pouët.net

Consider two voting systems for a party with three teams, each having one member and each releasing one prod (as there is one member per team, the vote by a team is the vote of its single member):

(i) weak ranking: for each entry voters can choose any score from 1 to 3 stars;
(ii) strict ranking: each voter must strictly order entries (two entries can't have the same score).

Assumption:

A1: objectively, Prod 1 has better quality than Prod 2 which itself has better quality than Prod 3 (Prod 1 > Prod 2 > Prod 3).

Notation:

Vote triplet:=(voter, prod number, score from 1 to three stars)
Result triplet:=(score of the first prod, second prod, thirs prod)

1. Let us first consider objective voters in the sense that voters follow objectively the ranking in A1. In this case (i) and (ii) lead to the ranking we assumed in A1. We give the solution which will make our notation clearer.

A1
Weak ranking
(1,1,***)
(1,2,**)
(1,3,*)
(2,1,***)
(2,2,**)
(2,3,*)
(3,1,***)
(3,2,**)
(3,3,*)
Result
(9,6,3)

A1
Strict ranking
(1,1,***)
(1,2,**)
(1,3,*)
(2,1,***)
(2,2,**)
(2,3,*)
(3,1,***)
(3,2,**)
(3,3,*)
Result
(9,6,3)

2. What will happen in the case where one consider A1': Prod 1 = Prod 2 > Prod 3? With objective voters:

A1'
Weak ranking
(1,1,***)
(1,2,***)
(1,3,*)
(2,1,***)
(2,2,***)
(2,3,*)
(3,1,***)
(3,2,***)
(3,3,*)
Result
(9,9,3) as expected.

Under strict ranking there are two problems. Voter of team 1 (2) can't score team's 2 (1) prod with *** as it already used *** for its own prod. Ignoring altrusim, we can assume that team 1 scores its prod with *** and scores Prod 2 with **. This problem does not affect the final result because voters of team 1 and 2 will still give the lowest score to team 3, which makes the final result consistent. The problem second problem with strict ranking is that voter 3 has to arbitrarily choose to attribute *** either to team 1 or 2.

A1'
Strict ranking
(1,1,***)
(1,2,**)
(1,3,*)
(2,1,**)
(2,2,***)
(2,3,*)
(3,1,***) or (3,1,**)
(3,2,**) or (3,2,***)
(3,3,*)
Result
(8,7,3) or (7,8,3)

Strict ranking is not best in the case where A1' holds. Thus strict ranking should never be used!

From now on we ignore strict ranking and only consider weak ranking.

3. We consider again A1 and add a further assumption: A2 ('nationalism'): voter i gives a highest score (***) to prod i then rank objectively the remaining entries.

A1, A2
(1,1,***)
(1,2,**)
(1,3,*)
(2,1,***)
(2,2,***)
(2,3,*)
(3,1,***)
(3,2,**)
(3,3,***)
Result
(9,7,5)

As we can see, even in the case of some nationalism, the final result is good! Team 3 wins a little under A2. Under A2, weak ranking is also good when Prod 1 and 2 have the same quality, as one can see below.

A1', A2
(1,1,***)
(1,2,***)
(1,3,*)
(2,1,***)
(2,2,***)
(2,3,*)
(3,1,***)
(3,2,***)
(3,3,***)
Result
(9,9,5)

4. Finally, let us add A3: voter of Prod 3 is more supportive of team 2 because team 2 enjoys a higher reputation than team 1 for objective reasons, then he votes objectively. We capture this assumption by considering that under A1, team 3 gives the same score (***) to teams 1 and 2 and under A1', team 3 gives a higher score to team 2.

A1,A3
(1,1,***)
(1,2,**)
(1,3,*)
(2,1,***)
(2,2,**)
(2,3,*)
(3,1,***)
(3,2,***)
(3,3,*)
Result
(9,7,3) which is true, fine.

A1',A3
(1,1,***)
(1,2,***)
(1,3,*)
(2,1,***)
(2,2,***)
(2,3,*)
(3,1,**)
(3,2,***)
(3,3,*)
Result
(8,9,5) which is wrong.

A1,A2,A3
(1,1,***)
(1,2,**)
(1,3,*)
(2,1,***)
(2,2,***)
(2,3,*)
(3,1,***)
(3,2,***)
(3,3,***)
Result
(9,8,5) which is true, but the score of team 2 is slightly increased.

A1',A2,A3
(1,1,***)
(1,2,***)
(1,3,*)
(2,1,***)
(2,2,***)
(2,3,*)
(3,1,**)
(3,2,***)
(3,3,***)
Result
(8,9,5) which is wrong.

This is a quick demonstration that:

- strict ranking should never be used;

- the fact that teams tend to overrate themselves is not a problem;

- but in the case where two prods have identical qualities, being unfairly supportive of one prod can have detrimental effects, regardless whether each team overrates or not itself.

Everyone stay very still, his reactions are based upon movement.

Thanks.

what is this all about?!

Did you actually try making a demo about it?

?

A critic of voting systems used in demoparties?

Not a critic but a discussion about it. I have seen two systems. The one that I call "strict ranking" here, which I think should never be used. The one used for BP (I call weak ranking) is better. And I can't find a more appealing system. Some can be find there Voting system. But even in the case of weak ranking, pro-team behaviour can have undesired effects.

It's designed to hypnotise you if you scroll down too fast!

psonice: :).

How about dice voting. That should be unbiased.

knl: no :).

tomaes: I rate your idea a... 6!

When a party has more than 6 entries, your dice will need to have more than 6 sides, as for role-playing games :).

nonsense, you can have more than one die.

black birdie did it right. everybody came 1st.

Our theses are strong, our papers are long.

i think the only bad thing about vote systems is that the one at bp werent scaled to handle the pressure in the 1 hour after the last compo at bp.

this resulted in a bunch of lost voting points as the voting ended before I (and maybe others) got back to life on monday.

my personal vote wouldn't have tipped any glasses, however if say 10-20 % of the people who wanted to vote was discouraged ....

Thanks for making my day :)

The only problem with voting systems is that Germans are usually allowed to vote.

This is much more preferred.

Yeah, thumb system ftw.

The voting system using partymeister at demoparties is fine as it is.

Inventing a complicated voting system that needs you to be sober while voting will surely never work or create better results. :D

;)