[EM] STV can be hand counted...

[Steve Eppley]

Hi,

Stephane Rouillon a écrit :
> James Gilmour a écrit :
-snip-

> > Computerisation, with electronic voting, may be the way 
> > forward, but there are still some problems to be overcome. 
> > But these problems relate to the voting method (electronic
> > without an audit trail) and NOT to the voting system, 
> > ie STV-PR.

Except that if the election will be counted by hand, then the 
choice of variation of STV-PR is constrained: It must forbid 
each voter from ranking candidates as equally preferred (except 
at the bottom "truncated" part of her ordering).  Trying to 
handle indifference in a hand-count would be impractical.

(Similarly, the variation of STV-PR that "fractionally 
transfers" each over-quota vote from winners would be 
impractical if the election were hand-counted.)

Banning expressions of equally-preferred candidates in STV-PR 
isn't a serious problem, per se.  But in Instant Runoff it can 
make a big difference, by forcing voters to rank a compromise 
over (rather than equal to) their favorite candidates in order 
to defeat a "greater evil."  Since some of the people advocating 
Instant Runoff seem actually to be just trying to use it as a 
stepping stone to PR, they insist Instant Runoff be just like 
STV-PR, meaning they won't advocate the variation of Instant 
Runoff that allows the voter to express equal preference.  To 
get them to change their mind on this would probably require 
some country to adopt a variation of STV-PR that allows equal 
preference.  Which requires machine-counting.

By the way, according to Gary Cox (professor of political 
science at UC San Diego), Australia's system is more like closed 
party list than STV PR, since most voters utilize the option of 
picking a party rather than tediously ranking many candidates. 
(Gary prefers closed party list because he says it reduces the 
incentives for similar candidates to compete against each other 
by promising pork to their constituency.  Whether or not I'd 
agree with him that closed party list is better depends on 
whether there are significant barriers against new parties being 
able to compete successfully, if the elite in an old party go 
corrupt.)

A PR variation that I think might work better than existing PRs 
would be to let each voter rank the *parties* in order of 
preference.  These votes could be counted similarly to closed 
party list, by awarding seats to each party in proportion to the 
number of votes that rank the party highest.  But it has a few 
advantages: (1) If there's a minimum threshold (5%, for example) 
that a party must meet to win any seats, voters who prefer tiny 
parties could rank them highest without wasting their votes, 
since sub-threshold parties could be deleted one at a time, 
smallest-first, with votes transferring to next choices as in 
STV.  (2) The preference orders could simultaneously be tallied 
by a good Condorcetian method (e.g., MAM) to find the "most 
preferred" party, which could be rewarded with extra seats, 
and/or agenda control in the legislature, and/or major offices 
like prime minister.  This second tally would require machine-
counting to be practical, since if the rewards given to the 
"most preferred" party are significant, then many small centrist 
parties would compete to be that party.  (3) Assuming the second 
advantage is implemented, with a large reward going to the "most 
preferred" party, the voters' relative preferences regarding the 
plausible compromises on important issues that lay ahead would 
be elicited.

---Steve     (Steve Eppley    seppley at alumni.caltech.edu)