[ARIN-consult] Consultation on Implementing Single Transferrable Voting for ARIN Elections

[William Herrin]

On Thu, Feb 10, 2022 at 4:19 PM Adam Thompson <athompso at athompso.net> wrote:
> * The defeat of a plurality candidate who got the most votes, while very much a corner case, is possible - but it is neither unexpected nor undesirable.  That's a feature, not a bug.  STV resembles a (hoping I remember the terminology correctly...) weighted centroid model, where enough smaller clusters can, in aggregate, successfully pull the center of the graph away from a dominant cluster.

Hi Adam,

I consider that a reasonable basis for claiming fairness. I'm not
convinced STV reliably does that but if it does then, okay,
reasonable. I'm not sure it's the right answer either but I'll agree
it has a reasonable claim to fairness.


> * Unless the voting system allows for both "for" and "against" votes, no-one has voted *against* any candidate at all, ever, even if that was both their thought process and their intent.  That's logically equivalent to thinking you can prove a negative.

Here I have to completely disagree with you. A majority ranking a
candidate last offers only one credible inference: that the majority
disapproves of the candidate. That's what you call Sherlock Holmes
level proof - when no other explanations are reasonable, the one that
remains is probably true.

If a voting system is fair, you should be able to turn it upside down
and get more or less same result: for STV, eliminate the candidate
with the most last-rankings instead of the one with the least
first-rankings, then the next one and so on until the number of
candidates left is the number of posts to be filled. Try that with STV
and it goes completely off the rails.


> STV has by now had quite a bit of real-world testing, certainly on scales far, far larger than ARIN and with - IMO again - much more "real-world" impact to the average voter; it has shown itself to *not* regularly produce the bizarre results its detractors are worried about; and it has shown to produce more balanced, less lop-sided outcomes that do not completely ignore minority groups.

STV's real world testing has primarily been with single-post
elections. As far as I can tell, the math that allows the last-ranked
candidate to be elected is not possible when only a single post will
be filled by the candidates.

ARIN elections fill -two- seats on the board from the pool of
candidates. STV *as proposed for ARIN* suffers a math problem which
all that "real-world testing" largely didn't test.

Regards,
Bill Herrin


-- 
William Herrin
bill at herrin.us
https://bill.herrin.us/