[EM] strong defensive strategy criterion

[James Green-Armytage]

Dear election methods fans,

	I suggest that ordinary winning votes methods (beatpath, ranked pairs,
river, etc.) fails Mike Ossipoff's "strong defensive strategy criterion",
according to what I think is the most reasonable interpretation of that
criterion, whereas cardinal pairwise passes the criterion.
	Here is the definition, from electionmethods.org

Mike's strong defensive strategy criterion (SDSC):
"If a majority prefers one particular candidate to another, then they
should have a way of voting that will ensure that the other cannot win,
without any member of that majority reversing a preference for one
candidate over another or falsely voting two candidates equal."

	I suggest that if my sincere preferences are A>B>C>D>E, and I truncate my
ballot after candidate C, I am essentially voting A>B>C>D=E, that is,
"falsely voting two candidates equal." Hence, although Mike may not have
intended it this way, I suggest that the majority of voters in his
definition should not have to use either order-reversal, OR truncation, in
order to absolutely prevent the other candidate from winning.
	If we accept this version of the criterion, I think that ordinary winning
votes methods fail. It's extremely easy to find an example, since most
majority rule cycles seem to suffice... I'll scoop one up from my proposal.

46: A>B>C
44: B>C>A (sincerely B>A>C)
5: C>A>B
5: C>B>A

	So, a majority prefers A to B, but B wins with winning votes. The A>B
majority consists of 46 A>B>C voters and 5 C>A>B voters, and I don't see
any way that they can prevent B from winning without reversing a
preference or falsely voting two candidates equal (which includes
truncation). This goes for most majority rule cycles that I can think of.
	The point of this e-mail, the positive point, is that cardinal pairwise
actually does pass this criterion, even with the strict definition that I
have given it. Because, in cardinal pairwise, if a majority ranks X above
Y, and rates X at 100 and Y at 0, there is no way that Y can win. In doing
this, the majority does not need to make a reversal or a false
equalization.

	According to Steve Eppley's wording of the criterion, under the name
"minimal defense", winning votes Condorcet does indeed pass. This is
because Steve's definition says that the majority can't falsify a
preference regarding the higher-ranked candidate, but it doesn't say that
you can't manipulate other preferences.

Steve's minimal defense criterion:  "If more than half of the voters
prefer alternative y over 
        alternative x, then that majority must have some way of voting
that 
        ensures x will not be elected and does not require any of them to 
        rank y equal to or over any alternatives preferred over y.
(Another 
        wording is nearly equivalent: Any ordering of the alternatives
must be 
        an admissible vote, and if more than half of the voters rank y
over x 
        and x no higher than tied for bottom, then x must not be elected.  
        This criterion, in particular the first wording, is promoted by
Mike 
        Ossipoff under the name Strong Defensive Strategy Criterion.  
        Satisfaction means a majority can defeat 'greater evil'
alternatives 
        without having to pretend to prefer some compromise alternative 
        as much as or more than favored alternatives.  Since voters in
public 
        elections cannot be relied upon to misrepresent their preferences 
        in this way, non-satisfaction means that elites will sharply limit
the 
        set of nominees that voters are asked to vote on, by offering a
system 
        in which there are only two viable parties, each of which
nominates 
        only one alternative.)"

	Steve's definition is a weaker criterion. Winning votes meets Steve's
criterion but not my interpretation of Mike's criterion; cardinal pairwise
meets both. Other anti-strategy add-ons to the pairwise method, such as
Mike's AERLO and ATLO, will probably also meet the stricter definition of
the criterion.

my best,
James