TORN -
I'll be glad to share the runoffs with you. Cerin and I had a different count for one of the votes, and I need to go back now and see if someone changed their vote after I recorded it to confirm which count is correct. It didn't change the final result, btw, but I'm OCD about the voting and want the counts to be correct. I need to do that first, and then copy the IRV results onto something that can be imbedded in an email and send it to you. So, give me a bit of time to get to that.
Faramond and I had a discussion about the ups and downs of IRV when we did the very first committee vote ... the results are not skewed by going to second choices, unless you assume that people would vote differently if the runoff was not instantaneous. But that's the same as saying that if we voted twice we would get two different results, which is possible but not relevant.
One must be careful in one aspect, however. Say that out of 17 votes, 7 of the people have a first choice of A but that's not a simple majority. 5 have a first choice of C and 5 have a first choice of D. By luck, everyone's second choice is B. So when we eliminate either C or D, suddenly B goes from 0 to 5 and it is a contender. If we eliminate C and D, B goes from 0 to 10 and wins the vote. It is not the first choice of anyone but the second choice of everyone and wins for that reason. Is this good or bad?
Faramond and I were discussing first of all whether this could happen. He argued that it could not, because once B=0 it cannot appear again. But according to my understanding of IRV it could happen because you record the votes in order of rank, and an option could in theory accumulate enough second-rank votes to challenge the nearest contender.
I don't know what the "proper" procedure is in this case, and you're right, someone has probably studied it. So far, for us, it's only a hypothetical. Except for one tie on the Article 3 votes, all of our IRV's have been clean. Some were darn close, and especially when members don't vote I am uncomfortable with an option that wins by only one vote, but with pick-one voting where a plurality wins, you can also get ties, and sometimes something will win that the majority of people really, really do not want.
Jn
_________________
"All things considered, I'd rather be in Philadelphia."
Epigraph on the tombstone of W.C. Fields.