[EM] ICT doesn't meaningfully fail CC. CC isn't meaningfully incompatible with FBC.

[Michael Ossipoff]

On Sun, Jul 29, 2012 at 1:39 PM, Jameson Quinn <jameson.quinn at gmail.com> wrote:

> Can you give a specific example where ICT fails the conventional CC, and
> your argument why that isn't "meaningful"? The words below don't add up to
> anything for me, so I think I need a concrete example.

I'll give a numerical example, if that's what you want. But first,
maybe I should better explain the verbal scenario, if only to clarify
what to look for in the numerical example:

Because this is only a verbal scenario, I'm not saying that there are
necessarily only 3 candidates.

First, let's start with un-improved Condorcet:

Your preferences include these: F>C>>W.

Your rank them in this order: C>F>W

C is the CW, and thereby wins.

Then you say, "No, I'm going to rank sincerely.", and you move F up to
1st place with C.

In fact, let's say a few other people, whose preferences and votes are
like yours, do the same.

Now, it happens that C was barely the CW, and that depended on _your_
votes of C>F.

Without those, now (F>C) > (C>F). F pair-beats C. C is no longer the
CW. Now there is a cycle.

And it just happens that, by whatever completion method is in use, W wins.

By ranking F at top with C, you've changed the winner from C to W.

Now, suppose we'd used ICT instead:

ICT counts (C=F)T + (C>F), instead of just (C>F), for determining
whether F beats C. Therefore, by moving F to 1st place, you don't
change the fact that C wins as CW.

But, since that verbal scenario has unimproved Condorcet returning a
cycle, it isn't an example of ICT failing CC.

So, before I can answer about why ICT's CC-failure isn't meaningful, I
should first find an example of that CC-failure.

Right now, I don't even have a verbal scenario for that failure.

But, in my previous post, I told why, even if it can result in a
CC-failure, what ICT does is not a violation or a failure in any
meaningful sense. Why is that? Because ICT's way of counting equal top
ranking is in keeping with the intent and wishes of the voter who
votes equal-top rankings.

I can't stay on the computer right now, but if I can find an ICT CC
failure example later tonight, I'll post it, and then tell why it
isn't meaningful

But I've already told, in the paragraph before last, why any such
failure wouldn't be meaningful.

Well, here's a verbal scenario for the CC failure:

X is CW, as counted in unimproved Condorcet. A lot of voters have
ranked X and Y equal at top. X wins in unimproved Condorcet. But,
because of how ICT counts equal top rankings, when the ballots are
counted by ICT, X no longer beats Y. So now X and Y are both unbeaten.
Y is at top on more ballots than X is, and so now Y wins.

Unimproved Condorcet's CW loses. Therefore ICT has "failed" CC (where
CC is defined by unimproved Condorcet's definition of the verb
"beat").

But, that result, though it's called a "failure" or a "violation" in
terms of how unimproved Condorcet defines "beat", that's the only
sense in which it is a failure or a violation.

And, if unimproved Condorcet's way of counting isn't better than ICT's
way of counting, then there is no reason to define "beat" as
unimproved Condorcet defines it, and there's no reason to judge
failure or violation in terms of unimproved Condorcet's way of
counting.

In my previous post on this subject, I told why ICT's way of counting
equal top rankings is more in keeping with the equal top ranking
voter's intent and wishes (in comparison to unimproved Condorcet's way
of counting them). So unimproved Condorcet's definition of "beat", its
interpretation of top-rankings, and its CW, can't be said to be more
legitimate for defining "failure" or "violation".

I'm going to assume that you'll take my word for it that the
above-described CC failure can happen, and that a numerical example is
required. But, if it is, then tell me and I'll furnish one.

Mike Ossipoff










> Jameson
>
> 2012/7/29 Michael Ossipoff <email9648742 at gmail.com>
>>
>> Suppose you rank X and Y both in 1st place.
>>
>> You do so because you prefer that X or Y wins.
>>
>> If (in un-improved Condorcet) X beats Y, thereby giving the win to
>> someone you like much less than X and Y, you won't like that.
>>
>> You'd prefer that that not happen.
>>
>> In ICT, your equal top-ranking of X and Y counts as your use of your
>> voting power toward keeping X and Y from beating eachother.
>>
>> ICT counts equal top-rankings more in keeping with your wishes.
>>
>> When the verb "beat" is defined consistent with your wishes when you
>> equal-top rank X and Y, and when your equal top ranking is interpreted
>> as you'd like it to be, then ICT doesn't violate Condorcet's
>> Criterion, and Condorcet's Criterion is not incompatible with FBC.
>>
>> For the same reason, IC-Smith-T, IC-Schwartz-T, Double-Ended
>> IC-Smith-T, and Double-Ended IC-Schwartz-T meet the Smith Criterion
>> and Condorcet-Loser.
>>
>> Mike Ossipoff
>> ----
>> Election-Methods mailing list - see http://electorama.com/em for list info
>
>