The problem with going strict win-loss is that not all opponents are equal. If I as a 3.5 player happen to play most of my matches against 3.0s playing up, then I would be expected to have a very good win-loss record. This wouldn't mean that I should be bumped to 4.0. Using both players' dynamic rating and the actual match scores creates some opportunity to differentiate a "strong" win from a "weak" win. I think that the concept is sound even if the actual application is imperfect.
Your point that people who win all the time should be moved up is valid as well. I think that the step to prohibit ratings appeals for people playing at championship (district and higher) levels is a good start. Assigning a higher "weight" to championship matches such that participants generate larger dynamics wouldn't bother me either.
No matter what you do, though, there will be people that game the system. I think Javier is right - most of the time things work pretty well. Sandbaggers eventually get moved up and life goes on.