I’ve just returned—in the circadian rhythms sense of getting back up in the morning and going to bed at night—from GECCO 2006 in Seattle. Much to write about and discuss, but here are some bullet points I gleaned from the innumerable sessions attended and break-outs disrupted with enthusiastic banter:

• Genetic Programming and related things can be used to create complex human-competitive designs. That’s common knowledge, among the cognoscenti at least. But the leap from building a new kind of lens system or a new kind of analog circuit, to building a new kind of complex object like a word processor or a robot or a work of art or a living thing… you need people in the loop. User-centric design. Had beer with Ian Parmee and Chris Simons, for instance, and attended several tutorials and workshops on dealing with user fatigue in automated design systems, framing questions for gleaning user wishes, and saw the same questions coming up over and over in other sessions where people weren’t thinking yet about the role of the user (aka “customer”) in evolutionary design, search, and optimization. Engineers need to think about those people. About not what this stuff is for, but who it’s for.
• Symbolic regression is not for curve-fitting. It’s for model discovery. Analysis. Understanding. That’s another broad class of problems where user-centric thinking has started to percolate into the practitioners’ consciousness: symbolic regression is for exploratory data analysis, not design divorced from a human brain.
• Genetic algorithms, evolution strategies, neural networks, and other numerical optimization and search tools are quickly falling into the canon of operations research and the toolkit of optimization. I saw substantially fewer papers on over-abstract toy problems this time around; instead, where theory was explicitly the goal, there was much more focus on using results from GAs and ESs to explore the structure of abstract toy problems from other domains.
• Of 500+ participants expressly working on optimization, search and design with evolutionary algorithms and related approaches, I can’t think of a single one concerned with optimality criteria. Maybe one or two pay attention to it, but nobody gives a damn for theory, for application, for customers. Is this a good thing?
• Lots of people seem to be leaving Google, after working there for a while. How does that come to be the case?
• Startup time is upon us once again. There was lots of buzz about productive, successful high-tech startups. A substantial drop in the number of morose burnt-out entrepreneurs.
• All that stuff high-tech people are supposed to like doing at conferences, like having a back-channel and blogging and crap? Bah. Nobody has a clue. How can the culture of computer scientists be so backward? Oh wait… never mind. I get that one.
• What one wants is to be able to talk with a diverse club of smart people, arrange to do short one-off research projects and simulations, publish papers or capture intellectual property quickly and easily, and move on to another conversation. Quickly. Easily. For a living. Can’t do that in industry. Can’t do that in the Academy. Yet in my experience, scientists and engineers all want it. Maybe even a few mathematicians and social scientists do, too.