Friday, February 17, 2006

The Network Theory Stuff

Since Jon asked for it, here it is. Warning: it's messy because I decided to dump it before I'd really revised it. If anyone has ideas for how it might be integrated or where I might go with it, let me know.
What the web looks like

When all of these blogs and webpages are linked together, the result is a complex web of information that is constantly changing and growing. Researchers are beginning to try to make sense of this complex web, relying on theories taken from math and physics to help them understand how the web operates. Understanding how the web operates also helps researchers understand how other complex systems might operate, giving us a window into how human beings make connections. Though web sites can be loosely described as having human beings behind their creation, in most cases, web sites do not have a single human being connected to it. Instead, sites such as CNN.com and Cocacola.com are coroporate entities with very little about them that resembles an individual human personality. Still the networks created between CNN.com, Cocacola.com, and other corporate sites resemble the complexity of relationships between human beings. With blogs, there is more of a human element. Even group blogs have a kind of personality that one doesn't see with coroporate sites. Combine the human element with the complex network theory behind the Internet and the results have implications for transforming the way we use the Internet in the classroom.

In Six Degrees, Duncan Watts describes Stanley Milgrim's experiment where he attempted to show that people are no more that six degrees apart. In the experiment, Milgrim sent envelopes to people who were to try to get the envelopes to a businessman in Boston through their acquaintances. Each acquaintance would do the same, marking their name on the list, so that Milgrim could then count how many people the envelope had to go through in order to get to the final destination. What he found was that the average number of people the envelope traveled through was about six, giving us the now-famous phrase “six degrees of separation” (37-39). While everyone seemed to intuitively agree that six degrees separating everyone on the planet, researchers still didn't quite understand why this was the case and if it was always the case and what it all meant.

In his book Linked, Alberto Laszlo Barabasi describes what the web looks like. On the one hand, the web looks a lot like many other complex and emergent systems, both man made and naturally occurring. The web resembles social networks and electric power systems, both man made, and the networks found in cells and brains, which occur naturally. The web is a scale-free network, which simply means that it contains nodes that are much more highly connected than most other nodes, i.e. there are hubs in the network. Imagine the airport hub system, only much bigger. The system is also growing at a very rapid rate. Nodes (blogs or web pages) are constantly being added, hundreds of thousands at a time. When these nodes join the network, they are more likely to link to a hub rather than some other random, less-connected node. Thus, on the Internet, we have a kind of rich-get-richer phenomenon where those sites that already have many links coming into it are more likely to be the recipients of even more links from nodes just joining the network.

Barabasi's research made no distinction between static web sites and blogs, which as I've discussed above, are dynamic and have an underlying technology that essentially pushes them into the web in ways that static web sites couldn't. Later in Linked, after describing the Internet as a scale-free network with highly-connected hubs, Barabasi explains that this isn't a complete description. In fact, he explains, in looking at the sites that are indexed by major search engines, he discovered that at most 40 percent of the Internet is indexed and can be found using a search engine. There exist, he says, islands and areas of the Internet that are difficult to reach. This is why sticking up a web site at random will not draw traffic. Let's say you create an extensive resource site and publish that on your school's website. It's possible someone could find that site if a) the site is linked from another page on the school's site or b) the site can be found through a search of just the school's web site. However, Barabasi explains that there is a less than 10 percent chance of a web site with only one link being found through a search, so in essence, that resource site is isolated from much of the web, available most readily, perhaps, to those associated with the school (who would regularly view the schools web site), but not much beyond the school.

Not only is a lot of the web isolated because no one links to it, but another swath of the web is isolated in two ways. One area links to what Barabasi terms the core of the web, the area that is most highly interconnected and through which one can easily travel. These are often the news sites like CNN.com or NewYorkTimes.com or other popular sites. These sites, however, do not link back to the smaller sites that link to them. You can link in but you can't link out. Following a link from this area of the web into the core leaves you unable to go back (at least not via a direct link). Another similarly isolated area of the web are the corporate sites, advertisers usually, that the core sites link to. These sites rarely link back to the core.

Blogs change this dynamic a little. First, as I mentioned above, sites in the core (i.e. CNN, Washington Post) are including links back to the blogs that are linking to them [picture here of wapo links], there are now ways back out from the core to the area of the web that has linked to it. Blogs are also less likely to become islands since there are now so many blog-specific search engines and sites that are taking advantage of RSS feeds to include blog content on their site. It's likely, however, that blogs by themselves display similar features to the static web. There are some blogs, for example, that don't generally link back to other blogs that have linked to them. And blogs that don't include comments or trackback also do not have as many links coming out of it or give smaller sites the opportunity to share their link. Still, it's much more likely for blogs to reciprocate a link. Perhaps, then, the area of blogs that are linking in is smaller than the corresponding area on the “static” web. Blogs also carry advertising which links to corporate sites where one cannot return back to the blog. And there are blogs that are islands, many of these by choice. They may have opted not to have an RSS or be indexed and may not link to anyone because they don't want to be found. If a blogger wants to be found, however, there are more ways than ever to encourage being found. Understanding how to expose your blog from a technical standpoint provides a practical way to gain an audience for a blog. Knowing what the nature of the network you're working in and how it's changing and how it can be exploited to your benefit can provide further insight into how to prevent your blog from becoming isolated on the web and perhaps how to develop it into a small hub.

From what I have discussed above, one way of taking advantage of the nature of the web would be to try to get links from a hub site. This can be readily accomplished by linking to Washington Post articles, but that link doesn't remain there for long. Adding tags that are picked up by Technorati.com, a search engine, but also a hub, is another way to get a link from a hub. Again, though, the link to the tagged post will drift down the page as others tag their posts in the same way. A more effective way of attracting a large number of visitors is to get a more permanent link, either within a post or preferably, on a blogroll of a very popular blog. Unfortunately, it is difficult to get noticed by the blogs at the very top. Following the rules of the network, most popular blogs link to other popular blogs, creating a tight cluster of blogs, with very few links going outside of that cluster, but many, many links coming in. This is a dramatic display of the power law. Clay Shirky has documented this phenomenon on blogs and notes that when the power law becomes obvious in online communities, people within that community and outsiders who want to break in blame the ones at the top of the system. However, as Barabasi's research shows, the system follows certain mathematical laws which cannot be easily changed even by those at the top.

Clay Shirky has documented the existence of this phenomenon, the power law, on blogs as well. He notes that other Internet communities prior to the advent of blogs followed a similar pattern (Shirky, 1). Shirky notes that when the power law begins to manifest itself in an online community, people within that community begin to blame the ones who are moving to the top of that system. They demand that they be better citizens and try to alleviate this problematic distribution. However, as Barabasi's research has shown, the way this system evolves follows a mathematical formula and basically occurs naturally so there is nothing the people at the top can easily do to rectify the situation.

7 comments:

Jon Breitenbucher said...

What do you want to accomplish with this? I'm sort of jumping in in the middle, or the end, I don't know how close you are to the Grail. What you have explained about the two sources seems like a good synopsis. I've read Barabasi and am familiar with the six-degrees experiment and your treatment does a nice job. Are you unhappy with how they support your thesis? Do you feel they are too technical, not technical enough, ...

I was thinking you were getting into the math. It's pretty cool stuff.

coComment doesn't like the popup. I had to open in a separate tab, but it is working now.

Laura said...

I think my main problem is I don't really understand the math enough to get into it. And, I don't this fits into the chapter I'm working on now. I thought it would because the chapter is primarily a technical one.

What I might do is include this in the chapter on emergence where it can, maybe, tie in with the theoretical underpinning of my approach to blogging and teaching.

Jon Breitenbucher said...

If you want to talk through the math, drop me an e-mail. I'd be happy to try and help you sort it out.

P.D. said...

You say, "However, as Barabasi's research has shown, the way this system evolves follows a mathematical formula and basically occurs naturally so there is nothing the people at the top can easily do to rectify the situation."

This just seems like bad faith to me. The fact that the phenomenon follows predictable patterns without intervention does not mean that intervention could not be effective. Spread of disease is occurs naturally as well, but that doesn't mean that public health measures can't alleviate the problem.

Laura said...

p.d., you're right. Perhaps that's stated to strongly. But it's true that the system is complex enough that it's not as simple as the top bloggers deciding to link only to smaller bloggers (unlikely that they'll even do that; you can't really dictate such things). Because what happens is that when new bloggers come in, they tend to be drawn to the big bloggers and the lists they keep. I guess the real issue is that there are no controls on the internet, no supervising group who can come in set policies that might balance things out.

In terms of public health, certainly there are measures one can take. In fact, some of the research that network theorists rely on are drawn from the way public health agencies deal with epidemics.

Laura said...

jon, I might take you up on that sometime. I've decided I might include this in another chapter.

Collin said...

Laura,

given what you've said here in the last couple of weeks, this may be too little too late, but a couple of months ago, I had an essay come out that grew out of the networked rhetorics course that I did last spring, an essay that tries to make some use of Watts, Shirky, et al.

I almost hesitate to mention it, bc I don't want to tempt you down an avenue that you've already abandoned, but I find the connections among the network stuff and writing classes suggestive, too, and there aren't a lot of other people working on it.

So if it's something you want to return to, and want to chat about this more, just drop me a note.

And good luck on the rest of the project...

cgb