Bucklin

[Mike Ossipoff]

Though I don't advocate discussing un-proposed methods, I should
maybe say a little about Bucklin. Bruce & I have both said that
it's the best method for electorates who want to big hand-counts,
and want the hand-count to be easy. Of course the U.S. electorate
isn't such an electorate, which is why neither of us is proposing
Bucklin as the method to propose to ER.

Bruce, as I understand him, prefers a version of Buckling different
from the one that was actually used. I forgot to mention: Bucklin
was used in this country earlier in this century, in several U.S.
cities, including San Francisco. Considering that computers weren't
available at that time, Bucklin was a good choice. I consider the
traditional Bucklin, the one that was actually used, to be the
better version. Let me describe it in English, as opposed to
formulese--just as a matter of curiosity, since it's not something
that anyone's proposing for our recommendation:

Voters rank any number of candidates they wish in order of preference.

Each ballot gives a vote to its 1st choice.

If any alternative has a majority then it wins.

If no alternative has a majority then each ballot gives a vote
to its next choice.

If any alternative has a vote total equal to at least half of the
number of ballots, then it wins. If not, then each ballot again
gives a vote to its next choice.

So, repeatedly, each ballot gives a vote to its highest ranked
alternative to which it hasn't yet given a vote, and this continues
till either 1 or more alternatives have a vote total equal to at
least half of the number of ballots, or till every ballot has
given a vote to everything that it lists, whichever happens 1st.
At that time, the alternative with the most votes wins.

***

Actually, of course, that last paragraph tells the whole story
of the Bucklin count, more briefly than the preceding wording.

I don't know about you, but I consider that clearer than formulese.

***

The price of having an easily hand-counted method is, of course
that Bucklin doesn't get rid of the need for defensive stratgegy
as well as does Condorcet. But it does a better job of it than
Copeland or Regular-Champion.

Bucklin can be considered an improvement on Approval. Bucklin
guarantees that a majority preferring A to B can ensure that
B won't win, without ever having to vote A equal to or over
anything that they prefer to A. 

That can't be said of Regular-Champion, or any other version of
Copeland.

Of course part of what that majority would do is to rank A over B
(perhaps by not ranking B). But, as I said, Bucklin's price for
its easy hand-count is a need for defensive strategy that Condorcet
doesn't require: In Bucklin, try to avoid ranking any more alternatives
than necessary. In other words, include in your ranking the alternative
that you're likely to need as a compromise, and everything you like
better than it--but nothing that you like less than it. That's like
the simple Approval strategy, except that alternatives are ranked.

Bucklin is one of the methods whick, like Condorcet, will work
perfectly in a 3-alternative election. That can't be said for
Copeland, which is completely indecisive in a 3-alternative
circular tie, and is always decided by its tie-breaker in that
situation. 

MPV & Copeland are 2 methods that can screw up even in a 3-
candidate election.

Some MPV proponents have claimed that Bucklin failed, that the
voters rejected it, when they tended to vote only for their
1st choice ("bullet-voting"). Actually, in keeping with what I
said above, bullet-voting is a valid strategy if one believes
that one's favorite has a win. 

But I should add that, in a 3-candidate race, the extreme
voters have no need to bullet-vote. If their candidate has
a majority he wins immediately anyway. If neither extreme
has a majority then Middle is the Condorcet winner.

For those reasons, too, the Middle voters have no reason to vote
a 2nd choice in a 3-candidate race. If an extreme has a majority
then it makes no difference what Middle voters do. If not, then
Middle is Condorecet winner.

***

I've discussed Bucklin, but not the other methods named by
Bruce, because Bucklin is a good method, the best easy hand-count
method, and has been used in several cities. While the other
methods that Bruce named are just things that various academics
have come up with.

***

Mike






--