The Newbie 2d3 stats thread (upd. 2019-05-28, 87g)

[popsofctown]

Stacking the setup in favor of town achieves "eventually branching out to other queues/areas" but not "retaining them long term". If we're giving out majority wins because some players only have fun when they win, they will be the sort to replace out of games they are losing, leading to full site flake or, worse, not leaving the site and replacing out all the time.
Keeping powerful roles like sane cop in the newbie setup isn't a good strategy either. I just got out of Krazy's large theme and it was super wacky and yada yada but there was 1 three shot investigative in a large theme with only one of the abilities as strong as sane cop, and that is in a large. So running 9 player games with an ungated cop is not actually representative of site play even if these players are going to move onto our stuff on the role-madnessier end of the spectrum. The simplicity of these setups works against them, we can't have millers because it's not a closed game and the miller would become named, so as a result we are literally putting a role in the game where every game after that the player is never going to experience a role as powerful as that. And it's supposed to be representative for discerning which players are going to fit well here long term?

[popsofctown]

Or the next six month's challenges, it's a really worthwhile endeavor.

[popsofctown]

There's definitely conceivable factors where more time and also less time can be worse or better for scum. There's lots of psychological biases that will operate differently at different day lengths.


Here's a question I have though: Are lynches made at or near the deadline to avoid a no lynch hitting with a different accuracy than lynches made because a town is satisfied with their day? And how do those two outcomes impact subsequent days?

Intuitively one would assume that towns are lynching scum more accurately, and winning more often, when they do not even need to skate along the deadline and find a lynch they are confident about in less time. But I think it's worthwhile to stay open to a counterintuitive result. Maybe these deadliners are compromise lynches with just a little bit of the right thinking from everyone in the town and therefore more accurate, instead of 1 idea 1 player had in the town and managed to spread virally - it's easier for 3 out of a group of 5 ideas to be valid than for 1 idea to be valid, if that makes sense. Or maybe the relative inability to defend oneself from a deadline lynch is important, and mafia are outperforming town at talking people out of lynching them (which is not a particularly counterintuitive result, it serves the mafia wincon more dramatically to pull the endeavor off).

Or maybe it informs future lynches better when this occurs D1 or D2. You can vanity wagon more safely in games where lynches are regularly getting achieved well before deadline, if you vanity wagon when the town needs to build consensus to avoid a no-lynch it starts to become more alignment indicative. And there might be tons of other gameplay mechanics that follow that pattern that are just harder to put your finger on.

[popsofctown]

I think towns just being better than they used to be is a big deal

This statement in the other thread had me musing, but I will post in this one because I am not suggesting a setup and this is statisticsy


I was curious how much newbie towns would be winning at mountainous mafia if they played it about as well as they are playing 2d3. Are they so good they could even win at mountainous mafia (maybe because bombtastic SE is always around, and newbscums are always around?) maybe?

So here's what I did. I wrote a program that simulated C2, because it seemed like the simplest setup, and though is still pretty complicated. I debugged it until the theoretical winrates seemed sane, 60% scum winrate with random lynching and night action usage and correct application of the investigative impact of jailkeeper (which I approximated as roleblocking a lone scum outside of MYLO instantly wins you the game and doing it in MYLO 50/50s the game, I believe that's right or close to it).
Then I added a skill factor to it, a dice roll before every lynch and jailkeep to attempt to "read correctly" and guarantee the desired outcome. And I adjusted this dice roll's chance to succeed until the results of my sim matched how much newbies are actually winning C2. It ended up at 11%.
Then I went over and did a different program and just did mountainous mafia, with the same amount of skill factor, 11% chance to ignore randomization and just lynch correctly.
It came out to a 55% winrate for scum.
There's only a sample size of 9 for C2, but C2 doesn't seem like an outlier on winrate so I can treat it as a somewhat larger sample. Unless something special happens in setups that are easy to write programs of but there's no reason to believe that.
If we are shooting for a 50% winrate for scum there may be very very little power that can safely be added to mountainous newbie games in the next revision. A lot of people seem to want a matrix for at least the possibility for 2 protown PRs so that more people get to play a protown PR, but I doubt there's actually enough legroom for that.

[popsofctown]

I mean your conclusion simply matches what has already been suggested in this thread: strengthen scum to match the use of TPRs

Ok. I apologize for trying to contribute. I promise not to try again.

[popsofctown]

I mean your conclusion simply matches what has already been suggested in this thread: strengthen scum to match the use of TPRs

Ok. I apologize for trying to contribute. I promise not to try again.

Oh no that’s not what I mean at all! I think the question of “could newbies win mountainous” is fascinating and I liked your experiment very much!
I just thought it was worth adding that “town PRs might be too much for these newbies” is one solution to the problem - and another is “if town should have power, scum should have more power”, which is something that’s already been suggested here.

Your contribution shows me we’re on the right track

Ok.
Let's try to brainstorm more ways for scum to gain power. I think there's some design space that isn't getting used that could be uniquely helpful for newbie semi-opens.