The problems are those things where there's a trade-off.
For example, a highly productive member may be very abrasive.
Are they too abrasive for their productivity or not? In other words, are they of net benefit to the project or not?
That's a trade-off.
If you don't consider it as a trade-off then bad things happen, you can lose the most productive members.
Basically whenever there's a tradeoff, consensus on any individual thing (e.g. 'civility') is highly likely to fail- that everyone should be civil will normally be consensus, but what about other factors surrounding contributors? Consensus on civility would be that everyone uncivil MUST be banned!!!!
That's where leadership of one form or another comes in; you have to say that civility is important, and how important other things are as WELL. It's the relative importance that matters.
What arbcom does is that the candidates state what they stand for on the areas that have to be traded off and then they get elected and make decisions (hopefully) along the lines that they were voted in for. That's why arbcom more or less works.
What other trade offs are there in/around the Wikipedia?
Other areas might be things like policies, there very much are areas where people are deliberately writing the policies differently in different parts so that they can delete things they don't like, even though the policies, on the whole, probably don't permit them to do that; if you write something into the corner of a policy somewhere and then edit war that to stick with a group, then it's very hard to remove, even if people in general looked at them wouldn't agree with it.
So the Wikipedia could go to more of a parliamentary type system where parliament writes the policies and tries to keep them consistent.
What other trade-offs do people see?