Condorcet, Thin Ice, and Battle Cries

Hugh Tobin htobin at
Sun Apr 27 19:16:53 PDT 1997

John De Lasaux wrote:
> At 12:40 AM 4/27/97 -0700, htobin at wrote:
> >John De Lasaux wrote:
> >>
> >
> >> =======================
> >> The *only* way that voters would rationally mark his second or lower
> >> preferences is if they are *guaranteed* that their lower prefences don't
> >> influence the outcome of the count until their first preference is
> eliminated.
> >>
> >
> >Please demonstrate why this is so, taking into account that in Condorcet
> >truncating one's ballot tends to reduce the chances that one's first
> >choice will prevail, as well as to increase the chances that one's least
> >favorite will prevail.
> COMMENT============
> I guess you are presupposing that a voter would mark his favorite choice at
> all levels of preference, else how could truncating his ballot *reduce* his
> favorite's chances? 

Note that I requested a demonstration.
In Condorcet, truncation can reduce the favorite's chances by making it
more likely that, in case there is no Condorcet winner, the voter's
third (or lower) choice will be "least beaten" and therefore win the
tiebreak.  By voting for B over C, the supporter of A not only reduces
the chances that C will be the beats-each winner, but also increases the
chances that C will be more beaten than A in case of a circular tie. 
Suppose my wife and I support A, but prefer B to C.  The election method
is Condorcet with the "margins-of-defeat" tiebreaker.  Suppose that the
votes of the other voters are such that, without our votes, A will beat
B by 1 vote, B will beat C by 1 vote, and C will beat A by 4 votes.  If
we both vote our true second preference (A,B) then A will win the
election (least beaten, by C by 2 votes); if we truncate, then our least
favorite, C, will win (still beaten only by 1 vote, by B).  (Of course,
on different facts it is also possible that B would be least-beaten as a
result of our second-place selections, but the question here was how the
second choice could help A). A similar example could be given with the
"least votes against in worst loss" tiebreaker; indeed that system is
harsher on truncation.

Incidentally, I think marking the same candidate at a both a higher and
lower level of preference would result in the all rankings of that
candidate except the highest being disregarded.  There could be other
rules, such as counting the ballot as invalid, but I think the rule
stated would most likely correspond to the confused voter's intent,
especially if no other candidate is ranked at all.  Thus truncating and
selecting the same candidate for all rankings would be the same.

> I thought the idea of preference voting was to give 
> voters the chance to pick a second or third choice in the event their first
> choice lost in an eary round. Is that true, or not?

No. I think the basic purposes are to give the voters the right to vote
all of their preferences among the candidates, without pressure to
discard their first choices in favor of a lesser evil, and to elect the
candidate, if any, who is preferred over each other candidate by a
majority of those who have a preference between the two. (If there is
such a candidate, why should we accept the election of someone else,
i.e. someone who loses to him or her?)  
As I mentioned, if you think it essential to have a first round in which
only first choices count, then you may consider that the first round in
Condorcet is to determine whether anyone has a majority of the first
place votes, and that only if nobody does, do the lower choices count. 
Adding this to the description of Condorcet makes no change in the
system and makes the description longer, but if it would help people
feel more comfortable with Condorcet, then perhaps it should be
considered.  However, it is not a question of one's lower choices
counting only if one's first choice has "lost" in that first round;
rather the the second and lower choices of all voters figure in
determining whether there is a Condorcet winner (which I sometimes call
a "beats-each" winner). Only if there is no Condorcet winner do we have
elimination, in Smith//Condorcet, by limiting the tiebreaker to the
members of the Smith set.  One can then say that those whose first
choices "lost" (to each member of the Smith set) have their lower
choices used in a later round to help select within that set, but this
is not the first use of those lower choices.
The Condorcet system does not fit neatly into your way of thinking of
decision methods.  Please consider that this does not necessarily mean
that Condorcet was wrong and you are right.  I highly recommend that you
read his Essai, or at least those portions directly addressing the
choice of election methods.

> ==================
> They will tell their supporters to "vote all the boxes for X to make sure
> your favorite wins". The "smart" voters will prevail, and the value of any
> sophisticated voting system will be lost.

No, in Condorcet this would be stupid advice, and could lose the
election for X.

> ==================
> >Neutral experts will tell voters, truthfully, that truncating is not
> >smart strategy to elect their favorites.  So why assume the voter will
> >fixate on the absence of this type of a "guarantee" to the point of
> >acting against his own interest?
> >
> >In instant run-off, there is no guarantee that casting a first-place
> >vote for one's first choice, instead of one's last choice, will not
> >cause the first choice candidate to lose.  So nobody would vote at all?
> ===================
> Why do you think we have such low turn-outs now? Politicians blame it on
> "voter apathy", as though it's the voters' fault ... they're "lazy" and they
> "don't care".

You do not respond to the point, but I agree with your new one
entirely.  It is not the voters' fault that they are given such limited
choices, in a system where voting for a minor candidate wastes one's
vote, while voting for the lesser evil tends to increase the "mandate"
for someone the voter regards as a fool or a crook.  Condorcet is an
excellent solution, as it allows the disaffected voter to vote for
numerous minor candidates above those from the major parties, but still
choose the lesser evil between them.  Unlike instant runoff, Condorcet
also will not pressure those who do support major candidates to cast
insincere first-place votes.
> The system has to clearly stand on its obvious merits (in the voters' eyes),
> and it must be immune to fallacious explanations by "spin doctors".

But it need not be immune to cynical tactical manipulation, as may be
encouraged by instant run-off?  And it need not guarantee that it is in
the voter's interest to name his or her first choice first?

It seems your argument boils down to the notion that the people are too
stupid to have  a truly democratic system.  Of course, similar arguments
have been made against having republican government at all (the issues
that the voters must decide in an election may be infinitely more
complex and difficult than the mechanics of Condorcet).  But as one of
the people, I cannot accept this notion.


Hugh Tobin

> John D

More information about the Election-Methods mailing list