THE RESEARCH Assessment Exercise announcements on grading and funding seem to have been followed by a certain tristesse which, combined with the angst of waiting for the Dearing report, has led to questions about whether the exercise should ever be repeated. A consultation on its future is imminent.

If we ask how it was for the various partners, the previous government must have liked it: performance grades were ratchetted up without extra expenditure. For the funding councils the exercise allowed for informed allocation of funds, so that the councils became strategic actors controlling, if not steering, the system. It gave senior managers in institutions a lever for internal review and change. Even many of the poor infantry liked the exercise in principle - it is the way it was done that causes them problems.

The favoured alternative to another research assessment exercise (being urged by the post-Thatcherite fat cat universities) is to transfer money from the funding councils to the research councils. I want to argue against that, but also urge some changes to the way the exercise is run.

The nature of universities as defined in the United Kingdom demands that they do some research. I do not argue that all individual academics should do research but institutions must. If not, the link to Ernest Boyer's scholarship of transmission of knowledge is broken. This has hitherto been seen as essential and, indeed, is vaunted in the publicity for their undergraduate admissions by the very universities now trying to deny funds to the majority of the university system.

ÖйúAƬ

Of course, it is done differently elsewhere, but in the UK only about 30 per cent of government research and development goes to universities; only about 10 per cent of total R&D goes to them. That is not true elsewhere and you cannot pick bits of an alternative model.

There is a risk that if all the research funding were transferred to the research councils, fewer universities would gain access to some funds. The RAE, for all its faults, rewarded non-traditional entrants who had succeeded. It was also, in 1996, more open in many ways in its processes than research council schemes. There is a risk too, that under the DTI, the research councils' view of research may narrow to fit a political agenda based on a view of universities as tools of the state economy. A diverse system will benefit from, and be protected by, diverse funding channels.

ÖйúAƬ

A second reason to keep the research assessment exercise and its funding is related to the individual, professional academic. New knowledge is like a butterfly - you do not run in straight lines to catch it. Yet there is pressure - over PhD completion rates for instance - for straight-line progress through projects. The serendipity that independent curiosity allows is a valued element of research. Local judgements by informed professionals are part of a research enterprise. If that self- determination is removed, the denigration academics have suffered over recent decades will be manifested by another action to diminish their practice. Many of us still preserve intrinsic commitment to disinterested work which does not fit well with the extrinsic incentives of competitive bidding against someone else's agenda priorities. A contract culture is a control culture and there is a risk of state corporatism subordinating creative individuals - who will either be insubordinate or be less productive, or opt out.

I want also to argue against even more concentration of funds via the RAE-linked formula. The 1996 results show how much added value the modern universities produce in research with relatively meagre funds. Some work last year in Australia suggested that there, the modern universities gave better value for money than traditional ones on a prestige publication productivity count. It depends what you want: the marginal gain from increased investment in 5-star departments may be negligible, but, as shown, can raise some newcomers to grade 5. The risk of over-concentration is that most future researchers are selected on the basis of A levels that get them onto the ground floor of the favoured few and stay within that ring-fenced laager throughout their career.

Some things do, though, need improving: * the RAE needs to relate to other quality assurance activities, particularly of teaching.

* the difference between the RAE and other research initiatives needs articulation; * in England, more openness is needed, as last-minute changes to funding undermines planning; * there needs to be increased recognition of a wider view of research; * some universities need to diversify their research strategy.

ÖйúAƬ

The 1996 RAE was better than in 1992. It can still be improved, but has given rewards for effort as well as elite status. There cannot be a major change without damaging consequences.

Ian McNay, head of the Centre for ÖйúAƬ Management at Anglia Polytechnic University, has just published a HEFCE-funded review of the impact of the 1992 RAE.