[EM] Preferential voting system where a candidate may win multiple seats

[Juho Laatu]

Some random notes. Please treat them as such. Just trying to point out what PAL representation looks like from different angles.

I guess the key feature of PAL representation is the dynamic size of the districts. In this thread one central theme has been practical reforms in the Norwegian (or similar) system. Therefore of course one starts from districts that are similar to the current ones. But I think it could be possible that also a working and very accurately proportional multiparty system would use PAL representation like dynamic districts one day. (For current single member district based systems the promise of PAL representation is of course quite different, and the reasons to support or oppose it are quite different.)

I note that the linked PAL representation article recommends use of super-districts. This means that voters would be (almost) forced to vote for the candidates of their own region (as in typical multiwinner systems force voters to do today). There was also the write-in option (to support the traditions of the USA I guess). If such write-in votes are rare, then they have also only small impact on the results. If they are common, then candidates of large centres would probably get more votes, and that would reduce regional proportionality somewhat.

If some party gets only one seat (nationally), then I wonder if that seat would go to the biggest super-district. Is that how the system works? I guess the candidates of the biggest super-district typically delegate votes to each others. I wonder if they are allowed to delegate to others too (which might mean delegation to some central places, central figures, and that migh distort regional proportionality to some extent). I also note that the negotiation process of candidate vote delegation might be a complex one, possibly even involving money and party coercion.

In large parties that will get several seats in each super-district this system could lead to more accurate regional representation than in traditional multimember district based systems. The problem of voters voting for the most central candidates of their (super-)district however remains (=> makes the regional representation less proportional). Same with candidate delegation.

One alternative to this kind of dynamic district oriented approach to providing good regional proportionality are e.g. ranked votes that allow voters to support all the candidates of their own sub-district (if they so wish), but just let the representatives represent whatever they do represent (region or ideology). I note that one key idea behind the PAL representation approach to districts is to follow the traditional single member district idea as closely as possible (maybe partly for marketability reasons, partly to guarantee good regional proportionality). This may be important in current single member district countries. When one looks at the system from current multmember district country point of view, the super-districts seem to correspond to the current districts. Then the question is, how much more will the dynamic districts offer when compared to traditional bullet votes to a party or to a candidate withins a (super-)district, or to a ranked vote based system (that allows voters to vote in terms of smaller sub-districts, but does not force them to do so (like super-districts do not force voters to vote locally)).

One more topic in my mind is the distribution of candidates to the atomic districts. I guess in PAL representation it is possible that representatives of all (dynamic) districts that cover one particular atomic district may come from that single atomic district. Each atomic district thus gets its own nominated representative (for each party) but it is possible that the geographical distribution of the candidates is not very balanced.

I guess that's enough (maybe even too much :) for now.

Juho


On 20.7.2013, at 17.40, Jameson Quinn wrote:

> I have kept up with this thread only intermittently. It seems to have strayed significantly far away from its subject line, and while I've been interested in some of the points that have been made, it's hard to summarize the thread as a whole.
> 
> There is one point I've wanted to make, which seems a bit off-topic, but no more so than the rest of the thread. That is that a least remainders approach is not the only way to get something "biproportional". You can also approach that ideal through delegation. Asset, for instance, is arguably perfectly proportional in all salient dimensions. And PAL representation is a biproportional system that works with a simple vote-for-one ballot. I encourage the people to follow that link because I think that the ideas in that system might enrich the conversation in this thread.
> 
> Jameson

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20130721/f6eb394c/attachment-0004.htm>