Tuesday, October 23, 2007

How valuable are social networks?

An interesting article in the Economist on the Facebook phenomenon. It mainly discussed from the business perspective. But it also has an interesting side-question worth elaborating: what are the returns on social networks? Forget the what or how much for a while, are the returns increasing, decreasing or constant?

From one perspective, investment in social networks seem to have increasing returns. Think about your friends as possible sources of information (on job, prospective boy/girlfriend, where to buy new car, new technology or other cool stuffs). Or sources of 'ideas' (how to solve the Hamiltonian homework, who to make fun in the column in the campus billboard). Then the benefit of having ten friends is more than twice of having five.

If this is the case, then companies or any other organizations that base their activities on social networks will have an ever-increasing values if their customers' networks are expanding. Think about cellphone companies. It worth having a cellphone if I have a certain number of people in my networks (the one I will be very likely to call). That's why we see a lot of promotion package like member-get-member, family plan, group package, etc.

The article also wrote the implications for social network websites like Facebook, Friendster, Flickr and so on. But here's the catch, as mentioned in the article:
But unlike other networks, social networks lose value once they go beyond a certain size. “The value of a social network is defined not only by who's on it, but by who's excluded,” says Paul Saffo, a Silicon Valley forecaster. Despite their name, therefore, they do not benefit from the network effect. Already, social networks such as “aSmallWorld”, an exclusive site for the rich and famous, are proliferating. Such networks recognise that people want to hobnob with a chosen few, not to be spammed by random friend-requests.
So which one is it; do social networks have increasing or decreasing returns? To be honest, I don't know the answer. As usual, when you can't give a firm answer, in the conclusion you'd say "this should lead to further studies."

Nevertheless, an old paper by sociologist Mark Granovetter provides a valuable direction. We need to distinguish between strong ties and weak ties. The paper explained the power of weak or impersonal ties -- in which an individual friend may become a hub to other people outside of our network.

So, the optimal strategy may not be investing in as many friends, but in a limited number of friends, who individually have other friends in their own networks.


7 comments:

  1. from the individual perspective, economists would argue that it's diminishing because maintaining connections is costly.

    from the structural perspective, even if there are short connections to almost everyone, finding and accessing these connections are different matters.
    e.g, a CEO knows a lot of other CEOs and he/she can readily access these CEOs. a limo driver may know a lot of CEOs too but the driver's access is limited.

    on weak ties.

    weak ties are good for obtaining diverse information. in other words, if you want to reach to as many nodes as possible, weak ties are useful.

    however, if you actively search for information, weak ties can be not very useful. although people who are weakly connected might have better information but they might be reluctant to help.

    strong ties i.e. people who are close to you, are more likely to put extra efforts that results in higher probability of successful search and hence higher returns.

    so it seems to me that the context of the process operating on the network is important here. the number of friends a person has is only part of the story.

    in short, i don't have the answer either; have to finish the dissertation first :|

    ReplyDelete
  2. Hmm... the conundrum between size and privacy could be avoided by having multiple setting.

    So some updates, events or photos could be seen by all in the network and the more personal posts only by few.

    Of course it could be more than 2 settings. Office colleague only, family only, friend only. You named it.

    ReplyDelete
  3. from the psychological perspective, the connection need not be reciprocal. a driver might think he has limited access to a ceo, but perhaps the ceo thinks differently.

    so i guess structure and context are important, but one's belief about the degree of his or her relationship with others would be crucial -- and it would be interesting to see how individual beliefs are accommodated in modelling social networks.

    ReplyDelete
  4. tirta: there is research on 'cognitive social structure' by sociologists and social psychologists. actually i was thinking of doing this kind of experiments with colleagues from UI's social psychology dept. i have some preliminary data about the effect of perceptions from my experiment, which i think worth to be pursued.

    in this area, i think experiment is more interesting than modeling.

    Btw, in addition to the $15bn facebook phenomenon, here is another social networking phenomenon:

    USA Today:"... the terror watch list has swollen to 755,000 with 200,000 people per year being added since 2004. Adding about 548 people daily every day ..."

    context is important :D

    ReplyDelete
  5. roby: care to share the findings of your experiment, in a nutshell?

    ReplyDelete
  6. tirta: the question is to what extent the perception one has about network affects the ability to utilize the network.

    from my current data: it seems not very much: at most only about 6% of all failures to use network can be attributed to perception.

    but the experiment i'm using is not specifically designed to answer that question. so it's a crude result.

    i think we can design a clean experiment just to address precisely the question about perception on networks.

    i won't be surprised if perception matters; but by how much?

    do you have any thought about this?

    ReplyDelete
  7. Roby and Tirta -- indeed I expected this post will trigger your responses. Please do share and elaborate your studies further sometimes.

    ReplyDelete