorcising the devil.

Well, far out. Pretty hardcore. You must be seriously irritated. Now, how about recognizing that people think for themselves, and don't need you to tell them what opinions to hold. But I'd like to see someone tell me "I won't talk to you because Brian warned me", just to warm your heart. Your heart seems to need this.

-- Flint (flintc@mindspring.com), November 22, 2000.


Flint,

>You are seriously telling me that should I write "IF you are going to insist on being such a jerk, why should I respond", that I never suggested you were a jerk?

Your inclusion of "such" is a back-reference implying a previous mention of "jerk". So your example does not parallel what I wrote.

If, instead, you wrote, "IF you are going to insist on being a jerk, why should I respond", then you would have suggested the possibility of being a jerk, but not made an accusation of that. I could respond, "Well, I'm not going to be a jerk", and the second part of your if-sentence would be irrelevant.

But note in the non-"such" case that in order for your example to parallel what I wrote, "being a jerk" would have to be as objectively determinable as being drunk, and "why should I respond" should be replaced by a non-question.

>That by writing "IF". I absolved myself of any insinuation?

No. Actually, "such" implied a previous assertion of "jerk".

>I can only laugh,

Not a surprise.

>and suggest that IF you think you are making an honest argument here, THEN you need to re-consult your copy of the dishonest debater's handbook, beginner's edition.

Your suggestion is unwarranted.

>I don't think people will consider Gore's customized recount legitimate, and already there is a suspicion that this is a trick to steal the election.

Gore has, you know, proposed extending the recount to the entire state.

According to what I've read, Bush rejected Gore's offer because of certain accompanying conditions.

But if Bush continues to reject proposals for statewide recount, Republican accusations of "tricks" will be seen through. A lot of us remember Watergate.

>I'm just trying to say that since punching IS voting, there being no practical difference between the two,

... so all those voters not using punch-card ballots did not cast votes? My pen-marked ballot, counted by optical scan, did not represent a valid manifestation of my intent to vote?

If you agree with thet, you have a clearly untenable position. If you disagree, then you're admitting that punching and voting are not synonymous. Which is it?

>Since I wasn't able to see a nonexistent difference, I was obliged to look elsewhere for the source of my "misrepresentation" and I took my best guess as to what that might be.

Bull. You're not that dumb.

>In practice, your recommendation *requires* the legislature and/or the courts to either individually ponder every ballot, or predict unknowable details about procedures.

No, it doesn't. It just requires a sufficiently clear definition of what constitutes a valid ballot.

E.g., the definition _might_ say (though this example might not be the best choice, or best wording) something like: (1) if any of the various pieces of card between the pre-voting perforations is visibly broken, then that shall be considered to represent a voter's intent to register a vote in that position. (2) But if there is no visible break in any of those pieces of card, then there shall be considered to be no vote in that position. (3) The appropriate local election officials, and those they designate, are to make the decisions on whether there is a visible break or not. (4) Appeals of such decisions are to be handled by so-and-so in thus-and-such manner ...

Now, this example definition rules out the so-called "pregnant chads", but that's only because the purpose of my example is to illustrate what level of detail the law can reasonably specify to guide the election officials in their decisions during a manual recount.

>Neither of these is possible.

... or necessary.

>You have bucked the responsibility for ballot interpretation from the counters to the courts or the legislature.

No, I didn't. The courts and legislatures should provide sufficiently clear definitions and guidelines for the local counters to be able to properly interpret ballots as often as possible, while also providing workable procedures for handling cases found not to be determinable under the law.

What we have seen recently is that existing Florida law has been sloppy about providing those definitions. I'm disappointed. I've seen other state laws that are sufficiently well-written to allow any reasonable person to determine what to do in a similar situation.

So now, it's the job of the Florida courts and legislature to provide some things that should have been in that state's election laws, but weren't.

>When I asked YOU how you would decide, you kept saying you'd decide according to properly written and interpreted law designed and written by someone else other than you.

That's not the wording I used to express my intent.

>I admit I assumed that someone MUST have the ultimate responsibility for determining the disposition of ambiguous ballots,

Well, that would be the collective responsibility of the Florida state legal structure and government representing the citizens of Florida.

>I tried to ask you how you would phrase any such rules if it were your responsibility to do so.

Why didn't you just ask it plainly that way?

The example I give above applies to your question, I think. But I can't provide a more detailed or definitive phrasing unless I were familiar with the details of Florida state law.

But why, Flint, are you so interested in how _I_ would write or administer the law? Why don't you trust the state of Florida to do the job of handling this, including the cleanup of legislative oversights, in accordance with its own legal structure, of which I am not a part?

>Instead, you just said you would follow the rules laid down by *somebody else", like the courts or the legislature. Don't dodge this, please.

What I said did not include your quoted phrase, "somebody else". Mhy do you dodge the issue of your multiple misrepresentations of what I've actually said?

I'm not dodging the issue. As a non-Florida resident, I have no official say in their legal matters. You seem to be trying to get me to act as though I were -- why?

>My point was that WHOEVER has such ultimate responsibility cannot avoid ambiguity such that partisan interpreters cannot disagree, no matter WHAT they write into the law.

So? That itself doesn't prevent adequate functioning of the electoral process.

>Saying you will follow what someone else writes into law fails to address the intended issue, which is the core ambiguity that cannot be avoided no matter who does what.

But that ambiguity can be reduced to levels such that the electoral process functions proerly despite sincere and honest disagreements.

>Going back to Glaeser, he knows (and so do I) that sometimes, improving the accuracy of ONLY those data points you don't like, makes them EVEN WORSE in terns of your desired preselected outcome. In that case, typically you choose not to publish [grin].

Everyone else reading this thread can see that I publish about a great deal. Please specify what you are talking about, with specific quotations.

What is your direct answer to my question about dishonesty of your postings, Flint?

>My whole point was that depending on our partisan positions, what we think is right would flip sides.

No, it wouldn't. Following Florida state legal procedure is right. Going to federal court to override those before even trying state court is wrong.

I would not "flip sides".

>The Florida law is clearly ambiguous,

... in some respects ...

>So you'd favor following *your interpretation* of Florida legal procedures.

I favor following the interpretation worked out by the people of Florida in accordance with their legal structures and procedures.

>But what are these procedures when even the courts can't agree?

The courts are part of the legal structure. The possibility of disagreements has been anticipated -- that's why there are appellate courts and a supreme court. It's their decision, not that of the federal courts.

NSP>[Do you continue to be so blinded by partisan fervor that you cannot recognize a nonpartisan argument? Or is it that you just couldn't bring yourself to make a nonpartisan statement about what I've written?]

F>This is a misinterpretation on your part, not intentional. Your argument is nonpartisan, since the error in procedure doesn't depend *in general* on which candidate makes the error.

Please explain in detail (a) which part of my arguments you are referring to, and (b) how that part is nonpartisan.

>I do not believe selective recounts should be permitted by *anyone*, but I recognize that the law permits candidates to do so,

Oh. So you disagree with the law in this regard. Thank you for clearing that up.

>Bad law, in my opinion.

What, then, do you think would be good law in this case?

NSP>Your if-clause is false. Bush had equal rights and opportunity to request recounts. Will you plainly acknowledge that?

F>How can an if-clause be false? It's a statement contrary to fact, hence the "if".

No, that's not what the "if" means. The if-clause is a conditional statement, which might be true or not true.

I think you know that very well. Are you trying to confuse the readers?

>Yes, either candidate could have demanded a recount.

Thanks for acknowledging that.

>Nominally, the deadline for demanding recounts has now expired, with only Gore having met that deadline.

Thanks for acknowledging that.

>I suspect the courts would in fact decide to permit a statewide hand recount if it were requested, nonetheless.

Probably so, in light of the extraordinary situation.

I don't mind a statewide recount. I mind whining because one side failed to foresee the folly of its inaction. I mind false accusations of illegality by people who know better.

-- No Spam Please (nos_pam_please@hotmail.com), November 22, 2000.


My apologies to the players in this thread for not posting here that a continuation thread has been started here because this one is taking forever to load now.

Particular apologies to No_Spam because I didn't specify in the continuation thread enough information to indicate WHICH op-ed piece of Eve's was being continued. [Truth of the matter was, I couldn't remember how to spell Glaeser's name at the time.]

It's up to y'all to post wherever you want, but I won't be following this thread anymore due to the slow load.

-- Anita (Anita_S3@hotmail.com), November 22, 2000.


NSP>"So? The choice was legal. Bush had the same rights and opportunity to request recounts as Gore in counties favoring him, and I think county officials could easily have found precincts favoring Bush by 10:1."

But see, here's the point of all this - you've now fallen back to what's LEGAL! *Not* what makes logical or statistical sense... it's LEGAL for Gore to distort the data in his favor - it would have been LEGAL for Bush to distort the data equally (or maybe a little *more* than equally) in his favor had he chosen to do so...

But at this point we've left Glaeser and his still valid point behind.

"The sloppy thinking of folks following that particular train of thought does not constitute a valid objection to a legal procedure."

Quid pro quo. Ipso facto. Cogito ergo sum. Etc.

NSP>"By "slanted", do you mean a dishonest count? If so, please say that."

I most vehemently do NOT mean dishonest.

NSP> By "distorts the outcome", do you mean "portrays other than a correct outcome"? If so, please say just what kind of distortion you mean.

Yes, pretty much - I mean statistical distortion. Logical distortion. Glaeserian distortion.

NSP>"Otherwise, I've already posted a numerical example of how the lead in a close race can switch back and forth according to various partial recounts, without "slanting" or "distortion" in a perjorative sense."

I grant you that, but your example is still not applicable to the Florida recount situation, where the the partial recounts are clearly slanted in a statistical, non-perjorative sense.

Footnote - I still stand by a statement I made last week that the failure of the Bush folks to request hand recounts will, if he winds up losing the election, be seen as one of the bonehead moves in all of history. And I still think he's gonna lose.

-- RC (randyxpher@aol.com), November 22, 2000.


"...Shaky-handedness could contribute to not holding the punch tool within the tolerance of verticality that produces a clean complete separation of chad from ballot. Also, the pressure exerted downward by a shaky voting hand might let up just at the moment when the punch tool has separated some, but not all, of the perforations around a chad...

...Keeping in mind that the shakiness (i.e., involutary muscle tremors) can produce split-second variations in angle, position, and pressure of the punch tool, imagine an irregular descent of the tool upon the ballot...

...>One could argue that maybe the strident thrust of the hulking bodybuilder's stylus is more likely to leave a dangling chad than the deliberate soft pressure of a little old lady...

Well, there you have provided your own example of how different people can get different results. But don't confuse shakiness, which is the result of involuntary muscle tremors, with weakness. I didn't specify weakness...

...>So I still maintain that errors found in the hand recounts as they have been done up to this point should be randomly distributed.

When you finally understand what shakiness in using the stylus implies, you'll see where nonrandomness can arise..."

No Spam, I quote all this just to show that, and I'm sure you realise this, that any discussion of the effects of shakiness/weakness/whatever on the likelihood of hanging a chad is pure speculation. Your explanation of this effect is as much guesswork as my goofy bodybuilder example...

Like you said, "Well, there you have provided your own example of how different people can get different results." Except that I refuse to speculate on those different results in the absence of some good industrial test data on punch card ballots.

-- RC (randyxpher@aol.com), November 22, 2000.


Flint,

I am not going to talk to you because Brian has warned me about your tricks.

Sorry, nothing personal, but I've been warned.

-- Uncle Deedah (unkeed@yahoo.com), November 22, 2000.


To whom it may concern (other than Flint)

Someone who I am not speaking to wrote; "If only recounts that favor Gore are permitted because of this deadline, this seems quite opposed to the intent of the deadline itself. Selective recounts can indeed help the trailing party. In effect, you are recommending the legal technicality of waiting until the last minute to file for a partial recount, forcing it to be one-sided in your favor from sheer timing. I don't think this ruse will pass muster in any court.

I think an intelligent court would rule that recounts cannot be "structured" so as to preferentially benefit one side. It's possible that the law will be tossed out as it currently reads. But nobody is going to allow the US Presidency to hinge on such a technicality.

No matter WHAT the deadline is someone can always wait until the last moment to file for a recount. Thus, IF this is indeed an unfair situation we are left with only two choices to rectify this unfair advantage.

One, no deadline.

Two, a hand recount MUST take place across the entire state even if problems are shown to exist in only ONE SINGLE polling place, maybe if even only ONE SINGLE district recounts by hand due to mechanical failure, the entire state must do a hand recount.

I am not sure that either one is a better solution than what the law already says now.

Ya pays your money and ya takes your chances.

-- Uncle Deedah (unkeed@yahoo.com), November 22, 2000.


No Spam:

[No. Actually, "such" implied a previous assertion of "jerk".]

No, your tortuous misunderstanding of "such" is an *inference* on your part, which you are working very hard to justify. It should be clear that if anything, the "such" is referring to the post being replied to, that is, the current instance.

[Your suggestion is unwarranted.]

We must agree to disagree about this.

[... so all those voters not using punch-card ballots did not cast votes? My pen-marked ballot, counted by optical scan, did not represent a valid manifestation of my intent to vote?]

Hey, have you been taking lessons from me? If you go back and read, you will find 3 or 4 cases where I was careful to specify "in this implementation", and "as implemented in these counties", and words to that effect. Now you carefully omit this qualification, and THEN come back and attack me for my failure to MAKE this qualification! Now, earlier you were irritated when I quoted part of your statements and (in your reading) failed to include what you considered critical. When *I* did this, it was a misrepresentation. Do you claim it's honest when YOU do it?

[Bull. You're not that dumb.]

I cannot find differences where they do not exist. Your insistence that such difference *would* exist if only they existed, and that therefore they DO exist, defies belief just as much. We are recounting punched ballots. A punch IS a vote when such ballots are used. Voters using punched ballots did not have the option of voting any other way.

[It just requires a sufficiently clear definition of what constitutes a valid ballot.]

If you mean, a single broken perforation by one candidate's name while the entire chad is gone next to another candidate's name MUST be interpreted as a vote for two candidates, then I agree with your proposed definition. If such a law were created before punched ballots were developed, we'd potentially have a great deal of obsolete verbiage (for the previous, now abandoned method) and no clear guidance for the current method. To avoid this, the legislature must (a) write volumes closely tied to current technology; and (b) they (or the courts) would need to update all that verbiage with every technological improvement. I suggest this is less possible in practice than you imply it is in theory.

[So now, it's the job of the Florida courts and legislature to provide some things that should have been in that state's election laws, but weren't.]

I almost agree. I think this kind of anal detail isn't so much the responsibility of the legislative and judicial branches as it is of the executive branch. There should have been a body of administrative regulation addressing this exact problem. That regulation should have been clear BOTH as to just what counts as a vote, and just how to resolve disagreements as to what counts as a vote. I agree Florida dropped the ball on this one.

[Why didn't you just ask it plainly that way?]

Arrgh! It was plain to me when I wrote it. *I* knew what I meant. But I won't say your failure to understand my intent was anyone's fault but my own. All I can say is, if I wrote what I read, I would know what I meant to say. But this cuts both ways. *You* know what you intend to say as well, and (if I understand correctly) you think my misreading of your intent implies distortion on my part. Maybe Brian's cheerleading influenced you? I continue to respond to my best understanding of your intent, and sooner or later my phrasing "works" in the sense that your understanding matches my intent. A communications breakthrough!

[I'm not dodging the issue. As a non-Florida resident, I have no official say in their legal matters. You seem to be trying to get me to act as though I were -- why?]

This is a way of thinking drilled into me, I think. It says, "Put yourself in the other guy's position. What does the world look like to him? What problems does he face?" So I try to point out that ambiguities are inherent in any process, and we must recognize this to deal with it. The best way to recognize it is to try to define a process so as to eliminate ambiguities, and see if someone can find some anyway. And sure enough, they always can! I don't want to demand that someone do the impossible, and when they can't do it, blame THEM because "it's not MY job."

[That itself doesn't prevent adequate functioning of the electoral process.]

Sigh. There are many ways the process can function "adequately" depending on what "adequate" means. Corruption is "adequate" if nobody complains.

[But that ambiguity can be reduced to levels such that the electoral process functions properly despite sincere and honest disagreements.]

OK, here again we must agree to disagree. There is an intractable minimum of measurement error. The more we reduce ambiguity, the closer the election must be before that ambiguity becomes critical. I do not believe it can be entirely eliminated. People WANT some of it, I think. What use is power if you can't apply it to get your preferences?

[What is your direct answer to my question about dishonesty of your postings, Flint?]

That it is only rarely intentional. That I am responding to my best understanding of your intent. As I pointed out in this post, you do some of the same things I do, that you complain about when I do them. I don't believe this is intentional on your part; you're doing the same thing I am. Also, it appears to me now that neither of us is particularly good at gleaning the intended meaning from what the other writes.

[The courts are part of the legal structure. The possibility of disagreements has been anticipated -- that's why there are appellate courts and a supreme court. It's their decision, not that of the federal courts.]

Now here is where I can't help feeling exasperated. You say (1) You would follow the law; (2) which is ambiguous, the courts disagree as to its meaning; but (3) there are procedures for clarifying the meaning. Yes, fine. We are currently learning that one court says Harris has the discretion the legislature gave her, and another one says she does not. We learn that one court feels a legislated deadline is THE deadline, and another feels that the legislature's deadline isn't the "right" one, THEIRS is the right one. Do we have a decision yet about dimples?

Now, in the absence of decisions from these appellate courts, what does the law mean? Opposing lawyers are arguing for different meanings. OK, if you were one of these lawyers, which side would you choose to represent? You have taken a side in this case before these decisions have been handed down, yet you claim your side is the "right" one by law!

You are the most frustrating person I've tried to talk to yet. I never know how to phrase my questions so that you will answer what I meant to ask. Let me try again. If Bush's lawyers win the case, would you "flip sides" because the CURRENT judicial definition has flipped?

[What, then, do you think would be good law in this case?]

"Count 'em all, or none at all". Remember Glaeser's original title? That's what I think a good law would say.

[I mind whining because one side failed to foresee the folly of its inaction.]

But that's not what I think is happening here. Perhaps Bush made a dumb decision, which would not surprise me. Perhaps he made a smart one. We don't know yet. But I am opposed to PARTIAL recounts, by either side, especially according to flexible rules.

And forgive typos in these posts. I'm at someone else's computer, and this keyboard is terrible.

-- Flint (flintc@mindspring.com), November 22, 2000.


Unk:

[Someone who I am not speaking to...]

No no no. You mean, "someone to whom I am not speaking"

[One, no deadline.

Two, a hand recount MUST take place across the entire state even if problems are shown to exist in only ONE SINGLE polling place, maybe if even only ONE SINGLE district recounts by hand due to mechanical failure, the entire state must do a hand recount.

I am not sure that either one is a better solution than what the law already says now.]

I must agree with No Spam that the regulations controlling this process could and should have been much more clearly defined. At least, this is my best understanding of his position, which I may have misunderstood despite a good faith effort to understand it.

As for recounts, not all elections cover the entire state, or even county. Recounts might happen only within the relevant jurisdiction, but should probably cover the ENTIRE jursidiction when they happen. But at the very least, we should try to avoid making up the rules AFTER they become necessary, *especially* in circumstances where the precise wordings can make a critical difference in the single most important election we have.

-- Flint (flintc@mindspring.com), November 22, 2000.


>> [Unk] but no matter how thin you slice it "if you're drunk or high, maybe you ought to sober up before posting again in this thread." implies drunkeness.

> [NSP] No, it implies the _possibility_ of drunkenness.

Actually, it implies a state (Florida?) that could be mistaken for drunkenness.

-- David L (bumpkin@dnet.net), November 22, 2000.


Brian,

Would you please tell Flint that I have decided to...oh...crap, you're aren't talking to him either. Let me see, David L, are you still talking to Flint? If you are, would you tell him that I have redoubled my effort to refuse talking to him now that he has corrected my english?

Also, while you are at it, tell him that I have painted a white line down the middle of the forum, and he is to stay on his side of the forum, and I will stay on my side of the forum. However, since I am a reasonable guy I have set the coffeemaker in the middle of the white line, and we can both still use it, but he must make coffee on the hour only, and I will make coffee on the half-hour.

Thank you.

-- Uncle Deedah (unkeed@yahoo.com), November 22, 2000.


ROTFL! Unk, this thread is beginning to feel like a hybrid of Get Smart and The Munsters.

With regard to slavish adherence to overrated grammatical rules, I am reminded of the Churchill quote, "That is the sort of English up with which I shall not put."

How does one discern the locus of a white line from the white background against which it is superimposed. (Wouldn't Flint have said something like that, that is, if he were on speaking terms with you.)

-- David L (bumpkin@dnet.net), November 22, 2000.


Cone of Silence, anyone?

(I miss that stupid show.)

-- Patricia (PatriciaS@lasvegas.com), November 22, 2000.


David L:

Hey, I'm on speaking terms with everyone. I think whoever is patient enough will eventually either reach agreement on something, or reach an agreement to disagree. Brian sounds like a bridge player at a poker game, aghast that bluffing is allowed. Why, the *nerve* of someone to bluff and take a pot with a weaker hand! Hey everyone, DON'T PLAY WITH HIM! HE BLUFFS!

Well, nobody can bluff all the time. But *never* bluffing is poor poker. Brian might prefer that his own rules be followed, and if so, good for him. But we all have fun differently, not all by my rules nor by Brian's. I'll talk to anyone.

-- Flint (flintc@mindspring.com), November 22, 2000.


Patricia,

May we infer that you miss it by that much.

-- David L (bumpkin@dnet.net), November 23, 2000.


Yes but it sets a precedent, and not one that would be good for America as a whole.

-- James Lockabrie (jlockabrie@lakecamajan.net), November 23, 2000.

Folks,

I'm switching to the continuation thread Continuation of thread on that op-ed piece from Eve. at http://hv.greenspun.com/bboard/q-and-a-fetch-msg.tcl?msg_id=004707 for posting my future responses.

-- No Spam Please (nos_pam_please@hotmail.com), November 24, 2000.


Moderation questions? read the FAQ