[EM] Ridiculous order-reversal precaution

[MIKE OSSIPOFF]

>MIKE OSSIPOFF wrote:
>
>
>>But can you give us a good reason for opposing giving people a
>>chance to thwart offensive order-reversal?
>
>Laziness and Ignorance. I'm too lazy to want to calculate the results of
>the poll

Actually it isn't necessary to know the results. If you notice that
someone has order-reversed against your favorite, you might well want
to drop their favorite from your list, as a defensive strategy against
the order-reversal, so that the order-reversal can't succeed, so
that the reverser will regret it.

It wouldn't be necessary for each voter to calculate the result though,
if he wanted to consider the result in deciding whether to truncate
in the 48 hours. The person counting the ballots could post it.
If Rob L.G. can count the ballots once, it's only slightly more
time-consuming to post pairwise-count results right after the balloting,
and then count and post them again after the 2-day truncation period
if anyone has truncated during that period.


>as is to decide whether to truncate or not, and certainly too
>lazy to test out the effects of all the possible truncations I could 
>choose.

No need to do that. If you feel that someone has used offensive
order-reversal against your favorite, then drop from your ranking
the alternatives that they like better than your favorite.

>
>Strategic Order Reversal and Truncation is a genuine problem with
>Condorcet - it's not a major one, but it exists.

Correction: Trunacation is a genuine problem with margins methods,
but isn't a problem at all with wv methods. Regrettably, margins
seems to be in fashion on EM right now, in spite of its strategic
faults.

Offensive order reversal is _a_ problem with margins methods, but
with wv methods, offensive order-reversal is _the_ problem, the
only possible problem.

It won't be any problem in public political elections, but it
has a good chance of being a problem in a committee poll such as this.
A worse problem with margins methods than with wv methods.

If you're going to
>introduce special rules to try and patch up that problem, why not
>introduce special rules to patch up IRV, Borda, Approval, Plurality, and
>the rest?

I'll tell you why: No patch can make them anywhere near as good as
the wv methods.

>
>If you don't think Condorcet unfixed is a good method for picking a
>winner in this kind of election because of the strategical element,
>there's a simple solution: pick another method - or pick "Manual".

Yes, that's what I'll do if for some reason we aren't allowed to
truncate ballots after the end of balloting. I'll then designate
Approval. Rank methods are basically unstable, and Approval's stability
is unequalled. But I believe that with the 48-hour truncation period
the wv methods are still the best designation choice. Cloneproof SSD
in particular.

>
>Besides, people could equally well want to strategically truncate when
>there has been no offensive order reversal - so your fix is somewhat
>wider than you imply.

Yes, with the margins methods they could need to defensively truncate
even when there's been no offensive order-reversal. But trying to
help the margins methods would be a hopeless task. They can't be
fixed. The 48 hour truncation period is for the methods that can
benefit from it--the wv methods.

With the wv methods, there's absolutely no reason to strategically
truncate unless to thwart offensive order-reversal.

People might truncate nonstrategically, lazily, and there's nothing
wrong with that, but that has nothing to do with the 48 hour truncation
period, because lazy truncation would be done in the initial balloting
anyway.

Mike Ossipoff




_________________________________________________________________
Get your FREE download of MSN Explorer at http://explorer.msn.com