[EM] Ridiculous order-reversal precaution

[MIKE OSSIPOFF]

It will be another day or 2 before I have a chance to reply to
messages of any length, but since this message is breif, and since
it relates to the poll, I'm replying now:


Rob LeGrand wrote:

I hope Mike won't take it personally when I disagree with him, but I think 
this
is ridiculous.

I reply:

Rob, how about doing me a favor by more carefully asking yourself
about the reason for a suggestion, before trying to label it "ridiculous". 
What you really meant was that you don't understand
the reason for that suggestion, and you request an explanation.

Rob continues:

Why should we allow truncation but not order-reversal (or for
that matter, ties higher in the ranking) for defense?

I reply:

There now, see how easy that was? That's all you have to say.

I thought I'd explained the point of the truncation period.
Offensive order-reversal is an offensive strategy that's possible with
any pairwise-count method. With the wv methods, offensive order-reversal
can be thwarted by mere truncation.

I clearly stated that the purpose of my suggestion was so that
people could apply counterstrategy against order-reversal. So
it would be rather stupid to suggest that during that 48 hours
people should also be able to order-reverse. Rob, the whole purpose
of the 48 hour truncation period suggestion is so that order-reversal
during the balloting period can be thwarted. Order-reversal is not
needed to thwart order-reversal with the wv methods.

Yes, with the margins methods, sometimes order-reversal is the
only defense agains order-reversal. But if you'll take a more careful
look, hopefully you'll see that it wouldn't be much help to have
a 48 hour period after the balloting in which people can order-reverse.
Because then an offensive order-reverser could do so at the end of that
period, and it couldn't be countered. So the 48 hour order-reversal
counterstrategy period doesn't work for margins, only for wv.
I clearly stated that it was for wv counts.

Rob continues:

Truncation is nothing
but ties at the bottom of the ballot.


I reply:

That's nice. So what? Truncation is sufficient to thwart offensive
order-reversal. Saying that someone could also vote other insincere ties
would serve no purpose, and would be rather silly.

Rob continues:

In fact, I don't see why ballots should be posted to the list before the 
vote
is closed.  Please just send ballots to Mike and me.

I reply:

You're counting the poll, and so of course you can make whatever
rules you want. In this instance you're completely changing the way
we've always done polls, but that's something that you can do, as
the person who's counting the poll.

1. If we aren't going to post the  ballots, then you should, after
each ballot that's sent to you, post to EM the new number of ballots
that have been voted. That way, even though the voters don't have
the actual ballots on the EM list, at least they know how many there
have been.

2. For me, the reason for posting the ballots is because it
makes the poll more interesting, and is likely to result in a better
turnout. Posting the ballots inclines people to participate more than
just reporting the number of ballots, or doing nothing, which would
be worst of all.

3. Although posting the ballots makes offensive order-reversal easier
(unless it's thwarted as I suggested), not posting the ballots
doesn't even come close to ensuring that offensive order-reversal won't
happen. And so it doesn't eliminate the usefulness of the 48 hour
truncation period.

Now, do you now understand the reason for the 48 hour truncation
period, and why it isn't also an order-reversal period? :-)

I really didn't expect opposition to that harmless-seeming suggestion.

Do you actually opppose having a 48 hour truncation period, after the
balloting is ended and the ballots and results posted, for the purpose
of responding to possible offensive order-reversal? If so, then
on what grounds do you oppose it?

There won't be enough offensive order-reversal to do anything in a public
election, but it's entirely possible here.

Rob continues:

Between the two of us,
there won't be any cheating,

I reply:

No, the reason why there won't be any cheating is because the ballots
will be posted, with the names of those who have voted them. Again,
that's how we've always done it. Or are you changing that too?

Rob continues:

and I doubt anyone will be able to strategize
successfully without seeing the ballots.

I reply:

Yes, offensive strategy certainly isn't as easy without seeing the
ballots. But it still can happen. One doesn't need a prediction to
use offensive order-reversal. One doesn't need assurance of reliability
before one tries something.

Rob continues:

(Mike and I should send each other
our votes before we see other votes.)  And when the vote is closed, let's 
not
allow ballots to be changed.

I reply:

On what grounds do you oppose the 48 hour truncation period?

Rob continues:

That might be a fun thing to try for a less
serious election

I reply:

What? It's because the topic is taken seriously that offensive 
order-reversal becomes likely. It won't happen when voting on ice-cream
flavors, but it wouldn't be surprising if it happened in this poll.
It's because of the seriousness of the topic that I made the suggestion.
(I believe I made that clear also).


Rob continues:

, but let's make this one realistic

I reply:

In a public election a truncation period wouldn't be needed because
order-reversal wouldn't be a likely problem. But if it somehow became
one, then a truncation period would be desirable in a public election
too. The "realism" that you want is a false realism, because just
because we leave out the order-reversal precaution doesn't mean that
we won't have order-reversal. This isn't a public election, and
you can't make it be like one by disallowing that precaution.

I'm not going to make an issue about it. Since you're counting the
poll, I'm inclined to not dispute how you want to conduct the poll.
But can you give us a good reason for opposing giving people a
chance to thwart offensive order-reversal?

Rob continues:

and as definitive as
possible.

I reply:

Of course the truncation period has no effect on the definitiveness
of the poll.

I like Approval because it has no offensive strategy.

Disallowing the 48 hour truncation period could make a difference in
what method designation would be the best.

Mike Ossipoff


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