Page 266 of 357

Posted: Sat Dec 21, 2013 8:54 am
by FakeGod
We can rename the role Innocent Person. I dunno. That's an easy fix.

Who's in charge of newbie queue setups?

Posted: Sat Dec 21, 2013 12:42 pm
by quadz08
Newbie listmod, ultimately.

Posted: Sat Dec 21, 2013 6:52 pm
by N
Fakegod, that setup is shit.

Posted: Sat Dec 21, 2013 6:54 pm
by Faraday
Why?

Posted: Sat Dec 21, 2013 6:57 pm
by Faraday
In post 6611, quadz08 wrote:
In post 6609, quadz08 wrote:Town
power roles
night actions are part of the game unless you're playing a mountainous. Newbie games need to reflect and include that.
As in, I think this is a bad argument. Knowing what roles work together and what makes sense in a setup is part of the game in all closed setups (the majority of games) yet it's not something we try to teach in the newbie setup.

I do think it suffers from a problem in that it's going to get repetitive for IC's and such, which is probably an actual issue. (no one really wants to play the exact same open setup game after game, do they?)

Posted: Sat Dec 21, 2013 7:15 pm
by Katsuki
In post 6629, Faraday wrote:
In post 6611, quadz08 wrote:
In post 6609, quadz08 wrote:Town
power roles
night actions are part of the game unless you're playing a mountainous. Newbie games need to reflect and include that.
As in, I think this is a bad argument. Knowing what roles work together and what makes sense in a setup is part of the game in all closed setups (the majority of games) yet it's not something we try to teach in the newbie setup.

I do think it suffers from a problem in that it's going to get repetitive for IC's and such, which is probably an actual issue. (no one really wants to play the exact same open setup game after game, do they?)
I dunno, newbie setups are pretty fun in f2f at least. :P

Posted: Sat Dec 21, 2013 7:19 pm
by FakeGod
In post 6627, N wrote:Fakegod, that setup is shit.
thx brah I knew I could count on u
In post 6629, Faraday wrote: I do think it suffers from a problem in that it's going to get repetitive for IC's and such, which is probably an actual issue. (no one really wants to play the exact same open setup game after game, do they?)
Ok. That's actually a good point. I don't mind playing the same stuff, but I obviously can't speak for others' preferences.

I do think this setup is an improvement over the one we currently have though.

Posted: Sun Dec 22, 2013 7:00 am
by Plessiez
(I'm going to say CI for Confirmed Innocent instead of IC for Innocent Child here, to avoid confusion with the newbie-specific IC.)
In post 6605, FakeGod wrote:I have a new setup proposal for newbie games. [...] My simulation shows that this setup has town win rate of
48.5%
(very close to 50%) given that town were to randomly lynch, and mafia were to randomly shoot. (unless doc claimed of course)
I'd be interested to see the details of your simulation.

For instance, what do you have happening if a day starts with 1 confirmed innocent, 1 doctor, 1 goon and 1 VT? The optimal strategy is, I think, for the doctor to claim here, since whether or not the goon counter-claims town now has a 50% chance of winning (you either lynch one of the two unconfirmed people not claiming doctor, or you lynch one of the two people claiming doctor). That's obviously better odds than you'd get just randomly lynching one of the people who isn't the mod-confirmed innocent.

Does your simulation take things like this into account?

More generally, do you really have the doctor protecting the confirmed innocent every night? Because I'm sure that isn't optimal either. There's a basic game theory thing going on here, no? Doctor should protect the CI some fraction p of the time, and the mafia should try to shoot the CI some fraction q of the time. Would be very surprised if either p or q was 1 or 0 in a Nash equilibrium strategy. If the doctor should always protect the confirmed innocent, mafia should never target him. But then the doctor shouldn't bother to protect him, so... (The actual value of p and q on any given night will depend on the number of VTs and goons alive, presumably, so would be a bit tedious to calculate.)

Basically, it seems to me that town should be able to win more than 50% of the time even without any sort of non-random scumhunting, though I'm too lazy to try to prove it. Can you show otherwise?
In post 6617, FakeGod wrote:
In post 6615, BBmolla wrote:But people know he's town! So if they follow him they'll be following someone who is 100% town
If the town follow the IC, then, assuming IC is about good as a random generator at landing a scum, town loses on average. (
48.5%
)
First, what you say here doesn't mean that following the CI isn't the town's best bet. If the town can't reasonably expect to do better than they'd do following the CI, but can definitely do worse, then they'll surely follow the CI. And if you assume an inexperienced town, and possibly a more experienced/talented than average CI, this is exactly the case.

Also, it seems that your claim could easily be restated as "if the CI is even slightly better at picking scum than a random generator, then the town will probably win more often than they lose when they follow him". This isn't a great argument against following the CI. Even if your original claim is true, which, really, I'm not convinced it is (unless your simulation accounts for things you've suggested it doesn't account for). I mean, you have to assume that there are people who can scumhunt better than a random number generator, or why bother having newbie games at all?

(For
learning mafia
following the CI is definitely subpar. But a good newbie set-up shouldn't make the town choose between playing to maximise their chances of winning and playing to maximise their chances of learning something useful.)

Thinking about it a bit, I suppose you could always arrange things so that the confirmed innocent player was never the game's IC, actually. Maybe not even a SE. That seems like a good way to keep the set-up fit for newbies, assuming that the underlying numbers work.

Posted: Sun Dec 22, 2013 10:12 am
by FakeGod
The simulation assumes that town is randomly lynching from the pool minus the Confirmed Innocent.

If doctor is chosen to be the lynch of the day, he claims immediately, and someone else is chosen instead.

Mafia kills at random, unless doctor is claimed, in which case they shoot the doctor, and shoot the Confirmed Innocent next night.

Posted: Sun Dec 22, 2013 10:14 am
by FakeGod
1 confirmed innocent, 1 doctor, 1 goon and 1 VT isn't a possible game-state during the day, if we assume that doctor is always protecting the Confirmed Innocent and scum is always killing someone else other than CI, and town always lynches.

Posted: Sun Dec 22, 2013 2:53 pm
by agi102
Trying this "setup making" thing again

Follow Which Clear (9p)


1 Innocent Child Doctor
1 Innocent Child Tracker
1 Innocent Child Vigilante
1 Innocent Child Watcher
1 Doctor
1 1-shot Bulletproof Townie
1 2-shot Cop
1 Mafia Goon
1 Criminal Child (Is revealed as "town" at game start, but is actually scum)

Doctors cannot cross-protect.

Posted: Sun Dec 22, 2013 3:11 pm
by GuyInFreezer
massclaim. gg.

Posted: Sun Dec 22, 2013 5:36 pm
by Ms Marangal
holy shit, that game is town-sided.

the IC's need to have a non-pr, otherwise Criminal child is fucked

the non-IC's need to have at least one VT as well, as it would fuck the mafia goon over.

I think all non-IC's should be VT's, the IC's are a power to themselves, as is

maybe keep doc or cop

give the goon a RB, and have them not able to submit an action as well as kill unless RB is the only living scum left.

I also say get rid of IC-vig

Posted: Sun Dec 22, 2013 5:39 pm
by BBmolla
Not to mention that if the IC scum is lynched d1 it clears 4 people.

Posted: Sun Dec 22, 2013 5:44 pm
by BBmolla
You also need to understand that the reason VTs exist is that mass claiming in a setup where everyone is a different role and it's open breaks it because it forces mafia to counter claim a role, and numbers wise loses them the game.

I think before each setup you need to ask yourself "What is my goal with this setup?" And also that in most games not everyone is a power role for a reason. And if they are, it's a closed setup.

Posted: Sun Dec 22, 2013 5:46 pm
by Ms Marangal
stratagem, Lynch through IC's commence!

Posted: Sun Dec 22, 2013 5:47 pm
by Ms Marangal
actually, Cop isn't even neccessary to have in this set-up

Posted: Sun Dec 22, 2013 9:09 pm
by Bulbazak
In post 6615, BBmolla wrote:But people know he's town! So if they follow him they'll be following someone who is 100% town so it's their best bet.

Plus, Doc will be on him so he won't die!
What makes that different than town following the IC (Inexperienced Challenged) in current Newbie games? The IC generally tends to be the slot most trusted by new players and the one most often protected the first night.

Posted: Sun Dec 22, 2013 9:58 pm
by BBmolla
IC can be scum obviously.

Posted: Mon Dec 23, 2013 5:22 am
by Plessiez
In post 6633, FakeGod wrote:If doctor is chosen to be the lynch of the day, he claims immediately, and someone else is chosen instead.
Well, this seems definitely wrong.

Why wouldn't scum fake-claim doctor when run up to L-1 if this is the strategy town adopts? Worst case is the real doctor counter-claims, the fake claimer is lynched (which was happening
anyway
) and the real doctor definitely dies next.
In post 6634, FakeGod wrote:1 confirmed innocent, 1 doctor, 1 goon and 1 VT isn't a possible game-state during the day, if
we assume that doctor is always protecting the Confirmed Innocent and scum is always killing someone else other than CI
, and town always lynches.
Still don't see a good reason to assume the bolded. (If you're the doctor and you know for a fact the scum will try to kill somebody who isn't the CI, why protect the CI? In fact, if you also know that scum are always killing, then if you protect somebody and there's no kill, you've created a second CI. But even if not, protecting somebody who might be targeted beats protecting somebody who definitely won't be targeted.)

Still think town wins more than half the time here. Tempted to try to prove it, but maybe after Christmas.

Posted: Mon Dec 23, 2013 5:39 am
by FakeGod
All assumptions made were for sake of easier simulation.

Posted: Mon Dec 23, 2013 5:46 am
by FakeGod
Game theory problems.

Scum will fake-claim doctor some fraction of the time if they get run up to lynch. Doctor will counterclaim in this case, and things follow.

Doctor choosing to protect between CI and someone else is also a game theory problem. In reality, doctors will probably choose to protect the CI only some fraction of the time.

Scum will also only shoot the CI some fraction of the time, obviously.

The assumptions are not perfect, but I hoped that they wouldn't throw the simulation off too much.

Posted: Mon Dec 23, 2013 9:41 am
by BBmolla
Workout Session

9 Players

1 Mafia 1-shot Strongman
1 Mafia Docblocker

1 Cop
2 Docs
4 VTs

  • Daystart
  • If both Docs target the same person, their protection will fail.


I didn't really think all of this through but I wanted to use Docblocker so this came to fruition. Lemme know if there is a breaking strategy I'm missing.

I think it's townsided atm. Maybe make it 7p?

Posted: Mon Dec 23, 2013 9:54 am
by Tierce
What if both target the same person and the Docblocker targets one of the Doctors?

Posted: Mon Dec 23, 2013 9:59 am
by BBmolla
In post 6648, Tierce wrote:What if both target the same person and the Docblocker targets one of the Doctors?
Docblocker blocks Doc from targeting, thus the other Doc's protection succeeds.

Double Doc On The Block

13 Players

1 Mafia 2-shot Strongman
1 Mafia Docblocker
1 Mafia Backup

1 Cop
2 Docs
7 VTs

  • Daystart
  • If both Docs target the same person, their protection will fail.
  • If both Docs target the same person, and one is targeted by Docblocker, the protection succeeds.
  • The Backup is only backup to the Mafia roles, becoming whichever of the other roles dies first.
  • Backup who takes on Strongman role does not replenish the used shots.


Sharing is Caring

13 Players

1 Mafia Y-shot Rolecop
1 Mafia Y-shot Doctor
1 Mafia Y-shot Watcher

1 Town X-shot Cop
1 Town X-shot Jailkeeper
1 Town X-shot Vigilante
7 VTs

  • Daystart
  • "X" starts as 4, and decreases with each shot spent by one of the power roles.
  • "Y" starts as 4, and decreases with each shot spent by one of the power roles.