[EM] Re: Blake's Margins Arguments

[MIKE OSSIPOFF]

I'd said:

>Aside from the fact that a majority of all the voters is the
>widely-understood meaning of "electoral majority", such a majority

But you were arguing that this is what "majority" means, not what
"electoral majority" means. I don't really know what "electoral
majority" means. A search on Google suggests it is usually used to mean
a majority of the Electoral College.

I reply:

I was using "electoral" to mean of or relating to an election.

But you really thought I was referring to the electoral college? :-)

By "electoral majority", then, I meant a majority of the voters in
an election, specifically in a particular race in an election.

Blake continued:

>Majority matters because it's a group of people whose need for defensive 
>strategy can be minimized to a degree qualitatively better than
>can be said for a submajority group of people. As described by the
>definitions of the majority defensive strategy criteria.

That's a strategic, not a majoritarian argument.

You wanted a majoritarian argument to justify majority?

Though many agree with me that the pairwise preferences of a majority
group matter, as a fundamental standard, it needn't be considered only
a fundamental standard. For instance, when you violate such a majority
preference, that can't be good for social utility. It could and often
would result in a lowering of SU much greater than the miniscule difference 
that margins advocates claim for margins in the questionable
zero-info simulations that they refer to. If SU counts as a fundamental
standard, then it's a reason to not avoidably violate majority pairwise
preferences.

A majority is a group big enough to get its way. If its members all
share a favorite, they can easily make it win. If they all share a
pairwise preference, they can enforce that pairwise preference. It's
merely a matter of how drastic a strategy they need in order to do so.
With the better methods, that strategy need for them can be minimized
better than it can with other methods.

Margins advocates say that wv also has strategy incentives. Amazing.
Gibbard & Satterthwaite showed that all nonprobabilistic methods do.
But which is worse--a possibility of gaining by insincere equal ranking
in certain kings of sincere circular ties, or a need to reverse a sincere 
preference in order to protect the win of a CW?

I can guarantee that, for certain methods, a majority won't have
certain strategy needs. Those guarantees can't be made for a submajority
group, and can't be made at all for some methods, such as Margins.
With Margins, it isn't even possible to protect that powerfully large
group that's more than half of the voters from those strategy needs.

Margins fails if you care about not making people need to vote drastically 
insincerely in order to protect a Condorcet winner, or
if you want to reduce the need for drastic insincerity as much as possible 
by not avoidably imposing it on majority-size groups.

Why does it matter that the method doesn't force people to drastically
misrepresent their preferences? When that happens, we don't know what
people want. The voting system can't respond to what people want if
people are voting opposite to what they want, or otherwise misrepresenting 
their genuine preferences. That surely qualifies as
a fundamental standard. I've told here why Blake's trifling strategy
incentives in wv don't compare to the avoidable genuine strategy needs of 
Margins.

I'd said:

>In fact, that's the usual form of the familiar lesser-of-2-evils
>problem: A majority prefer B to A, but they're split between factions
>who consider the middle candidate B or the nonmiddle candidate C their
>favorites.

I think the Mutual Majority Criterion that was discussed a while back is
a good way of viewing this problem.

I reply:

It isn't. It's about a fortuitous special case. It doesn't typically
apply to the problem that you're referring to, the one described in
my paragraph that you quoted above, in which the majority referred to
don't agree that both B & C are better than C. MMC doesn't apply
generally to the classic lesser-of-2-evils problem example.

Here's Blake's definition of MMC. It's reasonably right. Presumably
he's referring to actual votes rather than sincere preferences. Plurality 
meets MMC then, except that maybe Blake specifies that
the criterion mustn't be applied to Plurality, so that Plurality won't
pass. If the criterion referred to sincere preferences and stipulated
sincere voting, Plurality would fail for the reason why we'd expect
Plurality to fail. Likewise with Condorcet's Criterion, Smith Criterion,
Condorcet Loser, etc. It's often said that Plurality fails Condorcet
Loser. No it doesn't, when that criterion is defined in terms of
actual votes.

Blake's MMC:

A majority of voters are agreed
that they prefer any one of a set of candidates over any candidate
outside that set, but disagreee over who should win from inside the
set. It seems to follow from majoritarian principles that someone from
inside the set should win. Further, this principle can't lead to a
contradiction.

When I criticize Margins for majority rule violation, that
criticism is based on;

1. Many agree that in principle it isn't good to violate voted majority
   wishes. Voted wishes of most of the voters. To many, though          
apparently not to Blake, that's a
   fundamental standard.
2. Doing so can obviously result in a big loss of SU. SU probably
   qualifies as a fundamental standard.

Aside from majority rule, I criticize Margins for its failures of
the fundamental standard of not making people need to vote drastically
insincerely, concealing their actual wishes from the voting system
and from eachother thereby.


Mike Ossipoff





_________________________________________________________________
MSN 8 helps eliminate e-mail viruses. Get 2 months FREE*.  
http://join.msn.com/?page=features/virus

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