count the keys on the keyservers, apply some multiplier, chances are that it's still less than signal users today.
> ..., apply some multiplier, ...
You glossed over that like it was nothing. Let me rephrase GP: "... with no possible way of knowing the multiplier".
Obviously there exist k such that for n keys on "the keyservers" (we'll have fun enumerating those too) and s signal users, k*n > s.