On Wed, Jun 13, 2012 at 2:21 AM, Juho Laatu <[email protected]> wrote: > On 13.6.2012, at 3.06, Michael Ossipoff wrote: > >> Though census-based re-districting is usually discussed for >> single-member districts, there's no reason why it couldn't be used for >> multi-member districts. > > That's true if the district borders are just random borders drawn on the map. > The borderlines may however be also natural borderlines. For example in > >Finland the districts are for the most part historical areas further back > from the history than the country's independence. I guess many people would > >feel upset if they would be forced to be represented by other than the > representatives of their own historical area.
Ok, that's a good reason to keep the historical districts. In that case, that explains why PR allocation of seats to districts is the best solution. > > One key element in democracy is to structure administration (families, > municipalities counties, states, countries) and representation based on > natural >borders. Well, if you mean rivers and ridges, gullies, etc., if that's what people want, it too would work fine with PR. The problem there would be that different people could have different opinions about which parts of which natural line would be used for which part of a district boundary, and it could be gerrymandered. But you were talking about _well established historical boundaries_ that might be based largely on natural lines such as rivers, ridges and gullies. > > I also note that of course the differences in regional disproportionality due > to the sizes of multi-member districts are much smaller than regional > >disproportionality caused by single-member districts. Not a valid comparison. Single member districts aren't intended or meant to be proportional. No one claims that they're proportional. They're knowingly,intentionally used instead of PR. >In the case of Finland the main problem with regard to district sizes is the >problem that it is very >difficult for the smallest parties to get any >>representatives in the smallest districts. There will thus be bias in >political proportionality (not so much in >regional proportionality). If political proportionality and inclusion is considered important, then maybe some of the existing districts could combine, to improve that. Anyway, I understand why automated districting isn't used there, and the reason makes good sense. A small randomly varying difference in district representation, neutral when averaged over time, might be a small price for keeping the historical district boundaries that are meaningful to people. But, if optimally equal district representation per person is desired, then you want Sainte-Lague for allocating seats to districts. Mike Ossipoff > > Juho > > > > > ---- > Election-Methods mailing list - see http://electorama.com/em for list info ---- Election-Methods mailing list - see http://electorama.com/em for list info
