Linked #2

I’m reading Albert-László Barabási’s Linked: The New Science of Networks now – I borrowed the copy I bought as a present for a friend, and I promise it wasn’t part of any plan, whatsoever; see here for my first post about this book. I’m on chapter eight so far, and it gets more interesting with each page I read. One thing I particularly like is that Barabasi doesn’t reveal too much of what comes further ahead in the book; I can’t explain why, but I both feel led by Barabasi and a few steps ahead of him.

To put it briefly, Barabasi and his fellows were trying to find out if there are laws that govern the growth of networks. After the first seven chapters, they have proved that in the presence of growth and preferential attachment, you get a network that grows organically, maintaining a structure consisting of nodes and hubs – distinguished by the number of links. At some stage, Barabasi asks whether networks, like many things in nature, experience sudden phase transitions from disorder to order, or if they are in a continuous state of transition from disorder to order. Later, he finds that this is not the case; he writes: “No matter how large and complex a network becomes, as long as preferential attachment and growth are present it will maintain its hub-dominated scale-free topology.”

“Scale-free” means that there are no typical, or average, nodes. Instead, the entire network consists of many nodes with few links, and a few nodes with many links (the hubs). The presence of hubs makes a “small world” of a network: that each node is connected to every other node via a small number of other nodes. In other words, because of the simple rule of preferential attachment – that new links are more likely to be made with nodes that already have a large number of links – a structure emerges as the network grows.

Later, Barabasi’s team finds that there’s more to attachment than the number of links, but I think it’s nevertheless very interesting, in its simplicity. One of the reasons I’m reading this book is to see what of what relevance this thinking is to software architecture; if it gives any hints for evolutionary software architecture (as opposed to planned architecture). There’s a trend (at least in the agile community) to try to discover the simple rules behind evolutionary design and architecture. I know of two books in the making that will cover this; and Mary Poppendieck identifies, in her forthcoming book on Lean Development (which I wrote about here), seven rules or guidelines for the entire software development process, but I guess they apply equally for architecture and design. But it’s still a new field, and I hope that many more books and articles will be written about this.

One thing I have been thinking about as I read the book, was if scientists always ignore what they haven’t yet proved, or if Barabasi only does this in the book to not confuse the reader (which might be a good strategy). For instance, it seems unlikely that they – as they sent a robot out on the Internet to gather data about nodes and links – really expected to find nodes to be randomly connected with each other (as Erdös and Rényi’s model said). I got that impression, at least. Also, in the context of creativity, I find the problem solving strategies of scientists to be interesting. For example, how does a mathematician go about solving problems? I guess that it’s a lot less rational than you’d expect.