[EM] Score Voting and Approval Voting not practically substantially different from Plurality?

[Juho Laatu]

On 25.6.2013, at 18.00, Benjamin Grant wrote:

> On Tue, Jun 25, 2013 at 2:53 AM, Juho Laatu <juho.laatu at gmail.com> wrote:
> On 24.6.2013, at 16.06, Benjamin Grant wrote:
> 
>> So, as far as *I* can see, this converts Score Voting into Approval voting.  The only people who would bother to vote sincerely are:
>> 1)      Those who truly prefer Gore highest and Bush lowest (or vice versa), because there’s no strategic downside.
> 
> 
> You seem to assume that voters with opinion 'Gore:75, Nader: 90, Bush: 10’ are not strategic when they vote 'Gore:75, Nader: 100, Bush: 0’. There are thus two possible levels of sincereness, either people who think that all candidates are about equally good should vote that way, or if they should exaggerate and tell that the worst one of them is worth 0 points and the best one is worth 100 points.
> 
> But in the instance where someone's highest priority is to stop Bush, and a distant second level priority is to see Nader elected over Gore, it seems unavoidable to admit that if they vote 'Gore:75, Nader: 100, Bush: 0’ they will be harming their first priority by withholding support from Gore. Isn't this correct?

Yes. Those voters already have some strategic thoughts like "I must maximize the power of my vote". If they sincerely feel strongly that way (Bush is worth 0 points etc.), this can be classified as sincere voting. But if they think that all politicians are actually quite equal, maybe they should vote sincerely 'Gore:75, Nader: 80, Bush: 70’.

> So then that is a non-strategic vote in comparison to 100/whatever/0, yes?

I guess whether we call some vote strategic or not depends also on what the voters were requested to do. If they were requested to evaluate candidates so that 0 points means "worst imaginable" and 100 points means "perfect", then 0 points should be reserved for Hitler and Stalin and similar. Bush is certainly above that level for most voters. But if they were asked to spred the candidates on a scale from 0 to 100, then voters should use also numbers 0 and 100. (The latter approach of course has problems like someone nominating a Republican candidate that is much worse than Bush and thereby lifts Bush to at least 25 points in all ballots.)

> 
> That's what makes strategic voting different from sincere voting, isn't it: that strategic voting has a greater chance of creating a more preferred outcome?

The voters can either try to influence the outcome of the election as much in their own favour as possible, or they can simply indicate their opininion sincerely, as requested by the election organizers. In competitive elections (e.g. political elections) voters tend to adopt the first approach. I some more peaceful elections and polls they may adopt the second approach.

> So long as the strategic vote and the sincere vote are not the same, a sincere vote is a vote against your preferences.

If the election is not competitive, then your sincere votes is also ideal for you, even if you caould change the result to better from your point of view by falsifying your preferences. A typical example situation could be e.g. a vote within a family on what food to make today. In such environments the voters typically want to seek a balanced result rather than get their own preferences implemented every day by using some strategic tricks. Political elections are of course usually more competitive.

> That is why it seems so important to me to favor system where those two kinds of voting coincide as often as possible, right?

Yes. It is one of the key targets to find an election method that would sufficiently discourage strategic voting. In some methods like Approval people (on the EM list) usually expect voters not to vote sincerely (= approve those that you approve) but to cast their best strategic vote, which typically includes approving some of the frontrunners and not approving some of the frontrunners. From this point of view there are at least two categories of practical methods 1) methods where people are expected to express their sincere opinions on the ballot and 2) methods where all voters are expected to follow some strategy that is available to all and that least to balanced results despite of all being strategic. In both cases we want to avoid situations where some voters can cast a stronger strategic vote than others, and where strategic voting would somehow make the end result bad, or make the election more random, or allow the plotters to win, or make it difficult for the regular voters to vote.

>  
>> It’s days like these that I feel that there *is* no way to elect people that is fair and right. L
> All methods have some problems. But the problems are not always so bad that they would invalidate the method. I'd propose to study also the Condorcet compliant methods. I note that they already popped up in the later discussions and you more or less already promised to study them.
> 
> When compared to Range style utility measuring style Condorcet methods take another approach by allowing majorities to decide. With sincere (Range) preferences 55: A=100 B=90, 45: B=100 A=0 majority based methods allow A to win. Althoug B has clearly higher sum of utiliy, it is also a fact that if one would elect B, B would be opposed by 55% majority. A would be supported by 55% majority. Not a pretty sight to watch, but that's how majority oriented systems are suposed to work. Maybe the majority philosophy is that you will get a ruler that can rule (and there is no mutiny), instead of getting a ruler whose proposals would be voted against every time by 55% majority in the parliament or in public elections.
> 
> Juho
> 
> Interesting observation. Personally, in the above example, my gut tells me that B "ought" to win.

B might be the best option for the society, if B would be allowed to rule. But often the majority will not allow this to happen.

On the other hand people tend to give some weight also to how good the candidates are to other voters and to the whole society. I mean that often the majority understands that it does not make sense to e.g. elect a candidate whose election would lead to serious riots within the country. Parties also nominate candidates that are likely to get support also from the supporters of other parties.

So, yes, maybe B ought to win, but in competitive environmnets majority rule seems to be the rule (we may want it to be the rule, or it may be the rule since the majority will force it to be anyway). And in practice also majority decisions may take general interest into account to some extent, assuming that the voters are "civilized".

> However, start tweaking B's numbers downwards, and at some point we will find a level in which A actually looks better that B, for example:
> 
> 55: A=100 B=40
> 45: B=50 A=0 
> 
> Now B isn't looking so good compared to A. So there is obviously some threshold - which may be different for each of us - at which A is the better choice. Perhaps score voting (when everyone does vote sincerely) captures that threshold - maybe B "ought" to win when his numbers are highest. Problem is, a lot of people - perhaps even most - will soon get wise how to push their preferences, and suddenly the Ballots start to look a lot like this:
> 
> 55: A=100 B=0
> 45: B=100 A=0 
> 
> And then we are back where we started.

Yes. Strategic voting will probably be common in (competitive) political elections whenever good and working strategies are available.

Juho


> 
> -Benn

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20130626/6d259305/attachment-0004.htm>