One of the most popular aspects of networks is this notion of degrees. Now there is a few ways to look at the degrees the most popular aspect is the degrees of separation. The popular history behind this starts with Stanley Milgram http://en.wikipedia.org/wiki/Stanley_Milgram and his small world experiment http://en.wikipedia.org/wiki/Small_world_experiment . The original history starts with a short story call “chains” by Frigyes Karinthy http://en.wikipedia.org/wiki/Frigyes_Karinthy . What Stanely said is that the world and all the people in it make up a network and this network connects everyone through a series of connections via your firends and aquatenances. This became the basis for his small world experiment and for the popularization of the degrees of separation. The most popular of these degrees is the six degrees of Kevin Bacon in which a few college student gained there fame by stating that they can link any actor with Kevin in 6 steps or less. There are two great books to read on this topic “Six Degrees” by Duncan Watts and “Linked” by Laszlo Barabasi. The important point here is not that we can reach any person in six steps or less, it is interesting but not the critical bit of knowledge that was gained from this research. What was found is that in these networks not everyone is a equidistance and some people have more links to them then others. This concept has lead to the notion of hubs. So our world is small but not equal. Our networks are not all represented by random distributions and two of the most popular (world population and the web) are not equal from the point of view of the links to every node.
This brings us to another degree and this degree is small in nature but very very important this is the in and out degree of each node in a network. If you wanted a sucessful web page you would want bi-directional link and the degree of your site would be the number of links that come in to your site. What becomes important is how many links to where they go. Obviously if they go to another web site that has no outbound links and a very small number of in bound links that would not be very good for your site but on the other hand if you had a link to a large active (large number of in and out bound links) web site that would be very very good for you don’t you think???
4 Comments
Comments RSS TrackBack Identifier URI
Leave a comment
I think it’s an excellent idea to apply some sort of link ‘degree counter’ engine as a metric for a website, a la LinkedIn’s network graphing. As a website author, you may be able to gauge your site’s popularity and potential success by building such a network graph.
In building such a tool, or performing such analysis, however, we must be careful not to categorize links or sites too broadly. For example, having a large number of inbound links to a given Wikipedia entry (or the Wikipedia itself) would be natural and expected- but the outbound links would be few, and should end only in single-source reference material, not in ever-expanding link patterns. Special care must also be taken in advertising, as a given site may subscribe to an advertising engine, which could provide a completely irrelevant link, rendering the graph useless.
Great point Paul I totally agree in fact I think that many link analysis that are done are incorrect due to the “peripheral linking ” that goes on. Thanks
Cyril
Interesting link on degrees of separation on instnat messaging http://www.washingtonpost.com/wp-dyn/content/article/2008/08/01/AR2008080103718_pf.html
Interesting, but fatally flawed article. The research was limited to the MSN Messenger network, which does not represent more than a chunk of humanity- and a relatively small chunk at that. (I cannot cite a reference right now to the number of MSN/Windows Live Messenger subscribers, but I’d venture a guess that it’s only in the 10s of millions. Large by any network standard- but insignificant in the totality of humanity). It also follows that subscribers to any given computer based network have something in common- if nothing else, the resources to own a computer and network subscription.
I’d be willing to bet that the reporters of this article, not the researchers themselves, are the ones making the flawed statements here- but the reporting is awful, failing to link to the source research, or even approximate the number of nodes in the network. Commenting on the number of messages that were reviewed actually raises my concern some- I know of at least one major corporation that uses MSN Messenger as a corporate IM client- and I’m certain they would have a conniption if they even thought some non-NDA’d researchers were browsing through their message traffic. Again, this is likely the reporting giving the misconception- the data was likely sanitized, but that somehow got missed in the reporting.