Forum Brawl #3 - Suggestions and Feedback

[OldTrees]

How about:

-There are no public standings
-BMs inform via secret section the positions of each submission, after the round is over, as well as any expert votes they recieved?

This imo removes strategic voting, while still giving Brawlers a feedback about how well received their submissions were.

Nope. Strategic voting still easily exists.


I know that it may be hard to believe off hand. However consider what you know of me and the time I devoted to looking for a solution. There is no way to eliminate the possibility of strategic voting without exerting fictional levels of mental control.

Edit: I take it back. I just figured a way to strategically vote around even mindwipes. The only way to eliminate strategic voting is by eliminating the voter. Brawl cannot exist without voters.



No, a better solution is to focus on removing/reducing the desire to strategically vote. Look at me. Even in Brawl 2 when I was a brawler, I mentioned ways people could strategically vote. Yet I did not strategically vote and people trusted I would not strategically vote. Why? Because a strategical vote would defeat the purpose of the event in my mind.

[Zawadx]

How? It can't be strategic, as you have no idea how other teams are doing.

[andretimpa]

How about:

-There are no public standings
-BMs inform via secret section the positions of each submission, after the round is over, as well as any expert votes they recieved?

This imo removes strategic voting, while still giving Brawlers a feedback about how well received their submissions were.

Nope. Strategic voting still easily exists.


I know that it may be hard to believe off hand. However consider what you know of me and the time I devoted to looking for a solution. There is no way to eliminate the possibility of strategic voting without exerting fictional levels of mental control.

Edit: I take it back. I just figured a way to strategically vote around even mindwipes. The only way to eliminate strategic voting is by eliminating the voter. Brawl cannot exist without voters.



No, a better solution is to focus on removing/reducing the desire to strategically vote. Look at me. Even in Brawl 2 when I was a brawler, I mentioned ways people could strategically vote. Yet I did not strategically vote and people trusted I would not strategically vote. Why? Because a strategical vote would defeat the purpose of the event in my mind.

It depends on what you are willig to call strategic voting. On my system, thinking in a purely game theoretical way the best strategy is
Vote on your team and randomize your other vote. If everyone does this then voting becomes completely random and the rounds get decided by outside voters most of the time. That means that voters can only use their votes to introduce noise in the system. Since wagers can amplify the results of the noise (and you have no idea where they happened), this seems like a very poor strategy for me.

If we are willinv to believe that people would do that we mayas well distribute forum icons based on cointoss results.

Edit: I just realised that since outside votes are visible they are an information source. So the best strategy is actually to bias your randomization in favour of those voted the least by the rest of the community.

An interesting (curious imo) problem becomes then that we have no idea of how many honest and strategic voters we have and they would tend to vote in opposite ways (assuming the votes of the rest of the community reflect quality). So the best strategy could be sending average submissions (not good nor bad), which sounds closer to a minority game.

So yes, this system is not worth the bother (unless the community votes by PMs, which is very inconvenient).

[theelkspeaks]

When I (and many others) discuss strategic voting, we're primarily talking about players who deliberately withhold votes from the first place team (but still vote legitimately among all remaining teams).  I don't think that full blown strategic optimization is likely, but I am willing to believe that some players would deliberately avoid voting for their most direct competitors. It's a minor form of strategic voting, but it's one that can have a large impact.

To be perfectly honest, I don't think this sort of voting affected this Brawl much, if at all - most if not all of the votes my team lost after being in the lead I can understand after looking at the submissions.  I think a couple of our submissions were better than they finished, but I also think that's some combination of bias based on liking "what belongs to my team" and a difference in taste with other members leading to a submission I liked less finishing high.

That said, the possibility of strategic voting and the bad feelings associated with it certainly made our team rather unhappy during the Brawl.  I'm trying to come up with suggestions that would reduce the appearance of strategic voting, even if it's impossible to eliminate or prevent it altogether.  Under my anonymous submissions model in the later rounds, it would be much harder for accusations of strategic voting to have any merit, whatever mathematical possibilities of strategic voting might still exist.

[Calambar]

For me the only way to prevent strategic voting is to have an honest, unbiased, unrelated to any of the teams expert panel. But it has already been suggested here and not approved by Nae as far as I remember. Yet I still support this idea.

[farscape]

Expert panel vs public voting can have huge differences and it is hard to determine which is better. Both can be manipulated by playing onto certain bias factors in taste and both can be corrupted too.

Just a striking real-life example:
http://www.bbc.com/news/entertainment-arts-27371264

The latest Eurovision figures have revealed the differences between the public phone vote and the jury vote at Saturday night's competition.

The UK phone vote gave Poland's Donatan and Cleo top marks but the jury thought they were the worst act of the night.

[Calambar]

Well, there is a difference between strategic voting and unfair (for most people) voting. And I don't think something must be "better" because most of people think so. Depends on the criteria. When it comes to that Polish song (as I am Pole myself) I can say that this song is lame, lot of boobs and simple (in a negative way) lyrics. So I kind of agree with that "experts".

[OldTrees]

@andretimpa
Yes, anti-quality is the easiest strategic voting technique and it requires no information outside of the voter's head (so it works even if all votes are hidden). Since it is a simple technique to derive and requires no outside information. This technique can be preformed even during fictional levels of anti-strategic voting efforts.

But how small a percentage of brawler votes are strategic votes? Now, I do not pretend to read minds nor do I pretend to be an insider on the culture of the other roles. However I do know that the CIA breeds a meritocracy culture. From this I would assume that most votes from Card Designers would be merit based votes rather than strategic votes.

So while strategic voting can exist, I think we might be able to trust the brawlers to vote honestly.

@theelkspeaks
I agree. Strategic voting is mostly theoretical possible rather than actually impacting the event. Personally I attribute it to 2 factors:
1) Voters not seeing the voting results until votes are locked in. [The lack of this caused problems in Brawl 2]
2) Votes are not anonymous. [There is a Psychological bias towards honesty when being observed]

@Krzysiekxd
I too like the idea of an expert panel combined with popular voting. But I do not know if we have enough people to have a full expert panel (that only votes on their areas of expertise) and have enough brawlers for brawl teams. I think the Artist role was truly blessed when Vrt was on the expert panel(Great feedback was given).

[OldTrees]

Pardon the double post but I felt this needed to be separated.

Raw Point ranges (difference between highest scorer and lowest scorer of that role) ranged from 18 to 32 with an average of 24.
The Auction should be reformed. A 10 point cost for the highest scorer is not sufficient.

andretimpa cost 10, was the highest earner in his role, and still beat the 3rd earner that cost 1 and the 2nd highest earner that cost 4.
eljoemo cost 1 and was the highest earner in his role.
darkripper cost 2 and was the highest earner in his role, and still beat the 3rd earner that cost 1 and the 2nd highest earner that cost 3.
theelkspeaks cost 4 and was the highest earner in his role, and still beat the 3rd earner that cost 1 and the 2nd highest earner that cost 6.
sasuk3 cost 1 and was the highest earner in his role.
Absol cost 1 and was the highest earner in his role.

In most of these cases the highest earner didn't cost anywhere near 10 point.

[Naesala]

Pardon the double post but I felt this needed to be separated.

Raw Point ranges (difference between highest scorer and lowest scorer of that role) ranged from 18 to 32 with an average of 24.
The Auction should be reformed. A 10 point cost for the highest scorer is not sufficient.

andretimpa cost 10, was the highest earner in his role, and still beat the 3rd earner that cost 1 and the 2nd highest earner that cost 4.
eljoemo cost 1 and was the highest earner in his role. This just shows Boss's didn't have a good idea of value
darkripper cost 2 and was the highest earner in his role, and still beat the 3rd earner that cost 1 and the 2nd highest earner that cost 3.
theelkspeaks cost 4 and was the highest earner in his role, and still beat the 3rd earner that cost 1 and the 2nd highest earner that cost 6.
sasuk3 cost 1 and was the highest earner in his role. Dropped out, was covered by team effort
Absol cost 1 and was the highest earner in his role. Dropped out, was covered by team effort

In most of these cases the highest earner didn't cost anywhere near 10 point.

Most of your examples did not even come close to the 10 point cost bargain, so I don't quite understand. And the cap for bidding shouldn't make it so even the highest earners come out to earning the same as everyone else. We could try a higher cap, but what would you suggest, given the only person who's stats we can really say matters are andre's?

[MyNameIsJoey]

Pardon the double post but I felt this needed to be separated.

Raw Point ranges (difference between highest scorer and lowest scorer of that role) ranged from 18 to 32 with an average of 24.
The Auction should be reformed. A 10 point cost for the highest scorer is not sufficient.

andretimpa cost 10, was the highest earner in his role, and still beat the 3rd earner that cost 1 and the 2nd highest earner that cost 4.
eljoemo cost 1 and was the highest earner in his role. This just shows Boss's didn't have a good idea of value
darkripper cost 2 and was the highest earner in his role, and still beat the 3rd earner that cost 1 and the 2nd highest earner that cost 3.
theelkspeaks cost 4 and was the highest earner in his role, and still beat the 3rd earner that cost 1 and the 2nd highest earner that cost 6.
sasuk3 cost 1 and was the highest earner in his role. Dropped out, was covered by team effort
Absol cost 1 and was the highest earner in his role. Dropped out, was covered by team effort

In most of these cases the highest earner didn't cost anywhere near 10 point.

Most of your examples did not even come close to the 10 point cost bargain, so I don't quite understand. And the cap for bidding shouldn't make it so even the highest earners come out to earning the same as everyone else. We could try a higher cap, but what would you suggest, given the only person who's stats we can really say matters are andre's?

cough cough, bosses except me

[OldTrees]

Pardon the double post but I felt this needed to be separated.

Raw Point ranges (difference between highest scorer and lowest scorer of that role) ranged from 18 to 32 with an average of 24.
The Auction should be reformed. A 10 point cost for the highest scorer is not sufficient.

andretimpa cost 10, was the highest earner in his role, and still beat the 3rd earner that cost 1 and the 2nd highest earner that cost 4.
eljoemo cost 1 and was the highest earner in his role. This just shows Boss's didn't have a good idea of value
darkripper cost 2 and was the highest earner in his role, and still beat the 3rd earner that cost 1 and the 2nd highest earner that cost 3.
theelkspeaks cost 4 and was the highest earner in his role, and still beat the 3rd earner that cost 1 and the 2nd highest earner that cost 6.
sasuk3 cost 1 and was the highest earner in his role. Dropped out, was covered by team effort
Absol cost 1 and was the highest earner in his role. Dropped out, was covered by team effort

In most of these cases the highest earner didn't cost anywhere near 10 point.

Most of your examples did not even come close to the 10 point cost bargain, so I don't quite understand. And the cap for bidding shouldn't make it so even the highest earners come out to earning the same as everyone else. We could try a higher cap, but what would you suggest, given the only person who's stats we can really say matters are andre's?

I have not been paying attention to who dropped out. That is a good point.
However I do think that the auction should be restructured to discourage 1pt bids on top players (see eljoemo) and a team should still be able to compete if it chooses a 1pt/brawler team (unlike andretimpa's opponents).

The problem is I do not have enough experience with auction based systems to be able to make a specific suggestion on how it can be improved. War has been using an auction for awhile. Maybe some War veterans will have more ideas on how it can be improved.

Brainstorming ideas (I do not have the experience to predict their effects)
1) Raise the cap
2) Have the auction not impact the starting team score
3) ???