Semantic Werks

Talk at the University of Toronto CS department’s distinguished lecture series today by Alan Kay. Kay doesn’t believe there is a discipline of computer science. There’s a slurry of incremental, cul-de-sac research aimed at generating maximum publications (according to Kay, the UC system only cares about absolute numbers of papers published). And there’s a bunch of vendor junk, like SAP and Microsoft Office. This point really resonates: how can taking direction from IBM (autonomic computing, Eclipse, Jazz) be any different than taking direction from Pfizer, or Monsanto, or DARPA?

There’s an intuitive appeal to the argument against CS as a scientific discipline: if you look at what Sutherland, PARC, etc. built in the 1970s, or LISP in the 1950s, there’s not a lot of progress since then on the fundamental questions they were addressing. I think there’s a lot of room for criticism of Kay’s comparisons, but strong reactions are obviously what he was going for.

Interesting point: if we consider IQ, knowledge, and outlook (which I think of as perspective), the first one dominates among computer scientists. He characterizes this as making piles of bricks, e.g., the Pyramids. However, being able to solve small problems does not address the nature of those problems; hence his comment that architecture dominates materials. The second aspect is knowledge, which is akin to making good cars with available parts, re-engineering the available to address the issue. Henry Ford typifies this aspect. Finally, he uses Einstein as evidence of someone who (on top of having a high IQ and a lot of knowledge) shifted the perspective (paradigm, as Kuhnians would say) in a new direction, e.g., that Truth is relative, that time is relative, etc.

His main work seems to be focused on implementing work that is shifting perspectives in CS, rethinking problems in novel and unseen ways. His Viewpoints Research Institute is the main place he currently works in.

Caution: arguing that you are shifting perspectives or doing radical new work is also often associated with crackpot science. Not that this is what Kay is doing.