Why we all hated Clippy
I brought this up in a discussion for one of my courses, thought I might share it here:
I’ve kept on thinking about Clippy, and social behavior of software. The way I see things, the problem with Clippy was not that they tried to do a social actor into an assistant, the problem is that they got it wrong.
It’s fine to try to create an assistant that acts in a social manner, just as long as it plays by the social rules, the problem with Clippy is that it was actually rude. Software that tries to help you should not interrupt you but help, and Clippy would interrupt your workflow again and again, which is going to be interpreted as rude, irrespectful behavior.
I remember long ago using an illustration application (Corel Draw) that had some tools that learned from your actions and interpreted them, so when you were duplicating an object it read your history in order to predict the size and position of your next object, it usually got it right, and when it got it wrong the work you had to make in order to correct the error did not involve more effort than would have been required if the application was repeating dumbly.
I once gave this example on the very first lecture I gave. A smart woman in the audience asked me how this was different from the Clippy we all hate. At the time I was too nervous to think clearly and could not really provide an answer on time, but now I’ve thought about this a lot more, and have read more, and can say that the problem here is that social behavior is very complex, and there’s a very fine line between being helpful and being impolite, and Clippy’s worst sin was that in kept on interrupting people’s workflow, and that’s quite a rude thing to do.
…Interaction design can be very tricky indeed.