Orlando's Traffic Lights Lock Up. So, Re-set Them to 1972

greenspun.com : LUSENET : TimeBomb 2000 (Y2000) : One Thread

Here is a good read directly from the web-site of Gary North, dated 31-Aug-98

Orlando's Traffic Lights Lock Up. So, Re-set Them to 1972

Link: http://www.orlandosentinel.com/news/y2k/y2k0829.htm

-- Buddy Y. (buddy@bellatlantic.net), September 03, 1998


For anyone who doubts that there is a "Y2K embedded chip" problem, take heed.

-- Joe (shar@pei.com), September 04, 1998.

There is not an "embedded chip" problem. There is an "embedded system" problem.

The point of my posting this link is that the article does not support Gary North's views, even though he put the link on his web-site.

In the case of Orlando's traffic lights system, their solution makes perfect sense.

-- Buddy Y. (buddy@bellatlantic.net), September 04, 1998.

Notice what happened: They aknowledged the problem, tried out a solution and worked out an alternative. ("... if the vender can't fix it, then we will reset the computer..) Most important, Orlando tested their current computer program, (it failed), made arrangemetns with the vender for a repair, and then tested their fall-back solution in case the vender can't perform.

Now granted, their solution is at best a patch work fix, but it shows the right process is going on. Multiple their progress now (Sept 1998) with those who aren't testing, fixing, planning.

-- Robert A. Cook, P.E. (cook.r@csaatl.com), September 04, 1998.

It's interesting how the optimists and pessimists view the same event differently. The pessimists say, "See, I told you that embedded systems can cause problems, although the optimists said that embedded systems don't use date sensitive data".

The optimists say, "See how easy it is to come up with a workaround for any embedded systems problem. No big deal."

Jim Lord came up with the analogy that fixing Y2k problems is as easy as polishing marbles - except that the marbles fill up the Grand Canyon. Yes, Orlando came up with a reasonable patchwork workaround. But how many man hours went into the effort to find, test, develop a solution and implement the solution? Let's see, that many hours times 5 billion non-compliant embedded systems equals how many hours? And I bet that the Orlando traffic light embedded systems weren't 1000 feet down in the ocean under an oil drilling platform.


-- Dan Hunt (dhunt@hostscorp.com), September 04, 1998.

I think a definition is in order here.

The number 5 billion in the recent Datamation article refers to the number of non-compliant chips manufactured since 1972. Repeat "chips" not "systems." One "embedded system" can contain anywhere from one chip to hundreds, even thousands. Do not confuse "chip" with "system."

-- Buddy Y. (buddy@bellatlantic.net), September 04, 1998.

Everyone seems to agree that Orlando did the right thing. They found what seemed to be a potential Y2K problem, they did testing to verify that indeed it would be a Y2K problem, and they implemented a successful workaround. Truly a Y2K success. However, there are plenty of embedded systems/devices/chips running around planet that control life-sustaining functions for which there is no hope in even finding much less fixing in the time remaining. Thats reality, folks.

-- Joe (shar@pei.com), September 04, 1998.

Call me a cheerleader if you like, but I don't believe in "impossible" and "no hope."

Remember Apollo 13?

-- Buddy Y. (buddy@bellatlantic.net), September 04, 1998.

The second parapgraph of the article is as follows:

"The city has discovered its *central* [emphasis mine] traffic-light computer is clueless when tested with dates beyond Dec. 31, 1999."

Yes, there are a lot of embedded systems in most modern traffic control systems, but at least on the surface this would seem to be more indicative of a problem in a more traditional (i.e. software driven) system.

Now, let's see the possible spins on this. Some will say "See, the embedded system problem isn't as bad as people think. This wasn't even an embedded system." Others will say "See, I told you how hard it is to find an embedded system." Me, I just say "Another problem, another workaround that buys time until a real fix can be found. Hope to seem more like it."

-- Paul Neuhardt (neuhardt@ultranet.com), September 04, 1998.

No Orlando did NOT come up with a solution. They were very diligent, and started early, and had a system they could get to (on the street corner), and tested it, and still only found out that they could come up with a work-around - an interim fix to POSTPONE one problem.

Its a start - give them credit for what they did, and encourage them fully - but it isn't a solution yet. The solution is when their street lights are working (still need power) continously AFTER the vender finishes replacing the chips (boards) in every controller in every street corner.

-- Robert A. Cook, P.E. (cook.r@csaatl.com), September 04, 1998.

I beg to differ Mr. Cook. Traffic signal systems don't have date-sensitive controllers on every street corner. The date is only in the central computer system. Washington, DC has pretty much the exact same problem. This is not an embedded systems problem. It is simply the fact that the traffic signal computers are antiquated (in computer years) and have been overdue for replacement for years.

What is interesting here is noone is commenting on the statements made later in the article concerning the Big 3 automakers, the airlines, and CSX.

-- Buddy Y. (buddy@bellatlantic.net), September 04, 1998.

What is there to comment about? "We are working on it, we will get there." "Sure, may have a few glitches ... but nothing major or anything." And this will be claimed for as long as it can be. (Aside: anyone taking bets as to whether the Fed Gov't will meet their 4/1/1999 deadline?)

-- Joe (shar@pei.com), September 04, 1998.

My own quote was a mistake:


I apologize.

I meant to say that in the case of Orlando, as described in the article, and in the case of Washington, DC which I am familiar with, the traffic signal systems don't have date-sensitive controllers on every street corner. These are systems that have been in place for a long time. I would guess that many cities are similar. Of course, they will all have to check them.

-- Buddy Y. (buddy@bellatlantic.net), September 04, 1998.

"Traffic signal systems don't have date-sensitive controllers on every street corner."

is the missing quote from my previous post.


-- buddy (buddy@bellatlantic.net), September 04, 1998.

Buddy Y pointed out that the number 5 billion refers to the number of non-compliant chips, not the number of systems, stating that up to hundreds of chips might be in a single embedded system. I appreciate the clarification - but would like others to comment on any knowledge of the average number of chips in an embedded systems.

OTOH, doesn't Buddy's observation raise the odds that any given embedded system will have a problem? If 7% of the chips have a problem, then isn't there over about a 19.6% chance of an embedded system problem in a system containing 3 chips? I think the probablity math for non-failure is (1.00-.07) cubed, which equals 80.4% OK, 19.6% not OK. Any statisticians out there, feel free to correct my math. Extending the logic would give a 68.7% chance of failure in a system with 16 embedded chips.

To me, this would make the embedded systems problem a bigger problem, not a smaller one. <<<<<<<<<<>>>>>>>>

-- Dan Hunt (dhunt@hostscorp.com), September 04, 1998.

The problem with the 5 billion out of 70 billion number is that we don't know the definition of "chip" they are referring to. Does it include all types of chips? There are microprocessors, memory chips, clock chips, arithmetic-logic chips, ROM chips, PROM chips, EPROM chips, EEPROM chips, etc. etc. It behooves me now to go try to find the DATAMATION article.

-- Buddy Y. (buddy@bellatlantic.net), September 04, 1998.

The only clue as to what type of chips they are talking about is "top logic chip manufacturers." If that means all chips, then the number of embedded systems is actually much smaller than those numbers. If by "logic chip manufacturers" they mean only processors and controllers then the number of embedded systems would be higher but still not 70 billion. There is no question that in either case it is still a large number.

Here's the URL and quote. You have to subscribe to read the article.


" From interviews with the top logic chip manufacturers, Harden estimates that approximately 5 billion of the 70 billion chips produced since 1972 are subject to Y2K problems. "The question is, how do you go through the 70 billion to find the 5 billion that will have a problem? It's the quintessential needle in a haystack," he says. "

-- Buddy Y. (buddy@bellatlantic.net), September 04, 1998.

Or, in other words, turn out the lights...the party's over!

-- It's too late (Good@night.com), September 04, 1998.

I fear the lights will be going out by themselves...the party will be trying to turn them back on again.

-- Robert A. Cook, P.E. (cook.r@csaatl.com), September 05, 1998.

This thread illustrates the uncertainty surrounding embedded systems. 1. If - 5 billion chips is an accurate number 2. If - Those chips are installed singly or in multiples in a single system. 3. If - Those chips are still in service 4. If - Those chips are used in "mission critical" systems (not our office copier) By the time you get to "If #4" any human being can only guess how many systems will fail. I guess all we can do is be ready for something ugly.

-- Mike (gartner@execpc.com), September 07, 1998.

Moderation questions? read the FAQ