"falsification won't change outcome"

[Hugh R. Tobin]

Mike Ossipoff wrote:
> 
> In one posting, Tobin used my statement that falsification won't
> happen on a scale sufficient to change the election result
> to mean something other than what I'd meant by the statement:
> 
> I meant literally that falsification won't be widespread enough
> to change an election result. But Tobin was saying, in 1 letter,
> that my statement about that confirms his claim that the
> order-reversing Dole voters will never be able to make Clinton
> more beaten than Dole.
> 

I did not claim that, nor did I make many of the other assumptions or 
assertions attributed to me in commentary on earlier postings.  Indeed, 
I have argued that the possibility of successful order-reversal by a 
plurality and the absence of a convincing defense to that strategy is a 
non-trivial issue for Condorcet, in part because polling data might 
enable the plurality voters to predict a high probability of success.  
But voters are always dealing with uncertainty.  Nobody can be sure that 
an order-reversal strategy will or will not succeed, or that a natural 
circular tie will or will not be created.  My point was simply that
faced 
with this uncertainty, even if it seemed highly likely that the Dole 
voters would make Clinton "more beaten" than Dole, a Clinton voter who 
was indifferent as to whether Dole or Nader would be elected if Clinton 
were not, would want his equal ranking of those two candidates counted
as 
one-half vote for and against each, not as zero each way.  This is 
because there would be some chance that the half-votes of that voter and 
other like-minded voters would make Clinton least-beaten in case of a 
circular tie. (I have discussed elsewhere the strategic options to which
the Clinton voter should resort if not permitted half-votes.) 

> No, I said that the order-reversal won't affect the election
> at all, not that it merely won't make Clinton more beaten
> than Dole.
>

In my example, the order-reversal by a minority of the Dole voters, or 
their true preference for Nader over Dole, was enough to create a 
circular tie but not enough to change the election result if the 
half-votes were counted for the Clinton voters (who chose no second 
preference).  I think it illustrates that the basic thrust of Ossipoff's
statement would be correct in a slightly greater range of possible 
distributions of votes, if the half-votes for equal rankings were 
counted under the rule I proposed. 

> If any kind of defense against the order-reversal is needed,
> including the falsification strategy that Tobin suggests
> for the Clinton voters, that means that the order-reversal
> has happend on a scale sufficient to change the election
> result, which is what I say won't happen.
> 

I did not suggest any "falsification" strategy for the Clinton voters. I 
give Mr. Ossipoff full credit for educating us all about the potential 
for order-reversal strategy and for proposing the truncation deterrent 
(which is a falsification strategy in his terminology, in which
truncation implies hiding a true preference, though I did not use
"truncation" in that sense). 

> If, on the other hand, it's necessary to use strategy against
> the order-reversal, because it's widespread enough to change
> the election result (by making Clinton be beaten), then it
> very much runs the risk of backfiring for the Dole voters
> and electing Nader, if the Clinton voters know what they're
> doing.
>
The Dole voters' strategy is certainly risky if they prefer Clinton 
significantly to Nader, which is why it may be difficult to get enough
of 
them to participate, though there may be a minority who regard Clinton
as 
almost as bad, and therefore are willing to take the chance (if they 
think it is the only way that Dole might win).  The Dole voters even run 
a risk of electing Nader outright, as false Condorcet winner, if they 
have miscalculated.
The question, I think, is not whether the Clinton voters know what they 
are doing but what their preferences are.  It is possible that the mere 
fact that Dole supporters' attempt to create a circular tie in order to 
win the election would cause Clinton supporters to prefer a Nader 
election to a Dole victory when previously this would not have been
true. 
 But this is not necessarily the case, and one cannot assume they would 
falsify their preferences at the polls, with no benefit (and possible
harm) to their own candidate, just on the principle that voters who
attempt to manipulate the Condorcet system must be punished. 
 
> For that reason, since making Clinton more beaten than Dole
> is an essential part of the Dole voters' falsification strategy,
> something without which it wouldn't be a good idea to falsify,
> then the Clinton voters are making a mistake if they assume that
> the Dole falsification, while widespread enough to change
> the election result, isn't going to succeed in making Clinton
> more beaten than Dole. It has to, or it shouldn't be done in
> the first place.
> 

I agree that one should not try something that is not going to work.
This 
is true not only of the order-reversal strategy, but also of the 
candidacy itself!  Why run if you aren't going to get enough votes?  Why 
organize strategic voting if you can't get enough supporters to go
along?
In hindsight many electoral gambits look foolish (no offense to Steve 
Forbes).  The risks that order-reversal strategy would fall short and 
leave Clinton as Condorcet winner, or as least-beaten in a circular 
tie, are more tolerable than the risks discussed above, because Clinton 
probably would win if no order-reversal were tried.  So if Dole has set 
the strategy in motion I would not assume he would call it off at the 
last minute just because he is not sure that enough of his voters will
go 
along with it to make Clinton more-beaten.

My point, again, was that the circumstance in which the Dole strategy 
falls short of making Clinton more beaten by Nader than is Dole by 
Clinton is the only case in which the Clinton voter who sincerely 
declined to rank Dole and Nader could be made better or worse off by the 
method of counting equal rankings for the tiebreak.  In that case, the 
half-vote against Nader could win the election for Clinton.  Therefore, 
if one accepts that a voter should be allowed to express equal dislike
of candidates without forfeiting one of his alloted "votes against" in
the case of a circular tie -- i.e., that half-votes should at least be
an option -- then in order to make a case for even allowing the Clinton
voter an option to have his equal rankings at the bottom of his ballot
count zero each way, one must establish the value of strategic
truncation (i.e., falsely ranking candidates equally when one really
prefers one over the other), and explain why the truncating voter would
not want half-votes counted.

-- Hugh Tobin

> Mike
> 
> --