> Was it just things we couldn't observe at all without modern scanning electron micrography?
Perhaps also that we couldn't imagine, much less accept, that a dumb semi-random process tweaking bits could, over time, organize those bits into higher level abstract structures. We still mostly talk about evolution mutating genes, where perhaps if you zoom out a little, it's actually working at a higher level of abstraction.
Incidentally, this is the same outlook as some people have today wrt. LLMs - they can't accept the idea that backprop running on a blob of weights representing a simple (if large) graph can start encoding increasingly high-level, abstract organizational structures.
Perhaps also that we couldn't imagine, much less accept, that a dumb semi-random process tweaking bits could, over time, organize those bits into higher level abstract structures. We still mostly talk about evolution mutating genes, where perhaps if you zoom out a little, it's actually working at a higher level of abstraction.
Incidentally, this is the same outlook as some people have today wrt. LLMs - they can't accept the idea that backprop running on a blob of weights representing a simple (if large) graph can start encoding increasingly high-level, abstract organizational structures.