Can anyone explain the math to me that is used in the following Y2KNewswire article?

Y2K: A systematic problem

Thanks, it's been a long time since I was studying math in school.....

-- Jon Williamson (jwilliamson003@sprintmail.com), July 12, 1999

Answers

Jon, I assume you're talking of the calculations such as:

The answer is (1-(1-2.5%)^100), or a 92% chance of failure

The basis is simplified Probability & Statistics. In essence, it is calculating the probability that nothing will fail.

This "math" has been used at various times by Infomagic, Loserwire previously, and others. The underlying assumption is that, if any one thing fails, the whole fails. While perhaps valid in some instances, applying it as a rule is false.

I wrote a more extensive answer over on csy2k, last winter:

Link to Dejanews

-- Hoffmeister (hoff_meister@my-deja.com), July 12, 1999.

Jon, thanks for the link, here is a snip I found interesting:

""In some cases, we've been told the Y2K lying rate is as high as 50 percent," said Tom McGurk, who is also secretary of the Department of Management Services.

BellSouth Corp., for example, told state officials that more than half of the compliance claims by its vendors turned out to be incorrect when they were independently tested, he said."

The lying starts at the TOP of the political power heap and woks its way down to the lowest echelons of business and industry.

Ray

-- Ray (ray@totacc.com), July 12, 1999.

Hoffmeister:

Thanks, I knew it was related to probability theory. I did not make myself clear. It was the operation that the "^" symbol referred to that I was unsure of.

In this case, of course, it depends on whether by fail the author means truly "shut down", catastrophic failure, or a silly date error on a report.

In the case of this article, it was made VERY clear that they meant fail outright, as shown by the quote below:

(snip)

Out of the 25% that are non-compliant, suppose that 10% of those fail outright, while the other 90% continue to operate, but churn out bad data. So we have 2.5% of all the computers failing (in this scenario), with 22.5% churning out bad data in one form or another. (2.5% is 10% of the 25%.)

They took NO account of their defined 22.5% of systems producing some degree of bad data, to be caught more or less later.

So, you don't have to agree with their chosen percentages, but they made it very clear what the percentages were supposed to represent.

-- Jon Williamson (jwilliamson003@sprintmail.com), July 12, 1999.

2.5% chance of failure is 97.5% of success (1-.025)=.975
10 in a chain
.975 * .975 * .975 ... [10 times] = .757 chance of success
1-.757 = .243 = 24.3% chance of failure

-- A (A@AisA.com), July 12, 1999.

It's not the math, it's the assumptions. There should be a companion volume to "How To Lie With Statistics" entitled "How To Speculate Wildly With Probability Theory".

-- Trollyanna (r@t.com), July 12, 1999.

Hoff: don't continue to be an idiot. The issue has never been "if one thing fails, the system fails". The issue has ALWAYS been, "will enough fail to cause the following:"

A. no significant problems
B. recession
C. depression
D. total collapse

A side issue has ALWAYS been "Do you define TEOTWAWKI as B, C or D?"

-- a (a@a.a), July 12, 1999.

the "^" means "power of" -- in the case above, multiply the .975 (or 1-2.5%) by itself 10 times.

The point is, that while any system by itself may have a small prob of failure (large prob of success) -- string a bunch together (create dependencies) and you'll be surprised how fast the prob of success decreases for the overall (combined) system.

-- A (A@AisA.com), July 12, 1999.

A:

It is truly amazing how knowledge "reappears" once someone explains it to you again. (Remember Mark Twain's "Prince and the Pauper"??

Also, to note that this article did not deal at all with catastrophic failures within the companies from their own software, only from vendors of undefined products.

Hoff, no matter how you parse it, the odds of more damage than I'm comfortable with next year are higher than I like. NOT an Infomagic, but I've gotten soft and lazy in my middle age.

-- Jon Williamson (jwilliamson003@sprintmail.com), July 12, 1999.

Yes, but once again, the underlying assumption is that a failure in any one system causes the whole to fail.

With this assumption, the more complex you make the system, the more likely it is to fail. The more components you add, etc.

Again, in some limited circumstances, this assumption is probably valid. But not across the board, by any means. And when you raise the number of components required to bring the system down, as is the case in most instances, the probabilities drop dramatically.

-- Hoffmeister (hoff_meister@my-deja.com), July 12, 1999.

Regarding the % who are lying: I have found that vendors are also likely to lie to say that their systems are non-compliant. They have an interest in scaring us into purchasing new systems to replace supposedly non-compliant ones. I have done some of my own testing, following vendor guidelines as well as my own knowledge the products, and have discovered that the alledged non-compliance was not there. Or in some cases the non-compliance was of an insignificant nature.

-- elaine lyons (elainelyons@netscape.com), July 12, 1999.

Hoff, where is your information to back your statement:

"And when you raise the number of components required to bring the system down, as is the case in most instances, the probabilities drop dramatically."

First, you are unilaterally declaring that the number of components failing needed to bring down "the system" must be raised. Second, you state this is the case in most instances.

Upon what facts do you base these two statements? Or are these your opinions, based on a common sense appreciation of how the world works?

Please understand, I am neither criticizing nor arguing, just trying to define the positions. You ask the "doomers" to provide historical facts to back up their concerns about a unique event to occur in 172 days. Do you know the true reasons behind every business failure in the last decade? Can you prove that none of them were due to systems failures initiated by the loss of one critical system at a critical time.

If not, then I propose that my concerns are based on as valid a basis as your optimism. You are most welcome to said optimism, I'll stick to the pessimism I'm used to.

-- Jon Williamson (jwilliamson003@sprintmail.com), July 12, 1999.

I have to agree with Hoff to some extent. Redundancies and parallelisms can mitigate failure even in systems with components of high failure rates (e.g., human and other animal bodies).

But, face it, our systems right now, without Y2K problems, are running on the ragged edge. Add some additional Y2K stress and -- I think -- blooie!

-- A (A@AisA.com), July 12, 1999.

Sir Hoffmeister - you're not really being fair to the guy asking the question. To understand, he'll need a better explananion. See if this is more clear.

They are talking about probability of a system remaining in operation if various parts of it fail.

generally, for any system to work properly, a large number of things have to all be working at the same time. For example, to run your household heater in the winter: you MUST have: 120v AC power to the house to run the blower motor, the high temp sensor in the heater being "set" and reading correctly = not being sensing "too hot", 12v/24v power to the little thermostat on the wall, natural gas pressure to the burner, the pilot light being "on", the sensor to the pilot light working, and the thermostat being set to "heat" and not "cool" and not "off".

This is a single system - it needs 7 things correct. If any fail "off" (wrong position) the furnace will not work. Also, of course, the thermostat mus t"sense" a cold house - but we will assume the house is cold - and you want heat. All of these are reliable - but let's assume that they are only 99% reliable. The probablity then that all will operate correctly is .99 x .99 x .99 x .99 x .99 x .99 x .99 = .93 or 93.2% reliable for the system overall.

Let's assume that your thermostat has 75% chance of failing next year, and your power company has a 15% chance of failing. Their individual reliablilites are .25 and .85 so the accumulated reliability of the heater next January is .99 x. 99 x .99 x .99 x .99 x .25 x .85 = .202 Thus, you now only have a 20% chance of your heater working. Now if you test your thermostat, and find out that it will definitely fail - your reliability next January is 0.00 - you will never get heat until you fix the thermostat.

If you replace the thermostat with a y2k compliant thermostat, now your overall system reliability goes back up to .99 x .99 x .99 x .99 x .99 x .99 x .85 = .800 or 80% chance of getting hot air. The most likely failure mode is getting power to the house, but other things could still fail: mechanically or electronically. The new thermostat could even fail!

More typically: the real ratios are like .99999 x .9975 x .95 x .9999 etc, but I hope you get the point. Each different thing has a different reliability and a different affect if it fails.

Obviously, right now, MOST things operate correctly MOST of the time, so the world keeps going around with today's modern world of artificial power, water, heat, and light. (Only death and taxes, I've been told, are absolutely reliable).

Anyway, this is where the equation comes from: and there are many variations. The number of systems that must operate correctly all at once for the whole process to operate properly can be determined by multiplying their reliabilities. If there are alternate paths (other ways to get the job done), the equation gets more difficult, but is still calculable.

The fundemental problem with this theorectical approach come from many assumptions: what is the system exactly, did the user include all possible work-arounds, did he include all required operating systems (or did he skip a few?), what are the actual reliablilties (now, and under the stress of year 2000-induced failures?), what are the alternates under emergency and under normal conditions? If soemthing fails - does the whole process fail, or is it merely reduced? If reduced, what is the effect?

If this process works, will others fail that are equally important? More important? Not important at all? Arguing reliablility is a nice way to discuss the world and appear knowlegeable - but my concern is that it may be as important only as the number of angels falling off of the head of a pin.

If we could agree to the number on the pin in the first place, and how many are required to remain on the pin after year 2000, then we can argue about how many might fall off if the pin is used by a left-handed seamtress.

-- Robert A Cook, PE (Kennesaw, GA) (cook.r@csaatl.com), July 12, 1999.

Hoff said:

And when you raise the number of components required to bring the system down, as is the case in most instances, the probabilities drop dramatically.

uh, you mean like when

- All rail activity is coordinated from a central facility in Jacksonville?

- The entire power grid is coordinated by an unprotected, unencrypted leased line?

- One company produces the seeds for the worlds harvests?

Your argument reminds me of the old timers where I work who warned us that they still needed knobs on their gizmos. We told them "the software IS the knob now", and we replaced all their instruments with boxes with smooth panels and blinking lights. The problem is, if the software doesn't work, they can't use the instruments manually any longer.

The way I see it, the machinery that drives world is more like a GM plant that shuts down if two key suppliers strike, than it is like the Internet, where the system automatically reroutes around failures.

-- a (a@a.a), July 12, 1999.

So the conclusion from this math is that the probability of at least one failure (the compliment of all variables having success) is close to one (or a certainty). Well duh! Taking the logic from the other direction, we find that the probability that all systems will fail is extremely small. (.025 taken to the nth power - zero). Of course we know things will fail but how that translates to a 1 or a 10 on the scale is anyone's guess. The question is how many failures can we live with. Can we live with exactly two or more failures? Can we live with exactly three or more failures? The binomial is a better prediction. I believe that's what Hoff wrote in his article.

-- Maria (anon@ymous.com), July 12, 1999.

Hoff:

Actually, I agree with you that most things have work arounds to some greater or lesser degree. But you see that the demands for "evidence" can run in both directions.......

Robert, thanks for the clarification. College calculus and such was an awful long time ago.

-- Jon Williamson (jwilliamson003@sprintmail.com), July 12, 1999.

A@a.a - the case you are describing is exactly what is happening => 1) the number of alternate paths (to a successful process) are dropping as things become centralized.

2) that single process - which MUST work for the system to sustain itself - is increasingly vunerable to a single-point failure mode (the computer MUST work)

3) that single point failure mode is in itself more vunerable to multiple single point failures "upstream" - thus, the one computer that MUST run to operate the warehouse MUST have reliable power through the whole shift so load instructions can be passed to the trucks. It MUST have reliable power overnight to get routine oprder from the different stores who are 'building" the orders based on cash register processing that is on-line through the day.

A four hour power failure in one city thus affects orders being shipped the next day from another city - assuming the second computer is correctly programmed and has a valid oeprating system in the first place. Assuming the trucking company has gas available. Assuming the automated forklifts in the warehouse are properly programmed and have compliant oeprating systems.......

-- Robert A Cook, PE (Kennesaw, GA) (cook.r@csaatl.com), July 12, 1999.

Folks, I'm going home for the night. I might pick up on this from the house later.

Everybody, (and I mean optimists and pessimists alike), thanks for a clean, non flaming thread so far. Can we keep it up?

Although it is theoretical and we will NEVER have enough data to put into the equation to make it foolproof, this is really the heart of the entire issue.

Hoff, Maria, I most fervently hope you are entirely correct.

-- Jon Williamson (jwilliamson003@sprintmail.com), July 12, 1999.

Jon,

I appreciate you asking a question about the math. While I think Hoff has answered the math question, if I might point out a few issues. First, Alan Greenspan is wrong about the requirement for 100% of computers to "work." Computer hardware and software fail every day. This should give you your first hint about the utility of "go, no-go" probability applied to a "wicked" problem.

Let's move past the hypothesis that 50% of compliance claims are false. (And all of the assumptions based on this.) If 2.5% of computer systems (whatever that means) fail, what makes this number interesting? Is is above the "normal" rate of failure? Well, I'd like one of the resident techs to provide a normal rate of failure.... Now, of "normal failures" how many are resolved without human intervention? There are often "autofix" programs running. Of those programs requiring human attention, how many are critical? How many are fixed within an hour? a day?

Unlike a computer virus, if a computer system fails on rollover we have a pretty good idea of what's going on. (Although I feel for the inevitable techs who happen to have a systems failure unrelated to Y2K on Jan. 1st and spend countless hours looking for the wrong problem.) Does knowing where the problem is save time? Probably. Will some problems be readily "fixable?" Probably. Will some companies have contingency plans? Probably. Can most problems be fixed in a week or two? Maybe.

Aside from the math, the article simply restates the old "want of a nail" argument. The economy does not run like a tightly written computer program (if such a thing exists anymore). It is a large, messy affair with missed deadlines, strikes, outages, mistakes, glitches, etc.

Like the flight of bumblebee, it doesn't look as if it ought to work on paper... but it does. Oh, here's a cut from one of my earlier essays:

"4. The False Positive Y2K Statistical Analysis

"Go, No-Go" analysis leads to false statistical results when applied to macroeconomic phenomena. Social and economic activity is not a coin toss. Computer or embedded chip failures reduce efficiency. In extreme cases, efficiency can be reduced to zero, but this is an exception rather than a rule. Some level of activity will continue in all economic sectors, at least with some level of reduced efficiency.

Bad things can happen. There is a reasonably high probability the earth will be struck by an interstellar object. Of course, it may take another billion years. I am not losing sleep over this, even though it is a "non-zero probability."

Much of the Y2K writing on the Internet has been dedicated to the proposition that one failure will domino into a systemic failure. The rate of failure argument is logically flawed. Take a complex set of devices like those in the Apollo space program (1960s technology). Based on the incredible number of components and the mean time between failure, on paper the Apollo missions never make it off the pad. This is the danger of misapplied statistical analysis. While Y2K will cause some failures, very few are mission critical for the survival of civilization. (Do you think wed even notice if some Federal departments did not operate?)

Some problems will be fixed ahead of time, some will be fixed on failure and some will just not make much difference. No one can produce an accurate simulation to determine the number of computer failures that result in people looting the local thrifty mart."

Regards,

-- Mr. Decker (kcdecker@worldnet.att.net), July 12, 1999.

Maria - absolutely not!

<>

If a system is properly remediated, it reliability will remain near 100% - let us assume if something is properly remediated and properly tested it is as likely to remain operable as it is now: 99.999 reliable, or 1 defect in 10,000 products. (if you have a terrible system, perhaps one defect in 1000 or worse.

What we are concerned about is the reverse: if a system is not remediated, will it fail? If not remediated, it is a near certainty to fail!

If remediated and not thoroughly tested, perhaps the success rate will be only 50%, 75%, or even 90% - we don't know. (And apparently, judging from the repeated failures in sewage system testing, even if a system is remediated, its failure rate may be high!)

Lastly, if remediated and if properly tested, will the "support" systems needed also work? Those assumptions and analysis is not properly a binomial, because as Hoff pointed out, there can be partial work-arounds and temporary fixes.

You have to append everything with too many cautions to be useful - other thatn as A@A said, there is a reduced number of alternatives available now to bypass trouble. And computers require more accuate data than people.

Smart, sensible people who can think on their own, people other than bureacrats, anyway.

-- Robert A Cook, PE (Kennesaw, GA) (cook.r@csaatl.com), July 12, 1999.

Jon:

My statement is based on experience. That is, I've seen numerous instances of systems identified as "mission-critical" failing, abending, etc. None of these have caused the business to fail.

The fact that even after exhaustive searches via csy2k, few if any business failures caused by system failures have been uncovered further bolsters this point. I even contributed one possibility, though an SAP implementation usually encompasses many of what previously would be considered "mission-critical" systems.

The resulting probability is a matter of mathematical fact. The link to the Deja posting explains the math in more detail.

a

Please, 'a', try to read before exposing yourself. The LoserWire story is calculating exactly the probability that nothing fails, assuming that if one component fails, the whole fails.

Robert

You are in essence repeating the equation, using different probabilities for different components. I acknowledged that in certain circumstances, all components must function. But especially in complex systems, redundancies and work-arounds usually exist, or can be utilized, that keep a single source of failure from collapsing a system. You are correct, the calculations become more complex. I tried to demonstrate the somewhat more complex in the Deja post.

-- Hoffmeister (hoff_meister@my-deja.com), July 12, 1999.

Robert, give me a break! Do you know what a binomial is? Lets try this example: the probability that any one of these instances occuring is .01: you get out of bed in the morning and twist your ankle, you find that your toothbrush bristles broke off in your mouth and you choke on them, your hot water heater went beserk and you burned yourself, you spell hot coffee on your hand and receive second degree burns, you start a grease fire in the kitchen while cooking bacon, you got into a car accident on your way to work, you missed a step going up to your office and fell down a flight of stairs, you received an electrical shock from faulty wiring on your PC, your co- worker finds you annoying and decided to shoot you. So whats the probability of failure here, 1-.99 to the 9 or 9.5%. The list could go on and on but we still make it out of bed in the morning.

The bottom line: the more dependencies you put into the equation, the more chances you'll find for failure (of course!). Based on that, nothing can succeed but somehow we manage. Mathematical models are great (I know I use them everyday in work) but applying a model to Y2K if fruitless; it just can't be done.

-- Maria (anon@ymous.com), July 12, 1999.

BTW I want to add that stochastic independent variables give the same results as for dependent variables for large samples. Thus the normal distributrion can be used as an approximation to the binomial.

-- Maria (anon@ymous.com), July 12, 1999.

If all mission-critical systems work at 100% (ie, as well as they did before), an impossibility given the nature of software maintenance and the reality that < 80% will be remediated in U.S., less worldwide, what is the impact of the far greater number of "non mission-critical" systems that were completely unremediated but many of which are data input-output requirements for the mission-critical systems?

A few companies have remediated all systems, but the vast majority are explicitly only remediating "mission-critical systems" (there is no standard for defining them, so one company in an industry may remediate an "application type" that is not judged mission-critical by another company in the same industry).

It is emphatically not a case of one failure causing bizarre domino effects but of a web of failures and PRIMARILY OF NON-MISSION CRITICAL systems and applications, IMO. For these reasons:

1. Some unknown percentage of them (5%? 25%) will turn out to have been mission-critical enough (required for business to operate) and will require weeks, months or more to remediate, depending on size and breakdown type.

2. Too many of them (not all) will corrupt mission-critical systems and/or force abends or system halt. We don't know how long it will take to fix and whether or not an entire non-mission critical app will need to be remediated or not, a new application written, work-arounds, etc.

These effects will be propagated to an unknown degree across industries (again, due to the lack of standard in defining what to remediate) and application types (banking, utilities, etc).

Overlooking the occasional possibility that remediated systems will be somewhat more efficient than previously (given the scut work, who is assigned to work on Y2K, etc., that will be minimal), the world post-Y2K CANNOT work as efficiently as before for 'x' period of time. Robert Cook has this nailed better than anyone I have ever read.

Y2K is utterly unique from an IT perspective, not on an intra-application basis (that's tedious but mind-numbingly simple) but on a "world info architecture" basis. Extrapolating from previous statistics (which are just one step beyond meaningless, given that programming is at the same cultural stage as law and prostitution) won't help.

Infomagic is unlikely but a hell of a lot more likely than BITR or meteor impacts. The latter is just total ignorance speaking.

People still fail to distinguish between what we can be optimistic or pessimistic about. From the point of view of remediation, it's over. We lost. From the point of view of human response after rollover, we don't know. I'm hoping for tremendous human flexibility, courage and ingenuity and doing my best to organize for that.

It's hard watching people whose technical knowledge I do respect (Hoffmeister) unwittingly (I hope) create false hope about remediation, but that's life. If it's that hard for people in the industry to understand the reality, it's no wonder that non-IT folks on this forum are misled.

-- BigDog (BigDog@duffer.com), July 12, 1999.

But Miss Maria - I agree with you. Applying percentages based on current knowlege to predict future y2k-induced failure modes is no more accuate than arguing about the number of angels falling off of a pin.

Reliability forecasts are only valid given data or good assumptions. In certain cases, the "guesses" we have been arbitrairily using can be replaced with real numbers, and thus false conclusions eliminated.

For example - one could argue that since I replace my toothbrush every six months (thus eliminating life-cycle fatigue failures), and since no toothbrush has failed in the past 45-odd years, the true probability of failure is 1/(45*365) = 6.088 x 10^-5 or a reliability of .9999804

However, I have had 2 accidents on the way to work in ten years: or on a per-day basis, 2 failures/(10*300 working days/year) = .000666 failure rate. (.999334 reliability.) Your equation and your assumption are perfectly correct - only the assumed reliablities are slightly off. Let us hope the probability of a co-worker shooting me is slightly lower than 10%!

Rather - let us hope that the probability of the co-worker hitting me are much less than 10%. A miss doesn't matter.

Baes on four known failures in four attempts at installing the new FAA y2k-compliant air traffic control system, I'd say the reliability of it working next year (unless they fix something) is 0.00

reliability then (for

-- Robert A Cook, PE (Kennesaw, GA) (cook.r@csaatl.com), July 12, 1999.

BigDog It is emphatically not a case of one failure causing bizarre domino effects but of a web of failures and PRIMARILY OF NON-MISSION CRITICAL systems and applications, IMO....

Thank you for making exactly the point I tried to make on this thread; that assumption was the underlying assumption of the LoserWire story.

Infomagic has always been bizarre sci-fi to me. The one possibility of it occurring was a long-term loss of power, which is why I've focussed so much on the utilities. To my mind at least, it is no longer within the realm of possibility, except perhaps alongside meteors, etc.

I'll agree with the loss of efficiency, and think a recession, Y2k or not, is inevitable. But as you like to say, "It's still Y2k", and the question is just what are you preparing for? Or again, just how "do" you prepare for a recession?

-- Hoffmeister (hoff_meister@my-deja.com), July 12, 1999.

Good post BD. Hoff, Decker and Maria seem to be having trouble visualizing that we are talking multiple failures, occurring in parallel. This is unprecedented in the entire history of the world. And, I hate to sound like a broken record, but it will unfold amidst all of the other problem we are likely to incur during the same period.

-- a (a@a.a), July 12, 1999.

Hoff -- I didn't stop to read the article, so I can't argue to whether it is "one things breaks all" or not. IMO, "infomagic" is not just a function of utilities, though their successful operation is crucial, of course. We'll just differ on that. As you know, I would consider a recession a wonderful result.

Most of my preps are tied to "weirdnesses" that might result from supply chain breakdowns. Obviously, one can't prepare for everything/anything but if I had a diabetic in the family, I might want to be prepared for a recession in which insulin was unavailable for the amount of time that would have me die but most other things were. And so it goes.

-- BigDog (BigDog@duffer.com), July 12, 1999.

Here Hoff:

A. no significant problems (Poole)
    B. recession (Decker, Hoff)
    C. depression (Yourdon, 'a', BigDog)
    D. total collapse (Milne, North)
E. devolutionary spiral (Infomagic)

OK, lets throw out those with 1% or less chance

B. recession (Decker, Hoff)
    C. depression (Yourdon, 'a', BigDog)
    D. total collapse (Milne, North)

As someone said on another thread, compared to 99% of the population, you and I are essentially on the same sheet of music. :)

-- a (a@a.a), July 12, 1999.

Hoff: Your first name isn't Jack, is it????

-- George Leroy Tirebiter (lifeon@theTitanic.com), July 12, 1999.

Hoff,

Don't ask BD about a recession! He might have to refer you to my posts. (laughter) C'mon, Hoff. Preparing for a recession is boring. Pay down your debts, have a large pool of savings, develop a modest lifestyle, learn the keys to frugal living, have flexible job skills and a positive attitude. Where's the cool stuff like preparation "gear."

Regards,

-- Mr. Decker (kcdecker@worldnet.att.net), July 12, 1999.

Great reading. Thanks all.

-- Mike Lang (webflier@erols.com), July 12, 1999.

Decker:

1972: World plunged into serious RECESSION because a few rich oilmen think they should be getting more money for a barrel of oil. Effects last for years.

2000: World is beset by severe shortages in many industries [CIA]. Ripple effect from Y2K computer problems causes cascading, long lasting failures in many systems. [Gartner Group] Severe power outages occur throughout the year. [Koskinen, Tava/Beck/European News Agency] Stock bubble bursts. [Yardeni] World currency system is strained to breaking point. [Soros] Nuclear, biological, chemical and/or cyber terrorism is unleashed. [CIA] Solar and other environmental effects further exacerbate the situation. [NASA]

Business as usual? (laughter)

-- a (a@a.a), July 12, 1999.

"a,"

When I "retire" to the academic life, I'm saving a seat for you in my Econ 101 class... front row.

Regards,

-- Mr. Decker (kcdecker@worldnet.att.net), July 12, 1999.

Hoff,

you are both very knowledgeable and also very articulate. So it is a sheer challenge to prove you wrong. Anyway, I'll try these first shots off the top of my head. Warning: this means that I may dare to try even harder later! (if others on this forum help me!!)

Reading through your posts a critical concept transpires: you are apparently convinced that fail-safe systems exist. But what we have called "fail-safe" systems are systems whose probability of failing is so low that we decided, for practical purposes, to call them that way. But they are NOT. Otherwise Three Mile Island-type accidents or just plain regular modern aircraft accidents would NEVER-EVER happen. And they do. The reasons are:

(1) The human factor, human reactions, human fear, human FAILURE, etc., which in y2k might be triggered easily to encompass tons of new, unprecedented things simply because y2k has no precedents Hoff, and y2k circumstances are very different from anything known. I've seen people react pretty bizarre not with outright failure but just with loss of productivity in their equipment simply because they are used to higher performance. Bank runs, mediacl device failures, looting, gas shortage...(hope not, but...)

(2) Redundancy?: Yes Hoff, redundancy is there allright. But not everywhere. It ends up and boils down into nodes, as Critical Path Method (CPM) or PERT diagrams would clearly show. Nodes are MUSTS. You can design and build the most expensive house on the surface of this planet, but if doors are not delivered, for all intent and purposes it can't be called a 'house' simply because nobody can live in it. Gartner Group assigns vital, mission critical failures on this basis.

(3) Incompatibilities: Interfaces, no international standards, different 'windowing' solutions, etc., etc.

Furthermore, y2k in more than one sense will mean a simultaneous, international test of world wide technology, with no attempt to remediate and test what have been called "non-critical systems', which supply data to mission critical systems! No engineer or CIO wants that you know, because it necessarily spells trouble. Y2K trouble.

Also, there is no way to logistically foresee reasonably reassuring Fix On Failure programs under Y2K circumstances. Think of it.

Y2K is a change of paradigm Hoff, and anytime that you try to apply known philosophies to unknown phenomena, the model doesn't work right.

Recession? Yes, if we are very very lucky

Depression? Yes, for sure, plus beware 'cause there are more than a couple of wild cards in the deck which might mean far worse depending where you are located.

-- George (jvilches@sminter.com.ar), July 12, 1999.

"Y2K is a change of paradigm"

Says who? please qualify that statement

-- just (asking@to.beclear), July 12, 1999.

Just,

I say so, because of what is explained in points (2) and (3) above, plus the two subsequent paragraphs, which are things we always counted on having and have never experienced the lack of before. That is all spanking new in the history of engineering, technology, human resources, management, you name it Just. That's why IMHO it is a change in paradigm.

By the way Just, your comment has triggered off something I forgot to mention in relation to the myth that computers already break down all the time, JUST LIKE IN Y2K. Ooohhh no you don't !!

'Cause although computers DO break down "all the time" nowadays, they screw up DESPITE the fact that they are 99.99% bug free thanks to thorough testing and use sometimes after decades of de-bugging. That means that today's computers contain only o.01% of all possible bugs. (you can change the percentage points but not the concept). But under current y2k international circumstances (including the US problem of its sheer size), computer systems will have more like 10% of all possible bugs still inside them, which is THOUSANDS of times higher bug content than today. Not all showstoppers, but many troublesome (big trouble) and some Kaputten. The troublesome (under y2k circumstances) are also far worse than what they would be today, don't forget, because of points (1) (2) and (3) above.

Take care

-- George (jvilches@sminter.com.ar), July 12, 1999.

I believe the technical name for the equation you are currently involved in is called Circular Manual Ejaculation Theory first put forth by Ricardo Cabesa and Phil McCracken.

C'mon guys this shit's a waste of time and brainpower. Do you really think you're gonna come up with a logical failure rate from a problem this complex?

Get a grip. You couldn't even count the variables. That's why we're in this situation in the first place.

-- Simple Simon (logic@doesntfithere.com), July 12, 1999.

I sure do wish people would stop saying that non-critical unremediated systems will FEED critical remediated systems. Any systems analyst worth their salt realizes that a critical system BY DEFINITION includes all systems that feed it. Do you REALLY believe that mission-critical systems were defined by throwing a dart or selecting every 4th system, or eenie-meenie-m LOL...how do you spell that?

-- Anita (spoonera@msn.com), July 12, 1999.

I'll post this book again, I encourage everyone on this forum to read it. It will give you a fresh perspective on Y2K and may alter your position (both doomers and pollies)about the potential for successful HUMAN respones to multiple (common mode failure) Y2K induced problems.

The Logic of Failure: Recognizing and Avoiding Failure in Complex Situations by Deitrich Dorner.

It relates to Y2K and the potential HUMAN responses to cascading failures. Failures that originally stemmed from a Y2K glitch that a human incorrectly responds which now introduces a new problem, etc., etc.

It is about how 95% of people have very poor cognitive abilities (i.e. understanding cause and effect relationships over time) and the studies that show this to be the case. TMI was one of the examples.

VERY INTERESTING READING!!!

-- MarktheFart (quke@ix.netcom.com), July 12, 1999.

Anita, you're as wrong as wrong can be. The federal government classified around 6600 systems as mission critical, and there are damn sure plenty of other systems that feed into those 6600 that can affect their proper operation.

How can you be so stupid Anita? I mean seriously. I can't believe you posted that crap.

-- a (a@a.a), July 12, 1999.

Anita -- I am also flabbergasted. In real life, orgs are triaging like crazy. Sure, some critical enterprise systems (you know, dearie, the ones with > 10 languages, five primary databases and 5M lines of code) have had all/some of their data inputs/outputs mapped over the past thirty years but NOT FOR DATE CALCULATIONS. No one cared.

So, in remediation (I'm oversimplifying), you have three choices:

1. Diagnose date feeds from non-mission critical systems and remediate the portions (?) of those systems that affect the mission-critical application. How you do that without basically remediating the non-mission critical systems is beyond me. And people are simply not doing this.

2. Trap for date feeds from non-mission critical systems and block input. This assumes you know you don't need what is being supplied by the other system, of course.

3. Trap for date feeds and supply the needed data wrapped with the bad inputs from the non-mission critical systems in some other way (new app, new routine that bridges the two systems, etc).

Is some of this being done? I'm sure. But if you think systems analysts have been sitting around plotting DATE inputs from non-mission critical systems, you're smoking something. Most of them have been desperately trying to figure out how to distinguish the date thingies in the systems that HAVE BEEN defined as mission-critical, so the programmer guys can fix them.

-- BigDog (BigDog@duffer.com), July 12, 1999.

Quick scan down the thread reveals NOT A T***L IN SIGHT!

This'll be fun to read when we get back from dinner, but too tempting to start in on right now (wife eruption threatens.)

Something about intelligent exchanges acting as T***LBANE? (Perhaps a few entered herein and suffered a little-brain implosion? We can only hope.)

Of course we're only trying to rough in the outline of our mental pictures of events to come -- detail CAN'T help but overwhelm; useful only as f'rinstances. Back later....

-- jor-el (jor-el@krypton.uni), July 12, 1999.

Anita,

Never forget...All systems are critical to someone's mission!!!

-- K. Stevens (kstevens@It's ALL going away in January.com), July 12, 1999.

George

First, no, I don't think we've built any completely fail-safe systems. I brought up redundancies to illustrate the error in the model used by LoserWire, and others.

From an IT perspective, Y2k is anything but an unknown phenomenon. It has been analyzed and dealt with in far greater detail and scope than any other system problem I'm aware of. Certain infrastructure failures, like electricity, could indeed pose extreme risk. Which is why I've spent so much effort attempting to analyze them.

I also think your estimates of error rates for Y2k is extremely overblown. See this thread for more discussion.

-- Hoffmeister (hoff_meister@my-deja.com), July 12, 1999.

'a' and "Bigdog":

I just don't buy it. The first thing we do once a system has been determined to be mission-critical is to scan the program libraries to determine what files/databases are used. Once that scan is complete, we determine what programs TOUCH those files/databases. Once THAT scan is complete, we eliminate programs and systems that are defunct. Have EITHER of you worked on remediation projects? I'm not saying that in a sarcastic way. You both seem to assume that Daffy Duck is at the helm here.

-- Anita (spoonera@msn.com), July 12, 1999.

Folks:

This has to have been one of the *politest* sets of disagreements we've had here in a LONG time. Thanks to you all.

Would I be wrong in comparing this issue to "modeling" of complex phenomena such as climate or weather?

I say this because, in both cases, the number of variables and the lack of definitive information are both mind numbing. As I understand it, as a side consequence of chaos theory, very few if any meteorologists today expect to ever extend truly accurate weather forecasting for more than about 7 days into the future.

This will be the ultimate non linear system folks. A nudge, a shortage, from a supplier not realized to have been critical, one who might have had a "breakthrough" in their remediation if they had had only another month, could send multiple industries into writhing knots.

A proper and intelligent decision by a creative, out of the box thinker could save a situation that would have been totally lost by a mediocre programmer.

Just plain good or bad luck, the "fortunes of war", will quite possibly be the deciding factor.

So, let slip the dogs of war, sacrifice to whatever gods you believe in, and tip your hat to Murphy, we need him on our side.

-- Jon Williamson (pssomerville@sprintmail.com), July 12, 1999.

Anita:

In an ideal world, with plenty of time, no conflicting demands upon limited resources, and NO unmovable deadline, I'm sure that the process you describe is followed.

In a remediation that is started in plenty of time, with adequate resources, source code documentation, etc., again I'm sure that is how it is done.

But in the case of too many companies that started too late, or for whom the effort was more of a token than not, or for whom critical decisions were made by accountants or attorneys (read "due diligence") rather than by qualified IT people, I don't believe that was always done properly.

Folks, our current "system" is very reactive. Has anyone here NOT worked somewhere that ran on the "Oh, S**t" method?? You know, they need it WHEN?!?! Panic, run out for materials, grab temps, forget profits, just get the job done to keep the customer. Or jam in the overtime to keep the (machine, computer, network, you name it) up and running. Produce the report, crank out those 30 special order parts.

Even at 99.9% fixed, it is normally pretty klunky. This is the real world we are dealing with, not idealized systems of any kind.

-- Jon Williamson (jwilliamson003@sprintmail.com), July 13, 1999.

Jon:

I've not seen ANY case, nor heard of any case wherein the decisions were made by any other than folks in IT. You ALSO must remember that we've been working on this problem for a LONG time, as many date problems were resolved when systems were changed for other reasons.

I elaborated on my experience in IT in my response to Zoobie and Bryce in this thread and fully expected to get flamed for it. I now realize (since a very small subset of my experience has received so many reactions in THIS thread) that the entire reason I wasn't flamed was that no one read that thread (or...[grin] they fell asleep before they got to the end.) I confess that I was in rant-mode that day, but I've just not SEEN the incompetence that the few in IT on this board have seen, nor has anyone else that I know seen this incompetence, save the sleeping Jacks that I mention in the linked thread. I'm aware of a few "last minute projects from Hell" mostly from folks I've met on the internet in the several years in which I've been exploring Y2k via the internet. My objection is to assume that this is the norm.

-- Anita (spoonera@msn.com), July 13, 1999.

Anita:

Impressive experience you have. I'm a recruiter and always like to have candidates of your apparent caliber.

Back to the question at hand: There are still so darn many subsets of "completion". My experiences have been different. I've had too darn many conversations with candidates who HAD been involved in "due diligence" remediation work. Go through the motions. With people whose word I trust implicitly, who are personal friends with directors of Y2K work at Fortune 500 companies, who say that they are by no means sure they can even run production lines next year.

Anita, I have no doubt that you have seen and heard as you describe. I do know that the inside information I have is in stark contrast to your's. One *possible* explanation could be the difference between companies who use contractors versus companies that are struggling to do the work with their own staff.

Anyway, bottom line, I'd much rather YOU be right than ME. **GRIN** Still, I have to trust my own instincts and my own information, as you must trust yours.

-- Jon Williamson (jwilliamson003@sprintmail.com), July 13, 1999.

Hoff,

Gotcha! Noooo siiiirr. You won4t convince many with the argument that (a) we4ve known about y2k all along (b) we4ve been working for a 4long time4 now

So what? Because, first of all we4ve started LATE, most are behind, and many haven4t even started like SMBs and the rest of the world pretty much, which we need badly.

The falacy is thinking that "knowing" about y2k is the same as solving it. Awareness helps, but it4s only 1% of the problem. Hoff, awareness is not a "good luck" charm that protects people and businesses from y2k4s impact. ACTION is needed, and at this late in the game it4s obvious that we have done a lousy job.

If you think that the US Federal government is behind, just wait till you find out how behind Customs, etc., etc., are elsewhere in the world. If this is the best you can argue in support of your expectations, IMHO I believe you are dead wrong.

-- George (jvilches@sminter.com.ar), July 13, 1999.

Jon:

I agree completely. It certainly serves no purpose for ME to extrapolate on MY experiences to include the entire world, just as it serves no purpose for YOU to extrapolate on YOUR experiences to include the entire world. I've mentioned previously on this forum that if I were on a last minute project from Hell, I would see things differently. My point in bringing this up at all is that it seems that ONLY the folks that are on the last-minute projects from Hell are frequenting Y2k fora....outside of myself, who is simply curious about the whole thing, having been a founder of the MSNBC Year 2000 Issues forum at one time after discussing the issue with folks on the MSNBC Technology forum. I left the Year 2000 Issues forum when I found myself the only remediator/founder left. The rest had left in disgust. The original intent of that forum was to share remediation experiences so that others could have a head-start. It quickly moved on to other discussions regarding Y2k, as non-IT folks wandered in.

There remain very FEW fora that consist completely of remediators, yet my curiosity gets the best of me on occasion, so I lurk here at the Yourdon-founded forum, as well as the Debunker forum, as well as Csy2k on usenet (which ALSO began with remediators sharing information.) I rarely lurk on the old MSNBC Year 2000 Issues board, as it rarely offers anything of consequence these days. I post infrequently EVERYWHERE I lurk if I have questions or concerns, or feel the need to clarify an issue.

-- Anita (spoonera@msn.com), July 13, 1999.

The math is fine; it's the assumptions and the model I question.

The real macroeconomic model is likely to be far more complex when it is studied by the future economists of 2020. Y2kneswire illustrates a good point on how interdependent we are, but their approach is too linear. It doesn't depict an economic web but rather an economic chain.

Assumptions they make:

1) "Downstream" failures or successes can't simultaneously be upstream ones.

2) Effects of redundant systems and good contingency planning are negligible.

3) There is no "friction" or "decay" in the propagation of the error through the supply chain.

4) There is no amplification of the errors as they cascade through the supply chain.

All in all, NOBODY KNOWS for sure how bad it's gonna get. (Most likely your level of suspicion in y2k repairs has something to do with your overall faith in the System before your moment of GI-Nirvanna.) To me, it seems like it COULD get bad--hell, it will probably at least quite ANNOYING. Why not start preparing for a "rainy day?"

-- coprolith (coprolith@rocketship.com), July 13, 1999.

*phew* Too much math, not enough common sense. Some very peculiar notions about systemic problems floating around...

Look at it this way: your heart probably makes up about 2.5% percent of your total body "system" (quick, some med student out there give me an "average %" stat). Anyway... your heart stops, you stop... 100%.

Of course a purely biological analogy isn't perfect, but it is definitely a helluva lot closer to realistic than the absurd notion that 97.5% of everything else in our highly interconnected world will go chugging merrily along when the 2.5% fails.

-- M.C. Hicks (mhicks@greenwich.com), July 13, 1999.

Anita:

yep, we all have ahold of different parts of the elephant. Doesn't it make sense in a way that those involved in the "Project from Hell" would be more likely to show up wondering what the heck is going on than those working on a well managed project. We all deal with more or less "self-selected" groups everyday.

Coprolith:

Excellent points. Their model is fairly simplistic. As I discussed above, however, ALL models of complex, non linear systems are, of necessity, more simplistic than the systems they model.

The first computer model of weather systems, I believe, included the assumption that the surface of the planet was perfectly smooth, no mountains or valleys. From it, more precise models developed.

-- Jon Williamson (jwilliamson003@sprintmail.com), July 13, 1999.

I've tried using a "weather" analogy when casually analysing Y2K systems with friends. There is this major difference: We know all the components, structures and relationships of our digital technology. Well, someone knows, somewhere. We built it, didn't we? We don't have to research, hypothesize and learn about it as we do with natural systems.

Through generations of research, we are only lately coming to a useful understanding of climatic systems, compared to which Y2K is checkers compared to meteorological chess.

Whereas most natural systems are profligately redundant, (Gaia don't do JIT) a salient feature of our technological efficiency is simplicity and exiguity. If a forest were designed like our computer systems, the first woodpecker to come along would knock it down.

Hallyx

"This whole problem is boring and stupid. So are hippos....until they charge your boat."

-- (Hallyx@aol.com), July 13, 1999.

Jon:

You said: "yep, we all have ahold of different parts of the elephant. Doesn't it make sense in a way that those involved in the "Project from Hell" would be more likely to show up wondering what the heck is going on than those working on a well managed project. We all deal with more or less "self-selected" groups everyday."

Yes, I would agree to this completely. Unfortunately, the assumptions made by those outside IT are based on the information provided by those in this situation. I'd even take this one step further and suggest that those IN IT are gaining confirmation that their experiences are widespread.

-- Anita (spoonera@msn.com), July 13, 1999.

Hi!

-- George (jvilches@sminter.com.ar), July 13, 1999.

Hoff,

Gotcha! Noooo siiiiirr. You won4t convince many people that things will be pretty much okay because of the fact that (a) we4ve known about y2k for many years (b) we4ve been working on y2k for a long time

So what? "Knowing" about y2k is not a good luck charm! ACTION was what we always needed but everybody started too late or didn4t start at all, like many SMBs in the USA or foreign countries that we still badly need for supply chain logistics.

Working on y2k for a long time? Weeelll, obviously the world needed to be working on y2k for a MUCH LONGER time, or we wouldn4t both be here exchanging ideas on the subject, would we?

Hoff, you wouldn4t know how far behind Brazil, Russia and China still are, besides the fact that their pirated software and firmware won4t allow it. Hoff, Anita is right in suggesting that Mickey Mouse is at the helm... I mean Bill Clinton, sorry, 4cause we never had a clear strategy. We still don4t have it and time is up.

"Knowing" about y2k (awareness) is only 1% of the problem Hoff. The code is broken, right? There are 50,000 mainframes out there Hoff, plus 100 million computer systems, and 50 b-b-b-billion embedded chips for Crissake!! And you mean to tell me that we should all rest assured that they are pretty much fixed and adequately tested simply because of (a) and (b) above?? C4mon, get real guys. You are too smart for anything else.

I agree with Y2Knewswire: y2k will be systemic, with concurrent paralellel flaws world-wide, logistically pretty much impossible to Fix On Failure under y2k circumstances which you know well.

As soon as the US government or any of its agencies (CIA, State Dept. or whatever) starts issuing Y2K scorecards on foreign countries (as promised), the Y2K ball game will heat up to unbelievable temperatures. I promise. Make me accountable on this one. I4ll be part of it you know.

-- George (jvilches@sminter.com.ar), July 13, 1999.

Hallyx:

I've got to disagree about "someone" knowing all about the technology. I know I'm not an IT person (I'm a headhunter), but, again, I've talked to folks.

An example is my brother in law. Apparently quite a good programmer. Helping with a SAP conversion right now. He asked for the source code for an application he was working on. No source code. NO ONE left at the company had any idea what the computer was actually "doing" inside. They just knew what reports came out.

If you ask around, I'm sure you can come up with stories like these yourself. I've certainly seen them on this forum. Of course, many companies do have source code, some even have documented changes.....

-- Jon Williamson (jwilliamson003@sprintmail.com), July 13, 1999.

Sorry guys, but as you could guess, I had a 4technical4 problem here and I posted two similar posts without really wanting to. Apologies to everyone. Still, my second post is different 4cause I added more comments.

Another acknowledgement: It4s not Mickey Mouse at the helm, it4s Bill Cl... I mean Daffy Duck.

Furthermore Hoff, I would like to hear your comments on points 2 and 3 on my first post above on the human factor and the CPM nodes 4cause you swiftly avoided the subject. Thank you.

Take care

-- George (jvilches@sminter.com.ar), July 13, 1999.

Yes, Jon, I know what you're saying. Perhaps, in attempting brevity, all I've achieved is obfuscation. Let me try it this way. As a headhunter with an unlimited employee pool, could you not hire a huge yet finite group of engineers who would, among their ranks, know all there is to know about all the programs in existence?

If this should not be the case, this may be a world first: A totally contrived, wholly artificial contemporary system beyond the knowledge of its creators. I can think of no other instance where this is or has ever been true.

Hallyx

"I have yet to see any problem, however complicated, which, when you looked at it in the right way, did not become still more complicated." ---Poul Anderson

-- (Hallyx@aol.com), July 13, 1999.

Hallyx:

Possibly. However, would even the person who wrote a 30+ year old application in a now obscure language that they had not used in 15+ years remember a useful amount of detail about that application if there was no source code to refer back to? And if the application developer could have trouble in that situation, would it be any easier for someone who had never seen it and was trying to reverse engineer the thing, while under budget and time restraints?

"this may be a world first: A totally contrived, wholly artificial contemporary system beyond the knowledge of its creators. I can think of no other instance where this is or has ever been true"

Yes, I believe that is exactly what is going on. A real world first. We live in interesting times.

-- Jon Williamson (jwilliamson003@sprintmail.com), July 13, 1999.

Hallyx,

Loved the hippo quote!

Another problem with the weather analogy is that weather events are usually self-terminating/self-correcting. Tornados pass, floods recede, droughts break...

Y2K events, OTOH, will require skilled human intervention to fix.

-- M.C. Hicks (mhicks@greenwich.com), July 13, 1999.

George, here's goes to your points 2 and 3 (point 1 refered to human factors)

Your second point is not correct. First PERTs apply to project schedules not the network diagrams you imply. Second name one critical node that doesnt have a backup. Name one single point of failure that truly is a single point. Dont point to the satellite that disabled thousands of cell phones last year. If an individual cell phone was affected, the owner had numerous other forms of communications available.

True, redundancy may not be everywhere but the free market is, thus making everything redundant to an extent. Lets say for example there exists only one vendor for a particular product. Its now 1/1/00 and that vendor fails. I would guess that another entrepreneurial spirit will seize the opportunity to provide this product based on the need. If no one does, then the users of this product would find alternatives or do without. Sorry I cant think of any product that has only one vendor but maybe you can.

And on your third point, tell me more about different windowing solutions. Each application may pick a different pivot year but that doesnt make it a different windowing solution. The solution requires coordination with all interfaces. No one, I know of, is using this solution without the proper coordination. Please name one. Further I can think of many international standards and protocols for interfaces. Go to the NIST (or have they changed their name again) web site for more info.

Jon, my company has an application without any source code or personnel who know the language used. Fortunately, the application is non-critical (it prints reports for execs not critical to the business). We are trashing the code and re-writing the application.

-- Maria (anon@ymous.com), July 13, 1999.

Hallyx: you said

"this may be a world first: A totally contrived, wholly artificial contemporary system beyond the knowledge of its creators. I can think of no other instance where this is or has ever been true"

Absolutely!

Maria: First of all George's ref to PERT referred to critical path, honey, not your Netscape icon.

Second, entrepreneurial spirit will indeed be a factor, but we're talking about during a recession/depression/collapse, girl. Reebok USA ain't just gonna spring up like a mushroom. Hell, GM couldn't even switch suppliers in two months.

Thirdly, its very well known that windowing is not the best way to remediate. It's bad enough in house, but across 10/100/100/10,000 interfaces it's suicide. And, its only temporary. Which means all these brilliant companies will be expanding fields or rewindowing shortly, with probably even greater confusion.

And lastly, if you think rewriting the application is a solution at this late date, you are SERIOUSLY misguided.

-- a (a@a.a), July 13, 1999.

Jon said:

"Anita, I have no doubt that you have seen and heard as you describe. I do know that the inside information I have is in stark contrast to your's. One *possible* explanation could be the difference between companies who use contractors versus companies that are struggling to do the work with their own staff."

There is some, though not complete, truth here. Most companies who bring in contractors have made two decisions: they can't do it in-house; they are willing to let the contractors "do it right" or, at least, "do it." Too expensive not to once you've made the commitment.

OTOH, there are two other types of companies:

... Really top-notch ones who can and want to do it right in-house. Not a whole bunch, from my experience (20%?)

... The bozos who patch their Rube Goldberg systems together year after year (40%?)

The idea that IT is an orderly process is wrong, IMO and to my experience with mainframes. It is a whirlpool of technique, budgets, politics and management expertise (or lack thereof). Not engineering. Black art with odd, scintillating moments of logic interspersed (not code logic at the module level, architecture and systems logic).

Anita, I hope you're right. But, to paraphrase Barry Goldwater out of context, "in my heart, I know she's wrong."

Ed Yourdon has spent three decades trying to coax architecturally based design, test and maintenance out of heavy iron IT with minimal success. His pessimism stemmed directly from the implications of that for Y2K, as does mine.

-- BigDog (BigDog@duffer.com), July 13, 1999.

a, so nice of you to respond. Please explain this Netscape icon. I don't use netscape. I know about PERT and network diagrams through project management. The critical path of a PERT deals with the longest time from project start to end. One can "crash" the network by shortening the critical path. This does not include communications networks or single points of failure. Please explain your definition.

So reebok and GM provide a unique products with no other alternatives, since when? Try to stick to the point being made.

Windowing is not the ideal solution but it works, my dear. So what they will need to fix in a few more years. You are living in a dream world if you don't think bandaids fix a lot of problems. We don't live in the ideal world, that's just the point, nothing is perfect.

And lastly, rewriting is as good a solution as finding alternatives when you can't find source and no expertise exists to fix it, my dear. And yes we are doing just that with plenty of time to do it. This "time has run out" chant is wearing thin.

-- Maria (anon@ymous.com), July 13, 1999.

Alright, George, let's have a look. Maria beat me to some of this:

(2) Redundancy?: Yes Hoff, redundancy is there allright. But not everywhere. It ends up and boils down into nodes, as Critical Path Method (CPM) or PERT diagrams would clearly show. Nodes are MUSTS. You can design and build the most expensive house on the surface of this planet, but if doors are not delivered, for all intent and purposes it can't be called a 'house' simply because nobody can live in it. Gartner Group assigns vital, mission critical failures on this basis.

Once again, the point of bringing up "redundancies" was to point out the fallacy of the linear model used by LoserWire.

Most systems within companies can and do survive failures of the components. Sometimes through automated work-arounds, sometimes through manual. The fact is, systems in most large corporations just do not function in any sort of "failure-proof" mode. Sh*t happens, and is dealt with.

Again, I'm sure there are exceptions.

(3) Incompatibilities: Interfaces, no international standards, different 'windowing' solutions, etc., etc.

Here is where you are really off-base. Interfaces are typically point-to-point file transfers, with an explicitly defined definition. International standards have no bearing; the definition is shared from point to point. There do exist a set of standards for EDI definitions. Typically, the bigger fish pushes the smaller one; a WalMart will require all suppliers to utilize a new standard definition, and everyone jumps. Changing the interface definition on just one side isn't done; if it was, it would fail immediately, and won't wait for the rollover.

As for windowing, this "problem" is greatly overblown. With few exceptions, different pivot years have absolutely no effect on the interface, and won't have for another 20-30 years. For the overwhelming majority of dates passed, it just doesn't matter whether a pivot year of 20, 30, 40, or 50 is used. Maybe you consider this pushing the problem further off; but if we truly are facing a crisis on the rollover, then the immediate task is to get thru the crisis. Period.

For the rest, you continually exaggerate the problem. With available metrics, the expected error rate in unremediated code can be estimated at 3 to 6 times normal, definitely not "thousands" of times greater. There may be 50,000 mainframes out there, but they all have staff maintaining, supporting and yes fixing them. And just what fraction of those 50 b-b-b-billion chips even know what the date is, much less have potential problems?

-- Hoffmeister (hoff_meister@my-deja.com), July 13, 1999.

Marma, Hoff, thanks for your reply. I4m in a hurry right now but I can still try to put together a couple of comments.

(1) Marma, an example of a single point of failure possibility is the 10,000,000 barrels of daily oil imports required by the US. This will be the mother of all y2k imported problems. It will be cold in Jan- Feb 2000 throughout the northern latitudes, you can count on that! If you think that going back to 4manual4 or that 4workarounds4 will solve such a problem it means ignoring the very nature of the oil business today, including politics. The 1973 oil crisis will be a champagne garden party in comparison to this event. The oil industry is not even y2k 4ready4 ANYWHERE in the world. Let alone oil-tankers, Customs, docks, pipes... Read the Chevron white paper on Y2K and we4ll continue discussing the oil issue. Throw in a couple of the Texaco refineries reports while you are at it too...An the Amerada Hess incident also... By the way, Amerada Hess shut down for several weeks and still has not solved the original problem... The strategic oil reserves are very poorly distributed (refinery-wise) and they would only last two weeks. Still you need working refineries to get what consumkers need...

Phones is another interesting item. Gartner Group confirmed that in Asia and many countries in Europe phone companies are still clueless about y2k. What4s the workaround here: donkeys and mules with schoolchildren as 21st. century Pony Express?? Oh, surely you'll argue back that doesn4t affect us in the almighty USA, right? (Don4t even think of it)

(2) Windowing not a problem?? You should hear the nasty words that International Settlements Bank (Basle) people shout out loud in fancy meeting rooms concerning the y2k windowing 4solutions4 adopted by Latin American banks for international SWIFT transactions. Yes, Hoff, we4ve all known abut y2k a long time ago but we never got neither leadership, nor international remediation and testing standards, nor national standards. Were they not needed Hoff?

(3) I don4t feel that either of you have addressed the CPM nodes problem I described above.

(4) The 4manual workarounds4 do not resist further analysis quite frankly. I can4t believe what you are suggesting. Just where does all of your y2k line of thought leave robustness, quality assurance, total quality, just in time (sounds more like just in sh=t to me!), the EPA, etc., etc., plus, PLUS the human factor which IMHO you also left unanswered. Sorry. Tyhanks for your input anyway. Bye (wife upheaval in the cookin4)

-- George (jvilches@sminter.com.ar), July 13, 1999.

(2) Windowing not a problem?? You should hear the nasty words that International Settlements Bank (Basle) people shout out loud in fancy meeting rooms concerning the y2k windowing 4solutions4 adopted by Latin American banks for international SWIFT transactions. Yes, Hoff, we4ve all known abut y2k a long time ago but we never got neither leadership, nor international remediation and testing standards, nor national standards. Were they not needed Hoff?

Since I'm not privy to those discussions, you'll have to enlighten me. Interface file dates either use 2 digit or 4 digit years. If 4, the correct century is supplied, or it isn't. While possible that an invalid pivot year could be selected, I would think it extremely unlikely.

Whether or not international standards would have helped, corporate application development just doesn't work that way. For good or bad, the standards that do exist are developed in house, and vary from site to site. The nature of the beast. In general, corporations tend not to respond to governments telling them how to run their business.

Which leads me to really the basis of my optimism. Yes, I've worked with a few large corporations, know what they've done, and know they are not going to fail due to Y2k problems. I also can see the 20,000+ SAP installations of R/3, and realize quite a few other corporations have dealt with the problem as well.

But that is not the corporate universe. In essence, it gets back to the point you were harping on previously, Awareness. Companies are in business to make money, and survive. The survival instinct is very strong. Once made "aware" that Y2k posed a potential risk, that survival instinct takes over. I have seen it firsthand, and have no reason to believe that other companies are significantly different. Companies have spent the money to fix Y2k, and are fixing it, because it is in their best interest to do so. It's in noone's interest to let the company fail, and not fix it.

-- Hoffmeister (hoff_meister@my-deja.com), July 13, 1999.

Sure Hoff, it4s no one4s interest to have the company fail...

So according to your logic no company would ever fail and World Wars would never happen... because it4s no one4s interest, right?

By the same token, it doesn4t matter if y2k awareness was not there, if leadership was not there, if everybody started late, if the Federal and State and city governments are way behnd still... It doesn4t matter. You are betting the farm on their " instincts", right? Cut it out Hoff, will ya? (as I4ve said, you sound intelligent)

-- George (jvilches@sminter.com.ar), July 13, 1999.

Sure Hoff, it4s no one4s interest to have the company fail...

So according to your logic no company would ever fail and World Wars would never happen... because it4s no one4s interest, right?

Never said companies don't fail. Never said some companies won't fail because of Y2k. But this isn't some altruistic notion, that companies do things for the greater good. The profit-motive, and survival, keep most companies from failing. The same is true for Y2k. Those that do, will do so because of bad business decisions.

As for Wars, you're wrong. Most wars start because at least one side thinks it's in their best interest to fight it.

By the same token, it doesn4t matter if y2k awareness was not there, if leadership was not there, if everybody started late, if the Federal and State and city governments are way behnd still... It doesn4t matter. You are betting the farm on their " instincts", right? Cut it out Hoff, will ya? (as I4ve said, you sound intelligent)

You really should define what "leadership" you would expect. You're starting to sound, well, almost socialistic.

I am actually more concerned about governments, for exactly the same reason I am optimistic about businesses.

-- Hoffmeister (hoff_meister@my-deja.com), July 13, 1999.

What leadership I'm talking about? Every sort of leadership, political, business, labor unions (why not?), NGOs, everybody for Crissake. What's socialistic about leadership Hoff? You are not making as much sense now Hoff with this type of silly question. Don't tell me now that you believe that Y2K is small enough a subject as not to deserve leadership from management, Congress, etc.

Sh*t happens, you agree. Well, BIG SH*T also happens Hoff, it's just that you don't want to admit the mere possibility. It just can't be you say. Well, it definetly can be and surely will be if Y2K or any other equivalent phenomenon is approached the way it has, worldwide.

If you can't see that, let's drop the subject and goooood luuuck!

-- George (jvilches@sminter.com.ar), July 13, 1999.

I guess you and I have different views on the effectiveness of government leadership on business. In most cases, I feel the less government intervention, the better. And yes, this includes Y2k. Provide assistance, if required. But the decisions are best left to the business.

Yes, sh*t happens. And yes, probably more sh*t than normal will happen with Y2k. But it's not the apocalypse, and we'll deal with it.

-- Hoffmeister (hoff_meister@my-deja.com), July 13, 1999.

Hoff, the less government intervention the better??????

For fixing FEDERAL, STATE and MUNICIPAL I.T. systems from the y2k bug the LESS GOVERMENT INTERVENTION THE BETTER???????

ARE YOU NUTS Hoffmeister????

Just what are you talking about???? Are you suggesting that private businesses should have taken the leadership in fixing government systems??? Worldwide???????? Are you feeling O.Kay Hoff???????

You've just earned yourself a thread Hoff. Smile.

-- George (jvilches@sminter.com.ar), July 13, 1999.

Marma (and Hoff),

Your silence and swift by-passing of the "single point failure" examples you requested (please see my post above) apparently should be interpreted as your acknowledgement of the serious node problems I pointed out in a post further behind concerning pertinent y2k CPM considerations.

Hope to hear from you at the new thread I posted today:

"HOFFMEISTER: y2k leadership attitudes socialistic(??)//the less government intervention (for fixing Federal, State, and Municipal y2k problems) the better (!!)

-- George (jvilches@sminter.com.ar), July 13, 1999.

George, Unlike you, I can not spend 7x24 support to this forum. I do have a life. My comment on your post above takes some understanding. First what's your definition of CPM? Second, have you analyzed the global network involving Y2K that you can determine that oil is on a critical path? If so, please elaborate. I contend that even though oil distribution may be critical, it does not represent a single point of failure. I've studied communciations networks for the military and know a single point of failure can bring down the system. That's when we start to build alternatives and backups to reduce the probability of failure.

Also your example of phones as a single of failure does make it. Pony express may not be a viable alternative but there are alternatives (satellite phones, cell phones, the internet, and LEO comm) that are extremely viable.

And lastly, George, Hoff is right the gov should stay out of business (commercial business that is). The states have been doing Y2K remediation since 1995; I know because I worked on a contract back then for the state of Nebraska and our company went after many more contracts after that one.

George, stop your shouting; this is a gentle forum.

-- Maria (anon@ymous.com), July 14, 1999.

Marma:

I also have a life Marma, a very intense life by the way. And a lap- top, which gives you the false impression that I4m a full-time EY forum participant, which I4m not. I am more y2k knowledgeable than John Q. Public though and pretty energetic people say. I actually post very little Marma. By the way, how come you never place the accent on " Marma" ? Beautiful name you know.

Marma, please re-read Hoff. You and him don4t agree on government intervention. He is saying something completely different.

Your views on single points of failure are very personal Marma. I can4t help you there. As far as satellite phones availability, etc., in countries like Russia, Brazil, China, Indonesia, Argentina, etc., I also leave without comment.

Warm regards

-- George (jvilches@sminter.com.ar), July 14, 1999.

Alright, George, you've got me curious. I thought you just sort of lost it, and were grasping at straws. (Still do, actually)

But after rereading the posts, just don't see it. So enlighten me here. Which post of mine was I not talking of businesses, where I said less government intervention the better?

-- Hoffmeister (hoff_meister@my-deja.com), July 14, 1999.