|
Zarepath I think you don't give town enough credit.
It's slightly understandable given that it is Newbie Mafia, but to say our only options are randomly lynching or no-lynching seems a bit too much in my opinion. I base this on the other newbie minis where a plan (such as pressuring inactives) lead to discussion and therefore more information upon which the town made better-than-random lynch cases
I don't like this defaulting to random lynch. It allows for WIFOM and general chaos in general as it is not built on something solid such as post analysis but , as its name suggests, something completely random. I don't think this is a pro - town idea.
Promoting a good town atmosphere will lead to educated guesses, so I say we wait until people had a chance to read and post, then we pressure inactives and go from there.
|
Poor Qatol, had such potential 
4/13 = 30.7% chance of killing mafia so random lynch seems good for me
|
Ehh... well its 9/13 = 69% chance of hitting town so random lynch seems like a bad idea to me
|
FakePromise, I feel as though saying that you are willing to take a 70% chance of killing an innocent man seems like you might have something to hide. Criminals tend to be fine with killing off innocent people, and you seem to fit that profile. Normal people would not be willing to take such a risk.
|
And we're off! I'm looking forward to an exciting and educating game. I'd like to begin Day 1 with earnest and open discussion, with high activity and clear lines of reasoning. My style of scumhunting is to use Bayesian inference on evidence derived from the character and reasoning of peoples' posts. The base rate of being mafia is 4/13.
Considering that I can confirm myself as innocent, the chance of anyone else being mafia, a priori, is 1/3. The probability distribution will change as people post and reveal their motivations, until I'm willing to lynch someone on even odds or better.
Regardless, in the interests of objectivity, it is important to keep in mind the base rate, the false positive rate, and the true positive rate. Any statement can be made by either an innocent or a mafia, and the prior rate is fairly low. Thus, it is important to isolate statements that exclusively mafia would make, or at least to find examples of reasoning that lean towards indicating mafia motivations. This way, we can try to prevent suffering from confirmation bias and scumhunt more effectively.
On the topic of zarepath's proposal, I find it alarming. Lynching randomly has a 1/3 chance of success, which is far too low for me to support. Your reasoning, especially, suffers from confirmation bias. In response to this plan, anyone can either a) support the lynch, b) argue against the lynch, and c) say nothing. Since saying nothing leads to modkills, we can eliminate the case.
Supporting the lynch, in your argument, is evidence of being mafia. However, you claim that arguing against the lynch is also evidence of being mafia. That is inconsistent, unless you would like to claim which case is more indicative of being mafia.
To continue, the "random" process used to select zelblade is also odd. Why not use either an entirely random process (RNG), or one that is motivated by initial evidence (on the basis of inactivity or level of contribution)?
In conclusion, I have to defer my support for zarepath's proposal, and I am also inclined to raise suspicions towards him. His plan has a low probability of success, and his procedure for deriving information from ensuing discussion is flawed with confirmation bias.
Could we instead opt to lynch on the basis of inactivity and low levels of contribution? The odds are no worse than a random selection, so long as everyone is participating properly, and this encourages the generation of more useful information than zarepath's choice of testimony would.
|
In extension, I feel as though the proposed plan by zarepath is simply too great of a risk. Why take nearly a 70% chance of taking an innocent person's life? No, we should wait to find the lurkers, I agree with slOosh, we need to wait for a little more discussion to happen in order to make a decision.
|
United States2186 Posts
Just letting you guys know that I might not be able to reply timely to everything. I'll do what I can. Nothing personal just at times very busy.
|
I disagree with zarepath's plan to "randomly" lynch me. Of course i am being a litte biased here, but how does randomly lynching me provide any sort of information? How does me/others defending myself = mafia? Any player, regardless of aglinment, is going to defend themselves either way, as getting lynched in either case is bad for their team, and chances to win.
Thus, i believe that lynching me or anyone else "randomly" is a horrible idea.
As for what to do with our day 1 lynch, i suggest that we lynch a lurker unless a better canditate comes up. Remember that we still have alot of time left before the deadline, so we should utilize this time well to discuss and find a better lynch candidate than lynching by reverse-alphabetical order.
|
On January 25 2012 14:11 Ver wrote:
Just letting you guys know that I might not be able to reply timely to everything. I'll do what I can. Nothing personal just at times very busy.
If Ver is busy, please PM Incognito, who has also agreed to help coach for this game.
As for what to ask coaches, you can ask them if you are arguing about the right kinds of things or going about things in the right kinds of ways. Please do not ask them to do an analysis for you or if they think so-and-so is mafia.
|
Zarepath's decision to lynch someone at random does sound like the calculated mind of a mobster. However, despite several conclusions that we all wish to draw, we need to wait, the mafia will all expose themselves at some point in time.
on a side note
Fakepromise agreed with him at 30% odds -_-
|
+ Show Spoiler +On January 25 2012 13:28 zarepath wrote:
So, the first question I think has to be this:
What do we want to accomplish on Day 1?
We either no-lynch or randomly lynch, as far as I can tell. We don't have much game history to draw upon as far as deciding who to cut early.
I see several possible outcomes:
Randomly lynch a mafia/red: This is clearly the best possible outcome. If we have worthwhile hunches to go on, it's possible we can lynch a mafia on the first Day. From what I've read it's unlikely, but still possible; it all depends on what kind of discussion we see today.
Randomly lynch a townie/blue: Bad. The only upside is that we can then use the information from that lynching to determine likely mafia. Who was most excited about randomly lynching that particular person? Who bandwagonned on? Who were the last few votes who made the lynch possible?
No-lynch: We don't get a mafia, or confirmation of who isn't mafia. Because there's no risk of lynching a townie/blue, this seems to be very safe.
But what I wonder is this: if we don't lynch anyone today, our information for Day 2 is limited, and we give the mafia a night to kill someone. We end up with 1 dead townie before we're willing to make deductions. Maybe it's riskier to not lynch someone on the first night, because we'd have less information to go on.
So what I'm suggesting is this: we lynch someone randomly. Say, by reverse-alphabetical order. There's no reason at all for us to suspect this person. But by deciding to lynch them, we'll either see a lot of defense (ie, they're mafia), or a lot of people agreeing with killing that person (ie, those people are mafia.) Once that person flips blue or red, we'll know who was who.
TLDR; I suggest that we randomly lynch zelblade, or seriously discuss it, and see what happens from there.
Gentlemen,
First post and we are already talking about Random Lynches. I am definitely not a fan of it. Part of lynches, even if it's a miss-lynch, is that it still has the possibility of giving you information. Every little detail can give you information. I will not endorse a random lynch, especially so early in the game.
+ Show Spoiler +On January 25 2012 14:05 MidnightGladius wrote:
And we're off! I'm looking forward to an exciting and educating game. I'd like to begin Day 1 with earnest and open discussion, with high activity and clear lines of reasoning. My style of scumhunting is to use Bayesian inference on evidence derived from the character and reasoning of peoples' posts. The base rate of being mafia is 4/13.
Considering that I can confirm myself as innocent, the chance of anyone else being mafia, a priori, is 1/3. The probability distribution will change as people post and reveal their motivations, until I'm willing to lynch someone on even odds or better.
Regardless, in the interests of objectivity, it is important to keep in mind the base rate, the false positive rate, and the true positive rate. Any statement can be made by either an innocent or a mafia, and the prior rate is fairly low. Thus, it is important to isolate statements that exclusively mafia would make, or at least to find examples of reasoning that lean towards indicating mafia motivations. This way, we can try to prevent suffering from confirmation bias and scumhunt more effectively.
On the topic of zarepath's proposal, I find it alarming. Lynching randomly has a 1/3 chance of success, which is far too low for me to support. Your reasoning, especially, suffers from confirmation bias. In response to this plan, anyone can either a) support the lynch, b) argue against the lynch, and c) say nothing. Since saying nothing leads to modkills, we can eliminate the case.
Supporting the lynch, in your argument, is evidence of being mafia. However, you claim that arguing against the lynch is also evidence of being mafia. That is inconsistent, unless you would like to claim which case is more indicative of being mafia.
To continue, the "random" process used to select zelblade is also odd. Why not use either an entirely random process (RNG), or one that is motivated by initial evidence (on the basis of inactivity or level of contribution)?
In conclusion, I have to defer my support for zarepath's proposal, and I am also inclined to raise suspicions towards him. His plan has a low probability of success, and his procedure for deriving information from ensuing discussion is flawed with confirmation bias.
Could we instead opt to lynch on the basis of inactivity and low levels of contribution? The odds are no worse than a random selection, so long as everyone is participating properly, and this encourages the generation of more useful information than zarepath's choice of testimony would.
I'm not quite sure what you are trying to say, because it's really not clear in anyway. You're talking about Bayesian inference, and yet, i'm quite sure a small majority of this game would even know what that is, let alone understand how it works. You don't even bother to explain it - you just continue on describing the in's and out's how the Theory works. But at the end of it all, it comes down to the "importance of isolating statements mafia would make," which is what happens in pretty much every mafia game.
So instead of having you fall into you're own confirmation bias, are you going to let go of you're statistics theory and play the game? If you want to scum hunt more effectively, be my guest, but don't take two paragraphs describing a system of mathematical reasoning that the majority of this game probably don't understand. It amounts to nothing.
You want you're reads, you're analysis, and everything you have, to be 100% clear, transparent, and easy to read for everyone to see in the thread. Trying to confuse people Day 1 and fog up everyone's view about how you are going to hunt scum isn't the way of going about it.
With all that being said, on to my personal ideas about lynch targets this early. I don't think a random lynch, as I stated before, to be anything good. Information can be gleaned even from a miss-lynch. As for lynching lurkers/inactive players, it is also not effective in aiding us. What do we gain from killing a lurker? I see little gain coming from it, except with the off-chance that they flip red. I'd much rather use logical reasoning to determine who would should hang Day 1, and there is still plenty of time to do just that.
|
I'm sorry if my first post came across as unhelpful, but I want to establish first principles before getting down to the nitty-gritty. I'll try to be more clear and direct.
In simpler terms, players who suggest courses of action that hurt the town's chances are suspicious, as innocents should never be making these kinds of proposals unless they have much more information than they're letting on. As it's Day 1, this is clearly impossible, so I look askance at zarepath and FakePromise, who both advocate a plan with very low expected value.
The other part of my methods take a bit longer to develop, as I need to see more posts before picking up any trends. Scummy behavior is such that it betrays access to hidden information, and then does not adjust accordingly as information is made public to the town. Players who behave in this way are either not updating their beliefs properly (tunnel-vision, confirmation bias, or ignorance are common causes), or updating them according to hidden information (perhaps a blue investigative role, but overwhelmingly likely to be mafia).
Does that explain my position more clearly?
|
Regarding the setup, 4 scum to 9 town seems like a lot of scum to me. This would lead me to believe that scum KP is probably 1, as anything else will probably be excessive. As such, we probably have only 2 mislynches before LYLO, unless there is a medic prot of some sort of course. Thus we need to make sure that we use these lynches well, and use logical reasoning to pin down the lynch onto the scum.
To town, we need to post more, as more posts = more contributions, and would allow us to make analysis and thus help to pin down who the scum are.
|
With regards to your lynch proposal, I don't agree with not pressuring lurking/inactive players with the threat of a first-day lynch. Truly non-participating players can be excluded from analysis due to modkilling, and it stands to reason that more posting of positions will increase the sample size of data.
What logic would you claim to use on the first day? There is insufficient data to make any valid attempts at deduction, induction, or hypothesis. With what is available to you now, would you take even odds or better on any accusation?
I see the point of the first day as creating the pool of data required for later analysis, and the simplest incentive for players to do that is to lynch them if they don't.
|
There is no reason our first lynch should be random. We need to use the threat of lynching to put pressure on lurkers and suspicious people. This needs to start today. If we seriously consider a no lynch today all we do is encourage passive play. We should be creating opportunities to gain information and put pressure. The worst thing we can give the mafia is time.
|
EBWOP: Ninja'd, the "you" in the above post still refers to Bromancipate.
|
On January 25 2012 15:21 zelblade wrote:
Regarding the setup, 4 scum to 9 town seems like a lot of scum to me. This would lead me to believe that scum KP is probably 1, as anything else will probably be excessive. As such, we probably have only 2 mislynches before LYLO, unless there is a medic prot of some sort of course. Thus we need to make sure that we use these lynches well, and use logical reasoning to pin down the lynch onto the scum.
To town, we need to post more, as more posts = more contributions, and would allow us to make analysis and thus help to pin down who the scum are.
Ok this post is really weird.
1) OP clearly states Mafia KP.
On January 24 2012 08:35 dreamflower wrote:
Mafia Goon
Your goal is to eliminate everyone else in the town. Your ability, as a group, is killing off whomever you decide on at night and knowing the role of each other player in your mafia. You may kill your own members. Mafia killing power is always 1 until there are no mafia remaining.
Of course this is a newbie game so y'know, people make mistakes.
2) This post doesn't actually give us anything new or that helpful. We can do the math and figure how many mislynches we have till Lylo, which isn't information you typically use day 1. I mean, unless you push for no lynch, how's that info help?
3) This is the part that got me from giving benefit of the doubt to feeling weird:
On January 25 2012 15:21 zelblade wrote:
To town, we need to post more, as more posts = more contributions, and would allow us to make analysis and thus help to pin down who the scum are.
What the heck? Who does this? Who addresses town? A town post would have started at "we need to post more ...".
#FOS zelblade
|
On January 25 2012 15:40 slOosh wrote:Show nested quote +On January 25 2012 15:21 zelblade wrote:
Regarding the setup, 4 scum to 9 town seems like a lot of scum to me. This would lead me to believe that scum KP is probably 1, as anything else will probably be excessive. As such, we probably have only 2 mislynches before LYLO, unless there is a medic prot of some sort of course. Thus we need to make sure that we use these lynches well, and use logical reasoning to pin down the lynch onto the scum.
To town, we need to post more, as more posts = more contributions, and would allow us to make analysis and thus help to pin down who the scum are.
Ok this post is really weird. 1) OP clearly states Mafia KP. Show nested quote +On January 24 2012 08:35 dreamflower wrote:
Mafia Goon
Your goal is to eliminate everyone else in the town. Your ability, as a group, is killing off whomever you decide on at night and knowing the role of each other player in your mafia. You may kill your own members. Mafia killing power is always 1 until there are no mafia remaining.
Of course this is a newbie game so y'know, people make mistakes. 2) This post doesn't actually give us anything new or that helpful. We can do the math and figure how many mislynches we have till Lylo, which isn't information you typically use day 1. I mean, unless you push for no lynch, how's that info help? 3) This is the part that got me from giving benefit of the doubt to feeling weird: Show nested quote +On January 25 2012 15:21 zelblade wrote:
To town, we need to post more, as more posts = more contributions, and would allow us to make analysis and thus help to pin down who the scum are.
What the heck? Who does this? Who addresses town? A town post would have started at "we need to post more ...". #FOS zelblade
1) Opps my bad, didnt see that part.
2) I was simply trying to generate some discussion as it would seem that no one was talking much, and trying to emphasise that we need to use our lynches carefully.
3) I am sorry if that came off at scummy, and what i simply meant by that part was that town needs to post more.
|
Ok, my take on this:
First of, "randomly" lynching in the way zarepath proposed sounds strange. First of all, it is not truly random if we discuss the person we are lynching together with the random lynch. This would actually give mafia a very, very easy day 1, which is not something we want. They either just stay middle profile and lynch the person at hand with low danger to themselves, because they would only have to support the system of a random lynch, and not the actual lynch of a person. And if the "random" targeted person is one of theirs, they just have to state that they are against random lynching in general.
If we would want to lynch randomly, we should first decide that we do that, and then somehow randomize whom we vote for afterwards. However, since we don't have any reliable way of randomizing as a group, it would always ultimately be up to one single person, so when we say we lynch randomly, we basically give the decision of whom we lynch to one single man, who might or might not randomize. That does not really sound extraordinarely smart in my opinion, especially since we even give him a very good excuse should he lynch an innocent. So instead of having to guess one mafia, we now have to guess one townie only to even get a random lynch.
Next, we have the possibility of not lynching. While this gives additional information for our first lynch, it also gives the mafia a free kill. I don't really like doing that.
Then we have a policy lynch on a hard lurker. I think this is actually not a really good idea. All the mafia needs to do to avoid this is post a few more or less interesting posts to not be hardlurker, and they get a free kill with nearly no information for us out of it. They don't even need to bandwagon or use their combined influence.
So, we are left with talking, and finding a good target for a lynch. This is generally a good idea, but we have to keep in mind that mafia is probably trying to influence our aggression onto one of ours (which also gives us more information lateron)
And despite the general strangeness (scumminess?) of zelblades post, he still has a point in that bar extraordinary circumstances (which i think are actually not that unlikely, the mafia density leads me to believe that we probably have some blues, as others have pointed out before, too), we only have 2 mislynches. So if we lynch, we should lynch in a way that gives us maximum information, so no random or policy lynch.
At the moment, i would say that zelblade looks pretty strange, as sloosh has pointed out. Of course he has to defend himself, but afterwards he first argues for a lurker lynch, and then in his next post points out that we need to use our mislynches very carefully, since we only got a very limited amount of those. A policy lynch is not a lynch that gives a lot of information, so those both posts contradict each other very much.
And the "everyone needs to post more" part looks like trying to post more without actually saying much in my opinion.
However, i would also like to hear more from the people who have not yet posted anything.
|
On January 25 2012 14:19 zelblade wrote:
I disagree with zarepath's plan to "randomly" lynch me. Of course i am being a litte biased here, but how does randomly lynching me provide any sort of information? How does me/others defending myself = mafia? Any player, regardless of aglinment, is going to defend themselves either way, as getting lynched in either case is bad for their team, and chances to win.
Thus, i believe that lynching me or anyone else "randomly" is a horrible idea.
As for what to do with our day 1 lynch, i suggest that we lynch a lurker unless a better canditate comes up. Remember that we still have alot of time left before the deadline, so we should utilize this time well to discuss and find a better lynch candidate than lynching by reverse-alphabetical order.
The point is that upon flip, we have much more to go on. Anyone irrationally defending you now, if it turns out you're mafia, has a solid chance of also being mafia. So no, not anyone defending you is necessarily mafia, and not necessarily anyone accusing you, but upon flip, we can figure out which side of that we can throw our suspicions.
After reading others' responses to the idea: no, a RNG lynch would NOT be as useful to us, because we don't learn anything from it (as has been pointed out). By me randomly accusing someone, that person is forced to speak, and others are forced to defend if mafia/bandwagon if he's not.
I argue that the process has already begun, based on the content of some of these posts.
I think the strongest argument against doing this is that, because of the high mafia/town ratio, we only have 2 miss-lynches before it's LYLO, as zelblade pointed out.
On January 25 2012 15:24 DoYouHas wrote:
There is no reason our first lynch should be random. We need to use the threat of lynching to put pressure on lurkers and suspicious people. This needs to start today. If we seriously consider a no lynch today all we do is encourage passive play. We should be creating opportunities to gain information and put pressure. The worst thing we can give the mafia is time.
Lynching lurkers is not as great a strategy as it sounds. As others pointed out, mafia can just decide to post more, and then suddenly we're looking at lynching the less-active townies just because they don't talk enough. People who don't post at all get modkilled anyway, and seeing as how this is a newbie game, there are probably several lurkers who just don't know where to get started.
I also agree that we should go after suspicious people. And I think we'll find out who they are by continuing to pressure zelblade.
|
|
|
|
|
|
EPT
