Know the rules well, so you can break them effectively.– Dalai Lama XIV The golden rule is that there are no golden rules.– George Bernard Shaw

Of course my rules are necessary, proportionate and flexible, it is your rules that are the problem, being arbitrary in nature, excessive in force and of course far too numerous, all of which lead to unintended consequences.

Rules like so much else in medicine are a quantitative science and until the far off day of every fact being known allows reliable rules to be formulated, then like an elevator in a skyscraper, the trick is knowing on which floor to get off when the current known rules stop applying to a situation. Hence this article opposes at least in part the paper in this issue by Blakey et al.1

PubMed yields 18 962 results where ‘rule’ or rules’ are in the title and in our comparatively enlightened trust there are about 160 clinical guidelines varying from 2 to 274 pages and some 475 other trust policies and guidelines. Despite this, there are so many situations where there simply are no rules.

The first reference by Blakey et al concluding that algorithmic predictions are superior to clinical judgement2 is based on studies from 1944 to 1989 only and covers everything from the prediction of coupon utilisation from mail order catalogues via the risk of malingering to the ‘diagnosis’ of homosexuality! Of the few papers cited published between 1980 and 1989, clinicians predicted intensive care unit mortality and the diagnosis of abdominal pain better than algorithms but clinicians were worse on diagnosis of myocardial infarction and chest pain. The enthusiasm of Blakey et al for new technology and the papers they cite undoubtedly shows better capture of data nowadays but the outcomes remain the …