I think what was meant was that the semantics should be optimized to human. Those semantics matter more then the syntax, since you can quickly get used to syntax.

So think of it as learning to recognize and pronounce an alphabet used to form words, over understanding the meaning of sentences in a given language.

Having said that, once you get used to the lisp syntax, as I did, I find it is better suited to humans. But to believe it you need to try out structural editing.

Hmm, that seems to imply that you can't optimize for semantics and syntax at the same time.

It can optimize for both, in fact that's what Lisp is all about. Take Hy, its got Python's semantics, supports all of its features and paradigms, but its syntax is simpler, more uniform, homoiconic and more powerful. This is turn allows you to extend, modify or refine Python's semantics, since the Lisp syntax will always be able to express them.

> they should be optimized for the humans using them, not for the machines that run the programs or the compiler authors who create them.

Almost every high-level language espouses this goal -- even something as seemingly esoteric as J specifically does. Which is to say, your argument assumes that python's syntax is somehow more suited to humans, but I think you'd be hard pressed to back that up with evidence. Clearly it feels more natural to you, but why? It's hard to control for familiarity even in your own personal case, let alone programmers in general.