Newbie Game Stats, and Solution?

[Fiasco]

I agree that there's a problem and that using Pie C9 is probably the right solution. Pie C9 has only been tried once on the forums, though; maybe it should be tried at least once more before we make it the standard newbie game. You can all nominate it in the Open Setups thread.

I find this statistic VERY interesting:

Number of times town was lynched (or modkilled) Day One:

-80 (75.4%)

Based on random chance, you'd expect 5/7 here. That means the average town is doing slightly WORSE than it could do by just using dice tags to lynch. (OK, they're probably generating more information for use on future days than if they used dice tags, but I doubt it's enough compensation. And then I suppose there's always the option to use dice tags as a surprise, AFTER watching people's behavior. Except for hammer votes, which are probably important. Hmm.)

Also, doctors seem to harm the town. Or at least they don't help much.

BTW, here's an earlier discussion of randomization in newbie games.

[Fiasco]

(Unless I've forgotten how to do a chi-squared test properly, this sample is only a p value of .1698, which generally isn't statistically significant)

I checked and got something much lower (<.001 IIRC). If it's 11 vs 38 (no powers vs both) then IMHO there's clearly something wrong. I'm guessing at most half of all mods randomize properly.

4 to Lynch on a Townsperson Day One: 72

-2 Scum: 27 (37.5%)
-1 Scum: 34 (47.2%). Town won 5 of these games (14.7%).
-0 Scum: 11 (15.3%). Town won 3 of these games (27.3%).

Fraction of scum among players other than the lynched townie: 1/3
Expected number of scum on a townie lynch if scum and townies are equally likely to be on such a lynch: 4 * 1/3 = 4/3
Average observed number of scum on a townie lynch: 11/9

Townies are more likely than scum to be on a townie lynch!

So not only do towns lynch worse than random chance, it's not clear they get any compensation in the form of information from past votes. Though of course the story could be different if you looked at third and fourth votes only.

[Fiasco]

Beat People to Death with a First Aid Kit Mafia

2 mafia
1 doctor
1 cop
3 townies

When the mafia nightkills the doctor, they immediately get to use the doctor's first aid kit to beat someone else to death and make a little speech about irony. That is, they get another nightkill.

Optional rule: the doctor survives. In either case, if the doctor is targeted, the doctor's protective ability doesn't work that night, and the second nightkill target dies.

Probably too abnormal to make for a good newbie standard, but the logic behind it is to make a minimal change to the old newbie setup that makes mass claim a bad strategy.

Would a mass cop (but not doc) claim still break this? I wasn't around back in those times. It probably would. Bah.

[Fiasco]

I guess a better description would be:

Beat People to Death with a First Aid Kit Mafia

2 mafia
1 doctor
1 cop
3 townies

Day start. If the doctor is nightkilled, skip the next day. (Optional rule: Resurrect the doctor the morning after.)

Unless people think this is broken, I'll nominate it as an Open.

[Fiasco]

Mith: I see. I'm not sure a cop-only claim is a no-brainer in this modified setup. By claiming D1, the cop runs some risk (usually 1/3) of getting himself killed N1.5. In return he probably gets some credibility. Not sure what I'd do myself.

1.) The town can lynch a Town D1, Mafia kill the Doctor N1, get an extra kill, kill anybody, making it 2 v 2 and a scum win.
2.) The town can lynch a Town D1, the Doctor can protect correctly N1, and the town could:
-

A

: Have the Doctor claim who they protected, clearing somebody, which could end with lynching scum, scum kill Doctor N2, and somebody else N2.5, leaving the town at LyLo instead of normal 4-player endgame with a possible Cop alive.
-

B

: No-Lynch, and then scum might kill Doctor N2, get an extra kill, kill a townsperson N2.5, making it 2 v 2 and a scum win.
3.) Even if the town manages to lynch scum D1, if the remaining scum can kill the Doctor N1 and another townsperson N1.5, instead of the town getting two chances to lynch scum, they will only get

one

because it will be 3 v 1, even possibly without the Cop.

Those are all exceptional cases. Usually the scum aren't going to hit the doctor, and usually the doctor isn't going to protect correctly. Remember that there will probably be a cop claim on day 1 or day 2 that will help the town; usually the cop should get 1 or 2 investigations in, I think.

You may still be right that the double kill is too harsh on the town. If that's true, you could still use the rule where the doctor lives.

I think that set-up would only scare newbies away from ever claiming or hinting at Doctor. Also, it would frighten people out of claiming

anything

, because even claiming townie narrows down who the Doctor is. The best strategy for the Doctor would probably be to just claim Townie!

But that's a good thing! Pro-town roles should usually want to keep their exact role a secret. That's already true to an extent in other games, only there it's limited because the mafia takes more time to kill off power roles once they're out.

Further, this set-up essentially makes the Doctor role completely useless: a successful block means little to nothing if the Doctor dies the next night, because the scum will simply get their kill back, and be able to use it nearer to the end of the game, when it could be crucial.

The double kill rule would make the doctor more dangerous to the town, of course, but I don't see how it would make his healing ability useless. Getting a protect in may make the difference between 3 vs 2 and 2 vs 2 if the doctor is killed later.

I also agree with Brutal Assassin - strange rules only applicable to Newbie games are probably not the way to go.

That's true. Again, it would probably do better as an Open. But I'm not convinced it's a bad setup, especially with the "doctor lives" rule.

Come to think of it: if you use the "doctor lives" rule, it's almost like a version of Pie C9 where the blocker can only block the doctor and not the cop. Which would be a viable setup in itself, IMHO.

[Fiasco]

Basically, the theory is that the day 1 cop claim means that the doc will protect the cop, which prevents a night 1 cop death.

Doh! Of course. The cop claim is so the doc knows who to protect. It's not obvious to me that it would help (what if the scum claim doc when in danger of lynch?), but apparently people have gone through the analysis.

My setup doesn't have this problem; claiming cop just changes the mafia's problem from finding the cop to finding the doc (with one candidate, the cop, now ruled out).

[Fiasco]

Not enough -- each

individual

townie is more likely to be on a town lynch than each

individual

scum (though not by much).

[Fiasco]

Yosarian: that does look like it'd work, but what if you disallowed no-lynch?

[Fiasco]

My gut says it's optimal to investigate the player whose probability of being scum is closest to 50%. Almost always, that means the scummiest player alive.

[Fiasco]

I forwarded your post to my gut, which asked me to say that (I quote) "'guilty' results are usually more valuable than 'innocent' results, so you should prefer investigating scummy players, at least up to the point where the investigation is wasted because you know they're probably scum anyway, but people are almost never that sure of someone's scumminess, except because of overconfidence".