A couple of years ago, a large list of phished Myspace accounts was leaked on the internet.
I stumpled upon them and ran a very simple analysis. Check it out:

root@a:/# cat myspace.hackedlist | cut -f 2 -d : | sort | uniq -c | sort -n | tail -n 20
14 qwerty1
15 123456a
15 babygirl1
15 blink182
16 123456
16 123abc
16 iloveyou2
17 football1
17 nicole1
18 number1
19 password
23 myspace1
24 fuckyou1
28 iloveyou1
28 monkey1
29 fuckyou
54 abc123
74 password1

The file was in the form of “Username:password”, so the first part of that command “cuts” the second column, with the colon as the delimiter. Then it is piped through sort, which sorts the list alphabetcially, then the uniq -c command, which counts the number of times that a word shows up, then I sort it again to get the most freqent passwords, and tail the last 20 lines.

It is interesting to see that a lot of these passwords just tack “1″ on to them. And of course blink182 was all the rage back then aparently…

But the Myspace chain bullitens or picture comments are not what makes Myspace.com interesting. Its is the social network map that is what is interesting. But how can I see the big picture? And what does it look like? In my search on how to do this, I came across Ben Discoe’s website and read about his work with a similar website: friendster. I decided on a similar approach.

First, because I do not have direct access to Myspace.com’s database of friends, I would have to create my own local database, and then write a spider to crawl around and populate it. So I did. I wrote it in what I know best, are you ready people: bash scripting. You can look at it for yourself in the download below, its called myspace.robot. Basically it parses a myspace profile, grabs friends, pictures, music, interests, as much information that can be harvested, and then inserts it into the database. This robot is called by other scripts that say oh, pick a random profile. Or I have a robot caller that calls in a tree like fashion, starting at a given place and traversing so many levels (myspace.tree).

Once I have the information (friends in particular). We can use this information to create a map. At first I used imagemagick to generate an image. It sucked.

It took me a while to find graphviz.

Graphviz is the perfect tool for the job. I feed it the nodes, and weights between nodes. It calculated the best positions to put the nodes. It is very configurable, including different ploting algorithms, graph styles, and output formats. It take all of this and put it into a php-mysql interface for others to input their graph-making requests. My server on the backend takes the top one on the queue, grabs the appropriate profiles, creates the friend-map, then sends an email to let the user know that the map is ready!

A simple map that includes you and your friends turns out to be pretty large. If you want to go any deeper than that you have to use svg. (bitmaps turn very large.) But how does the program know where to put each friend, realative to the other friends in the map? Answer: It uses the “howgoodofafriendareyou” metric. The more friends you have in common with another person, the better a friend you are.
I’ll have the link up in a while for the interface for others to make their own maps. You might be interested to see who myspace thinks their closest friend is.