Talk:Independence of clone alternatives

This is the discussion page (the "Talk:" page) for Independence of clone alternatives. Please use this page to discuss the topic described in the corresponding page in the main namespace (i.e. the "Independence of clone alternatives" page here on electowiki), or visit Help:Talk to learn more about talk pages.

"Strictly-interpreted definition of ranked clones" point

According to Independence_of_clone_alternatives#Strictly-interpreted_definition_of_ranked-clones, using the ranked ballot definition for clones doesn't make sense because {A, B} can be considered a clone set relative to C, and {B, C} can be considered a clone set relative to B, but {A, C} is not a clone set because B is scored between A and C on one of the ballots. And therefore, the section argues, the notion of clones is self-contradictory in a scored context because, say, A is simultaneous a clone and not a clone.

But this doesn't seem exclusive to the scored context. Consider the ranked version

1: A>B>C
1: C>B>A

Here, the same argument holds: {A, B} could be a clone set, and {B, C} could be a clone set, but {A, C} is definitely not one. So this doesn't as such invalidate the ranked-clone concept. Indeed there's no reason why the clone sets should be consistent in this sense. What independence of clones does say is that if we start with an election

1: A>B
1: B>A

and A happens to win with 50% probability, then cloning B into {B, C}, the winner should come from the clone set {B, C} with 50% probability, and from A with 50% probability. And that's what a random ballot tiebreaker does, but not say random candidate, proving the latter isn't cloneproof.

So I'm inclined to just replace this section with a reference to Tideman saying that he considers the +/- epsilon rating to be a good generalization of the ranked clone concept. Kristomun (talk) 14:57, 8 July 2023 (UTC)

Name change

FWIW, I have no objections to the proposed name change :-) Kristomun (talk) 13:37, 14 July 2023 (UTC)