The closed models aren't giving you the real thinking traces, though.