[EL] SF excludes 28% of participating voters from runoff

[David A. Holtzman]

Slow down a bit guys.

You're talking about an interesting concept, but let's get the terms right.

"IRV" - instant runoff voting - is a term ordinarily only applied to 
one-round, one-winner elections.The "instant runoff" part implies 
something quick: one election day, instead of more than one election day.

When speaking of election methods, IRV ordinarily means the 
single-winner case of the system of the single transferable vote ("STV").

"RCV" - ranked-choice voting - is a term that does a better job than 
"IRV" of telling voters what they're expected to do in an IRV 
election.In San Francisco, RCV also means the election method, IRV.But 
neither of you is using the term "RCV" below.Thanks for that.

Another term you're not using is "choice voting" - sometimes abbreviated 
"CV" - which ordinarily refers to any of the multiple-winner versions of 
STV.

So when you're talking about using ranked-choice balloting to select two 
nominees (or two "winners" of the first round of an election), you 
really shouldn't be saying "IRV."It's two-winner STV (with ranked-choice 
ballots), or two-winner choice voting.


In a multiple winner STV election, the goal is the same as with IRV: to 
get over a threshold that is an unbeatable fraction (or percentage) of 
the vote.With IRV, the fraction is 1/2, or 50%, because once you've 
gotten over 50% of the vote, no coalition of other candidates can 
combine their support and beat you.In a *two*-winner STV election, the 
fraction is 1/3, or 33.333...%, because once you've got over 1/3 of the 
vote, there's not enough vote support left to put together *two* 
coalitions that can both beat your vote strength, so you're guaranteed 
to get one of the *two* winning spots. (With three winners, the 
unbeatable threshold is 1/4; with four, it's 1/5; and so on.)

In a two (or more)-winner STV election, voters' second choices can count 
in two ways.(By the way, second choices are not "second place" votes - 
but it's an understandable mistake to say so.)As with IRV, if a ballot's 
top choice is a candidate who's been eliminated from an election, the 
ballot is transferred to count for the ballot's next choice.The other 
way a second choice can count is if a candidate gets more votes than 
needed to be a winner (or a nominee, if there will be another round of 
balloting).If a candidate has received more votes than needed to win, 
each ballot can be treated as having two pieces - the fraction of the 
ballot needed to give the candidate a win, and the excess fraction.So 
the excess fraction of each ballot's vote is transferred (yes, as a 
fraction of a vote) from the winning candidate to the next choice marked 
on the ballot.Those vote fractions can add up and push another candidate 
over the winning threshold to victory.


In the Hahn-Huey-Bowen-Winograd-et al. contest, no candidate got more 
than 24.4%, so there would have been no excess votes to transfer in the 
first round of counting if the ballots had been ranked-choice, and the 
primary STV.And I don't think that in any round Hahn's votes would have 
exceeded the winning threshold (33.333...%) by enough for a "rank Huey 
second" strategy to have made much difference.

There are plenty of reasons that a "rank-the-Republican second" strategy 
could have backfired.It could have seemed like a pretty silly 
move:Remember that in a single-winner STV (IRV) election, your voters' 
second choices count for nothing until you're eliminated!It could have 
seemed treasonous: Supporting a Republican, in any way, for any reason, 
would have been a huge turn-off.And what if Hahn needed Bowen's support 
for something later?Or vice-versa?And oh, as it turned out, it wasn't 
exactly a cakewalk for Hahn to beat Huey, the Republican.Hahn only got 
54.7% in the one-on-one runoff.

- dah






On 11/13/2011 9:01 AM, Larry Levine wrote:
>
> Interesting concept. I'll ponder. Gut level, however, I'm uneasy with 
> the notion of people voting for multiple candidates. I think it has 
> the potential to distort the results. In the example you cite wouldn't 
> it have been to the advantage of both Hahn and Bowen to urge their 
> supporters to cast their second place vote for the Republican to 
> ensure a weaker run off opponent. The top two primary next June would 
> hold out the same potential if IRV were part of the equation. Take the 
> Berman-Sherman situation. Wouldn't each of them rather run against a 
> Republican in November than have to run against each other again. On 
> the other hand, it seems to make more sense than elimination of the 
> run off.
>
> Larry
>
> *From:*Thomas J. Cares [mailto:Tom at tomcares.com]
> *Sent:* Sunday, November 13, 2011 7:15 AM
> *To:* Larry Levine
> *Cc:* Dan Johnson; Douglas Johnson; law-election at uci.edu
> *Subject:* Re: [EL] SF excludes 28% of participating voters from runoff
>
> Larry, if that's how you feel, would you at least concede that it 
> would be an improvement on traditional run-offs, to use IRV to 
> determine the top 2 candidates, and then have them go head-to-head? 
> (Assume it'd be possible to rank unlimited candidates like in 
> Portland, ME)
>
> Wouldn't it be nice to know for sure that CA-36 voters really 
> preferred Hahn over Bowen, and that the candidacy of Winograd and 
> others didn't distort the outcome (or to have gotten the 
> more-legitimate result, if it did distort the outcome)?
>
> You can almost look at this algebraically.
>
> *Let a =* a traditional top two run-off
>
> *Let b =* using IRV (with unlimited rankings allowed) to determine the 
> top two candidates, and then doing a head-to-head run-off
>
> *Let c =* just having one IRV election (let's just say with unlimited 
> allowed rankings, to try to put that issue aside)
>
> *Let d = *the percentage of times 'b' would produce a different winner 
> than 'c'
>
> *Let e = *the average value of the change in outcome when 'b' would 
> result in a different outcome than 'c'
>
> *Let f =* the burden on voters and government to carry out an 
> additional election
>
> Your email advocates 'a', but clearly, isn't b>a?
>
> Then, isn't c>b+d*e-f?
>
> Some might take it further and let g = kemeny-young condorcet and say g>c
>
> My view is that Condorcet gets you the least-objectionable candidate; 
> plurality (and top-two runoffs with lots of initial candidates, which 
> is fairly similar to plurality) gets you the candidate who appeals to 
> the largest niche, and IRV strikes the appropriate balance (I'd go so 
> far to say that it could be the corner-stone of /really good government/).
>
> I say the right answer is *C*.
>
> I think some people unjustifiably prefer 'a' over 'b' or 'c' because 
> they're used to campaigns that are focused on appealing to the largest 
> niche. I think Don Perata would fall into this category. My guess is 
> that Perata would have liked a traditional runoff where he would have 
> had a large plurality margin in the first election and then hoped that 
> this would depress the turnout of his opponents in the run-off, so 
> that even while more people voting in the first election, as the RCV 
> election showed us, would have preferred Quan over Perata than 
> vice-versa, the benefit of depressed turnout in the run-off may have 
> enabled him to win. The only way I think you could say a>b and a>c is 
> if you think it's /a good thing/ to preserve that kind of path to victory.
>
> Perhaps there's one more reason - to preserve a special ability for 
> insiders to try to determine election outcomes by deciding early 
> on whose campaigns should and shouldn't be embraced, hoping to 
> manipulate a voronoi diagram of candidates' ideological coordinates, 
> where only certain candidates (or 1 machine-chosen candidate) are left 
> with a chance of getting the largest area.
>
> (To be fair, alternatively, one might just think that 'b' and/or 'c' 
> are negligible improvements that aren't worth implementing, but I 
> think RCV outcomes are demonstrating that that's not true, and that 
> the improvement is significant).
>
> Thomas Cares
>
> Tom at TomCares.com <mailto:Tom at TomCares.com>
>
> On Fri, Nov 11, 2011 at 2:32 PM, Larry Levine 
> <larrylevine at earthlink.net <mailto:larrylevine at earthlink.net>> wrote:
>
> Yep. That would work. Until it didn't. What's wrong with a runoff between
> the two top finishers. Let them discuss and debate the issues and let 
> those
> voters who wish to participate pick the one for whom they want to 
> vote. Oh,
> yeah, that's not reform.
> Larry
>
> -----Original Message-----
> From: law-election-bounces at department-lists.uci.edu 
> <mailto:law-election-bounces at department-lists.uci.edu>
> [mailto:law-election-bounces at department-lists.uci.edu 
> <mailto:law-election-bounces at department-lists.uci.edu>] On Behalf Of Dan
> Johnson
> Sent: Friday, November 11, 2011 9:51 AM
> To: Douglas Johnson
> Cc: law-election at uci.edu <mailto:law-election at uci.edu>
>
> Subject: Re: [EL] SF excludes 28% of participating voters from runoff
>
> That would suggest the proper response is to drop the limit of three
> rankings from the SF ballot. Then the number of exhausted ballots would
> fall.
>
> And then again, some voters really didn't have any preference between the
> two of them and prefer to exhaust their ballot.
>
> On Fri, Nov 11, 2011 at 11:46 AM, Douglas Johnson 
> <djohnson at ndcresearch.com <mailto:djohnson at ndcresearch.com>>
> wrote:
> > A different view on whether the of ranked-choice voting in San
> > Francisco was
> > "effective":
> >
> >
> >
> > According to the November 10 numbers from the Department of Elections,
> > the final round tally in the San Francisco Mayoral election was 79,147
> > votes for Ed Lee, 51,788 for John Avalos, and 48,983 "exhausted" 
> ballots.
> "Exhausted"
> > means the ballot did not contain a vote for either Lee or Avalos, thus
> > the voter was excluded from sharing his/her preference in the final
> runoff.
> >
> >
> >
> > Percentage-wise, Ed Lee won the vote of 43.4% of voters participating
> > in the Mayoral election. John Avalos received the final vote of 28.4%
> > of voters participating in the election. And 28.2% of voters casting
> > ballots in the Mayoral primary were blocked from expressing their
> > preference in the final runoff (26.9% were exhausted and 1.3% were
> over/under votes).
> >
> >
> >
> > In fact, less than half of those not voting for Lee or Avalos in the
> > first round listed either of them as their #2 or #3 choices. In the
> > first round,
> > 89,681 voters cast ballots for Lee and Avalos, while 90,431 voters
> > preferred other candidates as their first choice. As those other
> > candidates were eliminated, 41,254 additional votes were added to Lee
> > and/or Avalos. But
> > 48,983 ballots were "exhausted" and dropped from the counts.
> >
> >
> >
> > By a 48,983 to 41,254 margin, San Francisco's ranked-choice runoff
> > system excluded the views of more participating voters than it added.
> >
> >
> >
> > No system is perfect: without any runoff, Lee would have won 31% to
> > 19%, with 50% of the voters participating not casting a vote for
> > either of the top two. With a traditional runoff, the lower turnout
> > that sometimes occurs would also mean some of the primary voters would
> > not cast ballots in the runoff, though I would argue that is different
> > because that would be by their choice, not by the design of the
> > election system (and note that in some local CA elections, runoff
> > turnout is higher than primary turnout). In SF, it is the election
> > system that dictates the exclusion of some voters from the final 
> decision
> whenever the counting goes more than three rounds.
> > [I should acknowledge what's surely going through Larry Levine's mind
> > right
> > now: the election system in place influences campaign decisions, so
> > this paragraph's comparisons to alternative systems are imperfect
> > because candidates made decisions knowing they were in a RCV system.]
> >
> >
> >
> > Amidst the cheerleading for ranked-choice voting, I believe it is
> > important to remember that the RCV system has substantial drawbacks
> > too. I welcome the discussion of whether the drawbacks of RCV are less
> > than the drawbacks of traditional no-runoff or later-runoff elections,
> > but I would encourage all debaters to acknowledge that RCV is also far
> from perfect.
> >
> >
> >
> > - Doug
> >
> >
> >
> > Douglas Johnson
> >
> > Fellow
> >
> > Rose Institute of State and Local Government
> >
> > m 310-200-2058 <tel:310-200-2058>
> >
> > o 909-621-8159 <tel:909-621-8159>
> >
> > douglas.johnson at cmc.edu <mailto:douglas.johnson at cmc.edu>
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> > From: law-election-bounces at department-lists.uci.edu 
> <mailto:law-election-bounces at department-lists.uci.edu>
> > [mailto:law-election-bounces at department-lists.uci.edu 
> <mailto:law-election-bounces at department-lists.uci.edu>] On Behalf Of
> > Rick Hasen
> > Sent: Friday, November 11, 2011 8:52 AM
> > To: law-election at UCI.EDU <mailto:law-election at UCI.EDU>
> > Subject: [EL] ELB News and Commentary 11/11/11
> >
> > "San Francisco Voters Effectively Used Rank Choice Voting"
> >
> > Posted on November 11, 2011 9:33 am by Rick Hasen
> >
> > FairVote has issued this press release.
> >
> > _______________________________________________
> > Law-election mailing list
> > Law-election at department-lists.uci.edu 
> <mailto:Law-election at department-lists.uci.edu>
> > http://department-lists.uci.edu/mailman/listinfo/law-election
> >
>
>
>
> --
> Dan Johnson
>
> Attorney at Law
> 111 West Washington, Suite 1920
> Chicago, Illinois 60602
>
> 312.867.5377 <tel:312.867.5377> (office)
> 312.933.4890 <tel:312.933.4890> (mobile)
> 312.794.7064 <tel:312.794.7064> (fax)
> _______________________________________________
> Law-election mailing list
> Law-election at department-lists.uci.edu 
> <mailto:Law-election at department-lists.uci.edu>
> http://department-lists.uci.edu/mailman/listinfo/law-election
>
> _______________________________________________
> Law-election mailing list
> Law-election at department-lists.uci.edu 
> <mailto:Law-election at department-lists.uci.edu>
> http://department-lists.uci.edu/mailman/listinfo/law-election
>
>
>
> _______________________________________________
> Law-election mailing list
> Law-election at department-lists.uci.edu
> http://department-lists.uci.edu/mailman/listinfo/law-election
>
> -- 
> David A. Holtzman, M.P.H., J.D.
> david at holtzmanlaw.com
>
> Notice: This email (including any files transmitted with it) may be 
> confidential, for use only by intended recipients.  If you are not an 
> intended recipient or a person responsible for delivering this email 
> to an intended recipient, be advised that you have received this email 
> in error and that any use, dissemination, forwarding, printing or 
> copying of this email is strictly prohibited.  If you have received 
> this email in error, please immediately notify the sender and discard 
> all copies.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://webshare.law.ucla.edu/Listservs/law-election/attachments/20111113/3fcf700b/attachment.html>