[msla]

> It's a common idea that evolution propagates useful mutations, and the non-useful disappear.

Common ideas are commonly wrong. Non-useful ones only disappear reliably if they're selected against. Otherwise, you get goose bumps, or piloerection in humans: We don't have enough hair over most of our bodies for erecting it to do any good, but we still have all the structures to erect it anyway, because none of our ancestors lost any of them even as they became useless.

[mjburgess]

"Useful" is also context-dependent in two ways: it is relative to the environment of adaption; and also! relative to the existing genotype.

ie., evolution is starting with a bike and turning it into a car

it is not "starting with raw materials" and shaping them

So pretty much every life form is highly non-optimal wrt their environments.

[msla]

My point is that, as an optimizer, evolution is a very simplistic one. People try to explain why it does or doesn't do things, often falling into "just-so stories" as a result, and fail to grasp just how inhuman the process is.

Also:

> ie., evolution is starting with a bike and turning it into a car

> it is not "starting with raw materials" and shaping them

I'm not sure if you're drawing a distinction between evolution and abiogenesis here. You're right, in that technically evolution doesn't come into play until you have self-replicators of some form, but in colloquial terminology "evolution" is used to encompass the entire process all the way back to whatever nonliving organic chemicals formed the basis for the first life.

[TeMPOraL]

I understood them to mean that evolution evolves things on an ongoing basis, and doesn't start from abiogenesis each time. So the current form of a bird species is very much constrained by the form of what it evolved from.

I'd also add that it seems evolution isn't even turning bicycles to cars bolt by bolt, but is adding whole subsystems at a time. It's as if each bike was made of a set of standard ACME parts, and came with a set of blueprints for many other kinds of ACME parts - and evolution, at the large animal level, is just replacing one standardized part for another, or adding new ones where they didn't exist, or altering the manufacturing timings, etc.

I'm of course referring to the findings of evolutionary developmental biology - https://en.wikipedia.org/wiki/Evolutionary_developmental_bio....