Open Side Menu Go to the Top
Register
Academic Werewolf Discussions Go Here Academic Werewolf Discussions Go Here

12-05-2010 , 10:22 AM
I feel like my post about wolves always outing themselves being optimal should be in here but it's not and I can't find it.
Academic Werewolf Discussions Go Here Quote
12-05-2010 , 10:28 AM
Quote:
Originally Posted by BitchiBee
Usually in f3 the seer is moot (all peeks dead), and we stil expect vllage to win mst of the time except when the last wolf is wn
1. I'd like to see the stats.

2. zomg, you are leaving the world of GTO, should I explain GTO to you?

I think it's an interesting result, it's satisfying because it's a way to get the odds up to rand() at F3, despite the problem that the wolf always votes for a villager.
Academic Werewolf Discussions Go Here Quote
12-05-2010 , 10:31 AM
Quote:
Originally Posted by Chips Ahoy
Your theory on that is genius.
Yeah that's what I think- but it doesnt seem that it get the appropiate level of attention I think it deserves :O so im thinking ppl dont like it... im gonna try to make it a part of my game in the future - it think it will be hard work, but worth it
Academic Werewolf Discussions Go Here Quote
12-05-2010 , 10:31 AM
Quote:
Originally Posted by kokiri
I'd have to think to see what spots there are where it's worth a wolves outing themselves to guarantee a wolf/villager swap, because it might be that they can still do slightly better than rand if they do that in some spots on the game tree.
I want to say no because the chances of lynching a wolf is always less than half. Must lynch is obviously a special case.
Academic Werewolf Discussions Go Here Quote
12-05-2010 , 10:32 AM
please explain gto to me- i just started reading this thread
Academic Werewolf Discussions Go Here Quote
12-05-2010 , 10:34 AM
Quote:
Originally Posted by Toganim
Yeah that's what I think- but it doesnt seem that it get the appropiate level of attention I think it deserves :O so im thinking ppl dont like it... im gonna try to make it a part of my game in the future - it think it will be hard work, but worth it
It is awesome.

It has made rethink how to have those roles in games.

bsball needed to be snap lynched for his reaction.

Toganim's Idea:

There are 2 seers, 1 got a n0 peek, 1 didn't. Instead of leaving normal seer cover, villagers leave 2 sets of seer cover: one set for if they got the n0 peek, one set for if they didn't. Now you need both sets of peeks to be wrong before you are eliminated as seer. Tough on wolves.
Academic Werewolf Discussions Go Here Quote
12-05-2010 , 10:36 AM
Quote:
Originally Posted by Toganim
please explain gto to me- i just started reading this thread
A Game Theory Optimal strategy gives the best outcome no matter what your opponents do.
Academic Werewolf Discussions Go Here Quote
12-05-2010 , 10:47 AM
Quote:
Originally Posted by BitchiBee
are you using that 2/3 figure to prove that gto is bad?

because we only expect the wolf to wn less than 50% in our games because our wolves play very imperfectly
Quote:
Originally Posted by Chips Ahoy
I blame the seer. See above.
i don't think the seer is (all of) it. It's all the information accumulated from prior days, voting patters, reads, etc. Even with a no seer game, i would expect that there would be info, because it's impossible to not know what the wolves know. I guess the question is how much the difference would be with no seers. (and there would be no seer hunting either, which i guess is huge. hmmmm)
Academic Werewolf Discussions Go Here Quote
12-05-2010 , 10:50 AM
Quote:
Originally Posted by Chips Ahoy
It is awesome.

It has made rethink how to have those roles in games.

bsball needed to be snap lynched for his reaction.

Toganim's Idea:

There are 2 seers, 1 got a n0 peek, 1 didn't. Instead of leaving normal seer cover, villagers leave 2 sets of seer cover: one set for if they got the n0 peek, one set for if they didn't. Now you need both sets of peeks to be wrong before you are eliminated as seer. Tough on wolves.
Yeah, also if the n0 was randed/not randed etc. Im very happy you like it

Quote:
Originally Posted by Neil S
A Game Theory Optimal strategy gives the best outcome no matter what your opponents do.
Ty, GTO should include TTT
Academic Werewolf Discussions Go Here Quote
12-05-2010 , 10:55 AM
Quote:
Originally Posted by clowntable
I feel like my post about wolves always outing themselves being optimal should be in here but it's not and I can't find it.
Perhaps it's in this thread?
Academic Werewolf Discussions Go Here Quote
12-05-2010 , 11:03 AM
No, I'll have to try and find it. I roughly remember the context of the discussion and some people who posted ITT...advanced searching mode activated
Edit: Seems like the search is capped. Searching for all threads I posted in only goes back to 2009 (last 250). Any way to circumvent that, the post may be in the archives

Might as well paste other interesting post I found while searching:
Quote:
Originally Posted by lastchance
This post is about how to win 9ers. It can be applied to other werewolf games, but in these games, the power of the seer is far less and far different than this particular setup, with one seer with a n0 peek and only 2 wolves.

This post is all about maximizing the power of the seer as village, and minimizing it as wolf. Yes, you need to make good reads to win. But this makes the game easier, and I think it will maximize your win %.

I will start at D2, because that is the easiest day to play in most 9er turbos, and it has so very, very little to do with reads, and almost everything to do with tactics.

If you are the seer, your play will depend on the peeks you have. You generally will have these 3 kinds of peeks:
If a villager is lynched on d1:
Seer with a wolf and villager:
You want to leave the villager peek out there, and get the wolf you have peeked lynched. Wait to come out, don't jump the gun, but try to put as much pressure on the wolf as possible while still not looking like the seer. Do leave out the villager peek. I think here, you should come out before nightfall if you can't get your wolf peek lynched. You absolutely have to come out if your villager peek is getting lynched or you are under pressure. I think you should come out if a random is getting lynched in this particular spot, because most likely, it will be a villager.

Seer with two villagers:
You want to get out both villager peeks well before nightfall. If either of your villager peeks is in trouble, you should absolutely come out. I think here, you don't have to come out if a random is being lynched, if you think a wolf is being lynched (and you are generally good at this game). There are 2 villagers and 2 wolves in the pile of randoms, and you can put a lot of pressure on the wolves by not coming out (possibly leading to a fake seer you can shoot down ftw?).

Seer with one villager:
Don't come out. Freeroll. List your peek, don't let them get lynched. But don't come out. Never say I am the seer unless you are getting lynched. If your peek is under pressure, I think you should also keep your mouth shut in this case, but really try to keep your peek alive. It is just not a good spot to have only one villager in this situation.

As villagers, you should be patient to majority night. Do not lynch the seer (LDO). And you should make sure that wolves cannot read you as seer or vanillager by stating fake peeks, like a seer in your spot would do. And make sure these fake peeks don't get lynched. Try to avoid contradicting those peeks until someone actually makes an "I am the seer" post. Just make sure that you are NKable if the seer does not come out.

As a wolf, unlike d1, you can be a little more aggressive bussing your partner, but a seer that does not come out is so, so dangerous. Having a wolf get lynched without outing the seer here is a loss, so you want to have a good idea who the seer is before you lynch your wolf partner, or you can just have your wolf partner fake seer under trouble if the real seer isn't coming out, and that's not a terrible way to play it. In fact, if the real seer isn't coming out and you're under pressure, I would fake very, very quickly (and I think you should fake, though possibly less quickly).

If a wolf gets lynched:
Seer:
Make a post like this (assuming you do not have the last wolf peeked):

or


Try not to let these peeks get lynched. Don't vote them, don't push that hard, but yeah, try to lynch somebody else. This is called freerolling. You get your peeks out without getting auto-NKed. It is effective.

Villagers:
Make a post to the effect of this:

or


And act like you do not want these peeks lynched. Generally, peek people you actually do not want to be lynched (or people who won't be lynched), and then you will look more like the seer.

Wolf:
You let your partner be lynched on d1. Not smart. Find and kill the seer. If you don't, you lose. Eliminate people who don't look like the seer. Yeah, there's not a lot to be done here, which is why you can't lynch your wolf partner (and it's dangerous to bus them) if you don't have a good idea of who the seer is.

D1 strategy:
Seers:
With villa peek:
If you have a villager post, you can make a post like the above on d1, or you can be quiet. Either way, it works. I do prefer making a "freeroll" type post, but w/e. Don't let your peek get lynched, etc. Claim 2 or 3 mins before night if you're leading wagon.

With wofl peek:
Don't get lynched, especially don't die with valuable information. Claim seer earlier than normal under pressure because your info is so much more valuable, and you shouldn't get lynched here. There are many ways you can play this. You shouldn't come out until 5 mins before night, and you shouldn't come out with your peeked wolf getting wagoned hard, but really, I think you can play it super aggressively and lynch the wolf while dying, or you can be quiet and live to d2 most of the time with an extra peek.

Wolf strategy:
As you can see, having a wolf lynched without finding the seer is very, very bad. If you want to bus your wolf partner, you should be quite confident that you can kill the seer that night. You should probably try to bus villagers instead, because they might lynch the seer.

Finally, d3 with 2 wolves and 1 seer alive is also an interesting scenario I should touch on, and I feel is probably misplayed quite a bit.

You need to able to counter the seer effectively. If your wolf game consists of waiting until the seer comes out and calls you out before counterclaiming, that's bad, because oftentimes in this scenario, the seer will lock it up (w/ 2 villa peeks, possibly 2 wofl peeks), so you want to counterclaim. You need to counterclaim, believably. For many players, that means coming out as soon as day start so you are the one claiming first. You also must make sure that you can accuse the seer of being a wolf, because if you peek the seer as a villager, you're busted. In any case, understand your seer game, or at least people's image of your seer game. Contesting the real seer is awfully important in this spot.

Seers can either claim at start of day, or they can wait to bust counterclaims (but if someone peeks u as wolf, that's bad IMO). I think you do a mix of both, and try to confuse wolves.

So yeah, that's a long and slightly confusing post. But yeah, that's how 9ers work.

tl;dr: If you're a wolf, don't get lynched without outing the seer. Seers should list their peeks without stating that they're the seer. Villagers should act like the seer so as to confuse wolves and make them miskill (which is bad for them).

Note: this is only for 9ers. Do not follow this specific strategy for any other version of werewolf, but concepts in this post can be universalized effectively.

Last edited by clowntable; 12-05-2010 at 11:13 AM.
Academic Werewolf Discussions Go Here Quote
12-05-2010 , 06:24 PM
Quote:
Originally Posted by Nichlemn
...no villager has any information or skill at deduction that any other villager lacks
lol
Academic Werewolf Discussions Go Here Quote
12-05-2010 , 06:30 PM
Quote:
Originally Posted by BitchiBee
because we only expect the wolf to wn less than 50% in our games because our wolves play very imperfectly
it's probably optimal to expect the wolf to WN every time
Academic Werewolf Discussions Go Here Quote
12-05-2010 , 06:41 PM
information is king in ww

the idea that any GTO strategy of ww would involve intentionally getting very little information seems absurd

even if wolves played perfectly and never gave off any tells and always acted super-villagery, and all villagers also acted super villagery, it would still never make sense to just stop posting altogether
Academic Werewolf Discussions Go Here Quote
12-05-2010 , 06:55 PM
man, this thread is dangerously close to becoming useful

i thought it'd die in three posts :P
Academic Werewolf Discussions Go Here Quote
12-05-2010 , 07:00 PM
GTO ww from the villages perspective is (which I almost don't want to admit because its pretty much how you play which can be frustrating at times) is to wagon someone, see how they react and see how others also react to the wagon, then wagon someone else. rinse and repeat.

Even even wolves always played perfectly nothing will change the fact that villagers should try and use all of the time that they have to get as much information as possible
Academic Werewolf Discussions Go Here Quote
12-05-2010 , 07:17 PM
If wolves are playing perfectly wagoning wouldn't have any usefulness since whatever voting analysis you're trying to pick up can be easily balanced.
Academic Werewolf Discussions Go Here Quote
12-05-2010 , 07:20 PM
but that doesn't mean that it still isn't GTO for the village to try
Academic Werewolf Discussions Go Here Quote
12-05-2010 , 07:21 PM
Also, if everyone knew that someone only got wagoned to test people before moving on to another target, all reactions to said wagoning, would be pretty useless.
Academic Werewolf Discussions Go Here Quote
12-05-2010 , 07:23 PM
what I'm saying is that there is never going to be a situation where less information benefits the village
Academic Werewolf Discussions Go Here Quote
12-05-2010 , 07:33 PM
not every wolf is WN...
Academic Werewolf Discussions Go Here Quote
12-05-2010 , 07:34 PM
Quote:
Originally Posted by Luckbox Inc
what I'm saying is that there is never going to be a situation where less information benefits the village
this might not be completely true when you take into account that wolves need to kill the seer

of course in GTO ww all villagers always fake seer perfectly— but hunting the seer is definitely the one area where wolves need information too

but if you conceive of ww as essentially a battle of information— where the village thrives on it and where its the wolves job to negate it's effects, then i don't see how it could ever be the case where it would be gto to not post

even if posting doesn't actually help at all, it will always be the wolves responsibility to make that the case
Academic Werewolf Discussions Go Here Quote
12-05-2010 , 07:52 PM
obviously in the game it is always hugely +EV to post, but speaking strictly from a GTO perspective the various posts people make that help our reads are meaningless.
Academic Werewolf Discussions Go Here Quote
12-05-2010 , 08:03 PM
Quote:
Originally Posted by ibavly
obviously in the game it is always hugely +EV to post, but speaking strictly from a GTO perspective the various posts people make that help our reads are meaningless if the wolves live up to their GTO-expectations
How is it GTO for the village to not force the wolves to do that?

Quote:
I've been thinking about game-theoretic optimal werewolf. With 1 wolf, no seer and no voting history, it seems trivial: everyone votes randomly at the last possible second. Anything else and the wolf has an opportunity to gain an advantage. There is no talking, because no villager has any information or skill at deduction that any other villager lacks.
The bolded is what I understand the least about Nichelmn's premise. If everything is all GTO why should posting gain the wolves an advantage but not the village?

Last edited by Luckbox Inc; 12-05-2010 at 08:18 PM.
Academic Werewolf Discussions Go Here Quote
12-05-2010 , 08:21 PM
Quote:
Game Theory Optimal (GTO): A strategy that yields the highest possible EV (or: “is optimal”) if your opponent always chooses the best possible counter-strategy.
Everything you say is accurate but you are using the wrong terms
Academic Werewolf Discussions Go Here Quote

      
m