Made out of people

On Handling and Teaching Hardware

(This entry was written on a non-wired computer, during the HCIC conference.)

I just saw a talk in which the system was using a touch board that--they thought--had been modified to support several people. Unfortunately, it can only handle one touch at a time: if two users pressed at the same time, the system would kind of register a point somewhere in the middle.

Now, I've seen this before at a smartboard, and I've seen it fail before. But they see a serious disadvantage of the system as the fact that users need to derive--on their own--rules about how the board works.

The people talking about the system have built a long, complex set of alternatives: converted electronic chessboards linked to cameras; RF transmitters; etc. But this seems like an awful lot of work around a known-to-be-simple problem.

Why not use a more explicit "teaching" moment? It can be subtle--perhaps as the board starts up, it encourages users to "place a house here. good! now place a house here. good! now click here. oops, you just got two clicks. you didn't mean to do that, did you?" Heck, turn the averaging thing into a FEATURE: "if you want to click halfway in between two people, just have both press at the same time."

Now, that's not actually a feature anyone would ever use. But by training users to use it and play with it, they will be more ready to handle it when they see the failure occur. Rather than make it an implicit learning situation, why not make it explicit?

It turns out that the new system has enough odd features that they need such a teaching mode anyway. So they train users--here's how to click. Here's how to move an object. Why not allow them to learn how to not click poorly, and save yourselves the redesign? (In the new model, they can only have 15 pieces per board region and have to be aligned over orthogonal corners, due to the fact that they've built it over a chessboard.)