Incentives for Innovation

I attended Mobile HCI in Stockholm, Sweden last week. My research group had two accepted papers, both from interns (Kim Weaver and Patti Bao) who had worked with us last summer. While sitting through yet another study of users interacting while walking it struck me that the proceedings were full of Science (lots of scientific studies examining, often in minute detail, how users interact with mobile devices), it was very short on Innovation: they weren’t really any new applications or services that researchers had created. (And I’ll note that my team was guilty of contributing to that tilt: our papers were studies, not systems, because it’s easier for interns to concentrate on studies with only 3 months available.)

While it’s entirely possible that I’m just getting cranky with my advancing years (hey you kids, get off of my lawn!), I encounter more and more research papers that are studies of what other people have built rather than descriptions of things the researchers themselves built. And after thinking about it a bit in Stockholm, I think a large part of the reason comes down to the incentive structures we have in place.

What’s the incentive for doing really innovative work in research? You can publish a paper. What about if you do mediocre that’s an incremental addition to the field’s knowledge? You can publish a paper. Funny how whether you’re doing awesome work or mediocre work the outcomes are strikingly similar. And what’s the worst that can happen? Your paper gets rejected.

Lets contrast that with the start-up scene in Silicon Valley. What’s the incentive for doing really innovative work at a start-up? Your company becomes the next big thing and you make a bazillion dollars. What if you do mediocre, incremental work? Your company vanishes into the dustbin of history. What’s the worst that can happen? Your company goes down in flames and you need to look for a new job.

Call me crazy, but given that the incentives to be innovative are so much stronger in industry, it’s hard to be surprised that industry is now really driving the innovation in CS. In fact, I’d suggest that the disparity in incentives and outcomes raises the question of how long academic research, as it currently stands, will be regarded as valuable and interesting. The writing is arguably already on the wall in industry: existing industrial computer science labs are slowly disappearing, and new computer science labs aren’t really taking their place. So what is the future of academic research? Will we reinvent and reinvigorate the field? Or will the drift of the innovative center of the field continue its shift from academia to research?