[EM] Re: Definite Majority Choice

[Russ Paielli]

Araucaria Araucana araucaria.araucana-at-gmail.com |EMlist| wrote:
> On  4 Apr 2005 at 23:39 UTC-0700, Russ Paielli wrote:
> 
>>I was just looking at the wiki page for DMC:
>>
>>http://wiki.electorama.com/wiki/Definite_Majority_Choice
>>
>>I saw this statement:
>>
>>"DMC chooses the same winner as (and could be considered equivalent in
>>most respects to) Ranked Approval Voting (RAV) (also known as Approval 
>>Ranked Concorcet), and Pairwise Sorted Approval (PSA)."
>>
>>Do we know for sure that DMC always chooses the same winner as RAV and
>>PSA? If so, then in what respects are they *not* equivalent?
> 
> 
> Before commenting on a wiki page, please note that you might have to
> purge your browser's cache and reload the page to get the latest
> version.  I've run into this problem recently myself.
> 
> The excerpt you cite is due to my suffering from the same affliction
> as old what's-his-name, uh, John Kerry: I can't resist inserting every
> possible mathematical qualification into what I'm describing.  I
> should probably take out the first parenthetic remark, as it serves no
> purpose.  I may have done so by the time you refresh your browser,
> in fact.
> 
> DMC, RAV and PSA are all equivalent in the sense that they choose the
> same winner.  Period.  This is the same kind of equivalence as CSSD
> being equivalent to Beatpath.  The way that they are not equivalent is
> that they don't follow exactly the same path to get to that winner.
> 
> I happen to think that DMC is the simplest-to-grasp version of all
> three methods.  Here is one way to find the winner:
> 
>       Eliminate any candidate defeated by another candidate with
>       higher total approval.
> 
>       Among the remaining candidates, the candidate with the lowest
>       approval defeats all others and is the DMC winner.
> 
> Everyone is familiar with the idea of most or least points, so a voter
> looking at the pairwise array could find the winner in a few seconds,
> by inspection.  No mention of Smith sets, no ranked pairs, no fancy
> algorithms, clean and simple.

Well, I don't think that's quite the simplest explanation. I think 
simply saying, "eliminate the least-approved candidate until a CW is 
found" is simpler (assuming that the concept of CW has been explained, 
which is unavoidable).

My version is also very easy to visualize. Put the approval scores on 
the diagonal and arrange them in decreasing (or non-increasing) order. 
Then simply cover the last row and column until you find a CW in the 
remaining part of the matrix. If you can find a CW, you can visualize 
the procedure.

It's a matter of opinion, of course. Someone should try both 
explanations on a sample group and see which one the group understands 
faster. I'll bet on mine, but maybe I'll be surprized. Of course, having 
two simple explanations doesn't hurt either.

> 
>>If these methods are equivalent, then I think we need to eventually
>>try to somehow agree on a common name for public promotion. We might
>>also be wise to agree on the simplest explanation, with the more
>>complicated explanations used as "backup" material for those who are
>>intellectually curious.
> 
> 
> Yes, of course.  But see above -- can you get simpler than that?

Yes, I think I did.

> 
>>The actual name and acronym may be critical to the public salability
> 
> 
> 'marketability' might be the word you're after.

Bingo!

> 
>>of the method, so we need to be very careful in selecting it. We
>>shouldn't rush into it. "Definite Majority Choice" seems too generic
>>and not descriptive enough to me, but I am not necessarily opposed
>>to it. I like RAV (which I proposed myself), but I don't consider it
>>an ideal name either. In any case, we must avoid at all cost using
>>the word "dropping" in the name (it sounds too much like something
>>birds do).
> 
> 
> Well, Condorcet was called 'true majority rule' in the March 2004
> Scientific American article.  The DMC winner is chosen from candidates
> remaining after eliminating definitively defeated candidates.  So if
> you want to quibble, Definitive Majority Choice might be more
> accurate.  But I think we want to avoid having more than one
> 4-syllable word in the name ;-).

Yes, but SciAm is not a marketing journal, and academic names are not 
always the best popular names.

> Yes, the name can be important.  But you have to watch out for the
> initials also.  For example, I was thinking of something called
> Pairwise Ordered Sorting a while back and realized POS would be an
> unfortunate acronym.

I agree that the acronym is important too. It should be catchy if 
possible and it shouldn't bring up a million irrelevant hits in a search 
engine.

Let me just explain a couple of reasons I like the name Ranked Approval 
Voting. It leverages the name Approval Voting, which is already somewhat 
well known. It also gives a clue as to what it adds to AV. The name 
alone gives people who have heard of AV a good clue as to what it is. 
DMC does not do that. The acronym RAV is not too bad either. Again, the 
naming decision is hardly urgent. I'm just planting some ideas in your mind.

--Russ