I purposely did not read Adam’s post before writing mine, so apologies for anything repeated. I wanted to preserve my ideas about the book while writing.
Before I vent my frustration at the oversights of this book, I’ll give you a fairly unbiased summary of Barabasi’s book. Linked gives readers a detailed account of the discoveries that lead to our current understanding of networks and how they work. The discoveries that were most emphasized by the author were that of the Random network and of the Scale Free network model.
The Random network was the child of Mathematicians Paul Erdos and Alfred Renyi, who were searching for a way to explain how relationships are formed in complex systems, and it is exactly what it sounds like. They decided to look for the simplest answer, which was to connect nodes by rolling a set of dice; choose two nodes, then take a roll. If you roll a six, connect them. They found that from this strategy, they saw a wide variation in the number of connections for the nodes, but, generally, all nodes had about the same number. If you graphed the number of connections per node, you’d get a bell curve. Later study of random networks gave credibility to the idea of six degrees of separation/Kevin Bacon, stating that every person is only six connections away from any other person in the world, including Kevin Bacon.
While some networks follow this pattern, most display hubs, or nodes with a disproportionate number of connections. Mathematically, these are almost impossible in a Random network. So came about the Scale-Free model, in which the majority of the nodes found have about the same number of links, but a few(hubs) acquire a disproportionate number. This model follows a “power law.” To explain this distribution, though, it required more than rolling dice. They began adding factors to the likelihood that two nodes would be connected. These were growth, the fact that a network is growing in size and therefore a node can only link to existing nodes, preferential attachment, or the idea that a node’s preference for other nodes is proportional to the number of nodes they already have, and fitness, a node’s ability to compete for links. Barabasi goes on to apply these models to a number of examples, from social networks, to religion, to the internet, to terrorist networks
So, while all this is interesting, I found myself having this reaction when I finished:
Or in other words, I’m not surprised. These are great models, but the ideas they illustrate seemed obvious to me. When Barabasi stated that most networks displayed power laws in terms of distribution, I made a note in my book saying, “wouldn’t historians(or anyone) find this obvious?” Yes, they would. Barabasi notes that we didn’t really have this kind of understanding of networks until we had the web, because it’s so highly mapped out that we could study it. But what about the networks that historians have been studying for a very, very long time? Throughout history, there’s been a pretty consistent distribution of power- most people have about the same amount, almost none, and a few have a whole lot. It’s great that we have models for these things now, but it seems like in most of the examples, the author just seemed to let the details that other people put before him slip by. Oh wait, that’s exactly what happened.
“To look at the networks behind such complex systems as the cell or society, we concealed all the details. By seeing only nodes and links, we were privileged to observe the architecture of complexity. By distancing ourselves from the particulars, we glimpsed the universal organizing principles behind these complex systems.”(225)
The author purposefully ignored the details that were available in order to generalize about these systems. If we’re not being particular, then, shouldn’t we extract ourselves even more, to view the entire universe and all networks as clusters of nodes along a graph? Or maybe the details are what allow us to have a true understanding of how these networks work, instead of giving wide statements that can barely be argued against. He goes into a few specific examples of how certain properties of networks are demonstrated, but brings them straight out to the quantitative form, without addressing the cultural or historical context of the situation. It seems that this removal of details, while simplifying things, also blocked the obvious answers to his questions.
“How could these costly and heavy copper and optical connections follow the same rules as humans do when establishing their weightless social links or adding URLS to their Webpage?”(151, referring to the internet)
Because they’re created by humans. Is it surprising that humans would create something that would act the same way as they do when interacting with each other? In the quest for quantitative objectivity, people often lose sight of the fact that context exists for everything, that more information is not a bad thing. Even information that you don’t like, or see a value for. This is why it’s really easy to tell when a website has been over-designed or over-engineered. A broader knowledge base allows us to make better decisions and observations because we’re seeing problems from more angles. At least when we’re talking about something like networks. Barabasi, in my opinion, treated networks as something disconnected from qualitative areas of study, and as a result, gave observations that fell short of the revelations I hoped to find based on the title.