Strictly speaking, we can exhibit any definition of "good", even one that doesn't make any of the errors you pointed out, and still ask "Is it good?".

Moore was correct that no alternate concrete meaning is identical to 'good,' his mistake was thinking that 'good' had its own concrete meaning. As Paul Ziff put it, good means 'answers to an interest' where the interest is semantically variable.

In math terms, the open question argument would be like asking the value of f(z) and when someone answers f(3), pointing out that f(z) is not the same thing as f(3).

I think the 'huge and complicated' X that Luke mentions is supposed to be the set all of inputs to f(z) that a given person is disposed to use. Or maybe the aggregate of all such sets for people.