Talk:Ranked Robin
This is the discussion page (the "Talk:" page) for the page named "Ranked Robin". Please use this page to discuss the topic described in the corresponding page in the main namespace (i.e. the "Ranked Robin" page here on electowiki), or visit Help:Talk to learn more about talk pages.
Misleading Title
The title is misleading. The term Robin voting has frequently been used for Condorcet methods in general. There is no reason why this term should refer to Copeland-Borda exclusively. MarkusSchulze (talk) 18:28, 17 November 2021 (UTC)
- I think the "#History" section of this article is (perhaps) a bit misleading, but I'm okay with a rebranding effort around Llull/Copeland/Condorcet methods, along with a descriptive name that doesn't borrow from the last name of a white guy. I think we need to get out of the habit of naming alternative voting systems after the people that "invented" them, since we're all standing on the shoulders of giants anyway. I may just make some changes to the history section to make it more chronological (starting with Ramon Llull's work rather than the most recent rebranding efforts of Condorcet-winner compliant methods). -- RobLa (talk) 20:06, 20 November 2021 (UTC)
- I don't think the problem is (not) borrowing names from white guys, but instead that both "ranked" and "robin" are very general terms, so that the name seems more like "ranked choice voting" than "instant-runoff voting". That is, Markus argues that it describes something that could (if you look at it right) apply to every Condorcet method, and thus it shouldn't have the sole right to the term. Kristomun (talk) 00:06, 23 November 2021 (UTC)
Comparison to Black's method
I'm curious how this method compares to the method called "Black's method" here on electowiki and on English Wikipedia? I seem to recall a discussion about this on the electionscience Discord server (or perhaps in one of the reddit threads), but I can't remember if it was Black or one of the others that Ranked Robin was most similar to. How should we categorize this method (other than as a generic system that meets the Condorcet winner criterion)? -- RobLa (talk) 02:12, 22 November 2021 (UTC)
- I seem to recall that this method is Copeland//Borda, whereas Black is Condorcet//Borda (or Condorcet,Borda; those are the same because there's only one CW). If I'm right, then that should probably be pointed out in the article. It lacks a quick summary of just what kind of method it is, like the first sentence of Black's method. Kristomun (talk) 00:06, 23 November 2021 (UTC)
- Ranked Robin uses ordinal ballots, so it can't be Score. The first degree tiebreaker states: "Declare the tied candidates finalists. For each finalist, subtract the number of votes preferring each other finalist from the number of votes preferring them over each other finalist. The finalist with the greatest total difference is elected." That is, for each finalist X, X's score is sum over other finalists Y: (number of voters preferring X to Y) - (number of voters preferring Y to X).
- Suppose that a candidate is ranked first on a complete (untruncated) ballot, among the finalists. Then it gets one point from that ballot for each other finalist. If the candidate is ranked kth, it would get one point for each finalist, minus (k-1) points for the ones ranked ahead. So its score from a ballot is linearly related to its position, which describes (I think) the weighted positional system of Borda. Kristomun (talk) 11:14, 12 January 2022 (UTC)
- Kristomun says: "Ranked Robin uses ordinal ballots, so it can't be Score". As I read through the description of Ranked Robin more closely, I see there's a lot that I don't understand about the proposal. As written, it describes a "ranked ballot" (where "1" is at the top) but it seems possible to use a STAR voting-style ballot (the way my old electowidget voting mechanism worked). That said, as a drop-in replacement for instant-runoff voting, this seems fine, and moreover, it seems to me that cycles will be rare in practice, so the Copeland set seems likely to only have one winning candidate in 99.9% of the cases. Regardless, since the system uses pairwise matrices (like most Condorcet methods do), it's hard to see how this method has much in common with Borda, since adding candidates to the bottom of the ranking shouldn't make any difference in the pairwise comparison between the candidates above the newly added candidate.-- RobLa (talk) 04:46, 13 January 2022 (UTC)
- True, if the 99.9% Condorcet rate is a fundamental fact, then it doesn't much matter what Condorcet completion rule you use. It is possible that it's a side effect of the two and a half party domination that IRV encourages, though, or the two-party domination that using Plurality everywhere else does.
- I'd probably say the two main drawbacks of Ranked Robin are that it's more susceptible to strategy (than say minmax), and that it's not actually cloneproof. OTOH, Borda's honest VSE is not all that shabby, so at least you're getting something in return :-)
- As for Condorcet and Borda, there's a definite link between the two. Suppose there are no truncated or equal-rank ballots. Then the sum of A>X over all X is A's Borda score, so you can do a Borda count by using the pairwise matrix. That's what makes methods like Smith//Borda summable. Adding another candidate adds another X where A>X can be added to A's sum. In a more general sense, Borda is the "mean" while Condorcet is the "median": see http://lists.electorama.com/pipermail/election-methods-electorama.com/2011-January/092327.html. This gives a good intuition of why Borda is more strategically susceptible: because the median is robust and the mean isn't. Kristomun (talk) 10:26, 13 January 2022 (UTC)
Clone dependence
Is the bit about the limited range of clone failures true? It'd seem to me that there are two types of clone failures that this could be subjected to: crowding (failing the Copeland component) and ordinary teaming inside the Smith set (failing the Borda component). For the latter, something along the lines of
12: A>B>C>D>E>F 11: B>C>A>D>E>F 10: C>A>B>D>E>F
The Copeland set is {A,B,C}. A and B tie for Borda score, and now the usual teaming tricks work, e.g.
12: A1>A2>B>C>D>E>F 11: B>C>A1>A2>D>E>F 10: C>A1>A2>B>D>E>F
and A wins. Kristomun (talk) 12:50, 24 November 2021 (UTC)
Improvement to Ranked Robin method
On the Election-Methods discussion forum I'm suggesting that the Borda count tie breaker be replaced with "pairwise support counts" so that the pairwise matrix does not need to be presented -- to voters or to media reporters or to people who take the time to learn how the counting is done. This improvement will yield results that will be similar to the current version, but with less vulnerability to tactical voting.
The resulting improved Ranked Robin method can be described as follows:
"The [improved] Ranked Robin method elects the candidate who has the most number of one-on-one wins against every other candidate. [or whichever Copeland variation is desired] If more than one candidate has the same largest number, the method elects from those tied candidates the candidate with the highest pairwise support count. The pairwise support count for each of the tied candidates is the sum, across all the ballots, of the number of tied candidates who are ranked lower than the candidate whose pairwise support count is being counted."
The words "tied candidate" are important because the pairwise support count should not involve candidates who have already been eliminated using the Copeland method.
Several other election-method experts have given their responses to this improved method. Their responses can be accessed on these two pages:
http://lists.electorama.com/pipermail/election-methods-electorama.com/2022-March/thread.html (bottom of page)
http://lists.electorama.com/pipermail/election-methods-electorama.com/2022-April/thread.html (top of page)
The full suggestion is at the following location:
http://lists.electorama.com/pipermail/election-methods-electorama.com/2022-March/003709.html
The responses can be accessed using the "Next message (by thread): [EM] Improvement to Ranked Robin method" link.
This improvement to the Ranked Robin method will dramatically simplify the explanation of the method, which is expressed as the goal behind creating this method.
VoteFair (talk) 17:44, 5 April 2022 (UTC)
Code to implement it?
Thanks - interesting. Where can we get code to implement this? Python preferred. Nealmcb (talk) 01:46, 22 October 2022 (UTC)
Relationship to Total Vote Runoff?
Thanks! How does this compare to another new method: Total Vote Runoff: A Majority-Maximizing Form of Ranked Choice Voting by Edward B. Fole - SSRN - https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4328946 Nealmcb (talk) 18:59, 22 January 2023 (UTC)