Why Networks Matter I am thinking about something much more important than bombs. I am thinking about computers. --John von Neumann Network design is destiny. Networks are the organizing framework that enables billions of people to intelligibly interact. They decide the world''s winners and losers. Their algorithms decide where money and attention will flow. The structure of a network guides how that network will evolve and where wealth and power accumulate. Given the scale of the internet today, software design decisions up front, regardless of how seemingly small, can have cascading downstream consequences.
Who controls a given network is the central question when analyzing power on the internet. This is why critics who knock the tech startup industry for placing more emphasis on the digital world than the physical world--on "bits" versus "atoms"--miss the mark. The internet''s influence extends far beyond the digital realm. It intersects, permeates, and shapes large-scale social and economic landscapes. Even pro-tech investors play up the idea. As Peter Thiel, the venture capitalist and PayPal co-founder, once mused, "We wanted flying cars, instead we got 140 characters." The dig takes aim at Twitter, which originally limited tweets to 140 characters, but it''s intended to pan the perceived frivolity of the software-obsessed tech industry at large. Tweets may seem frivolous, but they affect everything from personal thoughts and opinions to the outcomes of elections and pandemics.
People who claim technologists aren''t focusing enough on problems like energy, food, transportation, and housing overlook that the digital and the physical worlds are interconnected and entwined. Internet networks mediate most people''s interactions with the "real world." The merging of the physical and the digital happens discreetly. Science fiction sometimes portrays automation as a visible process, where one physical thing gets replaced, one for one, by another as a direct substitution. In reality, most automation happens indirectly, where physical objects transmute into digital networks. Robo-travel agents didn''t replace human travel agents. Rather, search engines and travel websites absorbed their tasks. Mail rooms and postboxes still exist, but they handle far lower volumes of correspondence since the rise of email.
Personal aircraft haven''t upended physical transportation, but internet services like videoconferencing have, in many cases, obviated the need for travel. We wanted flying cars, instead we got Zoom. People tend to underestimate the digital world due to the internet''s newness. Consider the language people use. Subordinating prefixes like "e-" in "email" and "e-commerce" diminish digital activities'' value as compared with their "real world" counterparts of "mail" and "commerce." Yet, increasingly, mail is email and commerce is e-commerce. When people refer to the physical world as the real world, they fail to appreciate where they spend more and more of their time. Innovations like social media that were initially dismissed as nonserious can now shape everything from global politics, business, and culture to the worldview of any one person.
New technologies will further fuse the digital and the physical worlds. Artificial intelligence will make computers vastly smarter. Virtual and augmented reality headsets will enhance digital experiences, making them more immersive. Internet-connected computers embedded in objects and places--also called Internet of Things devices--will permeate our environments. Everything around us will have sensors to understand the world as well as actuators to alter it. All of this will be mediated through internet networks. So yes, networks matter. At their most basic level, networks are lists of connections between people or things.
Online, they often catalog what people might direct their attention toward. They also inform algorithms that further curate attention. If you visit your social media feeds, algorithms churn up all manner of content and advertisements based on your presumed interests. "Likes" on media networks and ratings on marketplaces direct the flow of ideas, interests, and impulses. Without this curation the internet would be a deluge--unstructured, overwhelming, unusable. The internet economy turbocharges networks. In an industrial economy, corporations accrue power mainly through economies of scope and scale; that is, ways of decreasing production costs. The diminishing marginal cost of producing more steel, cars, pharmaceutical drugs, fizzy sugar water, or whatever other widget lends an advantage to whoever owns and invests in the means of production.
On the internet, the marginal costs of distribution are negligible, so power primarily accrues another way: through network effects. Network effects dictate that the value of a network grows with the addition of each new node, or connection point. Nodes can be telephone lines, transportation hubs like airports, connection-oriented technologies like computers, or even people. Metcalfe''s law, one well-known formulation of the network effect, stipulates that the value of a network grows quadratically, meaning proportional to the number of nodes squared (that is, raising by an exponent of 2). For the mathematically minded, a network with ten nodes would be twenty-five times as valuable as a network with two nodes, while a network with a hundred nodes would be a hundred times as valuable as one with ten nodes, and so on. The law takes its name from Robert Metcalfe, a co-creator of Ethernet and the electronics maker 3Com who popularized the idea in the 1980s. Because not all network connections may be equally useful, some argue for variations to the law. In 1999, David Reed, another computer scientist, put forward his own self-named spin: Reed''s law, which states that the value of large networks can scale exponentially with the size of the network.
The formula best applies to social networks, where people are the nodes. Facebook has nearly 3 billion monthly active users. According to Reed''s law, that means Facebook''s network value is 2 to the 3 billionth power--a number so eye-blisteringly large that it would take 3 million pages just to print it. Whichever approximation of network value you prefer, one thing is clear: the numbers get big, fast. It makes sense that network effects would dominate the internet, the ultimate network of networks. People cluster around other people. Services such as Twitter, Instagram, and TikTok are valuable because hundreds of millions of people use them. The same is true of many networks that make up the internet.
The more people exchange ideas on the web, the richer that information network. The more people message over email and WhatsApp, the more relevant these communication networks. The more people conduct business across Venmo, Square, Uber, and Amazon, the more valuable these marketplaces. As a rule: more people, more value. Network effects take small advantages and snowball them into avalanches. When corporations are in control, they tend to guard their advantages jealously, making it difficult for anyone to leave. If you build an audience on a corporate network, leaving means forfeiting your audience, so you''re discouraged from doing so. This partly explains why power has consolidated into the hands of a few large tech companies.
If this trend continues, the internet could end up even more centralized, commandeered by powerful intermediaries that use their might to crowd out innovation and creativity. Left unchecked, this will lead to economic stasis, homogeneity, unproductivity, and inequality. Some policymakers seek to defang the largest internet companies with regulation. Their remedies include blocking acquisition attempts and proposing to split companies into parts. Other regulatory proposals require companies to interoperate, allowing easy integrations between networks. Users could then bring their connections wherever they like, and they could read and post content across networks according to their preferences. Some of these proposals could rein in incumbents and make room for competitors, but the best long-term solution is to build new networks from the ground up that won''t lead to concentrations of power for the simple reason that they can''t. Many well-funded startups are trying to build new corporate networks.
If they succeed, they''ll inevitably re-create the same problems with today''s large corporate networks. What we need are new challengers that can win in the market against corporate networks but provide greater societal benefits. Specifically, we need networks that provide benefits like those afforded by the open and permissionless protocol networks that characterized the early internet.