[EM] Competitive Districting Rule

[James Gilmour]

> > > > James Gilmour wrote:
> > > > Of course, you cannot have single-member districts and PR, ... ...
Juho wrote:
> > > I think there are methods that allow even this. It is 
> > > possible for example to first count nation wide the votes of 
> > > each party and decide the number of seats each party will get 
> > > based on the number of votes they got.
James Gilmour wrote:
> > This approach is fundamentally flawed because the nation-wide
> > totals of party votes are distorted by local tactical
> > voting in the single-member district contests.
> > There are no clever algorithms available that could possibly "correct"
> > for this inherent defect.  And of course, the PR you could obtain
> > by such a (flawed) voting system would only be PR of
> > registered political parties, with all that that implies.  Some of us want
> > PR of the voters  -  we do not want to give even more power to the parties.

Juho asked:
> Could you explain a bit more what you mean by "flawed" and
> by "correcting the defect". My first guess is that you refer
> e.g.  to the tendency of people (at least in current two-party countries)
> to vote on the two major parties and not to vote e.g. a smaller "Green" party.

> In two-party countries this could (at least initially) be the case.
> Note however that in countries with multi-party tradition
> the viewpoint of the voters might be different.

No, I the distortion I specifically had in mind occurs particularly when you have at least three parties.  We have four
"large" parties here in Scotland that have a serious chance of winning FPTP single-member district elections for the UK
(Westminster) Parliament.  The are also four "large" parties in Wales. In England (our large neighbour, which elects
most of the MPs at Westminster)  there are three "large" parties.  In these situations it is quite common for electors
not to vote for the candidate of the party they support, but to vote for another party, either to defeat an MP they want
rid of or to ensure that a candidate they don't want isn't elected.

So in 1997 in Scotland, significant numbers of SNP supporters voted either Labour or Liberal Democrat, significant
numbers of Labour supporters voted either SNP or Liberal Democrat, and significant numbers of Liberal Democrat
supporters voted either Labour or SNP.  Why?  To unseat sitting Conservative MPs and to make sure that no previously
second-placed Conservative candidates were elected.  This campaign for local tactical voting was called "Make Scotland
Tory free!", and it worked extremely well.  The Conservatives did not elect a single MP from Scotland in the 1997
election, even though they got more votes (17.5%) than the Liberal Democrats (13.0%) who elected 10 out of 72 MPs.  A
similar campaign in Wales also prevented the Conservatives from electing any MPs for constituencies in Wales in 1997.
(Both of these results are damning indictments of the highly defective voting system we use to elect the UK Parliament,
but that's another story.)

Where such tactical voting occurs, and electors deliberately vote for parties they do not really support, it would be
seriously misleading to use the national totals of the parties' votes to determine overall party representation.  You
could do it, but it would not be any kind of fair reflection of the real party PR.

> If your term "correct" referred to guessing afterwards
> what people would have voted even if they didn't, that
> surely would be quite impractical. I didn't assume that.
> Rather I expected the voting situation and attitude of voters
> to provide (sufficiently) sincere votes in the first place.

But in the actual situations we have seen at successive elections, significant numbers of electors have had very strong
incentives not to vote sincerely.  The numbers of voters who do this vary from election to election, and from place to
place around the country, depending on local circumstances and the practical opportunity for effective tactical voting.

> The question if political parties are a better way to implement
>  ideological PR than STV style is interesting. I'd just say that
> both are usable. And there are also other ways than STV to 
> reduce the "party power" if one so wants (e.g. opening up
> subdivisions within a party).

Of course both party PR and STV-PR are usable, because we can see practical examples from different countries around the
world.  Political parties are essential for effective representative democracy, but I want to see PR of voters (whatever
they want, as expressed by their preferences for all the candidates), not just PR of registered political parties.
Also, I think STV-PR provides a mechanism that allows an automatic balance to be struck between local representation and
national PR  -  something that is very important for voting reform campaigns in countries (like the UK and USA) where
representation has been locality-based for more than two centuries.

James Gilmour