X-Message-Number: 10470 Date: Fri, 25 Sep 1998 08:34:04 -0700 From: Peter Merel <> Subject: Y2K: Hard Facts, Fire, and The Marines Perry Metzger duplicates one of the many articles that are circulating lampooning Y2K. One of the biggest problems in understanding the situation is that there's so *much* written about it, good, bad, and mostly, like Perry's article, indifferent, that hard facts are exceedingly tough to find. There have been a couple of publications recently, however, that ought to give anyone with a technology background pause. First and foremost is the Dept. Of Energy's NERC report at ftp://ftp.nerc.com/pub/sys/all_updl/docs/y2k/y2kreport-doe.pdf Although this report draws optimistic conclusions, the data tells the real story. NERC surveyed just about every electrical utility in the US over Q3 98; almost all responded. Dig the crazy curves in sections 4.2-4.8. See how they all magically reach 100% readiness in Q4 99? Now remember what we know for a fact: 80% of all technology projects are delivered late. They're delivered late regardless of size, importance, or resources. Charting technological development is not even a black art - we simply don't know how to do it reliably. The relationship between progress, quality, resources, requirements and time is highly non-linear. Even if we accept the wildly unrealistic curves here on face value - and this is hard to do with projects jumping magically in completeness by 25% every 3 months - the cold fact is that 80% of these projects are going to run late. This is still more certain when you realize that most electric utilities won't be adequately tested before 1/1/00 because they're 24/7 systems, and that according to the report 36% of these organizations don't even have a written plan on they base their magic estimates. You can substantiate this horrible thing with more detailed if less comprehensive data: check out Minnesota State's recent study of its utilities at http://www.dpsv.state.mn.us/docs/infocntr/year2000/survey.htm . On the strength of this, the US power grid seems certain to go out. When we think about for how long the power will go out, the important factor is on the generation end of things. Failing hydro and nuclear plants, if their remediation can be effected instantly, will come back up within a matter of hours. Coal-fired plants, however, take a week or more to come back online because they need to bring boilers back to heat. More than half the electricity in the US comes from these coal-fired plants. Add to that the implausibility of the hydro and nuclear plants doing a magic instant remediation, and it's certain that the blackouts will last for at least a week across the US. What does this mean? Water and telephone networks rely on electricity for their function; so they go out too. In the absence of light and TV, most city folk will use candles and oil lamps to read by. One in a thousand bump their candle or lamp onto something flamable. Dial 911! Whoops, can't do that, no phones. Use the garden hose! Uh, sorry, no electric pumps means no pressure. Without water the fires spread fast, and even if the fire department could get to them they don't have the manpower for this kind of load. It's midwinter: supplies of heating oil turn to sheets of flame. Folk can see the fires sweeping toward them and panic. Jump in the car! Get outta there! Whoops, gridlock. It's midwinter: those that don't burn to death in their houses freeze to death in their cars. Okay, let's wish this nightmare away. Nothing rational about doing that, but it's just too horrible to take seriously. Now what? http://www.prepare4y2k.com/embed1.htm http://www.senate.gov/~y2k/statements/091098miller.html http://www.theaustralian.com.au/extras/007/4051126.htm These are typical of embedded systems remediation anecdotes, but they're not exactly hard facts; if you want these check http://www.mot-sps.com/y2k/black_prod.html - Motorola's list of "black" date-dependent chips. Intel's list is at http://support.intel.com/support/year2000/status.htm . About 90% of the 70 billion chips - http://www.datamation.com/PlugIn/newissue/09y2k.html - out in the field go into embedded systems, and estimates for the gross percentage of y2k-failing chips overall run between 1 and 10% depending on who you believe. So then, looking at where these chips go, http://www.ragts.com/webstuff/y2k.nsf/Pages/Brands-Allen-Bradley?OpenDocument is Allen Bradley's list of its date dependent manufacturing controls. Foxboro's is at http://www.foxboro.com/y2000/index.htm What's most interesting about the Foxboro list is that 30/580 or about 5% of their controls have problems. Of the systems based on these controls, however, 111 of 261 or 42% have problems. This is a little fuzzy because they're still studying some of their systems, but it'll do for now. http://ourworld.compuserve.com/homepages/roleigh_martin/y2k_com.htm provides links to a great many other manufacturers, so you can verify that these are typical numbers. Why are the more complex systems here failing so much more often? it's the Beach/Oleson Pain Index - http://www.cio.com/marketing/releases/y2kchart.html - at work. Simply put, the idea is that the more interconnected a process is, the more prone it is to catastrophic failure. Now the systems Foxboro produce are generic; they don't represent installed control systems, which will have plenty of extra site-specific interconnections in addition to these generic ones. According to http://www.shell.co.uk/news/speech/spe_beatbug.htm a typical oil rig connects up about 100 mission-critical control systems comprising 10,000 chips. If that's the case, then if the Beach/Oleson curves are at all reliable, we have to expect the rigs, as they stand, are going down. If you're a programmer, think of it by analogy with software; if 10% of the objects in your program were corrupted by some bug scrawling on memory, how likely would it be that your program would happily continue? The bigger the program, the more likely that that 10% is going to hit something vital. Now this is going to happen in every factory, every distribution network, every oil rig, every industrial plant of any sort all over the world, all at once. Modern industry is based on cyclic dependency and JIT manufacturing. Cycles and JIT amplify profit. A coal-fired generator, for example, is the cheapest source of the electricity that's used to run the trains that bring it coal - a cycle. The coal stores are kept low because that maximizes profit - JIT. When you break the cycles, when you have no stockpile of supplies because of JIT, the problem gets magnified. It's Beach/Oleson again, but now on a large scale. The effect now isn't broken factories; it's big business bankruptcies. So even if we wish away the too-deadly utility problems we're looking at a global manufacturing shutdown, big bankruptcies across the board, social dislocation as bad as 1930, and a simultaneous oil shortage worse than 1973. That's the best case. I invite Perry or anyone else to please cast doubt on the sources, the logic and the conclusions I've drawn here. This is not a happy prospect even for a paranoid twit like me. I don't believe the cities will stay safe places to live, but neither do I relish the prospect of taking to the hills with a bunch of uzi-armed Hatfields and McCoys. I'm a software architect, not a homesteader, damn it. So my strategy on surviving Y2k is different to the popular wisdom. Sure, I'll stock up, but I think there is at least one US institution that can keep internal and external order, and be there to commence rebuilding, and I aim to ingratiate myself with this organization. I aim to get associated with the US military somehow - no mean feat for a non-citizen - and have the marines on my side. If I were young and eligible I'd actually sign up for active service - even though my belly is yellow as yolk. Sure, the military is lousy with bugs too, but they have the mindset, the MREs and the guns needed to tough this out. There's a risk of war, I'm afraid, but I still suspect there's no place safer in Y2K than a military software group. Cheers, Pete. Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=10470