On Criticism

In Monday's talk in Brussels, I tried to build on my late-2003 lecture (H2TPM, Paris) on software aesthetics. Back then, I tried to list things we value -- and things we should value -- in software, and explored what software could learn from literary hypertext. This time, I asked why we don't seem to be learning -- specifically, why it seems that all our software is bad.

Have you noticed how deeply flawed we seem to consider all our software? The best software, if you listen to what people tell you, is invisible: software that always works, consistently does what its told, software that doesn't make you think. It's as if the only painters we admired were painting walls.

I fumbled one section of the talk when I improvised an excursion on brush-marks in software. I tried to say that we're all tied up in surface and in polish, in making software appear perfectly seamless and clothed in radiant brushed metal draperies. Someday, maybe we'll be able to remember that software is made, it's an artifact, that it naturally carries the equivalent of chisel marks or brush strokes and that these are not flaws.

Memo: interesting position. Quite possibly worth pursuing. But way too much for an improvised intermezzo.

Intimidated by an audience of philosophers of science, I fumbled the best question, which asked how software aesthetics relate to the aesthetics of physical science. My initial impression is that this was a difficult question. It's not -- not right now, anyway. Scientists know when they're wrong, and can feel reasonably confident that they're right. I don't think software creators are ever very confident that they're right. We're better off than novelists or politicians, but on the whole it's a lot easier for chemists (and cooks) to know when the theory is properly cooked.