[EM] wv & margins definitions & difference

[Elisabeth Varin/Stephane Rouillon]

I still find Mr. Ossipoff analysis not objective
in the sense he claims margins and relative margins
can sometimes produce equilibria only when defensive
order-reversal is used.

Winning votes methods can sometimes produce different result
the same way depending on the fact voters decided to express
their full preferences or not.

In the same way truncating your preferences can harm your favorite
using margins, giving your full preferences could harm your favorite
using winning votes.

Thus I see no relevance in verifying those "defensive strategies"
criterias,
when in the same time a method would not satisfy an "offensive strategy"

criteria. Matching the pair ranking to the highest probability then
seems the fairest objective, so it leads to relative margins.

Steph.

------------
Mike wrote:

In reply to the question about the difference between winning-votes
and margins, it's as Forest said: The difference occurs when
not everyone ranks all of the candidates. Truncation is common
in rank elections, and will always happen in public elections.

Say A beats B. Winning votes refers to the number of voters who
rank A over B. Margins refers to the number ranking A over B,
minus the number who rank B over A.

There are very important differences in strategy properties of the
methods using those 2 defeat-measures. The website
http://www.electionmethods.org lists some criteria, at its
technical evaluation page. All the criteria for which Condorcet is
listed as complying are criteria that BeatpathWinner(wv),
Cloneproof SSD, Plain Condorcet(wv), and Ranked-Pairs(wv) comply with.
The margins methods fail all of those criteria.

In addition, the margins methods are "falsifying", as I defined that
term here. With those methods, there are situations in which the
only equilibria are ones in which defensive order-reversal is used.

Mike Ossipoff

----
For more information about this list (subscribe, unsubscribe, FAQ, etc), 
please see http://www.eskimo.com/~robla/em