Captains1911, your point that we don't have hard data is quite correct, and I don't mean to dump on you, but assuming that there are thousands of good ones for every one with problems reveals a bias in the other direction. When you see properly functioning ones at competitions and classes, your sample is biased towards working ones because ones with problems won't make it to competition and might not be brought to a class, if the student is aware of the problem ahead of time. I am perhaps more sensitive to such things because I see similar reasoning all the time in medicine; specialists report large proportions of complex diagnoses and forget that the general practitioners are solving the simple problems without referral.
All we really know is that the problem is more than zero and less than 100%. (Actually, we also know that Glock made recommendations for hot ammo and break-in periods that they never did before, and changed parts multiple times, for a problem that they never openly acknowledged, either. I would have had more confidence in the company if they had handled it better.)