This post is almost absolutely perfect. I mean, I do not think I know any non-metagame information that is not in this post (except for seer-hunting information?). Perhaps more emphasis on the fact that the goal of the seer isn't to peek wolves?

Yeah, this stuff is awesome.

this time I'm remembering to save this stuff from being archived

also, here are some other strategy threads I have posted:

http://forumserver.twoplustwo.com/59...agers-1072558/

http://forumserver.twoplustwo.com/59...sings-1156721/

When are things getting archived? What should I save?

you still have a month or so, perhaps more

Do you have a link to the announcement handy? This troubles me.

You will find this less than helpful: http://forumserver.twoplustwo.com/50...month-1161213/

True. What does it mean to be archived, do you know?

The archived threads would be in a different database, perhaps on a different server entirely, would no longer be accessible for posting, and would have their own search page.

This stuff was first written when I had about 9 months of experience. Now, I have 7 years of experience. My wolf-hunting techniques have evolved considerably during that time, and I no longer feel that the advice from my original article is the best that I can give. The general principles are sound, but overall they serve better to clear some players as villagers than they do to actually catch wolves among the remaining suspects.

Much of my improvement as a villager for the first several years of my career involved developing and refining heuristics, or rules of thumb. That is, I would take note of specific types of behaviors, and keep track of which ones were displayed often by wolves, and which ones where rarely displayed by wolves. As I gained more experience, I gained a bigger mental database and was able to discard "tells" that turned out to be meaningless, and I gained increasing confidence in the ones that continued to work.

In the past few years, I've attempted to move beyond the usage of "tells" and rules of thumb, and instead look deeper and more holistically at each player's posting. In doing so, I attempt to answer one question: "Do they care about solving the game?" This is something that I first posted about here:

In the 18 months since writing that post, I have continued to focus on that line of thinking while hunting wolves, and generally with good success. This technique relies on looking past the literal content of a person's posts, and looking instead at the unspoken communication. It is easy for a wolf to tell you what reads he has. It is much more difficult for a wolf to write posts which convey a sense of veiled suspicion or trust. However, villagers write these posts all the time, because those are their natural reactions. I search for people whose reactions to various events don't seem to show any judgements. I search for people who tell us their reads, but don't show us in their posting that they really believe in those reads. (For example, when I have made a case against a wolf, I find that very often they will respond to me fairly extensively, but in a way that shows absolutely no attempt to discern my motives.) I attempt to determine, at the most basic level, whether a player is attempting to solve the game, and whether he cares about what he is doing. The villager can be expected to have lots of thoughts which are put together to form conclusions. The wolf will usually be missing pieces.

Another area where I have made recent improvements in my wolf-hunting is identifying deceptive behavior. Villagers are expected to be honest, while wolves are forced to behave deceptively. Therefore, identifying deceptive behavior can be a very powerful wolf-hunting tool. This is something that comes up frequently in the way in which wolves defend themselves against attacks.

The key to understanding deceptive behavior is realizing that most people hate lying, even in a text-based game on the internet. Instead, they will search for ways to make truthful statements to mislead you. (When you catch someone lying in a werewolf game, it's almost always an innocent mistake, and should be ignored. People's memories are just not very reliable, and there is a lot of stuff to remember in werewolf games.) This nifty graphic that I found on Wikipedia shows a hierarchy of disagreement, which also works pretty well as a hierarchy of villagery defenses:



Many of the signs of deception that are found in werewolf revolve in one way or another around a refusal to address the case that has been made against oneself. Argumentum ad hominem is a common way for wolves to avoid explaining their actions. Instead, they tell you why the person making the attacks isn't trustworthy and should be ignored. You also have wolves who simply pretend to be unaware that a case has been made against them, or who address only the minor points made against them while ignoring the majority of it. In one game, I even had a wolf flat-out refuse to explain the statements that he had made previously, after I had demonstrated that they made no sense. Being vague is also an indicator of deception. People being truthful can easily explain things in detail, while liars cannot.

There are a couple of reasons that detecting deceptive behavior is very useful. The first is that the behaviors are relatively consistent for everyone. When innocent, people are hardwired to emphatically deny false charges and be honest and forthcoming to clear their name, while the impulses to behave deceptively are very powerful and difficult to overcome even for people who are aware of them (although, it's obviously much easier in a text-based game where you have time to think about your response and don't have to worry about body language -- but it's still difficult to overcome the problem of not having the facts on your side, and thus the actual necessity to deceive in one way or another). The other good thing about detecting deceptive behavior is that it's something that generally comes up after you've already made a case against someone. If you already have a good reason to believe that someone is a wolf, and then they react deceptively, you can now pursue that lynch with much greater confidence. If someone responds well, then you know that you may need to rethink your read.

These topics are very difficult to explain concisely. I hope that this post was useful.

moar articles about wolfing, pls

That was an excellent post, soah.

Moar like this please.

Eventually I am hoping to put out a completely revised edition of Werewolf For Advanced Players, but for now I'm just going to use this thread to write some essays on topics which will eventually be incorporated into the final draft. Feedback welcome.

On seer strategy:

I wrote originally some stuff about not leaving peeks, and about making sure your peeks are clear once you die. This is something I've thought more about over the last few years, and I think that the advice is too simplistic for advanced players. (For inexperienced players, I'd still recommend playing seer in a rather straightforward manner, as I described originally.)

Even at the endgame, not all village peeks have the same value. We will define the value on a scale of 0 to 1. This value is equal to the probability that that villager would be lynched if not for the peek. A peek of someone who was certain to have been lynched has a value of 1, while someone who was never going to be lynched has a value of 0.

Wolf peeks could be measured similarly, with a value of 1 being given to a wolf who was never going to be lynched, and 0 for one who was certain to be lynched.

When you are the seer, you can then estimate the value of your peeks, and incorporate that into your strategy. The most obvious spot for this to happen is on Day 1, if you have had a pregame peek. Since you were forced to make a peek with no knowledge of how events would unfold, you will often end up with a peek that has low value. In these circumstances, you have additional incentive to not leave any hints of your peek, because the downside risk is low: If you are nightkilled and the wolves subsequently nightkill your peeked villager or the village soon lynches your peeked wolf, then it made no difference to the outcome of the game that the village never knew who was peeked.

For example, in a 17er I once peeked a wolf on n0. On d1, he did not show up until late, so I did not get much opportunity to try to lynch him. So, I just made sure none of my other posts might be misinterpreted as peeks, I made sure it was clear that I had a strong wolf read on him, lynched someone else, and then hoped for the best. Then, something interesting happened. On n1, the wolves killed the same player who I peeked. And at the start of d2, lots of people wanted to lynch the wolf that I had peeked. At this point, I realized that I had no value from my peeks. My village peek was dead and my wolf peek was going to be lynched. So, once again I was careful that nothing I posted might be construed as a seer peek -- including posts about my peeked wolf. As it turns out, he was lynched that day. On d3, I had a valuable village peek, but I left no hints of it, because I believed that the wolves would rule me out as a seer candidate if I'm still leaving no peeks after three game days. They are expecting the seer to have lots of peeks that should be made clear. Instead, I had very little information, and it was not worth risking a night kill just to make me one peek clear. (I survived the night. The following day, I was forced to claim seer, as a villager chose to fake claim seer with wrong peeks, putting both of my peeked villagers in immediate danger of being lynched.)

Had I been killed on n1 in that game, I feel that the village would probably have lynched my peeked wolf, even though I did not make it clear that I had peeked him. Since I had left no clear peeks, some villagers would have speculated that I had peeked that player wolf, since I had made a couple of strong posts about him. There would not have been any agreement on it, but it would accomplish something important: It would force players to look carefully at my peeked wolf. The wolves who are hard to catch are usually the ones who manage to avoid attention. Some players already considered that player wolfy, and my death as the seer would add some additional heat. Had I left a clear hint on him, he would have been lynched with probably very little discussion. With the added ambiguity, he probably still gets lynched soon, but the village gains the opportunity to gather more information by having him properly discussed, wagoned, and lynched.

In another game, I peeked a villager on n0. On d1, he was attacked by only one person, leading me to believe that the value of the peek was below average. The person making the attacks was someone that I already had as one of my top wolf suspects. Rather than defending my peeked villager, I instead just led a strong charge to lynch the attacker, which succeeded. I made it clear that his attacks on my peeked villager were wolfy (which was true). I hoped to accomplish a couple of things. First, I conditioned the village to think that attacking my peeked villager is bad. Second, I remove a player from the game who wants to lynch my peeked villager. This decreases the chances of my peeked villager ever being lynched, even if my peek is never known. The player I lynched was a wolf, I survived the night, and my peeked villager was getting cleared as a result of the d1 events. I peeked a wolf n1, led his lynch without leaving any peeks, and had the misfortune of getting nightkilled despite my lack of peeks, as the remaining wolves were both inactive and had no time to seer hunt. No one figured out my peeked villager aside from the wolf who I had lynched on d1, but he was nightkilled anyway for being clear, and the village won.

For the record, I did leave a clear but very buried peek of my villager on d1, hoping the wolves would overlook it in a wall of text. As it turns out, the village overlooked it too. This is one of the risks of not being clear. But the story also illustrates the good things that can happen: Given my reputation, I was probably much more likely to have died n1 if I had left a clear peek, but instead I was able to live and peek a hidden wolf, gaining the village tons of information on d2 as the hidden wolf was successfully wagoned over a villager, ultimately leading to a village victory. (And I think I was unlucky to have been nightkilled. The wolves who killed me had no idea that I was the seer, and the wolf I lynched on d2 wasn't sure that I had peeked him even in the postgame discussion. The village had assumed that I did, though, so he would have been lynched quickly even if he had evaded the d2 lynch prior to my death.)

These examples show just a few of the many ways that you might adjust your play based on the value of your peeks, and the benefits to sometimes not leaving too clear of peeks too early. I think that there is even merit to posting things such as, "I peeked either X or Y wolf." In my experience, the village does well when it knows for sure that one of two people is a wolf (for instance, I believe the village lynches correctly a high percentage of the time when two people claim the same role). The benefit from making this play is that it increases your ability to give seer cover as a villager, if people believe you would make that post as a real seer. Or, if they don't believe you would do so as a real seer, it would allow you to live longer as the seer while still leaving behind useful information and avoiding any chance that your peeked wolf survives all game to win. As long as you are presenting the village with the choice of a peeked wolf and someone else who could also be a wolf with your peeked wolf, there is limited downside to the strategy: the worst that happens is that the village lynches the wrong player first, but if that player was likely to be lynched regardless, then the outcome of the game remains the same. (I suppose the true worst case scenario is that the other player is lynched first, is a wolf, and your peeked wolf is never lynched. But this hopefully wouldn't happen often, and the village did at least benefit from an immediate wolf lynch, just the same as if you had left a clear wolf peek. You gave the village an opportunity to glean a lot of information from having w/w wagons, and it's not your fault if the village misreads it.)

I present this strategy not so much as a direct endorsement of it, but rather to further illustrate that the goal of the seer, as with any villager, is to increase the chances that the village wins, and that this goal necessarily requires the seer to affect as many lynches as possible, especially the future lynches. If the seer's information does not affect any lynches, then the seer has been useless. This important principle should help guide the seer's decisions each day. As long as the peeked villagers don't get lynched, it is not very relevant why they did not get lynched. The same is mostly true of getting the peeked wolves lynched, although here the goal should be to maximize the information that is obtained in the process of their lynch, hopefully relatively quickly.

If you take a risk as the seer and the village loses, lots of people are likely to be upset with you. However, that in itself is not a good reason to play a very risk-averse seer strategy. The value of the seer and his peeks increases dramatically with each day that passes. Aside from just the sheer value of the peeks, there is also much value from forcing the wolves to seer hunt each night, leaving them worried each day about who might have been peeked, and just generally delaying their ability to plan for endgame with perfect knowledge of the situation. It is often a good gamble to take some chances at losing small amounts of information early in order to improve your chances of being alive later in a situation with much greater leverage. When you have information that's of low value, you should prioritize survival and acquisition of additional information. With valuable information, you should play more conservatively.

This essay has focused primarily on situations where the seer has information of low value. I don't wish to imply that all situations are like this. There will still be plenty of times where straightforward play is called for. Even when your peeked villagers seem safe, you can't be sure that the situation will be the same at endgame. Don't gamble in spots where your information is already valuable. Just leave your peeks clearly, and then hope that the wolves kill someone else.

Surprised that this of all posts hasn't generated some controversy.

like anyone has read that post

so many words. maybe you can summarize it in like a sentence or two

I read and enjoyed reading both of the recent posts. It's good stuff

how about a werewolf strategy for casual players

tl;dr, for now

the villager article was very good(the recent post)

haven't gotten around to the seer one yet

#YOLO

i agree w/ soah's writeup. the obviousness of my peeks is directly correlated to how valuable they are to the village.

more people should seer with this in mind.

seems that too many people choose peeks based on whose role they'd like to know as opposed to which peeks would be the most useful at end game

personally i prefer to peek my strong villager leans who i think will be easily mislynched at end game

somebody should write a wolfing article about how wolves massively overrated being difficult to read and showing up on villager lists. you gotta be willing to be blatantly wolfy from time to time if it advances your team's disposition.

this.

pretty sure dean and i have the same seer strat. i want to peek someone who i already gave a village lean to the day before so that my defense of them will seem natural, as if i'm a villager trying to just drop a correct peek based on a strong read.

the problem with peeking out of your wolf reads is not only that peeking villagers is generally an easier way to seer, but that if you get a village peek of them you will have a tough time defending the next day without looking both seery to the wolves and strange to the villagers(they don't seem to like strange).

seer strat is tough to nail down as far as giving advice to people in general though because it is so player and lineup specific. obv my seer advice for charry vs someone like UD would be completely different.