See my edit to my last post. Given the realities of buffer-based audio streaming and disk-based recording, there's no way the written file can initially start and stop exactly where the user punches (or autopunches) in and out. The corrections have to be made after the fact, and the Bakers have just chosen not to apply them destructively... until now. I think there are arguments to be made on both sides as to whether they should extend destructive compensation to the other cases.