Universities are understandably anxious to pin down the precise criteria that will be applied in the 2008 research assessment exercise.
Millions of pounds will rest on the outcome, with the future of whole departments at stake, as well as the outline of the UK's primary research effort. But should consistency between panels (and sub-panels) be a top priority for the funding councils?
One legitimate criticism of the last RAE was that too little scope existed for adjusting the methodology to take account of the differences between subjects. Some attempt has been made to put this right for 2008, but one panel's flexibility will inevitably be another's inconsistency.
The quest for transparency and consistency in other forms of assessment to ward off possible legal challenges - notably in A-levels and first-degree exams, but even in teaching quality assessments - has been a prime cause of grade inflation. Those responsible for the RAE should ensure that it does not go the same way.
University administrators like nothing better than a system that can be turned to their advantage and they naturally feel cheated when it does not operate as they expected. Recent examples include the underfunding of grade 4 departments after the last RAE and the shortlived 6* category sprung on universities in the Government's 2003 White Paper. But here lies the uneasy tension at the heart of the RAE - and the reason that it will surely change radically after the next exercise. The results have to be made to fit funding requirements, regardless of the excellence of research.
The criteria to be published next month must not be misleading, but neither should they purport to be much of a guide to the practical outcome for universities. The new grading system will make it more difficult than ever to predict the impact, even after the Treasury has completed its spending review. But it would be naive to expect panels to ignore the possible effects of their judgments on their subject. However tightly the definition of "world-leading" research is drawn, there will be a temptation to find a representative share in most areas.
Few good judges expect the RAE to disappear entirely after 2008, but there is a growing consensus that a less bloated system will have to be found. Of course, there was a similar movement in 2001, but the more that universities looked at the alternatives, the more they liked the devil they knew.