The story so far

The short version

The dynamics that drive Internet fragmentation are fundamentally the same as those that drive the current backlash around “big tech:” state and non-state actors are competing for control of the Internet.

The question becomes: how can we observe these conflicts as they play out? As geographic borders are sometimes contested (think: China and India; Syria and Rojava), boundaries on the global Internet are contested too. The problem is that, for the Internet, we don't have great ways of seeing who controls what, or how “lines” of control are changing.

Our goal with the Internet Atlas Project is to get a better sense of all that: to create a “political geography” of the Internet.

The long version

You probably know that states are competing via the Internet. Fake news. State-backed hacking. Ransomware. Now maybe deepfakes.1 But there’s another conflict brewing. States (and others) are competing for control over the Internet itself.

The stakes are enormous. The conflict itself could cause wide-scale disruption, and once it ends—if it ends—the Internet we’re left with could be fragmented into national siloes. Or it could come out global—and globally censored. The range of possible futures is wide.

But wait—what does it mean to have “control of the Internet?” What does state power look like on the Internet?

The answer turns out to be complex. Here’s one observation: the Internet is so highly centralized that the US could effectively turn it off. This begs the question: if the US has so much power, why is the Internet as open as it is? One explanation—one I'm not sure I believe, but which I haven't heard anyone else propose—is that the Internet may be open because a hegemon dominates it. This runs contrary to the received wisdom that the Internet is at its most stable when it’s decentralized. In fact, some forms of decentralization have come at the expense of interoperability, creating measurable Internet fragmentation.

The idea here is to understand how state power works online in as much detail as I can manage. With that understanding, we could make better sense of, and perhaps even help to manage this conflict, helping states use data (rather than anecdote) to move the Internet toward shared goals.

What makes a state online?

But states—as we understand them—aren’t the only ones competing for power over the Internet. While they may not openly declare themselves challengers to state power, large multinational corporations are quietly challenging traditional states online.2 Where “Internet fragmentation” describes the current state of conflict between states on the Internet, “techlash” describes the current state of conflict between traditional and emergent ones. Traditional states, which exert an effective monopoly in terrestrial space, are facing serious power challenges on the Internet.

Here’s one example. Along with running Internet infrastructures like servers, domain names, governments blocking websites, one way to exert power on the Internet is to manage identity systems. Authentication systems, codified as passwords and usernames and SSNs and KYC rules, create governable persons. They’re an under-appreciated mechanism by which states exert power online, and a domain in which non-state actors are very actively competing with states for influence.3 As cryptographic personhood emerges, and as states and other actors compete to provide robust identity systems competitive or equivalent with national systems like passports, identity is emerging as a new front in the Internet power struggle.4

States of knowing

The idea here is: tech companies are states. My core thesis is that any entity that provisions identity systems and performs (what I call) the ‘epistemic roles of the stateis a state. I call them states of knowing. These states have a different relationship to territorial space than do territorial states. In the parlance of war, they’re disadvantaged in the land, air, and sea domains—but have an advantage in the cyber domain.

As new kinds of online states emerge to challenge existing states' power, new questions arise:

  • If tech companies are our Louis XVI, what does our French Revolution look like?

  • If the battle between traditional states and emergent online states is our Thirty Years War, what is our Peace of Westphalia?

At least, that’s the plan for now.


Meanwhile, advances in deep learning are about to change the nature of cybersecurity and design work dramatically. Deep learning is already changing the nature of state influence online—think of deepfakes. It may exacerbate conflict as states (and others) apply a fragmented patchwork of remediation strategies. And remember: the Internet is the substrate on which modern deep learning collects its data and reaches its users.


Less quietly, cryptocurrencies are loudly declaring themselves challengers to state power. Both new and emergent states are actively experimenting with blockchains.


This perspective tilts a seemingly “traditional” topic in cybersecurity—authentication (think: logging into things)—in a different direction.


Some states are adopting the specific designs that tech companies have created for identity systems. Estonia’s and Canada’s come to mind, as does India’s biometric identity system and Pakistan’s genealogical identity system.