[EM] True Ranked Choice - for Condorcet

Mon Aug 23 17:17:58 PDT 2010

Dave,

I like all your ideas here.  It is amazing that people continue to
misinform by making the false claim that IRV finds majority winners
and solves the spoiler problem, when IRV does neither, and in fact
fails more of Arrow's fairness criteria than plurality voting by
exhibiting nonmonotonicity, eviscerating election transparency in the
complex centralized counting process.

I like your approach for selling Condorcet as "True Rank Choice" as a
method that fairly counts all the choices of all voters, rather than
the wholly inequitable method of counting only some voters' 2nd, 3rd
choices and not others.  The nxn matrix makes Condorcet simple to
count, precinct-summable, and more easily auditable (although I'm
uncertain exactly how one does risk-limiting audits of it yet.)

I agree that it would be great if election methods experts could agree
on a best algorithm for resolving any of the unlikely Condorcet
cycles.

Kathy

> Date: Mon, 23 Aug 2010 13:53:19 -0400
> From: Dave Ketchum <davek at clarityconnect.com>
> To: electionsciencefoundation Foundation
>        <electionsciencefoundation at googlegroups.com>,   EM
>        <election-methods at lists.electorama.com>
> Cc: leeswalker <lee at rankedvoting.us>, RangeVoting at yahoogroups.com
>
> I see below that leeswalker is doing his best for IRV.
>
> Would be useful if some of us could do better for Condorcet - which I
> see as a competitor that should win.
>
> TRC - True Ranked Choice  - my thought for a possible label for
> Condorcet, based on:
>      Like IRV, let's voters rank their choices of candidates.
>      UNLIKE IRV, counters READ all that the voters vote.
>
>> IRV is good for letting voters state their preferences, has
>> determined backers, and is getting adopted some places.
>>
>> IRV is also getting rejected as experience shows its weaknesses.
>> While IRV voters can express their desires, vote counters are not
>> required (in fact not even permitted) to count all that they have
>> voted.  Steps in normal rounds - some may do variations:
>> 1. Count the top rank from each ballot.
>> 2. If this indicates a winner, counting is done.
>> 3. For candidate with fewest top rank votes, remove those, exposing
>> next ranks.
>> 4. Ballots with no remaining ranks are done.
>> 5. Back to step 1 with remaining ballots.
>
> Note that, of the ballots surviving to the last round, the counters
> have seen only down to the remaining top rank and know nothing of
> whatever the voters ranked below these.
>>
>> For an extreme example with IRV - easy to get near enough to this to
>> be in trouble:
>>      Suppose EVERYONE likes Joe and ranks him on their ballot.
>>      But some like Tom better and rank him higher; ditto for Dick;
>> ditto for Harry.  If any rank Joe as top, they are so few that Joe
>> loses with least top rank votes.
>>
>> Condorcet counting can read ballots voted for IRV.  Condorcet also
>> permits assigning equal ranks to multiple candidates.  Here counting
>> is for each candidate ranked on a ballot - except for candidates
>> ranked as high or higher by this voter, this candidate gets counted
>> as winning against all others.
>      Minimizes ranking failures - permitting equality and not
> requiring that every rank number get used simplifies for all.
>      Precinct summability both simplifies error checking and speeds
> up having usable counts - do not have IRVs problems with counting for
> a few delayed for any reason and when counted changing order of rounds).
>
>>      Usually the counts will show one winning against all others.
>> When a few win over the remainder, but not over each other,
>> resolution is doable but takes more effort among the few.
>      We NEED to agree on simple, explainable, cycle resolution - plus
> noting that this can be changed if/when there is agreement as to what
> is better - it should use the N*N array for data.
>
> Note that TRC voters can combine two goals, with truly minimum conflict:
>      Rank high their preferences among the likely winners.
>      Rank high their preferences among all candidates.  If these
> include possible winners - fine.  Even if not, these will still get
> included in the N*N array that shows the votes between EVERY pair of
> candidates.
>      Vote ala Plurality - fine if this expresses a voter's desires -
> such as voting for one of likely winners.
>      Vote ala Approval - simply vote all at same rank - TRC will see
> this as among the normal rankings.
>
> Note that the N*N array deserves more bragging than it sometimes
> gets.  It shows  all how they are progressing and thus if they may
> need to compete better.
>      Further, counting all should encourage more to vote and thus get
> visibly counted.
>>
>> Dave Ketchum
>>
>
> On Aug 21, 2010, at 2:58 PM, leeswalker wrote to:
> http://sfbay.craigslist.org/forums/?forumID=4123
> Re: [InstantRunoffNYS] Greatest Majority Voting -the purpose of IRV
>
>> For over a decade, election reform advocates have struggled for a
>> better term for ranked/IRV election reform, but none of these terms
>> has been well received:
>> Preference, Choice, Alternative Vote, Ranked, Ranked-Choice, Instant
>> Runoff, and Inclusive Ranked Voting--all such names cause confusion
>> instead of reassuring the newbies.
>>
>>  I think that Greatest Majority Voting is our white knight.  The
>> effect of IRV is to ensure that the opinion of the greatest majority
>> of voters will be heeded.  The term can be all inclusive of some
>> minor reforms also such as redistricting and multi-day voting, or
>> making election day a holiday for most workers.
>>
>>  Also, IRV supporters control most of the appropriate urls.
>>
>>  This election season is high-time to establisg GMV as inclusive
>> term for IRV and Alternative Vote.  I would like previously elected
>> politicians to get published ASAP explaining that it is now time for
>> GMV, after Alternative Vote and IRV are spreading so much this year.
>> _________
>> forum http://4gmv.org
>>
>>   Enough urls have been/will be reserved to allow various IRV groups
>> to have their own free GMV url.  I hope that Fairvote,
>> InstantRunoff.com and  more IRV-supporting groups will establish GMV-
>> friendly webpages asap so that one or more GMV-related urls can be
>> pointed to that page immediately, and some of the url ownership
>> rights can be transferred also.
>> -lee

-- 

Kathy Dopp
http://electionmathematics.org
Town of Colonie, NY 12304
"One of the best ways to keep any conversation civil is to support the
discussion with true facts."

Realities Mar Instant Runoff Voting
http://electionmathematics.org/ucvAnalysis/US/RCV-IRV/InstantRunoffVotingFlaws.pdf

View some of my research on my SSRN Author page:
http://ssrn.com/author=1451051