[EM] What is the ideal election method for sincere voters?

[David Cary]

--- Juho <juho4880 at yahoo.co.uk> wrote:

> On Mar 9, 2007, at 1:12 , David Cary wrote:
> 
> > I describe a sincere vote as one which reflects as accurately as
> > possible the person's personal preferences.  The preferences are
> > ones that are independent of the election method
> 
> Although the preferences may be independent of the election method 
> the description of the method may have impact on how one votes. You
> discussed the Range vs. Approval case. In Approval voters could be 
> requested/advised/recommended/guided 1) to mark candidates that
> they  approve or 2) to mark one/some of the front runners and
> candidates that they prefer to these. Are both cases sincere 
> according to your definition?

Probably so.  As long as the approval vote preserves non-strict
ordinal preferences (Y is at least as preferred as X and X is
approved, then Y is also approved), they would be sincere based on
the description I gave and the normative specification I have for
approval voting.

> > , independent of information about how other voters will vote
> 
> My second approval "advice" above referred to the outcome of the  
> election and the anticipated voting behaviour of other voters. So  
> also that may have some impact.
> 
> It is also possible that a voter changes his sincere opinion of a  
> candidate when he learns that that candidate has wide support. I'd 
> say this is still sincere also in the "make your decisions without 
> considering how your vote impacts the outcome of the election"
> sense.
 
My description of a sincere vote did not preclude a voter using
preferences formed from knowledge about how the various alternatives
would impact others, how others might generally react, or even what
the preferences of others might be.   My description only excluded
knowledge about how others might vote, including which other people,
if any, would vote.  I know that is a thin line of distinction to
make, but it captures best what has traditionally been denoted by a
sincere vote.

My description of a sincere vote is dependent on what is taken to be
the design intent or normative specification.  Some highly strategic
normative specifications would not be valid to the extent they
require a voter to consider how others will vote, for example approve
1 of the 2 front-runners.  Other normative specifications introduce
an element of strategy without directly requiring voter knowledge of
how other voters would vote, for example, approve those candidates
with above average personal utility.  I’m not sure to what extent
that is a loophole or just a feature, or if there is a good way to
eliminate that problem.  Approval and Range both have reasonable
non-strategic normative specifications.
 
Of course in a broader and perhaps more real sense, any vote
voluntarily cast is a sincere vote, no matter how much strategy might
have gone into calculating that vote.  In that sense, it is a
misnomer to call a strategic or tactical vote an insincere vote, and
especially to call it a dishonest vote.

> In Condorcet corresponding recommendations could be 1) to rank all 
> candidates in the order of preference or 2) to check the guidance  
> given by the experts of their party / candidate just before the  
> election day. I wrote the second case intentionally vague since to 
> this day I have not seen any clear generic guidelines on how to
> vote strategically in Condorcet elections. (Well, to be exact there
> is at least one quite generic strategy, to rank the (equally bad
and  
> equally strong) candidates of a competing party in the given order 
> but starting from a random candidate and continuing from the  
> beginning when the last candidate is reached. This way the  
> probability of a loop among them increases.)

The challenge is not finding generic ways to vote strategically, but
finding ways to vote that are in some sense strategically optimal. 
Many so-called counter-examples presented on this list are really
just illustrations that it is sometimes easy to use sub-optimal
strategy.  The optimal rules are likely vary depending on which
Condorcet method is being used.

> If the second definitions of the election methods are used, then
> the elections become more like games that all voters may play. A 
> good strategy brings good results. My question is, when does 
> "sincere voting" stop and when does "strategic voting" start in 
> this scale?

I distinguished between a sincere vote vs. sincere voting.  When
there are multiple sincere votes available to a voter, using
strategic considerations to choose which sincere vote to actually
cast would not be a form of sincere voting.  The linquistic
distinction is perhaps too confusing. 

Generally speaking, strategic voting will overlap sincere voting only
to the extent that the normative specification is strategic.  But if
a non-strategic normative specification is used, strategic voting can
mix with casting a sincere vote when a voter has multiple sincere
votes to choose from.

-- David Cary


 
____________________________________________________________________________________
Expecting? Get great news right away with email Auto-Check. 
Try the Yahoo! Mail Beta.
http://advision.webevents.yahoo.com/mailbeta/newmail_tools.html