Linked: the New Science of Networks

            Throughout the course of the semester we have been exploring the idea of emergence, the idea of how new behaviors and properties arise, often, in the absence of a creator. In doing so, we have been trying to make sense of certain phenomena and attempting to gain a better understanding of many complex questions that we face today relating to the idea of complexity, order, disorder, and randomness. In his recent book, Linked: the New Science of Networks, Albert-Laszlo Barabasi describes his experience with networks both the development of real-world network theory and scale-free networks. Barabasi starts by introducing the aim of the book saying that the point is to “get you to think about networks and to “explore, link by link, the next scientific revolution: the new science of networks” (8). He argues that there is an important value in studying networks since they offer a powerful abstraction for viewing the world around us. In addition, Barabasi states that networks are present everywhere and that all one really needs is an eye for them.

            By placing importance on the interconnectedness of living matter, Barabasi makes the point that nothing happens in isolation and that diverse systems such as the economy, the internet and the cell share amazing similarities in the way that they function. These shared similarities have made it possible to notice a trend of interconnectedness in networks. Understanding the interconnected nature of networks is key if one wants to make sense of emerging properties or behaviors such as failures. A real life example of the value of understanding networks is evident in the health research world in which “from manic depression to cancer we are realizing that most illnesses are not caused by a single malfunctioning gene, rather there are several genes interacting through a complex hidden network within our cells that are simultaneously responsible (180). By exploring networks and the properties that govern the way in which they function, one gains a better understanding of how and why things such as attack vulnerability, system failures, are not random, but rather a result of the development and evolution of networks. Linked is a book about networks, how they emerge, what they look like and how they evolve.

            Important to the topic of networks is the fact that networks are self-organized. Barabasi uses the example of a spider web that develops without a spider, which can be understood as the existence of a network that runs without a central creator. In this way, networks offer a “vivid example of how independent actions of millions of nodes and links lead to a spectacular emergent behavior” (221). To a large extent this is an idea that we have been exploring in the course of Emergence, the idea of how many simple interactions over time can lead to seemingly complex emergent behavior and properties. In the beginning Linked, Barabasi begins by introducing Random Network Theory, which was introduced by two mathematicians Erdos and Renyi in 1959. This theory, which dominated the understanding of networks until recently, is based on the notion that the universe is random and that networks are also random, meaning that the probability of being connected to any node is the same for all nodes. However, through the study of networks across diverse systems such as the economy, the cell, and the Internet, we have come to understand that most networks do not fall under this type of classification. Real networks are not as random or static as Erdos and Renyi had envisioned. We know now that factors such as chance and randomness do play an important role in the construction of networks. However, we now know that growth plays a key role in the development of a network’s topology. Also, real networks are not as centralized as they once were thought to be; a hierarchy of hubs that keep the network together instead, characterizes them. The idea of a hierarchy among hubs can be understood as the following: heavily connected nodes followed by less connected ones, followed by several less connected, etc. Barabasi states that people are attracted to hubs in social networks and that attacking hubs is more destructive to a system because it affects the nodes that are connected to it. Although some would argue that this is a weakness of a scale-free network, Barabasi argues that the vulnerability to attack is an inherent property of scale-free networks. If one attack

            Based on the notion that scale-free networks are vulnerable to attack and that this is inherent property of this type of network, it seems that this would be very problematic in real-life examples. However, the Internet, the cell, the WWW, and social networks are all scale-free and are known to have a resilience or error tolerance. How is this possible? Barabasi explains that certain networks have a topological robustness, which allows for them to tolerate error without the network totally becoming destroyed. Furthermore, decentralized and distributed networks have a noticeable advantage over centralized networks, because attacking a node or a hub in either a decentralized or distributed network will not automatically destroy the system. However, attacking several hubs in a decentralized network can be very destructive. Furthermore, although network systems in which the nodes have power law distributed connections that can be resistant to random failures, there is vulnerability to carefully targeted attack. Informed hackers and agents that attempt to deliberately damage a network typically eliminate nodes carefully and not randomly, in search of the most connected nodes.

            In addition to Barabasi’s discussion of attacks on networks, he also discusses the phenomenon of cascading failures and suggests that large scale cascading failures may be the result of failure spreading from highly connected hub to highly connected hub. A reali life example of this is in  blackouts (power outages) in which a highly interconnected system makes sense for a more efficient use of natural resources, but when something goes wrong there is a cascading failure.

            Throughout the course of the semester we have been focusing on emergence, trying to make sense of this idea by using computer models and by exploring different theories on randomness and complexity. In the case of Barabasi, he has been working with people from different backgrounds such as physics and mathematics in order to try to gain a better understanding of networks. This is very similar to what we are doing, except we have focused primarily on using Netlogo to increase our knowledge of emergence. In relation to the Emergence Course, Linked explores the idea of complexity by mentioning that complexity cannot be equated with randomness as it is under random network theory. Barabasi states that in complex systems components can fit in so many different ways that it would take a billion years for us to try them all. This reminds me a lot of our question of limits and our understanding of certain things such as cellular automata,in which changing the size of the world resulted in different outcomes. Being very zoomed in sometimes did not allow for us to see the bigger picture or pattern. Luckily, we have technology to be able to play with issues like this. Just as we have learned, Barabasi acknowledges that advances in technology have allowed us to gain a better understanding of networks to a much greater extent than was possible years ago.

            Perhaps what I found to be the most fascinating about Linked is the fact that it touches upon many of the important points that we posed at the beginning of the course such as the following basic ideas:

• Distributed systems, interactions as critical as component properties, and frequently bidirectional
• No "conductor" nor other external director (see Game of Life)
•  No "architect" nor other external designer (see Game of Life) Self organization
• Even simple distributed systems may require simulation to determine what properties they will exhibit
• Randomness may play an important role in both exhibited properties and in the evolution of systems
• Computer modeling as a new microscope/telescope, allowing one to see things impossible to see otherwise

I also enjoyed reading this book and found it to be very useful for thinking about emergence because it examines the following motivations for our course:

• Similar patterns/properties exist in systems at quite different scales and having quite different constituents
• It frequently proves difficult to account for the patterns/properties of systems in terms of the properties of the individual constituents alone
• Many systems seem to display patterns/properties despite significant disturbance of individual constituents
• Many systems display variations in patterns/properties in lieu of external causes (cf traffic jams) or internal directors (cf slime molds and flocking). The search for either may lead one down blind research alleys.
• Many things seem, on exploration, to be more complex than they "need to be"

In conclusion, Linked is an excellent book that explores networks. Reading this book has just like the course of Emergence, shaped the way in which I view everything. In the end, I realize that we make things more complicated than they need to be and that what seems complex can alternatively be viewed as many things interacting in simple ways in different levels giving rise to what we perceive as complex. Networks are self-organized and this is type f organization is something that we have explored in quite some depth in trying to understand emerging properties. We live in a small world and from this book I am reminded of the fact that most events are connected, cause by, and interacting with a huge number of other pieces of a complex universal puzzle. We do not have all the answers to all of the questions on emergence or networks, but we do have a greater understanding of networks than we did before the Internet. I definitely recommend the book to anyone who is intrigued by this idea of networks and the element of interconnectedness.