[EM] Re: A few compliances & noncompliances of DMC

[Araucaria Araucana]

On 14 Apr 2005 at 19:53 UTC-0700, MIKE OSSIPOFF wrote:
> Strictly speaking, DMC meets SDSC. But it meets it in a weaker sense
> than do the single-mode methods SD, SSD, BeatpathWinner/CSSD, MAM, &
> RP.
>
> For comparing methods that use mixed balloting modes to single-mode
> methods, one could define "thorough" versions of the criteria, in
> which "vote(s) X over Y" is replaced with "thoroughly vote(s) X over
> Y"

I like "strongly prefer(s) X over y", though it uses an overloaded
term.

>
> Informally, a voter thoroughly votes X over Y if s/he votes X over Y
> with respect to each balloting mode that that the method in use uses
> to choose its winner.
>
> Probably a precise definition could be written. Maybe later, by me,
> or maybe by someone else.
>
> DMC doesn't meet SFC.

DMC doesn't meet your *new* *special* definition of SFC.  But that
definition isn't SFC.  You could create a new criterion,
SFC-multi-mode if you want.  DMC does satisfy SFC if viewed as a
Condorcet method.  Approval is used as the defeat strength and is
measured directly instead of being inferred as in RP(wv) or CSSD(wv).

>
> DMC, being a pairwise-count method, can benefit from ATLO, to avoid
> the co-operation/defection dilemma that James spoke of.

Most of the time ATLO is unnecessary with DMC's approval cutoff, and
no recount is required.  Have you found a way to make ATLO summable?
Otherwise ATLO/AERLO add expense to an election.

>
> Somone pointed out to me that James' co-operation/defection dilemma,
> the bad-example of Approval, is not as bad as the
> co-operation/defection dilemma known as the "prisoner's dilemma".
>
> We've been hearing a lot about wv not meeting LNH (Later No Harm). As I've
> many times pointed out on EM, the way that IRV protects your favorite from 
> being harmed by votes for your lower choices is, by elimnating your favorite 
> before it lets you help anyone else. I call that electoral euthanasia.
>
> But LNH does have one use: It's one way of avoiding James'
> co-operation/defection dilemma. But that problem can also be well
> dealt-with via ATLO, and so there's no justification for giving up
> what needs to be given up in order to get LNH, as a way of avoiding
> the co-operation/defection dilemma.

Ted
-- 
araucaria dot araucana at gmail dot com