[EM] Chain Climbing

[Kristofer Munsterhjelm]

On 04/27/2014 06:24 AM, Michael Allan wrote:
> Forest Simmons said:
>> You are absolutely right about the heart of the Chicken Dilemma;
>> there is no way to tell from the above ballot set alone who the real
>> winner should be, because different sincere scenarios lead to the
>> same set of ballots. ...
>>
>> Where do we go from here?
>
> I suspect that a system (here an electoral decision system) cannot be
> both rational (sincere, anomaly free) and decisive at the same time.

I think I read of something like that as well, but I can't remember just 
where. The closest I *can* remember is a paper about computational 
hardness in manipulating election methods, where the authors said that 
all methods that pass a certain weak type of monotonicity can be easily 
manipulated. But I think it was Warren who said that these results 
aren't important in any case because in most real world elections, there 
are only a few candidates and thus every method is easily manipulable if 
you know every other vote.

> I have no formal proof.  But I think we must step outside the decision
> system.  It takes a combination of systems - one rational, say, and
> one decisive - brought into a mutual relation in order to act (elect)
> both rationally and decisively.

Another option is to weaken the impact. If the election is multi-winner, 
it might be possible for a group of voters to manipulate the method for 
some of the winners, but not all of them. In contrast, in a 
single-winner election, if the strategizers succeed, they win it all.

Although Sainte-Laguë degrades to Plurality when there's only one seat 
and thus is based on it, talk of strategy isn't nearly as prevalent here 
as it is in the US (which has a Plurality single-winner election and 
single-member districts).

Perhaps one could consider votes in assembly to constitute the decisive 
method and votes for the representatives to be the rational method. I've 
also seen papers argue that in-assembly votes work more like game theory 
predicts (at least for smaller assemblies) and so that equilibrium 
results that predict the Condorcet winner of a proposal will be elected 
are more applicable there.

I'm unsure of how to generalize this to say, delegate proxy systems, 
however.