Getting past the falsification issue

[Hugh R. Tobin]

Mike Ossipoff wrote:
> 
> First, when I use the term "truncation", I mean voting a short
> ranking that doesn't express all of the preferences that one
> has. I don't call it truncation if one really has no preference
> between one's unranked alternatives.
> 
> Tobin said that the Dole voters wouldn't be deterred from
> falsification unless they have a preference between Clinton
> & Nader. True enough. If there's a natural circular tie,
> and no Condorcet winner, then that could be true. But anything
> that only holds when there's no Condorcet winner only holds
> under less important conditions. What happens when each
> alternative is beaten, by the sincere rankings, is less
> important than what happens when there's a Condorcet winner.
> 
> But I'm not saying that what happens in a "natural circular
> tie" isn't important: It's important, I claim, because
> majority rule can still be upheld or violated, and the
> lesser-of-2-evils problem can still be gotten rid of or
> not gotten rid of--depending on whether we choose to
> use compulsory falsification or default falsification, or
> whether we do neither, and let non-falsification be the
> default assumption. To me the considerations that I named
> in this paragraph are important. I can't expect everyone
> to agree with me on that, but I believe that most do.
> 
> But if Tobin doesn't consider those considerations important,
> then what does his version offer to the voter, in terms
> of standards or criteria? Is it something important enough
> to justify giving up those considerations?
> 
> ***
> 

I confess I do not understand the objections.  One would think I 
had advocated monarchy rather than a tweak to the tiebreak 
system.  Surely majority rule is not compromised by counting equal 
rankings as one-half vote each way under my proposal.  Nor have I seen 
any demonstration that voters would somehow be coerced into voting other 
than their true preferences.  On the other hand, not counting equal 
rankings at all in the tiebreak provides an incentive to vote false 
preferences as between two equally disfavored candidates, when one is 
actually indifferent. 

The LO2E standard, if I understand it, seems to me a special case of a 
broader concept.  LO2E seems to refer to the ability to vote one's true 
first choice, without wasting one's vote or impairing the chances of a
second choice over a lower one.  (I like Lanphier's recent
statement of the principle and agree with Eppley that it cannot be fully
implemented under Condorcet or any system, per Arrow).  I think it is
also important that, to the extent possible, the selection, or not, of
lower rankings should not adversely affect one's first place vote. 
(Again, we cannot assure this in all cases, of course.)

To the greatest extent possible, the voter should be able to vote all
his 
true preferences, and non-preferences, without having to understand the 
nuances of the system or to make tactical decisions, and the system 
should take full account of the preferences indicated on the ballot,
without under-weighting or over-weighting the votes of any individual
(one-person, one vote).  

> Another thing: As soon as you count preference votes that
> were never voted, you're parting ways with preference counting.
> And when you do that, there's no longer any real justifiction
> to continue to use the rankings.
> 
> What we've then got is a point system: You can say that it's
> to the advantage of the anyone who'd indifferent to the
> other two candidates other than his favorite to give them
> both bad-points. Fine. A points system, but not a preference
> count.
> 
> But why limit it to half prefernces. Why not give everyone
> but your favorite a _whole_ bad-point?

Perhaps this question is only rhetorical, but I'll bite.  A feature of 
the Condorcet system is that it uses "votes-against" in the tiebreaker
as 
a measure, I believe Steve Eppley said, of how "unhappy" the voters
would 
be with a particular victor.  If we believe that the unhappiness of each 
voter counts equally, then each voter should have the same number of 
votes-against to cast: one for each pairwise race.  Allowing a voter to 
cast one vote against each would tend to give undue weight to the 
unhappiness of the truncating voters, compared to one who made a choice 
between two evils.  It would exaggerate how "beaten" was the loser as 
between the two equally ranked candidates, and thus would give an unfair 
advantage to the favored candidate of those voters casting the equal 
rankings.  It would thereby tend to discourage voters from voting their 
true preferences, for they would be throwing away votes-against that
could help their higher-ranked candidates in a tiebreak.  Counting
one-half vote each way is consistent with the egalitarian idea that each
voter gets one unit of unhappiness to allocate in each pairwise contest,
and comports with the common-sense notion that two voters who rank a
candidate equal second/third represent no more or less opposition to
that candidate than one who ranks him second and another who ranks him
third.  Thus, when the equal ranking is below the other Smith set
candidate(s) on the voter's ballot, the half-votes give effect to the
voter's preferences to the full extent that he is entitled to have them
counted.  Of course, in that situation the counting of half votes
involves no falsification of preferences.

I thought it was already conceded that the voter at least should be
allowed to choose half-votes when he ranks candidates equally.  If so,
open questions include whether half-votes should simply be counted in
all cases; whether it should matter if equal rankings are at the top or
bottom of a ballot; whether zero should be an option; and what should be
the default if options are allowed.  I have suggested, in the interests
of simplicity, a system that counts half-votes in all cases where we can
be sure that a sincere, well-informed and rational voter would choose
them if they were an option.

-- Hugh Tobin

> 
> Mike
> 
> --