On 17.6.2012, at 7.59, Michael Ossipoff wrote: > Juhos: > > Ok, you're right in a way: LR _does_ have something right it does: LR puts > the allocation as close as possible to the ideal fractional allocation. It > does that when it preferentially gives seats to parties or districts with > largest remainders. > > Who's to say that that isn't what you should want. And, if that's what you > want, then it's more important than the paradoxes, and it justifies the > paradoxes and makes them excusable. > > So why isn't it my favorite? When LR puts the allocation as close as possible > to the ideal fractional allocation, it's making the allocation as right as > possible. But Sainte-Lague looks at something else: Sainte-Lague instead > looks at it from the point of view of the individual voters or > district-residents.Each person has a right to say that s/he deserves the same > party representation or district representation as anyone else. The same > seats per person.
Ok, and the same with Largest Remainder. Also it can be said to do what the voters or distrrict-residents want. There are different possible criteria that are not compatible. > > So, Sainte-Lague minimizes differences among different people's s/p (seats > per person). Sainte-Lague differs from Largest-Remainder by looking at it > from the rights point of view of the individual voters or district residents. > > But there are two ways that we could measure how the different people's s/p > differs. You could minimize the differences--subtractive differences, by > which districts' or parties' s/p differ; or you could compare the _factors_ > by which the districts' or parties' s/p differ. Sainte-Lague does the former, > and Hall's method (currently in use for HR apportionment here) does the > latter. > > As I was saying, subtractive difference is more like what matters in > Congressional voting.That's why I prefer Sainte-Lague/Webster to Hall. I > presume tha that's why B&L prefer Webster too, but you'd have to look at > their book to be sure. > > You said: > > LR thus focuses more on the number of voters whose rights are violated while > SL focuses on the proportion (proportion changes in a small group means less > people/voters than in a large group). > > [endquote] > > I don't think that fairness is a about need to move voters around, from one > district or party or another. I've never heard of anyone's notion of fairness > to be about that. That is a quite natural measure and criterion. Maybe more so than the idea of differences in the proportion of seats per person. Election/allocation methods are supposed to do what the people say, and leaving some people unnoticed or noticing non-existing ghost people sounds like exactly what the methods should not do. > > I don't see how LR counts how many voter (or district members') rights are > violated. SL looks at _individual_ persons' rights to equal representation. > LR instead look at how overall right the allocation looks--how close it is to > the ideal fractional allocation. The resulting allocation can be compared with the ideal (fractional) allocation, and the difference can be given as number of people (easier to understand than e.g. difference in quota). > > I'm not saying that LR's standard is wrong. I'm merely saying that I'm more > interested in the individual persons' rights. I guess the same can be said about people who adopt LR for some use. Juho > > Mike Ossipoff > ---- Election-Methods mailing list - see http://electorama.com/em for list info
