Talk:Smith,IRV

From electowiki

This is not the same thing as Smith//IRV. The "comma and slash" terminology is:

X//Y: Eliminate everybody outside the X set. Run Y on whatever remains. The winner according to Y is the over-all winner.

X,Y: Use Y (evaluated on all candidates) as a tie-breaker for X.

E.g. suppose The X set is {A, B, C}, and the outcome according to the election using method Y is D>B>C>A, and if D is eliminated from every ballot, Y gives the ordering C>B>A. Then X//Y is C>B>A (C wins), and X,Y is B>C>A>D.

The explanation should probably go on some other page, and then we should find an election where Smith,IRV and Smith//IRV differ. Kristomun (talk) 22:52, 27 February 2020 (UTC)

Ok. What does single slash Smith/IRV mean?
Smith//IRV is based on https://rangevoting.org/SmithIRV.html which is "Smith,IRV", so I thought they were the same thing.
And where did this slash notation come from? Is there some less ambiguous terminology that we could use? — Psephomancy (talk) 23:52, 27 February 2020 (UTC)
Smith/IRV doesn't have a meaning according to that notation. Perhaps it's just a redirect for a misspelling.
As far as I know, the // notation was first proposed by Bruce Anderson on the EM list (http://lists.electorama.com/pipermail/election-methods-electorama.com/1996-April/065516.html). Note that in the post, "Condorcet" means "Plain Condorcet" which was an early list name for Minmax. The comma notation first appears in a post by Chris Benham (http://lists.electorama.com/pipermail/election-methods-electorama.com/2004-December/112367.html) who refers a draft paper by Woodall. That paper doesn't specifically explain what the comma notation is; apparently Benham's notation is a generalization of this process for IRV:

The same situation arises when the Alternative Vote (AV) is used to elect a candidate from T_N . The natural method is to apply AV, stopping when all but one of the candidates in T N have been excluded, at which point the remaining candidate is declared elected.

which is equivalent to taking the T_N (Smith) set candidate closest to being the winner in IRV (i.e. eliminated last, closest to the top of the social ordering). Woodall 'does' use the X,Y notation elsewhere in the paper (e.g. "CNTT,DAC"), but he doesn't define it outright. In particular, it's not obvious if he's referring to what the list members call X,Y or what they call X//Y. Since the draft paper isn't publicly available and the comma notation seems to be ambiguously defined there, I think we can more unambiguously attribute the notation to Benham. Kristomun (talk) 10:23, 28 February 2020 (UTC)
Smith/IRV is equivalent to Smith//IRV. Whoever wrote the Condorcet Wikipedia page mentions "One possible method is to apply instant-runoff voting to the candidates of the Smith set. This method has been described as 'Smith/IRV'." BetterVotingAdvocacy (talk) 16:53, 28 February 2020 (UTC)