YIKES!! Do the doomsday computer??greenspun.com : LUSENET : TimeBomb 2000 (Y2000) : One Thread
YIKES!! Did anyone out there do the doomsday computer on y2k wired??? I did it twice..both times using very conservative numbers...and still landed in the 87.5 range for total collapse. The last time I only used 5% in most of the catagories...which is a ridiculously low %.........Wow...I'll check back with you all later...Have to go buy some more Dinty Moore!!!
-- More Dinty Moore (Not @this time.com), December 05, 1998
And, even if you use "0" per cent chance of failure as your figure for EVERY category EXCEPT "impacts of everything above" (last box), set to just "1", you STILL get a 1% chance of total collapse.
-- runway cat (firstname.lastname@example.org), December 05, 1998.
I set most of the primary failure rates to 5% except for foreign banks causing us to go under (15%) and the secondaries later in the list at 1-5 % (0% for the solar flares, etc) and STILL came up with 46-65% chance for total system collapse. Looks like TEOTWAWKI is an even money proposition. Dirty Harry to CEO's and politicians worldwide: "Considering this software design flaw is the most pervasive and potentially devasting screwup in the entire history of mankind and could blow your civilization to kingdom come I gotta ask ya one question. Do ya feel lucky, punk? Well, do ya?" Apparently, most of them said yes.
-- nope (email@example.com), December 05, 1998.
Where is this y2k wired thing? I'd like to try it out myself..
-- Leo (firstname.lastname@example.org), December 05, 1998.
See my comments on this in a post to this thread here: Newswire's Y2k Domino Engine
-- Tom Carey (email@example.com), December 06, 1998.
GIGO - garbage in garbage out. Even worse when the program is garbage. How the devil can you get a 1% chance of total collapse when everything is set to zero. This obviously does not work.
-- Paul Davis (firstname.lastname@example.org), December 06, 1998.
I think the 1% is a built-in factor based on our current Commander-in-Chief...therefore the percentage could be optimistically low!
-- Texas Terri (TYSYM@AOL.com), December 06, 1998.
sorry, found the other threads on the subject after I had posted it. The 1% could possible come from the domino effect they describe in the survey.
Still shakey though...wired even said they were shocked at what they came up with when they did it.....
-- More Dinty Moore (Not @this time.com), December 06, 1998.
You DON'T get 1% with everything zero. Operator error is the thing making the garbage here. Of course, you would have discovered that yourself if you had really done the deal, which you obviously did not.
Are you a recent public school graduate by any chance?
The statistical algorithm here is based on 'or', not 'and'. Therefore any non-zero number in any field will result in a non-zero probability.
-- Will Huett (email@example.com), December 06, 1998.
Amazingly enough, I have to side with Paul Davis (for once) and agree that this is based on hocus-pocus. I hope that people will base their beliefs on Y2K on a lot better than this!
-- Jack (firstname.lastname@example.org), December 06, 1998.
Will, try setting all to 0 EXCEPT "impacts of the above" the last category, which, if all the concretes (above) are zero you wouldn't expect to have much effect. Naive viewpoint, admitedly.
-- runway cat (email@example.com), December 06, 1998.
"Impact items include the potential failure of the IRS, the possibility of accidental nuclear launches, the GPS failure, failure of water treatment facilities, panic, the effects of Europe and Asia, and other factors."
If water treatment facilities failed world-wide or the IRS failed here, which would be the de-facto end of the US government,just to name two, then a 1% probability of world wide collapse is not beyond the realm of possibility. Without the algorithm being made public, more in depth analysis is not possible. But, what the hell, even if it were made known, the number of people out here who could understand the math is vanishingly small. His explanation is valid, and the math therein is correct.
The point is that the interconnected nature of this problem means that very small percentages on individual failure points combine to make very GREAT probabilities of complete failure.
This point is made and made well.
Any one who argues that this is invalid on its face value does not understand even rudimentary statistics and is merely being argumentative for the hell of it.
Just because you don't understand something does not make it untrue.
But you know what? Those out there that refuse to accept y2k for what it potentially is are now irrelevant. Who cares anymore? If the statistics in this site are correct, anyone who refuses to
1) admit the possibilty of life altering disruptions and 2) prepare for them
is going to die if these problems manifest.
I don't care what the pollyannas think, neither do the computers. Any one still arguing that y2k will not be noticed isn't going to live through it if they are wrong.
Now is the time to do all you can for yourself and your families and whomever God puts in your path afterwards. Milne is right, at this point if someone insists on arguing minutae such as whether or not this 1% is valid when they lack the expertise to do so, then they do deserve their fate. In fact they are begging it.
Prepare. BUY FOOD.
-- Will Huett (firstname.lastname@example.org), December 06, 1998.
Just when I was starting to feel better too!! :P I hope the computer is wrong but I agree that some of the high statistics are from the interconnectedness.
-- More Dinty Moore (Not @this time.com), December 07, 1998.
OK, here's the deal with the Y2k engine.
The impact items (last box) is not programmed as described. Instead of using the figure entered to multiply with the running total probability and then adding this product to the running total probability, the impact item percentage is simply added to the running total probability.
That is, if total probability is 30% and impact items is 10%, then total probability, as he described, is supposed to be 33%:
(30% * 10%) + 30% = 33%
Instead, he's just adding the impact item percentage:
30% + 10% = 40%
That's why when you set all other factor to zero, but impact items to 1% you get 1% on the output. Try setting all other factors to zero and setting impact item to 50%. The result will be 50%. You can even set the impact item to 110% or -20% and get some even stranger answers.
Now, the lesson learned here is two-fold. The idea behind this crude engine is valid, its implementation just needs (much) work. The second lesson is that software require TESTING, lots and lots of TESTING -- testing on boundary conditions and testing that insure the stated specifications are indeed implemented. And since Y2k involves lots and lots of software, Y2k requires lots and lots and lots and lots of TESTING. Let's hope we get it!
Another observation: every time I've run this engine, "power" always gets the worst score. This is because power has more factors than the other two core industries. In fact, I believe the relative weights of the three main factors with their supporting sub-factors can be determined by setting each factor to one and setting the last item to zero. If one does this, we have power = 8.6%, financial = 3.9%, and telecom = 3.0%. Therefore, all things equal, it appears that financial has about 25% more weight than telecom, and power 186% more weight than telecom, which about matches the impression I have been getting. Maybe this was the author's intent, but I doubt it.
The model can and likely will be improved -- don't throw the baby out with the bath water, just yet.
-- Nathan (email@example.com), December 07, 1998.
Good analysis Nathan, thanks.
If you assume the whole calculator is an exercise to shock people by showing that small persents of failure quickly add up to system failure - it does its job.
If you assume the calculator is a valid predictive tool of anything in general, or any single service in particular - it can't do that.
-- Robert A. Cook, P.E. (Kennesaw, GA) (firstname.lastname@example.org), December 08, 1998.
I guess this means that no one has discovered a reliable crystal ball for me yet, eh? sigh......... ;-)
-- Tricia the Canuck (email@example.com), December 09, 1998.
Tricia, unfortunately, no - there is no crystal ball. But I hear that there still exists, in limited supply, the magic 8 ball. ;)
-- Christine A. Newbie (firstname.lastname@example.org), December 09, 1998.