[EM] [CES #6985] Re: Wow: new, simple Bucklin motivation for CMJ. So renaming to Graduated MJ.

[Abd ul-Rahman Lomax]

At 01:03 PM 1/8/2013, Warren Smith wrote:
>So should this realization by Jameson Quinn tell us that all 
>previous historical
>examples of Bucklin voting should be regarded as examples of the
>"Majority-Judgment" median-based system,
>and hence can be used to help evaluate how the latter behaves in practice?
>
>Unfortunately I think not because I think Bucklin voters historically
>were urged to provide rank-orderings not ratings.

That's a bit ... narrow-minded. It's true that the Bucklin 
implementations only allowed equal ranking in the third rank. 
However, given that limitation, we can expect that voters would have 
voted according to their relative utilities, out of which information 
falls ranking. That is, those Bucklin votes, for real voters, 
probably roughly expressed relative ratings. We did see skipped 
ranks, and that is a clear indication of rating rather than pure 
ranking. It indicates preference strength. It made sense.

There was one attempted Bucklin implementation that explicitly 
assigned fractional votes. (2nd rank was 1/2 vote, 3rd rank was 1/3 
vote, as I recall.) Unfortunately, the Oklahoma Supreme Court ruled 
it unconstitutional, not because of the voting system, but because of 
a rule requiring additional rankings under some conditions, or the 
ballot would be invalidated.

Reformers sometimes over-reach! Good lesson: one step at a time! If 
you try two, you might not get any.

That was a well-meaning effort, I assume, to help produce a majority 
result. For some reason, an obvious way was neglected, using runoff 
voting. Bucklin was actually replaced with runoff voting in some 
places, and using Bucklin as a first poll in a runoff system would 
have been a very effective reform, reducing -- but not eliminating -- 
the need for runoffs.

To reiterate, the "encouraged to provide rankings" comment is 
inaccurate, for two reasons: the 3rd rank category was pure Approval, 
one could approve of as many candidates as desired. Those are 
*ratings,* an absolute category of "approval for election." And empty 
ranks were available, and were actually used. Empty ranks are 
meaningless in a ranked system.

Bucklin was a ratings system, effectively using a range ballot that 
was restricted in the first two ratings to one candidate, and that 
was missing the normal linear-distribution Range vote of 1/4 vote. It 
was analyzed by a descending search for a majority, using 
sum-of-votes. Because a voter was represented in that process by more 
than one vote, it's essentially an Approval method. The descending 
analysis is similar to MJ. However, the "ties" are resolved by 
plurality of votes, that's the difference.

Stepwise reforms to this older Bucklin are obvious.