Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Teproc]

My point was that the "results" we would get from this plan should be ignored as far as actual scumhunting goes. ie we wouldn't take the targets out of the lynch pool entirely. As I said, it would be used mostly to be able to get something out of a dead person's flip.

Based on your previous post, I thought you had understood that because you mentioned the fact that we would only get something out of this when the claimants died.

[AHoppy]

My apologies, I see how my word choice was poor.

[yuma]

My point was that the "results" we would get from this plan should be ignored as far as actual scumhunting goes. ie we wouldn't take the targets out of the lynch pool entirely. As I said, it would be used mostly to be able to get something out of a dead person's flip.

Based on your previous post, I thought you had understood that because you mentioned the fact that we would only get something out of this when the claimants died.

and I say that is extremely dumb! Because there very well could be a real cop result in there and then we are going to just go ahead and ignore it and lynch the player.

If I was a cop and had a town result and then town decided to up and lynch that very player I would be furious!

[Jimmmmm]

Jimmmm : to be fair, he is trying to end the theory talk with an actual decision, rather than just getting past it without actually reaching consensus. I don't think there's actually a contradiction here.

Yeah, I know that's what he was trying to do. But he's asking for opinions on something that most or all of us have already expressed opinions on. Why not just check everyone's last expressed opinion and post the results?

[AHoppy]

Because people's opinions change.  Mine changed multiple times over the discussion.  I want people to commit to one plan so that we can move on.

[scott_pilgrim]

I was referring to the "some weird number rolling thing". You do see how that doesn't seem like a particularly attractive thing to vote for ?

I don't think calling it weird says anything about how good of an option it is, if I thought it was the best plan I would totally vote for it as "some weird number rolling thing".

vote: only cops claim with results

[Twistedarcher]

I agree with Yuma. Docs should just stay quiet.

I disagree with saying "let's not talk about theory anymore". We will talk about what we want to talk about, and shouldn't try to artificially end any conversations. Personally i don't have a lot more to say, as it's clear we aren't doing any claiming, but people can talk about what they want to talk about -- natural conversation is a good thing, not a bad thing, even if it's about theory.

[yuma]

only cops claim with results

[Jimmmmm]

only cops claim with results

[Jimmmmm]

I agree with Yuma. Docs should just stay quiet.

I disagree with saying "let's not talk about theory anymore". We will talk about what we want to talk about, and shouldn't try to artificially end any conversations. Personally i don't have a lot more to say, as it's clear we aren't doing any claiming, but people can talk about what they want to talk about -- natural conversation is a good thing, not a bad thing, even if it's about theory.

Natural conversation is a good thing. A huge Day 1 filled with 30 pages of theory is a bad thing.

[Archetype]

only cops claim with results

As much as I love theory talk (and really, I do), after we reach the consensus I figured we'd make in the first place, I'll do a reread to see if I can get some legit reads on players.

[faust]

Now I finally had time, and ran some statstics. I implemented mafia games with this setup with the following strategies:

- Standard:
Cops claim with a guilty result. If there is at least one guilty result, town lynches them. Else town lynches randomly. Scum never fakeclaims and kils randomly each night. Docs and Cops target randomly.

- No-Lynch:
Cops claim with a guilty result. If there is at least one guilty result, town lynches them. Else, if there is at least one unlcaimed Cop alive, town no-lynches. Else town lynches randomly. Scum never fakeclaims and kills randomly each night. Docs and Cops target randomly.

I ran 100.000 games with each strategy. Results are this:

Standard wins 12,1% of the games.
No-Lynch wins 29,6% of the games.

This is some quite strong evidence I think that we are better off no-lynching until 4 Cops are outed.

Vote: No Lynch

[Voltaire]

Pretty sure I don't care whether it was right or wrong...

You don't care if it's right or wrong? Wouldn't that mean you don't care if it's actually useful in finding scum?

[Voltaire]

Now I finally had time, and ran some statstics. I implemented mafia games with this setup with the following strategies:

- Standard:
Cops claim with a guilty result. If there is at least one guilty result, town lynches them. Else town lynches randomly. Scum never fakeclaims and kils randomly each night. Docs and Cops target randomly.

- No-Lynch:
Cops claim with a guilty result. If there is at least one guilty result, town lynches them. Else, if there is at least one unlcaimed Cop alive, town no-lynches. Else town lynches randomly. Scum never fakeclaims and kills randomly each night. Docs and Cops target randomly.

I ran 100.000 games with each strategy. Results are this:

Standard wins 12,1% of the games.
No-Lynch wins 29,6% of the games.

This is some quite strong evidence I think that we are better off no-lynching until 4 Cops are outed.

Vote: No Lynch

That's tempting except for one thing - human judgement certainly ups these odds. Because if I'm reading your numbers correctly, town only wins 12% of the time? That would make this setup classified as "unbalanced" and changed. Since I assume this setup is as close to 50/50 as can be considered reasonable, judgement in PR direction (ie not "random") must help a lot.

Which makes me wonder, if the "Standard" % can be raised so much, what happens to the No Lynch %?

Really though, I'm just disheartened overall right now because REDACTED.

[faust]

Well, the random lynching and random targeting certainly leads to town doing worse than with human judgment. I think both percentages are significantly raised by this. The question is: which profits more from it? "Standard" profits mostly from lynches being directed by humans. But that is two-fold, as there is also scum trying to get a mislynch, so it's unclear how big this benefit actually is. The targeting of the town PRs might be the more significant factor. But since PRs live longer with the "No-Lynch" strategy, it might even be that this plan is stronger positively influenced.

Note also that both strategies do not include Cops claiming with a town result, as that is hard to implement. After all, there's no reason for scum not to fake-claim something here, and the program can't deal with fake claims. I think in the "No-Lynch" strategy, Cops shouldn't claim a town result, because it narrows down the kill pool for scum. In the "Standard" strategy however, they probably want to claim. So that would mean that there's a possible positive impact on the "Standard" strategy that I have not taken into account. But since the numbers are that far apart ("No-Lynch" being more than two times as effective as "Standard"), I do not think it swings the outcome in favor of "Standard".

[Teproc]

As Voltaire said, if your simulation says that town has a 12% chance of winning normally, I think there's something wrong with your simulation. Which isn't particularly surprising, since this is a social deduction game, ie not the most easy thing to simulate mathematically.

The no lynch plan would need to be incredibly good for me to go along with it, just because I like playing the game more than, you know, not playing it.

[yuma]

Has the forum been down or was today just the most inactive
 day ever?

[scott_pilgrim]

Given that currently no one stands out to me as likely to be scum, I think I would pretty much be taking a shot in the dark if I voted for someone.  Given faust's statistics (which I really hope I can trust), it is better to not lynch than to lynch randomly, so

Vote: No Lynch

I honestly have no idea whether that's the right call, my understanding is that no lynch is uncommon even on the first day, but it makes sense that it would be a lot better in this set-up given that everyone has a PR.

[Teproc]

Even if it ends up being no better than random in terms of probability to hit scum, it's the interactions that matter. That's the problem with no lynch, it makes day 1 basically useless. Like, no lynch is pretty close to being as bad as a day 1 mislynch, and it's obviously infinitely worse than a day 1 correct lynch.

[yuma]

Even if it ends up being no better than random in terms of probability to hit scum, it's the interactions that matter. That's the problem with no lynch, it makes day 1 basically useless. Like, no lynch is pretty close to being as bad as a day 1 mislynch, and it's obviously infinitely worse than a day 1 correct lynch.

That isn't the way to think of it. Although i see why you are...

The way to think of it is to think of day1 just never occurring (from a stats point of view) Because then day2 starts over as a new day1, but with one less player. Basically what it turns into is that we start the game over with 12 players, but potentially have a % (someone else can do the math), but probably around 1/12 (so not that high) of someone being /outed as scum and a approx 1/12 chance that the night kill will be prevented...

now that might not be worth the trade off... but to say we will be lacking the interactions isn't exactly true because the next day just becomes the new day1 and the next day (day3) becomes the next day2...

[Teproc]

I see, that's an interesting way of framing it. 

I just thought of an argument in favor of no lynch by the way : people who rolled night 5 have a better chance of using their power. I don't know how good an argument it is, and I still think I'm against it, but it's worth considering I guess.

[scott_pilgrim]

I see, that's an interesting way of framing it. 

I just thought of an argument in favor of no lynch by the way : people who rolled night 5 have a better chance of using their power. I don't know how good an argument it is, and I still think I'm against it, but it's worth considering I guess.

Yeah, I think this is the main reason why in faust's calcs no lynch comes out as better.  Well, not specifically night 5, but people with later nights in general all have slightly better chance to live until their night to use their power.

[yuma]

Pretty sure I don't care whether it was right or wrong...

You don't care if it's right or wrong? Wouldn't that mean you don't care if it's actually useful in finding scum?

No, I mean that those sort of statements generally aren't useful to find scum to begin with, so I don't really care if it is true or not.

As an example, TA's statement perhaps could be true generally (I kinda doubt if it is, even if he has provided a few cherry picked examples), but because I am town I know it is not true in this instance... my statement is the same... I don't know if it true or not and I don't care because it is so easy to make such a statement that it holds very little water to me.

[pingpongsam]

I engaged the claim theory talk early and gained the most votes of anyone in the game for decrying the theory that has since been universally abandoned and supporting the theory that has been universally accepted.

I pointed out to the main proponent and progenitor of the motion to no-lynch that no-lynch is generally a scum suggestion and the game has devolved into a wait for D1 to end so we can see who dies first tonight.

Meanwhile I stopped posting, because I was legitimately busy, and suddenly no one is interested in painting me scum for engaging the topics at hand.

I suspect scum is controlling this day quite well.

[Jimmmmm]

I engaged the claim theory talk early and gained the most votes of anyone in the game for decrying the theory that has since been universally abandoned and supporting the theory that has been universally accepted.

I think this is misleading, and it comes across as a grab for Towncred. I didn't vote for you, but I definitely could have, and it certainly would not have been for rejecting or supporting theories. I highly doubt that is why anyone has voted for you.