[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: GRs, irrelevant amendments, and insincere voting

> On Sat, Nov 08, 2003 at 08:21:18PM -0500, Raul Miller wrote:
> > In general, people who wish to vote insincerely need to have highly
> > accurate predictions of the outcome of the vote to make sure their
> > insincere vote doesn't result in an outcome less desirable than a
> > sincere vote.

On Tue, Nov 11, 2003 at 07:28:09PM -0500, Branden Robinson wrote:
> Can you support this with some references to the literature?

Hmm.... Markus Schulze recently published a paper in "Voting Matters"
on the efficient implementation of the voting mechanisms we use,
which also happens to "prove that this method satisfies Pareto,
monotonicity, resolvability, independence of clones, reversal symmetry,
and plurality. From the definition of this method, it is clear that it
also satisfies anonymity, neutrality, Smith-IIA, and Schwartz."

If you write him and asked, I'm sure he'd send you a copy --

However, I'm not certain that that properly answers the question in the
sense you've phrased it.

Then again, I'm not sure I properly understand the question you're asking.
Can you phrase your statement about what you see is the problem in a
formal fashion?

> > Definition: "insincere vote" -- ranking option A above option B to ensure
> > some outcome which doesn't involve A winning.
> /me frowns
> That's considerably stricter than my definition.  By your definition,
> the only insincere vote is one which doesn't rank one's most-preferred
> option first.

I disagree with this assessment of my definition.

If I vote 1234 on a ballot ABCD, I have ranked option B above option C.
The reason I've ranked option B above option C is because I'd prefer
option B to win more than I'd prefer option C to win.

If I had ranked option C above option B, even though I prefer option B
to win over option C, then that vote would have been insincere -- this
would even be the case where I ranked option A first and option A was
my most preferred option.  Here, it's my treatment of option C which is
insincere, not my treatment of option A.

> Since our system only allows one winner per set of options, we can
> reasonably conclude that the voter wouldn't mind seeing any options not
> ranked first losing.

I think you're missing context, here.

Consider the ballot:
A - I like blueberry pie
B - I like steak
C - I like rotten maggots
D - I like cheeze blintzes
E - I don't like anything
F - I like blueberry pie and steak

Let's assign E the role of the default option (it's what we'd be stuck
with until we can reach a decision).

Now, let's say we have a hypothetical voter who votes 326451

Here, while it's true that our hypothetical voter wouldn't mind seeing
any of A, B, C, D or E losing to option F, it's also true that the voter
would mind seeing options A or B lose to option C.

> The strategic insincerity I find of interest involves ranking "further
> discussion" over a non-most-preferred option even though one would
> actually rather see a given issue settled with the non-most-preferred
> option winning.


> As Manoj said, artificially inflating your preference of "further
> discussion" may be indicative of a non-team-player mindset, at least
> when there are at least three options on a ballot.  (It's difficult for
> me to imagine a simple "Proposal,Further Discussion" ballot as anything
> more complex than a straight referendum.)

You've lost me, here.

> Who knows -- maybe a straight referendum would be the best way to
> resolve my pending GR, given its omnibus nature and some of the subject
> matter.  I'm really not sure yet.

I don't know what kind of issues you're grappling with.


Reply to: