This is kind of an aside, but nowadays we could at least be counting (lexer) tokens of code instead of lines of code. Or even number of AST edges.