I think it's pattern matching and intuition. If I see y = k/x, I can immediately process what that looks like. Further, I can take advantage of tons of conventions that make it read differently from a = b/c or even y = z/x. The conventions take some getting used to, but after a while writing out dependentVariable = divideBy(constant, independentVariable) would get super tedious and it's absolutely not instantly pattern matchable. Especially when you're working out some manipulation where you have to write the same variable a zillion times, and maybe even depending on your brain's pattern matching and inspiration to kick in. Hell, at this point I even use a shorthand for cos and sin for those reasons.

I’m sure part of the reason is that mathematics was, and largely still is, written by hand. You can manipulate “code” more easily on paper with a terse, implicit notation. It can also have the advantage of engaging your visual pattern-matching abilities more than prose or prose-like notation.

Part of the reason is that most variables do not have much meaning.

What name do you ascribe to "a continuous function from R to R"?

Another detail is that you are rarely juggling more than 10 or so names at once.

A lot of math at a higher level becomes abstract enough to where elements no longer have a concrete meaning. What does the x in integral(f(x)dx) mean? It's more about some things being harder to write in point free style rather than anything else.

This is perhaps why it's common to have short names for arguments in functional languages as well.