St. Louis and Pushover (Re: Reply to Rob regarding RCV)
Hi Chris
Regarding former St. Louis resident Lewis Reed's preference of voting
system, I'm not sure. I suspect he was selling his preference to the
highest bidder, and since he was still the President of the St. Louis Board
of Aldermen, he presumably had influence in STL politics (until the bribery
case in court made him politically radioactive). Whatever remaining
influence Reed may still have is waning with every day he spends in an
Arkansas prison, where I believe he lives today. Something tells me that
electoral reform is the least of Reed's worries in 2023.
Regarding approval's potential vulnerability to pushover when used in a
top-two primary, I'm personally not very concerned about the theoretical
possibility. Perhaps in the far future, we'll have really sophisticated
voters who understand how to strategically influence the primary in a way
that causes top-two approval to fail in a way that causes problems. My
sense is that St. Louis (which uses top-two approval) has bigger problems
than sophisticated attempts at "pushover"
I believe that St. Louis voters just wanted a good mayor in 2021, and
approved many competitors to Lewis Reed (including Tishaura Jones and Cara
Spencer). From my analysis, it looks like the wealthier voters in the
southern portion of St. Louis (i.e. the type that pay handsomely for
produce at farmers' markets in mall parking lots on the weekends to assuage
their guilt for the collapse of the independent farming in the United
States) voted for Cara Spencer. The voters in the northern part of the
city (in the floodplain at the confluence of the Missouri and Mississippi
Rivers, which USUALLY doesn't flood very much) voted for Tishaura Jones.
Given the demographics of St. Louis (43% Black, 42.9% non-Hispanic White,
5.1% Hispanic, 4.1% Asian, 1.0% Hispanic White), it seems like a Tishaura
Jones was a great choice to succeed their longtime White mayor. Northern
St. Louis is largely Black and Hispanic, and it's in really rough shape,
despite having some fantastic old brick houses and lots of fantastic real
estate that is available for very little money (see <
https://youtu.be/SPyjYQALnrE>). Jones is the first Black mayor of St.
Louis since 2001, and third Black mayor that the city has ever had (and the
first Black woman in the role).
Based on my cursory research, I'm not sure the "pushover" phenomenon is
well known outside of the jargon-speaking electoral reform community (and
perhaps not even here). Note that the electowiki section about pushover
has no citations, and the "push over" section of a similar article on
English Wikipedia even has the infamous "citation needed" tag:
https://en.wikipedia.org/wiki/Tactical_manipulation_of_runoff_voting#Push_over
Note: the "Push over" section appears to have had the "citation needed"
banner on it since 2009, which means that said banner is almost old enough
to get its driver's licence in many places. I have a hard time taking the
criterion too seriously given that it doesn't seem to warrant its own
article on English Wikipedia. It seems all of the important ones have
articles:
A question for Chris (anyone who cares to answer), what's the best
explanation of pushover at a public URL that seems reasonably academically
rigorous (e.g. something that seems like it would pass muster as a citation
on English Wikipedia)?
Rob
p.s. having driven through St. Louis many times (and even stayed there a
couple of nights), it's not hard to guess where the powers-that-be drew the
red lines:
https://storymaps.arcgis.com/stories/366759e8b76c46efbf6ff9e8fff3ac0b
On Mon, Sep 25, 2023 at 6:53 PM C.Benham cbenham@adam.com.au wrote:
Rob,
I was recently made aware of a paper published by researchers at Yale that
suggests that RCV makes polarization and extremism even more likely:https://isps.yale.edu/sites/default/files/files/di-pb-2-3-23-v3.pdf
I haven't read the paper yet, and I could be summarizing it incorrectly,
but it matches my intuition about RCV.
Unfortunately that is probably true. The conclusion of the paper includes
this weird absurd statement:
When a Condorcet Loser exists, RCV may weaken her electoral prospects for
some primitives, but we unearth contexts in which the candidates’ strategic
choices offset RCV’s benefits from second preferences to such a degree
that a Condorcet loser’s victory prospects may increase, relative to
plurality.
Lewis Reed was one of the first on the bandwagon to repeal approval voting:https://www.stltoday.com/news/local/govt-and-politics/effort-underway-to-repeal-approval-voting-in-st-louis-replace-it-with-new-system/article_2c3bad65-1e46-58b6-8b9f-1d7f49d0aaeb.html
I got to read that Lewis Reed was in favour of repealing Approval voting
in favour of "a new system", but not any more unless I wanted to subscribe.
Is that "new system" RCV?
If so, that would be a big improvement over Approval Top-Two-Runoff.
That is much more vulnerable to Push-over incentive than RCV. It loses
compliance with Favorite Betrayal, which is one of
plain Approval's main selling points.
It would be much less bad if the runoff was between the Approval winner
and the candidate most approved on ballots that don't approve the Approval
winner.
Chris Benham
[EM] Reply to Rob regarding RCV Rob Lanphier roblan at gmail.com
<election-methods%40lists.electorama.com?Subject=Re%3A%20%5BEM%5D%20Reply%20to%20Rob%20regarding%20RCV&In-Reply-To=%3CCAK9hOYk%2BKZJ4Ms5ygvaLXw9iKxc_GQAQhjAvaaK%3DB8-0PgG-ag%40mail.gmail.com%3E>
Sun Sep 24 23:00:58 PDT 2023
-
Hi Michael,
I've been letting this thread go without me, even though my first name is
in the title (and presumably doesn't refer to the ED of Fairvote or the CEO
of RealNetworks).
I think I largely agree with Kristofer Munsterhjelm and Forest Simmons in
this thread, but admittedly, I've only skimmed the conversation
(which...has a lot of words). Earlier this weekend, was inspired to copy
over some prose from English Wikipedia to create the following page because
of your mention of "L02E":https://electowiki.org/wiki/LO2E
It would seem that you are deep into the community/tribe of the Greens
based on this quote:
"Greens win all of our non-mass-media polls."
I don't think the Greens are a mainstream party. They seem to be quite
happy being a niche party complaining about how the Democrats are not as
pro-socialism and pro-environment as they are. I'm going to guess that the
Greens are fighting for fully compostable voting booths right now, and for
all of the weapons used by the police and the military to be fully
compostable. You know, the important stuff ;-)
I have two sets of recommendations based on whether discussing
single-winner reform or mult-winner reform.
- STV for multi-winner elections -- I like the proportional aspect of
STV, and the biggest problems of STV are masked by its strength at
selecting multiple candidates proportionally. I suspect that's why the
Greens are super into it. The Greens probably look at European democracies
and at Australian/New Zealand politics and believe that STV is FANTASTIC.
But if one imagines a circular, 2D space where voters' beliefs are
scattered throughout it, then I would suggest that STV is great at
selecting candidates near the perimeter of the space (proportionally
equidistantly around the edge), but does a poor job of picking candidates
in the middle of the space.
- Approval, STAR, or a Condorcet-consistent method for single-winner
reform -- this is how we plug the donut hole created by STV, and get .
We're polarized enough as a country (here in the U.S.) such that it's going
to be difficult for people who identify as Democrat/liberal/left-wing are
going to have a difficult time trusting anyone who is
Republican/conservative/right-wing. I'm cautiously optimistic that Dr.
Nicolaus Tideman's Condorcet group will actually start a well-funded
organization to match the organizations behind Approval (the Center for
Election Science) and STAR (the Equal.Vote Coalition). Very cautiously
optimistic. I think that ANYONE starting a new group may underestimate the
challenges of doing so. I also generally trust that Dr. Ka-Ping Yee did
the analysis correctly in 2005 (<https://electowiki.org/wiki/Yee_diagram>),
and that Approval and Condorcet-consistent methods perform roughly
equivalently.
Having met some folks I like from FairVote (they exist...or at least, they
did in 2018 or so), I believe their analysis regarding
Condorcet-consistency of past IRV/RCV (i.e. the vast majority have picked
the Condorcet winner). However, my fear is that we're going to see
more-and-more close RCV elections in the very near future. Most
responsible political organizations hire quants to crunch numbers (if they
have the money to do it), and/or listen to their quant volunteers who offer
compelling data-backed advice. EVERYONE is against corruption (right-wing
voters and left-wing voters), and hence why I suspect Tishuara Jones and
Cara Spencer both clobbered Lewis Reed in the first St. Louis mayoral
primary using approval (https://electowiki.org/wiki/2021_St._Louis_mayoral_election).
Lewis Reed was one of the first on the bandwagon to repeal approval voting:https://www.stltoday.com/news/local/govt-and-politics/effort-underway-to-repeal-approval-voting-in-st-louis-replace-it-with-new-system/article_2c3bad65-1e46-58b6-8b9f-1d7f49d0aaeb.html
However, I wouldn't be surprised if it was an open secret among St. Louis
voters that Lewis Reed was on the take (including Republican, Democratic,
Green, and Libertarian voters among others):https://www.ksdk.com/article/news/politics/newly-released-photos-show-st-louis-aldermen-lewis-reed-jeffrey-boyd-john-collins-muhammad-taking-thousands-in-bribes/63-e17024ae-eba0-4f6a-af9c-adeaa1d2a04c
I have no idea if it was corruption, incompetence, or both when it comes to
Bob Kiss in Burlington Vermont in 2009. I'm virtually certain, though,
that the numbers show he primarily appealed to his base, and only threw his
his opponents (even a substantial number of Democrats) a bone:https://electowiki.org/wiki/2009_Burlington_mayoral_election
And mind you, this was with all of the advantages of incumbency. If Kiss
was any good at his job from 2006 until 2009, he should have won the 2009
election in a landslide. The fact that he won in 2006 (and was presumably
Condorcet winner in that election), and then wasn't able to pull together
THE SAME VOTERS in 2009 who ranked him first place in 2006 just goes to
show that there was a lot of regret about him. The biggest problem in this
election was that many Republican voters ranked Montroll (a Democrat), Dan
Smith (an independent), and even Jason Simpson (a Green) higher than Kiss.
But because of the way the rules worked, their preferences with respect to
all of the candidates other than Kiss were ignored BECAUSE Republican Kurt
Wright made it to the last round. They would have gotten one of their
compromise candidates (e.g. Andy Montroll) if Kiss had not been as popular.
I've thought about that election a lot, because I've put myself in the
shoes of the Wright voters, and just imagine if Kurt Wright was the
vaguely-left-of-center Democrat, and Montroll was a vaguely right-of-center
Republican, and Bob Kiss was a far-right extremist who managed to get
elected to his first term because of a new election method. I would have
been pissed if the same system RE-ELECTED the crazy right-winger, even
though Kiss's support CLEARLY eroded since the initial election. But
(flipping the electoral spectrum back around) it would seem that Burlington
repealed IRV because it didn't do its job (i.e. letting the opposition to
the incumbent have some say).
I was recently made aware of a paper published by researchers at Yale that
suggests that RCV makes polarization and extremism even more likely:https://isps.yale.edu/sites/default/files/files/di-pb-2-3-23-v3.pdf
I haven't read the paper yet, and I could be summarizing it incorrectly,
but it matches my intuition about RCV. We discussed the paper quite a bit
at Sass's weekly Open Democracy Discussion a week or two ago:https://electowiki.org/wiki/Sass_Open_Democracy_Discussion
I'll probably be attending again this coming Tuesday as I often do. All
y'all are welcome to join, and if you join, you can learn more about it
from some of the people who likely have read it (or you might learn about
other research).
Rob
On 9/27/23 06:39, Rob Lanphier wrote:
Based on my cursory research, I'm not sure the "pushover" phenomenon is
well known outside of the jargon-speaking electoral reform community
(and perhaps not even here). Note that the electowiki section about
pushover has no citations, and the "push over" section of a similar
article on English Wikipedia even has the infamous "citation needed" tag ...
A question for Chris (anyone who cares to answer), what's the best
explanation of pushover at a public URL that seems reasonably
academically rigorous (e.g. something that seems like it would pass
muster as a citation on English Wikipedia)?
That seems to be Stensholt in 2010:
https://link.springer.com/article/10.1007/s00355-010-0442-0
Quoting:
Three types of strategic voting that are often discussed are the main
concerns of this article. How common are they? What is their impact
on the political landscape? How may they be mended to avoid
unfortunate consequences? In a three-candidate single seat election
they work as follows:
Strategy (1) Some voters switch ranking from ABC to BAC so that B can win instead of C.
Strategy (2) Some voters switch ranking from ABC to ACB so that A can win instead of B.
Strategy (3) Some voters switch ranking from ABC to BAC so that A can win instead of C.
The strategies are popularly called, respectively,
‘compromising,’‘burying,’and‘push-over.’ They all show a violation
of the IIA-axiom (Independence of Irrelevant Alternatives), as
candidate X may pass Y in the social ranking without passing Y in any
ballot ranking.
From what I gather, "pushover" is mainly an EM term and Stensholt
picked it up from here. Most academic articles call it something like
"nonmonotonicity" or "exploitation of nonmonotonicity", e.g. Nagel's
article on the Burr dilemma:
https://www.journals.uchicago.edu/doi/abs/10.1111/j.1468-2508.2007.00493.x
p. 56.
It's been a favored point of FairVote's to discount nonmonotonicity by
saying that exploiting it requires a very particular kind of preparation
and it's way too easy to overshoot and harm oneself. This is partly why
I've been repeating that the problem with nonmonotonicity isn't that it
can be exploited, but that it indicates that the method is getting one
of its elections wrong. I.e. it's an honest voting problem, not a
strategic voting problem.
-km