Benefits of having "trusted / high quality" user recognition
I've had a go (per Piotrus/Sue Gardner) at designing a hybrid approach. Rough concept is here, and please comment on the talk page if it's too off-topic for this thread.
It doesn't have to be "perfect", but it should be hard to game and fairly good for identifying good quality content editors, simple, and low overhead on individuals and community.
The key aims are automation, low gameability, simplicity of experience to users, very low scope for politicking/dramatizing/popularity contests, and low time needed by participants. I feel very strongly that automation alone (metrics for "trusted users") aren't viable, despite Sue's valid point. What we can easily do with existing tools is streamline it so far, that it's almost as efficient and substantively keeps all the benefits of both.
This one's a concept (rough only I'm afraid) - a hybrid of enwiki Mediation Committee's nomination method (demands filtering of good quality users and operates historically with no drama whatsoever) and a modification of the SecurePoll tool already in place.
That's the direction I'm thinking. It's a bit more involved than 100% automation, but it is simple (once set up) and keeps almost all the benefits of automation, all the benefits of user involvement, and very little of the drawbacks of either, when merged.
One thing I strongly encourage people to keep in mind: any system that is set up will be gameable. All of them. When I was doing corporate training, we had a rule that you "train to the norm, not to the exception." The idea was, of course, that you write a process or a training scheme that will work MOST of the time. Someone's always going to be an exception. Someone's always going to game the system, but if we can make it work 90% of the time, that's good enough.
The perfect is the enemy of the good.
Covered. I think I said almost the same above - you design it to be 90 - 95% good,which means it's slightly gameable. But you counter that by making sure removal is also to-the-point, and some kind of scrutineers exist for "surprising" results where there is a widespread suspicion of gaming or undue conclusion.