[EM] SARA voting: easier-to-describe MAS

[Jameson Quinn]

Here's a new definition of SARA
<http://wiki.electorama.com/wiki/Support_Accept_Reject_Abstain_voting>:

Support Accept Reject Abstain (SARA) voting is very similar to Majority
Acceptable Score voting
<http://wiki.electorama.com/wiki/Majority_Acceptable_Score_voting>, which
is the graded Bucklin <http://wiki.electorama.com/wiki/Graded_Bucklin> method
which uses 3 grade levels <http://wiki.electorama.com/wiki/3_grade_levels>and
breaks median ties using Score voting
<http://wiki.electorama.com/wiki/Score_voting>. SARA works as follows:

   1. *Voters can support, accept, reject, or abstain on each candidate.
   Default is abstain. Candidates get 2 points for each percent of "support"
   and 1 point for each percent of "accept", for a total of 0-200 points.*
      - *"Support" helps a candidate the most, "reject" hurts the most.
      "Accept" helps less than support, and is good for helping a second-rate
      candidate beat a third-rate one. "Abstain" helps a candidate avoid
      elimination in step 3 but hurts them in steps 2 and 4; it's good for a
      second-rate candidate when there's a first-rate one who can beat them, or
      for a candidate you're not sure about.*
   2. *Eliminate any candidates with under 50 points.*
      - *50 points would be the score of a candidate supported by 1/4 or
      accepted by 1/2. Anything less than that probably indicates a
lesser-known
      candidate who got a lot of 'abstains'; eliminating them here ensures they
      won't be the only candidate remaining after step 3.*
   3. *Eliminate any candidates rejected by over 50%, unless that leaves
   nobody.*
      - *If possible, the winner shouldn't be somebody opposed by a
      majority.*
   4. *Highest points wins. In case of a tie, fewest rejections wins.*
      - *This finds the candidate with the widest and deepest support.*



*...*

This does not pass the CD criterion; in the scenario (expressed as
Support>Accept>Abstain>Reject)

28: A>>B>C
27: B>>>AC (honest is B>>A>C)
45: C>>>AB

... the truncation by the B voters successfully elects B.

However, it does avoid a slippery slope. That is to say, in the above
scenario, even though A leads B by just one vote, a mere two strategic B
voters cannot elect B by truncating:

28: A>>B>C
2: B>>>AC
25: B>>A>C
45: C>>>AB

A wins, even though B would win with the analogous truncation in score or
Approval.

I think that this "no slippery slope" property is sufficient to give good
CD performance in the real world. If you assume that most other voters in
your faction will not be trying CD strategy, there is no incentive to try
it yourself, so the ball never gets rolling.

On the other hand, it can avoid center squeeze:

28: A>>B>C
27: B>>A>C
3: C>B>>A
42: C>>>AB

B, the CW, correctly wins.

If I'm right that this kind of system avoids CD strategy, the only
effective cognitive burden of this method is the decision between "accept"
and "abstain". You should "accept" a lesser evil against a greater evil;
you should "abstain" if you think that your favorite can win. This is, I
think, the minimum possible cognitive burden for a method that does well on
both CD and center squeeze; and it's distilled to its purest form.

2016-10-22 18:24 GMT-04:00 Jameson Quinn <jameson.quinn at gmail.com>:

> Support Accept Reject Abstain voting works as follows:
>
>
>    - Voters can support, accept, reject, or abstain on each candidate.
>    Default is abstain.
>    - Call a candidate "acceptable" if they are rejected by 50% or less
>    and supported or accepted by over 25%. If any candidates are acceptable,
>    eliminate all who aren't.
>    - Give remaining candidates 2 points for each "support", 1 point for
>    each "accept", and half a point for each "abstain". Highest points wins.
>
>
> This moves a bit away from the Bucklin roots of MAS, but it further
> reduces the instability of cooperation in a CD scenario.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20161023/e05a8604/attachment.htm>