Continuing the long tradition of thinking out loud on the EM list to make enough blunders to keep things interesting, I wonder about the following idea:
Use a Proportional Representation method in each round of a runoff to see which subset of candidates survives to the next round. The motivation for this idea is elimination of primaries. If there are no primaries, then there are lots of clones, especially from the big parties. IRV, approval runoff, and other common sequential elimination methods allow some of these clones (the ones from the biggest parties) to survive to the late stages of the game, while the little party candidates wipe each other out early in the game. It seems to me that if PR methods were used for deciding the survivors at each stage, then the clones would be eliminated earlier, and the choices would be clearer at the later stages. In a sense, IRV is a crude attempt to do something like this, because it uses plurality runoff, and plurality can be used as a crude PR method when careful attention is given to strategy. But in order to carry out IRV you have to collect enough information to do full scale PR, so why not take advantage of that? Forest
