from today's c.s.y2k post by Cory Hamasaki:

I'm sure the source code looks very similar. And in a language like the early BASICs where variables are single letters, a JAE algorithm and your "Pilot bid" would be exactly the same. However, this is not exactly the Jo Anne Effect as defined by Jo Anne.

This is the second pathological case of a date bounding algorithm going wrong in December 1998. Bill Hoyt's insurance co. in Baltimore (he named the building, -bks-) was the first. I want to remind SHMUEL that he was certain that the cases would not start until January 1999. Well, SHMUEL my commando pal, something's started.

My guess, and SHMUEL this is based on mulling over the timing of the 000197AF event, trying to figure out how early a date bounding algorithm can fail as well as the information in Jo Anne's articles, ... my guess was that a small percentage of the failures occur prior to the bounding period. Two reports are enough for me. You may want a few more, patience, my multi-lingual friend, they will come.

I suspect that most of the events have not yet been recognized by the system owners, they're losing the data but think, oh, business might have been a little off due to the ice storm or vacations.

You geeks and system operators out there who have been sitting on the wall, not sure if JAE is real or even if Y2K is real, fair warning. Take a close look at your applications. Maybe I'll dig out my papers and put together a "What to look for guide". for junior programmers, print some copies up.

The formal definition of the JAE might not include Bill Hoyt's record purger or this scheduling application but, fair warning, the systems are failing in strange and wonderful ways.

Let's assume that Tom's report is accurate, (it certainly is convincing to me) the -bks- questions are, 1) will the clueless press pick up this story, 2) will the airline itself go public or will they surpress this event, 3) will this turn into an Internet rumor and morph into other forms?

On Wed, 30 Dec 1998 03:59:06, "Tom C." wrote:

> A major airline that must remain nameless is unable to run a "pilot > bid" that extends into the year 2000. Every time the bid is run the > computer generates errors because it doesn't see the year 2000 > correctly. > > This airline must run a pilot bid every 3 months or so due to the > rapid expansion in this industry. The bid must run perfectly. Over > 10,000 pilots are affected by this bid. The cost to the airline in > training expenses alone are quite high if this bid can't be run > correctly. A manual backup will be painstaking at best. > > This to me is an example of the JoAnne effect. > > Tom

I've kept a file of the professional denial-heads, not -bks- who is in a league unto himself and practices a pure scientific/skeptic method but the real butt-heads out there, such as the so-called "Computer Professionals for Social Responsibility" who have screwed up royally.

CPSR had a chance to run pure research on the Y2K problem, like Egan did, and surveys like Capers Jones does, combined with analysis of live date algorithms, asking the questions how and when will the algorithms fail. Then they could have sampled the algorithms in industry and produced estimates on the failures using the analysis. But noooo, that's too much work for the lazy academics. (some CPSR members are academics.)

Those butt-heads in academia dropped the ball big time and chose to snipe at alleged rumors, I've seen their trite little rumors table. Is that "research"? I don't think so.

It would have been an excellent use of Computer Science graduate student horsepower, live code, analysis, samples from industry, publication quality papers estimating the failures in December 1998 and forward.

But noooo, we got idiocy like the cookie lady article in Scientific American, how shabby. How pathetic. How misleading for the public. And yes, people will die because those idiots ran their keyboards. I just want to s-s-s-shake s-s-s-some s-s-s-sense into the them.

Well, it's too late now. It's started. It would take a year, two classes of graduate students to gather the information and perform the analysis.

Sorry people, Ed Yourdon knows, so does Capers Jones, this mess is unraveling, the good news is, problems like the "Pilot bid" and Bill Hoyt's Baltimore insurance companies can be fixed, but:

1) It takes pros, not clueless nubies (review Bill's report, the problem persisted until they pulled in the old hand, the nattering geezer who should have been right-sized out of there, clueless old fogger, doesn't know how to click a webpage.)

2) This is just the beginning, it gets worse.

3) When does the magnitude of the problems overwhelm the capacity of the fixers to repair them.

4) How long before the sheeple notice that -mooOOOooo- something different is happening, -mOOOooooOOO-, I'm afraid.

paul milne says he wants to see the butt-heads eyes bug out; paul is fooling himself. They'll forget that we've had these running arguments and say, "how could anyone have known, we heard rumors but no one believed them." or "I was challenging YOU to produce the evidence and YOU let them down."

cory hamasaki 366 Days, 8,800 Hours.

--

More to come. Take care people, this ride is just starting.

-- a (a@a.a), December 30, 1998

Answers

Cory too? Didn't know that B***h*** was such a popular geek vocabulary staple. Has Cory bugged out? Have all the hard-hitting geeks bugged out? a, please keep us updated. Very curious to see if 1999 brings it all out into the open in such a way that ppl will be more willing to prepare.

xxxxxxx xxxxxxx xxxxxxx xxxxxx

-- Leska (allaha@earthlink.net), December 30, 1998.

I've said it before, I'll say it again. The "Jo Anne Effect" started long ago in many places. My company's finance dept. has been dealing with it since the start of FY1999 on 7/1/98. The program looks ahead to FY2000 and won't close out the month correctly. They have a work-around in place. We also have a problem where we can't store expiration dates past 12/31/99. We've been dealing with this since 1998. I suspect this must be true of many companies.

-- Buddy (DC) (buddy@bellatlantic.net), December 30, 1998.

I believe Buddy is on the mark here. The JoAnne Effect has been around for a long time.

C.H. is spending so much time and bandwidth yacking on about a formal definition of the JoAnne Effect that he overlooks the obvious, namely that the JAE occurs anytime a system tries to process a date on or after 1/1/2000 and does so incorrectly. JoAnne initially described the effect for financial systems that forcast ahead into 2000, but it occurs anywhere that date math is used. Therefore, it can occur at any time. For example, mortgage payment routines could have (and in fact some did) display the JAE in 1970 when calculating payment schedules going out into 2000.

-- Paul Neuhardt (neuhardt@ultranet.com), December 30, 1998.

Yes, of course Y2K problems have been encountered in the past, that's obvious. But you're missing the whole point of the JAE. It's in the sheer number and complexity of the failures, the timing, and the fact that many will be misinterpreted until the data becomes irretrievably corrupt.

-- a (a@a.a), December 30, 1998.

OK, aaa...but my point is that Cory says, "This is the second pathological case of a date bounding algorithm going wrong in December 1998. Bill Hoyt's insurance co. in Baltimore (he named the building, -bks-) was the first. I want to remind SHMUEL that he was certain that the cases would not start until January 1999. Well, SHMUEL my commando pal, something's started." He and his cohorts who are so sure they are such superprogrammers and authorities totally missed the call on this one and have overestimated its effect. My guess is that there is not a corporation in the U.S. that hasn't noticed the Jo Anne Effect yet if their FY1999 has already started, and many probably encountered it sooner than that.

-- Buddy (DC) (buddy@bellatlantic.net), December 30, 1998.

Buddy: I don't know if you're a programmer or not, and I can't speak for Cory, but...

The difference here is that this effect is based solely on the one-year look ahead. Other software problems can occur because of FY calculations, mortgage schedules, reservations into 2000, etc. JAE is the first example of miscalculations because the calendar has progressed beyond a certain point.

This problem is different simply because it was not anticipated. It's a subtle thing...maybe you have to be a gearhead to appreciate it.

-- a (a@a.a), December 30, 1998.

A, you said:

"The difference here is that this effect is based solely on the one-year look ahead."

Not true. I haven't read Ms. Slavin's original description that led to the term "JoAnne Effect" in quite some time but I don't remember her limiting the problem to one year look-aheads. She did use a one-year look ahead as her example since it was the situation she was most familiar with. If she did limit her concerns to one-year look aheads, then she has defined the problem too narrowly. There are systems all over the place that look much farther than one year in to the future, just as there are systems that take shorter looks that may still experience failures before 1/1/2000. Those systems with longer look-ahead periods have all had their chance to experience the JoAnne Effect by the day this message was posted.

Of course, from this day forward, only look aheads of one year or less will cause new instances of the problem, so I suppose this discussion could be considered a moot point. Still, the fact remains that there have been many systems in many places that have already taken a peek past 1/1/2000 and have therefore experienced the JoAnne Effect. Some survived, some didn't. Such is life.

-- Paul Neuhardt (neuhardt@ultranet.com), December 31, 1998.

You guys are still missing the point! The JAE is centered on Jan 99. It is a CALENDAR effect, not so much a software/FY/etc. thing. Cory was saying that the effect is actually wider than the Jan 99 window, that it overlaps into Dec 98.

At least that is my understanding of Cory and Jo Anne.

-- a (a@a.a), December 31, 1998.

"Cory was saying that the effect is actually wider than the Jan 99 window, that it overlaps into Dec 98. "

That is precisely the point a.a.a...it overlapped back at the beginning of FY1999. In my company's case it was 7/1/99. In other cases it was before that.

Paul N. is right. It is a moot point now. The "Jo Anne Effect" is and has been occurring already.

-- Buddy (DC) (buddy@bellatlantic.net), December 31, 1998.

"You guys are still missing the point! The JAE is centered on Jan 99. It is a CALENDAR effect, not so much a software/FY/etc. thing. Cory was saying that the effect is actually wider than the Jan 99 window, that it overlaps into Dec 98. At least that is my understanding of Cory and Jo Anne."

Not to put too fine a point on it, but "Duh!" That is exactly what Buddy and I have been saying all along, except that we are pointing out that it overlaps a lot farther back than just 12/1998.

-- Paul Neuhardt (neuhardt@ultranet.com), January 01, 1999.

OK maybe I am missing something, and maybe we need to ask Cory or Jo Anne for clarification, but...

what you guys are saying is that ALL pre Jan-99 failues are JAE. That was not my understanding of JAE from the original articles. I mean come on guys, I thought EVERYBODY knew that there would be pre-2000 snafus!

-- a (a@a.a), January 01, 1999.

# # # 19990101--Jo Anne Effect ( JAE ) == Event Horizon ( EH )

The EH date of failure does not matter ( e.g., 110 years, 30 years, 18 months, 1 month, 1 day, even 1 nanosecond )!

Regards, Bob Mangus # # #

-- Robert Mangus (rmangus@mail.netquest.com), January 01, 1999.

A,

"what you guys are saying is that ALL pre Jan-99 failues are JAE."

Yes, that is exactly what we are saying, assuming we are talking about failures related to a system's inability to handle dates on or after 1/1/2000.

"That was not my understanding of JAE from the original articles."

Fair enough. You should remember that JoAnne Slavin is a financial specialist (as I recall she is an accountant or some other related profession) and not a technician. When she first described the problem that has since been given her name, she did so from the point of view of a finince person who was worried about financial systems being able to forcast and post beyond 1/1/2000 during 1999. She defined her concerns along those specific lines since that is what she was most familiar with. But the "effect" she was worried about in those original postings can occur in many other systems outside of the few finincial ones she examined, and they occured long before 1999. It's the exact problem she describes, just on a wider scale.

Don't let the original description of a problem limit your thinking. Keep examining it and always ask yourself, "Is there more to this than meets the eye?"

-- Paul Neuhardt (neuhardt@ultranet.com), January 04, 1999.

Oops, missed again. Darn!

-- Where (are@ll.theproblems?), September 10, 1999.