Joshua, The wording is correct and the English seems plain to me. You just keep eliminating the candidate that has (among remaining candidates) the worst (as measured by losing votes) pairwise loss. 49 A 24 B 27 C>B A>C 49-27,   C>B 27-24,   B>A 51>49 The worst loss (as measured by losing votes) is B's (to C) so we eliminate B and then A  (as Forest put it) "is undefeated among those still in play" so A wins. Like Hare and Benham it fails Mono-raise and I think I eventually concluded that it was more vulnerable to Push-over strategy than them. It has the advantage of being super-simple and only needing the information in the pairwise matrix.  It's alternative name is Raynaud(Gross Loser). https://electowiki.org/wiki/Raynaud In the example I gave above, the three different versions of Raynaud all elect different winners. The worst loss by Winning Votes is A's so that elects C (in defiance of the Plurality criterion) and the worst by Margins is C's so that elects B. Chris > Could you elaborate on this a bit more? To me it sounds like: > > > while (no Condorcet winner) { eliminate the Minimax winner (!) } > > elect the Condorcet winner > > > > Or should that have been "most" instead of "fewest" losing votes? > > > On 4/8/24 02:57, Forest Simmons wrote: > >/Has anybody submitted Benham's GLE (Gross Loser Elimination)? Another > name />/could be Min Fuss Elimination: />//>/Until some candidate is undefeated among those still in play ... > eliminate />/the one showing the fewest losing votes in its worst remaining pairwise />/contest./

A  B  C

In an A versus B comparison, since A's "winning votes" are the ones for A, A's "losing votes" would be the ones for B. Eliminating the candidate with the worst pairwise defeat is completely unambiguous, though. Interestingly this doesn't always resolve a three-way cycle in the typical way (margin matrix is row candidate over column candidate)... A B C A 0 2 -4 B -2 0 6 C 4 -6 0 C gets eliminated, leaving A as the winner over B. Most methods would break A's weak win over B and make B the winner. If B did *worse* against C (B beats C by less than 4), B becomes the winner under this method. On 4/20/24 04:05, Chris Benham wrote: > > Joshua, > > The wording is correct and the English seems plain to me. > > You just keep eliminating the candidate that has (among remaining candidates) the worst (as measured by losing votes) pairwise loss. > > 49 A > 24 B > 27 C>B > > A>C 49-27,   C>B 27-24,   B>A 51>49 > > The worst loss (as measured by losing votes) is B's (to C) so we eliminate B and then A  (as Forest put it) "is undefeated among those still in play" so A wins. > > Like Hare and Benham it fails Mono-raise and I think I eventually concluded that it was more vulnerable to Push-over strategy than them. > > It has the advantage of being super-simple and only needing the information in the pairwise matrix.  It's alternative name is Raynaud(Gross Loser). > > https://electowiki.org/wiki/Raynaud > > In the example I gave above, the three different versions of Raynaud all elect different winners. > > The worst loss by Winning Votes is A's so that elects C (in defiance of the Plurality criterion) and the worst by Margins is C's so that elects B. > > Chris > > > >> Could you elaborate on this a bit more? To me it sounds like: >> >> >> while (no Condorcet winner) { eliminate the Minimax winner (!) } >> >> elect the Condorcet winner >> >> >> >> Or should that have been "most" instead of "fewest" losing votes? >> >> >> On 4/8/24 02:57, Forest Simmons wrote: >> >/Has anybody submitted Benham's GLE (Gross Loser Elimination)? Another name />/could be Min Fuss Elimination: />//>/Until some candidate is undefeated among those still in play ... eliminate />/the one showing the fewest losing votes in its worst remaining pairwise />/contest./ >