[EM] Pairwise Median Rating

Chris Benham cbenhamau at yahoo.com.au
Mon Jan 22 18:55:20 PST 2024


Ted,

> ...you didn't comment on whether ballots with all Smith candidates 
> below top rating should have their ratings bumped up: i.e., D > E > A 
> > blank > B  (A and B in Smith) would be recounted as A > blank > B. 

I don't think I left anything ambiguous.

Assuming in your example we are using say 5-slot ratings ballots then we 
interpret it as a score ballot thus:  D5, E4, A3, B0.

If A and B are in Smith then the average score of candidates in the 
Smith set is  3+0/2 = 1.5.   Only A is scored above 1.5 so only A is 
approved.

> I also noticed that there were cases where Smith//ASM(implicit) would 
> get different results (better, IMO) than Smith//Implicit-approval.
>

What does "ASM" stand for?

Chris


On 23/01/2024 10:56 am, Ted Stern wrote:
> Chris:
>
> Thanks for the clarifications, though you didn't comment on whether 
> ballots with all Smith candidates below top rating should have their 
> ratings bumped up: i.e., D > E > A > blank > B  (A and B in Smith) 
> would be recounted as A > blank > B. I think this makes the most sense 
> as a voter whose favorites are eliminated would want to ensure that 
> their highest ranked Smith candidate is counted as approved.
>
> In general I agree with your comments, though I think 
> Condorcet//Approval with all ranked ballots approved is probably not 
> optimal, and Approval Sorted Margins with explicit approval would be 
> too complex for a public proposal. I'd be happy with 
> Condorcet//Top-ratings as a public proposal.
>
> Smith//Implicit-approval seems to perform well in a number of 
> situations, but not appreciably better enough to make it worth the 
> effort of trying to get people to accept something more complicated 
> than Condorcet/Top-ratings. I also noticed that there were cases where 
> Smith//ASM(implicit) would get different results (better, IMO) than 
> Smith//Implicit-approval.
>
>
> On Fri, Jan 19, 2024 at 4:05 PM C.Benham <cbenham at adam.com.au> wrote:
>
>>     How is the average calculated?
>
>     We interpret the ratings ballots as score ballots, giving zero
>     points for the bottom rating (which is default for unrated),
>     1 point for the next highest, 2 points for the next above that and
>     so on.
>
>     Then for any given ballot we add up the scores of the candidates
>     in the Smith set and divide that by the number of candidates
>     in the Smith set and interpret that ballot as approving those
>     Smith set candidates it scores higher than that average score.
>
>     That simulates the best approval strategy if the voters only know
>     which candidates are in the Smith set.
>
>>     What advantage does Approval Sorted Margins have over
>>     Smith//Implicit-Approval?
>
>     Do you mean Approval Sorted Margins using ranking ballots with an
>     explicit approval cutoff?
>
>     Assuming yes, it uses a more expressive ballot, it is less
>     vulnerable to Defection strategy, and burial strategies are more
>     likely to have no effect rather than backfire.
>
>     In the method I proposed, omitting the ASM step and just electing
>     the candidate with the highest approval score (derived
>     as specified) would I concede make for a simpler method that is
>     nearly as good.
>
>     I worry a bit that with all methods that begin with eliminating or
>     disqualifying all candidates who aren't in the Smith set or
>     just "elect the CW if there is one", over time if there is never a
>     top cycle then the top-cycle resolution method could stop
>     being taken seriously.
>
>     An attractive feature of ASM is that it is a Condorcet method that
>     a lot of the time would work fine without anyone needing
>     to know if there is top cycle or not.
>
>     If the Approval order is  A>B>C  and A pairwise beats B and B
>     pairwise beats C no-one needs to enquire about the pairwise
>     result between A and C.
>
>     If we want something super simple to explain and sell, then
>     Condorcet//Top Ratings and Condorcet//Approval (voted above bottom)
>     are both not bad and much better than STAR.
>
>     Chris Benham
>
>
>
>     On 18/01/2024 10:13 am, Ted Stern wrote:
>>
>>
>>     On Tue, Jan 16, 2024 at 7:27 AM C.Benham <cbenham at adam.com.au> wrote:
>>
>>         Ted,
>>
>>>           3. Otherwise, drop ballots that don't contain ranks above last for
>>>                any member of the Smith Set.
>>
>>         Why not simply drop all ballots that make no distinction
>>         among members of the Smith set?
>>
>>>         I believe it passes LNHelp.
>>
>>         Douglas Woodall showed some time ago that Condorcet and
>>         LNHelp are incompatible.  I can't find
>>         his proof, but it says so here:
>>
>>         https://en.wikipedia.org/wiki/Later-no-help_criterion
>>
>>>         TheCondorcet criterion
>>>         <https://en.wikipedia.org/wiki/Condorcet_criterion>is
>>>         incompatible with later-no-help.
>>
>>         From your post again:
>>>         It probably fails Participation ..
>>
>>         It has been known (for a longer time) that Condorcet and
>>         Participation are incompatible.
>>
>>         So since we know for sure that your method meets Condorcet,
>>         we also know that it doesn't meet
>>         Later-no-Help or Participation.
>>
>>         Using a multi-slot ratings ballot for a Condorcet method of
>>         similar complexity I like:
>>
>>         *Eliminate  all candidates not in the Smith set.
>>
>>         Interpret each ballot as giving  approval to those remaining
>>         candidates they rate above average (mean
>>         of the ratings given to Smith-set members).
>>
>>         Now, using these approvals, elect the Margins-Sorted Approval
>>         winner.*
>>
>>
>>     It seems to me that Smith//Implicit-Approval or
>>     Smith//implicit-approval-sorted-margins would be affected by a
>>     couple of factors:
>>
>>       * How is the average calculated? Do you normalize scores? In
>>         other words, if a ballot has non-Smith candidates in the
>>         first, say, three ranks, do you up-rank the Smith candidate
>>         scores on that ballot by three? Also, if there are ranks
>>         below the top that contain only non-Smith candidates, do you
>>         collapse those ranks or leave the relative rank spacing on
>>         the ballot between Smith candidates untouched?
>>       * Approving Smith Candidates with scores above the mean has
>>         similarities to Median Ratings. It would be more similar and
>>         probably more stable to use the trimmed mean -- drop the top
>>         and bottom 25% of scores. This would give you an average
>>         score closer to the median.
>>
>>     What advantage does Approval Sorted Margins have over
>>     Smith//Implicit-Approval? I like ASM but fear it is probably too
>>     complex for any advantage it gives you.
>>
>>
>>>         Has Smith//Median Rating been proposed before?
>>         Not that I know of.
>>
>>         Chris Benham
>>
>>
>>
>>>
>>>         *Ted Stern*dodecatheon at gmail.com
>>>         <mailto:election-methods%40lists.electorama.com?Subject=Re%3A%20%5BEM%5D%20Pairwise%20Median%20Rating&In-Reply-To=%3CCAHGFzOSm%3Deni2SuD5YRMrYBu4Gn9%2BYQ2NrqC_sXG8QFPrcVApQ%40mail.gmail.com%3E>
>>>         /Tue Jan 2 15:12:26 PST 2024/
>>>
>>>           * Previous message (by thread):[EM] [Game Theory] Iterated
>>>             Prisoners' Dilemma as a voting method metric
>>>             <http://lists.electorama.com/pipermail/election-methods-electorama.com/2024-January/005215.html>
>>>           * Next message (by thread):[EM] Pairwise Median Rating
>>>             <http://lists.electorama.com/pipermail/election-methods-electorama.com/2024-January/005218.html>
>>>           * *Messages sorted by:*[ date ]
>>>             <http://lists.electorama.com/pipermail/election-methods-electorama.com/2024-January/date.html#5216>[
>>>             thread ]
>>>             <http://lists.electorama.com/pipermail/election-methods-electorama.com/2024-January/thread.html#5216>[
>>>             subject ]
>>>             <http://lists.electorama.com/pipermail/election-methods-electorama.com/2024-January/subject.html#5216>[
>>>             author ]
>>>             <http://lists.electorama.com/pipermail/election-methods-electorama.com/2024-January/author.html#5216>
>>>
>>>         ------------------------------------------------------------------------
>>>         Continuing my search for a summable voting method that discourages burial
>>>         and defection, I've come across this hybrid of Condorcet and median ratings
>>>         that acts like Smith/Approval with an automatic approval cutoff. I'm
>>>         calling it Pairwise Median Rating (PMR), but it could also be described as
>>>         Smith//MR//Pairwise//MRScore:
>>>
>>>             1. Equal Ranking and ranking gap allowed (essentially a ratings method
>>>             with rank inferred). For purposes of this discussion, assume 6 slots (5
>>>             ranks above rejection).
>>>             2. In rank notation for this method, '>>' refers to a gap. So 'A >> B'
>>>             means A gets top rank while B gets 3rd place. Similarly '>>>' means a gap
>>>             of two slots: 'A>>>B' means A is top ranked while B is in 4th place.
>>>             3. [Smith]
>>>                1. Compute the pairwise preference array
>>>                2. The winner is the candidate who defeats each other candidate
>>>                pairwise.
>>>                3. Otherwise, drop ballots that don't contain ranks above last for
>>>                any member of the Smith Set.
>>>             4. [Median Rating]
>>>                1. Set the MR threshold to top rank.
>>>                2. While no Smith candidate has a majority of undropped ballots at or
>>>                above the threshold, set the threshold to the next lower rank,
>>>         until there
>>>                is no lower rank.
>>>                3. The winner is the single candidate that has a majority of
>>>                undropped ballots at or above the threshold.
>>>             5. [Pairwise]
>>>             1. Otherwise, if more than one candidate passes the threshold, look for
>>>                a pairwise beats-all candidate among candidates meeting the MR threshold.
>>>                (i.e. Condorcet on just the MR threshold set).
>>>                2.  If there is one, you have a winner.
>>>             6. [MR Score]
>>>             1. Otherwise, the winner is the Smith set candidate with the largest
>>>                number of ballots at or above the Median Rating threshold (their MRscore).
>>>
>>>         This method is essentially Smith//Approval(explicit) with the approval
>>>         cutoff automatically inferred via median ratings
>>>
>>>         Step Smith.3, dropping non-Smith-candidate-voting ballots, could be
>>>         considered optional, but by doing that, you ensure Immunity from Irrelevant
>>>         Ballots (IIB), aka the zero ballot problem that affects other Median Rating
>>>         / Majority Judgment methods. In other words, the majority threshold is
>>>         unaffected by ballots that do not rank a viable candidate. It is possible
>>>         to do this summably if need be.
>>>
>>>         PMR either passes the Chicken Dilemma criterion without adjustment, or
>>>         there is a downranking strategy for defending against defection.
>>>
>>>         Consider the following examples from Chris Benham's post re MinLV(erw)
>>>         Sorted Margins (
>>>         http://lists.electorama.com/pipermail/election-methods-electorama.com/2016-October/000599.html
>>>         ):
>>>
>>>         >/* 46 A>B /*>* 44 B>C (sincere is B or B>A)
>>>         *>* 05 C>A
>>>         *>* 05 C>B
>>>         *>>* A>B 51-49,    B>C  90-10,    C>A 54-46.
>>>         *
>>>
>>>         With sincere ballots, A is the Condorcet Winner (CW).  With B's defection,
>>>         there is a cycle, and there is no CW. The Smith set is {A, B, C}. The MR
>>>         threshold is 2nd place, and A and B both pass the threshold. A defeats B,
>>>         so A is the winner and B's defection/burial fails.
>>>
>>>         >/* 25 A>B /*>* 26 B>C
>>>         *>* 23 C>A
>>>         *>* 26 C
>>>         *>>* C>A  75-25,    A>B  48-26,   B>C  51-49*
>>>
>>>         C wins with PMR (MR threshold is first place). B would win with most
>>>         other Condorcet methods.
>>>
>>>         >/* 35 A /*>* 10 A=B
>>>         *>* 30 B>C  (sincere B > A)
>>>         *>* 25 C
>>>         *>>* C>A  55-45,     A>B  35-30 (10A=B not counted),   B>C 40-25.
>>>
>>>         *A wins with sincere voting. When B defects to try to win, which it
>>>         would do with most other Condorcet methods, B wins. With PMR, C wins,
>>>         an undesirable outcome for B.
>>>
>>>         Here is another example from Rob LeGrand
>>>         (https://www.cs.angelo.edu/~rlegrand/rbvote/calc.html). It's not a
>>>         good example for chicken dilemma resistance, but it does demonstrate
>>>         differences from Schulze, MMPO, RP and Bucklin:
>>>
>>>         # example from method description page
>>>           98:Abby>Cora>Erin>Dave>Brad
>>>           64:Brad>Abby>Erin>Cora>Dave
>>>           12:Brad>Abby>Erin>Dave>Cora
>>>           98:Brad>Erin>Abby>Cora>Dave
>>>           13:Brad>Erin>Abby>Dave>Cora
>>>         125:Brad>Erin>Dave>Abby>Cora
>>>         124:Cora>Abby>Erin>Dave>Brad
>>>           76:Cora>Erin>Abby>Dave>Brad
>>>           21:Dave>Abby>Brad>Erin>Cora
>>>           30:Dave>Brad>Abby>Erin>Cora
>>>           98:Dave>Brad>Erin>Cora>Abby
>>>         139:Dave>Cora>Abby>Brad>Erin
>>>           23:Dave>Cora>Brad>Abby>Erin
>>>
>>>         The pairwise matrix:
>>>
>>>         against
>>>         Abby Brad Cora Dave Erin
>>>         for Abby  458 461 485 511
>>>         Brad 463  461 312 623
>>>         Cora 460 460  460 460
>>>         Dave 436 609 461  311
>>>         Erin 410 298 461 610
>>>
>>>         There is no Condorcet winner.  The Smith set is {Abby, Brad, Dave, Erin}.
>>>
>>>         Abby wins with Schulze, MMPO, while Brad wins with Ranked Pairs, Erin
>>>         wins with Bucklin.
>>>
>>>         In PMR, with a threshold of 3rd place, Abby, Brad, and Erin all pass
>>>         the threshold. Brad defeats Abby and Erin to win. But Brad's threshold
>>>         score of 484 is only slightly over the 50% mark of 460.5, so the Dave
>>>         voters hold the balance of power. Dave defeats Brad pairwise, so Dave
>>>         voters might not be as happy with a Brad victory, and Abby might be
>>>         able to persuade Dave voters to downrank Brad but not Abby. If
>>>         successful, Brad drops 44 points in MRScore and is no longer in the MR
>>>         threshold set. Abby defeats Erin, so Abby wins.
>>>
>>>         * 21:Dave>Abby>>Brad>Erin>Cora (Brad -> 4th place)*
>>>           30:Dave>Brad>Abby>Erin>Cora
>>>           98:Dave>Brad>Erin>Cora>Abby
>>>         139:Dave>Cora>Abby>Brad>Erin
>>>         * 23:Dave>Cora>>Brad>Abby>Erin (Brad -> 4th place)*
>>>
>>>         PMR passes Condorcet Winner, Condorcet Loser, IIB, and is cloneproof.
>>>         I believe it passes LNHelp. It probably fails Participation and IIA.
>>>         There are probably weird examples where changing one vote changes the
>>>         MR threshold. But overall, I think it has a good balance of incentive
>>>         to deter burial and deliberate cycles.
>>>
>>>         Has Smith//Median Rating been proposed before? It seems like a simple
>>>         modification to MR on its own.
>>
>>         ----
>>         Election-Methods mailing list - see https://electorama.com/em
>>         for list info
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20240123/cffcb64c/attachment-0001.htm>


More information about the Election-Methods mailing list