From Cook: What percent of self-reported "survey data" will accurately reflect remedation successes?greenspun.com : LUSENET : TimeBomb 2000 (Y2000) : One Thread
Spent a couple of hours reading the government's latest quarterly report.
There is evidence of a lot of progress being reported, but the fundemental question always has been (since early 1998) what precent of the government, utilities, large business, and small business programs and processes actually be fixed?
Yourdon assumed 80% in his origianl editions, looking through the government's latest status report indicates that "80%" may be close even now: after grossly summarizing many of the industrial group reports, most appear to be reporting that 90% of the companies and agencies are reporting they are 90% complete.
This entire report - and the thus the nation's entire y2k-course-of-action - is, of course, all determined from self-reporting, self-survey data of various voluntary industrial groups and the government reports from the Clinton administration's agencies.
Ee must determine how accurate the fundemental assumption is: if an industry group of 1000 "anythings" is surveyed, and only 50% respond to the survey, how many of the ones NOT reporting are actually working on y2k issues, and just didn't answer? How many who didn't respond are embarressed by their potential/probable failures/lack of progress, and don't want to respond.
For consistancy: let us assume that ONLY end-of-November "success" is valid, because it is reasonable to assume that programs rushed into service in the last two weeks of december mayhave better nevr altered at all....
Of the 500 "anythings" who responded to this self-reported survey, if 90% report they expect to "finsih by Sept 30", how many will actually complete their remediation and testing before December 30?
Again, just because "most" software programs are delivered late, some do finish on schedule. some a "little" late, and some "a lot" late. Therefore, a few/ some/many/most/almost all of these 90% will get their programs finished, right?
Assme that the rest (10%) "expect to finish" between 30 Sept and 31 December. How many of these stragglers will actually get done in time?
Fourth Question: If only "critical systems" are repaired and tested, can a typical company remain profitable for a full quarter? If not, how long will it take for the remaining non-remediated-but-vital systems to be repaired/returned to service - assume power is restored (and stays up) within 7-10 days of Jan 1.
Fifth Question: reconsider your answers. Would your answers lead you increase or decrease the number of days/amount of preparations you expect to actually do?
-- Robert A. Cook, PE (Kennesaw, GA) (email@example.com), August 26, 1999
Darn, making me think in the middle of the afternoon.....
First question. As for those who chose not to respond, probably at least half are not doing well, or are not doing much at all. Maybe more.
Second question. Perhaps 40% of those reporting status as 90% or greater complete will miss the 12/31/1999 deadline.
Third question. The percentage of the stragglers who finish remediation by 12/31/1999 will be negligible.
Fourth question. Extremely unlikely. I believe this is one factor behind the rise in corporate bond offerings this quarter. They are seeking operating capital to keep core operations open while fixing on failure. Depending on the size of the company, whether the application "drops dead" or injects bad data, and sheer luck, good AND bad, it could take as long or longer than the original "mission critical" systems remediation.
Fifth question. No, Robert, I've done everything I can think of to move to self reliance. No little family is an island, but I've done what I can to provide for my four, one sister and her four, and two parents. Plus possible others.
-- Jon Williamson (firstname.lastname@example.org), August 26, 1999.
So, if 1000 in an industry are "surveyed, and 500 respond, then:
Of the 500 who respond to the survey:
450 (90%) will claim to be ready by 30 Sept, but only 400 will actually be compliant. Assume 50 will face major, unexpected problems or significant delays in the final phases of repairing their critical systems.
50 (10%) more "expect" to be compliant before December by know they cannot finish before Sept 30; these are the stragglers.
Of these 50 stragglers and 50 "unexpected failures", 50 might get finally done before December, and 50 will face failure in critical systems.
So of 500 who answered the survey, it's reasonable to assume 450 will have their critical systems operable next January/February.
Look at the 500 who failed to answer the survey:
Assume 1/3 are making the same "reasonable" progress as those who answered the survey but chose not to respond, 1/3 are making some progress but are far behind, and 1/3 intend to "fix on failure" - (either by deliberate ommision or by accident or by corporate stupidity) So of these 500, 167 are making progress, but only 90% are actually going to get good results = 150 will manage the rollover.
167 are making poor progress, and 50% will face critical system failure: so 83 will face rollover okay, and 83 will have critical system failure.
The last group of 167 may have 10-25% succeed in fixing these (see where your assumptions come in?) - so only 42 will succeed, and 125 will face critical system failure.
Add both groups up: 450 may succeed of those 500 responding, and (150+83+42 = 275) from those who ignored the survey.
Industry-wide: 725 (72%) will manage, and 275 (28%) are expected to face critical systems failure.
____ And this conclusion (admittedly rough, but defensible assumptions) is derived from an industry who has told the government it is already "90% ready for y2k"......
-- Robert A. Cook, PE (Kennesaw, GA) (email@example.com), August 26, 1999.
You also have the semantics problem. Since there has never been a Y2K before, the terms "Y2K Compliant" and "Y2K Ready" have no fixed meaning. Each company could have a comletely different take on what it means to have all mission critical systems ready for the rollover. Does it mean they have everything fixed? Does it mean they only have everything fixed that they think will cause a problem? Does it mean they haven't fixed much of anything, but believe they know just what to do, when things break down (FOF)?
I wouldn't begin to know how to assess the survey results. I think you'd first have to survey the same companies, to find out just what "Y2K Compliant" means to them.
-- Bokonon (bok0non@my-Deja.com), August 26, 1999.
-- mabel (firstname.lastname@example.org), August 27, 1999.
Robert - Related to your questions #2 and #3 is my pet peeve - How much of the work not yet done reflects software or hardware upgrades not yet received because *those* vendors aren't ready yet or are behind in their mfg/distribution processes??
-- Brooks (email@example.com), August 27, 1999.
All of you raise excellent questions and I agree with your estimations. For myself, I have always assumed the MAJOR reason for low numbers of respondents to surveys is negative information about themselves. I base this on life experience. I am sure that AT LEAST 50% of those who fail to respond do so deliberately and because they are not/will not be ready for rollover. BTW, I expect that percentage to be higher amongst government agencies.
-- j.r. (firstname.lastname@example.org), August 27, 1999.