### Probability of Systemic Failure: Another Point of View

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

There has been a number of posts that analyze the probability of systemic failure by concentrating on a few key industries and postulating that if any one of them were to suffer a critical failure, economic paralysis would result. While this thesis may indeed be true, the debate is diverted to the actual probabilities of success one ascribes to each of these key industries. The net result is that the debate quickly gets bogged down into a futile number crunching exercise because these probabilities are essentially unknown.

I think that the probability-of-systemic-failure calculation, may be of more practical value by searching for those independent economic (and even social) ACTIVITIES that form essential "spokes" that keep the economic wheel balanced and spinning without a wobble. So just like a wheel could keep on turning upon losing a few spokes, it would quickly become unbalanced and begin to wobble which would quickly lead to catastrophic physical (and economic) failure.

The benefit of approaching the calculation this way is that we put the greatest emphasis on the SYSTEMIC nature of the problem which, in fact, goes to the heart of y2k, and the least emphasis on the probabilistic aspects which, I think we would all agree, is the least understood. Then by assigning very optimistic success probabilities to each of these "spokes" it can be shown that we are facing a huge risk of severe economic deterioration and eventual collapse. There are only two selection criteria for picking an economic (or social) activity to become a "spoke"; it should be independent of all others, and should be considered a vital economic activity. Given below is my guess as to which activities would qualify together with my optimistic "Success Probabilities". Note that the chance of at least one critical failure is 62%. If I assign my "best guess" probabilities that number rises to over 99%.

CRITICAL FAILURE OPTIMISTIC SUCCESS PROBABILITY Financial 92.17 banking 98 Fed wire transfer 99 tax proceeds (IRS/Customs) 95

Power 96.05 generation 98 transmission 99 distribution 99

Telecom 98.5 Food Production 98 Water Delivery 98

Waste Disposal 97.02 sewage 98 solid waste 99

Transportation 88.52 rail 95 truck 98 surface docking facilities 98 air 98 pipelines 99

Government Oversight 91.23 Military/National guard /police/fire protection 99 FAA 97 All other support such as SS, Medicare, Medicaid, etc. 95

Extraction and Mining 96.06 coal 99 gas 99 oil (domestic) 99 metals 99

External Factors 66.34 oil importation 95 foreign banking 95 " manufacturing 95 " air traffic control 95 " docking facilities 95 " surface shipping 95 " ground shipping 95 other critical raw material imports 95

Public Support of Infrastructure 91.75 urban rioting 97 bank runs 97 job lockout (from bankruptcies or fear of leaving home) 98 war/terrorism/series of natural disasters 99.5

OVERALL PROBABILITY OF AT LEAST ONE CRITICAL FAILURE: 61.6%

-- Dr. Roger Altman (rogaltman@aol.com), January 15, 1999

Sorry, my table got all re-arranged upon transmission. To understand it now, read the first row entry as the general category. For instance, "External Factors 66.34" is the subheading together with the OVERALL "Success Probability" for all the items TAKEN TOGETHER UNDER THAT SUBHEADING. Following after that are each category for that subheading. So, continuing with this example, oil importation 95%, foreign banking 95%, foreign manufacturing 95%, etc.

Oh boy, computers, computers...

-- Dr. Roger Altman (rogaltman@aol.com), January 15, 1999.

All I see here is more BS from someone who likes to hear themselves talk. Is anyone else in favor of banning further posts on Time Bomb 2000 until September? I mean, there is only so much postulating we can take before we explode like a tick.

-- Mojo Rising (Yesca@HT.com), January 15, 1999.

Mojo, feel free to drop off the face of the earth. Everything is speculation which is one of the greatest benefits of this forum. Posts on either side of the y2k thin, red line are welcome and essential.

Your ability to lurk or post here is your decision and no one is twisting your arm or banging your head against the wall to force you to stay.

Dr. Altman, I appreciate your work and I thank you for your post.

Mike =================================================================

-- Michael Taylor (mtdesign3@aol.com), January 15, 1999.

oh heaven's no, Mike, I want to watch Mojo explode like a tick!

Arlin

"...the least emphasis [is] on the probabilistic aspects which, I think we would all agree, is the least understood."

Roger, I disagree. The nature of complex systems' failures is not well understood.

Consider the following from Margaret Wheatley's Leading Through the Unknowns of Y2K:

"Complex systems failures share a set of distinguishing features:

• the longer they unravel, the more extensive their effects
• costs always far exceed what has been budgeted for amelioration
• as effects materialize, unknown interdependencies become visible
• the more that problems come into focus, the fuzzier they appear
• past experiences with simple systems dont apply
• cause and effect are impossible to track
• consequently, there is no one to blame
• These features describe the most frightening realization about complex systemic problems: They are inherently uncontrollable. They cannot be understood sufficiently either before or even while they are occurring, therefore prediction and control are impossible. Traditional approaches to management simply dont work."

Also, the suggestion to search for those independent economic (and even social) ACTIVITIES that form essential "spokes" that keep the economic wheel balanced and spinning without a wobble, discounts the invisible interconnecting processes that structure an economy, an ecology.

So, I ask you, kind sir, "Might there be something missing from your algorithm?"

and

Is there an activity within the "ACTIVITIES" that allows order to unfold?

~C~

-- Critt Jarvis (Wilmington, NC) (critt@critt.com), January 16, 1999.

I am replying to Critt Jarvis' comments. First, thanks for pointing out Margaret Wheatley's article entitled, "Leading Through the Unknowns of Y2K". I think if you re-read her article you will find that in large part what she is referring are the unknown complexities and interrelationships related to SOLVING the Y2K problem. However, she does make some reference to economic activity as another example of such incomprehensibe complexity that no one expert can really understand the entire system.

So I cannot understand how you come away thinking that we know MORE about probabilities than I expressed in my original comments. Since her article barely mentions economics it is not particularly relevant to my point, and to the extent that she uses it to bolster her 'complexity' message, it boslsters mine (unknown probabilities) as well.

To get back to the point: probability is not the issue because, as I have demonstrated by setting obsurdly optimistic probability-of-success values, the SYSTEMIC nature of the y2k issue VIRTUALLY GUARANTEES that we will have critical failures. Now once a critical failure occurs then all those "hidden complexities and interrelationships" that Wheatly refers to will surely rear their ugly heads to exacerbate and compound existing problems which will RAPIDLY accelerate economic collapse.

To site just one example: if sewage is not removed from urban areas, how long would it take to precipitate a severe emergency through rioting, and disease? And because death will rapidly follow, how long will it take disease to spread to surrounding areas? Do we need to know what probability sewage plant have to succeed with their y2k remediation? No we don't because a calamity that I have just described can come from almost ANY SOURCE. The point is WE ARE EXTREMELY VULNERABLE TO A CALAMITY. Let that anthropologists of the twenty second century pick apart our remains to figure out how it all started.

-- Dr. Roger Altman (rogaltman@aol.com), January 16, 1999.

I have an idea... Why don't you all go and try to figure the probability of which one of you Fat Heads will hit the bottom of a canyon floor if you all jump off a thousand foot ledge at the same time? I guess the only real way to be sure is to have you do some "theoretical testing".

The Lemming bus is now leaving...

-- Mojo Rising (FatHeadsBegone@Y@K.com), January 16, 1999.

Mojo: thank you for your detailed analysis on why there won't be a problem. Now tuck your head back in your ass and go off to sleep.

-- a (a@a.a), January 16, 1999.

Mojo - they'd all hit the floor at the same time BOZOID (Newtonian Physics) - didn't you learn anything at school yesterday???

Mojo drooping more like...

Andy

Two digits. One mechanism. The smallest mistake.

"The conveniences and comforts of humanity in general will be linked up by one mechanism, which will produce comforts and conveniences beyond human imagination. But the smallest mistake will bring the whole mechanism to a certain collapse. In this way the end of the world will be brought about."

Pir-o-Murshid Inayat Khan, 1922 (Sufi Prophet)

-- Andy (2000EOD@prodigy.net), January 16, 1999.

Oh! Forgot to add, if you jumped off at the same time Mojo Drooping, you would hit the floor LAST because you are an AIR HEAD (Mojoian Physics.)

Andy the doombrooder :)

"We're doomed I tell ye, doomed!"

Private Frazer, Dad's Army, Walmington-On-Sea Home Guard, 1939 (Undertaker)

-- Andy (2000EOD@prodigy.net), January 16, 1999.

Double </b> to correct for earlier <b> where there should have been a </b>

-- No Spam Please (anon@ymous.com), January 16, 1999.

Whatever you said/did Thanks. Some people just won't pick up after themselves.

S.O.B.

-- sweetolebob (La) (buffgun@hotmail.com), January 16, 1999.

Thanks no spam, I'm new to this stuff!

-- Andy (2000EOD@prodigy.net), January 16, 1999.

No spam, can you help me out. You recently posted a very good example of a banking transaction, a very detailed one - I have been trying to find it, can you remember which thread it was on? Many thanks if you can.

Andy

-- Andy (2000EOD@prodigy.net), January 16, 1999.

S.O.B.,

Oh, Andy just made one teeny little typo.

- - -

Andy,

Two HTML-newbie suggestions from others in this forum:

(1) As soon as you post something with HTML, exit back to main page, then click on the thread again to see it with your new entry, and check whether looks all right. {I started doing this ... mmmmm ... sometime last month. :-}

(2) When you're going to post something with HTML, first write it in a file on your own disk with some editor, then open that file with your browser to see how it looks. If OK, copy to submission window.

- - - -

Andy (cont.),

Creating that example was fun.

Am now searching ... searching ... searching ...

-- No Spam Please (anon@ymous.com), January 17, 1999.

Andy,

>example of a banking transaction, a very detailed one - I have been trying to find it, can you remember which thread it was on?

Will noncompliant systems reinfect compliant systems at http://www.greenspun.com/bboard/q-and-a-fetch-msg.tcl?msg_id=000Llt

-- No Spam Please (anon@ymous.com), January 17, 1999.

Hey thanks No Spam - I'm resaerching an article I'm writing on the imported data problem and may use your example for discussion if that's ok with you.

Thanks, Andy

-- Andy (2000EOD@prodigy.net), January 18, 1999.

Andy -- Okay.

-- No Spam Please (anon@ymous.com), January 18, 1999.

There's another way of looking at this "Death of a Thousand Cuts" scenario. Notice that the analysis, as presented, makes this very unlikely (and hidden) assumption: none of the experienced failures will be repaired.

I don't think that's the case. Certainly some failures will be repaired. How many? How soon? Who knows?

Given this completely unquantifiable variable, the calculation provides no new information. The probability of a critical failure, somewhere, sometime, of unknown duration, remains indeterminate.

-- Tom Carey (tomcarey@mindspring.com), January 19, 1999.

Tom:

Thanks for your response. It's a good criticism, but don't be tempted to dismiss my attempt to quantify the "death-by-a-thousand-cuts" scenario too quickly. Remember, the model represents a "snapshot" of the "probability-dynamic" caused by the SYSTEMIC nature of y2k. Sure there will be repairs. After all, with every passing day the world is incrementally closer to fixing all y2k problems. But the key question you should think about is the following:

Given the overwhelming probability that there will be critical economic failures as we move into the next millennium, will the TREND of the probabilitiy-of-critical-failure FOR THE ECONOMY AS A WHOLE worsen or improve?

I submit to you that if I were to prepare new data similar to the type I prepared in my original comments, the average of the "Probability-of-Success" values in that table would be LOWER not higher (which is really what you are implying -- otherwise what you're saying makes no sense at all) for the following reasons:

1) Non-productive (meaning military, police, firefighting which are expense items) segments of our enonomy would be put into force at a much greater rate than normal for crowd control and security, which would create a much higher than average economic drain.

2) An initial economic dislocation of the magnitude I forsee would have a significant impact on bankruptcy, worker productivity, job absentia, all of which lower productivity.

3) To the extent that y2k hits computers directly, productivity will be directly affected ("just-in-time-inventory-control" is but one example).

4) Trade and commerce will slow down drastically, particularly those elements dependent upon overseas goods (and who isn't these days). Thus, vast sectors of our economy will be extremely vulnerable to failure.

5) Consumer confidence is likely to take a drastic hit (stock market plunge) which will restrict economic activity, raising unemployment and reducing corporate earnings.

6) Even a FEW key corporate/government agency failures are likely to have a domino effect on the rest of the economy ("division-of-labor" argument).

7) The ability to continue fixing y2k problems will be substantially more difficult because programmers will be forced to work under "fix-on-failue" conditions -- a repair mode with HUGE disadvantages.

8) Even successfully operating contingency plans will significantly reduce corporate/government agency productivity (by definition, at least) which will further hamper economic activity.

There are many more reasons I could list, but please don't miss the point here. While no one can predict the EXTENT to which the items I've enumerated will effect the probabilities in question, they WILL BE AFFECTED and to argue otherwise puts you in the wishful thinking camp. Now, if that's where you want to wait out the rest of 1999, well, that's your business, but an argument such as that serves no useful purpose on this thread. 5)

-- Dr. Roger Altman (rogaltman@aol.com), January 19, 1999.

Yes, but aside from that, Doctor...?

Actually I think you're right on this. I think stuff will happen, and in a serious way.

I just don't think it's possible to assign a numerical probability to any of these contingencies. And qualitative estimates, whether yours, or mine, or Milne's, or Bilbo Baggins', characterize the estimator more than they do the extrinsic situation. This is not a fault. And Pascal's Wager still holds.

-- Tom Carey (tomcarey@mindspring.com), January 21, 1999.

Tom:

I thought the uncertainty of the probabilities was the point. By assigning ridiculously high "Probability-of-Success" values to each economic sector, and still have the overall number come out indicating the likelihood for at least one of these sectors will fail IS BETTER than bouncing qualitative scenarios back and fourth.

-- Dr. Roger Altman (rogaltman@aol.com), January 21, 1999.