[EM] TACC (total approval chain climbing) example

[Kristofer Munsterhjelm]

On 04/21/2014 04:20 AM, Forest Simmons wrote:
> Michael,
>
> I agree with your assessment of the current possibilities, and I realize
> that Woodall and Benham have the advantage of being based on a familiar
> runoff idea.  It is also possible to do chain climbing on the IRV
> elimination order, but monotonicity and immunity to second place
> complaints would be sacrificed.
>
> Immunity to second place complaints says that if you re-run the election
> with the (old) winner removed, the new winner will be someone beaten
> pairwise by the old winner.  Neither Benham nor Woodall satisfies this
> nice property. In other words if X is the Woodall Winner (say) there
> might be another candidate Y that (1) becomes the Woodall winner when X
> is removed, and (2) is not beaten by X pairwise   This may seem fishy to
> the Y supporters.
>
> Steve Eppley's concern about all of the truncations being used up by one
> very bad candidate should not be a concern if the voters realize that no
> totally truncated candidate can win no matter how many additional
> truncations there are among the remaining candidates.  A Pareto
> Dominated candidate is covered by each of  its dominating candidates,
> but the TACC winner cannot be a covered candidate.
>
> Another way to address Eppley's concern: Instead of the implicit
> approval order for TACC we could (without sacrifice of any nice
> property) use the Bucklin Order according to the equal ranking (whole)
> version of Bucklin.  This order is determined primarily by the median
> rank of the candidate and secondarily by how much support it has at that
> rank or above.
>
> You probably already knew all of this, but I wanted to get it into the
> archives for the record.
>
> By the way, what happened to the regular election methods archives?

Apparently there's been some kind of mess-up by the hosting providers. 
Rob is talking to them and trying to get it fixed.

On a side note (to the side note, heh!), if you want to get really 
complex, you could arrange a majority runoff - i.e. a real election - 
between the winners of two TCC (total chain climbing) methods. Call 
these methods the "honest" and "strategic" method: the honest method 
does well when the voters are honest, and the strategic one is 
relatively unaffected by the kind of strategy one fears wherever this 
idea is employed. So the former could be, I don't know, minmax or Borda 
TCC, and the second could be Bucklin TCC or the method you already 
explained. The system would skip the second round if both methods' 
winners end up being the same.

I have, of course, mentioned this "honest + strategic" runoff idea 
before. With a common TCC framework, though, it would be less 
complicated than if the two methods were entirely different (e.g. 
Smith//IRV and Schulze). About the only risk (besides the complexity) 
would be the chance of getting two clones or a very boring second 
candidate, making turnout fall through the floor.