Last week's Pre-Budget Report was going to bring down the curtain on the research assessment exercise and set a new direction based firmly on metrics. In the event, it did (at most) only half the job. The RAE will go after 2008, but something that bears a passing resemblance to it will take its place outside science, technology, engineering and medicine (Stem). And even in those subjects, there is a long way to go before agreement is reached on the precise metrics to be used and the degree of oversight to be applied to the results. There is understandable frustration that the long deliberations that produce results in 2008 will so soon be overridden. But those who champion the current system are fighting a losing battle. The Higher Education Policy Institute is right to expose the inefficiencies in research council allocations, but the political wind is blowing in this direction. There is no prospect of restoring the RAE but every possibility of shaping its successor - even in the Stem subjects, where the advisory panels may become more influential than the Treasury statement implies. The concerns expressed by the learned societies this week - that the new arrangements might widen divisions among universities and channel efforts into "safe" areas of research - are serious. They can be addressed only by full engagement in the development of a system that might look unstoppable but is still far short of the finished article.