Forgive the ">" marks--I could only post this by forwarding it, because this computer doesn't have copying.
Warren-- > >Regarding the value of K, in the exponential approximation to the >state-size probability distribution, why should it be equal to the number >of states divided by the number of seats. The reason I ask is because >something like that would make things a lot easier, because finding A and B >(K and K') by least squares would be a lot of work. > >But why should the number of seats have a role in the formula that >approximates the probability density of state-sizes? > >As I said, I ask because any fact like that would greatly make easier the >task of fitting that curve. > >And I hope that you will add at least the first of the additions listed >below, to the quote from me at the website. If adding both of the additions >would give me too much space at the website, I'd understand, but at least >the brief 1st addition should be added. Surely I should have a little input >about what I say. > >Here's my request about that: > >Right after the quote from me, about equal representation expectation for >everyone being the natural goal, I'd like to add: > >"The original method can be derived from that goal." > >[end of first suggested addition] > >Then, optionally, if you don' mind adding this much: > >"For some particular two consecutive whole number numbers of population >quotas, and for the states in that population region, write expressions for >the total expected number of seats, and for the total expected number of >quotas possessed by those states. Set those two expressions equal, and >solve for the rounding point, R. That R, found in that way, makes the >expected s/q in that region =1. When R is so chosen in each such region, >between each pair of consecutive whole numbers, then all of those regions >have their expected s/q equal to 1, and therefore equal to eachother. > >Then, if we look at it with regard to those regions, without regard to >where in those regions the states are, everyone has the same representation >expectation. If we look at states' positions on a finer level than those >regions, there could be no such thing as unbias or equal representation >expectation. So this method makes everyone's represenation expection be >equal to the extent that that can be done." > > >[end of 2nd suggested addition] > >If the 2nd addition is more than you want to add, then just add the first >one. Otherwise, could you add both? > > >Mike > > ---- election-methods mailing list - see http://electorama.com/em for list info
