[EM] Competitive Districting Rule

[Juho]

Thanks for explaining what the situation in Scotland is. It really seems that Scotts have been working against the Duverge's law for some time now.

The MMP method of the Scottish Parliament certainly turns the Scottish party system in the direction of a multi-party system, but I guess the tendency to move in the multi-party direction existed already before the MMP started influencing.

But whatever the timing, at least now there seems to be a clearly identifiable conflict between two-party (Westminster Parliament) and multi-party (Scottish Parliament) political arrangements. Having a two-party EM in use at the same time with a multi-party style MMP EM is a conflict situation. It is hard to make the party system work sometimes as if it would consist of two parties and sometimes as in a multi-party configuration.

You have a campaign to reform the Scottish Parliament EM. Do you have plans to do something with the Wetminster Parliament EM too (i.e. the Scottish branch of it)? It is ok to improve one thing at a time, but on the long run I think some improvements would be good to alleviate the problems related to hanging between two and multiple parties.


> STV-PR is probably the best way of allowing voters to express their sincere preferences.

STV-PR allows the voters to express their opinions with extreme flexibility. There are also interesting party based and hybrid methods that are maybe less flexible but interesting and expressive. Btw, when talking about improved party based methods I typically think about open lists, not about the closed ones. And STV-PR may well be good for Scotland although I have interest in improving other methods too. :-)


> This sounds very like MMP, which we use to elect the Scottish Parliament

The method I described was at least close to a MMP method (as defined in wikipedia).
- it is a MMP method that fully compensates/balances the results so that full PR is achieved
- it is a MMP method where no additional seats are allocated to balance the results => single member results are (re)calculated (= some plurality winners changed to others) so that full PR will be achieved
- it differs at least (from the wikipedia definition) of MMP in that no separate party votes are given (each candidate of a single member district is assumed to belong to some party/list and PR is derived from this)

Maybe I should write here a concrete and very simple version of the type of EMs I had in mind.
1) Every candidate belongs to a party; we use single member districts (one representative elected from each, no others); voters give one vote to one candidate in their district
2) Count the votes of each party (in all districts) and decide the number of seats each party will get (using one of the well known algorithms)
3) Find the strongest plurality winner among all the districts that have not elected their representative yet
4a) Elect this candidate if his/her party doesn't yet have as many seats as decided at point 2
4b) Otherwise eliminate all remaining candidates of this party from the election
5) Return to point 3 (unless all representatives have already been elected)

- let's assume that all parties had candidates in every district
- I think this EM has single member districts and PR. You commented such methods in your reply to Claes Wallin. Please check how this one meets the definition.


Maybe I should also give one example of an improved party based EM to clarify what I meant with those.

It is possible to allow parties to create (hierarchical) alliances and to let them have internal (hierarchical) groupings. In such a situation my vote to candidate C would go primarily to her/him, and alternatively to her group, then to one step bigger group, then to her party, then to some alliance of parties etc. The highest level alliances will get as many seats as the number of votes in their "subtree" gives them. These seats will then be allocated to the next lower level proportionally etc. until individual representatives are found. (and let's assume also here that we did not run out of candidates at any branch during the process :-)

This actually resembles in some sense the free transferrable vote in STV, but now the order of inheritance of the vote has been set by the candidates, not by the voters. I think both approaches have their good points. Some interesting points in this method are: simple voting (just one candidate named); clear indication and commitment by candidate C to the policy/branch he/she names; voters need not study the policies of large number of different candidates themselves; ability of the voters to influence the internal balance and evolution of the policy of the parties (i.e. voters decide who and what themes/branches are strong within a party instead of the "central clique of the party" doing so in some hidden way); simple mathematics.

BR, Juho


_____ Original message _____
Subject:	Re: [EM] Competitive Districting Rule
Author:	"James Gilmour" <jgilmour at globalnet.co.uk>
Date:		15th July 2006 1:16:4 

Juho Sent: Friday, July 14, 2006 10:22 PM
> The Scottish situation sounds to me like a multi-party system 
> (that has emerged under different rules) has gotten trapped 
> in a two-party EM, and this kind of mixture is not a pretty 
> match (looks actually quite terrible).

No, not at all.  For UK (Westminster) parliament elections the whole of the UK has used only FPTP (simple plurality) in
single-member districts for many decades, in most case for more than 100 years.  We all also used the same voting system
for local government elections (the only other public elections we had until comparatively recently).  In the elections
after the 1939-45 war, the two main parties (Conservative and Labour) got around 90% of the total vote (96.8% in 1951).
The third party was the Liberals (now Liberal Democrats) who, in elections after 1945, got around 9% of the vote, but
only 1% of the seats.  Despite Duverger's "law", support for the Liberal Democrats has grown and in 2005 they had 25% of
the vote and 9.6% of the seats.  Eight smaller parties also have seats in the UK Parliament.  So there has been a
three-party system across the whole of the UK for some time.

In Scotland the SNP (Scottish National Party - campaigning for independence) became a significant force in 1970, gaining
11% of the vote, again despite Duverger's "law".  In the October 1974 election the SNP peaked at 30.5% of the vote, and
they now get around 20% of the vote (but far fewer seats).  So In Scotland we have had a four-party system for the past
30 years, all based on single-member districts and the simple plurality (FPTP) voting system.


> I guess the other three parties were fed up with 
> conservatives taking too many seats (more than PR would 
> allow?) and decided to join forces.

No, the Conservatives were UNDER-represented in Scotland.  In the 1992 election they got 26% of the votes but only 15%
of the seats.  But the Conservatives had formed the UK Government (the only government we then had) since 1979, a period
of 18 years when the 1997 election was called.  The Conservative Government was increasingly disliked (!!) in Scotland.


> The others obviously got 
> their revenge and now took more seats than PR would allow :-).

No, this was NOT a move by the other parties.  The "Scotland Tory-free" campaign was a grass-roots campaign among
electors.  There were some websites exchanging information among electors, but the parties all stayed very quiet on the
subject.


> I tend to think that if voters are clever enough to make the 
> 1997 tactic work, they would also be able to use full PR 
> right if they would be given the chance.

Yes indeed!  The Scottish Parliament is elected by MMP and our electors have (mostly) shown great ability to manipulate
that voting system (so far).


> A good voting system 
> is anyway such where one can vote based on one's sincere 
> preferences. And I do believe better EMs exist than the poor 
> multi/two-party combination of 1997.

STV-PR is probably the best way of allowing voters to express their sincere preferences.


> On STV:
> STV gives at least the option of giving first position to a 
> local candidate and second to some non-local alternative.

I don't understand this comment.  With STV-PR as I know it, ALL the candidates are local, ie all within one
locality-based multi-member district.  If any party is so stupid as to include some carpet-baggers in its team of
candidates, the voters will deal with them very severely.  When you have community-based multi-member districts for
STV-PR, the term "local" takes on a new meaning, especially in rural areas where there are likely to be several
significant clusters of population.  Then you want to make sure your team of candidates do not all live and work in the
same part of the district.  This applies even in densely populated cities.


> This scenario has however the problem that typically voters 
> tend to vote well known candidates, and they often come e.g. 
> from the capital, not from the local community. I guess this 
> is one reason why EMs often force people to vote only the 
> local candidates.

STV-PR encourages the electors to learn more about more candidates, but they can vote however they want, eg by party, by
locality, by women before men, by ethnic community, by some pressing local issue, by any combination of these, etc.  I
may not think much of the criteria you used to select your preferences, but they are your criteria and in a democracy,
your criteria are as valid as mine or anyone else's.

I know that by promoting STV-PR I shall help to secure the election of some candidates whose political views I oppose,
and I know that I shall be helping to empower some voters who, in my view, make very poor choices based on very poor
criteria.  But that's democracy and I do believe in democracy.


> STV style has the problem that voters need to have lots of 
> information about the candidates. Or maybe they'll just vote 
> the most famous ones. My ideal ballot (from the completeness 
> point of view, maybe not if need to avoid complexity is 
> there) would be one with both STV style ability to set voter 
> specific preferences but also party based groupings 
> (including subdivisions, areal criteria and party alliances). 
> I might e.g. vote "James > local conservatives > other conservatives".
> 
> (We could also force regional balance (regional PR) even if 
> people would not vote "regionally enough" in the same way the 
> method I describes forced changes in the "single winners" of 
> the districts to gaon national ideological PR.)
> 
> Note that subdivision of parties and their alliances and 
> whatever other groupings add tools to the voter to express 
> what she wants. Also models where STV like ordering is not 
> used but the vote to James automatically goes to the smallest 
> group that James belongs to, then to the next bigger group 
> etc. may work better than current more rough "vote party 
> only" or "vote party member only" arrangements.

This all sounds very like the "above-the-line" voting that is used in the Australian Federal Senate elections.  It has
perverted STV-PR very severely, so that that implementation is really nothing more than closed list party PR.


> In the party subdivision EMs I see potential for making the 
> votes influence the party opinions more than what the case 
> often is today (i.e. when parties often seem to be deaf and 
> push the party office line instead of the supporters' line.

That's why I favour STV-PR over party list PR  voting systems.  STV-PR puts the power in the hands of the local voters
and so makes the elected representatives much more accountable to those voters.


> I still have to say that the simple method that I described 
> (single member districts + national PR) would fulfil some of 
> the key targets you maybe had (PR, locality in the UK/US 
> traditional style, more sincere votes), although it had also 
> some "rounding errors" like giving victory in some districts 
> to other than the plurality winner.

This sounds very like MMP, which we use to elect the Scottish Parliament (implemented by region, not nationally).  It is
better than FPTP in single-member districts, but it is a very poor voting system compared to STV-PR.  For a full
presentation of the arguments in relation to the Scottish Parliament, see:
http://www.arbuthnottcommission.gov.uk/docs/Consultation%20Responses/Civic%20Organisations%20and%20Bodies/Fairshare.pdf

James

----
election-methods mailing list - see http://electorama.com/em for list info


		
___________________________________________________________ 
To help you stay safe and secure online, we've developed the all new Yahoo! Security Centre. http://uk.security.yahoo.com