Epicmafia

Odds versus Social Reads.

over 9 years

I've seen a lot of people suggesting that EpicMafia is a game that fundamentally relies on social interactions; I argue that it is isn't. This game adheres more to mathematics and picking the most likely condition to allow your victory, we can see this in effect on every instance the game enters mislynch and lose.

Let's say that on the day we have two civilians, one doctor, and one Mafia, for a total of four people. The Mafia in this instance should always hope to be considered part of the blue collection because it increases their odds of not being lynched. If they were claimed as a Doctor this would increase the chances of the village winning from 1/3, to 1/2 On the last day. A significant increase on a random pick.

The Mafia's odds of victory go up if the Doctor claims and doesn't receive a baited counter-claim, because if everyone simply claims blue, the Doctor that is cleared shall most likely die anyway; this means the odds of the village winning are only 1/3 (picking between blues).

A player should logically pick the actions which produce a consistent statistical advantage over the other options. Not doing so, is literally not playing to win. When the actions of power roles are taken into consideration, it matters whether they pick a random or not in some circumstances, here's why.

Let's say that the Cop has a situation on night one where they're completely safe to pick a person to investigate, among 6 Or 7, Or any arbitrary number of people, there is likely no bomb in this scenario, a granny, or anything else that'd educate a person to hindsight.

If they choose not to investigate anyone, they have potentially wasted a village resource; there was a decent chance they'd have gotten a guilty report, and if not that, they'd at least have clears for the village to choose. Counter-claiming the cop is often a poor decision for Mafia (statistically), by this example.

There are five people, 2 Are civilians again, one is Mafia, and one is the Cop. In this scenario the Cop always outs, because the village then immediately has a 1/2 Chance of killing Mafia by picking between the inevitable double Cop claim.

In this case, the Mafia should reproduce the exact same results as the Cop does if they choose two innocent, why so? Essentially, one of the civilians is dead, and by no amount is either Cop any more suspect than the other. Going from this, it's possible for Mafia to always devise a strategy which results in a situation where they're likely to end up with a 1/2 Chance of being hammered. The only possibility that they'll have for odds that are better than this is in a set-up which caters to them, in a balanced set-up you'll likely see people always ending up in the 50/50 Situation.

Because by playing the odds, instead of imagining that it is truly possible to understand things from other people's perspective, and by acting exactly the same in every single game, no matter your role, it would be unfeasible to end up in situations which didn't rely on luck.

Soft-claiming loses its ability to bait Mafia claims out.

Reads become redundant because they rely on dubious meta assumptions.

The game essentially relies on pot luck, usually from the first and second night, should a Doctor or some other protective role be neutered, the Mafia will usually always end up in a lylo.

Should the Doctor save, night one, the Cop may out sooner, all according to the odds being favourable to them rather than anything to do with social interactions. By this point the only thing that could make EpicMafia a social game rather than Russian Roulette, is by meta, trolling, etc. Should it be a normal thing for blues to soft-claim? Then the Mafia will find themselves soft-claiming in every game. The only reason to do otherwise is forming a clique which feeds you games because of a bias.

To reduce every game down, the method is simple, pick the first role in a game's slot and then write out the possibilities for that role depending on who dies each night and who could potentially claim. Example; if I am Bomb, and Cop dies night one, what does this mean for me?

If Doctor saves, how does this affect the roles like Tracker, Vigilante? How does the game look by Day 2? Would it be a viable tactic for the Cop to out here and get two reports out, potentially damning at least one Mafia to counter-claim? I always assess things based purely on the maths behind it, and always try to act as pro-town as possible, even if Mafia, it is in the best interests of everyone to play odds.

This thread will be used as reference for other threads in the future.

Heads or Tails?

Tails

Heads

bernin

over 9 years

Mafia is completely an odds based game. Starting with the first kill the game is going to swing town-sided or scum-sided based on whether a pr or blue dies etc. Being a good player is about maximizing your odds. As mafia that means killing power roles, killing players who can carry the town and claiming at the right times. As town its about factoring tells, reads and voting patterns based on experience to determine who is most likely to be mafia.

Trying hard as either alignment is a way to improve your odds. You can call it manipulation or playstyle, same thing.

Players do X as mafia, players will also do X as town. The game is all about determining odds. Players who play to maximize their odds win more games.

Arcbell

over 9 years

Odds are important, but don't make the mistake of thinking everyone who considers "social reads" or "manipulation" as important parts of the game is falling prey to confirmation bias. Some are, yes, but lie detection is a real thing and it's possible to be much better than random at it, even over text.

Kellyn

over 9 years

Just read it all and the comments, and felt inspired to say this.

As an extremely logically-minded person, I tend to dictate everything on what roles are left up, and what the odds are. Like you, I play to win by making the best guess possible. In the past, I didn't understand the point to needless talking or scumhunting, but I decided to dig deeper to try to understand their point of view. In my own experience, I've found that the main logical role of scumhunting is to gain an extra advantage over the numbers. Mafia, at its core, is based on psychological principles. Its whole basis is to deceive and and to find the deceivers. While it can be isolated to numbers, number can not be a final decision. After all, what makes you flip one way or the other in the end? Your decision, using the knowledge you've gained by talking and scumhunting. Also, I don't believe all assumtions are based on 'meta', but that's a moot point.

The just of what I'm trying to say is that I, too, am logical; but I've found the purpose of talking is to gain a bigger chance of winning by trying to eliminate or target certain people. Ever since then, I always love seeing the whole Logic vs. Psychology aspect of the game, and I love the balance they exhibit.

-Kellyn <3

FriedChckn

over 9 years

I read the beginning and then glossed over the rest, but I feel like this just isn't true. Many of the great plays of the game involves maneuvers that involve going against the "mathematically" correct action. If I am a mafia at nighttime and the only roles remaining are a clear and 2 villagers (I'm claiming blue as well), I may decide to kill one of the villagers instead of the clear. Simply playing "statistically" in your favor does not a great mafia player make.

The best players are charismatic and can convince the town that they're telling the truth, or maybe they're clever and can more easily identify when someone is lying.

It is in such cases that great players are made- those who defy convention and the odds or the circumstance in a game.

I disagree with the sentiment that the game doesn't fundamentally revolve around social interactions, because you could theoretically win in any scenario as a town (or mafia) no matter how statistically low your odds are if you can understand the intimacies of the social interactions in the game.

VictoriaEldor

over 9 years

Because not all games are classic, so the number for the Cop's value has to change for each set-up. The number was just supposed to represent how valuable the role was in that particular set-up and wasn't specific.

I'm also lazy.

Dibbun

over 9 years

Why is this a great mystery to be philosophized over?

In classic, a cop has a 1/3 chance of finding a guilty. They have a 1/5 chance of surviving until d1. They have a 1/6 chance of their target being killed n1, and a 1/5 chance of doc killed n1. Doc has a 1/7 chance to save.

So the odds of someone finding a guilty and there being a doc save is 1/21.

I have no idea why you need walls of texts and "arbitrary numbers" of things when you can just multiply fractions.

VictoriaEldor

over 9 years

animegayboy69 says

if you view the game as a functional random number generator, why would you even play it

e: I'm not sure if this is a troll thread or the OP has autism

If it was random the incentive would probably be 'not knowing how things will turn out'. I can make decent bets on outcomes so long as I take into account everything that has happened.

But there is still a chance things go hilariously bad.

It's the same question as asking someone why they get up in the morning, because at any moment a plane could crash into their home.

Aesop

over 9 years

VictoriaEldor says

Evil says

Sileath says

non-linear equations for what takes simple arithmetic

Perhaps, but why not find an absolutely concrete method? I'm fairly certain one exists, and even if it doesn't, I'm having a great time trying to find one!

The best way to handle it as a maths problem is to break it down individually.

For example. Take the role of Cop, and calculate first this; what are the odds that the Cop will guess correctly on Night 1? Put this aside.

Then ask, what are the odds that the Cop gained an innocent? Put that aside.

Two other possibilities, what if the person he investigated was shot? There are other possibilities.

Now that we have these, we calculate how beneficial each of those scenarios would be.

Once all of the possibilities of Night one are calculated for every town role, you surmise that there is a "chance that a beneficial thing happened for town!" On night one, such as, Doc could have saved, or Cop could have found a guilty, and you weigh their worth against the other possibilities.

We could arbitrarily say that finding an innocent report as cop is worth about 1, but finding a guilty is a solid 1.2.

Whereas if the Cop investigated the dead blue, he has only a 0.6 (It's not totally awful, because both he and the Doctor have survived, and there is now less people in the pool of suspects to investigate, leading to a higher likelihood of seeing a Mafia.)

If Doc dies, or Cop dies, that could be equated to a mere 0.4

So in the most ideal circumstances (the best outcome), Doc saves Cop who found a guilty claim, and town is currently 1.4x likely to win.

That's all it takes to measure game modes. It also adds an objective value to every role, for example, a powerful role adds 1.1 To the village chances, whilst other townies could potentially be harmful at 0.9

same

animegayboy69

over 9 years

if you view the game as a functional random number generator, why would you even play it

e: I'm not sure if this is a troll thread or the OP has autism

VictoriaEldor

over 9 years

Evil says

Sileath says

non-linear equations for what takes simple arithmetic

Perhaps, but why not find an absolutely concrete method? I'm fairly certain one exists, and even if it doesn't, I'm having a great time trying to find one!

The best way to handle it as a maths problem is to break it down individually.

For example. Take the role of Cop, and calculate first this; what are the odds that the Cop will guess correctly on Night 1? Put this aside.

Then ask, what are the odds that the Cop gained an innocent? Put that aside.

Two other possibilities, what if the person he investigated was shot? There are other possibilities.

Now that we have these, we calculate how beneficial each of those scenarios would be.

Once all of the possibilities of Night one are calculated for every town role, you surmise that there is a "chance that a beneficial thing happened for town!" On night one, such as, Doc could have saved, or Cop could have found a guilty, and you weigh their worth against the other possibilities.

We could arbitrarily say that finding an innocent report as cop is worth about 1, but finding a guilty is a solid 1.2.

Whereas if the Cop investigated the dead blue, he has only a 0.6 (It's not totally awful, because both he and the Doctor have survived, and there is now less people in the pool of suspects to investigate, leading to a higher likelihood of seeing a Mafia.)

If Doc dies, or Cop dies, that could be equated to a mere 0.4

So in the most ideal circumstances (the best outcome), Doc saves Cop who found a guilty claim, and town is currently 1.4x likely to win.

That's all it takes to measure game modes. It also adds an objective value to every role, for example, a powerful role adds 1.1 To the village chances, whilst other townies could potentially be harmful at 0.9

Mike

over 9 years

You've never experienced a social epicmafia.com so in that sense your post is valid.

Since you're doing this why not just come forward and out pranay's method of memorizing role alignments. In your next PR game take a screen shot of the list, refresh and take another screen shot. If the person at the top of the list didn't move, they are 100% a power role, no outliers. If they're second on the list they have 50% chance to be a power role. If they're second to last on the list they're >50% chance to be a power role. I'm including mafia as power role in this.

Zhuorb

over 9 years

I read 2 lines and decided I wasn't interested

SackAttack

over 9 years

Por que no los dos!

Evil

over 9 years

Sileath says

non-linear equations for what takes simple arithmetic

Perhaps, but why not find an absolutely concrete method? I'm fairly certain one exists, and even if it doesn't, I'm having a great time trying to find one!

Sileath

over 9 years

non-linear equations for what takes simple arithmetic

Evil

over 9 years

You just made my day. I've been trying to prove that it's entirely possible to use mathematical algorithms to predict mafia in a three way situation. So far I haven't done too well, but I think that's because I'm looking at it in black and white. I'll get one of my math savvy friends to explore non-linear equations. I'll post whatever he comes up with, but I probably won't be able to explain it until he catches me up on it.

Great post, like seriously, I'm impressed beyond measure.

VictoriaEldor

over 9 years

EmmerrLouise says