[EM] 02/06/02 - Adam's 02/02 example of three equal candidates:

Wed Feb 6 05:23:30 PST 2002

```>I have read your text on IRV in the 02/02 letter a number of times and it
>appears that the demon you fear is that the people will vote in a circular
>pattern with their lower choices while voting equally for all three
>candidates with their first choices, a rarity in a rarity.

That's not really what the example was about.  I assumed a
right,middle,left candidate (Bush, Gore, and Nader respectively in the
example).  Gore voters broke both ways in their second choice, but not Bush
or Nader voters.  The only time I mention a Bush or Nader voter voting for
the other is at the end when I mention that a Bush voter could vote for
Nader FIRST in an effort to win the election for Bush.

The example is not that far-fetched.  The only thing about it that was
"rare" was that all three candidates were very close to one another in
first preferences.

>This same circular pattern would cause a Circular Tie in the Condorcet method.

No.  In Condorcet, Gore would be the COndorcet winner by a wide margin.

>Getting back to your demons, I see your dilemma as being that you wish to
>vote for the one candidate of your two favorites that will receive the
>most second choice votes from your other favorite, but you don't know
>which way the circular pattern goes, clockwise or counter clockwise.

While I again reiterate that it's not circular (the second choices go in to
Gore or out from Gore) the solution by this argument is obvious... everyone
in the election should vote for Gore with their first choice, to hedge
against the candidate they like even less.  This is called defensive voting
or the lesser of two evils problem.

>The pre-election polls would need to be asking people if they plan to vote
>AB or AC or BA or BC or CA or CB.  From this information you should be
>able to decide which of your two favorites to rank first and second.  If
>your two favorites are B and C, and most B voters will be voting BA while
>most C voters will be voting CB, then you want to vote for B in hopes C
>will be eliminated first and send a large bloc of votes to B.

You make it sound like I have two even favorites.  (If this is the case,
wouldn't Approval be grand?)  What If I support Bush, and I find that all
sincere Bush voters are putting Gore second... by your logic I should vote
Gore first in order to send Bush votes to Gore.  Classic lesser-of-two-evils.

>You must realize that most other people will be looking at the polls in
>order to decide which of the top two candidates to vote for.  This will
>drop one candidate to a distant third

I agree the 3-way tie rarely happens in IRV.  IRV maintains the two-party
hold on things very effectively.  See Australia or Malta for proof.  On the
other hand, if IRV ever did allow a true multi-party situation to arise,
situations like this would be common,

>In the cases of the ABC methods, Approval, Borda, and Condorcet, you
>should only vote for your most favorite.  Let the other voters support your
>favorite with their lower choices while you will not be supporting their

In Condorcet, my second choice can't hurt my first choice, except in a far
rarer situation than this (the circular tie, which you correctly recognize
as a "rarity within a rarity").

In Borda, you are right that your second choice can hurt your first, but
leaving it off is not an effective way to combat this.  The best way is to
vote a sure loser in second and your real second place in third.  I agree
it's not a good method.

In Approval, your first may hurt your second, it's true, but the effect of
this is that the compromise candidate wins in Approval elections nearly
every time.  In the case of the two candidates that are far ahead (which
you imply is the only realistic one in IRV)  then this is not a major

>IRVing will be true to the rules of preferences, which are:
>
>One, The changing vote rule:  "Lower preferences shall only be used to
>allow a voter to change his vote in whole or in part."
>
>Two, The Golden Rule:  "A lower preference is not to harm nor help an
>earlier preference."
>
>Three:  The wine rule:  "We will use no preference before its time."

These are not true properties of voting systems as much as they are
attributes of IRV.  True properties should focus not on the methods of the
election but on the outcomes of the elections.  Example: if a majority
prefers candidate A to all others, can they make sure candidate A wins?