[EM] Re: alternate proportional method

[James Gilmour]

Stephane wrote (in part):
> STV only allows comparison between the candidates of a same 
> party running in the same multi-member constituency. You 
> cannot compare all candidates of the same party because, to 
> my knowledge, all candidates of the same party do not run in 
> the same constituency.

This betrays a misunderstanding of how STV-PR works, and indeed, a
misunderstanding of the whole purpose of STV-PR.  So far as the voting
system is concerned, the parties are totally irrelevant.  It is the
candidates as candidates that matter.  The purpose of an STV-PR election
(no matter how large or small the electoral district = number of members
elected together) is to elect the N candidates who are "most"
representative of those who voted, as expressed through the preferences
marked by those voters.

PR of political parties is obtained only to the extent that the voters
vote the parties' tickets.  When the voters do this, party PR results.
But when the voters are motivated to seek PR of something other than the
parties, party PR does not result.  Thus PR of political parties may be
an outcome of STV-PR, but it is never the objective.

> (I do not think their is a limit for 
> the number of candidates of a party, however it would not be 
> strategically optimal to have more than n+1 candidate I think 
> where n is the number of seats of the constituency, 

There is no reason why there should be a limit on the number of
candidates nominated by any party in one district, but for practicality
it is usually set at the number of seats in the district (e.g. in
Northern Ireland).  I cannot see any circumstances in which it could
possibly be an optimal strategy to nominate as many candidates as there
were seats in the district - no party is going to win all the seats in a
public election.  That would maximise voter choice (which I favour), but
the party would rarely have anything to gain from so indulging its
supporters.  The larger parties, ie those that expect to win two or more
seats in a typical STV-PR district, usually nominate at least one more
candidate than the number of seats they expect to win.  The small
parties that hope to win one seat, but often don't win any, usually
nominate only one candidate.  There is nothing to be gained by their
nominating two or more, and their may be something to loose  -  see
earlier posts on vote averaging and vote management.

> and I 
> think a candidate cannot run in several constituencies at the 
> same election).

This avoids the problem of a candidate being elected in more than one
district.  But electoral law allows one person to be the elected
representative for only one district.  So the problem is obviated at
nomination.

James