## Note 7 Probability and disasters

In the section on Public Predictions, I wrote 'The longer it has been since the last major air crash made headlines, the more likely it is that one will occur soon'. This prompted correspondence from several readers concerning probability theory. Chief among these was Alan Jackson, with whom I enjoyed some very fascinating correspondence. Alan's main point was that if an event is truly random, in the strict sense of the word, then the laws of probability tell us nothing about whether it will or will not occur at a specific time (this is what it means to be 'random').

This point may run counter to some people's intuition and 'common sense'. Some people who play roulette feel that if 'red' has not come up for a long time, it is some how 'more likely' to come up on the next spin. This is not true. The chances of red coming up remain 50/50, and previous spins makes no difference. If one regards airline crashes and similar disasters as being truly random events, then what has occurred in the past makes no difference to the probability of another such crash occurring soon.

However, it can be argued that some kinds of disasters are not strictly random. It can be argued that major airline crashes are the consequence of what must be relatively constant factors within the airline industry (operator error, parts failing, technological glitches etc.). If this is granted, then one can calculate the average number of major crashes which occur over a given period of time, and then look for statistical deviations. If the actual number is measurably below average for a given period, then an 'airline crash' prediction is more likely to come true than at other times. The further the deviation from an established average, the more likely it is that the prediction will come true. Of course, this view involves assumptions which are open to question.