Discussion about this post

User's avatar
Noah MacCallum's avatar

Also, another cool example: https://www.sciencealert.com/octopus-and-squid-evolution-is-weirder-than-we-could-have-ever-imagined

Octopus heavily rely on post-transcriptional RNA modification to adapt to their environment, an approach that's more error-prone than DNA editing but leads to faster iteration times. An intermediate data point between evolution of genes and the adaptation of brains!

Expand full comment
Noah MacCallum's avatar

Thanks Evelyn! I think you're pointing to an interesting distinction, where for human-produced app-platform systems, there is marked cost difference between the two, and this makes it harder to produce new hardware. For living systems, all genetic changes are just as "costly", but the more platform-y systems are just in very deep and wide local minima, requiring many genetic changes for improvement that are progressively improbable. I do look forward to humans having a way to evolve the whole stack, but that would likely require full programmability of matter itself (and, I suppose, some smart way of navigating this extremely large search space - see AI Drug discovery via search algorithms for some cool inspiration there, maybe fodder for a new post!).

Expand full comment
1 more comment...

No posts