Total Posts:17|Showing Posts:1-17
Jump to topic:

Mafia: Rated Games

tvellalott
Posts: 10,864
Add as Friend
Challenge to a Debate
Send a Message
8/4/2013 7:51:41 PM
Posted: 3 years ago
Hi guys,
As some of you might know, a group of us (I say us, but I actually haven't contributed any of the work) as in the process of doing mafia ELO scores based on as many completed games as possible (some games can't be included, like FREEDO's two games for example.)

Anyway, I don't think it's going to give an accurate depiction of who did well in a particular game, although it's a good start (and you'll see that players who are considered good are among the player who have the highest ELO score.) For one, you can play very well and be part of the losing faction or play poorly and be part of the winning faction. Then you have to take into account that some games are poorly balanced and that people replace in multiple times...

Mafia overall, is a clusterfvck. Some games are excellent and some games are atrocious.

That's where my idea of rated games come in. Rated games would work sort of like beginner games. You run a series of semi-open (that is, players have a list of possible roles but not specifically which ones) and possibly themed (this idea is still in infancy, fake claiming is definitely a mafia skill but it might be hard to balance) games. Voting and night actions would have static points. Townies get points for being part of a scum lynch for example. Cops get points for investigating scum, perhaps even make it so they get more points for discovering scum earlier. The mod then awards points subjectively based on performance. Good analysis, good scum hunting, good WIFOM...
That sort of thing.

What say you DDO?
"Caitlyn Jenner is an incredibly brave and stunningly beautiful woman."

Muh threads
Using mafia tactics in real-life: http://www.debate.org...
6 years of DDO: http://www.debate.org...
Lucky_Luciano
Posts: 4,350
Add as Friend
Challenge to a Debate
Send a Message
8/4/2013 8:02:17 PM
Posted: 3 years ago
I think elo is good enough. If you are a good player and deserve to win then you will win more than you lose and given enough games your elo will float to the top. If you are a bad player who deserves to lose then you will lose more than you win and given enough games your elo will sink to the bottom.
"Age is not important" - Airmax 2014
"Australia... is that a place?" - Airmax 2014
tvellalott
Posts: 10,864
Add as Friend
Challenge to a Debate
Send a Message
8/4/2013 8:10:22 PM
Posted: 3 years ago
At 8/4/2013 8:02:17 PM, Lucky_Luciano wrote:
I think elo is good enough. If you are a good player and deserve to win then you will win more than you lose and given enough games your elo will float to the top. If you are a bad player who deserves to lose then you will lose more than you win and given enough games your elo will sink to the bottom.

That's obviously not how it works.
ELO is a personal score. Mafia ELO based solely on win/loss doesn't factor in personal performance at all, thus it's not accurate. I'm proposing something which factors in personal performance on a much larger and more accurate scale.

Do you not like the idea of rated games?
"Caitlyn Jenner is an incredibly brave and stunningly beautiful woman."

Muh threads
Using mafia tactics in real-life: http://www.debate.org...
6 years of DDO: http://www.debate.org...
Lucky_Luciano
Posts: 4,350
Add as Friend
Challenge to a Debate
Send a Message
8/4/2013 8:12:28 PM
Posted: 3 years ago
At 8/4/2013 8:10:22 PM, tvellalott wrote:
At 8/4/2013 8:02:17 PM, Lucky_Luciano wrote:
I think elo is good enough. If you are a good player and deserve to win then you will win more than you lose and given enough games your elo will float to the top. If you are a bad player who deserves to lose then you will lose more than you win and given enough games your elo will sink to the bottom.

That's obviously not how it works.
ELO is a personal score. Mafia ELO based solely on win/loss doesn't factor in personal performance at all, thus it's not accurate. I'm proposing something which factors in personal performance on a much larger and more accurate scale.

Yes ELO is a personal score, but if you perform well on a consistent basis, i.e. you are better than the opposition, then you will win more often than you lose. This isn't always the case, but it will be more often than not.

Do you not like the idea of rated games?

It's not that I don't like the idea of rated games, it's that I don't like stagnation in creative development. As seen with the games I mod, I only really enjoy games that are innovative to some degree and I feel like standardized games will work against creative progress.
"Age is not important" - Airmax 2014
"Australia... is that a place?" - Airmax 2014
tvellalott
Posts: 10,864
Add as Friend
Challenge to a Debate
Send a Message
8/4/2013 9:01:01 PM
Posted: 3 years ago
At 8/4/2013 8:12:28 PM, Lucky_Luciano wrote:
At 8/4/2013 8:10:22 PM, tvellalott wrote:
At 8/4/2013 8:02:17 PM, Lucky_Luciano wrote:
I think elo is good enough. If you are a good player and deserve to win then you will win more than you lose and given enough games your elo will float to the top. If you are a bad player who deserves to lose then you will lose more than you win and given enough games your elo will sink to the bottom.

That's obviously not how it works.
ELO is a personal score. Mafia ELO based solely on win/loss doesn't factor in personal performance at all, thus it's not accurate. I'm proposing something which factors in personal performance on a much larger and more accurate scale.

Yes ELO is a personal score, but if you perform well on a consistent basis, i.e. you are better than the opposition, then you will win more often than you lose. This isn't always the case, but it will be more often than not.

I agree and as I said, the current state of the mafia ELO does reflect good players being higher than poor players. I'm not saying we SHOULDN'T do the ELO (especially when so much work as already been done). There are some concerns at the moment though about things like how to factor in replacements. I had an idea and I thought I'd come to the forums and see what people thought.

Do you not like the idea of rated games?

It's not that I don't like the idea of rated games, it's that I don't like stagnation in creative development. As seen with the games I mod, I only really enjoy games that are innovative to some degree and I feel like standardized games will work against creative progress.

I'm certainly not saying that ALL games should meet my rated format. Some of the best games are, as you say, the more innovative ones. I'm suggesting a series of games that runs outside of the mafia mod list that are thoroughly rated. I'm thinking 'mafia championships!' in the future. :P
"Caitlyn Jenner is an incredibly brave and stunningly beautiful woman."

Muh threads
Using mafia tactics in real-life: http://www.debate.org...
6 years of DDO: http://www.debate.org...
tvellalott
Posts: 10,864
Add as Friend
Challenge to a Debate
Send a Message
8/4/2013 9:02:19 PM
Posted: 3 years ago
At 8/4/2013 8:12:45 PM, daytonanerd wrote:
When, exactly, do you think we will get mafia ELO scores?

Don't hold your breath. I have no idea really. There are actually two groups of games; the ones in the 'Games' forum and the old ones in the 'Misc' forums. The misc games haven't even been touched, though I did a lot of work back in the day to bring them all together. The newer Games forum games are only partially complete.

Could still be several weeks.
"Caitlyn Jenner is an incredibly brave and stunningly beautiful woman."

Muh threads
Using mafia tactics in real-life: http://www.debate.org...
6 years of DDO: http://www.debate.org...
FourTrouble
Posts: 12,759
Add as Friend
Challenge to a Debate
Send a Message
8/4/2013 10:46:39 PM
Posted: 3 years ago
I think ELO would be a decent indication if there were some modifications involved. For example, if you were night-killed on N1, you had practically no influence on the game, so you can't really be held responsible for the outcome, win or lose. Stuff like that.
Buddamoose
Posts: 19,448
Add as Friend
Challenge to a Debate
Send a Message
8/4/2013 11:10:03 PM
Posted: 3 years ago
Anyone who was replaced was awarded a 0 to their ELO score
Anyone who forfeited a night action was automatically docked -.5 points(just once, even if multiple forfeits occurred)
Anyone who participated and played their part, but didn't necessarily stand out, was awarded a 1
Anyone who participated, played their part, and had a crucial night action that was beneficial to their side, was awarded +.5 points.
Anyone who participated, played their part, and provided solid analysis was given +.5 points.

Solid analysis being analysis that not only pinpointed mafia accurately, but led to said mafia's downfall as well.

Winner automatically recives +1 point.

This is the basic gist of how I rated my last game via ELO. I found that a 0-3 scale was far more better to judge then a 0-5, as 0-5, when a 1 is the base score, is far too much room on the +base side.
"Reality is an illusion created due to a lack of alcohol"
-Airmax1227

"You were the moon all this time, and he was always there to make you shine."

"Was he the sun?"

"No honey, he was the darkness"

-Kazekirion
Buddamoose
Posts: 19,448
Add as Friend
Challenge to a Debate
Send a Message
8/4/2013 11:11:47 PM
Posted: 3 years ago
^^ of course the above can be far more refined, but in my honest opinion, its a good start to at least a semi-objective scoring system, while still leaving some wiggle room for subjectivity on the mod's or outside specators/judges parts. Thus allowing for compensation for players that say, lost, but played a great game.
"Reality is an illusion created due to a lack of alcohol"
-Airmax1227

"You were the moon all this time, and he was always there to make you shine."

"Was he the sun?"

"No honey, he was the darkness"

-Kazekirion
bossyburrito
Posts: 14,075
Add as Friend
Challenge to a Debate
Send a Message
8/4/2013 11:21:47 PM
Posted: 3 years ago
At 8/4/2013 11:10:03 PM, Buddamoose wrote:
Anyone who was replaced was awarded a 0 to their ELO score
Anyone who forfeited a night action was automatically docked -.5 points(just once, even if multiple forfeits occurred)
Anyone who participated and played their part, but didn't necessarily stand out, was awarded a 1
Anyone who participated, played their part, and had a crucial night action that was beneficial to their side, was awarded +.5 points.
Anyone who participated, played their part, and provided solid analysis was given +.5 points.

Solid analysis being analysis that not only pinpointed mafia accurately, but led to said mafia's downfall as well.

Winner automatically recives +1 point.

This is the basic gist of how I rated my last game via ELO. I found that a 0-3 scale was far more better to judge then a 0-5, as 0-5, when a 1 is the base score, is far too much room on the +base side.

How is this considered to be Elo? I thought Elo was based off of wins and losses, with the wins and losses being given different weightings based on the opponent.
#UnbanTheMadman

"Some will sell their dreams for small desires
Or lose the race to rats
Get caught in ticking traps
And start to dream of somewhere
To relax their restless flight
Somewhere out of a memory of lighted streets on quiet nights..."

~ Rush
Buddamoose
Posts: 19,448
Add as Friend
Challenge to a Debate
Send a Message
8/4/2013 11:26:11 PM
Posted: 3 years ago
At 8/4/2013 11:21:47 PM, bossyburrito wrote:
At 8/4/2013 11:10:03 PM, Buddamoose wrote:
Anyone who was replaced was awarded a 0 to their ELO score
Anyone who forfeited a night action was automatically docked -.5 points(just once, even if multiple forfeits occurred)
Anyone who participated and played their part, but didn't necessarily stand out, was awarded a 1
Anyone who participated, played their part, and had a crucial night action that was beneficial to their side, was awarded +.5 points.
Anyone who participated, played their part, and provided solid analysis was given +.5 points.

Solid analysis being analysis that not only pinpointed mafia accurately, but led to said mafia's downfall as well.

Winner automatically recives +1 point.

This is the basic gist of how I rated my last game via ELO. I found that a 0-3 scale was far more better to judge then a 0-5, as 0-5, when a 1 is the base score, is far too much room on the +base side.

How is this considered to be Elo? I thought Elo was based off of wins and losses, with the wins and losses being given different weightings based on the opponent.

Well perhaps not "ELO" specifically, but a ratings system to be sure. I think that people are in part, like me, taking ELO to mean rating system, or taking ELO, to mean ELO like is present here on DDO.

For mafia a straight up ELO system like is present for debates here is wholly unfeasible. It works with debates because you have just one opponent and thus a clear way to judge how far above or below the skill level of your opponent you are. But in mafia games there are so many players, and multiple people on each side, that it would simply be too muddled to make such a system and have it work.

I'm calling it ELO, by namesake only, when truthfully, the actual basis of an ELO system, is, as I said, in my opinion, unfeasible.
"Reality is an illusion created due to a lack of alcohol"
-Airmax1227

"You were the moon all this time, and he was always there to make you shine."

"Was he the sun?"

"No honey, he was the darkness"

-Kazekirion
drafterman
Posts: 18,870
Add as Friend
Challenge to a Debate
Send a Message
8/5/2013 7:50:17 AM
Posted: 3 years ago
@TV:

I love the idea of rated games, though I am thinking of it more as distinct from the ELO effort going on and would have its own scoring system.

My thoughts:

1. There are no replacements. Period. A person is a player in the game until they die. If they are inactive, oh well.
Corollary to 1: Mafiosos can no longer direct the actions of inactive players. If a Mafioso is inactive, they can't use their role and can't perform the NK.

2. All possible roles will have to be codified, including how points are to be tallied for their use. (I.e. you get points for investigating scum, but how many?)

3. The game set-ups should be based on approved designs to reduce imbalance issues.
F-16_Fighting_Falcon
Posts: 18,324
Add as Friend
Challenge to a Debate
Send a Message
8/5/2013 9:41:36 AM
Posted: 3 years ago
At 8/5/2013 7:50:17 AM, drafterman wrote:
@TV:

I love the idea of rated games, though I am thinking of it more as distinct from the ELO effort going on and would have its own scoring system.

My thoughts:

1. There are no replacements. Period. A person is a player in the game until they die. If they are inactive, oh well.
Corollary to 1: Mafiosos can no longer direct the actions of inactive players. If a Mafioso is inactive, they can't use their role and can't perform the NK.

2. All possible roles will have to be codified, including how points are to be tallied for their use. (I.e. you get points for investigating scum, but how many?)

3. The game set-ups should be based on approved designs to reduce imbalance issues.

The game will die if there are no replacements. Why not penalize the player by kicking them out rather than let the game become sub-optimal?
drafterman
Posts: 18,870
Add as Friend
Challenge to a Debate
Send a Message
8/5/2013 10:08:12 AM
Posted: 3 years ago
At 8/5/2013 9:41:36 AM, F-16_Fighting_Falcon wrote:
At 8/5/2013 7:50:17 AM, drafterman wrote:
@TV:

I love the idea of rated games, though I am thinking of it more as distinct from the ELO effort going on and would have its own scoring system.

My thoughts:

1. There are no replacements. Period. A person is a player in the game until they die. If they are inactive, oh well.
Corollary to 1: Mafiosos can no longer direct the actions of inactive players. If a Mafioso is inactive, they can't use their role and can't perform the NK.

2. All possible roles will have to be codified, including how points are to be tallied for their use. (I.e. you get points for investigating scum, but how many?)

3. The game set-ups should be based on approved designs to reduce imbalance issues.

The game will die if there are no replacements. Why not penalize the player by kicking them out rather than let the game become sub-optimal?

The only alternative I see is to modkill them. Allowing replacements introduces as yet unresolved issues regarding how to grant a score to the player. So far I have yet to see an acceptable solution.

Given that most replacements happen as a result of a handful of players, I think this issue can be mostly avoided by denying those players entrance into such games.
tvellalott
Posts: 10,864
Add as Friend
Challenge to a Debate
Send a Message
8/6/2013 8:25:30 PM
Posted: 3 years ago
I think rated games would probably have to have significantly longer phases and smaller numbers of players. Like 240 hour DP's and 72-94 hour NP's with 9 players. I know personally that I can't play mafia because my ability to be active is so sporadic, to the point where even though I want to play, I don't, because I have had to be replaced so often.

The other problem with replacements in a rated game is that you bring with you unavoidable knowledge from your previous role. If you're a townie being replacing into town, unless you were vanilla (which, unfortunately, is a role we don't use as often as we should) you are gonna have some knowledge. Even worse if you're a townie being replaced into mafia.

I think lack of activity should be punished to some extent within the rated game. If you miss posting for a day phase, you should be penalised. This brings us back to a discussion we've been having about people playing a certain way once ELO is live.

The only solution I have to the replacement problem is that each player is rated on a phase by phase basis. Every player should have the opportunity to get the same number of points per round, regardless of their role. Vanilla's would have to have some sort of handicap on their voting to counter balance power roles and the points associated with their night actions.

I like the Mafia Scum Matrix6 setup (http://wiki.mafiascum.net...) as a possible way we could run our Rated Games (should they ever eventuate) but I've thrown in

So you have the following 9 possible scenarios; all 7 townies, 2 scum:
1 Townie Jail Keeper, 6 Vanilla, 2 Goons
1 Townie Cop, 1 Townie Doctor, 5 Vanilla, 1 Mafia Roleblocker, 1 Goon
1 1-Shot Bulletproof Townie, 1 Townie Tracker, 5 Vanilla, 2 Goons
1 Townie Jail Keeper, 1 1-Shot Bulletproof Townie, 5 Vanilla, 1 Mafia Roleblocker, 1 Goon
1 Townie Cop, 6 Vanilla, 2 Goons
1 Townie Doctor, 1 Townie Tracker, 5 Vanilla, 2 Goons

They play infinitely more games over there than we do here and this is considered the most balanced semi-open setup format conceived of yet.
"Caitlyn Jenner is an incredibly brave and stunningly beautiful woman."

Muh threads
Using mafia tactics in real-life: http://www.debate.org...
6 years of DDO: http://www.debate.org...