Alex wrote > OK, maybe I went out on a limb with some of my statements about PR. Let > me offer a more reasonable rationale for including single-member districts > with PR in a bicameral system: > > The electorate is made of many group of people, each with their own > preferences. It's rare that a single public policy will reflect the first > choice of a majority of the voters, but it's possible to come up with > policies that at least somewhat satisfy a majority, via compromise. > > There are 2 ways to find that compromise. The first is that we all > indicate our ideal policies, via the proxy of a party (assume that we have > a wide range of parties, with diverse platforms, so indicating your > favorite party gives an approximate indication of your ideal party > preferences). The parties get seats more or less in proportion to how > many voters support them, and then the legislators hammer out the > compromises via the legislative process.
Why you persist in talking only about "the parties"? Why do you not consider "the candidates"? PR restricted to parties does restrict the representation of the voters. But a more sensitive system of PR (like STV) that allows the voters to choose freely among the candidates as individuals as well as party nominees, gives more scope for representation of the variations of views among the electors. Then the voters can dictate the balance of views and the compromises, or at least they can so far as there are differences of views among the candidates - and those differences are often quite great. > The second is that We the People identify compromises. We do this with a > single-winner method that's better than plurality (since plurality makes > it almost impossible for more than 2 parties to compete, and hence fails > to offer a wide enough range of choices to successfully identify a > compromise that satisfies the largest number of voters). I would argue > that Approval or Condorcet is the best for that purpose, but that is a > point to discuss later. The point is that in each district, we identify > the candidate who satisfies the greatest number of voters, and we send him > to the legislature. But you still leave half of those who vote without any representation or say at all. How much better to give 85% or 90% of those who vote a say through the election of candidates who are their direct representatives. And your single-winner selection is not a very sensitive tool - no matter how good the method of voting. > Obviously, single-member districts have their problems, so in a bicameral > system the other house should be elected by PR. You would almost certainly get very different results in terms of party representation. Which would have the greater legitimacy? I know which is likely to be the more representative. > You could argue for > electing both houses by PR, but then the houses are more or less redundant > with one another. Not necessarily. For example, the structure of the two houses could be quite different, eg like the Australian House of Representatives and the Senate > The largest defect of single-member districts is > probably gerrymandering, and that can be fixed: No. The largest single defect of single-member districts is that they usually deny representation to half of those who vote. At best, they can guarantee representation to only half. The second largest defect of single-member districts is that they commonly distort the wishes of the voters in terms of overall representation. Gerrymandering is a defect of single-member districts, but it comes third on my list. > Very large districts and > very small districts are difficult to gerrymander. On the very large and > very small scales your ability to specifically include or exclude key > voter blocs is rather limited. Also, as discussed on this list about a > year ago, there are non-partisan districting algorithms that one can use. But in the world of real politics these will not be accepted. No matter how non-partisan the districting, there are still major problems when natural communities have to be split to give the required numerical equality. The location of the boundaries of single-member districts will always have major implications for representation, even with the most non-partisan system of allocation. Changing the boundaries may not be done for partisan reasons, but it will have partisan consequences. > Let me close with a question: For those who still think single-member > districts have no place in a legislature, what do you propose as a good > model for a bicameral legislature? This is not a subject to which I have given a great deal of thought because it is not at the top of my (UK) political agenda. And you may have noticed the mess we've got ourselves into trying to reform our Upper House (still unelected!!). But the Federal Parliaments of Australia and Germany might be good places to look for a start. James ---- Election-methods mailing list - see http://electorama.com/em for list info