|
Vatican City State1650 Posts
On May 06 2011 11:03 DropBear wrote:SUSPICIOUS PEOPLE UPDATED
Beneather
redtooth
VarpuliS
Cthsazsa
sinani206
ilovejonn
Lyter Yes I realise there are seven names there. PEOPLE I AM UNSURE ABOUT
chaos13
Rising_Phoenix
orgolove My feelings on Beneather have not changed. It does not seem that I can convince anyone to join me so I'll likely switch to Cthsazsa as he has the most votes of my suspects. On redtooth, I just wanted to point out this: Show nested quote +On May 06 2011 06:06 redtooth wrote:
Why KillerSOS is scum.I don't think he's as scummy as Chaoser. However, they're probably not both scum so I haven't pushed for a two pronged press in opposite directions. Let's start off by examining something someone else said (since my credibility seems to be shot at this point): On May 06 2011 00:50 Jackal58 wrote:
Could be all that. Could be just overwhelmed noobs. If Irish is scum he'll out himself. Cthsazsa already has. His response to my poke was over the top and quite extended. His post I poked him about was a pure scum agenda.
His response was pure scum response. He constantly evades the issue I poked him on and shifts it to a not 'knowing the definitions of lurker/inactive"
That's not the item in his post I FoSed him on and I told him that. Yet he still tries to deflect back to that issue. He's scum. Look at the second bolded part. That is typical noobscumtell - overreacting and defending through quantity, not quality. Jackal is right in that Cthsazsa has exhibited this behavior but so has KillerSOS and to a much greater extent IMO. His posts have acknowledged the viable arguments against him but simply ignored them. Here's the explanation for these guys: "They are scum who overreacted when called out. However, instead of addressing the arguments against them, they try to ignore them like its a non-story but continue to participate in the thread regardless." The first bolded part is there because I agree. But we'll have more time for that later.
Ultimately, there simply isn't enough time for me to argue a lynch on Chaoser, especially since I have a final tomorrow so won't be as active for the rest of the day. I'll just wait until post-lynch when I can start off every post with "Hey I'm redtooth. I told you so." You may not be a noob, but you're doing exactly what you say is scum play. You then go and derail things with a discussion about the chances of a day vig finding a target? I'm not sure how this helps at all. I can't tie you to either of the major lynch targets so I'm not as strong on you as I am the other candidates on the list.
Do note that I've been pushing on redtooth but abandoned it as no one had picked up on his scum tells early in the game.
|
Vatican City State1650 Posts
On May 06 2011 10:43 VarpuliS wrote:
@GGQ: I <3 your sig.
Could you link to the game w/ Serejai that you're referring to?
<3 could be a breadcrumb?
FoS VarpuliS
|
Vatican City State1650 Posts
Although I've initially suspected redtooth because of his (kinda dumb) initial post that asked people to share their mafia experiences...
The past ~10 pages have convinced me otherwise.
He's really putting serious effort into his posts, and he's trying to trick things without being afraid of being revealed. I can't detect any hint of red-ness (pun intended).
|
Vatican City State1650 Posts
On May 07 2011 13:49 DropBear wrote:
What redtooth is doing is derailing town. Don't listen to him.
Yeah. Actually, this was how I felt in the first place... he started the game by attempting to lead town's discussion into "sharing your mafia experiences"
And now he's doing shit like this.
ugh. You know what. Despite all my misgivings, I'm voting on him.
|
Vatican City State1650 Posts
I really think redtooth either needs to be checked by a dt or be lynched asap. He's been acting plenty scummy enough, and his kills will give us plenty of information both on Amber and others involved in this dispute.
|
Vatican City State1650 Posts
... Somehow the lynch is inexhorably going towards people who are completely different from our initial focus. I think we're being led by the nose into reds' preferred targets, as we hit the mark at the beginning of day 2...
Redtooth
|
Vatican City State1650 Posts
On May 08 2011 17:16 AirbladeOrange wrote:
I'm good if town wants Amber and Cthsazsa as two of the lynch candidates. Irish will probably be modkilled.
btw. Normal townies wouldn't be this certain about lynch targets. Only reds truly know who would be good for them to be lynched. The tone of this post almost 100% confirms to me that Airblade is another red.
|
Vatican City State1650 Posts
Sigh... looks like redtooth lives another day. I highly doubt his vanilla-ness at this point. Either he's a blue role or he's a red. If he lives by day 2~3, that in itself will practically prove that he's a red, as the mafia would've taken care of him otherwise.
Anyway, chaoser makes some pretty good points about amber, compared to Cth...
I really dislike how the votes are so close that the reds can easily manipulate the totals at this point.
|
Vatican City State1650 Posts
Lawl. Excellent.
Now we have a town lynch, with the voting patterns to show for it. It was a close one up to just a few hours before the end of the voting period = plenty of information.
We just need to look at the people:
1. stayed with someone else to the end
2. last-minute switched from another enemy to amber
4 mafia left - with good DT checks and/or vigi kills, we might win this next day.
|
Vatican City State1650 Posts
|
Vatican City State1650 Posts
An hour and 20 minutes huh. I'll post it later.
|
Vatican City State1650 Posts
This is in progress - I'm posting this just in case I'm killed in the night (though not likely)
[url blocked]
The excel file contains the vote time and target for the past 2 days. It's quite interesting. I'll post a full analysis after day breaks and I have a few more identifiers.
|
Vatican City State1650 Posts
Ugh. I refuse to believe that town medics were stupid enough to let chaoser die. A no medic setup, maybe?
|
Vatican City State1650 Posts
Ok. I'm done. First, the DL link:
[url blocked]
So, lets begin.
The Theory of Redness Factor and its Applications
Introduction
I am in firm belief that patterns can be derived from TL mafia (or most any forum mafia, for that matter) by looking at the voting patterns smartly and objectively. If one carefully deduces the characteristic voting patterns of mafia members, one can numerically convert such patterns into the "redness" of each particular voting behavior. If one can correctly identify the factors and calibrate the correct conversions, one can find who the reds are just from the voting data.
From this premise, I first went through the votes of each player, and listed them in the order they appeared in the public voting thread.
Day 1:
![[image loading]](http://i.imgur.com/vbTlk.png)
Day 2:
![[image loading]](http://i.imgur.com/ImKJ7.png)
Proposed Protocol of Data Analysis
Now that we had the voting patterns for both days, and especially the second day during which there was a closely fought red lynch, I decided to consider typical red voting patterns and its numerical weight in my scale. The below was what I came to, after careful deliberation:
![[image loading]](http://i.imgur.com/of1eg.png)
Some justifications:
general point values of each redness factor category
a: simple. Since reds have full information, they are more likely to vote for townies than one of their own.
b: reds have a greater motivation to change their votes according to the game situation at the time, since again, they have more information than the town.
c. since reds can PM each other in this particular game, they are more likely to agree on a vote when they PM each other and post it in close time periods.
d. reds have a greater motivation to lurk than townies.
e. same reason as a - reds have less chance to vote for their own than townies. Even simple probability shows it.
crunch time - when there's a closely fought red lynch near end of the day
f. when there's a tie or a red is leading the lynch vote total, reds have a much, much greater chance to vote away from the lead, even if they initially voted for each other to hide their identities.
g. again, red have a much greater chance to vote for the next closest target in effort to save their own.
h. reds have zero motivation to vote for their member that's closest to being lynched, so anyone who does so is most likely not a red.
i. again, reds have no motivation to lower the vote total of a townie that's the vote leader, and put their member that much closer to danger.
j. once the red lynch is inevitable, the reds will try to blend in the bandwagon by mass voting for the red vote leader. This is almost a guarantee, and thus is included in the consideration.
Further, as the reds would not fight so hard over the lynch if #2 is a red, any such hard fought lynch almost 100% guarantees the identity of the second most vote getter.
Driving from this revelation, we can then consider that anyone who stayed their votes on #2 despite the red lynch also has a higher chance of being a red than other voters.
Results
After setting up the parameters, I manually went through each vote and applied the point totals.
Day 1:
![[image loading]](http://i.imgur.com/eBtqR.png)
Day 2:
![[image loading]](http://i.imgur.com/OKadP.png)
As we can see, in day 2, when there were a lot of lead switches back and forth as time went on, which gave us quite a lot of data to work with. To give you a perspective, below is a graphical representation of how the vote totals changed as each vote was cast.
![[image loading]](http://i.imgur.com/ixeza.png)
Conclusion and discussion
Analysis of above patterns let us assign a "red factor" value to each player. A ranking of these players by the red factor gives us four principal suspects:
![[image loading]](http://i.imgur.com/xUx3K.png)
Of course, this process isn't completely infallible. For instance, I probably would not have caught GGQ with this method, as he voted for cthsazsa just once during the second day. This also shows the most important flaw of this method - lack of data. The more data I have, the more accurate I can get.
At this point, given all available data, I believe that the last 4 remaining reds are:
- DropBear
- ilovejonn
- AirbladeOrange
- Varpulis
... Next time I'm going to write a program instead of doing all this by hand.
|
Vatican City State1650 Posts
O WHOOPS
ok then 3 of the 4. Probably the first 3.
|
Vatican City State1650 Posts
I wish I can edit posts. The reasons why I need more data are:
1. randomness. If you only have 2 days to use as data, one day can be an anomaly and you wouldn't know. But if you have 10 days worth of data, patterns can be much more pronounced than temporary anomalies.
2. calibration. The numerical values I assigned to each behavioral patterns are not final. With more data, I can tweak those values and introduce new parameters to better predict the reds.
|
Vatican City State1650 Posts
All right. I got permission to edit two of the pics. See edited.
|
Vatican City State1650 Posts
On May 10 2011 17:16 Mig wrote:
orglove could you show how the numbers would look not taking into account the last minute jump onto amber. After chaoser claimed dt I feel like everyone was pretty much forced to vote amber regardless of being town/mafia and the only people who didn't were probably the ones who weren't online in time.
So basically I would like to know what the numbers would look like if voting had stopped at 11:15 with dropbears vote.
The numbers come out like this when I take away all the points from votes after 11:15:
![[image loading]](http://i.imgur.com/b3d2B.png)
|
Vatican City State1650 Posts
10 out of those 17 comes from the fact that he came second to a red lynch, and all the high red-value players were willing to vote on him and see him through before the votes turned against them...
It's not like Amber was the godfather or even the roleblocker. Amber was just a mafia goon, and the reds were willing to lynch Cthsazsa over Amber...
I think I can be very sure that Cthsazsa is town aligned.
|
Vatican City State1650 Posts
On May 10 2011 17:25 Forumite wrote:
Orgolove, could you throw up the calculations that resulted in those numbers for the following players? Jaminz, MIg, Orgolove and Cthsazsa.
I just want to know why they come so far down.
Jaminz:
day 1:
+1 from voting for a confirmed blue
+0.5 from voting within an hour of GGQ
day 2:
+0.5 from voting within an hour of GGQ
-1 from voting Amber
-3 from voting on a red when the lynch was on a non-red
-5 from making the vote to change the totals from a non-red lynch to tie
total: -7
Mig:
day 1:
+1 from voting a confirmed green
day 2:
-1 from voting Amber (confirmed red)
-3 from voting on a red when a non-red is leading the lynch
-5 from making the vote to change the totals from a non-red lynch to tie
Total: -8
Orgolove (lol)
day 1:
0.5 from changing vote once
day 2:
0.5 from changing vote once
-1 from voting a confirmed red
-3 from voting on a red when a non-red is leading the lynch
-5 from making the vote to change the totals from a non-red lynch to tie
Total: -8
Cthsazsa:
Day 2:
vote on May 9th, 4:52
+0.5 from changing vote once
vote on May 9th, 9:35
+1 from changing vote 2nd time
+0.5 from voting within an hour of a confirmed red (GGQ)
-1 from voting for a confirmed red
-3 from voting on a red when there's a tie
-5 from making the vote to change the totals from a tie to a red lynch
adjustment
-10 coming 2nd place in a closely fought red lynch
|
|
|
|
|
|
EPT
