Bay 12 Games Forum

Please login or register.

Login with username, password and session length
Advanced search  
Pages: 1 ... 8 9 [10] 11 12 ... 16

Author Topic: Mafia Theory  (Read 76154 times)

TheDarkStar

  • Bay Watcher
    • View Profile
Re: Mafia Theory
« Reply #135 on: April 17, 2015, 03:54:40 pm »

To be fair, I also was lurking because my motivation to play was critically low.  I never should have joined that game in the first place.  That said, I have in the past seen value in tactical (not strategic) lurking as scum.  The idea is "not be around" at a critical juncture when things are swinging (often day end) so you aren't there and forced to commit to a choice you don't want to make.  It's a bit underhanded, and better scum than I could be there anyway and make a convincing argument for others to take the path that benefited scum the most, but I will say I've done a tactical lurk in the past.

I've certainly used this in the past. In past games, I've stopped posting for a few hours or a day if I start to get overly scrutinized. It's effective as town, too, but less so if the scum are trying to push a lynch on me.
Logged
Don't die; it's bad for your health!

it happened it happened it happen im so hyped to actually get attacked now

Jack A T

  • Bay Watcher
  • Mafia is What Players Make of It
    • View Profile
Re: Mafia Theory
« Reply #136 on: April 19, 2015, 09:41:16 pm »

I've also used short-term lurking as town and scum.  Not often, but sometimes.  Sometimes, I like to see what people post without my intervention.  At other times, I just want to avoid too much attention.

As for rewriting the OP, we definitely could use a larger discussion.

Specific points:
*The earliest copy of Dakarian's guide I can find is from all the way back in 2009, so it is over half a decade old.  The last BM that included a full copy was XLI, after which the majority of the guide was accidentally cut off (which is probably an issue).  The guide is not without value, but it is quite distant from the current meta and it advocates a rather risky technique to use (with all the fake certainty involved, there's a good chance of getting yourself lynched).  It would be nice to see some of the old hard push scumhunting again, though, I think.
*The first BM with the cryptographic hashing rule was XLII.  There's no clear BM-specific discussion I can find that led to its addition, but there had been some discussion/criticism of cryptography in KotM 3, a couple months before BM XLII.
*The scumtell list disclaimer definitely needs strengthening, and the value of meta should be included.
*OMGUS really doesn't work all that well as a scumtell.
Oftentimes I find it's easiest to scumhunt people who are attacking me by pointing out the flaws in their arguments and challenging where their accusations are coming from.
Same here.  People tend to know a lot about their own actions, making it easier to identify flaws in other descriptions of their own actions.
*I've definitely seen scum try too hard to look town (good example: Hector13's overtownwashing in BM LIII).  It's really hard to clearly identify or prove, though, and of all the entries in that list, the trying to hard one has caused the most confusion.
*Lying is a general scumtell, though identifying the reason for lying is important.  It's not the super-scumtell some people think it is (there being the possibility of gambits or of certain PRs trying to hide), but it is generally scummy.
*Rolefishing should have a disclaimer added about it being a classic tell for some PRs (protective, especially) too.
*Argembarger's part doesn't help much, especially in BMs (full of inexperienced players, the most likely to break down like that whether town or scum).  Still, I have a vague feeling I've dealt with scum appealing to emotion like that.  Null tells, I think.
*The link to the Notable Games Archive should be updated.
« Last Edit: April 19, 2015, 09:43:01 pm by Jack A T »
Logged
Quote from: Pandarsenic, BYOR 6.3 deadchat
FUCK YOU JACK
Quote from: Urist Imiknorris, Witches' Coven 2 Elfchat
YOU TRAITOROUS SWINE.
Screw you, Jack.

4maskwolf

  • Bay Watcher
  • 4mask always angle, do figure theirs!
    • View Profile
Re: Mafia Theory
« Reply #137 on: April 19, 2015, 10:36:57 pm »

*Wow that guide goes all the way back to the fifth BM... that's why it's so out of date.  I read the whole thing and it isn't any better at describing the current meta.  I'm kind of interested in how the meta has evolved over time, though, from the hyper-aggressive tunneling behavior promoted in Dakarian's guide to what it is today.
*Having seen the discussion on cryptographic hashing,  I now see why it was forbidden in the game.  That was more of a curiosity than anything else, I didn't think it should be brought back.
*I personally would say that meta is THE strongest weapon there is, which is why so many BM's end in town loss.  Like, seriously, the one time I can remember pegging scum with scumhunting is by analyzing their behavior in terms of their past behavior (this was Toaster in FHPBYOR, and it didn't matter since he died overnight).  That's what makes new players so devastating as scum: if they don't trip over some silly thing and get lynched, they'll rarely be caught because nobody knows how they normally play.  Similarly, this is why new town players usually don't find scum, because they don't have a read on anyone else over the course of several games.
*Agreeing on the OMGUS thing.
*As we've both stated, the TTHTAT scumtell is incredibly difficult to prove and incredibly subjective.  hector13 is probably one of the few people I can actually apply it to, but it's a relatively rare strategy nowadays and again incredibly meta-dependent.
*I'm mostly arguing against the use of lying as a super-scumtell, as you put it.  It is definitely scum behavior under many circumstances, though.
*Agreed here.
*I have a vague recollection of somebody mentioning Wuba being fooled by and AtE done by Vector in one of the early BM's (and, if I'm recalling my facts correctly, Vector's first game).  Not sure what type of AtE, though.
*Hehehe yes it should be.

Now for some more points of my own:
*I've found that under a lot of circumstances, if you're in a tight spot and there's one lone person talking about how they believe you, and you are not-scum, the person defending you is probably scum looking to gain some major towncred upon your lynch and saying "see, told you so".  This isn't defending in the typical sense I see it used, because they defending you in front of everyone else: they're just expressing their opinion that you are town and leaving it at that.  Furthermore, townies will tend to be skeptical of even strong logic behind actions (heaven knows I am: Supernatural 7, anyone?  I'm not sure how much of it I voiced, but I was constantly going through every possible doomsday scenario in my head of "what if this person is lying").  This is the one example I can think of for TTHTAT that actually has some backing to it.
*Passivity extends beyond just not taking a stand on lynches.  It is, rather, the person's entire demeanor.  It isn't the kind of "meh, whatevs" passivity either that I'm talking about.  I'm talking about the "walking on eggshells and trying not to break them" passivity.
*Lurking in the sense of malign lurking (not being active at critical junctures in the game) is incredibly hard to prove, though.  I've occasionally seen reason to call normally active players on their absence, but it's not really a provable thing and more of a gut feeling where the situation can just as easily be "I was at the doctor's".  Active lurking is a little more clear, but that really ties directly in to passivity: and active lurker is passive, and a passive player active-lurks.  Those should probably be tied into one scumtell: posting without content, avoiding contentious issues, and shying away from voting.
*I've never been clear on why the D1 no-lynch thing is in there to begin with.  I'm not sure what the math for lynch vs. no-lynch is, but the only person I've seen bring up no-lynching outside of MYLO/LYLO with any frequency is Varee, but that whole passage seems to be taking up space where it could simply say something like "The D1 lynch provides concrete information that can be combined with D1 discussion to provide a foundation for future scumhunting.  Without the D1 lynch, the town has only one death onto which to pin the D1 discussion, and the WIFOM of the mafia kill makes even that an unreliable benchmark."
*On D1 reads lists: enough of them already.  I know I used to be one of the big proponents of getting reads out as quickly as possible, to the point where I've ticked off some of the old guard in games, but I've come to my senses in this regard.  The only purpose D1 reads lists serve is as excellent WIFOM and rabbit-chasing fodder on later days.  D1 reads are MEANINGLESS, particularly in a BM.  Unless you get your reads in right at the end of the game-day, having seen the actions of everyone and analyzed them according to their meta, you have an INCREDIBLY small base of information to go off of when making them.  This leads to reads lists that are biased by one individual post rather than taking the individual as a whole and are in no way reflective of the person making them's views on later days.  However, it is very easy to see someone making an incorrect read or seemingly pushing what in hindsight was a mislynch and jump on that with all your might, which means that D1 reads DO serve as a means of confusion for the town, and an unnecessary one at that.  So no more, please, they aren't needed and they aren't beneficial.
*On D1 extensions: as above, DON'T DO IT.  Even if somebody hasn't posted anything and promise to within the coming day, the odds are incredibly high that they won't get anything in and a RL day will be wasted with almost no content added.  Here's the thing about D1: pretty much every lynch is effectively pseudo-random, barring weird bastard-mod setups where things go wacky (I'm looking at you, Smster w/ Love).  Without a lynch result or a broad base of player posts to go on, it is near impossible to make a truly informed decision in your lynch, and who gets lynched really comes down to who was the last person to make a "mistake".  I will caveat this entire statement with this: if there is a strong discussion AND the extension will not make the game go over the weekend, ONE 24 hour extension can be helpful for later days, but no more.

Persus13

  • Bay Watcher
  • 6th King of the Mafia
    • View Profile
Re: Mafia Theory
« Reply #138 on: April 19, 2015, 11:15:55 pm »

I've seen a lot of new players coming in with the idea that a no-lynch day one is best. Lynching D1 seems counter-intuitive and many other places start with a N0 or no lynch day 1. That whole passage is designed to head off new players who think no lynch is a good idea.

As for extensions D1, I think there should be a middle ground between your suggestion and as it currently stands. Days have a sort of flow to them, and if they get cut off too early it can be annoying. Generally, ending the day when the majority of people have seriously voted is best. Ending before that is not.
Logged
Congratulations Persus, now you are forced to have the same personal text for an entire year!
Longbowmen horsearcher doomstacks that suffer no attrition and can navigate all major rivers without ships.
Sigtext

4maskwolf

  • Bay Watcher
  • 4mask always angle, do figure theirs!
    • View Profile
Re: Mafia Theory
« Reply #139 on: April 20, 2015, 01:07:55 pm »

I've seen a lot of new players coming in with the idea that a no-lynch day one is best. Lynching D1 seems counter-intuitive and many other places start with a N0 or no lynch day 1. That whole passage is designed to head off new players who think no lynch is a good idea.
I'm actually more complaining at the length of the passage, it can be easily shortened without any loss of content since it mostly seems to be there as a vanity badge for a player who is no longer here.

Edit:
4mask fun fact of the day: I'm currently working on a breakdown of the current mafia setup using certain simulation rules to get a rough estimate of the expected win chance of the town.  Knowing me, I'll likely never finish it, but suffice it to say that if the town lynches scum D1 the odds are DRASTICALLY shifted in their favor.  Also, I'm finding that the preferential option in situations such as 5 town v. 1 scum is to no-lynch now rather than later: it gives the town a higher win chance to no-lynch as soon as the situation reaches an even number of players rather than if they lynch until MYLO and THEN no-lynch.

Also, if the game is down to two vanilla town + one jailkeeper + one mafia, and the mafia does not know who the jailkeeper is, the statistical best option is to lynch (3/8 vs. 1/3).  Just thought that was a fun tidbit.
« Last Edit: April 20, 2015, 01:18:46 pm by 4maskwolf »
Logged

Jack A T

  • Bay Watcher
  • Mafia is What Players Make of It
    • View Profile
Re: Mafia Theory
« Reply #140 on: August 28, 2015, 10:31:08 pm »

As my vacation calmed down, I decided to look into something I'd been considering looking into for some time: D1 lynch effectiveness.

It is common knowledge that D1 lynches are almost always against inexperienced townies.  We expect mislynches D1.  We've spoken for years of how we almost always mislynch D1, and the consistency of our town lynches has been a major motivator of years of efforts to change how we play D1 (and, in fact, in general).  Everyone knows most D1 lynches are against townies, and we handle D1 accordingly.

There's just one problem with the common belief: it's wrong.  In non-BMs, we usually lynch non-town D1.  In the past 4 board pages of non-BM games (about 1 and a half to 2 years of games), we have lynched some sort of non-town D1 56% of the time.  Even omitting BYOR 13, where a scum lynch was forced where a town lynch would likely have occurred, and my BYOR, where my failure to give public wincons led to the lynch, it's 52%.

See the results of my search here.  I used all non-BMs (we already know BMs actually do almost always start with a mislynch) in the past 4 pages that completed D1, aside from the King/Prince of the Mafia games, non-Mafia games, games with a non-lynch-based D1 voting system (Orbfalls and Worse Things), Mini Mafia Madness (sorry, not untangling that, and I can't even figure out what the alignments are), and the Guardians (a rare bastard mod where the mod was lynched).

The 'Experienced?' entry:  I looked at each lynched player and checked how many previous games they had.  Many, they got a yes.  Few or none, they got a no.  About a half dozen or so, somewhat.  My method is quite debatable.  Big takeaway here, though, is that those lynched D1 vary widely in experience, and are certainly not just newbies (several full-on veterans can be found, in fact).
« Last Edit: August 28, 2015, 10:33:10 pm by Jack A T »
Logged
Quote from: Pandarsenic, BYOR 6.3 deadchat
FUCK YOU JACK
Quote from: Urist Imiknorris, Witches' Coven 2 Elfchat
YOU TRAITOROUS SWINE.
Screw you, Jack.

notquitethere

  • Bay Watcher
  • PIRATE
    • View Profile
Re: Mafia Theory
« Reply #141 on: August 31, 2015, 04:58:34 am »

This is a welcome surprise and something that can be used to jolt some enthusiasm into D1. Good work!

One thing I've noticed is that 3rd parties are particularly vulnerable: their play can be mistaken for mafia and yet mafia are just as likely to vote for them.



A theory question: is there any disincentive for survivors to claim right away? Barring siding with specific parties at lylo, no one has an incentive to kill a survivor and yet the game is very hard for them if they're mistaken for either town or the mafia.
Logged

Persus13

  • Bay Watcher
  • 6th King of the Mafia
    • View Profile
Re: Mafia Theory
« Reply #142 on: August 31, 2015, 09:49:24 am »

A theory question: is there any disincentive for survivors to claim right away? Barring siding with specific parties at lylo, no one has an incentive to kill a survivor and yet the game is very hard for them if they're mistaken for either town or the mafia.
Yes, the general distrust placed on survivors due to the possibility they may be fakeclaiming or will side with the mafia. I've never seen mafia claim survivor though, but I have seen SKs claim it (and the one game where I was an SK I claimed survivor as well).
Logged
Congratulations Persus, now you are forced to have the same personal text for an entire year!
Longbowmen horsearcher doomstacks that suffer no attrition and can navigate all major rivers without ships.
Sigtext

notquitethere

  • Bay Watcher
  • PIRATE
    • View Profile
Re: Mafia Theory
« Reply #143 on: August 31, 2015, 09:55:38 am »

This will be something I'll investigate: do survivor claimants usually survive?
Logged

Mephansteras

  • Bay Watcher
  • Forger of Civilizations
    • View Profile
Re: Mafia Theory
« Reply #144 on: August 31, 2015, 11:08:41 am »

This will be something I'll investigate: do survivor claimants usually survive?

It'll be interesting to see what the numbers end up showing. From my observations, the answer is that it's pretty variable. Some tend to die because they say things that lead one faction or another to see them as a threat (Sometimes even knowing that they'll be killed for saying it). Others die because a faction decides they will be a threat and no better candidate for a lynch shows up. Others live through the game just fine.
Logged
Civilization Forge Mod v2.80: Adding in new races, equipment, animals, plants, metals, etc. Now with Alchemy and Libraries! Variety to spice up DF! (For DF 0.34.10)
Come play Mafia with us!
"Let us maintain our chill composure." - Toady One

notquitethere

  • Bay Watcher
  • PIRATE
    • View Profile
Re: Mafia Theory
« Reply #145 on: August 31, 2015, 11:13:12 am »

I think regular survivors (and even serial killers to some extent) are anti-game, as the players have an incentive not to engage because if they're too visible the mafia might nightkill them.
Logged

Jack A T

  • Bay Watcher
  • Mafia is What Players Make of It
    • View Profile
Re: Mafia Theory
« Reply #146 on: August 31, 2015, 11:15:36 am »

This is a welcome surprise and something that can be used to jolt some enthusiasm into D1. Good work!
Thanks.  The single largest problem I've seen D1, aside from the much-maligned RVS, is morale.  This might help with that.
One thing I've noticed is that 3rd parties are particularly vulnerable: their play can be mistaken for mafia and yet mafia are just as likely to vote for them.
I definitely noticed that.  Third parties also lack the protection from lynches that teammates can provide.  The local playstyle may also be oriented towards catching third parties, somehow.  Worth looking into more.

A theory question: is there any disincentive for survivors to claim right away? Barring siding with specific parties at lylo, no one has an incentive to kill a survivor and yet the game is very hard for them if they're mistaken for either town or the mafia.
The main problem I see with that is that a claimed survivor shoots up the vigkill priority list.  At best for them, they're the fallback target, a safe target to kill when a better option is not available.  At worst, they're a priority target, known non-town and seen as potential anti-town.  SKs pretending to be vigs are likely to shoot them as well, claimed survivors being incredibly easy to justify killing.  They are also potential lynch targets, but that is less likely.

If a survivor has strong defensive powers, though, and perhaps something to offer the town, claiming might be a good idea.
Logged
Quote from: Pandarsenic, BYOR 6.3 deadchat
FUCK YOU JACK
Quote from: Urist Imiknorris, Witches' Coven 2 Elfchat
YOU TRAITOROUS SWINE.
Screw you, Jack.

4maskwolf

  • Bay Watcher
  • 4mask always angle, do figure theirs!
    • View Profile
Re: Mafia Theory
« Reply #147 on: August 31, 2015, 10:08:54 pm »

I remember the one survivor I put in a game of mine... they had strong defensive powers and were given a fake claim cover role, and Teneb took the hint that someone was gunning for them and never claimed survivor.  Survived until the end of the game too, although they could have easily claimed because the person who wanted them dead was lynched Day One (although they didn't know it).

The person who did claim survivor in that game, origamiscienceguy, was a Miller/Death Miller Serial Killer.

Jack A T

  • Bay Watcher
  • Mafia is What Players Make of It
    • View Profile
Re: Mafia Theory
« Reply #148 on: May 10, 2016, 05:47:46 pm »

Now that BM LIX is over, I feel comfortable posting this.
We have, what, like 50 beginner games to use? Wanna take a guess how many first Day Townie lynches had scum on them?

Let's see how far we can get...
[snip]
And this is the point where I got bored doing research.

I could probably do this for every game and I guarantee we will find, at a MAXIMUM, 1 game where both scum voted for the first Lynch. And honestly, I highly doubt that even exists. It's about a 10% chance of existing, which considering that this is for all BMs, is a really good guess.
I finished the research.  Data, as usual, is on Google Drive.

*Over the course of 59 games, we have had 49 D1 town lynches, 15 of which were under the current setup.
*There is usually at least one scum on any D1 town lynch.  10/15 (66.7%) of new setup D1 town lynches, and 36/49 (73.5%) overall, had at least one scum on them.
*Two-scum D1 lynches are not common, but not unheard of.  7 D1 town lynches (14.3% overall) have held two scum (37, 34, 15, 13, 11, 4, 2).  It is worth noting that none of these games occurred under the current setup (BM 41+), and most are clustered in the relatively early BMs.  It is safe to say, I think, that there has been a turn away from this behaviour over time.
*I feel comfortable saying that large wagons (5, 6, 7) are more likely to have scum on them at all than smaller wagons (2, 3, 4).  Aside from 2-person lynches (a tiny sample), though, every size has had scum most of the time:
I am not comfortable saying more than that about presence of scum by wagon size.

Basically, there's probably scum on any BM D1 town lynch, but 2 scum on the BM D1 town lynch is rare, and has never happened under the current setup.
Logged
Quote from: Pandarsenic, BYOR 6.3 deadchat
FUCK YOU JACK
Quote from: Urist Imiknorris, Witches' Coven 2 Elfchat
YOU TRAITOROUS SWINE.
Screw you, Jack.

TolyK

  • Bay Watcher
  • Nowan Ilfideme
    • View Profile
Re: Mafia Theory
« Reply #149 on: May 11, 2016, 04:54:17 pm »

So, statistically, day-one lynches contain only 1 mafia member. Interesting.
(Going by BM's, ofc)
Logged
My Mafia Stats
just do whatevery tolyK and blame it as a bastard mod
Shakerag: Who are you personally suspicious of?
At this point?  TolyK.
Pages: 1 ... 8 9 [10] 11 12 ... 16