Mafiascum.net

Hoopla wrote:

SCIENCE

is probably unbalanced given we used to run a 2 mafia, 2 Masons, 7/8 Townies variant of Friends and Enemies.

0.4ish EV is hardly imbalanced.

how did you arrive at .4 EV, Emp?

zoraster wrote:how did you arrive at .4 EV, Emp?

Firstly, I said ish. I wrote down the possibilities that lead to a town win and added them up.

Empking wrote:

zoraster wrote:how did you arrive at .4 EV, Emp?

Firstly, I said ish. I wrote down the possibilities that lead to a town win and added them up.

But what assumptions did you make? Because if it were just random lynch then you should have gotten 22.86% EV. Obviously an adjustment needs to be made for the masons, but how did you account for that?

Cogito Ergo Sum wrote:Um, no?

I was under the impression that "Small Open Games" was 8P or fewer. What happened to that?

zoraster wrote:

Empking wrote:

zoraster wrote:how did you arrive at .4 EV, Emp?

Firstly, I said ish. I wrote down the possibilities that lead to a town win and added them up.

But what assumptions did you make? Because if it were just random lynch then you should have gotten 22.86% EV. Obviously an adjustment needs to be made for the masons, but how did you account for that?

1. If a mason is run up both masons claim.
2. Nobody else claims.
3. Masons automatically claim D2.
4. Scum kill any any claimed masons.

Empking wrote:

zoraster wrote:

Empking wrote:

zoraster wrote:how did you arrive at .4 EV, Emp?

Firstly, I said ish. I wrote down the possibilities that lead to a town win and added them up.

But what assumptions did you make? Because if it were just random lynch then you should have gotten 22.86% EV. Obviously an adjustment needs to be made for the masons, but how did you account for that?

1. If a mason is run up both masons claim.
2. Nobody else claims.
3. Masons automatically claim D2.
4. Scum kill any any claimed masons.

I'm not ready to believe this is an optimal strategy. I'll test it later against:
1. If a Mason is run up, they claim.
2. If scum is run up, they will claim Mason.
3. If scum claims Mason, they will be counterclaimed.
4. A claimed Mason will not be lynched without a counterclaim.
5. A claimed Mason will die during the next Night (or as soon as possible in the case of multiple outed Masons)
6. Scum will not counterclaim Mason unless it would improve their win rate.
7. If there are multiple Masons alive, scum will only counterclaim at LyLo for a 50/50 shot at the win.

i actually think emp's makes more sense. I started to post what you did, Vi. But considering how short the game is, masons should probably either claim (if run up) as a pair or not. That way any player that claims mason without a pair is lynched.

Vi wrote:

Empking wrote:

zoraster wrote:

Empking wrote:

zoraster wrote:how did you arrive at .4 EV, Emp?

Firstly, I said ish. I wrote down the possibilities that lead to a town win and added them up.

But what assumptions did you make? Because if it were just random lynch then you should have gotten 22.86% EV. Obviously an adjustment needs to be made for the masons, but how did you account for that?

1. If a mason is run up both masons claim.
2. Nobody else claims.
3. Masons automatically claim D2.
4. Scum kill any any claimed masons.

I'm not ready to believe this is an optimal strategy. I'll test it later against:
1. If a Mason is run up, they claim.
2. If scum is run up, they will claim Mason.
3. If scum claims Mason, they will be counterclaimed.
4. A claimed Mason will not be lynched without a counterclaim.
5. A claimed Mason will die during the next Night (or as soon as possible in the case of multiple outed Masons)
6. Scum will not counterclaim Mason unless it would improve their win rate.
7. If there are multiple Masons alive, scum will only counterclaim at LyLo for a 50/50 shot at the win.

wrt 7, I'm pretty sure that that's not the optimum strategy. (2/3 compared to a 1/2 for scum).

In general though, I don't see that strategy you propose do much better then mine (It'll still be what I call 40ish).

zoraster wrote:i actually think emp's makes more sense. I started to post what you did, Vi. But considering how short the game is, masons should probably either claim (if run up) as a pair or not. That way any player that claims mason without a pair is lynched.

If both Masons claim D1, the Town is disadvantaged quite a bit if they actually do lynch scum that Day (though their chances are still pretty good). If they lynch Town, though, it doesn't make a difference.

wrt 7, I'm pretty sure that that's not the optimum strategy. (2/3 compared to a 1/2 for scum).

It's not, upon reflection. Scum have a much better chance of winning if they don't counterclaim, so they won't.

In general though, I don't see that strategy you propose do much better then mine (It'll still be what I call 40ish).

It does, actually.

See if you guarantee that a scum run up will claim mason, you should never lynch a claimed VT.
I think a lot of the EV page things (named role and mason sections) are broken because of that.
That stuff is where game theory happens.
Both sides have "I will do this X fraction of the time" variables they need to set and then we do some math and find the nash equilibrium (point where neither side wants to change their strategy).
And then we need to think about "what if we randomly select somebody and let them choose who to run up" and stuff like that which generate extra information while making scum/masons less likely to get run up and making the problem more complicated.
I really like thinking about small-number-of-people-remaining situations.
I still need to put up that lylo article and work more on my scumhunting software tools / stats research.
I need to play less games so I can do more science it's fun but I feel guilty spending scumhunting time on it...

I promise full analysis of 2 masons 2 scum 1 VT sometime soon.

Llamarble wrote:I promise full analysis of 2 masons 2 scum 1 VT sometime soon.

It should be simple, right? Either the scum claim masons for a 50%, or VT's for 1 - 2/3 * 1/2 = 67%, unless the townies' reads give them better chances.

I'm pretty sure you're assuming massclaim immediately, which town doesn't want on account of the whole 67% option for scum thing.
Basically not massclaiming creates the possibility of scumlynch-> scum shoot VT -> town autowin.

BTW by "strategy" I mean choices for variables like percent chance you lynch a claimed VT instead of running up somebody else, etc.

the probability of town winning without any immediate claims is:
P(lynching scum immediately) * ( P(VT killed) + P(mason killed) * P(town wins in 1/1/1) )
2/5 * (1/3 + 2/3 * 1/2) = 27%, so that's 73% for scum.

If you run up a mason, they and their buddy claim. Shenanigans ensue.
I assure you you are oversimplifying.
Be patient and I shall create pretties.
Possibly very patient unfortunately.

Okay, I set up 1VT 2SCUM 2MASONS assuming random run-ups as opposed to we-pick-somebody-to-choose-who-to-run-up or voting or whatnot.
I am hoping (and somewhat expecting) some stuff to simplify out when I start solving, but right now I have something like 15 terms with a total of 10 variables, 4 of which scum set and 6 of which town set.

I know the answer is between 60% (scum always claim VT) and 66% (town can't do worse than starting with massclaiming) for scum though. Ha.
The reason it's all so complex is that scum have claim / counterclaim options, and if you don't lynch the first person run up you create a lot of branches, etc.

What was the ruling on PieE7's balance? What is the ruling on SCIENCE's balance? I want to push on.

Hoopla wrote:What was the ruling on PieE7's balance? What is the ruling on SCIENCE's balance? I want to push on.

You're the ruler on balance!

Science is almost certainally balanced.

Hoopla wrote:What was the ruling on PieE7's balance? What is the ruling on SCIENCE's balance? I want to push on.

The ruling on balance is that we don't know what balance is, objectively speaking.

I'm definitely coming down against Pie E7 though, as it basically requires one of the scum to fakeclaim. SCIENCE is better, if not blatantly so.

What's the problem with Gurgi EC8?

Lord Gurgi wrote:What's the problem with Gurgi EC8?

The possibility of drawing no Weak Cops, and playing in a mountainous 2:6 setup. Make it 1-3 Weak Cops and it seems fine to me. What do you think?

Well, I haven't seen any obvious complaints in the games that have run it. But if you think that would be an improvement that's fine. I guess I just don't like a guarantees in the setup.

None of the games have featured a town with zero cops. The only guarantee is that there is one cop - that isn't going to break the game, and it fixes the 1/8 games where you're playing in an unbalanced setup.

Okay.

Science is kinda garbage, I dislike.

Adding 2 VTs probably balances it.

I guess we could all agree on this definition of "balance" : A game is considered balanced when there is no breaking strategies from the start, a reasonable town win probability calculated with basic routine (no more than 60% no less than 40% without 3rd party - no more than 55% with 3rd party) and an interesting mechanic that makes people want to play it (because it's pointless to discuss balance on a game nobody wants to play - like mountainous *cough*)

Gurgi EC8 should be with at least one weak cop, it allows for fakeclaiming and counterclaiming.

In SCIENCE, only the encryptor should have daytalk (rename it ?) or the town should have a decryptor specialist who can investigate at night and get at dusk who had been daytalking. He could even be mason too though it would remove the risk of masons being taken for scums