What price knowledge? Richard III p < Nazis y = research grant (1 - q)

Apparently Research Councils UK has changed its mind. It had held that "we want to get to the stage where we can say with an investment of x we get a return of y", developing an algorithm with which to test the likely economic benefits of research, including research in the arts and humanities. It is rather sad that it has lost this ambition, since it would certainly have added to the gaiety of nations to see the result.

I had imagined a scenario like an historian seeking a grant for time to research into whether Richard III indeed murdered the princes in the Tower. There would have been a probability, p, that she would find out that he did. Such a result could easily generate a TV programme, a heritage trail, film rights and associated merchandise. On the other hand, there would have been perhaps a higher probability of (1 - p) that she would find no such thing.

Unfortunately there is a confounding factor, another probability, q, that the creative people decide to make the programme, the heritage trail and the film regardless. If the creative people had decided this, then of course RCUK's outlay for the historian's research would have been entirely wasted.

Finding out credible values for p and q would not have been easy, but it should not have been beyond a well-resourced economist or mathematician to decide how probable it would have been that our historian got a result, even if to do so might have involved funding some preliminary meta-research from other historians. These would have been carefully chosen for a track record of success in attracting funding to their own projects, so the result would have been as accurate as possible. The algorithm would doubtless have been calibrated on past data, such as John Locke's responsibility for the economically productive American constitution, or Rousseau's role in the slightly more iffy French Revolution. Perhaps it was the need for this kind of meta-research that killed the project, and it would certainly have been much better if x and y could have been discovered without messy input from people in the same line of business as those being assessed.

A more tractable part of the problem would have been computing the benefits in question. Terms such as royalty, Nazi, murder or incest would naturally have carried a positive multiplier, and well-targeted research that involves more than one of them would have benefited correspondingly, and may still do so given whichever priorities remain now that the full scheme has been abandoned.

There was undoubtedly a problem, which is that in some areas other than the creative industries, the forthcoming benefits might have been thought to depend on the plausibility of some proposition or other, or the validity of some argument or some approach. But RCUK may have thought that this is less important in the arts and humanities than, say, in engineering.

Generally speaking, acceptability to the public, which can quite easily be anticipated by means of focus groups and online questionnaires, substitutes quite well for truth. It is widely accepted that fiction is as good as fact in most human affairs, as drug companies, for instance, well know. Indeed, proving this is expected to be one of the most exciting, economically fruitful results of a ten-year joint interdisciplinary collaboration between philosophy and divinity, although unfortunately the result may put the partnership under strain.

Sometimes a benefit has more to do with firefighting than media opportunities. Suppose, for instance, a philosopher had sought a grant in the mid-19th century for work opposing the social and economic constructions of Karl Marx, or a later grant to do the same in opposition to Lenin. It would have been quite difficult (although not, RCUK must have thought, impossible) to have run an algorithm on the benefits of averting any likely future impact of Marxist-Leninist thinking. It would have been quite hard, before the event, to see exactly what those impacts would turn out to be. At the time, it would probably have been impossible, and perhaps RCUK got cold feet because of that. It would certainly have taken very highly paid teams of consultants from the private sector to get this kind of calculation right.

I, for one, was certainly looking forward to the algorithm they came up with. And we must all feel a slight pang now that we are not going to be treated to it. It looked as though it would herald a breakthrough in computing science, going beyond yesterday's dreary old rule of "garbage in, garbage out" to generate robust results that would have been largely insensitive to the quality of the input.