Sir: George Szmukler makes some excellent points in his article (Psychiatric Bulletin, January 2000, 24, 6-10). It is indeed impossible to predict rare events like homicide, for the sound mathematical reasons he quotes. The phenomenon of retrospective distortion should be understood by everyone who is tempted to be wise with hindsight. Inquiries are very probably, as he says, a waste of time and money, repeated time and again and no more useful than an obsessional symptom to a patient with a neurosis.

In one sense, however, it is our own fault. It is easy for psychiatrists to fall into the trap, to collude with the illusion that we are effective in preventing individual tragic outcomes. The threat of such events is, after all, about the only shroud-waving potential that the subject possesses, in the battle for funds. The unit where I work is one of the most modern and attractive buildings in the country, yet its closure as part of the rationalisation of services, has produced scarcely a murmur of protest. I suspect it might be different, if the public believed it was full of dangerous people, who were only prevented from committing crimes by the skills of those looking after them.