[EM] Reply to Tom's IRV w/ties of 03/19/01

[Tom Ruen]

Don,

Thank you for your kind reply, but in the end, you are not quite right.

Don said:
>      Anytime you have two top candidates that you like equally, that will
> not be a problem with Irving.  You merely chose anyone of the two as your
> first choice and the other as your second choice, it will make no
> difference. Also, it will make no difference which of the two will be
> eliminated first, your full vote will automatically go to your other top
> candidate.

We agree that when one choice of a ranked-tied of two choices on my ballot
is eliminated, the other will gain my full vote. The problem is that my vote
in ranked IRV affects which will win. I may not WANT my vote to influence
which will win (between my favorites).

If my most favorite has more first rank support than my second favorite, but
doesn't gain enough second rank support from my second most favorite's
supporters, who will be the eliminated candidate, to win, then voting my
most favorite becomes a dangerous choice since both may lose.

In IRV, with 3 strong candidates, I am sometimes encouraged to vote
insincerely for my second most favorite (my compromise), just like people
now vote for their compromise among the expected top-two in plurality.

In IRV, voting with a tied-rank means I'm afraid of voting fully for my
favorite and without this option, I would compromise fully and vote for my
second favorite. Tied ranks is a way that I can support my most favorite
without helping him to win over my second most favorite.

Our election system is centered around 2 major parties and a number of minor
parties. IRV won't change this. People who are voting among the usual
top-two parties have little incentive to vote after their favorite. People
voting for a 3rd party candidate are VERY interested in considering support
for a second choice since they expect to be eliminated.

If I am a supporter of a 3rd party candidate, and my candidate rises (in the
polls) to near second place (catching up to a major party candidate I would
pick next), I must seriously consider if my candidate will gain the transfer
votes to win the election. 3rd party voters can't make this assumption,
especially if the 3rd party candidate is further from the political median
than the major party alternative. Most 3rd parties will be further from a
popular center since major parties are more interested in winning than
standing up for hard (unpopular) issues.

Within this situation I have to make a decision - do I vote 3rd party and
risk my worst choice to win, or compromise and hurt my 3rd party choice.
(Again, sounds like plurality, doesn't it!)

In this situation, I would give voters a better choice - to vote a tied
rank. A tie isn't quite sincere if I prefer one choice over the other, but a
tie is more sincere than voting for my compromise with a top rank.

I've compared this in my mind tied-votes in a (non-instant) runoff, like in
party endorsement elections. Party endorsement elections don't have forced
elimination each round. Therefore I can safely vote for my favorite once and
then if my favorite is not much ahead of my compromise in last place, I have
a chance to move (down) to my compromise who I believe will have a better
chance in the end to win. IRV, with forced elimination, doesn't allow me to
CHOOSE when and where to compromise. Therefore I consider tied-ranks as one
compromise which represents my indecision more sincerely.

This is one situation where offering a tie is a better choice to voters than
voting fully for a compromise. This is a strategic example where it can have
meaning for me as a voter.

Other voters may not have this strategic reason, but may also be interested
in voting for two choices equally.

The ranked ballots handle this choice (as Don's example shows). The meaning
and intent of this choice is clear to voters and to vote counters. Some
voters will choose this as most what they WANT. Under IRV these ballots will
be spoiled despite having clear voter intent.

This is why I support tied-ranks - because we CAN consider voting this way,
and some may WANT to vote this way.

As long as we use forced elimination, and I can't transfer my votes to my
second favorite after knowing my favorite can't win, I will support this
choice for voters.

Tom Ruen

----- Original Message -----
From: "I like Irving" <donald at mich.com>
To: "[EM]" <election-methods-list at eskimo.com>
Sent: Thursday, March 22, 2001 9:05 AM
Subject: [EM] Reply to Tom's IRV w/ties of 03/19/01


> - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 03/22/01
> Greetings List,
>
> Tom Ruen wrote:
>
> Tom:  "Approval is N one-vote elections for N candidates. Every voter gets
> up to N total votes."
>
> Don: Yes Tom, that is correct, Approval is the same as Plurality-at-Large,
> but that is nothing to brag about.
>      The only groups that should ever use Approval Voting are `Mutual
> Admiration Societies'.
>
> Tom:  "IRV penalizes people for voting for a favorite first who can reach
> second place but can't win."
>
> Don:  In order for your favorite to win he needs to get the votes.  Irving
> is not some scheme for bypassing that requirement.
>
> Tom:  "I would be happier with IRV w/ties than without, so long as my vote
> becomes a full vote for one after my other choice is eliminated. Divided
> tie votes work in IRV because only one choice is eliminated at a time so
my
> vote is guaranteed to transfer. A divided vote in a single round plurality
> election is not useful since both choices can lose without my full
> support."
>
> Don:  Why must we be concerned if you are happy?
>       If we must, we must. Let's see if we can design an IRV w/ties ballot
> and the math that will make Tom happy:
>
>          Candidates           Choices
>       - - - - - - - - - + - + - + - + - + - + - +
>                         | 1 | 2 | 3 | 4 | 5 | 6 |
>       - - - - - - - - - + - + - + - + - + - + - +
>              A          | X |   |   |   |   |   |
>       - - - - - - - - - + - + - + - + - + - + - +
>              B          | X |   |   |   |   |   |
>       - - - - - - - - - + - + - + - + - + - + - +
>              C          |   | X |   |   |   |   |
>       - - - - - - - - - + - + - + - + - + - + - +
>              D          |   |   | X |   |   |   |
>       - - - - - - - - - + - + - + - + - + - + - +
>              E          |   |   |   | X |   |   |
>       - - - - - - - - - + - + - + - + - + - + - +
>              F          |   |   |   |   |   |   |
>       - - - - - - - - - + - + - + - + - + - + - +
>              G          |   |   |   |   |   |   |
>       - - - - - - - - - + - + - + - + - + - + - +
>
>      This proposed ballot for Tom will be recorded in the computer as:
>
>               0.50 ABCDE  and  0.50 BACDE
>
>     Please notice that Tom's two `tied' candidates are both in the top two
> ranks of each half of a vote.  This will allow his two half votes to
become
> a full vote for one of his top candidates in the event, God forbid, one of
> his top two candidates gets eliminated.  This should rest Tom's fears.
>
>     Let's suppose one of his top two candidates is eliminated.
>     That will give us the following two possible results:
>
>           1.0 ACDE    or   1.0 BCDE
>
>     His full vote does end up on one or the other of his top two
> candidates, all's well that ends well.  This should make Tom happy.
>
>      Just for the record, we should check and see what end results we will
> get when we use plain Irving.
>      Tom will be forced to vote either ABCDE  or  BACDE, poor Tom.
>      Again, let us suppose one of his top two candidates is eliminated.
>      That will give us the following two possible results:
>
>           1.0 ACDE    or   1.0 BCDE
>
>      My my, we seem to have the exact same results, there is no
difference.
>      Tom's insistence of IRV with ties is nothing more than unnecessary
> rigmarole, which is par for the course on the Election Methods list, but
in
> a real election in the real world, we don't need rigmarole.
>
>      It's like I keep saying, there are going to be people who will insist
> that they can improve on Irving, but they always seem to come up short,
> Irving is still the best method.
>      Anytime you have two top candidates that you like equally, that will
> not be a problem with Irving.  You merely chose anyone of the two as your
> first choice and the other as your second choice, it will make no
> difference. Also, it will make no difference which of the two will be
> eliminated first, your full vote will automatically go to your other top
> candidate.
>     But you must accept the fact that at least one of your top two
> candidates will be eliminated, sorry but that's the way it is.  I know
it's
> hard, but you must let go.
>
> Donald Davison