3 ms·

This isn't even remotely close to true. The only perspective in society from which it appears that Academia does more R&D than the private sector is academia i

by ke88y 3y ago

This isn't even remotely close to true. The only perspective in society from which it appears that Academia does more R&D than the private sector is academia itself. Academics in this respect have their eyes wide shut. At mot, publishing in academic conferences at private companies is namesake. But academic publishing in general is a shadow of its former self in terms of dissemination power, both because most academics are horrible writers and also because a product you can buy (with a support contract) or a well-documented software package is many orders of magnitude more impactful than a paper no one will read. And even then, many companies publish far more than even highly productive research universities. Consider AI. There are little pieces here and there that you can attribute to academia. But for the most part, over the last two decades, the enabling big steps happened at TSMC+NVIDIA+Google+OpenAI.

kevinventullo 3y ago
over the last two decades, the enabling big steps happened at TSMC+NVIDIA+Google+OpenAI Those big steps are standing on the shoulders of decades of academic work, and were carried out by people who had spent a great deal of time in academia.
- ke88y 3y ago
  > and were carried out by people who had spent a great deal of time in academia. Right. Who then promptly left academia when they needed to start doing R&D at a meaningful and consequential scale. The academic contribution to chip fab R&D today is nearly inconsequential compared to private sector investment, both in raw dollar amounts and especially in terms of impact per dollar spent. This is like saying that academia is responsible for Google because its founders spent a couple years on an NSF grant. Yes -- but then they left to do the real R&D work at scale. Academia plays and important role in training people. But the R&D component is -- with some exceptions -- mostly a matter of "training young people for the real work by messing around on toy problems".

by ke88y 3y ago

kevinventullo 3y ago

over the last two decades, the enabling big steps happened at TSMC+NVIDIA+Google+OpenAI Those big steps are standing on the shoulders of decades of academic work, and were carried out by people who had spent a great deal of time in academia.

ke88y 3y ago
> and were carried out by people who had spent a great deal of time in academia. Right. Who then promptly left academia when they needed to start doing R&D at a meaningful and consequential scale. The academic contribution to chip fab R&D today is nearly inconsequential compared to private sector investment, both in raw dollar amounts and especially in terms of impact per dollar spent. This is like saying that academia is responsible for Google because its founders spent a couple years on an NSF grant. Yes -- but then they left to do the real R&D work at scale. Academia plays and important role in training people. But the R&D component is -- with some exceptions -- mostly a matter of "training young people for the real work by messing around on toy problems".