It's really hard, though, to see this systemically with fresh eyes.
Today for instance REPLACE takes the thing to do the replacement in first:
>> replace [a b b a] 'b 'c
== [a c c a]
But if this philosophy were extended that would become either of:
>> replace 'c 'b [a b b a]
== [a c c a]
>> replace 'b 'c [a b b a]
== [a c c a]
It's hard to say which makes more sense. If the "how you would write it out in English" argument comes up, it seems briefest as:
replace 'b <with> 'c <in> [a b b a]
I'm not sure how that would hammer out in the other case:
replace 'c <wherever you find> 'b <in> [a b b a]
But once you move away from the "obvious-seeming" cases, the whole thing stops seeming obvious. There's a lot of inertia behind the historical choices...this is tough to reason about.