Devise a set of rules and academics will quickly find ways to play the system. It quickly became apparent that some disciplines were cannier than others at promoting their own self-interest. In 1996, for example, the Modern Languages panels were much more generous in judging their peers than were their colleagues on the English panel. The result was that Britain was deemed to boast world-leading research in French Literature but to be distinctly mediocre at English Literature, which seems a little counter-intuitive. What the result really meant was that the modern linguists were intent on protecting their little pot of gold whereas the EngLit lot couldn’t resist biting each others’ backs. Each round of the RAE has accordingly become more complicated than the last in the attempt to outwit the “game-players”.

As I write, the 2008 competition is in full swing. Getting on for 200,000 academic books and articles are being mailed out from a repository near Bristol to the thousand or more professors who form the 70 panels that are assessing the “quality profile” of every academic department in the country. Thanks to the conversion of polytechnics and higher-education colleges into universities, nearly 200 institutions are participating. The direct cost of the previous exercise, in 2001, was £5.2 million (against an initial budget of £3.6 million): a substantial sum, though only one per cent of the amount of money (£5 billion) distributed on the basis of the results. The indirect cost of the whole process — in the time spent on the exercise by the departments making submissions and the panellists reading all the work — is beyond calculation.

The current exercise will be the last in the present form. In a classic Gordon Brown move, the death-sentence of the RAE was pronounced in a half-sentence gabbled out in the middle of the 2006 budget, with no prior consultation of the profession or even, it seems, the quangocracy. Having been made progressively more complicated, the RAE will be replaced by a radically simplified funding distribution system based on “a basket of metrics”. The idea is to do away with the peer-review process and instead crunch some citation indexes through the ministerial computer.