[EM] Does IRV elect "majority winners?"

Terry Bouricius terryb at burlingtontelecom.net
Tue Dec 30 15:59:09 PST 2008

I take offense at Abd repeatedly suggesting I am a liar or am engaging in 
deception. We have a legitimate difference of opinion about the 
appropriate use of the term "majority" and interpretation of RRONR.

At the outset, we might all agree that no system can really assure a 
_true_ "majority winner" in an ultimate sense, since there may be a tie, 
or there may simply be no candidate running who a majority of voters can 
abide. The core of my argument is that if the winner of a traditional 
two-round runoff system (without write-ins) is appropriately called a 
"majority winner," so can the winner of an instant runoff election.

The term "majority" simply means more than half, and it is regularly 
applied to different denominators...a "majority" of the entire membership, 
"majority" of those present and voting, "majority" in the second round of 
a runoff system, etc. In governmental elections we generally use the 
short-hand "majority" without specifying all of the exclusions from the 
denominator. Abd is insisting that uniquely for IRV elections, we should 
list the exclusions (i.e.. "a majority of unexhausted ballots," or "a 
majority of those who expressed a preference between the final two 
candidates", etc.). It is acceptable to make this detailed explanation, 
but not necessary in normal speech. In a typical U.S. governmental runoff 
election we do not list the exclusions from the denominator when naming a 
majority winner...We do not say "Jane Smith won with a majority, excluding 
those who were eligible but did not register to vote, or registered but 
did not cast a ballot, or cast a ballot but skipped the race, or that 
ballot was blank, or spoiled, or illegal, or contained identifying marks, 
and excluding those who may have participated in the first round of voting 
but not the second." We just say she was the "majority winner."

Abd accepts that the winner of a top-two runoff (TTR) system (without 
write-ins) is appropriately called a "majority winner," but not 
necessarily the winner of an instant runoff. He treats the "majority 
winner" of a two-round runoff as somehow better, or more valid. 
Ironically, it is typical for the winner of an IRV runoff to have won more 
total votes within a given jurisdiction and have a larger majority 
threshold to reach, than a "majority winner" of a two-round runoff, simply 
due to higher voter participation resulting from a single trip to the 
polls (yes, I know this is not an absolute as separate runoffs do on rare 
occasions have higher turnout).  One essential difference between these 
two "majority winners" is simply the duration of time between the 
beginning and ending of he candidate marking process, in that the IRV 
ballot allows the voter to complete the task in a single visit to the 
polls. (Yes, I know voters also get to have a "second-look," etc. but that 
is irrelevant to the "majority" issue here).

A top two runoff system, JUST LIKE IRV, finds a "majority winner" by 
excluding from the denominator any voters who expressed a preference in 
the first round of counting, but whose preferred candidate gets eliminated 
and who express no preference between the two final candidates for the 
final round of counting.

Let me set out some thought experiments (these are silly, perhaps, but are 
presented to illustrate an underlying point.)...

1. First, I still think my interpretation of RRONR's use of the word 
"abstention" is sound (that not indicating any preference between 
finalists can reasonably be deemed an abstention that RRONR says should 
NOT be counted in the denominator). Note that on page 394 in the section 
on the "Right of Abstention" RRONR points out that in an at-large election 
a voter may "partially abstain" by marking fewer candidates than allowed. 
In other words, the voter has participated in the election with some marks 
for some candidates, but may still be said to "abstain" from some aspect 
of that contest. Thus Abd's insistence that under RRONR a ballot with some 
candidates marked cannot also be an abstention is incorrect.

Abd agrees that RRONR says to treat "each segment of a ballot" separately 
in considering blanks and abstentions. I have argued that each possible 
pairwise final runoff combination is functionally the same as a separate 
segment of the ballot.

To explain...imagine if the IRV ballots were inefficiently, but logically, 
divided into a series of questions or segments as follows:
Section A. Rank the candidates in the order you would prefer they be 
included as finalists in a runoff count.
Jane    1  2  3  4
Mary   1  2  3  4
Stan    1  2  3  4
Dave   1  2  3  4

Section B. In each possible final runoff pairing below mark an "x" for the 
one candidate you would prefer.
(1) Jane __  vs. Mary__
(2) Jane __ vs. Stan __
(3) Jane __ vs. Dave__
(4) Mary__ vs. Stan __
(5) Mary__ vs. Dave__
(6) Stan __ vs. Dave__

Now suppose a voter completely marks the ballot except in segment B. (4) 
of this unusual, but functional, IRV ballot where the voter abstains and 
votes for neither Mary nor Stan in that possible match-up. Further suppose 
the elimination process from the first segment of the ballot results in a 
final runoff tally between Mary and Stan. This voter has clearly abstained 
from that particular question or segment of the ballot. This is the same 
as the exhausted ballot situation on a normal ranked ballot, except that 
the series of pairwise questions is extrapolated from the initial 
rankings, rather than asked redundantly.

2. Secondly, I assert that the top-two-runoff generally functions, de 
facto, as a single election system with two stages, and not two completely 
independent elections. I suspect you will argue this point, but I think it 
is a reasonable assessment on my part, since the total purpose is to have 
voters fill an office. Voters may abstain in the first, second, both, or 
neither of the stages. In the first round, let's say all 10,000 eligible 
voters cast valid ballots. In the second round, perhaps 5,000 voters opt 
not to participate, and 5,000 cast valid ballots. the winner of the runoff 
might have 2,501 votes in this round. That is clearly a "majority" of the 
ballots cast in the final round, but clearly NOT a majority among voters 
who cast ballots and participated in the election process to fill the 
seat. Even though the winner might receive fewer votes in the runoff than 
the ultimate loser received in the first round (which can never happen 
with IRV), and a huge majority of all voters who participated in some way 
in this election process may consider the ultimate winner to be the worst 
choice, we generally say the winner is a "majority winner" because we 
exclude the voters who abstained in the final round. The denominator used 
is the number of voters who opted to express their preference between the 
finalists, not the number of voters who voted in the process and may have 
voted for losing candidates at the earlier stage. Abd has expressed 
agreement with this interpretation of a "majority winner" in the second 
stage of a traditional runoff.

Now what if the second stage of balloting followed immediately upon the 
first stage, while an association's members were still assembled? Most 
would agree that those who now don't participate in the runoff stage or 
express any preference between the finalists should not be included in 
calculating the majority threshold. A voter who passed in a ballot for the 
runoff without any marks showing her preference between the two finalists 
would  have her ballot counted as an abstention or blank and it would not 
be included in the denominator. Perhaps each voter could even have their 
original ballot redistributed to them to make their subsequent choice in 
some way -- so it need not be a "separate" ballot paper. Now what if all 
of the voters were voting simultaneously (perhaps on hand-held devices 
linked by wireless RF). As the voter is pressing the first-choice button 
for one candidate the voting device indicates (after reading all voters' 
simultaneous first choices,) that this candidate has just been eliminated. 
The voter proceeds to push a second choice, and the device shows that this 
candidate has also just been eliminated, etc. Once the voter decides not 
to give any more choices, that voter is "abstaining" from any further 
inclusion in the decision process, and the denominator for calculating a 
majority threshold. Now suppose the hand-held device only _appears_ to be 
linked to the master computer in real-time, and the device automatically 
reports to all voters that their first choice has been eliminated (without 
any actual factual basis), and all voters have the opportunity to indicate 
their second choice, and so on. Whether or not the device is linked, it 
can ask for the same information from the voters about next choices _as 
if_ their prior choice actually were eliminated. Now we have exactly 
replicated the essence of an IRV voting process, getting to it in a 
gradual transition from a two-round voting process, or from an exhaustive 
ballot voting system, showing that there is no FUNDAMENTAL difference, as 
long as the voter is entitled to rank as many choices as she wishes. We 
still have a "majority winner," as the term is generally used, whether the 
runoff choices are marked on a ballot by voters a month apart, or seconds 

The one point on which I long ago agreed with Abd is that if an 
implementation of IRV has imposed restrictions on the number of rankings 
the voter may make (due to inadequate voting technology, for example), 
there is a genuine difference between a two-round runoff "majority" and a 
possible winner of such an imperfect implementation of IRV. I can not go 
back and correct historic claims made by others, but I myself long ago 
stopped saying such imperfect implementations "assured" majority winners.

Terry Bouricius

----- Original Message ----- 
From: "Abd ul-Rahman Lomax" <abd at lomaxdesign.com>
To: "Terry Bouricius" <terryb at burlingtontelecom.net>; 
<kathy.dopp at gmail.com>; <election-methods at lists.electorama.com>
Sent: Monday, December 29, 2008 11:55 PM
Subject: Re: [EM] Why I think IRV isn't a serious alternative 2

At 08:50 PM 12/29/2008, Terry Bouricius wrote:
>Kathy Dopp wrote:
>since "abstentions or blanks" are from those who have not voted.
>I believe my interpretation of Robert's Rules of Order is correct. In
>order for a ballot being reviewed by a teller to be "blank," and thus
>excluded from the majority threshold calculation, as directed by Robert's
>Rule of order, the voter must certainly have submitted a ballot paper.

Bouricius, you are totally off, stretching, trying desperately to
find ways to interpret the words there to mean what you want them to mean.

A "blank" is a blank ballot with no mark on it. From p. 401: "All
blanks must be ignored as scrap paper."

There is no way to know if this was actually cast by a voter, or if
it was a piece of paper stuck to the underside of another ballot. It
is a *blank*.

There is another possibility: the ballot has multiple questions on
it. In this case, each section is treated as if it were a separate
piece of paper. In this case, if an election is in a section, and
there are no marks in the section, the ballot is considered, for that
election, as if blank. In this case, we may consider that the voter
has abstained.

But if the voter marks in the section, but the marks are ambiguous,
or do not cast a vote for an eligible candidate, in this case the
voter is considered to have voted, and is part of the basis for a 

Public election rules differ here. A voter must generally have cast a
vote for an eligible candidate, and the vote must not be spoiled, if
I'm correct, for it to count as part of the basis for a majority.

Robert's Rules of Order places particular emphasis on finding a
majority, and if a vote is doubtful, it may have been the intention
of the voter to participate, but not to vote for the otherwise-winner.

>RRONR is clearly not referring to hypothetical ballots from those members
>of an association who did not submit a ballot at all.

Not submitting a ballot at all -- or submitting an explicitly
abstaining ballot -- is an abstention.

>  Those who do not
>submit a ballot clearly did not vote, but those who cast ballots may
>abstain or leave the ballot blank, and thus not have their ballots
>included when calculating the majority threshold.

Casting a blank ballot is equivalent to an abstention, except it
isn't explicitly recorded as such, because the member pretended to
vote. However, all the member has to do is write on the ballot NO!
and it is a vote. Against all the candidates, effectively. (YES!
would have the same effect!)

>  The only question is
>whether an exhausted ballot should be interpreted as abstaining on the
>question of which finalist should win, or if that ballot should be
>interpreted as an "illegal vote," which RRONR says should be included in
>calculating the majority threshold.

There is no question. Bouricius, THERE IS NO QUESTION. Not for any
parliamentarian. Robert's Rules are quite clear, if you actually read
the whole section on preferential voting, that majority failure may
occur if voters don't fully rank candidates. This was utterly clear
from precedent, and the interpretation that you are making up here
does enormous violence to the very concept of majority vote.

Questions submitted to votes should be explicit. Voters don't
definitively know who the finalists are, with IRV. They may have
intended to vote for a finalist, but got it wrong as to who the
finalists were. They may detest both finalists and are unwilling to
support either. If a majority is required, truncation is a very
legitimate strategy, it means, please, if it is not one of the
candidates I have ranked, I want further process to determine a
winner, I want the chance to reconsider and maybe even to write in a
candidate on the runoff ballot. (*Which is allowed in many places.*)

>  One can think of the ranked ballot as
>a series of questions about pairwise contests...not unlike the way a
>Condorcet ballot is viewed...

You can. But that's not what's on the ballot.

>  one of the questions could be IF the race
>comes down to a final runoff between candidate C and candidate E, which
>should win?

Sure. Now, there are 23 candidates, as in San Francisco. There are
three ranks on the ballot. Further, I don't even recognize most of
the names. Maybe I know the frontrunners, but what if I don't? Should
I vote for someone who I don't know? No, I vote for the candidate or
candidates I know and trust. In a real runoff election, if no
majority is found, I am then presented with, usually, two candidates,
and I can pay particular attention to them. We see comeback elections
with real runoffs that we don't see with IRV, for several reasons,
all of which indicate that these comebacks improved results.

>  The difference between IRV and Condorcet is that IRV uses a
>sequential algorithm to determine  which candidates are finalists, while
>Condorcet does not reduce to "finalists" at all.

Condorcet could be conceptualized that way. In fact, so can
Plurality. Imagine an IRV ballot, only nobody adds additional ranks.
(Bucklin elections saw additional ranking of 11% in an Alabama
primary series, and my expectation is that IRV elections in the same
environment would probably have about the same truncation level. -- 
but this isn't the point here). Run the eliminations.

Presto! Majority vote, using Plurality voting. What has happened is
that all the ballots that do not contain votes for the top two are
not counted. It's totally arbitrary, actually, we could claim
Unanimity, just carry the rounds one more step with this Plurality
counting or with IRV.

Imagine the campaign! GUARANTEE UNANIMITY!

Of course it would be immediately seen as ridiculous. You don't get
unanimity by disregarding all disagreement, and you don't get a
majority by a similar approach.

>  However, if a voter has
>indicated no ranking for either C or E, that voter has effectively
>abstained from that particular question.

Sure. That particular question. But not the election.

Why is Bouricius torturing this point? Because, obviously, the claim
of "finds majority winners" is very important to the FairVote talking
points, it's been central for a long time, it is found in ballot
arguments, articles, all over the place. He's trying to find a way to
justify it.

Retroactively. Quite simply, those words did not mean what he is
claiming they *might* mean, to the people they were dumped on.
People, quite simply, didn't think carefully about the implications
of truncation. A lot of people have in mind Gore v. Bush, and
immediately think that those votes for Nader might have flipped that
election. They look at the numbers and think that this would have
given Gore a win. Maybe. There were also other minor party candidates
involved, and many of them would not have added preferences. Quite
simply, we don't know if the election outcome would have changed, and
we don't know if preferential voting would have found a majority.

But "majority" sounds like a very good goal, and communities which
have majority-required top two runoff clearly have valued it. "Find
majorities without expensive and inconvenient runoff elections" plays
well. I'm sure much of this occurred immediately to Richie when
"instant runoff voting" was suggested to him in the mid 1990s.

Quite simply, it doesn't happen. IRV, with nonpartisan elections, is
not electing anyone other than the plurality winner (first
preference), and when a majority is not found in the first round, it
usually isn't found after that.

"A majority of votes will continued to be required" is what the
voters of San Francisco were promised. The IRV "majority" is not a
requirement, it is a mathematical certainty, if it refers to
last-round majority, created by disregarding all those inconvenient
ballots with legal, sincere votes for the wrong candidates. Did they
notice that the IRV proposition actually struck the majority
requirement from the law? If it was still required, why was it being
struck? I could blame the voters, for sure, but I could also blame
the members of the committee that put together the "impartial
summary," the promoters of IRV who clearly encouraged that
misunderstanding, and the opponents who quite clearly didn't do their
own homework.

>  Since the voter who voluntarily
>truncates is de facto abstaining from deciding which finalist should be
>elected, if the voter has indicated no preference between them, I think 
>is reasonable to treat this abstention as an abstention as directed by

That is pure deception, turning Robert's Rules of Order on its head,
producing a result clearly contrary to the result that they
explicitly say will occur. A voter who has voted in an election has
not abstained, no matter how you slice it. The voter may have
abstained from a pairwise election, for sure, but that is a specific
abstention, ot an abstention from the election, and, in addition, may
not be voluntary, as Bouricius slips in. 3 ranks on the ballot. 23 

>While I agree that it may not be completely UNresonable to take the view
>that Abd and Kathy Dopp favor, I think it is contrary to the most usual
>interpretation of RRONR.

And now the kicker: we have explained -- and I could cite word for
word, and have in many places -- the explicit language of Robert's
Rules of Order on this. Bouricius has just said the exact opposite of
the truth. What he is proposing as the meaning of "abstention," and
the basis for majority, is totally contrary to the plain language of
RRONR, not to mention the "usual interpretation."

Usual interpretation by whom? By FairVote activists and those duped by 

I'm saddened, to tell you the truth. This is the absolute worst
argument I've ever seen from Bouricius, it's word manipulation to try
to take a text and make it say the exact opposite of what it plainly says.

I'd thought that he was above that, but, apparently not.

The public will *not* be fooled when the issues are made plain and clear.

More information about the Election-Methods mailing list