Results 1 to 9 of 9

Thread: #8: Macro Meta Crunching (by Helz)

  1. ISO #1
    Ruler of the Universe Thingyman's Avatar Administrator
    Join Date
    May 2015
    Location
    Denmark
    Posts
    9,600
    Community
    PokerNet / 2+2
    Gender

    #8: Macro Meta Crunching (by Helz)

    Article #8: Macro Meta Crunching
    - written by Helz

    Meta hunting could be summarized as getting to know a player's normal behavior, associating it to their alignment in that game, and comparing it to their current play in order to get an impression of what their current alignment is. This is a popular technique because of the inherent differences between town and non-town motivations that result in perceivable changes in behavior that are individual to each player. You could equate these differences to ?tells? in games of poker. A person will behave differently based upon their current hand. The effective use of statistics is crunching data in order to identify statistically significant information gained through enough replication to identify correlation. This correlation points to the sample data XXXXXX suggesting XXXXXX about (in this case) the player the data was taken from.

    By identifying ways to crunch large amounts of data in order to identify patterns that reflect differences in behavior or ?tells? players manipulate this technique to identify a player's alignment. By establishing a large pool of analyzed games players can then simply plug in their method of analysis to a game and apply a calculated level of mathematical certainty to read a player independent of any real analysis.

    Here would be 1 simple example:
    The ?meta? tell for Bob is that in games as scum he dissociates his perspective from his posts when analyzing a player. This is a ?scum tell? because it reflects him posting from a separate perspective as if attempting to hide his perspective and forgetting to translate his analysis. Player X breaks down the word count of some of Bobs games using online tools and identifies that when playing as a town player Bob frequently uses the words ?I, My, and Me? where as in games as scum the count for these words is much smaller in proportion to the total analyzed word count. In contrast in games where Bob was playing as a scum player Bob he more frequently uses the words ?You and We?. Player X has now established correlation between this word count data and Bobs alignment moving a simple meta tell into something that can be mathematically analyzed. With this in mind Player X breaks down the word count of Bobs analysis posts in the current game and identifies that Bob has a lower than normal use of the words ?I, My, and Me? than observed in Bobs ?town games? while the data for his use of ?You and We? match the analyzed data taken from Bobs ?scum games?. Using this tool Player X is able to apply simple statistical formulas to identify a calculated level of mathematical certainty that Bob has a scum alignment in the current game.

    This practice is not limited to breaking down word counts to identify perspective tells. The behavior targeted, method of data collection and different ways of separating sample pools allow for limitless different methods of analysis. Some players use it to identify accounts behind anon games while others use it to clear town or identify patterns that push them to look into a player more as opposed to using it as a substitute for scum hunting.

    I am presenting this information to level the playing field. I know of some players using it to build information on the regular players in their site which allows them to constantly pull out good reads and seem skilled. In my opinion some uses of this is unfair to other players and turns Mafia/Werewolf into something it was not meant to be. Although there are also many nice applications of this that allow for players to improve such as analyzing themselves and potentially prevent this from functioning against them or even counter it by manipulating their posts to throw off people who would use it against them. I will note that this is mostly functional against players who stick to 1 play style because by using different play styles you make analyzing the data of your posts much more complex to do with any level of accuracy.

    On a less related but associated subject this can also be applied to help determine the accuracy of your reads as a player in order to identify what you are good at finding and should therefore focus on in your analysis. To do this after each game just look at the reasoning behind your correct and incorrect reads and chart your progress. You will find patterns emerging over enough games pointing to you almost never being correct when analyzing one aspect or indicator while you may very often be dead on. Gaining understanding of your ability as a player allows for focused effort to improve. It would be equally easy to apply this to identify why you got lynched as scum, why you got lynched as town, or why you were targeted for some night action (As long as you get the player to explain his reasoning post game).

    Cliffs:

    • Data can be taken from a player's games and analyzed to identify meta trends.
    • These trends can then be compared to data taken from a current game to identify specific factors of a player such as alignment.
    • A player can analyze their own games to identify their own tells and improve as a player.
    Last edited by Thingyman; August 27th, 2016 at 05:07 PM.

  2. ISO #2
    Ruler of the Universe Thingyman's Avatar Administrator
    Join Date
    May 2015
    Location
    Denmark
    Posts
    9,600
    Community
    PokerNet / 2+2
    Gender
    Thank you for the article @Helz

    I think it's super interesting. Presumably, most of us do rely on meta to at least some degree when playing mafia, and I for one have never understood the severe dislike some people have for meta reads. But I have certainly never applied it in this way, i.e. used "macro meta reads" or whatever you would call it. I generally just try to get a feel for how game solvey a person usually is as either alignment, what their attitude usually is in certain situations, etc. when applying meta reads. I haven't ever looked at just one very specific data point that can be compared across all games in an easily quantifiable manner such as a word count analysis as you've exemplified here.

    I am curious to more examples of people actually having done this in games - and how successful it was?

  3. ISO #3
    Banned

    Join Date
    Nov 2015
    Location
    Texas
    Posts
    815
    Community
    Homeless
    Gender
    For a long time I was one of those that was very critical of Meta reads. My reasoning was that they are generally used in a way that can not really be verified and as such, becomes a powerful tool for scum. If you ask a player "Why did you say you think player X is town?" and they respond "Because of my meta read, I feel like this player generally more aggressive as town." what you are left with is useless to really read the player saying this. To get an idea of their mindset or how they came to this conclusion you would have to dig through a ton of games and make assumptions as to what games that player was looking at. Because you are unable to follow this line of reasoning you can not put value in either their read or gain anything from them giving that read.

    What changed for me is simply that when I was working on the meta portion of my guide I ended up with the first sentence in this article. "Meta hunting could be summarized as getting to know a player's normal behavior, associating it to their alignment in that game, and comparing it to their current play in order to get an impression of what their current alignment is." I had to acknowledge that it is a totally valid concept and it completely changed my view of meta hunting. I realized that it is not meta hunting that I hate, its how 90% of players use it in games.

    All that aside I would rather not break down specific situations of it being successfully used because it would also require explaining how it was used. I will say that it was really rare that I identified some "dead on" patterns in someones posting that allowed me to see something with near certainty but the times I did find this I felt like I was cheating and took the magic out of the game. I was not all that great at this but there are players that are very good at it and I felt like it was unfair that the majority of player had no idea this kind of thing exists. Gives you something to think about when 1 player constantly nails a specific players alignment every single game.

  4. ISO #4
    Season 6 Champion Beck's Avatar
    Join Date
    Jun 2015
    Location
    Where it's at!
    Posts
    29,611
    Timezone
    UTC-05:00
    Community
    NovoCiv
    Pronouns
    he/him/his/his/himself
    Gender
    This looks really good, gonna give it a read and share my thoughts -- a big thanks in advance.
    leaping up into the air getting juiced up beyond belief

  5. ISO #5
    JAMES MCCLOWN Frog's Avatar
    Join Date
    Nov 2015
    Posts
    28,195
    Community
    murdermystery@gmail
    Gender
    Excellent content here

    Many divergent applications can be drawn from this

    It's a tight rope to walk, explaining theory and divulging tells that, you walked perfectly to get players to start recognizing patterns in (possibly their own) behaviour

    Looking forward to whatever else you lay on the table

  6. ISO #6
    Amberguard Emberguard's Avatar
    Join Date
    Feb 2016
    Posts
    930
    Timezone
    UTC+11:00
    Community
    Neoseeker
    Gender
    It's like when you get close to someone in real life and really get to know them you can see things others can't. Everyone else might think they're fine but you might know them well enough to know they're not feeling too well some days or somethings playing on their mind.

    Meta reading is nothing more then getting to know the person. It's not cheating, it's just something that'll develop regardless of whether you're trying to develop it or not. The more time you spend with someone the better you know how they behave to certain situations. That's where anonymous accounts come in handy. It breaks that connection you have with the person preventing your brain from recognising their personal traits leaving you with pure reads on how people in general would act instead of how that specific person would act.

    The biggest problem people have with meta reading is sometimes they overthink it or try too hard to find some sort of meta on someone without fully understanding the psychology underneath what they're trying to read.
    Last edited by Emberguard; September 3rd, 2016 at 11:51 PM.

  7. ISO #7
    Ruler of the Universe Thingyman's Avatar Administrator
    Join Date
    May 2015
    Location
    Denmark
    Posts
    9,600
    Community
    PokerNet / 2+2
    Gender
    Embeeeeeer! Missed you

  8. ISO #8
    five people have died Apoc's Avatar Game Manager
    Join Date
    Aug 2015
    Location
    Neo-Storpeo
    Posts
    45,462
    Gender
    Quote Originally Posted by Emberguard (#6)
    It's like when you get close to someone in real life and really get to know them you can see things others can't. Everyone else might think they're fine but you might know them well enough to know they're not feeling too well some days or somethings playing on their mind.

    Meta reading is nothing more then getting to know the person. It's not cheating, it's just something that'll develop regardless of whether you're trying to develop it or not. The more time you spend with someone the better you know how they behave to certain situations. That's where anonymous accounts come in handy. It breaks that connection you have with the person preventing your brain from recognising their personal traits leaving you with pure reads on how people in general would act instead of how that specific person would act.

    The biggest problem people have with meta reading is sometimes they overthink it or try too hard to find some sort of meta on someone without fully understanding the psychology underneath what they're trying to read.
    When you use a computer simulation to find out everything about someone without ever talking to them, that's the "cheating" it part

  9. ISO #9
    Bandwagoner Primate's Avatar
    Join Date
    May 2017
    Posts
    159
    Timezone
    UTC-06:00
    Community
    FFA, FE, GLB
    Gender
    Yeah. Meta reads are a huge part of my game once I get familiar with those I am playing with. I don't use software, or even take notes. I just get to know people like someone mentioned above. Without it, we are stuck poking little tidbits for reactions and analyzing vote patterns, which is a little boring to me.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  •  

Keyboard Shortcuts

about us
Mafia Universe is a community hub for people who enjoy playing the forum variant of Mafia (also known as Werewolf). We offer fully automated Mafia games and a wide variety of customized features crafted to optimize your game experience. We also proudly host the Internet's only database of Mafia/Werewolf communities.

We hope you stick around!
Role of the Day
Caroler

The Caroler may each night visit a person and sing them a carol. The carol will inform the targeted player of 3 names, at least one of which is mafia. However, if the targeted person is visiting someone on that night, they will not hear the carol. The Caroler may not choose the same player on consecutive nights.