Newbie Data

[Mr. Flay]

Yeah, the biggest problem with Pie E7 is that it gets unwinnable for the Mafia if the Town happens to hit the Roleblocker D1 (which is essentially blind luck in a Day Start game). I think someone worked out an actual optimal strategy for that, but I don't recall it offhand. Faster game, but more random; I think F11 provides enough cover for a lone Goon to pull it out, at least occasionally.

[Mr. Flay]

Sure. How many times did scum hit a Power Role (relative to the chance of a power role being in that game) would be an interesting statistic. I don't think it'll vary much from random chance, but I might be wrong.

Also, since I didn't say it explicitly earlier, nice work kunkstar!

[Mr. Flay]

Yeah, the biggest problem with Pie E7 is that it gets unwinnable for the Mafia if the Town happens to hit the Roleblocker D1 (which is essentially blind luck in a Day Start game). I think someone worked out an actual optimal strategy for that, but I don't recall it offhand. Faster game, but more random; I think F11 provides enough cover for a lone Goon to pull it out, at least occasionally.

Even in F11, if the RB is lynched D1, the game is very close to over. If he tries to fakeclaim Cop then gets countered, the Town wins.

Yeah, but 20% chance (while slim) is better than 0%. I was in one of those early Newbie Games where there was literally NO way for me to win unless the Cop f*cked up; it was extremely demoralizing.

[Mr. Flay]

You'd expect some random drift, but it's still a bit suspect that the no-power-roles setup is the one that keeps coming up low, given the C9 trend. At least we're doing a better job there.

IIRC from earlier analyses, we're actually doing better than when we started with F11 (and obviously better than with C9). I think we're always going to have some residual, if we're doing

current

selection right. A time-based analysis would prove me right or wrong...

I'd be interested to see how cop deaths correlate with win rates.

You mean Cop deaths after D1? It already drops Town chances to 7.7% or so... not sure how much more you'd learn, though what Town chances are after a N1 Cop death could be interesting.

[Mr. Flay]

It's weird how Roleblockers are being lynched D1 significantly

more

than Goons in the C & D setups.

Seriously. Somebody needs lessons in "how to protect your faction's PRs".

[Mr. Flay]

I imagine it'd be too burdensome to find out how often a cop, who survives to day 2, and has a guilty manages to get that guilty person lynched.

Without coming out as Cop? I expect the sample size is relatively small, but it'd be pretty awesome to see, I agree.

Hoopla: I hadn't considered that, mostly because I find Newbie Games to be pretty bad at applying site-meta. Not enough newbies know it, and few ICs really want to apply it with an iron fist (not sure about SEs). I'm happy to be wrong, and the sample size of games seems small enough to scan for trends (if kunkstar could link those 22, plus maybe the 7 Pie E7s).

[Mr. Flay]

Those are the 22 games in which the Mafia Roleblocker was the D1 lynch in F11.

[Mr. Flay]

We should either expand the potential role combinations for newbie games to get newbies experience with things like watcher, tracker, and the like (and to balance out that incredibly scum-sided win percentage), or make newbie games completely vanilla and have it be a basic exercise in playstyle and theory.

I'm not opposed to 1, but we can't seem to win 2:10 Vanilla Games with

experienced

players. Upping the size makes them last WAY too long (and essentially turns them into Minis), and keeping them at 2:5 or 2:7 makes them uncomfortably hard for Town to win.

I'm tentatively okay with mith's idea in post 15 to remove the 2M7T option entirely, as it puts the knowledge of the # of Power Roles in the domain of the Mafia (where it kinda belongs, to teach them how to play). We're somehow winning an average number of those, as Elmo pointed out in post 17, but it's probably through blind luck. I know there was some refutation in an old MD/SI thread for why this might not work, but I'm too lazy today to look it up.

Where do you get that number from?

o_0 Are people just not reading the first post? F11 set, Day One lynch data: of the 13 games where the Cop is the first lynch, Mafia win the game 92.31% of the time.

[Mr. Flay]

*shrug* F11:C has the best tested value of any Newbie Setup we've run, 44% vs. 56%. And I still think (open) 2M7T is demoralizing and a good way to lose players/ICs. All kunkstar's data shows is that Town can manage not to do any worse than random in an environment where Mafia teams self-destruct on a fairly regular basis (most games, the IC is not going to be Mafia).

I'd be happy either eliminating F11:D or going to a straight F11:C (Pie E7++) setup.

Neto: I assume you're talking about in F11/Pie E7? It was pretty common in the original newbie setup.

[Mr. Flay]

Counterpoint: Mr. Flay's position seems to be that without power roles Towns are successful at finding scum more through blind luck than hunting. (Is that right?)

This is a statement that bothers me somewhat so I want to ask about it.

I think we are much worse at 'scumhunting' than we think we are (people like to claim absurdly high accuracy rates, when in truth most "good" players get maybe 5-10% better than random). I also think it's easier overall to scumhunt in larger games, like nhammen says. The lynches can make smaller games extremely swingy; look at the first post for perfect examples. In a bigger game, you get more 'misses', and it comes down to that 5-10% skill, more than random chance.

[Mr. Flay]

Argh; that reminds me that I desperately need to fix my SWA. I did something wrong when I calculated it the first time, then ran out of time to fix it...