Stephen Wolfram: A New Kind of Science

Also on Twitter: https://twitter.com/i/broadcasts/1mrGmabyNovGy

You know how every six months someone posts an impassioned screed about how programming needs to be visual? And then the collective community has to explain how this isn't a new idea, it's been used to do a couple of things, and shows no signs of moving out of that niche because as soon as you actually put the elbow grease in to try to make a general-purpose "visual language" a whole lot of details the beautiful model in your head didn't have to worry about stick their ugly faces right in and destroy the entire thing?

The belief that all of math can be reduced to cellular automata is a very similar belief, with I suspect very similar sources of appeal. There's these really beautiful pictures that cellular automata make, and they neatly explain how a couple of sea shells look, can produce some interesting patterns both visually and mathematically, and have an interesting handful of proofs like the Turing Completeness of some rules.

However, when you try to get them to do anything useful, it all just comes apart. Cellular automata are in general just too chaotic to be harnessed, and by the time you tame their chaoticness you've probably just badly reinvented something that already exists (e.g., yes, great, you embedded a Turing Machine in your cellular automata, but we already have Turing Machines that are way better).

I read a decent chunk of the book. It had pretty pictures and interesting ideas. But given its claims, I also read it with an engineering eye. From an engineering point of view, the book verges on vacuous. I like pure math for its own sake, so had it merely claimed to be an interesting dive into the pure math world of cellular automata I would hold nothing against it. But as a revolution it has been sorely lacking in much of anything, even by math standards, to say nothing of engineering standards. Given that it is now 20 years old, claims of how important it is going to be in the future also ring quite hollow.

But the dream never dies because my gosh are those pictures pretty. Surely the fact that the pictures are Beautiful means they must also be True?

Maybe in Philosophy it does, maybe it doesn't. But it certainly doesn't in Engineering.

From my recollection, it's a mix of interesting work (some of it actually done by him, lots not but not clearly attributed), semi-academic wankery, a pretty good demonstration of what happens when you don't have a good editor....

I think his premise is that cellular automata can model literally anything.

Simple rules do not necessarily mean simple behavior. Explore where observation leads in various different systems and rules.