I have no problem with that approach, but the cost-benefit analysis needs to take into account the likelihood that some huge percentage of your user-base can't

I have no problem with that approach, but the cost-benefit analysis needs to take into account the likelihood that some huge percentage of your user-base can't run the app. If 50% of my users get a crash at start-up, I'm going to try to make that go away. If no one is really having the problem, then it's silly to ask for Apple to block potentially legitimate feedback. If loads of people are having the problem, then the cost-benefit analysis ought to look very different.