Any idea can be ruined by a ton of people saying 'but what if this/that/the other happens'.
You have to just come up with the best idea/combination of ideas possible, and accept the possibility that things can go astray.
Somewhat relatedly, it's fine to be critical of ideas that are presented, but when all you do is criticize other people's thoughts, while not providing any of your own(or even corrections to what you consider flaws) then it doesn't really mean much.
I think that whatever scoring system is used ultimately needs to have a little complexity, but not too much.
The scoring method of dividing the 36 points evenly was for a number of reasons:
1) 36 is divisible into a lot of numbers easier (making points easy to count).
2) Basing the points on the team size accounts for small imbalances in the setup itself.
3) With the same 12 - 14 players playing in every game, you should be able to gain a strong meta on everyone, making winning as scum much more difficult, therefore rewarding more points to a scum win.
4) Enough games played will give everyone an equal opportunity to be every alignment, and nobody is going to be able to play perfectly enough to have a perfect score, meaning ZDS's fear of unfair advantages is unfounded. If you're afraid of random variance causing unfair advantages, then you shouldn't be playing Magic.
5) The scoring is easy to calculate.
6) If a game happens to be imbalanced, a smaller team will be rewarded with more points.
7) By giving out points only for a win, players can't game the system by using out-of-game reasons for not playing to their win condition.
8) Uncontrollably dying early in a game does not result in lost opportunities for points.
I think that whatever scoring system is used ultimately needs to have a little complexity, but not too much.
Then again (about the part I deleted) quality control doesn't need to know how to do something, just how to test something. I don't mean to be hypercritical. Just trying to help.
I agree with what I left here, though. An ELO system seems fantastic, but possibly overly complex. Seppel's idea (sorry) seems too simple and unfairly luck based.
As a suggestion, could one council member do ALL the randoming for every League game? That would eliminate arguments of tampering or bias (by this I am of course nominating Azrael or maybe DYH).
The scoring method of dividing the 36 points evenly was for a number of reasons:
1) 36 is divisible into a lot of numbers easier (making points easy to count).
2) Basing the points on the team size accounts for small imbalances in the setup itself.
3) With the same 12 - 14 players playing in every game, you should be able to gain a strong meta on everyone, making winning as scum much more difficult, therefore rewarding more points to a scum win.
4) Enough games played will give everyone an equal opportunity to be every alignment, and nobody is going to be able to play perfectly enough to have a perfect score, meaning ZDS's fear of unfair advantages is unfounded. If you're afraid of random variance causing unfair advantages, then you shouldn't be playing Magic.
5) The scoring is easy to calculate.
6) If a game happens to be imbalanced, a smaller team will be rewarded with more points.
7) By giving out points only for a win, players can't game the system by using out-of-game reasons for not playing to their win condition.
8) Uncontrollably dying early in a game does not result in lost opportunities for points.
There's my case on the 36-point method.
The bolded is exactly my point. We'd need to know the structure.
Are we talking 16 Basics bracketing down to two Normals? Or are we talking 4 Basics and a Specialty? Its the scope and frequency that each player gets access to points that make this fair or broken.
The main issue is you can't alternate players on being Scum or Town because it eliminates Gambler's Fallacy. You actually could calculate who's been town before.
So therefore in a 16 Basic system, bracketing to 8, then to 4, then 2 or a Normal, the player that randomed into Scum would have huge potential to have a ton more points. Someone else work up the stats from my model? Many of you are more well versed than I will ever be on that sort of thing.
Private Mod Note
():
Rollback Post to RevisionRollBack
My Clan Is Dead.... Long Live The Izzet! Johnny, born and raised. Always lookin' for the Next Level Combo. Thanks to Bornover of FHLS for the banner!
"As a suggestion, could one council member do ALL the randoming for every League game? That would eliminate arguments of tampering or bias (by this I am of course nominating Azrael or maybe DYH)." -Artifice
in response to this, I'd almost reccomend using a mod from another section of the forum, concidering this league should include both these people as players.
How many league players would be needed? 8 teams of 12 = 96 people!?! if its the same 12-14 for every game it's going to be super meta'd (especially with people as RL friends).
Timeframe is important. Maybe setting a Day time limit so games have a max timeframe.
I'm amused that [Shalako], of all people, are talking about terrible reasoning. "Hey, PF's RVS daykill must be serious, therefore I should shoot him to prove that I have a daykill!"
2011: Best Mafia Performance (Individual) - Best Newcomer
2012: Best (False?) Role Claim - Worst Town Performance (Group) - Best Mafia Performance (Group) - Best SK Performance - Best Overall Player
2013: Best Non-SK Neutral Performance
2014: Best Town Performance (Individual) - Best Town Performance (Group) - Most Interesting Role - Best Game - Best Overall Player
2015: Worst Mafia Performance (Group) - Best Read
2016: Best Town Performance (Group) - Best Town Player - Best Overall Player
2011: Best Mafia Performance (Individual) - Best Newcomer
2012: Best (False?) Role Claim - Worst Town Performance (Group) - Best Mafia Performance (Group) - Best SK Performance - Best Overall Player
2013: Best Non-SK Neutral Performance
2014: Best Town Performance (Individual) - Best Town Performance (Group) - Most Interesting Role - Best Game - Best Overall Player
2015: Worst Mafia Performance (Group) - Best Read
2016: Best Town Performance (Group) - Best Town Player - Best Overall Player
I would say using ELO is not gonna work. We have been trying to get ELO to work on the bot for years, but it just hasn't turned out pretty. Maybe I should ask Xyl how exactly his scoring system works.
2011: Best Mafia Performance (Individual) - Best Newcomer
2012: Best (False?) Role Claim - Worst Town Performance (Group) - Best Mafia Performance (Group) - Best SK Performance - Best Overall Player
2013: Best Non-SK Neutral Performance
2014: Best Town Performance (Individual) - Best Town Performance (Group) - Most Interesting Role - Best Game - Best Overall Player
2015: Worst Mafia Performance (Group) - Best Read
2016: Best Town Performance (Group) - Best Town Player - Best Overall Player
I would say using ELO is not gonna work. We have been trying to get ELO to work on the bot for years, but it just hasn't turned out pretty. Maybe I should ask Xyl how exactly his scoring system works.
While I understand that Xyls system works, it unequally gives points to quantity over quality. I recently reclaimed 1st in the setup ss3 (for non irc users, its a 3 player game with 1 mafia, 1 townie, and 1 super saint) in a close race of :
-XylBot- 1 | Creampuffeater | 745 points | 54 wins, 23 losses, 0 draws
-XylBot- 2 | WPPWAH | 744 points | 140 wins, 102 losses, 0 draws
and while I understand that having a ton of wins over a ton more games, my 70.1 win percentage is way higher than WPPWAH's 57.8 win percentage. Now obviously, quantity must have some importance (or people who are 1-0 or 2-0 would dominate every list), but I dont think Xyl's is the way to go.
Xyls total list (on irc, !xmafia top10, which I wont quote here) also has some instances of lower percentage being higher than a slightly higher percentage (everybody in the t10 basically has between a 57 and a 55% win rate).
Private Mod Note
():
Rollback Post to RevisionRollBack
Quote from hotshizzle »
<hotshizle> WINE IN FRONT OF MEAL
<hotshizle> i think
Would something like this work? (I haven't done the math to verify if it's fair)
+2 for winning
+2 for MVP
+1 after winning for each past victory in a row (eg: a player who wins three games in a row, then loses one, then wins another will obtain 2+3+4+0+2 = 11 points)
-1 after losing for each past loss in a row (eg: a player who loses three games in a row, then wins one, then loses another will end up with 0-1-2+2-0 = -1 point)
I've never liked losing points, so nix the last line, but everything else sounds fine. Maybe +1 for MVP so that it doesn't push a single player that far out in front. And actually the +1 streak multiplier doesn't sound fair because it punishes losses in the middle of the league more than ones at the beginning or end.
I share Zinda's/Yanni's chief complaints about this, as well as wondering if a situation will present itself that town will vig/scum+SK will kill high-point players for their own self-preservation in the season rankings.
I also feel like the division of ranked/unranked games might serve to divide the community. Maybe I'm panicking a bit prematurely, but situations like this always strike me as a hope for the best, expect the worst type thing.
Couple of factors that haven't been mentioned with regard to scoring, is the skewed probability of mafia/town winning depending on the design, and how little an individual players' skill often matters to affect that.
I remember seeing a spreadsheet that LJustus or Jsexton used to run, keeping tabs on every players' win ratio. It really didn't correlate to playskill, at all. So for that reason, I think eliminating points townie wins makes sense, as does emphasizing points for playing well despite the end result.
Adding points for being night-killed makes a lot of sense though.
Do we think an announcement not to factor the scoring system into your play as poor sportsmanship will curb the worst excesses, or will we have to keep folks' scores out of view to make it more difficult to calculate?
Not letting people see scores (other than their own?) wouldn't do anything. If we have a spread sheet/ranking thread with the players ranks (without points) it wouldn't stop Person X from vigging person Y simply because person Y is one or two higher in the overall rank than Person X.
EDIT: and if you don't have the ranking chart/scores public it defeats the entire point of having a league.
Private Mod Note
():
Rollback Post to RevisionRollBack
Originally Posted by Arcadic View Post
scumbag
Want Higher Level Card Evaluation? Visit Diestoremoval.com
Are we really that worried that all of us are just massive dicks?
I think the best thing is to start with the detailed point system and modify over time. Obviously this won't be perfect, but it's easier to remove point categories than add them on later. It's also easier to evaluate "remove this point" than judge the ramifications of adding a category to score.
Also, it's probably best to stick with basic balanced setups for league play anyways. I think it might be best to run semi-open games where all roles available to the mod are known beforehand as "common" roles. It does not hurt analysis, but it would help controlling balance of the games. Ironing out the oddball roles early would be a good thing.
I also think the games should be small to force a higher turnover and availability, but they should be varying size so players are in a wider variety of setups.
Wouldn't it be fairly easy to fix the "quantity over quality" problems with a lot of these ideas by just having a fixed number of games that each league player plays in as the season?
And a player who, for example, won 5 games out of 20 will have the same score as one who won 5 games out of 10 (which means quantity compensates for lack of quality, which is not something I want to see, personally).
As TFF above me said, give each player a fixed number of games. This is easily accomplished.
I would also like to point out large games in general and Specialties in particular tend to favour scum (as I showed here and here), whereas no type of game significantly favours the town.
Then if we run 12 - 14 player games and you get 1 point per win, that should be balanced, right?
It works for sports where the players are pros and have no other activities other than training, but for Mafia, making some games mandatory is going to be a pain in many players' asses (because they'll have to either skip other games, or play over their personal limit or under bad RL conditions).
Then don't make any particular game mandatory. For example, you could say each player must participate in 10 league games by the end of the year to complete their season. If a player plays more league games than that, only the first 10 count.
A suggestion: All Mini hosts are encouraged to host their games as a League game. This way, League games don't add further complexity to the queues, and hosts can choose whether or not they want their games to fit into the more structured game type that a League game will require.
I also agree that League games should have semi-open setups.
I think the best option is probably adding a league slot in the upcoming games of different sizes. You simply say "the next league game is a mini" and use the next mini slot with an approved game.
Fixed. We only ran 18 non-Basic games last year, and several of them were too fancy for the league, so 5 seems like a more realistic number. As for why I think we should pick the best of them, it's for entertainment value. The league won't be as fun to watch if a number of players have their scores set in stone early on.
The number wasn't a suggestion, just an example. If 5 sounds better, then that's fine.
But I don't like the "best of" idea at all, as that just goes back to quantity over quality. It's much easier to win 5 out of 20 than 5 out of 5.
We've already had veterans such as DYH and Arimnaes volunteer to begin designing League games, and we'd welcome further volunteers as well.
1. I volunteer. So hard. Please. But to design, not host.
2. I'm skeptical about running league games in addition to the queues. Making the queues even slower than they are sounds like a problem.
So far, we haven't pinned down what the criteria for game designs should be; we could spring for either innovative new setups, a more standardized base that makes the competition playing field more level and allows more participants, or some mixture of the two. If we're using new material, I've got a few ideas for a game myself...
3. Specialties are probably too complex. Most roles should be simple, but you can probably swing one or two "experimental" roles per setup. And I use that term very strictly.
Here's our initial draft for the League's scoring system:
Oooff.
4. I'd much prefer something on the simpler side. The principle of Seppel's 36 point system appeals to me, though in practice I think it favors scum too heavily.
5. I very much like the idea of mod-given points at the end of the game.
6. ELO would be perfect except not enough games get played to make it work. ELO needs at least 2 digits worth of games before it starts working, and preferably high in the 2 digits.
Overall: cool idea, tough to implement. I'll be watching.
2011: Best Mafia Performance (Individual) - Best Newcomer
2012: Best (False?) Role Claim - Worst Town Performance (Group) - Best Mafia Performance (Group) - Best SK Performance - Best Overall Player
2013: Best Non-SK Neutral Performance
2014: Best Town Performance (Individual) - Best Town Performance (Group) - Most Interesting Role - Best Game - Best Overall Player
2015: Worst Mafia Performance (Group) - Best Read
2016: Best Town Performance (Group) - Best Town Player - Best Overall Player
How about this for a scoring system: a point value corresponding to how many of the members of the other team you outlast. (Say, converted to a 20 point factor - so for every scum you outlast as town in a mini, you get 6 or 7 points; for every player you outlast as SK in a normal, you get 1 point; etc.) Since this score should correlate with victory, it obviates the need for that scoring measure. It glosses over individual player contributions a bit, but if a good player contributes to his team's victory and a bad player hinders it, that shouldn't matter so much.
Biggest problem I can see with this scoring system or any other is that the desire to score points throws off in-game considerations. Since it's hard to earn points when you're dead in most of these systems, if a player is trying to get ahead of another and they're scum/town respectively, the former may kill the latter despite it not being in his team's best interest.
I don't like the flag system. It encourages a really strange kind of behavior/would completely change how mafia is played. Unless that's the goal O_o
Also, I don't think getting lynched as a townie is always a negative thing. But I guess that's personal reasons and theories.
This system seems like it could punish people for running gambits which kind of removes some of the creativity you can apply to mafia.
I think you could include negative points for replacing out, getting modkilled etc. a player who wants to preserve their rating for the year (I recommend a rating for the year and a cumulative rating that doesn't get reset at the end of the year) could replace out or get modkilled without a negative impact on their rating.
+1 for winning +1 for outstanding individual play +1 for outstanding team play -1 for losing -1 for extremely bad play
There would be some type of "League Council" to decide who gets what. None of the bonuses or maluses are mutually exclusive uniess they logically have to (eg: you can't win and lose at the same time, but you can lose and play outstandingly), and there is no limit to how many players can get the bonuses for playing well or the malus for playing poorly. The objective is to make the league exiting and suspenseful, by allowing a variety of way to gain points and a variety of possible strategies for league players. Some can try to win a lot of games, some can focus on maximising their scores for each game, etc. The maluses for losing and playing poorly are here so players think twice before joining every league game they can, without actually setting a hard limit that will cause too many scores to freeze before the league is over.
Being a proponent of simple is better with this, I can approve of your scoring system.
Private Mod Note
():
Rollback Post to RevisionRollBack
If you're town and I'm mafia, you've already lost. You just don't know it yet.
I mostly like ZDS's suggestion. -Mafia is too complicated to have assigned bonuses for very specific things without drastically altering how a game is played. But just assigning points for winning or losing isn't enough either, because you can play well and still lose. So, this points-assigned-by-council idea is great; however, ZDS's point system doesn't seem like it would add up enough over the small number of games that can be reasonably played in a year. My suggestion:
The Moderator of each league game gets 4 points per player to distrubute among the players in that game (up to a maximum of 10 to one player). Two other spectators (possibly reviewers of the game) also get 3 points per player to distribute among the players (up to a maximum of 7). The points can be distributed either by each judge's discretion or based on more specific guidlines (maybe one spectator judges teamwork, the other evaluates individual performance, and the Mod scores overall play, for example). -This results in an average player earning 10 points in a game, an awful performance getting 0, and perfect play scoring a 24.
The end of year score could either be the total added up over a fixed number of games, or an average score taken over a variable number of games (with an assigned minimum number).
I like the idea of a "scoring council" a lot, but I thought it would be unpopular.
EDIT
I think what I really like is the idea of Bonus, Penalty, and Judging sections.
Bonus:
+3 for winning
+1 for replacing in
Penalty:
-2 for modkill
-2 for replacing out
Judging:
0-3 for individual effort - focus on role use/accuracy of cases.
0-3 for team effort - focus on pushing successful lynches/playing for the greater good.
The idea being that winning, benefiting the game, playing well, using your role well, and putting your team ahead will gain points while being selfish, breaking rules, dropping games, and just sucking will score low or lose points. I wanted to include modkills/replacements because they will happen.
+1 for winning
+1 for outstanding individual play
+1 for outstanding team play
-1 for losing
-1 for extremely bad play
There would be some type of "League Council" to decide who gets what. None of the bonuses or maluses are mutually exclusive uniess they logically have to (eg: you can't win and lose at the same time, but you can lose and play outstandingly), and there is no limit to how many players can get the bonuses for playing well or the malus for playing poorly.
The objective is to make the league exiting and suspenseful, by allowing a variety of way to gain points and a variety of possible strategies for league players. Some can try to win a lot of games, some can focus on maximising their scores for each game, etc. The maluses for losing and playing poorly are here so players think twice before joining every league game they can, without actually setting a hard limit that will cause too many scores to freeze before the league is over.
I think this is the best scoring system proposed so far.
I like the idea of a "scoring council" a lot, but I thought it would be unpopular.
If something like this is implemented, where scores are assigned by judges evaluating a player's performance (which I think is the only realistic way to fairly score a mafia game), then I have an idea that could benefit that:
-Each player in the game could have a quicktopic or something where they can keep notes and discuss any strategy they might be using in the game. The judges would have access to these (for reading them, not posting in them), so they could better evaluate what each player is doing or attempting to do and where each player's reads are. -It would be a place for power roles to explain their ability choices, for scum to explain any tactics they are using, and for townies to say things like, "I notice player X is suspiciously doing this, but I'm going to wait to see if it continues" or "I'm going to try drawing a reaction from Player Y by doing this" and other things like that. Since there are a lot of behind the scenes choices, info, and strategies in mafia, something like this could be very helpful to judges in scoring a performance.
2011: Best Mafia Performance (Individual) - Best Newcomer
2012: Best (False?) Role Claim - Worst Town Performance (Group) - Best Mafia Performance (Group) - Best SK Performance - Best Overall Player
2013: Best Non-SK Neutral Performance
2014: Best Town Performance (Individual) - Best Town Performance (Group) - Most Interesting Role - Best Game - Best Overall Player
2015: Worst Mafia Performance (Group) - Best Read
2016: Best Town Performance (Group) - Best Town Player - Best Overall Player
My only concern is who would be on the scoring council and what if (not saying anyone will if this comes to happen) someone has personal disdaon for another and refuses to acknowledge they played a good game?
A panel of judges is probably the best bet for a viable, exciting league. A strict mathematical formula is not going to give a result which correlates to the best overall play, just general goodness. You can game an automated system but aware judges can see and dock points for intentional violation of the rubric. It's credible enough for animal shows and the Olympics even though it's not a strict, pure mathematical system. I don't think that's a reasonable or sensible goal given the reasoning of actions being hard to weigh in a clean way.
In the few attempts I've seen do to this elsewhere, there's been concern about the system biasing play and how that would be generally awful. I don't think anyone is likely to play with a system in mind over their own alignment but it can bias neutrality where there was once no motivation. Choices for N1 kills and "tied levels of scumminess" suddenly have a tiebreaker. This is still going to exist in a subjective judging system based on a rubric because it may be advantageous to kill someone early/mid-game on to prevent them from earning points. It's going to be somewhat worse if a system is used that specifically targets out individual game actions as "+" or "-". A possible solution is delaying the updates of the system by a month or so or not revealing the league's point total until the end, but that adds more questions of credibility and kills the point of following your totals with excitement trying to catch up.
Some kind of system you opt into (in case you don't want to be judged) with a rubric that weeds out a split between success in the league and the game seems at least viable. The main problems would be 1) finding people willing to do all that work 2) finding people who are qualified 3) finding people who have integrity 4) still having people left over to actual play in the league. Those are serious problems, but if you had infinite/resources time it would actually work in comparison to a mathematical point system which probably will always turn out a mediocre result.
EDIT: Thinking about this under a 10 point system, you could also do something like have a minimum and a maximum amount of league games. Instead of adding up all your points ever your totals are averaged together. The person with the best average (overall body of work) is the highest rated. This would help against it being a system of "who played the most games?" You would need a sufficiently high minimum number of games that was not too far from the maximum to ensure someone didn't sit on two abnormally good performances and shut everyone else out though.
EDIT2: Instead of averaging all your games and a maximum, maybe something like a minimum and your best five games so you're still incentivized for continuing to try.
EDIT3: Here's what I'm currently playing around with mentally. Something like a 12 to 16 player league. It lasts for a year and the games are all twelve player, keeping them short. At the end of each game, three judges who did not participate in the game weigh in based on the rubric. Judges can be members of the league or outside observers, but they must be smart, credible, trustworthy people and have not participated in the game as a player. At the end of the year the averages are weighted up and the one with the best body of work finishes atop the league with the bragging rights.
2011: Best Mafia Performance (Individual) - Best Newcomer
2012: Best (False?) Role Claim - Worst Town Performance (Group) - Best Mafia Performance (Group) - Best SK Performance - Best Overall Player
2013: Best Non-SK Neutral Performance
2014: Best Town Performance (Individual) - Best Town Performance (Group) - Most Interesting Role - Best Game - Best Overall Player
2015: Worst Mafia Performance (Group) - Best Read
2016: Best Town Performance (Group) - Best Town Player - Best Overall Player
2011: Best Mafia Performance (Individual) - Best Newcomer
2012: Best (False?) Role Claim - Worst Town Performance (Group) - Best Mafia Performance (Group) - Best SK Performance - Best Overall Player
2013: Best Non-SK Neutral Performance
2014: Best Town Performance (Individual) - Best Town Performance (Group) - Most Interesting Role - Best Game - Best Overall Player
2015: Worst Mafia Performance (Group) - Best Read
2016: Best Town Performance (Group) - Best Town Player - Best Overall Player
Basically, what is more valuable: A rubric that everyone has access to with various ways to gain points, or a dynamic evaluation from league judges?
I do not believe there is really need for judges. The moderator of each game should know how the play is going, which player makes the crucial play that decides the game one way or the other. I think it could be 50% mod. and 50% player, as the players know who manipulated them, who they trusted, etc. and whether those assumptions were correct or not.
I also would like maybe just a simple award system where basically every year we give awards to players for various aspects of the game, where the players nominate and the mods of the forum determine the winner. Examples include: Best (Town/Scum/SK/Neutral) Player, Most Outrageous Fake-Claim, Best Town, Best Mafia, Best Game. The mod of each game can do a mini-award for the particular game and the forum votes/nominates for the year.
* 1 point for every win.
You have to just come up with the best idea/combination of ideas possible, and accept the possibility that things can go astray.
Somewhat relatedly, it's fine to be critical of ideas that are presented, but when all you do is criticize other people's thoughts, while not providing any of your own(or even corrections to what you consider flaws) then it doesn't really mean much.
I think that whatever scoring system is used ultimately needs to have a little complexity, but not too much.
1) 36 is divisible into a lot of numbers easier (making points easy to count).
2) Basing the points on the team size accounts for small imbalances in the setup itself.
3) With the same 12 - 14 players playing in every game, you should be able to gain a strong meta on everyone, making winning as scum much more difficult, therefore rewarding more points to a scum win.
4) Enough games played will give everyone an equal opportunity to be every alignment, and nobody is going to be able to play perfectly enough to have a perfect score, meaning ZDS's fear of unfair advantages is unfounded. If you're afraid of random variance causing unfair advantages, then you shouldn't be playing Magic.
5) The scoring is easy to calculate.
6) If a game happens to be imbalanced, a smaller team will be rewarded with more points.
7) By giving out points only for a win, players can't game the system by using out-of-game reasons for not playing to their win condition.
8) Uncontrollably dying early in a game does not result in lost opportunities for points.
There's my case on the 36-point method.
I agree with what I left here, though. An ELO system seems fantastic, but possibly overly complex. Seppel's idea (sorry) seems too simple and unfairly luck based.
As a suggestion, could one council member do ALL the randoming for every League game? That would eliminate arguments of tampering or bias (by this I am of course nominating Azrael or maybe DYH).
The bolded is exactly my point. We'd need to know the structure.
Are we talking 16 Basics bracketing down to two Normals? Or are we talking 4 Basics and a Specialty? Its the scope and frequency that each player gets access to points that make this fair or broken.
The main issue is you can't alternate players on being Scum or Town because it eliminates Gambler's Fallacy. You actually could calculate who's been town before.
So therefore in a 16 Basic system, bracketing to 8, then to 4, then 2 or a Normal, the player that randomed into Scum would have huge potential to have a ton more points. Someone else work up the stats from my model? Many of you are more well versed than I will ever be on that sort of thing.
Johnny, born and raised. Always lookin' for the Next Level Combo. Thanks to Bornover of FHLS for the banner!
Mafia Results, Links, and Stats
in response to this, I'd almost reccomend using a mod from another section of the forum, concidering this league should include both these people as players.
Millionaires, I hear it's good Music (Disclaimer: lyrics not PG-13) Thanks, CC
@All Why not have a mod and score keeper [league Tallier]
Also maybe the points should be used to punish bad "manners". A triple post penalty. A 1000+ word wall. Inactivity. Replacing out. etc.
Further a score keeper could be reviewing for who started what lynch keeping the mod free to tally votes and answer rules questions.
What does ELO Electric Light Opera have to do with all of this?
How many league players would be needed? 8 teams of 12 = 96 people!?! if its the same 12-14 for every game it's going to be super meta'd (especially with people as RL friends).
Timeframe is important. Maybe setting a Day time limit so games have a max timeframe.
12-11? I'm losing track
{мы, тьма}
2012: Best (False?) Role Claim - Worst Town Performance (Group) - Best Mafia Performance (Group) - Best SK Performance - Best Overall Player
2013: Best Non-SK Neutral Performance
2014: Best Town Performance (Individual) - Best Town Performance (Group) - Most Interesting Role - Best Game - Best Overall Player
2015: Worst Mafia Performance (Group) - Best Read
2016: Best Town Performance (Group) - Best Town Player - Best Overall Player
C'mon Iso, you play chess.
But, gotcha, thanks.
{мы, тьма}
2012: Best (False?) Role Claim - Worst Town Performance (Group) - Best Mafia Performance (Group) - Best SK Performance - Best Overall Player
2013: Best Non-SK Neutral Performance
2014: Best Town Performance (Individual) - Best Town Performance (Group) - Most Interesting Role - Best Game - Best Overall Player
2015: Worst Mafia Performance (Group) - Best Read
2016: Best Town Performance (Group) - Best Town Player - Best Overall Player
Come join us in the MTGSalvation chat ||| My trade thread. ||| My Personal Modern Blog: The Fetchlands
Stalker!I mean
{мы, тьма}
2012: Best (False?) Role Claim - Worst Town Performance (Group) - Best Mafia Performance (Group) - Best SK Performance - Best Overall Player
2013: Best Non-SK Neutral Performance
2014: Best Town Performance (Individual) - Best Town Performance (Group) - Most Interesting Role - Best Game - Best Overall Player
2015: Worst Mafia Performance (Group) - Best Read
2016: Best Town Performance (Group) - Best Town Player - Best Overall Player
While I understand that Xyls system works, it unequally gives points to quantity over quality. I recently reclaimed 1st in the setup ss3 (for non irc users, its a 3 player game with 1 mafia, 1 townie, and 1 super saint) in a close race of :
-XylBot- 1 | Creampuffeater | 745 points | 54 wins, 23 losses, 0 draws
-XylBot- 2 | WPPWAH | 744 points | 140 wins, 102 losses, 0 draws
and while I understand that having a ton of wins over a ton more games, my 70.1 win percentage is way higher than WPPWAH's 57.8 win percentage. Now obviously, quantity must have some importance (or people who are 1-0 or 2-0 would dominate every list), but I dont think Xyl's is the way to go.
Xyls total list (on irc, !xmafia top10, which I wont quote here) also has some instances of lower percentage being higher than a slightly higher percentage (everybody in the t10 basically has between a 57 and a 55% win rate).
I think of ELO every time someone mentions ELO.
I've never liked losing points, so nix the last line, but everything else sounds fine. Maybe +1 for MVP so that it doesn't push a single player that far out in front. And actually the +1 streak multiplier doesn't sound fair because it punishes losses in the middle of the league more than ones at the beginning or end.
I also feel like the division of ranked/unranked games might serve to divide the community. Maybe I'm panicking a bit prematurely, but situations like this always strike me as a hope for the best, expect the worst type thing.
P.S. That would be awesome.
I remember seeing a spreadsheet that LJustus or Jsexton used to run, keeping tabs on every players' win ratio. It really didn't correlate to playskill, at all. So for that reason, I think eliminating points townie wins makes sense, as does emphasizing points for playing well despite the end result.
Adding points for being night-killed makes a lot of sense though.
Do we think an announcement not to factor the scoring system into your play as poor sportsmanship will curb the worst excesses, or will we have to keep folks' scores out of view to make it more difficult to calculate?
EDIT: and if you don't have the ranking chart/scores public it defeats the entire point of having a league.
scumbag
Want Higher Level Card Evaluation? Visit Diestoremoval.com
I think the best thing is to start with the detailed point system and modify over time. Obviously this won't be perfect, but it's easier to remove point categories than add them on later. It's also easier to evaluate "remove this point" than judge the ramifications of adding a category to score.
Also, it's probably best to stick with basic balanced setups for league play anyways. I think it might be best to run semi-open games where all roles available to the mod are known beforehand as "common" roles. It does not hurt analysis, but it would help controlling balance of the games. Ironing out the oddball roles early would be a good thing.
I also think the games should be small to force a higher turnover and availability, but they should be varying size so players are in a wider variety of setups.
As TFF above me said, give each player a fixed number of games. This is easily accomplished.
Then if we run 12 - 14 player games and you get 1 point per win, that should be balanced, right?
Then don't make any particular game mandatory. For example, you could say each player must participate in 10 league games by the end of the year to complete their season. If a player plays more league games than that, only the first 10 count.
I also agree that League games should have semi-open setups.
The number wasn't a suggestion, just an example. If 5 sounds better, then that's fine.
But I don't like the "best of" idea at all, as that just goes back to quantity over quality. It's much easier to win 5 out of 20 than 5 out of 5.
1. I volunteer. So hard. Please. But to design, not host.
2. I'm skeptical about running league games in addition to the queues. Making the queues even slower than they are sounds like a problem.
3. Specialties are probably too complex. Most roles should be simple, but you can probably swing one or two "experimental" roles per setup. And I use that term very strictly.
Oooff.
4. I'd much prefer something on the simpler side. The principle of Seppel's 36 point system appeals to me, though in practice I think it favors scum too heavily.
5. I very much like the idea of mod-given points at the end of the game.
6. ELO would be perfect except not enough games get played to make it work. ELO needs at least 2 digits worth of games before it starts working, and preferably high in the 2 digits.
Overall: cool idea, tough to implement. I'll be watching.
{мы, тьма}
2012: Best (False?) Role Claim - Worst Town Performance (Group) - Best Mafia Performance (Group) - Best SK Performance - Best Overall Player
2013: Best Non-SK Neutral Performance
2014: Best Town Performance (Individual) - Best Town Performance (Group) - Most Interesting Role - Best Game - Best Overall Player
2015: Worst Mafia Performance (Group) - Best Read
2016: Best Town Performance (Group) - Best Town Player - Best Overall Player
Biggest problem I can see with this scoring system or any other is that the desire to score points throws off in-game considerations. Since it's hard to earn points when you're dead in most of these systems, if a player is trying to get ahead of another and they're scum/town respectively, the former may kill the latter despite it not being in his team's best interest.
Also, I don't think getting lynched as a townie is always a negative thing. But I guess that's personal reasons and theories.
This system seems like it could punish people for running gambits which kind of removes some of the creativity you can apply to mafia.
I think you could include negative points for replacing out, getting modkilled etc. a player who wants to preserve their rating for the year (I recommend a rating for the year and a cumulative rating that doesn't get reset at the end of the year) could replace out or get modkilled without a negative impact on their rating.
scumbag
Want Higher Level Card Evaluation? Visit Diestoremoval.com
Being a proponent of simple is better with this, I can approve of your scoring system.
The Moderator of each league game gets 4 points per player to distrubute among the players in that game (up to a maximum of 10 to one player). Two other spectators (possibly reviewers of the game) also get 3 points per player to distribute among the players (up to a maximum of 7). The points can be distributed either by each judge's discretion or based on more specific guidlines (maybe one spectator judges teamwork, the other evaluates individual performance, and the Mod scores overall play, for example). -This results in an average player earning 10 points in a game, an awful performance getting 0, and perfect play scoring a 24.
The end of year score could either be the total added up over a fixed number of games, or an average score taken over a variable number of games (with an assigned minimum number).
EDIT
I think what I really like is the idea of Bonus, Penalty, and Judging sections.
Bonus:
+3 for winning
+1 for replacing in
Penalty:
-2 for modkill
-2 for replacing out
Judging:
0-3 for individual effort - focus on role use/accuracy of cases.
0-3 for team effort - focus on pushing successful lynches/playing for the greater good.
The idea being that winning, benefiting the game, playing well, using your role well, and putting your team ahead will gain points while being selfish, breaking rules, dropping games, and just sucking will score low or lose points. I wanted to include modkills/replacements because they will happen.
I think this is the best scoring system proposed so far.
If something like this is implemented, where scores are assigned by judges evaluating a player's performance (which I think is the only realistic way to fairly score a mafia game), then I have an idea that could benefit that:
-Each player in the game could have a quicktopic or something where they can keep notes and discuss any strategy they might be using in the game. The judges would have access to these (for reading them, not posting in them), so they could better evaluate what each player is doing or attempting to do and where each player's reads are. -It would be a place for power roles to explain their ability choices, for scum to explain any tactics they are using, and for townies to say things like, "I notice player X is suspiciously doing this, but I'm going to wait to see if it continues" or "I'm going to try drawing a reaction from Player Y by doing this" and other things like that. Since there are a lot of behind the scenes choices, info, and strategies in mafia, something like this could be very helpful to judges in scoring a performance.
{мы, тьма}
2012: Best (False?) Role Claim - Worst Town Performance (Group) - Best Mafia Performance (Group) - Best SK Performance - Best Overall Player
2013: Best Non-SK Neutral Performance
2014: Best Town Performance (Individual) - Best Town Performance (Group) - Most Interesting Role - Best Game - Best Overall Player
2015: Worst Mafia Performance (Group) - Best Read
2016: Best Town Performance (Group) - Best Town Player - Best Overall Player
540 Peasant cube- Gold EditionSomething SpicyIn the few attempts I've seen do to this elsewhere, there's been concern about the system biasing play and how that would be generally awful. I don't think anyone is likely to play with a system in mind over their own alignment but it can bias neutrality where there was once no motivation. Choices for N1 kills and "tied levels of scumminess" suddenly have a tiebreaker. This is still going to exist in a subjective judging system based on a rubric because it may be advantageous to kill someone early/mid-game on to prevent them from earning points. It's going to be somewhat worse if a system is used that specifically targets out individual game actions as "+" or "-". A possible solution is delaying the updates of the system by a month or so or not revealing the league's point total until the end, but that adds more questions of credibility and kills the point of following your totals with excitement trying to catch up.
Some kind of system you opt into (in case you don't want to be judged) with a rubric that weeds out a split between success in the league and the game seems at least viable. The main problems would be 1) finding people willing to do all that work 2) finding people who are qualified 3) finding people who have integrity 4) still having people left over to actual play in the league. Those are serious problems, but if you had infinite/resources time it would actually work in comparison to a mathematical point system which probably will always turn out a mediocre result.
EDIT: Thinking about this under a 10 point system, you could also do something like have a minimum and a maximum amount of league games. Instead of adding up all your points ever your totals are averaged together. The person with the best average (overall body of work) is the highest rated. This would help against it being a system of "who played the most games?" You would need a sufficiently high minimum number of games that was not too far from the maximum to ensure someone didn't sit on two abnormally good performances and shut everyone else out though.
EDIT2: Instead of averaging all your games and a maximum, maybe something like a minimum and your best five games so you're still incentivized for continuing to try.
EDIT3: Here's what I'm currently playing around with mentally. Something like a 12 to 16 player league. It lasts for a year and the games are all twelve player, keeping them short. At the end of each game, three judges who did not participate in the game weigh in based on the rubric. Judges can be members of the league or outside observers, but they must be smart, credible, trustworthy people and have not participated in the game as a player. At the end of the year the averages are weighted up and the one with the best body of work finishes atop the league with the bragging rights.
I like ZDS' idea of having judges, but I also see nothing wrong with E_P's rubric.
Basically, what is more valuable: A rubric that everyone has access to with various ways to gain points, or a dynamic evaluation from league judges?
{мы, тьма}
2012: Best (False?) Role Claim - Worst Town Performance (Group) - Best Mafia Performance (Group) - Best SK Performance - Best Overall Player
2013: Best Non-SK Neutral Performance
2014: Best Town Performance (Individual) - Best Town Performance (Group) - Most Interesting Role - Best Game - Best Overall Player
2015: Worst Mafia Performance (Group) - Best Read
2016: Best Town Performance (Group) - Best Town Player - Best Overall Player
{мы, тьма}
2012: Best (False?) Role Claim - Worst Town Performance (Group) - Best Mafia Performance (Group) - Best SK Performance - Best Overall Player
2013: Best Non-SK Neutral Performance
2014: Best Town Performance (Individual) - Best Town Performance (Group) - Most Interesting Role - Best Game - Best Overall Player
2015: Worst Mafia Performance (Group) - Best Read
2016: Best Town Performance (Group) - Best Town Player - Best Overall Player
I do not believe there is really need for judges. The moderator of each game should know how the play is going, which player makes the crucial play that decides the game one way or the other. I think it could be 50% mod. and 50% player, as the players know who manipulated them, who they trusted, etc. and whether those assumptions were correct or not.
I also would like maybe just a simple award system where basically every year we give awards to players for various aspects of the game, where the players nominate and the mods of the forum determine the winner. Examples include: Best (Town/Scum/SK/Neutral) Player, Most Outrageous Fake-Claim, Best Town, Best Mafia, Best Game. The mod of each game can do a mini-award for the particular game and the forum votes/nominates for the year.—Lazav
_______________________________________________
Mafia Stats
Summary:
Total Win %: 40%
Total Scum Win %: 60%
Total Town Win %: 20%
Total Neutral Win %: 0%
Retract my last point then.
—Lazav
_______________________________________________
Mafia Stats
Summary:
Total Win %: 40%
Total Scum Win %: 60%
Total Town Win %: 20%
Total Neutral Win %: 0%
...and add in that everyone places the bet privately to an entity not in the game.