[EM] serious strategy problem in Condorcet, but not in IRV?

[Stephane Rouillon]

Kislanko at aol.com a écrit :

> In a message dated 8/19/03 4:06:08 PM Central Daylight Time,
> stephane.rouillon at sympatico.ca writes:
>
>
>> This can happen with plurality too in a three runners election...
>> Expected:
>> A: 34
>> B: 33
>> C: 32
>> D:   1
>> Reality:
>> A: 33
>> B: 30
>> C: 34
>> D:   1
>>
>> Supposes I am a C>B>A fan.
>> If with a friend, we move our vote from C to B to block the expected
>> election of A,
>>
>> we could end up making A win!!!
>> Expected after "strategy":
>> A: 34
>> B: 35
>> C: 30
>> D:   1
>> Unexpected reality:
>> A: 33
>> B: 32
>> C: 32
>> D:   1
>>
>> Classify strategies as much as you want, it is not their nature that
>> is important.
>> It is their probability of occurence and their consequences.
>
> Ok, for us logicians let's try to sort this out.
>
> All the example says to me is that 2 people who voted insincerely
> based upon an inaccurate assumption caused their candidate to lose
> instead of win.

Markus thinks they are sincere, just compromising.  He says it is not
manipulation.
I think they are sincere, but still trying to manipulate the outcome.
Sincere manipulation is still manipulation to me.

> That is not a fault of the METHOD, is it? Seems to me that this only
> says the 2 voters made a mistake.
>
> Paul Kislanko

I would not say it is the fault of the method.
Those people made the fault by their own decision.
But if a method could avoid people making that fault, would not it be a
BETTER method?
Actually, I think no method is strategy-proof. But some have less
probability of being
manipulated and less consequences to manipulation. Thus it gets us back
to the esperance
measure to compare methods, except we still have problems with the
combinatorial obstacle.

Finally, strategy is not only a probabilistic aspect, it includes
feeding false information to
your adversaries in order to affect their voters behavior. I know I
sound like playing a wargame, but we analyse electoral method using game
theory, don't we?
So if you still think that the previous voters were not manipulating the
electoral system by
getting the most of their ballot, do you agree that their could be
manipulated voters?
If there are methods that prevent manipulation of and by voters, I think
those
are BETTER methods.
That's an argument near what Mike Ossipoff was arguing when defending
winning votes over margins and relative margins... Do not let the system
give the opportunity
to voters to try to improve their result with modified votes based on
the expected ballot
distribution. Wether the manipulation is defined as sincere or unsincere
is not really
relevant to me...

Steph
PS: I hope you do not see anything wrong sharing this with the list...
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20030819/e9a804bb/attachment-0003.htm>