Saw this article a couple of days ago on slashdot. Perhaps I'm suffering from HCI-withdrawal, since I'm not at the CHI conference this year, so I kinda got into reading about this reasearch. It's also an area I know a bit about.
Unfortunately, there doesn't seem to be a published paper yet, which makes it hard to find out exactly what they did. I wonder how much of the effect they are seeing is spatial vs. attachment to a specific computer. I remember when I was in school I always preferred certain locations in the computer labs, much the same way that people always sit in the same place in meeting rooms.
The best-known (and best conducted, in my opinion) research in this area is by Nass and Reeves at Stanford. I love the way they describe their work: they went into the stacks at the Stanford Library, found books on personality theory, and wherever it said "whenever a person interacts with another person" their crossed out "another person" and wrote in "a computer"... and they magically had a whole new set of theories to prove or disprove. And it turns out that mostly they are true: people assign social characteristics to computers the same way they assign them to other people. They wrote this up in a great book called The Media Equation that I highly recommend.
Nass and Reeves were consultants on an infamous Microsoft product called (you guessed it) Bob. It's the most maligned, and least understood, product in the history of the company. The leads of the Bob project had a very simple premise: if people treat computers the same way they treat other people, then we should deal with that explicitly in the interface. So they set about designing a "social interface" that does exactly that.
Bob went wrong in two ways. First was just simple execution; Version 1.0 was designed around very simple household things, like writing a personal letter, filing recipes, etc. It was very limited and constrained, and people just didn't find it very useful. That was a result of the Bob team being small, young, under severe schedule pressure, and being forced to compete with Works and other consumer productivity tools.
The second reason is much more damning, and also applies to clippy and the dog in the Windows XP search pane. I think it was best explained by Bill Buxton at CHI 99 in Pittsburgh, so let me relate to you what he said.
The first time you and I meet, we are very polite, probably shake hands, and choose our words very carefully. Over the next several times we meet and get to know more about each other, our dialogue shifts to being more casual, and assumptions are made based upon past conversations and context that let us skip over entire interchanges of information. This is a natural progression; if I were to speak to you at our tenth meeting the same way I spoke to you at our first, you'd think I had a personality disorder. But that in fact is exactly what Bob (and clippy, and the search dog) do: they never adjust the way they speak to us. There are user studies that show that most people respond very positively to the character the first few times they interact with it; however, a few dozen times later many of them are sick of it. It's in fact the corollary of Nass and Reeves' basic theorem: when computers exhibit pathological social behavior, we respond the same way we would to humans that exhibit those behaviors.
There's one other important lesson here: the importance of doing longitudinal studies on your designs. Because things change.
7:57:06 PM  ; ;
|
