the brass doesn't crumble, it loses zinc and gets softer. That means it resizes differently for 1, for 2 you get a lot more flow of material from just above the web area which leads to pre-mature case head separation. Doing a small sample in a custom chamber without a control group and larger data set is hardly scientific or conclusive. It is the exception not the rule.
76 firings on 1 case with annealing. Also, probably the exception, but let me see the test where un-annealed brass goes to remotely that close.