[pnathan]

Okay, so there are issues with academia. How do those get solved?

The obvious answer is "more funding", but there's also cultural shifts, wherein academia is excorcised for not doing "Real World Work".

Is part of the solution to drive upward from K to PhD, focusing on critical thinking and humanities, educating more than training?

I don't know.

[_delirium]

I'm not sure it's more funding, as much as the way it's doled out. The idea with the way funding has shifted was indeed (in part) to make academia more "real-world" and competitive. Instead of a university getting a big block grant from the NSF, or researchers getting unrestricted funding directly from their university's budget, as used to be common, individual researchers apply for competitive 3-year grants to fund an individual project. In addition, researchers are increasingly expected to fund their students and student expenses (e.g. travel), whereas a larger percentage of that used to be funded by universities. Some of that (particularly at state universities) is outright decreases in funding for research, but some of it is just a change in how it's administered.

The idea, which was plausible, is that projects can then be judged on their merits, instead of one big slush fund that who knows what comes out of. But the unintended, if not unforeseeable side effect is that it adds a huge amount of extra overhead, and incentives to target only "fundable", aka "sellable" research. With NSF funding rates for projects currently running at 5-10% of proposals, and typical large research universities expecting you to have 1-3 of these 3-year grants going at any given time, you need to be submitting 10+ grant proposals a year! And ideally also working your networks to see if you can attract some corporate funding. That's a huge amount of overhead, and it also sucks a lot of the appeal from academia, since rather than the university setting giving you freedom, you're in some sense closer to an independent firm that has to bring in its own funding, constantly chasing the next round of financing lest your lab implode and students go hungry.

There are likely people who will thrive in that environment, but I think it's increasingly going to be people who are skilled at research management and sales. The #1 job is attracting external financing, and the #2 job is heading up a successful mini enterprise with that financing, ensuring the lab is operating well. This is becoming pretty close to explicit. One university I've been at now sends around an internal newsletter ranking faculty by number of dollars brought in so far this year! If that's what you're going to be judged on, why even be in academia?

[zeroonetwothree]

I wouldn't say it's unforseeable: http://en.wikipedia.org/wiki/Rent-seeking

[Lewisham]

Okay, so there are issues with academia. How do those get solved?

They don't. As far as I can tell, all the problems, bar those that stem directly from funding, have always been there. That's the nature of the institution.

What changed, for CS at least, is that Google provides much of what is attractive to academia, but has all the funding that is required to make the other roadblocks go away. Google is not like Xerox PARC or MERL or Microsoft Research or other industry research labs. It's a company that's built upon being a research lab. I can believe the hey-days of Sun and SGI were probably similar.

The easiest way I describe working at Google to my grad friends is "It's a giant grad lab, except grads are also the ones running it, and they're billionaires." Academia can't compete with that, and it never did. The open question is not how to fix academia (as it never will be), but whether the Google model is sustainable. For as long as the Google model does exist, you will always see a net loss of professors to industry, rather than the other way around.

[_delirium]

I dunno about the "never did" part. Google is doing some interesting stuff, but I don't think they're yet up to the peak of academia, like the amount of individual researcher freedom combined with innovation going on in the heyday of the MIT AI Lab. Even grad students were doing totally independent never-been-done research projects! Mere staff members like Richard Stallman had influential active side projects, too.

Google does seem like a great place if you're a senior enough researcher, though. People like Peter Norvig, Ken Thompson, and likely the author of this linked article, seem to get basically 100% freedom to work on whatever they want, with minimal management or job requirements, which is pretty much the ideal position to be in as a researcher. I suspect not all Google employees get Ken-Thompson-level freedom from having a boss, though.

[carterschonwald]

Hehe, this is where is gets interesting. There's a huge number of small to mid sized companies which give you huge creative latitude for projects. Perhaps the most famous one is Valve (read their handbook, it's amazing!)

I'm actually hoping to replicate some of those ideas of both the good bits of academe plus the modalities in aforementioned handbook as I start to pull people into my org, Wellposed. (because what's more exciting than working with fun nice interesting intelligent folks who do amazing work plus being paid insanely well? )

[_delirium]

It's an interesting model that I do hope catches on. My impression is that it's not currently widespread, but maybe I'm missing all the interesting action. Afaict, in the game industry (which I'm pretty familiar with, being an AI researcher in academia focusing on game design / game mechanics), the more common model is a very highly managed EA-style one. And among smaller firms, creative freedom in design is more common, but technical research isn't that common, because they don't have the resources/budget/interest. The only smallish game company I've seen produce actual published research is LittleTextPeople (some years ago, there was also Zoesis, but it's now defunct).

Perhaps it's better in other fields?

[carterschonwald]

who cares how other people do their biz, its how i'm structuring my (not game focused) operation to the full extent that I can :)

(and because there are no outside investors, it is so to the full extent that fiat authority can make there be no org chart :) )

[thronemonkey]

Keep in mind that academic science includes fields other than CS ;)

[arethuza]

When I worked in academic research (NB a long time ago) I worked on pretty large EU funded projects (30 to 40 people, 7 or 8 organisations in multiple countries) and they were a nightmare due to the amount of "management" that comes with projects spending millions or tens of millions.

My feeling was that taking the money (even less money as you wouldn't have the manager/bean-counter overhead) and splitting it amongst a large number of small projects would have been far more productive in terms of actual research output.

[xtracto]

I just commented this in reply to another person but I feel you will understand it more.

I was just part of an FP7 EU project in which 11 organisations of 7 different countries participated... and gosh, the admin burden of that is horrible. I got to be a work package leader and unfortunately it means you can't really do a lot of R&D. The amount of forms you have to fill , all the Brussels reporting is amazing. Added to that, I think the whole "publish or perish" attitude of academia just is not for me. So what if I only publish one paper from the project? yeah, everybody things I am a loser and that my participation was not successful (yeah, nobody says that but...). It feels as if research is just done with the objective of publishing papers... not only that, but colleagues actually say it "but you know, we have to see how are we going to translate this research into publications".

On the other hand, I also worked in a smaller 3 year EPSRC project with 3 Universities and 3 industry partners. It "felt" better, although the 3 industry partners did not really care (each meeting they sent someone new because the other person was either out or doing something else, so the new guy didn't know anything about the project and was just in the meeting to fill in). In this project, at least each academic partner was doing whatever they wanted and just presented their work, and a the end of the project, something "practical" came out of it.