I've been looking for an RP/fpA-fpC hybrid that would behave properly in all circumstances (Selective RP had some bad edge cases) and I believe I've finally found it. Here's the shortest definition: *Iteratively restrict the set of candidates to all candidates X such that for some XYZ contest:* *- the Ranked Pairs winner is the same as for all current candidates* *- the fpA-fpC winner is X* *until there's only 1 candidate left.* That's it. The logic behind it is quite simple - either we find buried DMT candidates (and we choose among them) or we choose the current RP winner. Here's the set of criteria I was looking for: - cloneproofness - monotonicity - ISDA - DMTBR - preferably: LIIA The simulations indicate that this method satisfies all of that, with no special tie-breaking rules needed. It indeed seems to have overall strategy resistance on Smith//IRV level. But, as always, I will be thankful for any independent checks. (I'm not sure about the name yet. Burial-Detecting Ranked Pairs?)

On 10/4/23 13:53, Filip Ejlak wrote: > I've been looking for an RP/fpA-fpC hybrid that would behave properly in > all circumstances (Selective RP had some bad edge cases) and I believe > I've finally found it. > > Here's the shortest definition: > > */Iteratively restrict the set of candidates to all candidates X such > that for some XYZ contest:/* > */- the Ranked Pairs winner is the same as for all current candidates/* > */- the fpA-fpC winner is X/* > */until there's only 1 candidate left./* > > That's it. The logic behind it is quite simple - either we find buried > DMT candidates (and we choose among them) or we choose the current RP > winner. I'm not quite sure what you mean by iteratively restricting the set. Could you give me an example of how the method works? Say, Schulze's five candidate example: 5: A>C>B>E>D 5: A>D>E>C>B 8: B>E>D>A>C 3: C>A>B>E>D 7: C>A>E>B>D 2: C>B>A>D>E 7: D>C>E>B>A 8: E>B>A>D>C if it wouldn't be too tedious? It's got a large Smith set. -km