Stoofer's 2nd Law

[bigAl]

It's very comprehensible:

raj - initials of my name

rhcp - red hot chili peppers

freak - me

[bigAl]

With a great use of my precious free time, I have determined

Stoofer's Constant, K, to be equal to 155

, in the equation P = 1 - (K/(K+n)). (Using experimental data from the latest five pages of Mafia Discusion - I did not include any topics after Mr. Stoofer stated his second law, as it seemed biased. Then I fiddled around in excel for a while, trying to find a way to determine the probability. My probably incomprehencible excel file can be downloaded here: http://members.shaw.ca/abphillips/new_o ... toofsK.xls)

Oddly enough (or perhaps not oddly), the MafiaScum community seems to go though phases of people voting or not. As I was skimming through the threads, I would often get many threads in a row with no votes and then later many threads in a row with votes.

Here's my awsome graph:



EDIT: Made the graph smaller. Dragging a corner of it didn't work; instead I just set the zoom a bit lower and took a screen snap shot.

[bigAl]

OK, new stats:

Here's the graph for the past several months (everything from about 24th of jan, 2006 (when stoofer first coined his law?) to feb 15th 2007)



the combined data from last time and this time:



Also, a graph of all the posts (I forgot axis lables - for the x: blue is how many posts it took for someone to vote for somelse, red is the length of the thread if no one voted in it yet , y is the number of threads that that's happened in - sorry if that's confusing):




Edit: By the way this was all taken from the mafia discussion forum. There was quite a peak in votes right after the Law was made known, and then it died down a lot.

[bigAl]

Yep, I took that into account. I sorted all the threads by length, then:

First data point = (number of threads that have a vote in the opening post) / (total number of threads)
Second data point = (number of threads that have either a vote in the second post) / (number of threads that are two posts or more long) + First Data Point
Third data point = (number of threads that have either a vote in the third post) / (number of threads that are three posts or more long) + Second Data Point
Etc.

I

think

that should work, but maybe not. (Not that it gives terribly accurate results anyway - it really needs a bigger sample size, especially for the longer threads. There were only six threads that didn't have a vote in the first 100 posts.)

[bigAl]

Where? I'll soon sort that out.

http://www.mafiascum.net/forum/viewtopic.php?t=526
http://www.mafiascum.net/forum/viewtopic.php?t=763
http://www.mafiascum.net/forum/viewtopic.php?t=2092
http://www.mafiascum.net/forum/viewtopic.php?t=2274
http://www.mafiascum.net/forum/viewtopic.php?t=3432
http://www.mafiascum.net/forum/viewtopic.php?t=2310
I'm not sure if the ms community would appriciate you bumping all those old threads though....

Could you send me the raw data? I'd like to play with this myself.

I have three excel files:
- One for the data from jan24/2006 to feb13/2006 (after the law was announced): http://www.ualberta.ca/~aphillip/StoofsK_new.xls
- One for the data from ???* to jan24/2006 (befure the law was announced): http://www.ualberta.ca/~aphillip/StoofsK.xls
- One for both data combined: http://www.ualberta.ca/~aphillip/Stoofs ... nd_old.xls

I know that they aren't very understandable spreadsheets; the important data is in the first three columns (please ask if they don't make any sense). Also, there may be a little bit of overlap between the two, since the forums (fora?) sort threads in order by the date the

last

post, rather than the first. So if someone bumps a thread that was more than a month or so old, it would might have gotten counted twice.

*I can't seem to find where I wrote down where I started checking the threads from, all I said was "about the past five pages from jan24/2006", which might put it around may/2004.

I'd have to agree with Fritzler's sentiment, at least in the case of this thread.