Sunday, April 03, 2005

Mark Kleiman points to a New York Times article about moves at DARPA, the Pentagon's research funding agency, to cut funding to long-term, academic research in computer science, in favor of shorter-term industrial development projects. A number of prominent computer scientists are quoted in the article, sounding quite incensed at DARPA's supposed shortsightedness in failing to renew and expand their usual grants. Buried towards the end, however, is this telling tidbit:
Still, a number of top scientists argue that the Pentagon's shift in priorities could not have come at a worse time. Most American companies have largely ended basic research and have begun to outsource product research and development extensively even as investments in Asia and Europe are rising quickly.
Now, those of us who were involved in computer science research back in the boom years of the '90's, when industry was spending like crazy on it, might scratch our heads a bit at this claim. You see, I'm trying to recall if any of these same professors back then used to chide DARPA for funding so much CS research, and reassure them that the government's money could best be spent elsewhere, since industry was easily taking up the slack in the CS area.

I clearly recall some academics arguing that industry's massive funding of CS research required a matching injection of government funds. I seem to remember the occasional claim that government funding had to keep pace with industry, to avoid private corporations "capturing" the expected massively lucrative fruits of basic CS research by claiming intellectual property rights on them. Maybe somebody might have argued that the surge in industrial funding of CS research was a sign that the research area had a lot of potential, and was hence deserving of government largesse. Or that it proved the economic importance of research to the hi-tech industry, which therefore needed still more government funding of research to help it stave off competition from other countries.

But I somehow can't seem to recall a single one of the professors quoted in the article arguing back then that government funding is most needed when industry isn't bothering to supply much funding of its own, and therefore that the abundance of industrial funding at the time was a sign that government research funds were less urgent, and could safely be cut back without harming the field or the economy.

Then again, perhaps it's just my poor memory.

3 comments:

Anonymous said...

I don't see the inconsistency here.

(1) You argue that the correct level of government research funding now is higher than the correct level in the past. Let's accept that for the sake of argument.

(2) The quoted researchers say the level is too low now. Let's accept that too, for the sake of argument.

It doesn't follow that the quoted researchers must think that the level was too high in the past. Maybe they think it was just right in the past, and too low now. Maybe they think it was too low in the past, and far too low now. You might disagree with either of these propositions; but they don't contradict (1) and (2).

-- Ed Felten

Anonymous said...

I don't see the inconsistency here.

(1) You argue that the correct level of government research funding now is higher than the correct level in the past. Let's accept that for the sake of argument.

(2) The quoted researchers say the level is too low now. Let's accept that too, for the sake of argument.

It doesn't follow that the quoted researchers must think that the level was too high in the past. Maybe they think it was just right in the past, and too low now. Maybe they think it was too low in the past, and far too low now. You might disagree with either of these propositions; but they don't contradict (1) and (2).

-- Ed Felten

Dan Simon said...

Actually, Ed, I'm personally more inclined to think that government and industry have both gotten the research funding curve just about right. Heavily funding research in an applied topic makes more sense when the corresponding industry is going through major growth and changes that require lots of new ideas, than when it's busy consolidating and absorbing the changes and ideas generated during the previous growth period. The former periods are also obviously the ones during which industrial research tends to get better funding. In fact, I've long argued that fields where industry spends a lot on research--pharmacology is a good example--need and deserve a lot of government research funding, because industry usually isn't very good at research, doesn't really like doing it, and only does it when the need for it is clearly very pressing.

During dynamic times, when government and industry are both generously funding research, the two communities can feed off each other, speeding innovation. This was a common justification for greater funding of CS research during the tech boom. (I seem to recall Ed Lazowska, in particular, making exactly this argument, although I might be mistaken.) However, the contrapositive of this position is that research funding is less productive during more sedate periods, when the industry is neither changing rapidly nor heavily funding research. It seems to me that researchers who advocated more government research funding during the boom on the grounds that it would be particularly valuable as a complement to industrial research and a spur to innovation, are disingenuous in advocating an even greater need for more research funds now that there's much less industrial research to complement, and less apparent need to spur innovation.

(Of course, I'm using a rather broad brush here. For all I know, most of the researchers quoted in the article never made the "government-funded research should complement industrial research during these times of rapid change" argument during the '90s. However, it's my impression that that was a common argument at the time.)