Hi Folks,

Charles Perrow, an expert on complex system failures and former Three Mile Island investigator, wrote a fascinating book back in 1984 called "Normal Accidents". In it, he proposes that systems have two primary characteristics that determine their accident propensity:

1. How much slack time they have to fix problems without catastrophe (he calls this "loose coupling" (slack available) vs "tight coupling" (no slack))

2. How complex/interrelated the systems are (linear vs. complex).

He maintains that complex, tightly coupled systems will have "normal accidents". In other words, there will be events, possibly rare, where multiple unexpected simultaneous failures will coincide (the complexity part) AND there will be insufficient time for the humans to figure it out (the "tight coupling" part). He calls them "normal" because they are fundamentally unavoidable.

Then, in the book, he suggests that those technologies that will have "normal accidents" should be assessed for their catastrophic potential. If the potential is high enough, and if their are reasonable alternatives, then these systems should be abandoned. In the 1984 book, he put nuclear power and recombinant DNA in these categories. (two years before Chernobyl)

He is apparently working on a new edition of his book which will include a chapter on Y2K. A draft of the new Y2K chapter is on the web at the address below. He isn't very big on predictions, but rather, anxiously anticipates the event with academic curiosity to see just how "tightly coupled" our society really is.

Worth the read if you are interested in the Y2K complexity problem. And the original book is excellent.

Here's his final paragraph from the new chapter (http://azstarnet.com/~nuu/Other_DOCS/charles_perrow_2.htm) :

"Y2K is that rare potential disaster that we can see coming, plan for, and prognosticate about. Predictions follow the expected pattern. Those at the helm are optimistic, social commentators are pessimistic, and most social scientists duck their heads and fail to exercise the predictive skills their learning should call for. Properly viewed, as an opportunity to test theories about organizational mind sets and organizational interests in all three cases, lining up the players and the predictions, and more important in the case of social scientists, to test our theories of crisis management (from denial through preparedness to coping), the importance of the newly enlarged global economy, and the "mechanical" and unforgiving coupling of society that runs on cheap oil and electronics, versus the "organic" coupling of a dense web of resilient devices and human relationships, this is an event we should profit from, even though we dread it. Its potential scale and scope dwarfs all other "normal accidents" discussed in the book. But the tools discussed in Normal Accidents would appear to be among the most relevant available."

-- Rick

-- Rick Stahlhut (stahlhut@net-link.net), June 18, 1999

Answers

Thanks for the link Rick. I just read it in it's entirety (longggg!). Although I believe Charles Perrow's original book would make an interesting read, I have to say I found the content of this link to be rather "unconnected" from what I would expect in a revised version "draft". What started out as a promising and interesting analysis of Y2k's "catastrophic potential", simply became a re- statement of what many of us already know (both Polly and Doomer versions). I would have really enjoyed it if he had followed through with some realistic predictions/conclusions. A "heads-up" would be appreciated if you happen across his final draft Rick. Thanks

-- CD (not@here.com), June 18, 1999.