More inputs, more ballotings

[MIKE OSSIPOFF]

Dave Ketchum wrote:

Mike seems to use "better" as approaching some ideal. Not clear
what that is, or whether I would agree with it being an ideal.

I reply:

For me the ideal would be no incentive for strategy. An impossible
ideal, of course, for nonprobabilistic methods.

Dave continued:

Do not understand "CR"
, "multi-input", or "(wv)"

I reply:

We should have a website for defining new terms that we use a lot.

CR stands for Cardinal Ratings. It's the familiar points system
in which the voter may give to each candidate any number of points
with some pre-specified range. For instance, in one version, the
voter may give each candidate any number of points between 0 and 10.
We sometimes hear also of a 0-100 version. I've sometimes called
the 0-10 version "Olympic 0-10".

CR is strategically equivalent to Approval, which is why I like CR.
In CR, one's best strategy is to give maximum points to every candidate
for whom one would vote in Approval, and give minimum points to every
candidate for whom one wouldn't vote in Approval.

CR doesn't have Approval's extreme elegant simplicity, or easy
implementation, but it's more familiar to people. I'll try Approval
first, and then try CR only if Approval isn't accepted.

By "multi-input", I just meant "having more than 1 kind of input (usually, 
but not necessarily, all from voters).

An example would be a balloting system that has been discussed here,
in which voters rank the candidates, but also indicate an Approval
cutoff point in their ranking, or list the candidates who get their
Approval vote.

Another multi-input method that's been proposed here was suggested
by Tom Round, and later by Steve Eppley--The standing-down option,
or candidate withdrawal option:

Whatever voting system is being used, after the count any candidate
may declare his withdrawal from the election, and call for a re-count
of the ballots from which his name has been deleted.

This was initially proposed, and is usually proposed, as a mitigation
for IRV's problem. It does a pretty good job of that, but the
IRV promoters reject it, just as they reject every mitigation proposal,
apparently because they want to impose on voters all of the
disadvantages of unmitigated IRV.

Note that with this system some of the input is from candidates.

Last summer we did some polling in which our voting system was
Voter's Choice, a method in which voters could make out whichever of
the following kinds of ballots they chose to: Ranking, CR, and
Approval. The actual count was by Approval, and the voter had the
option of designating any count rule using one of the above ballotings,
indicating that he wanted his final Approval set to extend down to
the candidate chosen by his designated voting system. One would
designate the voting system that one believes does the best job of
finding the best compromise that one can get. Voters also had the
option to designate "manual", to indicate that they want their
final Approval set to be the candidates that they voted for in
their Approval ballot.

The abbreviation "wv" stands for "winning votes". Condorcet's method
involves dropping weakest defeats, or keeping strong defeats. So
it's necessary to define how to measure the strength of a defeat.
There are 2 measures proposed by people here:

Winning votes means that if A beats B, then the strength of A's
defeat of B is measured by how many people ranked A over B.

Margins means that the strength of that defeat is measured by
the number of people who ranked A over B, minus the number who
ranked B over A.

Understandaby, margins at first might sound better, but winnging
votes confers compliance with a number of defensive strategy criteria.
I call them the majority defensive strategy criteria, and we
define them at http://www.electionmethods.org

Additionally, wv makes more sense ethically. All the pairwise
defeats initially found in the count are decisions voted by the
voting public. If A pairwise-beats B, and if we drop that defeat,
then we're overruling the people who voted A over B.

But don't we overrule the people who voted B over A if we don't drop
that defeat? No, because they were overruled when the public voted
for the decision "A beats B".

Margins advocates like margins because it's symmetrical. But A's
defeat of B intrinsically has a glaring asymmetry: The voters who
voted A over B won. The voters who voted B over A lost. Drop the
defeat and you overrule the A voters. Keep it and you overrule no one,
since the B voters have already lost and been overruled by the
greater number of B voters.

Another way to put it is: We overrule voters by dropping a defeat.
We overrule no one by keeping a defeat, because the defeats that
exist before we begin dropping them are the expressed will of the
people.

When winning votes is used, SSD, CSSD, & RP comply with SFC, GSFC,
WDSC, & SDSC. PC at least complies with SFC & WDSC. As I said, these
criteria are defined at our website.

SSD is defined there too. RP definitions have been posted in recent
days.

Let me briefly define CSSD.

First the Schwartz set:

1. An unbeaten set is a set of candidates none of whom is beaten
   by anyone outside that set.
2. An innermost unbeaten set is an umbeaten set that doesn't contain
   a smaller unbeaten set.
3. The Schwartz set is the set of candidates who are in innermost
   unbeaten sets.

[end of definition]

Cloneproof SSD (SSD stands for Schwartz Sequential Dropping):

1. Calculate the Schwartz set, counting only undropped defeats.
2. If there are no defeats among the members of that set, then its
   members win, and the count ends.
3. Otherwise, drop the weakest defeat among the members of that set,
   and go to 1.

[end of definition]

Of course typically there's only one winner. If there were pairwise
ties, there could be a Schwartz set with more than one member, and
with no defeats among that set. But pairwise ties are vanishingly
unlikely in public elections.

Mike Ossipoff






- I cannot spend
all my time here - some of my available time goes for promoting Condorcet.
BTW - is Condorcet in use anywhere? It should be.






_________________________________________________________________
Get your FREE download of MSN Explorer at http://explorer.msn.com/intl.asp.