Just make a program that spits out random strings, run it as code and see if it compiles. If not, repeat. That way, just like the apes typewriter thing, you find the most amazing code that is so complicated, nobody would have thought of, but it solves all of humanities problems, especially the power problem you created by running this code 1000000000000000000000 times parallel.

Yeah, but you'll also find the code that will break containment and launch all of the nukes. Safer to have it write random gibberish until it writes a thesis on a nuke safe monkeys and typewriters coding program 😋

Also known as: genetic algorithms
(That's the same way your code came to be!)

@Ashkin never heared of that. Sounds interesting

Given goals, you randomize the input (code, settings, environ, circuitry traces, ... whatever you're trying to perfect) a few thousand times, and run the simulation. Score them based on how close they came to achieving the goals.

Take whichever ones score the highest and use those as seeds for the next generation. Either by modifying them, by merging them together, or whatever approach you like.

What's interesting is the approach you use doesn't matter. The project will continue to evolve until eventually it achieves all of the goals (granted they're possible).

But... good luck understanding how.

@Ashkin oh yea, that. I love thr videos by carykh on that. But it only works because it is not entirely randomized, otherwise it would take too long.

Well of course ☺

(Though pure random would eventually work, too! Humans are proof of that.)

@Ashkin I would argue that human are not a result of pure randomness. Its not like random atoms are put together until a human comes out but rather millions of generations of evolution with small changes each generation

@jobylie
... which is exactly the process described above.

Each organism in each generation changing slightly due to arbitrarily different environments and mutations in dna/etc. Essentially random changes.

the original building blocks? Essentially random things coming together and happening to work.

But this is why it's called genetic algorithms: it very closely mimicks how DNA evolves over time.

It's fascinating!