Somewhere near the intersection of cognitive science, psychology, and computer science lies the subfield of Human-Computer Interaction (HCI) — sometimes called Computer-Human Interaction. It’s not a subject I’ve studied at all (and if you saw the user interfaces I’ve developed myself, all doubts would evaporate from your mind); indeed, my own personal preferences in this regard would be considered by some to be positively Luddite. But at work there is a regular HCI seminar series (running weekly during the term since 2003), and I take the opportunity to attend when my responsibilities allow.

Today I attended a talk by Alice Oh, a lab alum who now works at the Korea Advanced Institute of Science and Technology, which was entitled “Machine learning approaches for understanding social interactions on Twitter”, and it struck me that a significant fraction of HCI research today isn’t really about Human-Computer Interaction: it’s about Human-Human Interaction; the computer (or the network) serves only as a mechanism, a facilitator and an amplifier, allowing humans to interact in ways they otherwise would not be able to. Having the computer in the loop allows researchers to collect information about the humans and their interactions that would not otherwise be easily available outside of an NSA or Google data center, putting the research much more in the realm of sociology or psychology than what we think of as traditional computer science, even if the result of the research is a computer program. This was particularly the case for the final research topic Oh discussed in her talk, which was an (as yet unpublished) analysis of Twitter language use in multilingual communities, with regard to both the structure of the social graph and the choice of topics in various languages.

I can’t say whether this is an actual shift in direction or just recency illusion on my part, but it seems a particularly salutary one if so. (Not that there’s anything wrong with making better user interfaces! Like many people in my profession, I don’t particularly like computers, and I do like humans — but I find computers, even the 1980s user interfaces that I tend to prefer, to be much easier to interact with than actual humans, so I put a lot of value on research that helps us understand how computer-mediated human interaction actually works.)