Social Network

From 118wiki

Jump to: navigation, search


Social Networks

A Social Network is a term used by sociologists to describe structures in society; it is related to the colloquial use of "network" as a verb (as in, that guy really knows how to network with movers and shakers). We mean Online social network, which is actually just a social network in the traditional sense, but now facilitated by computer networks. Therefore, some of the theory from sociology, anthropology, psychology, political science, and communication studies can be applied to online social networks. However modern networks do more than just carry data, they can add services, databases, ways of pooling knowledge and forming communities in new ways.

Online social networks offer the promise of Metcalfe's, if not Reed's, Law. That is, we will see more value than intuitively expected when enough users participate in an online social network. How does this occur?

  1. The social network, very broadly defined (to include even people who just use search engines and react to what they find), can bring consumers and producers together more efficiently than otherwise possible. Prominent examples:
    1. Ebay.
    2. Froogle.
    3. Online Dating Services. (Itself, a topic of study.)
    4. Craig's List
    5. Yahoo Answers and Google Answers
    6. Zillow
    7. List by Time Magazine (of social network sites)
  2. While some of these are just searching and databases, others actually do more to make connections. A few people get caught up in the philosophy of networks and the relation to commerce (I have in mind the hype of The Cluetrain Manifesto, also Cluetrain, not recommended reading).
  3. But there is apparently more to the value of social networks than making simple connections for commerce or other purposes. If we study Emily Chang's list of social network applications we find the emphasis is on group collaboration. (Another list is Wikipedia's List of social networking websites.) The idea is that the "whole is greater than the sum of the parts" in a group. So it could be possible that an online social network such as Facebook or Myspace (see also here) could bring together a group (of more than two people) together. A group can potentially collaborate to create some common information, which has independent and enduring value even to new members who will join the group, so that value can accumulate and make even more "connections" (perhaps even in the sense of the semantic network).
  4. So how do members of an online "group" (or even something we call "group" without any real evidence) collaborate? Many students are familiar with social bookmarking, where people contribute tagged URLs -- this is what Clay Shirky claims by the tagging notion as something more practical than the full semantic web. Some users voluntarily, though very informally, label URLs, pictures, songs, videos, and anything else online with tags that can be searched. This activity feeds on itself. In some vague sense, these users are forging the links of something like a semantic web, though falling far short of the true vision of the semantic web. The result is a collaborative effort, the product of all the volunteers. Can we harness this volunteer collaboration for some productive, concentrated effort that increases the value of networks? One idea is "crowdsourcing". The idea seems optimistic, but there is business sense to it. We know that search engines, such as Google, are very effective advertising companies: who works for Google? The answer is that, perhaps, you do! That's right, if you've ever created a searchable web page, then maybe you have added value to Google as a company (by summing the long tail). The ultimate example is Ebay: somehow, Ebay manages to get over half a million people working for it without having to pay them salaries, health care or insurance, and yet Ebay pays nothing for the warehousing, shipping, and other messy details that more traditional companies like Walmart have to deal with.
  5. What are some techniques of collaborative, social network systems?
    1. Recommendation systems can be built-in to social network software, assisting users to pool opinions.
    2. Collaborative filtering can find patterns of user preferences, and automatically "cluster" users by these patterns: this can help generate recommended links or new connections.
    3. Reputation systems combine the evaluations of many users to suggest metrics for the level of "trust" (how trustworthy) some book, user, organization, or product has. Ebay uses this to establish the reputation of sellers and buyers, and this contributes greatly to Ebay's value (Amazon might be another example).
    4. Social searching attempts to let users "train" the search engine to do a better job; an example of this could be findory. Some say social search is overrated.
    5. Second life, and other simulated gameworlds give users a collaborative entertainment experience; the worldwide revenue for online, multiplayer games is evidence of considerable value. Users create and consume their own virtual worlds and can build on efforts of others. (A former student sent me the following second life email opinion.)
  6. Recommendation systems don't necessarily work well at present, as noted in this blog entry. They are gradually getting better, as more advanced techniques from machine learning algorithms, data mining, and information science improve things. The current generation of automatic "spam" filters for email aren't perfect, but without them email would be much worse. A recent "trendy" book on the Wisdom of Crowds compares the collective evaluation of many individuals to how markets set efficient prices. There are sometimes hostile reactions to this idea, such as Jaron Lanier's article called Digital Maoism.

This blog entry offers some way to distinguish social network systems from traditional online communities, which existed even before the web.

Social Network Structure

We saw in Information Network that social networks can be studied, perhaps showing that a small-world phenomenon holds. This article finds structure even in the way people make telephone calls (it's surprising). Some file-sharing systems (P2P file sharing) are part of social networks. Since we know that social networks may exhibit the small-world phenomenon, and a characteristic of such small world networks is that some of the nodes are highly connected, motivated an "attack" on this type of network. The RIAA efficiently attacked a file sharing system by bringing legal action against the most highly connected members: this had the effect of breaking apart the network (and that might not have been as easy for a network where everyone had the same number of connections).

There are now many illustrations of the topology of different social networks: a gallery of social structures, a tool to make a graph of your own social network from emails.

Social Software

The description of social networks and the systems that implement social networks or use themes from social networks points out the importance of computer networking and software. The term Social Software refers to these implementations. How important is social software in the overall picture of network applitions? It's likely that the impact of social networks is overestimated (one commentator says social software is just too much work).

Web 2.0

We can't cover social software without mentioning the hype surrounding "Web 2.0", defined by Tim O'Reilly. Read What is Web 2.0 to get details on the vision of Web 2.0. This map of Web 2.0 applications is a useful supplement to Tim O'Reilly's article.

There are now many lists of examples of Web 2.0 software, such as web2list, but the examples that claim to be "Web 2.0" are so varied that we don't really see a precise concept. Some people, like Paul Graham argue there isn't much to the whole Web 2.0 pattern; James Fallows tried using Web 2.0 applications for a few weeks and found the experience doesn't live up to the promise (Tim O'Reilly responds here). While we can't judge how social software will evolve (certainly most "Web 2.0" or Enterprise 2.0 companies will fail), it has generated some interesting protocols and web development frameworks: we'll take a look at these later in the course.

You could get additional pointers from this conference site, however it almost looks like a parody of the hype surrounding social networks, so I don't recommend it.

Supplemental (for Lecture)



Social distance

Dunbar's number


Personal tools