On Wednesday, January 29, 2003, at 02:24 PM, Eugen Leitl wrote:

Feds are sure inefficient, but the random dispersal of funds does tend to hit the far shots now and then. The private sector tends to ruthlessly optimize on the short run (because the long shot doesn't pay if you go broke before you can reap the possible benefits).

It's about the single most powerful reason for federally funded research to exist.

I should have mentioned in my first reply that you need to spend some time looking into evolutionary learning and markets. For example, the importance of quick feedback and correction, with profits determining which markets are explored. I have strong views on this, having studied the electronics/semiconductor market for many years, having studied carefully the role of intermediate products (such as RTL --> DTL --> TTL --> op amps --> MOS RAMs --> 4-bit microprocessors --> etc.). Products introduced in 1963, say, were generally making the bulk of a company's profits by 1965-66, paying for the 1965 R & D and the 1966 product rollouts, which then paid for the 1967-69 cycle, etc. I know this was true of the earlier technologies and it matched everything I saw in my years at Intel and thereafter. The "2-4 year payback cycle" in the electronics industry, from roughly 1955 to the present, was terribly important. Each generation of technology paid for the next generation, and costly mistakes resulted in companies ceasing to exist (Shockley Transistor, Rheem, Precision Monolithics, and so on...the list is long). Successful products led to the "genes" (or memes) propagating. Phenotypes and genotypes. This same model gave us, basically, the commercial automobile and aviation industries. "Moon shots," on the other hand, distort markets, suffer from a lack of evolutionary learning, and have almost no breakthroughs ("But what about Tang?"). "I am proud to announce, as your President, the goal of creating our national mechanical brain, a machine which will be built with one million relays and vacuum tubes. I am committing one billion dollars to this noble endeavour. We expect to have the mechanical brain operating by 1970." --President Dwight Eisenhower, 1958. Really, Eugene, you need to think deeply about this issue. Ask your lab associate, "A. G.," about why learning and success/failure is so important for so many industries. Read some Hayek, some von Mises, some Milton Friedman. And even some David Friedman. Ask why the U.S.S.R., which depended essentially solely on "federal funding," failed so completely. Hint: it wasn't just because of repression. It was largely because "picking winners" doesn't work, and command economies only know how to pick winners (they think). Think deeply about why this list is what it is.