3 ms·

Resource allocation is a zero sum game. If taxes are being levied to pay for a vast intellectual class living in ivory towers, that's less funding available to

Resource allocation is a zero sum game. If taxes are being levied to pay for a vast intellectual class living in ivory towers, that's less funding available to pay for industrial research. Academia love to claim that practical application should not be any concern of theirs. But why should this be so? After all, they eat practical food and live in practical houses. Someone has to pay for it all. And the idea, espoused in this essay, that eventually, on a long enough timescale, their research will all be adopted and it's just that industry is super slow to recognise the awesomeness of their research and productize it - well, this claim seems to be getting weaker all the time. We now have many areas of computer science that have spent decades going down very expensive rabbit holes with no serious adoption and none on the horizon. Not just formal methods, but pure FP languages like Haskell remain rarely adopted despite huge hype and massive academic investment. There was a few days ago, a post asking what happened to the semantic web with many commenters positing that the semantic web became an academic money pit that absorbed huge quantities of grant money and delivered nothing whatsoever. I'm sure there are many other examples lurking if you scratch the surface. If academic funding was cut significantly and returned to companies in the form of lower corporation taxes, perhaps connected to research incentives, would it be so bad? There'd probably be less research done overall but if lots of current research goes nowhere and is never adopted that'd be a net win.

int_19h 8y ago
What do you make of corporations subsidizing seemingly pure academic research, then? Take the aforementioned Haskell, for example - you speak of "massive academic investment", but Simon Peyton Jones is still a Microsoft [Research] employee, and Simon Marlow is a Facebook employee (and former MSR as well)! And yes, you don't see Microsoft use Haskell in its products - but the ongoing funding clearly indicates that there's substantial perceived value in that investment. Given the amount of time this has been going on, surely that's more than can be explained away by "hype"?
- repolfx 8y ago
  I think the corporate world is still learning how to do effective research. Too many companies think that doing research means replicating a tiny university within their own walls. I'm not saying industry always gets it more right than academia, but I think even the orgs like Microsoft Research that are basically just clones of academia, are still on a slightly better path. For instance MSR has done a lot of cutting edge research into operating system design. Still, the money was in the end wasted - nothing of Midori ever surfaced in real products. The gold standard for me is Google. They're one of the few firms that managed to completely fuse research and industrialisation into a totally cohesive whole. Whilst they technically speaking have a research department, the research they're known for is largely not done there and Google has managed in the past to routinely create new innovations from all over the company. It's worth realising that a lot of industrial and corporate research isn't labelled as such. It's just "innovative product development". Example: If we look at JetBrains, they developed a new programming language from scratch (Kotlin) that has several innovative and cutting edge features. Do we think of JetBrains as a sponsor of research or Kotlin as the product of research, probably not, yet it solves many problems that were previously unresolved, like how to make a new language that has seamless interop with Java, and how to improve the usability of features like generics and nullity checking. The work involved in creating it explored new frontiers that academia has ignored, simply because usability improvements and practical topics like interop with existing code is not considered 'smart' enough to be worth the title of research. You see this pattern crop up repeatedly. Academia, left to its own devices, researches ever more exotic extensions to Haskell, knowing full well most programmers don't use Haskell. The corporate funded academic research done by JKU Linz and Oracle has focused on how to make existing languages like Ruby and Python super fast to execute, super debuggable, cheap to develop runtimes for, how to solve the language interop problem once and for all and so on. They've produced something real - Graal - which looks on track to actually let Ruby call into C++ call into Java call into JavaScript call back into Ruby again, in a seamless and fully optimised manner. Now that is cutting edge but incredibly practical research that could have industry-shaking impact within the next 10 years. It isn't primarily funded by the NSF or EU though. It's funded by a corporation.
  - irundebian 8y ago
    It's just lame to play off practical against theoretical research and academic vs corporate research. Each of them has its advantages and disadvantages.

int_19h 8y ago

What do you make of corporations subsidizing seemingly pure academic research, then? Take the aforementioned Haskell, for example - you speak of "massive academic investment", but Simon Peyton Jones is still a Microsoft [Research] employee, and Simon Marlow is a Facebook employee (and former MSR as well)! And yes, you don't see Microsoft use Haskell in its products - but the ongoing funding clearly indicates that there's substantial perceived value in that investment. Given the amount of time this has been going on, surely that's more than can be explained away by "hype"?

repolfx 8y ago
I think the corporate world is still learning how to do effective research. Too many companies think that doing research means replicating a tiny university within their own walls. I'm not saying industry always gets it more right than academia, but I think even the orgs like Microsoft Research that are basically just clones of academia, are still on a slightly better path. For instance MSR has done a lot of cutting edge research into operating system design. Still, the money was in the end wasted - nothing of Midori ever surfaced in real products. The gold standard for me is Google. They're one of the few firms that managed to completely fuse research and industrialisation into a totally cohesive whole. Whilst they technically speaking have a research department, the research they're known for is largely not done there and Google has managed in the past to routinely create new innovations from all over the company. It's worth realising that a lot of industrial and corporate research isn't labelled as such. It's just "innovative product development". Example: If we look at JetBrains, they developed a new programming language from scratch (Kotlin) that has several innovative and cutting edge features. Do we think of JetBrains as a sponsor of research or Kotlin as the product of research, probably not, yet it solves many problems that were previously unresolved, like how to make a new language that has seamless interop with Java, and how to improve the usability of features like generics and nullity checking. The work involved in creating it explored new frontiers that academia has ignored, simply because usability improvements and practical topics like interop with existing code is not considered 'smart' enough to be worth the title of research. You see this pattern crop up repeatedly. Academia, left to its own devices, researches ever more exotic extensions to Haskell, knowing full well most programmers don't use Haskell. The corporate funded academic research done by JKU Linz and Oracle has focused on how to make existing languages like Ruby and Python super fast to execute, super debuggable, cheap to develop runtimes for, how to solve the language interop problem once and for all and so on. They've produced something real - Graal - which looks on track to actually let Ruby call into C++ call into Java call into JavaScript call back into Ruby again, in a seamless and fully optimised manner. Now that is cutting edge but incredibly practical research that could have industry-shaking impact within the next 10 years. It isn't primarily funded by the NSF or EU though. It's funded by a corporation.
- irundebian 8y ago
  It's just lame to play off practical against theoretical research and academic vs corporate research. Each of them has its advantages and disadvantages.