When you are making huge batches and you are running continuously, and if you are rich enough to scrap a marginal batch, well then, things are nearly perfect. It appears as if things are not near perfect.

Of course, this assumes that Marks process and data are spot on.