Bryan Caplan  

Deception, Detection, and Democracy at GenCon

Hayek in "Unbroken"... Nomadism...
At this year's GenCon, I discovered a thought-provoking game: "Are You a Werewolf?" - a.k.a. "Mafia."  The game's a brilliant exploration of cheating, cheater detection, and democratic information aggregation.  The basic set-up:

1. There are 15 players and a moderator. 

2. Each player gets one card, which you may not reveal until you leave the game.

3. Twelve of the cards say "Villager," two say "Werewolf," one says "Seer."

4. Each turn, the moderator tells the whole village to "go to sleep."  You're on your honor to keep your eyes shut until the moderator tells you to open them.

5. Next, the moderator tells the werewolves to wake up and silently choose a single victim to eat.  The werewolves then go back to sleep.

6. Then the moderator wakes up the seer.  The seer points to one other player and the moderator signals whether that one person is a werewolf.  The seer then goes back to sleep.

7. The moderator wakes up the village, revealing the latest victim.  The victim shows his card. 

8. The villagers then argue about "Who's a werewolf?" until a majority of the survivors pick a victim to lynch. The victim reveals his card.

9. Return to step 4.

The game ends once (a) both werewolves are dead, or (b) the number of werewolves equals the number of villagers.

With random elimination, villagers win 39.5% of the time.  In practice, though, the villagers won all six of the games I played.  My sense is that the villagers win about 90% of the time.  The fundamental reason is that villagers have a greater-than-chance ability to detect cheating - even when the cheaters are struggling to conceal their cheating.  How?  From demeanor, words, and behavior.  My werewolf status was almost immediately uncovered in my first game because I incompetently concealed my smug delight. 

Playing Werewolf engages basic human skills at a high level.  What exactly do cheaters look like?  Excessive happiness and unnatural poker forces are both strong signs.  But so is flattened - or extremely bimodal - affect.  Honest villagers tend to be moderately emotionally variable.  Of course, if you repeatedly play with the same people, you should add "fixed effects" - look for deviations from the person's normal demeanor rather than deviations from people's normal demeanor.

As I played, I thought that my cheater detection skills improved sharply.  My deception skills, not so much.  My knee-jerk conclusion was that detection is easier to improve than deception.  But this article points out a plausible alternative hypothesis: Practice matters.  You practice deception in 2/15 of games, and detection in the remaining 13/15 of games.

There are already small academic literatures on Werewolf in mathematics and psychology - see here and here for starters.  But a few hours of play convince me that it deserves far more attention. 

Has anyone else played?  Any social scientific insights you'd care to share?

COMMENTS (23 to date)
joeftansey writes:

In larger games with more werewolfs working together, discovering 1 werewolf often leads to the discovery of the rest based on who was cooperating with them. This needn't be the case, but it seems like its hard for werewolf players to sacrifice short term overt cooperation (like drawing attention away from real werewolfs or voting against their execution) in order to enhance long term team survival.

There's also an extent to which werewolf players feel like they lose if they die, even if the werewolf *team* wins. So this probably skews things as well.

Though with perfect strategy, I think the game does reduce to pure chance.

Julien Couvreur writes:

I played a few times and didn't get detected the one time I was a werewolf. Maybe it was a fluke ;-)
My strategy in terms of posture and behavior was to observe myself during games I was not a werewolf, so that I could imitate myself when I was trying to conceal being one.
In terms of chatter, you cannot really "convince" people that you are not a werewolf, because there is so little "real" information about who is who. So my strategy was to subtly shift suspicion, rather than defending myself. When someone would accuse me, I'd say: "I see why you could think that. There is so little information to go by, I'm having trouble myself figuring out who is a werewolf. I've been thinking about *that* guy which I found acted weird/etc.".

asg writes:

I offered to run this at Capla-Con but schedule did not permit. I recommend "The Resistance", a similar game but one which does not have player elimination, thus ensuring all players can remain engaged throughout the game.

In my experience playing Werewolf, I often played with the same group, and quickly became known as a high-payoff target. If I was not a werewolf, the werewolves would often eat me first, and even when I was not, I would get executed early on, since the villagers saw my execution as a safe play (this was the consequence of being one of the two systems designers among a bunch of writers and content makers at a game company). The takeaway was that repeated play and reputation matter a lot in social games like this.

AMW writes:

I've played intermittently with friends and family for a number of years. We call it Mafia (i.e., werewolves = "mafia", villagers = "townspeople" and seer = "night watchman"), but I'll use the werewolf nomenclature for consistency.

One strategy that I will often try to follow is to disagree with my fellow werewolf about who to lynch. In fact, if the villagers are dead certain that my partner in crime is guilty, I will vote right along with them to distance myself from suspicion.

I've noticed that, due to the low information availability, in a number of games a coalition of villagers will form without any substantial goal of actually rooting out the werewolves. Instead, lynching becomes an entertaining way of flaunting your in-group status. Accusations will then be made by coalition members against out-group participants with no real suspicion of guilt.

One variant of play that we typically use is to add a fourth role, called a doctor (in werewolf parlance "healer" might be preferred). The doctor makes his move after the seer and picks one player to heal. If that player was targetted by the werewolves he is saved. However, the moderator does not inform the group who was targetted, as that would divulge information of innocence.

Sol writes:

Discovered Werewolf years ago at game night. I've played it a few times, watched it played a few other times, and IMO it is a hateful game. There's nothing real to go on at all, so it just becomes a game of who you can convince your friends to kill. Very distasteful.

I wonder if part of my attitude might be because of the "friends" thing, though. It would have a very different feel played with 15 random strangers. Still suspect I wouldn't like it.

On the other hand, the Spagetti Western card game Bang! has a similar randomized "good guys and bad guys" secret assignment, and I love that game. I think maybe the difference is that who is who is usually determined by actions that happen in the game; for instance, if someone starts shooting at you, you generally assume they aren't on your side...

sabre51 writes:

I've played this game many times. My general strategy:

As a villager, I say almost nothing until I have some real information- heard someone move, find something seemingly contradictory in statements, etc. I do feel this makes it more likely that I will be executed since there is nothing to deflect suspicion from me, but since finding werewolves by chance is very rare I do not consider this a large sacrifice for the villager team. When I DO have concrete information, I will often immidiately say, "I am a villager, and I am almost sure. I volunteer to be executed next round to prove this if we can kill the werewolf now." Usually, when one werewolf is killed you can figure out the other one based on cooperation, and you only need to get lucky once (including the seer using my same strategy) to win.

As werewolf, I try to put on a display of thoughtfulness the first round or two, not accusing anyone or being too forceful defending myself- usually there are enough of those people that you can survive two rounds without participating too much. Then, when someone else makes an accusation with any sort of reasoning behind it, I come in as the clear-thinker and support them. I have found this almost always works to kill the person they suspect, and builds trust with that person so that I have an ally in the future. I am not sure the overall win % in general, but I would say I can win 45-50% as a werewolf and 70-75% as villager.

PrometheeFeu writes:

I would agree with asg about reputation effects matter across games. I once managed to win a very large game as a werewolf in a way which I found quite masterful. Everyone else agreed that I had been very clever and they therefore proceeded to lynch me first for the next half-dozen games.

In one variant of the game, it is revealed whether the mob's victim was a werewolf or not. When that is the case, I have found it a high-payoff strategy to establish credibility with the villagers by identifying my fellow werewolf and joining the mob against them.

Something else that I have found is that you must keep a delicate balance between throwing accusations recklessly and holding back. Both are often seen as evidence of guilt. I think this is more rational than it might seem. A werewolf might want to throw off suspicion. As such, they would want to ensure that someone else get picked. They have two options: 1) try to go unnoticed, 2) select a scapegoat and try to make sure that scapegoat gets it. So anyone who is too eager or too slow to accuse another is likely to draw suspicion to themselves.

The additional roles can make all of this very complicated and interesting. For instance, declaring yourself the seer might be a dangerous proposition. After all, the werewolves also have a strong incentive to claim the authority of the seer. As such, a good strategy is to lynch the self-declared seer. If they were lying, you most likely just got a werewolf. If they were not a werewolf, you can simply lynch the person they were accusing next time. Of course, that significantly decreases the chances of the seer ever manifesting themselves, not to mention the fact that the werewolves might kill them.

Finally, pre-existing social relationships make all of this very complicated. In all gamer groups (in my experience) there are a few people who have mock-feuds of sorts. These things come out quite obviously in this game. Similarly, some people have strong reputations for being adept or inept at deception and so might be implicitly trusted or not trusted.

Michael writes:

I played one game where the moderator made people leave the room after they had been killed (it was our last game of the day, after having played 4-5 rounds).

As it turns out, the moderator had fixed the deck and he was the only werewolf.

It is probably fun/funny when done 1 out of 100 games. All should give it a shot.

Tom West writes:

Werewolf is a lot of fun, and I thoroughly enjoyed "The Resistance", but in my limited playing of the latter, I found it too easy for the "bad guys".

In my opinion, the trade off for the greater fun of playing the "bad guy" should be that it's much harder for you to win.

And yes, I tended to be a leading figure in "find the werewolf" in many games, which worked to my advantage when I was finally the werewolf.

Unfortunately, if you win by killing all the people who were 'trusting your leadership' (so to speak), you're pretty much guaranteed to be lynch fodder for the next 20 games :-).

Skeptic writes:

Isn't a more likely reason for the better than chance performance of the villagers simply that one or more are cheating by keeping their eyes open while "asleep"?

asg writes:

Skeptic - in Werewolf there is a moderator who runs the game and keeps his/her own eyes open at all times, so there is some enforcement.

David writes:

When I have played this game in the past, nobody reveals their cards, even after they've been lynched/eaten. That way, a villager coalition can never be positive that they have pinpointed the werewolf coalition. This seems to increase the probability of werewolf victory.

Lee Herridge writes:

We always play the game as Mafia and what Julien said about convincing people is dead on. Those who first start playing usually get lynched when accused because they think they can give an 'in-character' (an explanation from within the fictional world) defence. In-character defences don't work because everyone is fully aware they are playing a game and so are unable to suspend their disbelief.

In my experience (which is mainly as someone who runs the game), the best defences are a combination of rational strategy (explaining how it wouldn't make sense to act the way the Mafia has), good old-fashioned deception (esp. if you're Mafia) and scapegoating (casting blame on real or perceived enemies). Some times, you hit a winning streak because you have been proven to be particularly trustworthy or circumstances allow you to escape; other times, no matter how convincing you are you still end up dead, usually because of bias.

I have only seen in-character defences work one time, only because we essentially ran a How to Host a Murder party with Mafia rules (so everyone role-played a particular character for the night). To role-play, you suspend your disbelief to act as the character would so in-character defences become plausible.

I guess this is probably analogous with pure politics. People aren't perfectly rational and because information is incomplete, we can end up siding with someone we otherwise wouldn't. Arguments are as much about emotional appeal as they are about facts and truth. There are both pre-existing power blocs (the Mafia) and ones that will evolve over the course of the game (players realising they can trust one another). It is a very interesting social experiment if you think about it.

Henry writes:

I haven't played a lot, but on a forum I post on there is a sizeable werewolf community, with many games played in this subforum. Here is an advanced strategy guide posted by a regular player.

Swimmy writes:

Try playing it on IRC. The social deception is a little bit easier, but the best strategy for werewolves is still to set themselves up as logical and inquisitive in the beginning. Sometimes, if playing with more than 15, it is a good idea to accuse your fellow werewolves to cover yourself up. Even if this turns into an argument, the villagers will probably only kill one of you, and might trust you on later rounds if they're not too practiced.

Tao of Gaming writes:

I second the recommendation for the Resistance, a game with no player elimination that also gives you concrete information much sooner (rather than, "He died, we should lynch X." Also, the Resistance is much faster, and scales from 5-10, which are easier to get to outside of a convention.

There is also "Ultimate Werewolf" or "Werewolves of Millers Hollow" (or variants online) that give many more of the villagers special abilities, which makes the game more interesting and provides more information. For example, one pair of villagers may absolutely know the other is human. The bodyguard can pick someone before the werewolves wake up (in secret, known only to the moderator) and if they werewolves lynch that person, they fail (but the bodyguard is wounded and can't use the special ability again). Etc etc. (To balance, there will often be a pro-werewolf villager). The game can get as complicated as you like. I've seen convention games with 20-odd players....personally, I prefer games with more actual data, but to each their own.

Also in the genre is Battlestar Galactica, but that has a lot of rules and can take 3 hours your first few plays. (With experience, it falls to 75-90 minutes).

Willem writes:

We play with the same people quite often and what you even see is that people build a reputation of being a better-than-average-liar get staked / eaten (even sacrificed) early.

So once you're a lot better at hiding your deviations from the rest, you basicly are out of the game soon and often.

It's funny how Sol thinks it's distasteful, since in my opinion it's not about killing but about strategizing. We had an off/online game that lasted for about two weeks. You just see a lot of different sides of people in such a game.

wesley writes:

If I was mafia or werewolf, I used to vote to kill (or eat) the other mafia guy (or werewolf), but I didn't broadcast my vote to everyone, just to a few people. That helped ensure that those people would be on my side for the last rounds, when it gets tough.

It can also be quite clever to secretly kill the person who's been accusing you during the open rounds, and then claim that it's a set-up. Risky move, but if you can pull this one off and then the kinslaying, you've got some serious credibility.

Lode Cossaer writes:

Usually, this game is played with more werewolves if there are 15 people. In Belgium it's actually quite popular. We also have other 'roles'.

- The witch (which can save someone once from the dead and can kill someone once)
- Cupid; who get's to pick 2 people who are 'in love'. They form their own coalition over and above the villagers and the werewolves. (So if we have a cupid relation between one werewolf and one villager, that's where the allegiance lies.) If one of the two loved one dies, the other dies as well, so they have to protect each other.
- The hunter; when he dies, he can 'take someone with him'. (Irrelevant of when he dies.)
- Peeping thomas; he can 'cheat' during the night. (Must be careful though, or the werewolves will kill him.)

It adds to the game tremendously though.

Garrett writes:

I'm surprised we didn't play this at Caplacon.

My favorite strategy is playing as a Mafia and then pretending to be the Watchman. You can sell out a fellow Mafia and no one will suspect you.

Tom West writes:

I never found the 'peeping thomas' character (otherwise known as the boy who cried wolf) worked particularly well.

If a person claiming to be the boy fingered a non-were-wolf, then he was next to be lynched, so there was no percentage in pretending to be the boy.

Consequently any time the boy successfully 'cheated' and spotted the wolves, they died pretty much immediately.

Jeremey Arnold writes:

Wait a minute... Bryan Caplan was at Gen Con in Indianapolis, and I DIDN'T KNOW???? I'm so disappointed... I wish I had known you were coming to Indianapolis... Please tell me you'll come next year! I'll tell you what, just to provide further incentive, I promise to take you to a very nice dinner and not ask you a bunch of annoying questions... just really good ones about Public Choice Research!

Sean Tompkins writes:

Here are two of the most interesting things I've observed while playing this game -

Staying quiet until called out on it keeps you out of the spotlight. The extroverts dive right in to talking about strategy and casting suspicion on others - but that suspicion quickly gets turned back toward them. The introverts sit back quietly observing, and are usually highly represented in the endgame.

The best way to gain a "known good" reputation is to come VERY CLOSE to being eliminated by the village. If someone is just a vote or two away from being eliminated, but consensus isn't reached, the general feel from the group becomes "we've already tried that, and it didn't pass" so suspicion moves on to someone else. It's always interesting to me that being moderately suspicious is better defense that being "pure of heart and mind" innocent...

Comments for this entry have been closed
Return to top