Six Degrees of Separation, Wallop and More

Last year, a girl from the Software Institute gave me a Wallop invite. I later invited a few friends, plus some Wallop friends I picked up along the way. The social network was laid out, but usage never took off. Not just for me — everywhere I looked was a wasteland, except for some scattered test posts.

Software like Wallop is collectively called “Social Network Service (SNS).” I love guessing English acronyms — I figure SNS stands for Social Network Service. “There’s a trust transfer relationship between people, and this trust transfer hierarchy is limited. An individual’s successful interaction with society must be built on this trust chain.” This is a Social Network model that once won a Nobel Prize, and it’s also the theoretical foundation for community software like Wallop.

Funny story — I first encountered this social network model last semester when I helped a teacher’s daughter (studying in the US) with her VB.NET homework. The assignment was roughly: if you want to find a connection between any actor and another actor in Hollywood, it takes at most six steps. Write a program to test it. This sounds hard to grasp (which is why the professor’s mom was still confused after I explained it three to five times). It was only later that I realized this little program was backed by a sociological theory — Six Degrees of Separation. Side note: you can see how US universities, when teaching理科 (science) students to code, always weave in some humanities knowledge. In contrast, domestic students learning C in college are mostly taught by Grandpa Tan writing “Student->Name = ‘LiMing’.”

Six Degrees of Separation is also called the Small World phenomenon. 35 years ago, American psychologist Stanley Milgram proposed the “Six Degrees of Separation” theory in Psychology Today. He believed any two strangers can connect through “a friend of a friend,” and the number of friends between them is about five. Through this chain of friends, the world is that small.

Wallop’s Network feature seems to demonstrate and validate this theory [see image below]. In the Flash interface, the expanded Network is quite pleasing to the eye. Everyone connected to you can be linked together, and you can “follow the vine” to find more. When browsing anyone’s content, you can use the “Point to me” feature in the shortcut menu to see their relationship to you (the algorithm is obviously the shortest path algorithm — thanks again to the great Dijkstra). But I think this still doesn’t fully represent human relationships.

First, relationships should be vectors, not scalars. That means if two people have a relationship, it could be one-way or two-way. As the saying goes: “Everyone knows the talented poet-painter Tang Bohu, he just doesn’t know me.” But Wallop currently uses a single line to represent relationships, so it can’t express “I know him, he doesn’t know me.”

Second, relationships should have categories — teacher-student, friends, classmates, relatives, colleagues, romantic rivals, etc. To put it plainly, Wallop’s Network should have grouping. QQ and MSN do this well, but Wallop doesn’t — a major shortcoming. Imagine when someone’s relationships get complex and all of them are displayed on the same plane, it’s no longer about aesthetics.

Third, relationships shouldn’t just be vectors — they should have weights and priorities. For example, friends should be divided into close friends, casual friends, acquaintances. Enemies can be divided into ordinary enemies, never-speak-again enemies, want-to-beat-up enemies, father-killing enemies…

In summary, I think a proper relationship model should be classifiable, weighted, directed vectors. As shown below.

My home machine is old — opening Wallop uses too many resources and always hangs. Can’t send mail to MS. If anyone kind would organize this article into English and send it as a proposal to MS, I’d be most grateful.