Re: Comments on the text for STV voting in the draft Process 2016 doc dated 18 July 2016

On Sunday 2016-07-24 11:17 -0700, Stephen Zilles wrote:
> The fifth paragraph of 2.5.2 begins,
> 
> "The shortest term is assigned to the elected candidate who received the
> least support,"
> 
> In this, the term "least support" is undefined and, as far as I can tell, it
> is not a term used in describing STV tabulation. If there is a referenceable
> source for the term, then that should be linked to. The approved text began,
> "If the tabulation system ranks candidates according to their level of
> support, the shortest term .". Thus, tying "level of support" to the
> ranking. Without this, I do not think the term, "least support" has much
> meaning. 
> 
>  
> 
> I am not at all sure how to fix this issue without making presumptions about
> the STV tabulation system.
> 
>  
> 
> One possible way to fix this is to say that the details of the "vote
> tabulation system" must specify how the "level of support for a candidate"
> is computed. Then the fifth paragraph would be just fine as is.  For
> example, using a Droop Quota and a fractional vote distribution system, the
> level of support is computed as, "the number of candidates minus the round
> in which the candidate is selected, and, within a given round, the number of
> votes (before distribution of the surplus votes) each candidate selected in
> that round received." This two part number ranks support first on the round
> in which the candidate was selected and then on the number of votes
> received. This, of course could still produce a tie.
> 
>  
> 
> This suggestion can be put into the proposed text by changing,
> 
> "details about the vote tabulation system selected by the Team for the
> election"
> 
> TO
> 
> "details about the vote tabulation system selected by the Team for the
> election, including a specification of how the level of support for each
> candidate is computed,"

Generally these systems maintain the invariant that if you run the
algorithm with a set of votes and the constraint that there are 3
seats available, and then run the algorithm with the same votes and
the constraint that there are 4 seats available, the 3 people
elected are a subset of the 4 people elected.  This allows assigning
the short term to the person in the second set but not in the first
set.  I think this is generally how short-term assignment works with
such systems.  (I think it's how it's done in Australian Senate
elections in a double-dissolution election like the one that just
happened.)

It's worth double-checking that this is true of the system that
we're using.

It might be worth skewing the wording a little bit towards that
concept, although it may well also be fine as-is.

-David

-- 
𝄞   L. David Baron                         http://dbaron.org/   𝄂
𝄢   Mozilla                          https://www.mozilla.org/   𝄂
             Before I built a wall I'd ask to know
             What I was walling in or walling out,
             And to whom I was like to give offense.
               - Robert Frost, Mending Wall (1914)