On Aug 23, 2011, at 9:06 PM, Richard Fobes wrote:
I very much agree with Jameson Quinn that the time has come to write,
sign, and widely distribute a formal statement of the election-method
principles that we agree upon. Yet instead of just providing a
checklist of what we approve, I suggest we take advantage of this
opportunity to ...
* ... inspire(!) policymakers, politically active citizens, and
frustrated voters to take action, and ...
* ... give them a clearly explained declaration they can use as
ammunition in their battles to implement election-method reforms.
To serve these purposes, I'm boldly suggesting an entirely new
wording. Keep in mind that one of my professions has been to work as a
technical writer specializing in translating especially complex
technology into clear English, and I also have experience writing
marketing materials.
This version incorporates the suggestions and refinements already
discussed, so the revision work already done is not being wasted.
Previously I too was thinking that the other version was too long.
Ironically this version is even longer. I now realize that the other
version went into too much detail about subtle issues, and that's what
made it seem long. In contrast, this version uses the extra words to
clearly explain fundamental voting concepts that most people do not
already understand, and to serve the above-listed purposes.
Also I think (or at least hope) that this version better identifies
our real areas of agreement.
My hope is that either this version, or a merging of this version with
other versions, will produce a declaration that we can sign with much
more enthusiasm.
----- The Declaration of Election-Method Experts -----
**
Unanimously we agree that the kind of ballot used in plurality voting
is not appropriate when there are more than two choices. Its
deficiency is that it does not collect enough preference information
from the voters in order to always correctly identify the most popular
candidate when there are more than two candidates.
The mention of "two choices" confuses. The kind of ballot to be used
must be decided, in at least most cases, before one can be certain how
many candidates will be chosen from. If nothing else, write-in
candidates can mean the possibility of more choices.
Unanimously we agree that there are three kinds of ballots that
collect enough preference information to always, or almost always,
correctly identify the most popular candidate. The names and
descriptions of these ballot types are, in alphabetical order:
* Approval ballot, on which a voter marks each candidate who the voter
approves as an acceptable choice, and leaves unmarked the candidates
who are not acceptable
* Ranked ballots (or 1-2-3 ballots), on which a voter indicates a
first choice, and optionally indicates a second choice, and optionally
indicates additional choices at lower preference levels
* Score ballots, on which a voter assigns a number for each candidate,
with the most familiar versions of such voting being to rate something
with 1 to 5 stars or rate a choice with a number from 1 to 10, but any
range of numbers can be used
The type of ballot used in plurality voting does not have an
academically recognized name, but the term "single-mark ballot" can be
used to refer to this primitive ballot type.
Why is the unfairness of plurality voting not better known?
Single-mark ballots do not collect enough information to reveal the
actual preferences of voters in elections that have three or more
reasonably popular candidates. This lack of full preference
information makes it nearly impossible for anyone to produce clear
proof, or even evidence, of unfair election results.
The Approval ballot allows selecting one or more, but does not allow
indicating preference among them. When this same ballot was used in
plurality it worked because the voter was only allowed to select one.
Perhaps the target of this discussion is desire to indicate more than
one AND which are more or less desired.
**
In addition to the four supported methods listed above, we also
support some combined methods. Specifically we support the use of the
Condorcet method to identify a "Condorcet winner" (who is
pairwise-preferred over all the other candidates) and then, if there
is no Condorcet winner, we support using either instant-runoff voting
(IRV) or approval voting to resolve the ambiguity and identify a
single winner.
Most, if not all, Condorcet methods use the same way to find the CW and,
if found, declare that to be the winner. If not found, different methods
have their own way to find a winner.
I question involving IRV here. Seems like, unless defending against
such, that it would have its home problem of wrong choices.
Note that each member of the cycle would be the CW if all other cycle
members were excluded.
Note also that they would have to be near ties to form a cycle. Look at
A>B>C>A. Be easy to have B much weaker than A, and C much weaker than B,
but then how could it happen that C was stronger than A.
In the list of signatures we indicate which voting method each signer
most strongly prefers.
Yet most of us also agree that we will support the adoption of any of
the supported methods, namely, in alphabetical order:
* Approval voting
* Bucklin voting
* Any of the Condorcet methods, which are, in alphabetical order:
* Condorcet-Approval
* Condorcet-IRV
Again I choke on IRV getting near Condorcet, even though they use the
same ballot.
* Condorcet-Kemeny
* Condorcet-Schulze
* Condorcet-Tideman
* Range voting
(The choice of counting method determines which kind of ballot is
needed.)
**
Finally I'll close with a suggested format for signatures (where the
third portion indicates affiliation):
Richard Fobes; Portland, Oregon, USA; VoteFair.org; prefers
Condorcet-Kemeny method
