[jamesjporter]

I think the programming industry is interesting because my impression (I have no idea if this is backed up by data) is that there's much more mixing between academia and the private sector (i.e. people move from academic to industry jobs and vice-versa more frequently/easily) than in other fields (natural science, social sciences). I wonder if this increases the productivity of either.

[jostmey]

I am in the Natural sciences. I see that people are trained in esoteric technologies or procedures that may only be useful to the immediate scientific question at hand. This would seem to make it hard for these individuals to transition to industry, unless their specific expertise is required.

The computer programming community has the advantage that its technologies and best practices are ubiquitous throughout the industry and academia (??). This may facilitate an easy transition from industry to academia and vice-versa.

[rayiner]

I think it's actually the opposite. Academia is more divorced from industry in the computer science world than say the aerospace world. When I was working with grad students as an aerospace major, they were by and large working directly on problems relevant to industry. Most computer science research seems much more mathematical/theoretical than practical, in comparison. I think this is also a relatively recent phenomenon. Compare say Preston Brigg's thesis on graph coloring register allocation (1992) to Sebastian Hack's thesis on register allocation for SSA form programs (2007). The former I think was much more written with implementers in mind, and the latter has a lot more math and proofs (though it's more readable than a lot of CS papers these days, which can tend towards a dense maze of proofs and esoteric notation).

Indeed, I think there is a language barrier between CS academia and CS industry that doesn't exist between AE academia and AE industry. In aerospace, everyone speaks the language of differential equations, finite element analysis, computational fluid dynamics, etc. But CS academia is deeply steeped in type theory, automated proofs, etc, while CS industry talks about object orientation and test frameworks, etc.

[javert]

I wonder if this increases the productivity of either.

My research as a CS grad student is very much motivated by interactions with industry people, and very much is intended to facilitate things they need.

So, yes, mixing in the computer science/software engineering field is very useful for both academia and industry.