The Science of Networks
There is a cartoon about social networking—a funeral of a man with just five people in attendance, and somebody saying, “ … but I don’t understand, he had over 2000 Facebook friends!” It’s funny when you see it through the eyes that believe a true friend, somebody who will come to your funeral, is worth more than hundreds of virtual, online, Facebook or effectively fake friends. I have an opposite view. I think it’s pretty amazing that a person who has five friends at his funeral would have had 2,000 people who he knew and could connect with.
On another note, my friends are split between those who publish their birthdays on social media and those who expect their true friends to remember them. Since there’s little chance of my remembering a thousand birthdays, I’m likely to depend on a diary or computer somewhere anyway, so it might as well be on Facebook as far as I’m concerned.
In both of these examples, we’re talking about social networks and comparing strong and weak ties within networks in some ways. It is now broadly understood that both types of ties matter and bring different values. Strong ties offer psychosocial support—people you can turn to at times of stress or trouble. People who have the time for you. People who will remember your birthday and attend your funeral. In a professional sense, these could be your organizational allies, or the partner you’ve worked with for years, the head hunter who got you your last few jobs, or your former boss who mentors you. But there is another type of bond altogether—a loose tie, which also fills our social networks. The people who just know you well enough to wish you on Facebook, but won’t attend your funeral. Or those who know you professionally but not well enough to follow your every move. Mark Granovetter’s seminal work on The Strength of Weak Ties1 changed our view of networks. He argued that in a number of areas, including marketing, politics, and other areas of persuasion such as getting a job, weak ties made a bigger contribution than strong ties. Strong ties usually form between like-minded people who share similar interests and often the same friends. By definition, therefore, the spread of new and radical ideas requires connections to people who aren’t already converted—that is, people from a different group or a weak connect. Moreover, because these weak connections tend to be from different clusters, they create bridges between cliques in a network. Your chances of getting a job or getting widespread traction for a new idea are higher via the weak ties in your networks who may have access to new and different opportunities and people, rather than your strong ties who may move in roughly similar circles as you and may not have new opportunities and ideas beyond what you know yourself.
It may surprise you to note that Granovetter’s work on Weak Ties predates Facebook or LinkedIn by over three decades. On the other hand, should it? Human social networks have existed from the days of cave dwellings of the hunter-gatherers hundreds of thousand years ago, and these dynamics have been largely true since then. Networks predate human beings as well. aspects of network theory, a more formalized branch of science in the past few decades, have been found to be true across the animal kingdom, evolutionary science, and even in microorganisms and individual cells. Albert-Laszlo Barabasi’s work on network theory2 started with the World Wide Web, but eventually encompassed biological networks in cells, physical networks in state transitions of substances as they moved from liquid to solid or vice versa, and many other human social networks, including the professional network of Hollywood films.
You may be familiar with the idea of the six degrees of separation—part parlor game, part myth—which looks to establish connections to the actor Kevin Bacon from any other actor, in less than six links. Although Kevin Bacon is irrevocably connected to this game, it was only a quirk of fate. Using now freely available data about movies and actors from IMDB, it’s possible to actually analyze the Hollywood actors network. When Barabasi and his colleagues did the analysis, they found Kevin Bacon only ranks 876th in the six-degrees game—he had an average separation of 2.79 from any other actor. On the top spot, Rod Steiger had an average separation of 2.53, with Christopher Lee and Martin Sheen following in the podium places. This idea of the average distance is one metric that can be used to understand networks. Networks have a clustering coefficient, which shows the denseness of networks. A denser network may have higher clustering coefficient, and correspondingly, a lower average distance between any two nodes. Networks also have size, which is the number of nodes in the network. Taken together you could start to understand the behavior of a network. Information would travel faster through a densely connected network and the speed at which it reaches the end would depend on the density and size. Think of gossip in a well-networked small town, where news of an affair might travel overnight. Whereas if you take the entire political system as a network that is much bigger and much less densely interconnected—news of a scandal may travel at varying speeds and take much longer to reach everybody.
Scale-free networks are networks where the distribution of connections between nodes is not just uneven, they are strongly skewed. Rather than a random distribution (which you might represent as a bell curve), they are asymptotic curves with a few nodes at a very high number and many in a long tail. Imagine the income distribution of all people in any capitalistic society arranged by levels of income. You’ll have a very long tail of people with low to medium incomes and a very high spike at one end where a very small number of people enjoy incredible wealth. A much-cited recent report by Oxfam reported that the eight richest people in the world have the same cumulative wealth as the bottom 50 percent of the population. If you plotted the world’s income distribution, therefore, you would see a scale free network, which displays this kind of power law distribution. There is a mathematical expression of the power law, and these kinds of power laws are exhibited all around us—in biological cells, social networks such as Facebook, and professional networks such as Hollywood. They are ubiquitous, and we need to understand them better.
Aether was the old word for the stuff that fills the sky, and over time, it came to mean air, as in vanished into the ether. Information that could travel at high speeds through the air was always a sought-after fantasy of scientists, so Robert Metcalfe was probably jumping the gun when he named his new invention Ethernet on May 22, 1973. What he meant by ether was actually coaxial cables, which would carry packets of information in a network. Nonetheless, Metcalfe’s Ethernet would go on to become the dominant model for networking between computers for the next few decades. If you’ve used a corporate LAN or WAN, you’ve probably used an Ethernet network. Over the 1980s and 1990s, corporate networks across the world adopted Ethernet networks at scale, and as we speak, this remains close to a $20 billion annual market.
Much of the Internet sits on top of Ethernet networks—by using TCP/IP protocol as a layer over Ethernet, which is the physical layer. The Internet connections don’t actually care what the physical layer is. They work over wireless, and wired connections and dramatically expand the reach of networks beyond LAN and WAN, to a universal and publicly accessible network. The Internet needs little introduction, but it isn’t the only network we use today, by far. We typically use Wi-Fi networks in our houses—which connect to the Internet but also increasingly connect our TVs, entertainment systems, music systems, computers, phones, and other devices such as Amazon Echo and Ring Doorbells. The cars we drive today include networks, which connect the many onboard computers to each other and to the outside world. Effectively, we move from network to network as we go from home to car to work. City centers and public places are also increasingly connected to the Internet via Wi-Fi networks, or we are forced to use more expensive mobile (3G and 4G) networks. One way or another, we are permanently connected in these layered networks. Which means information can flow through these networks in ever more efficient and creative ways.
The human body is a network too, our brains are seen as one of the most sophisticated neural networks, and something we are only recently able to model. As medicine embraces nanotechnology and we start to connect the body through implants and measuring tools, we will make ourselves into a more explicit network as well, and become another extension of the network of networks that our lives are increasingly starting to represent. As you can see, while for much of human history our social networks has been abstract concepts, the last 50 years have seen a physical manifestation of our social networks that has grown in unimaginable ways. And riding on the physical network is the increasingly sophisticated information, social, and professional networks, which make our worlds work.
Understanding the Value of Networks: Looking at the World Differently
Understanding how networks work therefore may be key to our being able to best use and exploit the networked world we find ourselves living in. Thanks to the hyper-connected nature of the world, we can now look at a lot of our major infrastructure and industrial systems as networks. For example, instead of the mechanical view of the asset intensive environments and moving parts in transport, we could look at our transport system as a network with nodes, links, and behaviors. This would allow us to better model how information flows through these networks. In any case, the interplay of all the elements of this network make it far too complex to model using traditional structural means.
We are seeing some of the network effects playing out today. One of the ideas Robert Metcalfe is also credited with is the eponymous Metcalfe’s law, which states that the value of a (telecommunication) network is proportional to the square of the number of connected users of the system. Putting aside the exact mathematical equation, which has been questioned and modified, this is intuitively easy to understand. The first person to have a telephone has a useless device as she can’t call anybody. The value jumps as a second and third person get phones. The more nodes on a network the more valuable it becomes. Today’s autonomous cars struggle because they also have to account for the random and unpredictable behavior of human-driven vehicles on the road. The value of autonomous vehicles goes up the more autonomous vehicles there are on the road—they become safer, they exchange information, and learn collectively and they update each other based on information received from their environment.
In their excellent 2001 HBR article Where Value Lives in a Networked World, Mohanbir Sawhney and Deval Parikh identified the idea of network intelligence as something to be harnessed by decision makers in complex environments. They argue that the value of the network is that the intelligence and tasks do not have to be colocated on a network. Intelligence can be pooled and centralized or decentralized as required. The way client server systems work or browsers work today is very indicative of this. You have the distributed, fragmented frontend intelligence that deals with users in each click, touch, and transaction. And you have the aggregated and centralized intelligence at the back end. In Sawhney’s words, the middle bits are the dumb conduits.
Malcom Gladwell talks about nodes or people who become hubs in a network3—they have the maximum number of connections, and he also talks about mavens, who attract information and knowledge. Connectors and mavens can only exist in networks, by definition. You might see plenty of this in the startup network in Silicon Valley or even inside your business. Usually, some people become very good at being connectors—they have been around longer and know everybody in the business. They have a lot of social capital and can be very powerful to any project or initiative in spreading ideas. You also have the mavens who everybody goes to, to discuss new ideas or information.
One of the great things about networks is their resilience. In a network, thanks to the presence of multiple and therefore individually redundant connections, the loss of a node usually does not significantly slow down the performance of the network. A server or a computer going down on the Internet would barely be noticed by a systems administrator today, because everything is usually backed up and distributed with built-in redundancy. The Internet’s underlying network architecture was in part a response to a need for a highly resilient system that could not be brought down by simply bombing a network brain, in a war. This kind of network intelligence that functions without a brain is plentiful in nature. You see it in the working of many species, including ant and bee colonies. You can see it work at breathtaking speed and motion in the murmurations of a starling cloud. Or you can see it in the glacially slow formations of slime mold, which are single cell organisms. The starling murmurations work in a way that each starling’s movements only impact the seven starlings closest to itself, but this is transmitted quickly. Because of this, any reaction to external stimulus such as a bird of prey can spread very quickly through a flock of starlings, and the whole cloud can react very quickly, but almost as an intelligent unit independent of the size of the flock. Slime mold naturally forms a line toward food, but despite no brain or advanced communication capability, it can collectively find the most efficient way out of a maze or replicate railway systems. These are both intelligent and resilient because removing one part of the slime or any starling from the flock will not change the behavior or sophistication of the network.
Thanks to the hyper-connectivity we’re seeing across the world, and the flow of abundant data, it would not be wrong to say businesses increasingly represent complex networks. Rather than terms such as supply chain, we should probably think supply network. In fact, my colleague Frank Diana, a futurist, has researched this at length and he looks at the emergence of ecosystems, rather than industries, in the way that business is increasingly organized. Ecosystems are essentially networks and they follow the laws of networks. This gives us entirely new ways in which to think of optimizing businesses. By opening up APIs, and enabling coupled functionality, technology firms are always looking to become ecosystem hubs and extract network value.
Tip: For any digital project, map out every stakeholder involved in ensuring the output you seek, and how they are interconnected. What binds these entities together and how can you move to the center of the network?
Networks in the Real World
We will see more and more application of network theory and insights across industries and problem domains, as ecosystems proliferate. Whether we are looking at payments networks or health care systems, network properties such as resilience, network intelligence, and the use of network metrics and behaviors will be used to model these industries and solutions rather than the traditional ones we’ve seen historically. Here are four interesting areas to consider.
Blockchain isn’t new anymore. It is technology that is clearly born of the network. In order to make transactions truly robust, blockchain uses the network to get over the two key challenges of fraud—either the computer that is recording a transaction is compromised and under the control of a malicious third party or the miscreant tries to go back and change previous records to show a different version of transactions.
In essence, Blockchain is a distributed ledger where transactions or records are stacked in blocks and then stored in every computer on the network. Moreover, the security is also distributed across the network as each block is encrypted more than once, and finally, it is put through a cryptographic process to which the answer can only be found by trial and error. All the networks in the network try to solve it, and when one does, the solution is verified by others in the network. This element of chance means there is no single computer in charge. Finally, the hash or encrypted value of each such solution is made the identifying string of the next block—so the chain ensures that you can’t tamper with previous blocks without changing all subsequent values. At present, blockchain is ideally suited for high-value transactions, where speed is not essential. Land Registry, tracking valuable jewels, and post-trade settlement are good examples. In future, I’ve no doubt that our identities will be stored in a blockchain-like network solution.
Future Organization Structures
The traditional organization structure has been under fire, as it can often appear to be out of step with the needs of organizations today. By creating a hierarchy, it forces a decision system that is not necessarily in line with its environment. For example, capital allocation is done by a senior financial expert who may not understand how emerging technology is changing the return on capital across range of initiatives. Investment in new disruptive products and solutions may often be derailed by owners of existing products, or senior people whose career is locked into the success of the current products and processes. A number of options have been mooted—most organizations have strived to grow flatter and some have even tried holacracies as a structural principle. The self-organizing principle of holacracy will undoubtedly have many critics and its share of failures. But ultimately, modeling the organization as a resilient network is a very powerful idea. This would allow the network intelligence to play a key role in the behavior of the firm, along with any individual decision makers.
Shifting Value in Broadcasting and Media
For the broadcast media industry, the network value model translates value moving to the two ends of the content network. Companies either need to own the content, or they need to own the consumer. Content owners are those who are making more investments into original content—HBO and the BBC have always been investors in new content, but today, every single company, from Netflix, to Amazon Prime, is following suit. At the other end, those who own the consumer are the ones who have a relationship with the individual consumer. These are the platforms with registered users, or providers with a billing relationship with customers. Set top box providers or OTT platforms all belong here. The value is leaving the middle bit—the world of traditional TV channels, for example—who have historically made money by connecting content providers to platforms. In the UK, Sky’s business model has, in a very large part, been driven by its monopolistic stranglehold on premier league football content. The pure content distribution and packaging model of channels (and magazines) is gone. If you’re going to be in the media business—you either need to own original content or the viewer relationship, or both.
Is there no value in distribution then? Well, tools like Paper.li and Del. icio.us allow the aggregation itself to be democratized. This is basically software-eating media-eating software! Also, players like Business Insider, Outbrain and Bleacher Report, or Quartz are trying hard to build a business that truly exploits the new distribution in different ways. Quartz has been described as an API, as fundamental to their business model is the assumption that people don’t directly visit their website.
Darwin’s Paradox is the phenomenon by which despite occupying under a thousandth of the words surface, the coral reefs support 25 percent of all marine species. A part of the answer lies in the cooperation and symbiotic relationships between species in the reefs. Some algae families that absorb carbon dioxide provide the corals with growth materials in exchange for CO2 and host services, protection, and access to sunlight. This cooperation is now recognized as being just as critical to the evolutionary system as competition.4 So, it is that all the tools and products that use Google’s map API actually make the maps stronger, and often better, by adding data back to it. Or the Twitter ecosystem, which works in a similar way, relies on third parties for many of its sustaining innovations.
Open-Source Software
Since the time of Linus Torvalds, and later Netscape, in the 1990s, the open-source movement has championed the cause of software that can be freely used and distributed. Not just to use, but to modify, and build on. This has led to some of the most commonly used software across the world, especially the Apache Foundation, which has over 227 million lines of code and over $22bn worth of software available to the world at large. This is only possible through the power of the network—in connecting and enabling collaboration across these thousands of software experts across the world and creating shared value across hundreds of problem areas of software development, deployment, and maintenance.
The Dark Side
Sadly, terrorism too, uses similar network efxfects—by creating media storms through acts of barbarism, they look to heighten fear and anxiety. Other network participants such as politicians sometimes use divisive policies, which sustains localized discrimination leading to alienation and strengthening the extremists. The media and politicians are unwitting cooperators in this network. Today’s terrorist networks are rarely command and control organizations. Instead, like platform organizations, they allow individuals and groups around the world to share a common platform for furthering their individual agendas.