pouët.net

AFAIK Partymeister uses Condorcet method

As far as I remember, main#4 used strict ranking which forces the voters to rank two prods whereas they can be of identical qualities.

Are you sure?
I though it was strict ranking @ numerica and weak ranking @ MAIN.

I trust you.

Darwinism.

ergon

I would not give a thought to that if ranking of prods on pouet was the same as ranking from demo parties (which one is more reliable?). As it is not always the case, in particular for prods of similar qualities, one should worry about the presence of some ranking bias in demo parties. It is not always a small sample bias (e.g., BP had hundreds of voters).
Ideally:
- strict ranking should never be used;
- voters should not be rushed when voting;
- team size should be corrected for;
- ...

I still think thumb up/piggie/thumb down works the best.

Voting is unfair.

Knoki: I agree with you in the sense that prizes are generally given to three prods. But when it comes to voting, we have to vote for more than three prods. Therefore, with more than three scores, voters are allowed to be fair. Assume that objectively Prod 1 > Prod 2 > Prod 3 > Prod 4. One will naturally thumb up Prod 1, pig Prod 2, then thumb down Prod 3 and 4 whereas Prod 3 would deserve some score between piggie and thumb down.

Zplex: as long as prices (money, material,...) are given, on what system one should rely? But I agree with you that voting can be (not "is") unfair. To have my prod ranked last whereas I put my heart and made efforts in doing it is not very friendly. Price giving forces teams to compete with each other and avoid having too bad prods. But as I said early, a demo is mainly technical. The winners are the innovators. It is thus easy to assess who innovates, who doesn't. Regarding the art aspect, which is more subjective, being badly ranked because your art is different may mean nothing.

I think you got it wrong. I implied everyone can give thumbs up/pig/thumb down for all releases, the one with the highest ratio wins.

The "silly little votesystem" I used in TMDC works this way:

Watch all demos.

Watch them again, giving each a score. It doesn't matter what scale you use, as long as they're in relation. So, if one is BILLION TIMES better than the rest, give it BILLION TIMES more points.

Then the votesystem scales your points down so that you have exactly 1.0 points of total vote value.

I know the system is imperfect, most likely 'gameable', especially if the competition is tight (just give 1 to one entry and 0 to the rest - assuming everyone elses' scores are divided equally between 10 entries, your choise will stand out).

On the other hand, it generates fancy values out of relatively small number of judges.

Regardless whether the system was fair or not, I think the order of demos ended up, if not completely, then almost correct. And that's a matter of opinion, of course..

Knoeki: I missread. Thanks. But my point is that if you have more than three prods that have different qualities, there mut be more than one grades available.

sol_hsa: you're right it is ordinal ranking, not cardinal. I don't get your "scaling down" approach. Is it a normalization? Do you mean that if there are three prods (Prod 1, 2, 3) and I rank Prod 1 higher than Prod 2 and Prod 2 higher than Prod 3, you will transform my vote as 1,0,0?