In standard management analysis of a decision, management tries to make an intelligent estimates of the risks and rewards of any particular decision. So today we take an elementary look at decision-tree analysis of the risks of war with Iran, using open-source literature, to try to get a handle on a problem that Sy Hersh's reporting indicates is very urgent.
We are still in the book Computers in Battle, will they work, edited by David Belkin et al, 1987. Today's installment focuses on a couple of insights from the chapter by Alan Borning, p. 101 ff, on "Computer Systems Reliability and Nuclear War." Check this out.
Borning notes that for all practical purposes both the US system and the Russian system must be viewed as one system, simply because the reaction times are so short, and that the risk of a malfunction rises as unintended mistakes enter either system. (cont'd)
Let's focus first on some details of the reliability of the US system. On page 129, he notes the unsettling statistic that in 1977, or 29 years ago, an exercise of the NORAD system (North American Defense) exhibited a reliability factor of only 38 percent. Put a different way, that translates into a 62 percent chance of failure. Hopefully almost three decades of technology has done something to increase the odds of success, but a look at the problem from the view of experts here is certainly enough to raise a concern among the technical community, as well as the general public.
Then there is the chance of problems with the system operated by Russia at the time (now the former Soviet Union states). Without giving exact statistics, author Borning notes that alcoholism, drug abuse and those with potential psychologica problems have a certain statistical probablilty of ending up in jobs with nuclear strike responsibility. Think Absolut vodka.
Add to this the complexity of the two systems, and trying to react, with decision times greatly shortened by the speed that computers can bring to first-strike and reaction strike capabilities. In 1987, the reaction time was as short as 12 minutes. In practical terms, this means that the decision period in a nuclear conflict could be over in 12 minutes. In any given hour, that translates to five times an hour, or in any given day, 120 periods per day, where the system could decide a conflict one way or another.
To raise a legal problem, Borning notes that the decisions have to be made so fast that it can operate actually without constitutional authorization. He details Jimmy Carter's Presidential Directive 59, and Reagan's Defense guidance (p. 154) to show that the policy was that nuclear forces MUST PREVAIL. But he also notes that in many circumstances Congressional approval is impossible, and even presidental approval may not be possible. Worth reading.