I was at a meeting of our research group, TLRG, today. I presented on future research directions for VLEs (rather like Elizabeth Taylor and marriage, I can't resist doing these crytsal ball gazing exercises, even though I know something always comes along and makes them redundant). Talking to my colleague Adam Joinson, he mentioned he was doing research on social software looking at issues of privacy and trust. One point in particular interested me and that was a degree of ambiguity is preferable in such systems. For instance, when creating avatars, it helps if the avatar doesn't resemble the individual too closely, as this frees up behaviour online, and also in location systems, users feel as though their privacy is invaded with excessive detail. Saying 'I'm in Cardiff' is okay, but not 'I'm at this address...'
This goes somewhat counter to the natural instincts of developers and designers, which is always to make systems more detailed, more precise. It is a brave piece of design that specifies a level of ambiguity in a system. Like many things though it seems that less is often more.
