My problem (and I think a lot of other people's) is not with the arbitrariness of lines so much as the fact that you can run a longer program through a very dumb minifier and get a "short" program just by sacrificing readability. It's still very much the same program; we're just rewarding people for making it harder for us to read. This problem isn't solved by switching the requirement to bytes. I would sooner accept a token limit than a byte limit, because it's harder to elide tokens to the point of unreadability in most languages.