At Goldsmiths I became interested in what makes "good" interactive art. Partly because I was so disappointed by the art works I saw in galleries.
(Later I was to exhibit equally bad and disappointing failures of interactive art. It's hard to get right.)
Inner life (Already moving) - The work should have its own internal dynamic, rather than wait for the viewer / user to push it around in order to see anything working. The user is thrown into interaction with it. (See also Gbloink!)
Proportionality - Promise of movement should correspond to actual movement. Don't show lots of things that look as though they can be interacted with but can't be. Input effort should correspond to complexity of output. - Or, at least, the work should exhibit a multiplier effect. Works that do little but seem thrown together as an effortless sketch can be beautiful. Works which seem like they were a "faff" to produce but don't do much feel like great failures. (I wonder is there's an HonestMaterials aspect to this?)
Historic forward motion - it should feel as though it's making progress somehow. Advancing the direction of development or revisiting to rethink existing moments in art history. Shouldn't just be A.N.Other use of a current technology. (See also RootsAndFuture)
Contextual appropriateness - fits with the other works in the gallery.
WabiSabi - shows imperfections and the passing of time.
See also : * I wonder how the interactive of the NewAesthetic fares compared to this?