lol, yes, that has been removed from my game forever

Again I'm not advocating coming to the conclusion in blue, at all.

To put example 2 into a real world context. Imagine in an anonymous game two players make their first posts as crossposts and they're both single word vote posts for the same player. The mistake in example two is giving one of them a wolf lean and one of them a villa lean. This is clearly a mistake but it's not one I think any person makes, which is why I think it's a bad example.

Possibly a better example would be two people in an anonymous game defending player C for the exact same reason at the exact same time. Giving them different reads for this would again be a mistake but it's one that is more likely to actually be made by players.

I'm fine with that, that's a better example.

In the OP you stated that Hero had no opinion on Player C's role.

In the world where Hero does know Player C's role your analysis of just the {actions} part is still problematic. Actions which are pro-village are not necessarily more likely to be done by villagers than wolves. Consider the scenario where a wolf is lynched on day 1. A wolf who is getting voted by many other wolves is much more likely to be lynched than a wolf who is getting saved by the other wolves. Thus, when a wolf gets lynched on day 1 you would expect a higher busing percentage than the times a villager is lynched on day 1. (conditional probabilities) If you are equating pro-village actions with a likelyhood to be a villager in this spot then you could be making a huge mistake. Also people routinely equate pro-village posting styles (logical, calm, coherent, rational) with being a strong likelihood of being a villager which in itself deserves to be very high on a list of classic mistakes.

You also don't necessarily know that wolves' voting accuracy will be worse than that of the villagers unless you're introducing metagame reads from outside of the vacuum. Villages that lynch at random will usually lose without help from the seer so in an equilibrium where wolves vote for random wagons and villages base all lynches on meaningless data (voting records) it's the wolves who have the best of it. And that's not even considering the potential for manipulation by the wolves in which they choose to be right about things where their vote didn't impact the outcome but choose to be wrong in very important spots, while still maintaining an average voting record overall.

Essentially, any basic criteria of who should be lynched which is derived off of things which are easy for the wolves to manipulate is not a recipe for long-term success. Wolves learn very quickly to exploit these things and the ongoing adjustments leads to wolf behavior which not exploitable with level 1 analysis. Which brings us full circle: Votes in a vacuum are meaningless.

I agree with CPHoya right now less than I agree with the Pope's position on condom use.

Does this say what you meant to say, or the opposite?

What I meant to say.

This is the most helpful phrase to pull from the entire thread. All this conditional "if A is a wolf, then B is a blah blah," or "because the NK was C, then A can't be a wolf" logic that goes on in games is just lazy thinking, and a poor replacement for real analytical and tone reads.

And as we see from this thread, conditional reads are easily misinterpreted, misapplied, or disagreed upon. When you make a long-winded analysis of other players or the game as a whole, your thoughts are transparent to everyone reading it. I'm struck by how much shorter most people's posts are nowadays. I suspect it has something to do with people merging their wolf and villager games, which is not something most players were concerned with in the early days.

Who are you?


this and the fact that most players give reason for their reads

If all we ever did was vote, this would all be fine, but we don't. But it's mostly right.

Back in the day, villagers wanted to look villagey, not ambiguous, and the better players didn't rely on getting a d1 pass to get away with not bothering. So the game is a lot easier for the few players who can sit back for a day or two but harder for everyone else.

i think i'm going to skip this thread

This is an insightful post.

It is prolly, unconsiously, why I prefer not voting day 1 until very late. i.e. when context becomes more important.

You are, in effect, championing a style of play that abandons some of the available data because of the possibility for wolf manipulation of that data. I don't know why, because that data just gets plugged into {history, experience, meta, tone, slips, psycholgy, etc.} to offset the {actions} component.

You've basically decided that {actions} is neutral, always, forever, and that you can play based solely on {history, experience, meta, tone, slips, psychology, etc.}. I think that's possible in your highly unique and specific case, but not true for most other players.

You also continue to state that I'm championing something that I'm not, which is a rule to be applied in real world applications. I don't think that lynching a wolf means that you shouldn't be lynched ever. I do think it suggests that there are better options for lynch in the early game, which can be offset by the equation I keep parroting. I still am not sure you really disagree with the use of the equation, since it obviously contemplates {history, experience, meta, tone, slips, psychology, etc.} offsetting a pro-village wagon or vote.

Finally, you have gone full circle only if you believe specifically that wolf manipulation is so prevalent as to cloud entirely the ability of a village to interpret {actions} at all. I don't believe that. You do. So I'm not surprised, in that context, that we disagree.

I do have to ask, though: if {actions} is always null, and pro-village wagons and votes are always null, aren't you just left with interactions and accumulated meta? If so, how can someone with your meta advocate that new players play the same way? They don't have your accumulated knowledge.

okay well i'll say ONE thing

If you vote for someone and there is no chance of them being lynched that day, I will ignore the object of the vote and its relationship to my reads as an important indicator of role. I will not ignore: the rationale for the vote, the timing of the vote, the fact that you could have voted someone else who IS a viable lynch but didn't.

I am sure there a lot of people whose only means of finding wolves is identifying the players that they disagree with - and all of those people are not good at being villagers. And when they start to figure out that their methods aren't leading to them finding many wolves, maybe they will realize that people who disagree with them are probably doing a better job of finding wolves.

I call this the "blind squirrel finding nuts" fallacy of werewolf. Example:

Player A thinks Player B is a villager.
Player C votes for player B.

Player C and B are both wolves, but Player A clears Player B and votes for Player C.

Yay, you were right! And you were also wrong. Also consider the related scenario:

Player A thinks Player B is a wolf.
Player C defends Player B.
Player A might vote Player C, if he thinks that players he disagrees with are wolfy and votes as such. Player C could be a better villager with a better read than player A. Or Player C could be a wolf trying to clear himself by defending a villager. Or Player C could be a player regarded as a good villager and wolf, who is defending a wolf as a wolf, and now what?


The point is, overanalyzing these relationships don't help you find wolves. The most important thing is who votes who when a lynch is on the line. That's why clear bandwagons are always important, and you should call out and identify AFK players and players who vote away from the main bandwagons later in the day. Doesn't mean there's always a wolf bandwagon available, but even then you get good reads later (would a wolf start a CFD on a wolf if the top two wagons were villagers? etc). So it's up to the village to make good reads and solid cases early in the day and start voting so we don't get caught up in all the stuff that doesn't matter for wolf hunting.

note: i did not reread this post

i disagree with soah on very many werewolf related topics but i know he and i agree on this particular one

Then help us out yo, show us why you think I'm completely wrong.

That's what I was trying to say here:

Let's put a little meat on why conditional reads are bad.

There is an action. Based on some internal rule of mine, I find the action wolfy. Now my internal rules are necessarily flawed and probabilistic. So my read is going to have some confidence level, or error rate, or what have you that expresses the value of the read. My experience is my earlyish wolf reads are approximately 50% wolf.

Now for a read on a read, I'm taking my initial read (which was only 50%!), assuming it is true, and applying another internal rule that is necessarily flawed and probabilistic. 50% of 50% is 25%. The probabilities are such that in my experience, a 2nd level (derived) read is just noise.

Which one?

That votes that matter should be counted more heavily in context?

I think everyone does. . .

I think Herbie and I are saying that it's probably not just noise, but it's not a whole lot better than just noise, so it's by definition very thin, so we're not going to push it.

Which ends up resulting in the same play you would prescribe anyway.

I want one of the pros in here to explain to me how new players are supposed to "find wolves" when they don't have the accumulated knowledge and experience that the pros obviously have.

EDIT: I mean, it may well be better to completely ignore wagons and **** and just figure out **** with sick tone reads, but you can't really tell me that newer players are supposed to mimic that. That'll just be a circus.

I would mostly just be paraphrasing soah at this point.

I really think the disagreement between Soah and me is philosophical or semantic. I don't think he's actually saying anything I'm not.

He just thinks that wolves can cloud {actions} so completely that it's not even worth considering.

I think you have to consider it, that it's not as clouded as he thinks, and that, in any case, you can resolve the clouding by using {history, experience, meta, tone, slips, psychology, etc.}.

Neither one of us is advocating a dumb level 0 follow the logic approach to WW.

Oh, and people are objecting strenuously to the heuristic, which is totally fair but I can't think of a better heuristic right now.

I don't want new players to "find wolves" I want them to try. I want them to use whatever faulty logic they like and come to perhaps completely wrong conclusions. I then want to analyse their train of thought as they attempt to come to conclusions and deduce whether it's genuine or fabricated.

like, here's how to toneread for newbies:
villagers have their detective face on and are mean.
wolves have their poker face on and are nice.

tell the newbie this, and get him to try and make some reads. You should be able to see a marked difference between a wolf newbie doing this and a villager newbie doing this, and for bonus points the wolf newbie will probably spew a couple of people's roles trying.

I'm afraid, day1 there isn't really anything but tonereads, you can make guesses as to what the wagons are and thus what the people on the wagon are, but you have to toneread the person being wagoned before you can say anything about who's on it.
but once they flip, obviously you can analyse wagons.