I can’t help but think that if computer science grants had more industry input into the writing process and more importantly the process of selecting grants to fund, we’d see a great wailing and gnashing of teeth and then some far more useful and relevant research being done.
For example, there has been research into “semantic wikis” and “semantic knowledge systems” for years. While fun, the sums invested seem to have been out of all proportion to what is basically quite a simple idea, with quite low uptake so far. (The main beneficiary of this particular line of research, other than the researchers themselves, may have been IBM, who I think received some EU funding to help it build its proprietary software. But I don’t know if that software was successful, so perhaps even IBM got hoodwinked.)
Also, the open source software created in those projects is frequently abandoned after production – in one case quite deliberately so, immediately after funding ceased, in order to focus on the next “latest, greatest” semantic knowledge system, which happened to involve a total rewrite. It’s great that there’s some open source code, but no-one seems to care about it or maintain it, which indicates something fundamentally wrong with the whole process.
It’s also possible that semantic wikis are actually brilliant, but not enough effort has been put into commercialising and disseminating the existing research. Again, industry advice could help here.
Another benefit of this idea might be that if grant assessors started to say, “No, this distributing computing calculus of yours has absolutely no relevance to bridging the digital divide in India”, funding bodies might realise that if they want to keep funding the amount of “basic research” that is being done, they ought to be honest about it and stop attaching spurious criteria to all funding tranches that grant writers then have to write nonsense to “fulfill”. By all means have practical criteria for more practical funding tranches, but don’t try and pretend that all research will have practical applications in the next 5 years, because some of it won’t.