Here's my idea...

The federal government of the great republic of the United States of Americe needs to do the following:

Establish Friday, 29 October, 1999, as the "end-to-end" test date. At 9:00AM PDT, all systems will be set to 5 minuted before 12:00PM 31 December, 1999. The systems will run for 5 hours without rebooting.

Since the "small" glitches we are to expect can be easily fixed, and we are not to have any problems that are greater than a "bump in the road", the recovery should be accomplished by the opening of business on Monday, 1 November, 1999.

The weekend could be used to bring all systems back on-line, unscramble any data, and determine any fixes needed.

It also gives our great Republic 61 days to contemplate the year 2000.

Feedback please...

-- Uncle Bob (UNCLB0B@Y2KOK.ORG), September 24, 1999

Answers

Uncle Bob,

While it really sounds great, it for the most part is undoable. Let me explain the way things are from my point of view. I have not been able to take any time off this year due to Y2K. I just finished pulling a 42 hour weekend pull (2 days) for my companies y2k testing. Right now if they told me we were going to do it all again right away, I think that I would just hold up my hands, shake my head, and walk out the door to never return. We really do need a break before the "fun" begins at new years evil. What do you mean to do... kill us before the said "fun" begins? We do need a break in order to rest up for the long run. Oh yeah, my company doesn't even begin to think about their contengency plan until December.

-- (cannot-say@this.time), September 24, 1999.

Great idea, never happen. A shame because this really is what should be done. It's just to simple, and there's to much up side. If the shtf you could just set the date back and the country could work like hell for 2 months and get things fixed. Oh well, we elect them.

-- (ericnamy@yahoo.com), September 24, 1999.

I dont think it likely, Uncle Bob. Herre's my guess why:

Theres too much of a chance of leaving systems open and vulnerable, of giving away our only advantage here: that the whole world is facing the same problem on the same day. If we do have weaknesses and problems, it leaves us open to others before they face the same. For example, DOD weapons systems-if theres glitches, it wouldnt take long before the decreased capability was exploited in that window of opportunity we would have provided.

While its great if something could be done, I just dont think it will happen.

-- LauraA (Laadedah@aol.com), September 24, 1999.

I asked the panelists a question like this back in February (I think it was) at the Washington DC Y2K Group meeting. I wanted to know if I understood correctly that the only time the grid would be tested would be New Year's Eve. After some bobbing and weaving and hemming and hawing Mr. Jon Arnold and Mr. Don Evans basically acknowledged that my understanding was correct. Evidently it is not possible to get regulatory approval to do this, and the engineers evidently cannot figure out how to do it anyway. There is NO plan to test the grid. All they intend to do is test the individual components (wire, transformers, generators, computers, etc. one at a time).

This, of course, is madness.

Don't agree?

Visualize your car, with bad brakes. The mechanic says he tested the pads. He tested the hoses. He tested the connections. He tested the bolts. He measured the disks and drums. He checked the brake light bulbs. He put everything together. But he did not start up your car, drive around the block, and see if the brakes will actually stop your car. He gives you the keys. Doesn't the thought cross your mind: "Hey, bozo, YOU did the repair, suppose YOU do the test drive!"???

And besides, suppose someone got hurt in your Y2K prequel? Think of the lawsuits!

And, anyway, Bob, what is the point of testing a system which is functioning perfectly? (g)

Too bad the Bump In The Road theorists do not have the courage of their convictions, or they would take you up on your excellent suggestion. What a great way to silence the doomsters two months early, and stop this Y2K hysteria in its tracks, right? The suspense is driving some folks nuts anyway, so let's have the rollover at the end of October and be done with it! Then we can enjoy the party on the Mall without any nagging doubt about whether we will get home alive! Wow!

-- Joseph R. Whaley (whaley@attorney-us.com), September 25, 1999.

We all know this is a mental exercise only because the logistics are impossibly complex, and it wouldn't be PC to demonstrate that the emperor has no clothes. But let's play along anyway:

Anyone who has been half awake and has his share of common sense knows that the 'network' is DOA, and an October test certifying its demise two months early wouldn't be very attractive to anyone but this tiny group of GI's.

Hell, if word got out it may even scare some of the slumbering masses, but since the media would never accurately report such a failed test, there really is NO technical or political benefit to do it.

-- Dr. Roger Altman (rogaltman@aol.com), September 25, 1999.

Won't happen. Focus is on calm normalcy to protect Christmas sales so many businesses depend on for a good portion of yearly sales.

-- Leslie (***@***.net), September 25, 1999.

Uncle Bob,

I'd like to improve your hypothetical in the sense of reducing the public safety risks. Instead of everyone doing the test, how about a truly random 5% sample across every industry in October. In this manner, if things go wrong, it won't break the nation, government can provide needed intelligence, support, and financing to the risk takers and those depending on the risk takers, and we'll have some actual lessons learned under the belt.

Yes, I realize that it's not going to happen. And that's a big bummer.

Sincerely, Stan Faryna

-- Stan Faryna (faryna@groupmail.com), September 25, 1999.

While Stan's suggestion tries to overcome some of the logistic impossibility, it would still be spotty at best. And very expensive, and we probably don't have the resources. But if it were possible, it would be extremely useful and informative for everyone.

Since the exact time and date of many transactions (financial all the way down to the level of temperature sensors) is critical, what you're talking about seems to be some gigantic network of interconnected time machines, and hooking a huge number of standalone enbeddeds into this network is problematical. And ALL of these transactions (probably billions per second) would need to be duplicated onto this time machine network. A daunting task.

As for the brakes analogy, this is what happens all the time. You complain that your brakes squeak and don't stop you well. The mechanic finds that the pads are worn away. He replaces the pads (and turns drums or straightens and aligns rotors or whatever). He doesn't really test them -- he found the problem and fixed it. The vehicle seems to stop quietly on the lot. Almost always, this is sufficient. Sometimes the mechanic misses something, and sometimes that something would have shown up in an extensive test. We live with this, and it works fine often enough for us to find it fully acceptable.

So even if the economy-numbing expense could be afforded, and even if we had the many years it would take to set it all up, all we really accomplish is to try to calm the fears of those who have already made it totally clear that they wouldn't trust it to begin with. So what's the point?

-- Flint (flintc@mindspring.com), September 25, 1999.