xkcd

by **webby** » Tue Mar 29, 2011 12:00 am UTC

And by same result, I mean repeated results (ie guilty on each night or innocent on each night).

by **webby** » Tue Mar 29, 2011 12:26 am UTC

One more thing - I think we have it certain that at least one out of more_people, mpolo and tasteslikecoke are scum. Very tempting to lynch one of more_people and mpolo today, find out all the information we can about the other in the night, then probably lynch them tomorrow if the first wasn't scum. I think the chances of us getting down to 5-2 or 4-2 with that strategy is at least greater than 95% + whatever information we get during the night, but there's that 5% chance of it being wrong and we lose.

Useful for deciding whether to pursue that:
1. What are our chances of winning if we just go four nights without lynching and get down to 4-3?
2. What are our chances of winning from 4-2 or 5-2?
3. What are the chances of tasteslikecoke being scum and us not getting enough information to reveal mpolo/more_people as town?
4. Have either mpolo or more_people seemed particularly scummy?

by **mpolo** » Tue Mar 29, 2011 6:18 am UTC

If we let it get to 4-3, then every lynch has to be good from then on. (Mislynch would be 2-3 the next day, so scum control the lynch. After one successful, we'd have 3-2, which is equivalent, then 2-1 in endgame.)

If we start lynching at 5-3, then a mislynch puts us at 3-3 (lynch 4-3, nk 3-3), which is a scum win, I believe (contrary to what I thought a couple of pages back when BigNose suggested this). Because two successful lynches brings us to 1-1, where we cannot lynch successfully for lack of a majority, and the last scum wins at night.

4-2 is equivalent to 5-3. (If we lynch successfully, we have 2-2 and lose.)

5-2 is very good for us: a mislynch puts us at 3-2, giving us a second chance.

Obviously, you have no real reason to believe me, but since I know my alignment, I also know that tastelikecoke is a sane cop and that more_people is scum. Which makes it pretty much impossible for me to comment on question 3 or 4.

by **John Citizen** » Tue Mar 29, 2011 7:15 am UTC

mpolo wrote:Obviously, you have no real reason to believe me, but since I know my alignment, I also know that tastelikecoke is a sane cop and that more_people is scum. Which makes it pretty much impossible for me to comment on question 3 or 4.

Unfortunately, that logic is flawed. It is impossible to eliminate the possibility of TLC being scum pretending to be sane, thus throwing m_p's alignment into doubt.

It is impossible for TLC, m_p and Mpolo to all be town.

I've also noticed that again, nobody was roleblocked.

by **John Citizen** » Tue Mar 29, 2011 7:30 am UTC

EBWOP: I would suggest -1 or -3 copping to establish more groups that cannot all be town.

by **webby** » Tue Mar 29, 2011 7:44 am UTC

The problem is that it's not enough to find groups that can't all be town, we need to find an individual who can't be town. This is the last night we can lynch and still get a second chance.

The question is whether we're going to be able to narrow anyone down completely as being scum in two nights, or if the assumption that tasteslikecoke is town is as close to a certainty we're ever going to get.

This is especially true if we're unlucky enough to have already lost two sane/insane cops - then we're stuck with only two people with useful information, as opposed to scum who have three people trying to spread misinformation. I'm not sure how close we are to identifying definite scum, because we have a lot of guilty/guilty and innocent/innocent results.

My personal view is that we should take the chance and lynch either more_people or mpolo today, then everyone gather information on the other (except the people who have already copped them, who should maybe cop tasteslikecoke?)

Would like to hear thoughts from everyone on this idea, but particularly more_people and tasteslikecoke.

by **more_people** » Tue Mar 29, 2011 9:04 am UTC

Yeah, I'm cool with everything that has been said expect for the part where mpolo claimed he was town in two sentences worth of words.
Claiming to be town doesn't help town because it makes people be more suspicious of you. It can cause people to mislynch you as a result unless you are scum.
I'm going to vote for you for the reason stated above because you should already know this.
Vote: mpolo

by **mpolo** » Tue Mar 29, 2011 9:53 am UTC

John Citizen wrote:
mpolo wrote:Obviously, you have no real reason to believe me, but since I know my alignment, I also know that tastelikecoke is a sane cop and that more_people is scum. Which makes it pretty much impossible for me to comment on question 3 or 4.

Unfortunately, that logic is flawed. It is impossible to eliminate the possibility of TLC being scum pretending to be sane, thus throwing m_p's alignment into doubt.

It is impossible for TLC, m_p and Mpolo to all be town.

I've also noticed that again, nobody was roleblocked.

Ah. You are correct here. Since we left off the possibility of "scum" in the table, it didn't click for me. I know that TLC isn't insane, but I don't know if he might be scum.

More_people made some kind of slip up in his last version of the table. He has me copped as Scum by TLC, but has it correct in his footnote.

The lack of roleblock is quite interesting, and I'm not sure why the scum are letting everything get through (I'm taking for granted that no town cop would just make up a result rather than saying that he is roleblocked.)

As has been said before, this is the last day that a lynch can be safely made, but if we take that option, we have one less day of results before LyLO. And now realizing that it is not just between more_people and myself, it seems to be an unacceptable risk to lynch one of the three in the hopes of exposing the scum among the others.

Lynching TLC and having him come up town would mean -- we know that either myself or more_people is scum. That is, no better than before -- actually worse, as we will have lost a sane cop (or insane if you believe more_people over me -- both are too important to sacrifice).
Lynching TLC and having him come up scum would be great. We would know nothing more about myself and more_people, but we couldn't complain about this result at all.

Lynching me means losing a townie. When we see that I was town, we confirm TLC as either sneaky scum or an insane cop. More_people's alignment remains uncertain.

Lynching more_people: if he turns up town, I would know that TLC is scum, but you all would have no reason to believe this. (Barring results from other investigations)
If he turns up scum, we rejoice, and TLC is either very sneaky scum or a sane cop.

I could theoretically risk lynching more_people, as I have a guaranteed scum hit the following day. More_people (if he is town) can theoretically risk lynching me, as he would then know that TLC was scum. However, the rest of the town would have no clear way to know which of TLC and more_people is scum.

I think that the only way forward is a No lynch vote. (The possibility of losing a sane cop is too great, in my opinion, to risk lynching TLC without other evidence of scumminess.)

If somebody wants to do the hypotheticals for if I were scum, go ahead. I am getting headaches from the layers of subjunctive involved in that.

by **more_people** » Tue Mar 29, 2011 10:20 am UTC

If tlc isn't mafia then they are dead anyway because if they are town then they are either sane or insane.

by **Gopher of Pern** » Tue Mar 29, 2011 10:39 am UTC

True, TLC is dead either tonight, or will be highly suspected for tomorrows lynch. As the only confirmed non-paranoid, non-naive cop, they are a high priority target for the scum. I also doubt scum will intentionally let their cop results play that way, as it does expose them.

At this stage i'm leaning closer to Mpolo as scum than more_people. The first kill was against VZ, who would have copped mpolo. As they knew who was copping who, they likely would have chosen someone who would be copping them. The less cop results on the scum, the better for the scum.

by **webby** » Tue Mar 29, 2011 11:07 am UTC

Mpolo, we have at least a 2/3 chance of getting one scum if we lynch up to two of the three of you. I think it's actually much better than that because the analysis of the results that we have which I posted earlier suggest a low probability for TLC being scum. Still waiting for someone to do an independent analysis and come up with probabilities.

I don't think we can afford a no-lynch tonight. If TLC dies, we're already left with somewhere between 1 and 3 useful cops against 3 scum.

Gopher of Pern wrote:True, TLC is dead either tonight, or will be highly suspected for tomorrows lynch. As the only confirmed non-paranoid, non-naive cop, they are a high priority target for the scum. I also doubt scum will intentionally let their cop results play that way, as it does expose them.

Actually, if we lynch more_people or Mpolo today, and they're town, then TLC can't be killed during the night without revealing that the other is scum.

Gopher of Pern wrote:At this stage i'm leaning closer to Mpolo as scum than more_people. The first kill was against VZ, who would have copped mpolo. As they knew who was copping who, they likely would have chosen someone who would be copping them. The less cop results on the scum, the better for the scum.

I was actually leaning towards more_people over Mpolo for a similar reason - that more_people has copped the dead people (known alignment) which are the most useful cop results. Also, I think his posts seem slightly more scummy, especially his quick jumping on the bandwagon of my idea without any real discussion beyond a reasonably weak attack on Mpolo. I'm still to be convinced either way, however:

FOS: more_people

I think we also need to come up with a suitable copping strategy for tonight before voting, haven't thought of one yet - any ideas?

by **tastelikecoke** » Tue Mar 29, 2011 12:14 pm UTC

So wait, I'm either sane or insane then.
So let's say I'm SANE

Spoiler:

if I'm INSANE

Spoiler:

The branches are balanced, so is the possibility that mpolo or m_p is scum are balanced, logically. Of course I assumed that I'm not scum.
Now everything's based on tells. I'm pretty sure I'll put a vote to either mpolo or m_p...

by **BigNose** » Tue Mar 29, 2011 4:26 pm UTC

OK, I have tried to sort out the logic as we have it.
My first step was to try to 'prove' TLC as either Town or Scum.

If TLC is TOWN then
..... as Ibarra copped TLC as Scum, then Ibarra is Insane
.............. If Ibarra is Insane, then Ibarra copped Mpolo as Scum, ergo Mpolo is Town
.............. Ibarra could be Paranoid
..... as GoP copped TLC as Scum, then GoP is Insane
.............. If GoP is Insane, then GoP copped Webby as Scum, ergo Webby is Town
.............. Ibarra could be Paranoid

If TLC is SCUM then
..... If Ibarra is Sane, the Ibarra copped Mpolo as Scum, ergo Mpolo is Scum
.............. Ibarra could be Paranoid
.............. If GoP is Sane, then GoP copped Webby as Scum, ergo Webby is Scum
.............. Ibarra could be Paranoid

If TLC is Scum, then Mpolo & Webby could be Scum.
It's a game-wrapper, but also very very surprising if it is true.

So, not 'proof', but a good indication that TLC is Town.
Also, would Scum make a 'deliberate' mistake?

Please, please check this over, as I started to compile a follow-on of TLC is Town, Sane or Insane, but found a mistake, hence no follow-on atm.

mpolo wrote:More_people made some kind of slip up in his last version of the table. He has me copped as Scum by TLC, but has it correct in his footnote.

Hmmmm. This isn't the first time that I have been unhappy with Mpolo. It was Ibarra that put you in as Scum from their copping, not a mis-quote from M_P.

mpolo wrote:Hmmm... In one of the strategy guides to normal dethy, they had people randomly choose their investigation target. Then we get some people with two or more results already on the first day, which could also help with analyzing. But someone might slip through uninvestigated as well.

This was the first one and I let it slip-by. We will win this game based on logic and while I can understand the need for 'Random', the potential for 'holes' could be great. As a potential option in N4 that would be understandable, we would have a lot of other information available and it would screw-up any Scum strategy, but as a N1 strategy . . .

If we are to go to a lynch, then my vote would be on Mpolo, however, I still think that we should consider a NL, as we still have a lot to learn.
As you will see from my follow-on (when I get it right), there are still LOTS of holes.

by **mpolo** » Tue Mar 29, 2011 6:16 pm UTC

BigNose wrote:
mpolo wrote:More_people made some kind of slip up in his last version of the table. He has me copped as Scum by TLC, but has it correct in his footnote.
Hmmmm. This isn't the first time that I have been unhappy with Mpolo. It was Ibarra that put you in as Scum from their copping, not a mis-quote from M_P.

mpolo wrote:Hmmm... In one of the strategy guides to normal dethy, they had people randomly choose their investigation target. Then we get some people with two or more results already on the first day, which could also help with analyzing. But someone might slip through uninvestigated as well.
This was the first one and I let it slip-by. We will win this game based on logic and while I can understand the need for 'Random', the potential for 'holes' could be great. As a potential option in N4 that would be understandable, we would have a lot of other information available and it would screw-up any Scum strategy, but as a N1 strategy . . .

If we are to go to a lynch, then my vote would be on Mpolo, however, I still think that we should consider a NL, as we still have a lot to learn.
As you will see from my follow-on (when I get it right), there are still LOTS of holes.

I misread the table (and neglected to go back and hunt down the origin of the error). I had forgotten that TLC's mistake meant that there were two results on me for last night.

The first quote that you take umbrage at was simply citing a strategy I found by Googling "Dethy strategy" and was presented as a possibility to consider, not as a recommendation. [url="http://mafiascum.net/wiki/index.php?title=Dethy/Analysis"]Here is the link.[/url]

by **mpolo** » Tue Mar 29, 2011 6:17 pm UTC

Failure to use the url tag correctly. How amusing. I will assume you can copy/paste if you want to verify it.

by **BigNose** » Tue Mar 29, 2011 9:28 pm UTC

The first point about the Random copping, I can sort of understand, it's just at the time it seemed out of place as we were well into discussing a linear strategy, plus you didn't seem to want to follow your strategy 'option' up.

But the attempt to find 'something' on M_P, who is your current direct adversary (lynch of M_P or Mpolo) is either stupid Town, or stupid Scum.

It's not like you had to look back very far (same page). I even put in a 'corrected' version with SCUM*.

My problem is, IF we are to lynch today and we have choices of M_P or Mpolo, then my vote is on you.

I'm not even sure that we need a lynch today.

ALL, for this night, one option is to go minus 2. That should avoid most people copping someone who has already copped them. Unfortunately, almost any strategy (except +2 & +4) is likely to lead to this 'cross-copping' very soon.

by **webby** » Tue Mar 29, 2011 10:37 pm UTC

BigNose wrote:If we are to go to a lynch, then my vote would be on Mpolo, however, I still think that we should consider a NL, as we still have a lot to learn.
As you will see from my follow-on (when I get it right), there are still LOTS of holes.

Yes there are a lot of holes, but what information are we going to gain on the last two nights? Assume TLC dies in the night. We then have somewhere between 1 and 3 sane/insane cops. It's very likely that at least 1 and possibly 2 are amongst the dead people, given the number of guilty/guilty and innocent/innocent results last night. That means we're left with 1 or 2 sane/insane cops, which gives us a maximum of three useful cop results, even assuming we can distinguish the legitimate ones from scum ones.

Overall, I'm not convinced that we'll get enough information to narrow it down for sure by night 4. No-lynch is a very risky option - clearly we should only take it if we're convinced the chances of finding scum on day 5 by the no-lynch strategy are greater than 2 in 3 + whatever credence you give to the ideas about tastelikecoke being town. I can envisage scenarios where we're going to be left with lots of 50-50 chances, but no definites.

by **Krong** » Wed Mar 30, 2011 4:35 am UTC

I've been trying to see the solid logic behind this discussion, but I just can't right now. There are a couple of guarantees, but it's mostly a lot of "probably"s, which are generally unreliable when scum can manipulate 3 of the results and who dies each night. This whole thing smells fishy to me right now... let me count the ways:

1. Discussion is focused on mpolo, more_people, and tlc, as I believe it's been proven at least 1 of them is scum. OK, why these 3? I'd be surprised if there aren't other triads out there that we can make the same argument about.

2. mpolo stating that he knows more_people is scum because he knows tlc is sane.
(a) This is faulty logic as stated before, as tlc could easily be scum.
(b) Anytime a player knows with logical certainty another is scum, they must vote for that player. There is no reason to withhold a vote on scum once you're certain they're scum, even if the town is politely calling for a nolynch. mpolo didn't cast that vote.

(Of course, I'm kind of having it both ways here, saying he should have voted when it would have been on faulty logic. But he either he didn't know that logic was faulty at the time and should have voted, or he did know and shouldn't have stated it as truth).

3. more_people missing the fact that the logic of mpolo's assertion was flawed.

4. webby treating tlc as if he's the most likely of the bunch to be town, when there simply isn't proof for it. (He is most likely to be useful if he is town, but that's not the same thing).

5. tlc copping mpolo last night out of order. We're assuming all along that this is a townie mistake, when it could be a great way for scum to further manipulate the results.

6. webby's been pushing hard for a lynch on a specific person today, when we don't have logical proof of that person's scumminess nor a real argument made based on posting.

7. I'm belatedly realizing that webby's been leading most of the discussion on seemingly arbitrary matters -- "should we post in order?", "what order do we post in?", etc. This was probably a mistake by the rest of us, as if webby is scum we can bet the choices he made were not arbitrary at all.

As webby said earlier, we've wound up in a situation where scum has really hurt us in terms of how much useful information we have. I'm pretty worried about both of our options -- nolynch twice and risk those nights' results being worthless, or lynch today and risk being and LYLO tomorrow. I'd prefer the second, I think, but at the same time I feel like this discussion is being pushed too hard by people I don't really trust.

Hmm... got to dig into these results a bit more.

by **webby** » Wed Mar 30, 2011 5:18 am UTC

Krong wrote:1. Discussion is focused on mpolo, more_people, and tlc, as I believe it's been proven at least 1 of them is scum. OK, why these 3? I'd be surprised if there aren't other triads out there that we can make the same argument about.

I can't find any other triads, that one is an obvious one that comes from the fact that tlc is the only one with useful cop information. If you can find another one, I'd be happy to use the same logic with that triad as with this one.

4. webby treating tlc as if he's the most likely of the bunch to be anthill, when there simply isn't proof for it. (He is most likely to be useful if he is anthill, but that's not the same thing).

My argument (and I've called for other people to do the calculation) is that he's the least likely to be scum because if he was scum, that puts the non-scum, non-naive/paranoid cop results that are guilty/guilty or innocent/innocent at either 2-0, 3-0 or 4-0, when we'd expect 50-50. More likely is that the true ratio is either 1-1, 2-1 or 3-1. Exactly how much more likely is up to someone else to calculate.

5. tlc copping mpolo last night out of order. We're assuming all along that this is a townie mistake, when it could be a great way for scum to further manipulate the results.

I can't see that that would be helpful unless mpolo is scum and wanted to avoid being copped.

6. webby's been pushing hard for a lynch on a specific person today, when we don't have logical proof of that person's scumminess nor a imaginary argument made based on posting.

I'd be happy with a more_people or mpolo lynch.

7. I'm belatedly realizing that webby's been leading most of the discussion on seemingly arbitrary matters -- "could we post in order?", "what order do we post in?", etc. This was probably a mistake by the rest of us, as if webby is scum we can bet the choices he made were not arbitrary at all.

I know this isn't going to be helpful to you right now, but after I've played a few games you'll find that this is how I generally play - as either town or scum, I like to lead the discussion. I also seem to be online more frequently than some of you, possibly because I tend to be at my computer all uniday and I leave a tab open.

As webby said earlier, we've wound up in a situation where scum has really hurt us in terms of how much useful information we have. I'm pretty worried about both of our options -- nolynch twice and risk those nights' results being worthless, or lynch today and risk being and LYLO tomorrow. I'd prefer the second, I think, but at the same time I feel like this discussion is being pushed too hard by people I don't really trust.

Ok, I can see why you wouldn't trust me, so I'll go along with lynching any one of a triad which contains at least one scum. I'm not, however, going to support a no-lynch today.

by **mpolo** » Wed Mar 30, 2011 6:25 am UTC

Krong wrote:2. mpolo stating that he knows more_people is scum because he knows tlc is sane.
(a) This is faulty logic as stated before, as tlc could easily be scum.
(b) Anytime a player knows with logical certainty another is scum, they must vote for that player. There is no reason to withhold a vote on scum once you're certain they're scum, even if the anthill is politely calling for a nolynch. mpolo didn't cast that vote.

(Of course, I'm kind of having it both ways here, saying he could have voted when it would have been on faulty logic. But he either he didn't know that logic was faulty at the time and could have voted, or he did know and shouldn't have stated it as truth).

As it is my first time playing dethy, I just presented what I was thinking. I also knew that there was no way anybody was going to be following my vote, as my analysis depended entirely on knowing my own alignment.

by **BigNose** » Wed Mar 30, 2011 9:44 am UTC

Next stage, is TLC Sane or Insane?

If TLC is Sane then
...... TLC copped M_P as Scum, ergo M_P is Scum
...... TLC copped Mpolo as Town, ergo Mpolo is Town
............ Mpolo is Sane, Insane or Paranoid
............ If Mpolo is Sane then
.................. Mpolo copped BigNose as Scum, ergo BN is Scum
.................. Mpolo copped Krong as Scum, ergo Krong is Scum
............ If Mpolo is Insane then
.................. Mpolo copped BigNose as Scum, ergo BN is Town
........................ BN is Insane or Paranoid
........................ If BN is Insane then,
.............................. BN copped Anchorman as Scum, ergo Anchorman is Town (confirmed)
.............................. BN copped JC as Scum, ergo JC is Town
.................................... JC is Sane, Insane or Naive
.................................... If JC is Sane then
.......................................... JC copped GoP as Town, ergo GoP is Town
................................................ GoP is Sane, Insane, or Paranoid
................................................ GoP cannot be Sane if TLC & JC are Sane
................................................ GoP cannot be Insane if Mpolo & BN are Insane
................................................ GoP could be Paranoid
.......................................... JC copped Ibarra as Town, ergo Ibarra is Town
................................................ Ibarra is Sane, Insane, or Paranoid
................................................ Ibarra cannot be Sane if TLC & JC are Sane
................................................ Ibarra cannot be Insane if Mpolo & BN are Insane
................................................ Ibarra could be Paranoid
.......................................... JC cannot be Insane if Mpolo and BN are Insane
.......................................... JC could be Naive
........................ BN could be Paranoid (who said that)
.................. Mpolo copped Krong as Scum, ergo Krong is Town
........................ Krong is Sane, Insane or Naive
........................ If Krong is Sane then
.............................. Krong copped JC as Town, ergo JC is Town
.................................... JC is Sane, Insane or Naive
.................................... JC cannot be Sane if TLC and Krong are Sane
.................................... If JC is Insane then
.......................................... JC copped GoP as Town, ergo GoP is Scum
.......................................... JC copped Ibarra as Town, ergo Ibarra is Scum
.......................................... JC could be Naive
.............................. Krong copped Webby as Town, ergo Webby is Scum
........................ If Krong is Insane then
.............................. Krong copped JC as Town, ergo JC is Scum
.............................. Krong copped Webby as Town, ergo Webby is Scum
........................ Krong could be Naive
............ Mpolo could be Paranoid

If TLC is Insane then
...... TLC copped Mpolo as Town, ergo Mpolo is Scum
...... TLC copped M_P as Scum, ergo M_P is Town
............ M_P is Insane or Paranoid
............ If M_P is Insane then
.................. M_P copped VZ as Scum, ergo VZ is Town (confirmed)
.................. M_P copped Anchorman as Scum, ergo AM is Town (confirmed)
............ M_P could be Paranoid

NOTE: Any indication of Naive, Paranoid or Scum must end the logic tree.

Do NOT take this as definitive. I have been going in circles trying to build this, so I expect atleast 1 mistake.
If there are any mistakes, then please strike-through and put in the corrected line(s).

PS Remember, this is based on an 80-90% confirmation that TLC is Town.

by **dotproduct** » Wed Mar 30, 2011 9:51 pm UTC

Votals:

mpolo (1) - more_people

9 players, 5 to make a decision

by **webby** » Wed Mar 30, 2011 10:27 pm UTC

dotproduct wrote:Votals:

mpolo (1) - more_people
more_people (1) - webby

9 players, 5 to make a decision

That was a FOS, not a vote.

by **webby** » Wed Mar 30, 2011 10:28 pm UTC

EBWOP: I mean mod madness has changed Finger of Suspicion to FOS.

by **dotproduct** » Wed Mar 30, 2011 10:42 pm UTC

fixed

by **dotproduct** » Thu Mar 31, 2011 1:24 am UTC

Mod Madness Rule:

During Mod Madness, votes will only count if they are preceded by "##", i.e.

## vote: person

by **Krong** » Thu Mar 31, 2011 4:23 am UTC

Response to webby:

webby wrote:
Krong wrote:1. Discussion is focused on mpolo, more_people, and tlc, as I believe it's been proven at least 1 of them is scum. OK, why these 3? I'd be surprised if Thor aren't other triads out Thor that we can make the same argument about.

I can't find any other triads, that half-two is a obvious half-two that comes from the fact that tlc is the only half-two with useful cop information. If you can find another half-two, I'd be excited, we so excited to use the same logic with that triad as with this half-two.

/me slaps self
Yeah, this is rather obvious now that I think about it for more than a second. Sorry about that -- that's the only triad we have now.

4. webby treating tlc as if he's the most likely of the bunch to be anthill, when Thor simply isn't proof for it. (He is most likely to be useful if he is anthill, but that's not the same thing).

My argument (and I've called for other people to do the calculation) is that he's the least likely to be scum because if he was scum, that puts the non-scum, non-naive/paranoid cop results that are guilty/guilty or innocent/innocent at either 2-0, 3-0 or 4-0, when we'd expect 50-50. More likely is that the true ratio is either 1-1, 2-1 or 3-1. Exactly how much more likely is up to someone else to calculate.

I don't think I understand your point here, so let's go into it deeper.

Suppose this was "Super Town Dethy", and there were 100 town players and 1 scum. On D1, Player A investigates scum, and no one else does. On D2, Player B investigates scum, having investigated town yesterday. Meanwhile, Player A is investigating town. Everyone else stays on town.

We see that only 3 players out of 101 could have results flipped: A, B, and Scum. A and B each have a 50% chance of being non-naive & non-paranoid, so the expected value of non-fake result flips is 1. With the scum, there's a max of 3 flips, and a minimum of 0. Very far from 50% of players flipping. And this all assumes the scum didn't kill either potential flipper each night.

Now... what's the probability for this game? After 2 days of results, I think there's a potential for 6 people to have flipped, if everyone of those 6 copped scum one night and town the other. But there could be someone copping scum twice, which means that only people copping the other scum could flip -- that would put us down to 4. Scum could have killed potential flippers twice. And we still have a chance that those potential flippers are naive or paranoid. We could calculate the probabilities for each of these, I think, but I don't think it's that unlikely to be at 0 flippers.

Actually, does someone want to make a go at that? If we can figure out the probability of having exactly 1 flipper, that's going to be the same probability that TLC is town, isn't it? Actually, that's not right... we'd want to compare against the probability of exactly 0 flippers; since we know for a fact that we have either 1 or 0, the probabilities for more are not very useful.

5. tlc copping mpolo last night out of order. We're assuming all along that this is a townie mistake, when it could be a great way for scum to further manipulate the results.

I can't see that that would be helpful unless mpolo is scum and wanted to avoid being copped.

mpolo was the one who was extra-copped. And TLC copping out of order produced this triad that's making us want to lynch someone, while it simultaneously makes us trust TLC more -- there's value in that.

I feel pretty comfortable with what you said on the rest of your points, though, so I'm not quite as worried about you being scum anymore.

by **Krong** » Thu Mar 31, 2011 4:32 am UTC

E B W O P: Sorry, I posted that before I'd given it a solid double-check... hope there's nothing too screwy in there.

I think our order of business should be:

1. Figure out the probability of getting exactly 1 result flip and exactly 0 result flips.
2. The ratio of those two will give us odds on whether TLC is scum. For instance, if the probability of 1 result flip is 25%, and of 0 is 15%, there's a 62.5% chance TLC is town and a 37.5% chance he's scum.
3. Figure out if we're willing to lynch TLC at that threshold, keeping in mind the benefits and costs (only benefits on a correct lynch, both benefits and costs on a mislynch.)
4. Lynch TLC, or not.
5. If we don't lynch TLC, lynch one of mpolo and more_people. It's a toss-up as far as the cop results go, I think, so we'd be back at regular post-analysis + vote on opinions.

I'll agree that a nolynch today seems undesirable.

by **BigNose** » Thu Mar 31, 2011 9:20 am UTC

We have a current choice of voting NL or voting for 1 of TLC, Mpolo or M_P.

My preference, as we are still at _6 vs _3 today and therefore will only be at _5 vs _3 tomorrow, plus we would have an extra nights volume of cops results, is a No Lynch today.
_There is also the probability that we would be in a better position of 'knowing' more and therefore less chance of a mis-lynch.

What I have noticed is that some people seem to be dropping into the background of this discourse. Obviouosly TLC, Mpolo and M_P need to speak out and thankfully we have contribution from Webby, Krong and myself, but it has gone light from a _number of others. That is not good play, because despite my _logic tree aiding some in identification of _their role and helps to potentially identify Scum, it does not take into account that anyone of those could be Scum and obviously, _their result could/would be a lie.

In brief: Lurking is a scum-tell. The _logic tree cannot pick out lies (yet).

YAY. Screw the _Fora Noobs and _thier word-filter / April Fools jape.

by **webby** » Thu Mar 31, 2011 9:54 am UTC

Ok, here we go - spoilered for long maths:

Spoiler:

My conclusion is that if we had no information to go on other than the cop results (ie nobody seeming particularly scummy), then we'd assign the following probabilities for town/scum:
mpolo = 64/36
more_people = 64/36
tastelikecoke = 72/28
everyone else = 67/33

That's actually much worse than I'd hoped.

72% chance of getting to 4-2 or better if we lynch more_people and mpolo, but a 28% chance of losing (assuming we can't get more information from copping tonight).

Of course that's just what the maths says, people will form their own opinion on people who are acting more scummy than others, however I provide this as a starting point. No guarantees that everything I've calculated is completely right though.

by **John Citizen** » Thu Mar 31, 2011 10:03 am UTC

6-3 today, mislynch puts us at 4-3 LyLO.
NL puts us at 5-3, MyLO.
Another NL puts us at 4-3, LyLO.

Essentially, it comes down to how sure we are of the lynch candidate. Lynching now gives us a second chance, but a no lynch gives us an extra day of results. This is the important choice we have to make.

Ninja'd

In lack of a scumtell, and going by webby's probabilities, I would suggest a no lynch, but I may change my mind.

by **webby** » Thu Mar 31, 2011 10:16 am UTC

Ok, so obviously it's a bad idea to lynch tomorrow unless we're certain (because if we mislynch, we lose). So we have 2 options.

Option 1:
Lynch 1 of more_people, mpolo and TLC tonight.
Gather as much information as we can about the other 2 during the night (anyone have any idea of a good copping strategy to do this?).
Lynch 1 of the 2 remaining tomorrow and hope we're right, else we lose.

Probability of getting to 5-2: 33-36%
Probability of getting to 4-2: 66-72%
Probability of losing: 28-33%

We may be able to improve these probabilities with smart copping.

Option 2:
Wait, hope we get useful cop results (I've shown that the most likely number of useful cops we have left is 3).

Advantages:
We potentially have more information before we have to make a lynch.

Disadvantages:
The scum are likely to start narrowing down who the sane/insane cops are and killing them off. I don't think we can get more than one useful result out of any cop, and as today has shown that tends to lead to sets of 3, at least 1 of which is scum. Maybe if we're lucky we get a choice between 2.

So I'm going to stick to my idea that a lynch today is best. The question is who...

And yes it would be nice to hear from Ibarra and Gopher of Pern on these issues, as well as the 3 in question, obviously.

by **John Citizen** » Thu Mar 31, 2011 10:23 am UTC

I have to ask, what is FUN?

by **John Citizen** » Thu Mar 31, 2011 10:56 am UTC

EBWOP: Never mind, found it.

by **mpolo** » Thu Mar 31, 2011 11:05 am UTC

losing currently word-filters to losing

It looks to me like we are going to need the results. I would say to no-lynch today, getting results to work with tomorrow. Hopefully we can identify some cop types. The nasty thing is that tastelikecoke will be lost to us (if he is not scum, unless scum goes for a mega-wine strategy).

Tomorrow is MyLO, so it is probably better to NoLynch again, and hopefully the complete table identifies at least 2-3 of the scum.

If we do this, whom should we cop? BigNose suggested Sequential -2, which for me would be copping tastelikecoke. Do we go with that or something else?

by **mpolo** » Thu Mar 31, 2011 11:07 am UTC

E.B.W.O.P. again wordfilters to again (I am only advocating two no-lynches, unless we have definite scum found tomorrow).

by **BigNose** » Thu Mar 31, 2011 2:10 pm UTC

mpolo wrote:If we do this, whom could we cop? BigNose suggested Sequential -2, which for me would be copping tastelikecoke. Do we go with that or something else?

Hmmm, missed that point.

PS If you insert _an underscore before any filtered word, it ain't filtered. (Damn Noobs and _their Fun-week)
FYI Did you know that _MOD gets filtered to Noob - GREAT

by **tastelikecoke** » Fri Apr 01, 2011 4:45 am UTC

Copping mpolo was a honest mistake. I mean logically, copping mpolo is the most possible mistake I could make. It would be illogical to cop John Citizen for example, since that is nowhere near sequential +3. The most likely mistakes I could make is to cop VectorZero(+2), Mpolo (+2), or Anchorman (+4).

That was no room for scum to intentionally make mistakes. VectorZero and Anchorman being dead and all.

by **BigNose** » Fri Apr 01, 2011 9:34 am UTC

TLC, I sort-of had you at 80% _Town until that last _post. It _just feels strange, but I'll ignore that and rely on the _logic.

Given that TLC, M_P & Mpolo are consecutively in order _of the _list and that we have already used +1 & +3 and (I) prefer not to use +/-2 because _of this phenomenon, I propose _an option of +4 or -4 for tonite's copping.

by **webby** » Fri Apr 01, 2011 11:53 am UTC

The talk of having a set +4 (or whatever) idea for copping makes me a little bit uneasy. This is for two reasons:
1. It made sense on the first 2 nights when we had no information to go on. Now, however, surely there's a better strategy? Basically we want results that add to the information we already have on TLC/mpolo/more_people. Not sure what the best way to do this is - one way is that everyone cops whoever out of those 3 they want?
2. It seems a way of changing the subject of the thread. Somehow it shifted from whether to go lynch or no lynch to assuming no lynch and asking how we should cop. For that, I'm slightly suspicious of mpolo and BigNose.

Haven't really formed an opinion on the TLC mistake thing - still trying to work out reasons it might be beneficial for him to say he's copping mpolo instead of BigNose. The only one I can think of would be that BigNose is scum, and that TLC was trying to remove cases where it came down to us knowing that at least one of TLC and BigNose is scum. A slight stretch, but not beyond imagination.

Who is online