@sparklefish, yes I love your approach to identifying issues and solving them. This is a great proposal. We definitely get a lot of theory crafting based on personal feelings, and attempts at fact finding for the purpose of proving or reaffirming these personal feelings. So I really appreciate this.
With some of these specific implementations, the reason for implementation for rematches for example as I explained was not to infer that people stopped having fun after 3 matches. I can go into this more again, but I might be just being redundant from my email. But it's tough, the system is compelled to straddle the social features, speed, and balance, and so if we want both balance, speed, and social features, we have to recognize that the system will never be the idea of perfection. Therefore, any device we design to collect and analyze data has to be unbiased. I can't have the mindset of designing a mechanism for feedback and data analysis for the purpose of proving a specific point right or wrong. With things we do with the match maker, we look at data yes, but we do look at what issues people are reporting. Not "I hate this," or "I love that," but what issues and use cases people are reporting. We try to use data to see how severe these issues are. And I think that this is the point you're making.
Great stuff.