Sorry, forgot to hit reply all. I think that people have enough trouble understanding the basic diagrams without adding more complexity.
OTOH, it would be interesting to be able to give the chance of failure for each method, rather than just showing a few examples of failures. Otherwise, you can be accused of cherry picking. Also, even with your suggestion, it would probably be worth having multiple diagrams with various standard diviations. On your pathologies, (1) non-convex I am not sure if this is actually an issue. (2) non-star like This is an issue. It is the global version of a monotonicity failure. If the centre of the voters moves towards a candidate, it shouldn't result in the candidate losing. (3) non-connected This cannot happen unless the method also fails (2) (4) candidate is not in own win region (5) no win region I would agree that these are issues. It can be caused if the method is exremist seeking (centre squeeze) or centerist seeking. Maybe it would be worth trying to distinguish between the 2 types of methods. ---- Election-Methods mailing list - see http://electorama.com/em for list info
