[EM] Fwd: Ranking unacceptable candidates.

[Juho Laatu]

This message that was intended to go also to the EM list.

Juho


Begin forwarded message:

> From: Juho Laatu <juho4880 at yahoo.co.uk>
> Date: 19. 10 2013 8.31.39 UTC+3.00
> To: Michael Ossipoff <email9648742 at gmail.com>
> Subject: Re: [EM] Ranking unacceptable candidates.
> 
> P.S. Should this go to the EM list too?
> 
> Juho
> 
> 
> On 19.10.2013, at 8.29, Juho Laatu wrote:
> 
>> On 19.10.2013, at 3.31, Michael Ossipoff wrote:
>> 
>>> Juho--
>>> 
>>> Ok, but you just have to make sure that ranking unacceptables can't
>>> increase the probablity that an unacceptable will win.
>> 
>> I would say that the probability of ranking unacceptables making one of them win is small enough so that voters need not worry. The benefits of sincere ranking should be bigger than the risks.
>> 
>>> 
>>> And you have to ensure that the rules you use to achieve that don't
>>> create any strategic need to rank all the unacceptables who have a
>>> chance to win. That can easily happen, given rules that make it safe
>>> to rank uncceptables.
>> 
>> Also here I'd say that the strategic needs and benefits of startegic voting shall be low enough (in many cases they are luckily negative) so that the voters need not worry about them.
>> 
>>> 
>>> IRV stands out, meeting Later-No-Harm (LNHa), and Later-No-Help (LNHe).
>> 
>> IRV is a good method the sense that one can prove that complete rankings cause no harm. It has some "randomish" properties that many do not like. And it favours large parties. IRV may be a good choice if one has to defend against real and claimed strategic problems. But generally I lean towards Condorcet style methods (in typical political elections where the target is to elect good compromise candidates). They have some well known strategic (theoretical) vulnerabilities (burial is the first one), but my guess is that in most political environments they are marginal enough to be overrun by the benefits of electing the "right" winner.
>> 
>> The ranking categories that we discussed could be an interesting informative addition to the Condorcet (and why not any ranked) methods. With "informative" I mean that they could be used even if they have no impact on which candidate will be elected. I.e. just to allow voters to carry messages, to make voting easier (in some ballot formats), and to collect additional statistical information.
>> 
>> Juho
>> 
>> 
>> 
>>> 
>>> Michael Ossipoff
>>> 
>>> On Fri, Oct 18, 2013 at 2:19 PM, Juho Laatu <juho4880 at yahoo.co.uk> wrote:
>>>> On 18.10.2013, at 19.20, Michael Ossipoff wrote:
>>>> 
>>>>> I don't deny the desirability of the better expressivity and more
>>>>> information gotten by encouraging complete rankings.
>>>>> 
>>>>> But, as you know, one must choose between different desiderata.
>>>>> 
>>>>> The problem is that, if completre ranking is encouraged, that can make
>>>>> incentive for complete ranking, and that can make strategic need of
>>>>> complete ranking. It can make a burial-war, as you yourself pointed
>>>>> out.
>>>> 
>>>> There's a risk that people start using strategic truncated votes or defensive strategic truncated votes. Both would delete valuable informaton. I don't believe burial-wars would be a major problem in typical ranked elections. If they are, they might be irrational burial-wars.
>>>> 
>>>> If complete rankings start causing problems, then we need some tricks. But I do believe that in most large public elections it makes sense to recmmend voters to rank all (but one) potential winners (at least). I thus believe that complete rankings is the way to go in most pubic elections. I have not seen convincing evidence that we must satisfy with less (because of strategic voting related problems in typical public elections).
>>>> 
>>>>> Maybe, for public elections it's better give incentive for
>>>>> unacceptable-truncated rankings, instead of for complete rankings.
>>>>> Largely because it makes voting easier. That could be important.
>>>> 
>>>> Ease of voting is important, but so is getting the correct results. If people truncate, it could be a typical voting pattern to truncate the candidates of "the opposite side". That's not good since in that case we could easily elect extremists of the biggest grouping (note plurality style here) instead of candidates that are acceptable also to the other groupings. One thus has to rank also the key candidates of the unliked groupings in order to make the election method work as intended (I'm assuming that the intention is to elect good compromise candidates).
>>>> 
>>>>> For example, in my most recent poll at Condorcet Internet Voting
>>>>> Service, there are 30 party platforms. No one wants to be
>>>>> strategically-compelled to rank all of them.
>>>> 
>>>> People should not be strategically compelled to rank all of them. I just want all voters with clear opinions to express them, in order to allow the method to elect the best candidate.
>>>> 
>>>> If there are so many candidates that people do not want to rank all of them, there are different approaches to making voting easier.
>>>> 
>>>> One obvious approach is to reduce the number of candidates. It does not make sense to have very many candiates with no chances to win (I assume that this is a typical political election with maybe 2-6 potential winners). It makes sense to have more canididates than just those 2-6. That means major party candidates that can be potential winners in the next election, and candidates of small parties that may grow and win in the future.
>>>> 
>>>> Another approach is to allow use of group names. I'm not sure if this is easy enough for all the voters in all elections. The voters could vote e.g. A1 > A2 > A_party > B1 > B_party > C1. This vote means that all unlisted candidates of the A_party are ranked equal between A2 and B1, etc. This would make it possble to give some opinions on candidates of multiple parties without ranking all of them (e.g. all the 20 candidates of A_party).
>>>> 
>>>> I also note that the ballots may be formated so that they make ranking of numerous candidates easy. Let's say that we have the names of the candidates at the beginning of each row. Canidates of one party and/or one grouping are grouped together. There are 15 columns that the voter can tick. Five first columns are green and titled "favourite", next five columns are white and titled "acceptable", and the last five colums are red and titled "unacceptable". It should be a rather quick operation to go through a long list of candidates, maybe ranking quickly all candidates of party X in the "unacceptable" section, etc. (This approach will not allow complete ranking if there are more candidates than there are comulms, but that may be a small penalty if we can get more complete information from the voters, and get the potential winners ranked completely this way.)
>>>> 
>>>> It is good if ranked elections have only so many candiates that voters can form an opinion of all of them (at least the potential winners) and rank all of them (at least the potential winners).
>>>> 
>>>>> You don't want to need to
>>>>> rank the 2nd worst over the worst, just so you can counter
>>>>> Worst'voters doing the same thing to your favorite.
>>>> 
>>>> Complete rankings should be used to give complete information, not for strategic reasons (hopefully those needs are very minor when compared to the need of getting correct results).
>>>> 
>>>> Juho
>>>> 
>>>> 
>>>> ----
>>>> Election-Methods mailing list - see http://electorama.com/em for list info
>> 
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20131021/b8416c86/attachment-0003.htm>