[EM] Candidate Withdrawal suggested by Tom Ruen

[Instant Runoff Voting supporter]

- - - - - - - - - - - - - - - - - - - - - - - - - 12/23/00
Does the IRV process retain the spoiler effect?
by: Tom Ruen <tomruen at itascacg.com>

Does the IRV process retain the spoiler effect?
   Short answer: It doesn't have to - if we do it right! (Stay tuned!)
   Let me offer a short story.
Disclaimer: I use real names to increase reader involvement, but I make no
claims that the numbers, people, or preferences represent reality. This is
only a story!
    Pretend that the 2000 presidential election had been done by IRV. (No
more electoral college - hurray!)
   Let's look at candidates Bush, Gore, and Nader alone for simplicity, and
consider Nader was more popular than expected, and also that he got his
support exclusively from people who would otherwise have chosen Gore. Just
pretend for simplicity's sake, okay?
   All throughout the campaigning, Nader wavering supporters whine, "We
can't support you, Ralph. Although we like you best, Gore seems stronger in
the polls, and we don't want Bush to win."
  My somewhat nicer than real Nader says, "No problem friends, we've got
IRV now. Vote me first and Gore second. If I don't win, your votes will
transfer for Gore."
   Meanwhile Gore was not so nice perhaps and says to his wavering
supporters,  "I'm the best liberal candidate. Nader is extreme! I can beat
Bush - sure I can!"
   So election day comes and the liberal majority is split:
Ballots: First-second count
Bush-????  48%, Nader-Gore  27%,  Gore-Nader  20%,  Gore-(none)  5%
  In the IRV process: Round 1: Bush 48, Nader 27%, Gore 25%
  Amazingly Nader was more popular than Gore realized, and Gore is
eliminated for being last place!
  Go Nader! Hurrah!
  Round 2: Bush 48%, Nader 47% [exhausted 5%]
  Oh, but alas Nader loses because 5% of supporter for Gore that didn't
rank Nader second.
  IRV winner (by plurality): Bush!
  Now Nader supporters cry, "Hey, look! We [liberals] collectively have the
majority of the vote - Gore can still win! I thought our votes will be
transferred to Gore and then we can still pick a winner!"
  Bush says, "Sorry folks, too late! Gore is a 3rd place loser - he can't
have your votes now."
  Then nice Nader, who knows he can't win says, "Oh, sure Gore can have my
votes. I withdraw from the election and now Gore's second rank ballots are
available to him!"
  Bush demands, "Oh no you don't! The rules of IRV are final. Gore was
eliminated. Votes can't be transferred backwards!"
  Nader says, "Sure they can - I promised my supporters. Gore was
eliminated because of my supporters, not yours."
  Bush reacts with surprise, "Did he say 'Up yours?'"
  Nader says, "The man's delusional! I'm outta here - get those votes
recounted pronto!"
  Hundreds of lawyers jump from behind the curtains, chiming in unison,
"Here we come to save the day..."
  Well, you can imagine the courts will not be happy to decide this one for us!
  So let's decide now before anything like this really happens!

  Questions about IRV election process:
1. Will a candidate be allowed to withdraw from the election after the
"counting" is completed and the "winner" known?
2. Is IRV elimination final and irreversible, or if something changes (like
allowing #1), can a candidate demand a "recount" of the IRV process, hoping
for a raise in his or her standing?
  These questions may vital when the declared "winner" doesn't have a true
majority due to exhausted ballots. A different candidate might take the
lead on those lost exhausted ballots plus second ranks!
  My example is simple and contrived, but the effects are real. Something
like this can happen whenever a third party gain unexpected support.
  I would like discussion about the fairness of these questions in theory
and the practicality of applying them in real elections.
  Have these questions been asked before?
  If so, what were the answers?

Sincerely,
Tom Ruen

- - - - - - - - - - - - - - - - - - - - - - - 12/27/00
Dear Tom Ruen,

     I have noticed that you have been advocating allowing a candidate the
right to withdraw after an IRV election, but the withdrawal is to be
effective before the IRV election.
     The first time I seen this proposal was about three years ago on the
Election Methods list, posted by a Steve Eppley. At that time an effort was
made by Steve and a Mike Ossipoff to sell it to Rob Richie of CVD, but no
sale. Since then the method has been mentioned a number of times on the EM
list.
     The details are the same for both Steve's and your design, but then
there are not many details. Each of you may have created the method on your
own, I don't need to know which of you was first, or maybe you both are
reading the same author, or maybe each of you have reinvented the wheel,
but I digress.

     What is important is for you to know that candidate withdrawal is a
form of Approval Voting, a better form in my judgement. In the case of your
Bush - Gore - Nader example, the results will be exactly the same as
Approval Voting.
     Your example:  48 Bush  27 Nader-Gore  20 Gore-Nader  5 Gore
     Approval results:    48 Bush  52 Gore  47 Nader
     Withdrawal results:  48 Bush  52 Gore  47 Nadar
      (Nader did have 47 before he withdrew)

     I rate IRV-plus Withdrawal the best of the Approval Methods, better
than Bucklin, better than Borda aka Approval Count, and of course better
than Approval Voting, which I rate the lowest, but I do not advocate any of
these methods.

     I favor runoff methods and in a runoff election the votes are not
counted of people who do not show up for the runoff election. In your
example the five percent Gore voters chose not to vote in the runoff. We
can only count the votes of the people who do show up, in this case only
the 48 Bush and the 47 Nadar voters appeared at the runoff. Bush wins by a
majority, yes a majority of 48+47.
     Why the five percent did not vote is unknown, but I would say that
they considered Bush and Nader to be equally good or equally bad. It is the
same as if their second choices were divided equally between Bush and
Nadar, in which case Bush again will be the winner with 50.5, a majority.
(So, you see the system works.)

     One more comment: Because the withdrawal will only be undertaken if it
will change the outcome of the election, this decision should not be made
by a single person, who is playing King maker, not even the candidate
involved.
     If you must advocate candidate withdrawal, then I suggest that the
withdrawal should be automatic with the election system. This is possible
by a redesign of Bucklin. The results produces by the redesign will come
very close to candidate withdrawal results. The method would work as
follows:
   1) If there is a candidate with a majority of the first choices, he is
the winner and the election is over.
   2) If no majority winner, then we add the second choices of all the
lower candidates to the first choices. This is about what you are trying to
do with candidate withdrawal. You are trying to collect enough choices,
among the lower candidates, to overtake the leading candidate.
   3) If there is a majority of the total voters, then we have a winner.
(note: I did not say a majority of the total choices being used)
   4) If no majority winner, then we add the next choices of all the
current lower candidates to the totals. (note: we may have a different
leading candidate at this time)
   This routine continues until we have a winner with a majority of the
number of voters. If we have more than one with a majority, the greater
majority is the winner.
   This method will give the results that you wish:
           48 Bush  52 Gore  47 Nadar
   We need a name for this method. How would you like to have your name on
an election method??   Let's call it:  The Ruen Approval method. (your 15
minutes of fame)

   I would rate the Ruen Approval method better than Candidate Withdrawal,
of course the method needs to be tested with a number of examples, you can
do that, after all the method does have your name on it.

My Current Standings of the Approval Methods:
     1) Ruen Approval
     2) Candidate Withdrawal
     3) Bucklin
     4) Borda Count aka Approval Count
     5) Approval Voting

Regards, Donald Davison - Host of New Democracy,  www.mich.com/~donald

   +=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+
   |                         Q U O T A T I O N                         |
   |  "Democracy is a beautiful thing,                                 |
   |       except that part about letting just any old yokel vote."    |
   |                            - Age 10                               |
   +=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+