Social interfaces

Comments on The Media Equation : How People Treat Computers, Television, and New Media Like Real People and Places, by Byron Reeves and Clifford Nass, Cambridge Univ Press, 1996.

The assigned reading was chapters one through four. This was a somewhat difficult reading to comment on... I found myself squirming a lot while reading Reeves and Nass. Anyway, here goes...

  1. Good point first. Some of these results are really quite thought-provoking. Supposedly this work is responsible for the irritatingly and obnoxiously cheerful paper-clip, but if that's true then it was clearly misunderstood. R&N recognize that "social" interaction can be overdone -- particularly with experts who can see through "deceit." What the work does say is that people do react to quite limited cues in any kind of interface, and that we as UI designers (ahem) need to understand how and why this happens.

  2. The results are supported by careful experiments. Without actually having read their papers, I assume that the book accurately reflects the results. Still, I would have felt more comfortable about the veracity of their results had the book included more details about the actual experiments and results. For example, they say that "the computer told each user what it thought of its own performance." Er, how, exactly, did it do that? To a certain degree, some of these experiments seem to me to be forcing the issue: if you could get an electric power drill to turn around to the person holding it and say "Hey, buddy, did I do good?," then it seems unlikely that you wouldn't get some kind of "social" response.

  3. There is also a certain degree of obviousness in a lot of this. After all, people cry in movies (except me <gruff> <gruff>), yet everybody "knows" they're just made up. I swear somewhat anthropomorphically at motorcycle parts that won't come unstuck, and I know people who pat their car on the dashboard to encourage it up a steep hill. Anybody that knows how to write can inject a certain degree of personality into even dry technical prose if they choose to (and if their professor doesn't make them "fix" it).

  4. Reeves and Nass tend to overstate their results. For example, people have "attributed to an animated line a personality as rich as that of their best friend." A NeXT computer "triggers rich scripts for social interaction." Bollocks. Their results are interesting and useful, but I think they get carried away with the pretence that they just carried out some experiments and gosh, look at these incredible results we got!
Which takes me to four points. Given my overall critical tone, I guess I am less likeable but I appear to be smarter. I don't wish to devalue this work, which I find fascinating, but reading Reeves and Nass gives me the uncomfortable feeling that you could never win an argument against them but you still wouldn't trust them enough to quite believe them. There's social responses for you :-)

John Reekie, March 17th, 1998.