An excerpt here:
I saw a great deal of checklisting in my career. For example, when the prison department grew alarmed at the number of suicides in prison, or perhaps I should say at the publicity being given to the number of suicides in prison, it developed a form that in effect was a checklist, to be applied to any prisoner thought to be in the least suicidal or likely to attempt suicide (not quite the same thing).
The purpose of this form, I soon discovered, was not so much to prevent suicide as to prevent criticism after the suicide had taken place: For if the form had been filled correctly, it was possible to argue that all that could have been done to prevent it was in fact done. This kind of magical thinking was accepted more or less wholesale by the courts investigating the suicides of prisoners.
In fact, the form was even more cunningly designed than this suggests. It was of such complexity that it was almost impossible to fill it in correctly, or at least sufficiently. So that if blame had to be apportioned, which is to say if negligence had been incontestable, it could be apportioned to the last person who had not filled his little bit of the form properly. It could then be claimed that, but for this omission, the suicide would not have happened; and the chief glory of the form was that the failure to fill it adequately was almost always by someone very lowly in the hierarchy.