[EM] (Possibly) new method/request for voting paradoxes. :)

[Juho]

On Oct 13, 2009, at 9:01 PM, robert bristow-johnson wrote:

>
> On Oct 13, 2009, at 1:58 AM, Juho wrote:
>
>> Welcome to the list!
>
> thanks.
>
>> On Oct 13, 2009, at 7:48 AM, robert bristow-johnson wrote:
>>
>>> it is also important to have a deterministic and monotonic measure  
>>> of voter support that is understandable to the less scholarly
>>
>> I have often promoted the measure of least additional votes  
>> required to become a Condorcet winner as one understandable and  
>> natural/fair (and simple, easy to display) measure of which  
>> candidate is the best. This measure leads to the minmax(margins)  
>> method. The "least additional votes" approach minimizes the  
>> strength of opposition to change the winner to any single  
>> alternative winner after the election (and thereby aims at making  
>> the society more stable).
>>
>> Not all on this list agree that minmax(margins) is a good method.  
>> It may in some extreme cases elect outside the Smith set. But in  
>> such cases the defeats within the Smith set are stronger than  
>> defeats of the winner outside the Smith set, so electing that  
>> candidate would not break my heart :-).
>>
>> When deciding which Condorcet method is best some people put more  
>> weight on resistance against strategic voting while some try to  
>> optimize the output with sincere votes. Different environments may  
>> have different needs with respect to strategic voting. In  
>> environments where strategies are not expected to be a problem (in  
>> Condorcet methods) one may use the latter type of criteria.
>>
>> Just wanted to point this out since you seemed to make the  
>> assumption that the best winner must come from the Smith set.
>
> yeah, i guess i have made that assumption (or conclusion).  i think  
> the same reasoning applies; to select a non-Smith candidate over a  
> Smith set candidate seems to violate the same really basic principle  
> of democracy as selecting someone else over the Condorcet winner if  
> a Condorcet winner exists.  anyone inside the Smith set clearly  
> beats anyone outside of the Smith set if the voters are asked to  
> simply choose between the two.

The Condorcet winner wins any other candidate. Smith set members do  
not, and they may have worse losses than candidates outside the Smith  
set.

I think the reason why the Smith criterion is so popular is that it  
offers a visually attractive way to "break the cycles" and make the  
social preferences look like they were transitive (although they in  
principle are not). There sure are cases where one of the Smith set  
members is the most natural winner, and that is by far the most  
typical case, but I think there are also cases where one can justify  
electing outside this set. The Smith criterion emphasizes the fact  
that some candidates lose to _all_ members of this set but does not  
put any weight e.g. on the _strength_ of defeats of the candidates.

Again, just trying to find weaknesses in criteria that are commonly  
taken as granted. Since group preferences are not total linear  
orderings one may have also other approaches to finding the best  
winner, e.g. the strength and position of each candidate if he/she  
will be elected.

>
> from a political POV (not the academic one of worrying about obscure  
> details of hypotheticals that seem to me to be very unlikely to ever  
> happen in the context of a simple political spectrum), the *most*  
> important concern in my agenda is to promote Condorcet over any  
> other method and to worry about which Condorcet less.

I agree that most members of the Condorcet family of methods are worth  
promoting, and are excellent for typical single winner elections.

>  *some* method to resolve a potential Condorcet cycle *does* have to  
> be determined, in advance, wherever Condorcet is adopted, but  
> *which* method (as long as it is meaningfully consistent with some  
> monotonic measure of voter preference) is less of a concern for me.   
> and, if some really good argument comes up to change one Condorcet  
> method to another, i would likely say "whatever" and i could vote  
> either way.

As already discussed above it is sufficient to determine one single  
winner, and resolving the cycles (to form a linear preference order)  
is not required. Group preferences are of different nature than the  
preferences of individuals (that are typically assumed to be linear).

>
> i really think that moving away from 2 dominant parties and Choice  
> Voting is extremely important in the political marketplace, and even  
> though i supported it in the past, that all of these people plugging  
> IRV should be plugging Condorcet instead or, at least, along with  
> IRV as the alternative to plurality and/or delayed runoff.  and i  
> really don't like the "happy talk" from the IRV activists (like  
> fairvote) about the "unmitigated successes" of IRV when i have seen  
> it to fail, first hand.

I'm not a friend of two-party systems (although I think also they have  
some good points and can be claimed to be one working model of  
democracy). I believe proportional systems are better for most modern  
democracies.

Condorcet is a single winner method (not a proportional multi-winner  
method), so it may not be very helpful in trying to get rid of a two- 
party system. Proper proportional multi-winner methods might be a good  
target. (Condorcet is excellent for most single-winner elections  
though, and it avoids some of the worst problems of IRV.)

>
> BTW, "hi" to Terry Bouricius and/or Rob Ritchie, if they be hangin'  
> out here.

I guess they are. Hello!  Btw, it has been said that the target of the  
US IRV community is to end up in a STV based proportional system. The  
STV style multi-winner approach would also alleviate the known  
problems of plain single-winner IRV. There are of course also other  
approaches to proportionality (= the most obvious alternative to a two- 
party system).

Juho



>
> --
>
> r b-j                  rbj at audioimagination.com
>
> "Imagination is more important than knowledge."
>
>
>
>
> ----
> Election-Methods mailing list - see http://electorama.com/em for  
> list info