[EM] Re: Thresholds, pairs approval

[Kevin Venzke]

Chris,

I'm thinking a threshold may not even be desirable if the first-preferences
are counted double.  Just this would convince some voters to raise second-
preferences to a tie in first, to express explicit "approval."

But if we did have a threshold, how about one based on the most votes received
by one candidate?  That is, with a threshold of 1/2, if A gets 100 first-place
votes, and B only gets 40, B can't win even if he's the CW.  There's a disaster
scenario here, too, but voters can use strategy (raising) to avoid it.

 --- Chris Benham <chrisbenham at bigpond.com> a écrit : > Kevin,
> Here is a maybe tolerable method:  
> Have an approval cutoff  in the rankings. Default cutoff  is between 
> first and second. Eliminate candidates who do not make a low  Approval 
> threshold which is related to the number of candidates. Among remaining 
> candidates: if no CW elect the most approved Smith set member.
> This way a voter who wants to give first preference to  a sure loser 
>  and  second prefernce to a maybe CW, can place approval cutoff between 
> second and third and so help second preference  make the threshold.

Could work.  My fear is still that an approval cutoff might be
under-utilized.

> Regarding Forest's  pairs approval method, I think that in practice it 
> might be that not all of  the major parties will field  2 candidates and 
> that if they do the voters might react against it  so that the most 
>  approved pair  might often not be 2 candidates from the same party.

I think this is probably not right.  Any party with a chance at having a
plurality is foolish not to field 2 candidates: If they don't, they could
easily lose on the ranked portion, when they had a chance at dominating it.

Which voters are going to react against it?  If party A runs two candidates,
A's supporters are silly not to vote for that pair.  The only thing every
other party can do is agree on a different pair to vote for.  This probably
amounts to agreeing on a candidate.

> But 
> even so  giving voters a choice between candidates of the same party and 
> getting all the sincere preferences (or even ratings) on the ballot (and 
> presumably strictly complying with Majority Favourite) to me makes it 
> better than plain Approval.

Approval permits a choice between candidates of the same party.  (Anyway, in
Forest's pairs approval, I doubt the "mandatory second candidate" would even
campaign.  His job is to lose.  It would be amusing, though, if the other
parties decided to rank this guy above the serious candidate.)

It might not strictly comply with Majority Favorite.  Conceivably, the
majority could miscalculate so badly on the pairs approval ballot that the
majority favorite doesn't even end up in the winning pair.

Kevin Venzke
stepjak at yahoo.fr


___________________________________________________________
Do You Yahoo!? -- Une adresse @yahoo.fr gratuite et en français !
Yahoo! Mail : http://fr.mail.yahoo.com