Adventures in Gnarly Computation

Rudy Rucker, entertaining as always. A sample:

If we suppose that many natural phenomena are in effect computations, the study of computer science can tell us about the kinds of natural phenomena that can occur. Starting in the 1980s, the scientist-entrepreneur Stephen Wolfram did a king-hell job of combing through vast seas of possible computations, getting a handle on the kinds of phenomena that can occur, exploring the computational universe.

Simplifying just a bit, we can say that Wolfram found three kinds of processes: the predictable, the random-looking, and what I term the gnarly. These three fall into a Goldilocks pattern:

•Too cold (predictable): Processes that produce no real surprises. This may be because they die out and become constant, or because they’re repetitive in some way. The repetitions can be spatial, temporal, or scaled so as to make fractally nested patterns that are nevertheless predictable.

•Too hot (random-looking): Processes that are completely scuzzy and messy and dull, like white noise or video snow. The programmer William Gosper used to refer to computational rules of this kind as “seething dog barf.”

•Just right (gnarly): Processes that are structured in interesting ways but nonetheless unpredictable. In computations of this kind we see coherent patterns moving around like gliders; these patterns produce large-scale information transport across the space of the computation. Gnarly processes often display patterns at several scales. We find them fun to watch because they tend to appear as if they’re alive.

Enjoy.