[EM] Re: majority rule vs. maximum approval (was: least additional votes)

[James Green-Armytage]

Dear Forest Simmons,
	Replies follow...

>What I mean is "the degree of consent."
>If you can get 85 percent of the people to agree that plan A is an 
>acceptable alternative, then I consider that an 85 percent consensus.

	I do understand what you're saying, but I'm still not totally convinced
that "acceptable" is sufficiently well-defined. I'm tempted to say that
one of the most helpful definitions of "consenting" to a course of action
is "preferring a course of action to the status quo". This brings us into
the territory of supermajority methods, which I explored to some extent in
December 2004, and which I am told has been explored in more detail
earlier in the history of the EM list.
http://lists.electorama.com/pipermail/election-methods-electorama.com/2004-December/date.html
	In some cases it may be possible to do an iterative procedure, where an
option is first sought that is preferred over the status quo by 90% (for
example). If no such option exists, the procedure could lower the bar to
80%, and so on towards 50%+. 
	Strategic voting may still be a vexing issue in such a procedure,
however. If you establish a Condorcet completion method winner first and
then ask whether that particular measure is preferable to the status quo
(referring to the same ranked ballots), there is little strategic
incentive in the second part of the process (placement of status quo in
the preference ordering), but you exclude possible solutions that are not
the CW.
	What are some alternative procedures? One would be to find the CW among
options that are preferred over the status quo by the required
supermajority. The strategic vulnerability here would be in false
placement of the status quo in the preference ordering, by which an
organized majority could prevent the selection of any candidate at the
supermajority phase, thus biding time until the bare majority phase is
reached.

>Approval voting by itself cannot solve the "tyranny of the majority" 
>problem, because a confident majority can always impose its will by 
>"bullet voting."

	Right. I suggest that supermajority methods are needed to prevent the
tyranny of determined and well-organized majorities.
>
>However, the spirit of approval voting is in the direction of increased 
>consensus, and voters that approve sincerely (whether from altruism or 
>lack of information for strategizing) increase the likely degree of 
>consensus with respect to the outcome.

	I understand what you're saying, although I'm not sure that I agree with
you. As you suggest, there are two fundamentally different possible causes
for the approval winner being distinct from the majority winner (i.e. the
winner given a Smith-efficient method).

(1) There is a meaningful and interpersonally constant phenomenon of voter
"approval" (or "acceptance" or "consent"), such that voters either approve
or disapprove of any given candidate. People will cast approval ballots
according to this inner feeling rather than according to strategic
concerns.
(2) Poor information and coordination.

	I suggest that explanation (2) is much more likely to be the primary
operative reason that approval election results will differ from the
results of a Smith-efficient method in contentious public elections.

my best,
James