[EM] MTA2, better than MTA or MCA. Dave comment. 3-slot ICT. MTA2 strategy.

MIKE OSSIPOFF nkklrp at hotmail.com
Wed Mar 28 13:38:06 PDT 2012

MTA2, better than MTA or MCA:

MTA2 is the same as MTA, except that, when there are 2 or more candidates with
top ratings from a majority of the voters (call them "top-majority candidates), the top-majority candidates whom you've
middle rated don't get your middle ratings if any of your top-rated candidates are
among those who have top ratings from a majority of the voters.

For anyone who hasn't seen the definition of MTA, I'll write the entire definition of
MTA2 here:

1. Voters mark each candidate "Top", "Middle", or unmarked.

2. If one candidate is rated Top by a majority of the voters, then s/he wins.

3. If no candidates are rated Top by a majority of the voters then the winner is the
candidate who is rated Top or Middle by the most voters.

4. If two or more candidates are rated Top by a majority of the voters (call them majority-top candidates), 
then each ballot that doesn't have any top-majority candidates among its top-rated candidates is said to
"still middle rate" its middle-rated candidates. The winner is the candidate who is top-rated or still middle-rated
by the most ballots.

[end of MTA2 definition]

As I said, I consider MTA2 to be better than MTA and MCA.

Dave comment:


You said:

Deciding on implementing would require decisions on ballot format and counting rules.


Exactly. And the more different the balloting, and the more different and complicated the count, the more
skeptical, doubtful, hesitant and resistant the public will be. The media and the opponents are sure to say,
"This proposal needs a lot more study."  In other words, table it forever. 

That's a big advantage of Approval. It's the most minimal change from Plurality. The balloting needn't be different
from what it now is with Plurality. Instead of "Vote for 1", the ballot would say, "Approve one or more".  That's the
only change in the ballot. 

As for the count, there's only one way to count approvals: Add them up. 

With all of the other proposals, someone could ask, "Why this method, why this count, instead of some different one?"

They can't ask that about Approval, because there's only one simplest, easiest improvement on Plurality.

At EM, we discuss some complicated methods, and some proposals that are really for proposing farther down the
line. That's fine, but it's clear that Approval is the proposable, enactable, feasible first proposal for voting system
reform. We can disagree on what's the ideal best, or on what should be proposed when automated counting
is fraud-secure, or when complicated methods are proposable, but we'll get significant new-method results
soonest if we propose what's most enactable now.

Though MTA2 is better, as a method, or (better yet) as an approval-management option in an Approval election,
there could also be a method "MTA/MCA", in which the voter could select, for each of hir middle-rated csndidates,
whether or not s/he wants to give hir middle rating for that candidate to count when 2 or more candidates, including that one, have top
rating from a majority. I consider MTA2 to be much better than MTA/MCA.

Strategy is similar for MTA2, MTA and MCA. Someone has already pointed out that you aren't helping your top-rated 
candidates very much against your middle-rated candidates.  --only if one of your top-rated has a top-majority.

So, give top ratings as you'd give approvals in Approval. Except, when you're uncertain as to whether you'd give candidate
C an approval, or if C is barely good enough to get an approval, or is almost good enough to get an approval, then give
to C a middle rating.

That wording is imprecise. No method can have anything like the clear and simple strategy of Approval. Still, what MTA2
adds is of value. Its strategy, as I suggest it, starts with Approval strategy.

You won't give many middle ratings. Kevin said that that kind of strategy seemed to get the best results in a simulation.

But I still say that the first proposal should be plain ordinary Approval. MTA2 could later be added as an option.

Of course so could the options of MTA, MCA and MTA/MCA. But, because MTA2 is better than all of those, I don't
know if there'd be any point in including those other options.

ABucklin could be included too, as an option. Though I like the conditional versions of all these methods, and would use
it in an actual election, I admit that the AOC kind of conditionality is too computation-intensive for a feasible handcount.
Maybe GMAT and MMT as well. Probably so. 

If you had to choose between using the MTA option or the MCA option, then I'd suggest that MTA might be better if
you expect for one of your middle rated candidates, but none of your top-rated ones, to get a top-majority. Or if you
like the middle rated candidate or trust hir voters.

Otherwise, if you expect that it's more likely that a top-rated and a middle-rated candidate of yours will get top-majority,
or if you like the middle rated candidate or distrust hir votes, then choose MCA.

But, with MTA2, you needn't bother with such a choice.

With a handcount, ABucklin would have to not use the delay that I've described here, because it makes too much
count labor. Just as  a matter of fact, I might add that, if automated counting were fraud-secure, the voter could
choose for hirself whether s/he wants the delay to operate in hir ballot, if s/he uses the ABucklin option.

Of course you might want to vote the same way in 3-slot RV. But it has more practical value in MTA2. MTA2 offers something
genuinely (a little)  more strategically varied and expressive than Approval. An opportunity to vote more expressively, and for that to have some
practical meaning, more than it has in 3-slot RV.

MTA2's provision could probably be somehow added to Chris's full-rankings generalization of MTA.

MTA2 is probably the most deluxe method of its type that meets FBC and isn't conditional or defection-resistant.

Of course MCA is 3-slot ABucklin. Though ICT is handcount infeasible, that probably isn't true of 3-slot ICT. It might be a good
method. It would be worthwhile to compare its merits to those of MCA2.

Mike Ossipoff

Mike Ossipoff

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20120328/332335fd/attachment-0002.htm>

More information about the Election-Methods mailing list