On Dec 30, 2007, at 2:14 AM, Dan Bishop wrote: > In the coming weeks/months/whenever-I-feel-like-it, I will be > performing > simulations to evaluate the performance of multi-winner methods. In > order to do this, I will make the assumptions that: > > * There is a uniform linear political spectrum. (Other models of > voter > behavior will be considered later.) > * Candidates are uniformly-distributed random variables in. > * All votes are sincere. (i.e., a voter at position V votes A>B iff > abs(A-V) < abs(B-V)) > > The last two assumptions, that no strategy is involved in either > nominations or voting, is admittedly unrealistic. But, as I see it, > in > order to know the best strategy to use with a method, you must first > know how it would behave without strategy, so that's a useful thing to > analyze.
The first assumption is pretty unrealistic as well. I look forward to seeing your other voter models. For example, I was thinking of the problem faced by US Democratic primary voters during the current cycle. Granted, they don't have to rank the candidates (except in Iowa, where there's a limited contingency process), but they still have to come up with a first choice. Looking at the candidates, though, it seems to me not terribly useful to try to order them in any one dimension, and to the extent that they can be ordered on multiple dimensions, very difficult to resolve those rankings in to a single ordering (or a first choice). Many voters appear to value "electability", for example. But we're faced with the difficulty not only of trying to rank (and perhaps weight) electability, but then to figure out how that weighting figures into some larger process that looks at how effective the candidate might be if elected, how we feel about their stands on various issues (Iraq, health care, etc). Then there are second- (third-?) order considerations. I may think that Iraq is a more important issue than health care, but question the extent to which any particular candidate, however much I might identify with their position, can do about it. So maybe I should weight health care plans more heavily, if I think that we're "ready" for serious reform. And so on. I tentatively conclude that it's not very helpful to model that kind of decision process, partly because it's impossibly difficult to model and, more important, to validate the model. And here's where Gilmour's 'contingency' becomes especially relevant. Voters aren't dispassionate computers with elaborate linear programming built in. What they *can* do (we know because they actually *do* it) is to decide, finally, on their first choice. And having accepted that, it's no great leap to assume that they can decide on a second choice contingent on the unavailability, for whatever reason, of their first choice. I'll try not to turn this into a pitch for STV, but it strikes me that one of its attractions may be that with its explicit reliance on contingent choices, it asks voters to solve a problem they're prepared to deal with, without solving impossible resolutions of conflicting-- and vague--personal value scales. As a voter, I prefer to think in terms of contingent choices. I don't *want* to solve the strategy problem presented by approval voting. With range voting, I don't feel that I'm in a position to assign cardinal weightings to the candidates (never mind for the moment how they're ultimately counted), nor do I want to be burdened with then turning those numbers into a strategic game to maximize my outcome. Sure, STV is manipulable. But so is everything else. And its implicit decision model conforms to something that I, as a voter, can make sense of. And in making my contingent choices, STV's later-no-harm property becomes important to me. Of course I'd like it to be monotonic, etc, but on *my* value scale, its advantages, in practice, outweigh its shortcomings--in practice. ---- Election-Methods mailing list - see http://electorama.com/em for list info
