Little detail not sure I reported before. In the many-voter election being studied (candidate's utilities for ABC are 210), I looked at the subset of the table that involves only Approval Votes, i.e., full scale or min (2 or 0).
The utility for sincere voting Range, stated as gain over not voting, as percentage of maximum possible (favorite guaranteed to win with this strategy, perhaps it is the stuff-the-ballot-box strategy): 48.15% Utility for Approval Voting strategy (it does not matter if the voter votes 200 or 210 with many voters): 44.44% Utility for Approval *election,* i.e., all votes are Approval style. 41.67% In this situation, zero-knowledge, 3 candidates, utilities evenly spaced (preference strength the same for A>B as B>C), it appears that not only does Approval *strategy* worsen the outcome for the voter, but Approval *voting* makes it even worse. I think this effect comes from the ties; ties are more common with Approval votes. This study only looks at the effect on utility of votes presented in a context where the vote is possibly not moot, i.e., there is a vote that the voter can cast that can affect the outcome. We then assume that the voter only actually casts votes that are designed to advance the voter's utilities; thus the favorite is rated max only, the least-preferred is only rated zero, and the middle candidate can be rated 0, 1, or 2. We are studying the effect, essentially, of how the middle candidate is rated. I am positing that this is a legitimate way to consider the utility of voting strategy with Range Voting, and it can be extended to other voting contexts, i.e., the same method could be used to study plurality, IRV, etc. However, if the conditions are not kept simple, the complexity of the study increases very rapidly. ---- Election-Methods mailing list - see http://electorama.com/em for list info
