DSC uses a somewhat interesting method - it effectively goes and excludes
the groups of candidates that the most people prefer a "solid coalition" to
until it finds a winner. However, what I am wondering is - what are the
primary flaws of these two methods (especially as compared with IRV, of
which I know quite a bit about the flaws)? I know all later-no-harm
satisfying methods have more flaws than methods that don't satisfy it, but I
am just curious... Frankly, I'm surprised more methods have not been
developed that satisfy this criterion, as it would seem like an important
one if you want people to vote honestly.
Also, does anyone have a clue as to how one would go about developing a
multi-winner PR version of these methods? THAT sounds like an interesting
prospect to me, as STV is currently the only method out there for pure
multi-winner PR that satisfies LNH (not counting party lists, asset voting,
and other nontraditional methods)
Tim
On 12/22/06, Kevin Venzke <[EMAIL PROTECTED]> wrote:
Tim,
--- Tim Hull <[EMAIL PROTECTED]> a écrit:
> At this point, I'd say the choice is between IRV/STV and some form of
> range/approval voting (for single and multi-winner elections). The
issue
> of
> later-no-harm may come into play, though, and cause IRV/STV to be the
> choice... As it is, students tend to bullet vote under the current
> system
> for multi-candidate elections, and it would be good to try to eliminate
> this
> (which systems failing later-no-harm won't do).
Another possibility is that the voters are bullet voting under STV
because they don't understand the method well enough to determine what
their vote is doing.
This possibility is a reason why I prefer not to have a strictly ranked
ballot format for a single-winner election.
> MMPO sounds like it has too many drawbacks to be a suitable IRV
> replacement
> in this regard... Does anyone have any suggestions as far as
> improvements
> to IRV that improve performance while maintaining later-no-harm OR
> systems
> that come close to satisfying later-no-harm (It seems like some
Condorcet
> methods might - especially those that fail later-no-help).
Actually all Condorcet methods fail LNHelp, too. MMPO is the best example
of what you're hinting at, I think.
LNHarm is extremely limiting. FPP and IRV get by just by looking at the
top of the ballot at any one time. MMPO looks at the lower rankings, but
guarantees to only use them to hurt candidates you rank even lower.
(You might guess, that this is where the LNHelp failure is.)
DSC works like this: Voters strictly rank. Sort every possible set of
candidates by the number of voters ranking exactly that set of candidates,
in some order, at the top of their ballots. Then go through the set,
most voters first, and for every set you encounter, if there are
"unflagged" candidates in the set and outside the set, "flag" those that
are outside the set. Stop once only one candidate is not flagged, and
elect him.
Examples:
49 A
24 B
27 C>B
sets and their voters:
49 {A}
27 {C}, {B,C}
24 {B}
A wins immediately. Not great, but LNHarm+Plurality criterion require it.
49 A>B
24 B>C
27 C>B
51 {B,C}
49 {A}, {A,B}
27 {C}
24 {B}
First A is disqualified and then C is, so B wins. Not bad. Better than
IRV here.
49 A
24 C>B
27 D>B
A wins immediately.
49 A
24 B>D>C
27 C>B
or
49 A
24 B>C
22 C>B
5 D>C>B
A wins these two as well. Quite poor I think; IRV is better here.
> Currently, the best option I have seen as far as IRV is the candidate
> withdrawal variant - where a candidate can withdraw and force a recount.
It would be tricky to come up with the withdrawal procedures, I think.
Another idea that has occurred to me is to have two rounds of voting.
In the first, use majority defeat disqualification and first preferences
to determine the two candidates who stand in the second round.
This mitigates the first method's risk of insisting on a strange winner.
It also makes burial strategy less effective: Instead of stealing an
election outright, you can only steal your way into the second round,
which you wouldn't expect to win if you were being shut out in the first
round.
Both rounds *individually* satisfy monotonicity and LNHarm.
Here's an example of the idea:
49 A
24 B
27 C>B
A has a defeat. The first method's result is C>B>A. The second round is
between B and C.
Kevin Venzke
___________________________________________________________________________
Yahoo! Mail réinvente le mail ! Découvrez le nouveau Yahoo! Mail et son
interface révolutionnaire.
http://fr.mail.yahoo.com
----
election-methods mailing list - see http://electorama.com/em for list info
----
election-methods mailing list - see http://electorama.com/em for list info
