Critique of Cursor's 'browser experiment' claims.