(To state it in AI lingo:)
It's not about the best measure for "amount of code".
It's about wether "amount of code" is a good metric to begin with.