Friday, January 29, 2010

Mail

I got this email from Kevan:

The following rolled-up message dropped into the wire tray of the postal system, marked with a Detective’s Seal:-

“I don’t know what this proves, but I got your email… you know who you are.”

Comments

Klisz:

29-01-2010 14:56:08 UTC

Also, Roujo has exactly a 34% chance of being a murderer. (That is, I used my role power to find out if Roujo was a murderer or not, and the answer was FALSE.)

Thrawn:

29-01-2010 15:37:33 UTC

Actually a 2.29% chance, assuming Alethiophile is a murderer.
You’d get the result you got if: He’s not a murderer (22/23 chance) and you’re right (66% chance) or he is the murderer (1/23 chance) and you’re wrong (34% chance). As you can see the probability of you being wrong is heavily outweighed by the probability of him not being a killer.
The odds of him being the murderer are (1/23*.34)/(1/23*.34+22/23*.66)
Interestingly, assuming we don’t know about Alethiophile, if your power detects that someone is a murderer you only have a 15% chance of being right (or 8% if we’re right about Alethiophile), since the probability of you being right is also heavily outweighed by the probability of them not being a murderer.
Don’t worry, your power will get more useful as more innocent guests are killed off and the possibility of false positives drops. False negatives will get more likely, but that’s life.

Hix:

29-01-2010 15:54:52 UTC

Great.  A scientist who can’t even apply Bayes’ Law to properly interpret data.  You really think that the slim chance that Roujo was a murder before you performed an investigation actually went UP to 34% because your investigation found extra evidence that e WASN’T a murderer?  Your conclusion doesn’t even pass that basic sanity check.

tecslicer:

29-01-2010 16:00:29 UTC

I wish I was better at math. I am so glad that my power is only making people calm down. But no, he is not a murder. We already tied up alethiophile and Orn just (In his flavor text) admitted to being a murder so Roujo can’t be.

Thrawn:

29-01-2010 16:02:28 UTC

To be fair he is a Professor of Philology. If we need to determine whether a supposedly ancient document is actually a later forgery, he’ll be very useful.

Klisz:

29-01-2010 16:09:03 UTC

@Hix: What I meant is, there si a 34% chance my data is incorrect.

Klisz:

29-01-2010 16:10:42 UTC

@tecslicer: That could easily be a typo, or he could be a detective.

tecslicer:

29-01-2010 16:17:00 UTC

When the town wakes up in a game of Mafia, it is generally considered a failure NOT to kill someone. And In the absence of real solid irrefutable evidence, Such as your brilliant and confusing equations, a slip of the tongue or slight sound in the night is all that is required to rise the mob.

We could at least tie him up…a little.

Uvthenfuv:

29-01-2010 16:22:04 UTC

I guess there’s no need for it, but I also got that message.

Also, another letter that arrived earlier than the other one:

“Did someone ask for applesauce and teddy grahams?”

...why didn’t anyone mention it?
I’m starting to feel paranoid O_O

Hix:

29-01-2010 17:20:12 UTC

Darth Cliche said: “What I meant is, there si a 34% chance my data is incorrect.”

No there is not a 34% chance that your data is incorrect.  Your data’s incorrect status is equivalent to Roujo’s Murderer status.  The chance that your data is incorrect is 2.29%, as calculated by Thrawn.

Before you recieved an answer, the chance that it would eventually be incorrect was 34%.  Now that you’ve got an answer of “X is not a murderer”, the chance that it is incorrect has dropped.