Re: [EM] the Mono-Raise criticism of IRV and its hybrids
There are at least two good answers to that criticism: 1. You know very well that, with voting systems, you get what you pay for, and you pay for what you get. An extra advantage or good property comes at a price. IRV meets MMC and is free of the chicken dilemma. Benham and Woodall likewise, and they additionally meet the Condorcet Criterion (CC). Though CC isn't really necessary, it often or usually is a good idea, for rank methods for official public elections, or even meetings or organizations, where IRV's compromise-elimination can make the method adversarial. Beatpath meets Mono-Raise, but has the chicken dilemma. It's a question of what has practical importance to voters, and what is merely an aesthetic criticism. 2. For a Beatpath advocate to criticize IRV or its hybrids for failing Mono-Raise is more than a little hypocritical, because Beatpath, like IRV, fails Participation and Mono-Add-Top. Additionally, Beatpath fails Mono-Add-Unique-Top, which IRV doesn't fail. As for the count difficulty or insecurity criticism, as I've said, an IRV count is little more than a series of Plurality counts. Each precinct can have its counts sent in to Central, and the Central results reported, just as in a Plurality election. The public imaging approach to security would work as well with IRV as with any method. As always, I disclaim that don't propose IRV , or any FBC-failing method, for current conditions, where the media have most people convinced that only the Democrat or Republican can win. The Greens offer IRV,and it would be fine for the Green scenario. Maybe the CW-preferrers won't like the elimination of their CWs (likely the Greens will be the CW-preferrers). If so, then there will be a majority who would like to change to a Condorcet-complying method. That will probably be Benham or Woodall. I've been told that Benham was actually first proposed by someone named Hill. I believe that Hill called that method "Condorcet-IRV". Condorcet-IRV is offered at the Condorcet Internet Voting Service (CIVS), as one of several rank-counts that CIVS offers. Condorcet-IRV: Do IRV till there's an uneliminated candidate who isn't pair-beaten by any other uneliminated candidate. Elect hir. Woodall: Do IRV till only one member of the Smith set remains uneliminated. Elect hir. Schwartz Woodall: Do IRV till only one member of the Schwartz set remains uneliminated. Elect hir. [end of definitions] I recommend Schwartz Woodall as my favorite general-purpose voting system for organizations, clubs, meetings, etc. But Condorcet-IRV is more briefly defined, and nearly as good. Sometimes Approval's social optimizations would make Approval the best choice. Voters often do better with Score than with Approval, because its fractional ratings mitigate strategic errors. Michael Ossipoff Election-Methods mailing list - see http://electorama.com/em for list info
[EM] List of primary voting projects & invite to mirroring network
Hi Richard, (cc Metagov) > > ... And clearly the method tries to [inform decisions] through > > consensus, even if it's not dynamic. Still I'm curious. > > Yes, through consensus. But I don't know what you mean by "not > dynamic" because, as indicated above, anything can be changed at any > time. That's a good design I think. But even if it *weren't* dynamic, then I think it would still belong on the list. That's all I meant to say. So I've added VoteFair here: http://zelea.com/w/Stuff:List_of_primary_voting_projects We have a Knight submission that concerns primary voting, the status of which is currently up the air. Knight is about to pick 50 submissions to proceed to the next phase (refinement). If we make the cut, then I imagine we could still modify the submission during the refinement phase, though I'm not 100% sure. Bearing that in mind, please consider joining us there. If you agree, we could try to add VoteFair to the list of providers (currently 3): https://www.newschallenge.org/open/open-government/submission/free-range-voting/ Mike Richard Fobes said: > On 3/28/2013 10:39 AM, Michael Allan wrote: > > Thanks for explaining, Richard. > > This is an important topic, so I want to help out. > > >> The algorithm does not attempt to identify when the negotiation > >> process is done. If the participants have a genuine desire to reach > >> a mutually satisfactory agreement, then the results will slowly > >> converge on an optimum set of approved proposals. ... > > > > What drives this change (convergence) in the results? > > Dissatisfaction with the current list of recommended proposals. > > Specifically it happens when a minority fails to get their most > important proposal into the list of recommended proposals, or when any > group sees a "very bad" proposal getting into the list of recommended > proposals. > > > Do some of the > > existing participants change their minds and re-rank the proposals? > > It's more like they are seeing which of their proposals are disliked, > and splitting those up so that some of the narrower proposals (hopefully > the ones they care about the most) will be approved. > > As an example from Oregon history, a proposed law to ban all animal > traps failed to get passed by the voters. But if the law only covered > trapping specific animals (wolves & such), then it might have passed. > The proposed law would have made it illegal to trap moles, and that's > the only way to deal with moles (that anyone in this neighborhood has > found to work). > > In other words, the animal-rights activists got too greedy. Greedy > proposals get ranked as "disliked". > > Projecting this issue onto the negotiation-tool usage, if there was one > proposal to ban traps that are set for wolves, that might pass. If > there was another proposal to ban the trapping of moles, that would not > have passed. That would have given the activists what they wanted most. > By overreaching, they lost out completely. > > > Or do new participants enter the process and rank the proposals, with > > those rankings not subsequently modifiable? > > Everything can be modified at any time. > > > In your case, it might not matter. Clearly your use cases show an > > intention to inform decisions. > > Yes. One of the basic ways to find a solution that "everyone" likes is > to learn more about what people really want. > > >And clearly the method tries to do so > > through consensus, even if it's not dynamic. Still I'm curious. > > Yes, through consensus. But I don't know what you mean by "not dynamic" > because, as indicated above, anything can be changed at any time. > > Richard Fobes Election-Methods mailing list - see http://electorama.com/em for list info
[EM] Approval vs Schwartz Woodall for organizations
FairVote has claimed that Approval is only good for non-contentious
elections. I've never agreed with that. Approval can work fine with any
kind of elections, including contentious ones.
When there is strong contention, when there are greately differing
alternatives, some of which are unacceptable to some voters, voters will
want to win if they have a majority. For that kind of elections, it's
desirable to use a voting system that automatically enforces majority rule.
In those strongly-contested elections in which people want majority rule,
Approval can work too. The chicken dilemma is about an {A,B} majority that
wants to defeat C, but who don't trust eachother to not take advantage
of eachother's cooperativeness.
In fact, maybe the A voters could have a majority, but, if they don't know
it, they could give it away by helping B.
Those problems aren't really problems in Approval, because they're well
dealt with by strategic fractional ratings, whilch I've much described.
There are additional reasons why those things aren't a problem, and I've
listed them at this forum.
Nevertheless, when it's a contentious election, where there are some
alternatives that are unacceptable to some voters, you'd rather have
automatic enforcement of majority rule. That measn easier and quicker
majoriity rule.
For that kind of elections, I'd recommend Schwartz Woodall as the voting
system for an organization, club, meeting, committee, family-voting, etc.
Under less contentious conditions, when the alternatives aren't so
different, and no winnable ones are unacceptable to anyone, then there's
little or no reason to not use Approval. Approval brings important and
desirable social optimaizations, choosing the most liked alternative, or
the one that is acceptable to the most people, and maximizes the number of
pleasalntly-surprised voters.
Returning to contentious elections for a moment:
The important majority is a mutual majority. IRV and its hybrids that I
recommend (I refer to them collectively as IRV etc. or I&c) meet the Mutual
Majority Criterion (MMC) and don't have the chicken dilemma. Some meet the
Condorcet Criterion (CC) and some don't.
I don't consider CC to be necessary. A mutual majority (MM) is the
important majority. But protecting other majorities too, by electing the
CW, is good for stability, and often or usually, that's a good idea. As
I've mentioned, not electing CW compromises probably makes a method too
adversarial and non-cooperative for amicable organizations, but not for
public political elections.
Still, the stability of electing the CW can be desirable for official
public elections too, and it might be demanded by a majority. I've spoken
of the possibility that the CW preferrers might not want their CW to get
eliminated, and might side with the non-MM voters in demanding a method
that better elects CWs.
The Greens offer IRV, and I fully support that. If, after IRV has been in
use for a while, they later find that they're the CW, and they want a more
Condorcet-efficient method, that's fine too. I myself don't consider
Condorcet efficiency to be necessary for official public elections in the
Green scenario, but CW-preferrers might (or might not) want it.
By the way, Bucklin has the chicken dilemma, and in its more
briefly-defined version, the one preferred by Lomax, Bucklin fails the
Mutual Majority Criterion.
IRV, and its hybrids that I recommend all meet the Mutual Majority
Criterion and don't have the chicken dilemma.
The IRV hybrids that I recommend are:
MM/Woodall, MM//Benham, AIRV, Woodall, Benham.
I've defined Woodall and Benham in recent posts.
I'll re-state my definitions of MM//Woodall and MM//Benham here:
MM//Woodall:
Choose from the innermost mutual majority preferred-set. Choose from it by
Woodall, counting only the rankings of the members of that innermost mutual
majority.
MM//Benham:
Same, but substitute "Benham" for "Woodall".
Michael Ossipoff
Election-Methods mailing list - see http://electorama.com/em for list info
