[EM] Elimination with Take-Down

[Forest Simmons]

On Fri, May 5, 2023, 1:09 PM Kristofer Munsterhjelm <km_elmet at t-online.de>
wrote:

> On 5/4/23 22:48, Forest Simmons wrote:
> > What started as Chain Climbing has evolved into a Banks efficient
> > take-down Elimination protocol.... whenever you eliminate an
> > alternative, let it take-down with it any alternative it defeats ... an
> > expedient for conferring Banks efficiency on elimination methods.
>
> A thought about this: Could we create a method that works in this way
> but also produces the same result as Smith//Range for a three-way contest?
>
> Suppose we have an ABCA cycle and everybody but that cycle has been
> eliminated. Then there are six possible Range orders: ABC, ACB, BAC,
> BCA, CAB, and CBA, where ABC means that A has the highest score, B is
> next, and C is last.
>
> If C is last, then C is eliminated and takes A with him. B wins.
> If B is last, then B is eliminated and takes C with him. A wins.
> If A is last, then A is eliminated and takes B with him. C wins.
>
> So this only produces the "proper" outcome if the Range order is ACB,
> BAC or CBA. The other half of the time, it'll be incorrect. So for
> maintaining the VSE of a high-VSE method, it might not be a good idea.
>
> It might be useful for strategy, though. An interesting question is if
> take-down elimination Benham loses any strategy resistance. With IRV
> being as chaotic as it is, we could hardly lose any VSE :-)
>
> And another thought: wanting Banks might be overkill


It's not Banks per se that seems important to me ... it just turns out that
this Banks efficient protocol is a very inexpensive, simple way to make any
elimination method Landau efficient ... which is important for immunity
against pairwise  domination complaints ... such cheap insurance it would
be irresponsible to pass it up!


-- at least for
> current versions of the "homotopy method". Although on the hand, if we
> do get large honest Condorcet cycles, that shows there's genuine
> competition. So who knows what could happen after Condorcet has been in
> use for a long time?
>
> -km
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20230509/3420f8cf/attachment.htm>