An interesting book this is… Sagan is a professor of political science at Stanford University and in this 1993 book (that actually was elected the Best Book from the Science, Technology and Environmental Studies Section of the APSA) he studies how safe the handling of nuclear weapons by the US armed forces actually is. There has never been a major accident or accidental explosion, so generally the handling of nuclear arms is regarded as an example of high reliability operations with an outstanding safety record. But is this optimist view justified? Based on new and formerly classified information Sagan’s book (he has done some awesome research by the way) sheds some very sobering light on this subject and is interesting from both a historical and safety perspective…

Baffling (but, to be honest, fully expected as well) was also the light that is shed on official reports to high placed politicians, Congress and (let’s face it) us: the tax payers and rest of the world. Incidents are brushed under the carpet for political or other reasons and formal evaluations are often reduced to clouded or very positive statements aborting the ability to actually learn anything (and improve safety)!

The book takes root in two known - and seemingly opposing - theories within safety (management): the Normal Accidents theory (Perrow et al) and the High Reliability theory (see e.g. Weick et al) and applies these to a number of accidents, incidents and near misses with regard to handling of nuclear weapons and response/warning systems, judging both theories on their merits and drawing some conclusions on the state of safety regarding nuclear weapons handling.

The funny thing - for me at least - is that I think that Sagan does a way better job of explaining, defending and making palusible the Normal Accidents theory than Perrow, and where I advised in my review of “Normal Accidents” to try to find a good summary of that book: here is one to be found in Chapter 1! Also I have the feeling that Sagan has a less fatalistic outlook on some things than Perrow has at times (whose advise is after all to quit some things altogether). What the book fails to mention (or at least acknowledge) in Sagan’s discussion (and application) of High Reliability is the ability of HRO’s to react well and quick on unexpected events and then return to a stable steady state. Good “firefighting skills” are after all and essential part in Weick’s book. I wonder if this would have changed the conclusions had Sagan spent more thought on this part of High Reliability.

Striking for me was that I have a greater relation to the Normal Accidents theory than I wanted to admit after having read Perrow’s book, but probably that’s all about agreeing with the spirit of an idea and not necessarily subscribing to the actual words it is brought with? After all, I could have known: most safety experts have after all the continuous feeling that if something can go wrong at one point of time it will go wrong…

This is no reason to act fatalistic, of course. Above I wrote that Normal Accidents theory and High reliability are seemingly opposing. Opposing because the former has a rather pessimistic view on safety, the latter a rather optimistic point of view. I chose, however, to call this “seemingly” because I am convinced that Normal Accidents theory and High Reliability theory are in fact not excluding each other. It’s rather a case of being two sides of the same coin. Because Normal Accidents theory is true (accidents can and will happen - a statement back up by evidence in this book) it needs High Reliability elements to keep the chances of bad things happening as small as possible (something that is backed by this book as well - no serious accidents with nuclear weapons have happened, after all. They could have, but they didn’t.). So my conclusion is that it’s not a case of choosing between the two, but using them both - with a healthy dose of common sense! It’s not that just one of the theories is true: they are both to a certain degree and therefore are rather complementary. One of the problems I have with Perrow’s Normal Accidents theory is that the theory does not provide many solutions (apart from the cautionary principle), but only problems (or, at least a warning). There comes the need for High Reliability theory. They have some solutions but should maybe retain a slightly larger skepticism towards the effectiveness of these…

I already praised the thoroughness of Sagan’s work. I would also like to give special mention to the clarity he gives when pointing out what part of the book is actually based on facts and where/when he begins to work with assumptions and speculations - something that is unavoidable, of course, because information is still partly classified and actual accidents (in the sense of nuclear detonations and wars) have not happened yet (thankfully). 

The book concludes with a very useful discussion of recommendations. Very sensibly done and overall I regard this book as highly recommended reading with some good and important thoughts on both safety theories and elements like the Garbage Can model (conflicting goals and politics within organizations), limitations to organizational learning and the like.

To quote another review: “This book is a significant contribution to … international security studies, organizational theory, and risk analysis.”

ISBN 0-691-02101-5