PDA

View Full Version : [Graph Theory] - Do Kirupa forum threads follow a "fat-tail" power law?



redrum87
July 25th, 2006, 04:41 AM
Let's get geeky... :nerd:

This is slightly talk/random, but it deals with computational science so I thought I'd post it here.

In my efforts to understand the Internet and the web, life, social networks, and more, I've found that graph theory (http://en.wikipedia.org/wiki/Graph_theory) comes in handy. I've noticed over my few years on kirupa forums that there are some unanswered threads, short threads, medium threads, long threads, and ridiculously epic threads. I'm guessing that the topology of the forums, from the perspective of thread-length, is a small world (http://en.wikipedia.org/wiki/Small-world_network) (as opposed to a social perspective, being that this isn't a social networking site and there isn't a "friends" feature, thus tracking who knows who would be very difficult).

My question to you all is, do you think that the length of threads follows a fat-tail distribution (http://en.wikipedia.org/wiki/Skewness)? That is, are most threads unanswered or very short and rarely does a thread reach epic status? To put it another way, as the length of the thread increases, does the number of those threads in existence drop by a predicable factor? Or do you think the distribution of threads follows a bell curve (in which few threads are short, most threads are medium, and few threads are long)?

Why does this matter? It doesn't really in the context of discussion, other than fueling curiosity.

chrisclick
July 27th, 2006, 11:24 AM
well it seems like no one knows what you are talking about... Sorry. :cap:

bwh2
July 27th, 2006, 11:33 AM
...being that this isn't a social networking site and there isn't a "friends" feature, thus tracking who knows who would be very difficult).well, there is a buddy list feature although i don't know if anyone uses it or if it's even enabled.

i would tend to think that the distribution varies by subforum. i would think that in the "value added" forums like flash, server-side, etc., it's more of a fat-tail distrubution. probably more bell shaped in random. but that's just a guess.

The_Vulcan
July 27th, 2006, 09:10 PM
Or do you think the distribution of threads follows a bell curve (in which few threads are short, most threads are medium, and few threads are long)?

Your bell curve sounds right to me.....

However that depends on how you define short, medium and long.
Depending on the numbers you assign them it could easily be a fat tail...

And thats what you gotta hate about stats...
You can just about make the numbers show whatever you want.

redrum87
July 28th, 2006, 01:51 AM
Whoa... I didn't expect this thread to come back to life... :pleased:

What would define short, medium, and long, would depend on the distribution of the threads, starting with 0 replies.

Maybe kirupa has some stats that could help out here, if he's willing to reveal them. Although, I'm not sure but I think that might be some sort of AdSense TOS violation...

kirupa
July 28th, 2006, 03:20 AM
I don't have any stats on which threads get viewed more, etc. ;)

redrum87
July 28th, 2006, 04:15 PM
I didn't think so, but I thought it would at least be worth a shot.

yoyo14
September 29th, 2006, 11:09 AM
hey red drum 87

SlowRoasted
September 29th, 2006, 12:10 PM
If an epic thread is one with 1000 or more posts, and a short thread is 10 posts or less, then I would say that the bell shape curve is skewed to the side of the short threads. It's obvious that there are WAY more short threads than epics.