First, if this is going to be at all successful, the implementers are going to watch, and see what happens with the system in real-world use. Unless they happened on the perfect system right at the gate, there will be room for improvement. Second, they may have more than one category, not visible to the user. Maybe "complainers" (people who always report on people who kill them) and "bullies" (campers and griefers) will both get a red rating over time. (Yes, the guy who responds to everyone who's better than him by labeling him a cheater is a bad actor, just of a different style than the bully.) And if they're put in different categories, then you'll have a bunch of red guys who all happen to be complainers accusing each other of having nice aimbots and another bunch of guys saying "fuck you!" when someone kills them and hunting each other down for the rest of the session. Couple this with a little bit of trust management, where the opinions of people who are higher in your levels of trust will make their opinions carry more weight for your experience, and therefore who you're likely to be playing with. I don't need to know someone very well to decide that I think they're a bit of a dick, and a good game will keep us separate just to make everyone's experience better. And my opinion might be shared by my friends for a number of reasons, some of them valid.
Really, you're thinking about this too simplistically. Most of these games aren't some seasonal league where you have to play a bunch of the other teams to see who's the best, and the people who want to play will generally be enough that you can use some discrimination in your grouping to keep conflicting personalities apart without diminishing gameplay for anyone, let alone the good actors.