07 · Political Philosophy

This chapter sits in Part I, the Court of Philosophy — the seventh hearing, where philosophy asks the machine who is actually governing when its code sets the terms.

Who governs when code is law?

On 16 August 2017, days after the violence in Charlottesville, Cloudflare CEO Matthew Prince terminated the account of The Daily Stormer, a neo-Nazi website, cutting off the DNS and content-delivery service that kept it reachable on the open internet. In an internal email later published by Gizmodo, Prince described his own reasoning with unusual honesty: "Literally, I woke up in a bad mood and decided someone shouldn't be allowed on the Internet," he wrote. "No one should have that power" (Gizmodo, "Cloudflare CEO on Terminating Service to Neo-Nazi Site," 2017). No court had ordered the takedown. No legislature had voted on it. One executive, at one infrastructure company, flipped a switch, and a publication that had survived years of public outrage disappeared from the reachable internet within hours. Prince was right that no one should have that power — and right, too, that in that moment, he had it.

Lawrence Lessig had already named the mechanism a decade earlier. Code and Other Laws of Cyberspace (1999, revised as Code 2.0 in 2006) argues that four forces regulate behavior — law, social norms, the market, and architecture — and that in digital environments, architecture usually wins, because it does not ask for compliance, it makes the disfavored action impossible rather than illegal (Lessig, Code and Other Laws of Cyberspace). A law against a website can be appealed, delayed, ignored pending review. A DNS provider withdrawing service is not a rule about the website; it is the website's disappearance, instantly and everywhere, the moment the change propagates. Cloudflare's decision is Lessig's thesis in its purest form: code did not enforce the law here, code was the only law that mattered, and it was written by a company's terms of service rather than by any legislature.

Every major platform now runs something that functions like a legal system without being one. Kate Klonick's "The New Governors" describes Facebook, Twitter, and YouTube as private jurisdictions complete with legislation (terms of service), adjudication (content moderation review), and — since October 2020 — an appellate layer: Facebook's Oversight Board, a $130-million-endowed panel of twenty members whose rulings on content decisions are contractually binding on the company, and which overturned four of the first five decisions it reviewed (Klonick, "The New Governors," Harvard Law Review 131, 2018). It is jurisprudence's full apparatus, minus the one thing that ordinarily justifies it: no user elected the legislature, and the constitution can be rewritten unilaterally by the company that wrote it the first time. Prince's discomfort with his own power was, in effect, an admission that he had briefly become one of Klonick's "new governors" without any of the checks a real government answers to.

The deeper stake, Shoshana Zuboff argues, is not who moderates speech but who extracts and owns the "behavioral surplus" generated by everyone's use of these systems — the data left over once a service has been rendered, repurposed to predict and shape future behavior and sold as a prediction product to advertisers and beyond. The Age of Surveillance Capitalism (2019) contends this is a genuinely new form of power, distinct from the industrial capitalism that classical liberal theory built its protections against, because it operates through unilateral claims on experience that the subject never consented to and typically cannot see, let alone exit (Zuboff, The Age of Surveillance Capitalism, PublicAffairs, 2019). Contract law and privacy torts were built to police visible transactions between two parties who both know an exchange occurred; they were not built for a form of extraction whose whole design depends on the other party not noticing.

Not every digital institution has defaulted to sovereign-or-corporate. Elinor Ostrom's Governing the Commons (1990) identified design principles by which real communities manage shared resources without either a state's coercion or a market's pricing — clearly bounded membership, rules matched to local conditions, graduated sanctions, cheap and accessible conflict resolution, and governance nested across scales — overturning the assumption, inherited from Garrett Hardin, that shared resources are doomed to overuse absent private ownership or state control (Ostrom, Governing the Commons, 1990). Ostrom and Charlotte Hess later extended the same principles explicitly to knowledge and digital commons, arguing that open-source projects, Wikipedia, and shared research infrastructure could be read and governed the same way (Hess & Ostrom, eds., Understanding Knowledge as a Commons, MIT Press, 2007). Linux and Wikipedia are the proof this third path is not merely theoretical: neither a state nor a corporation, and yet governed well enough to outlast most of both.

When a state itself delegates to code, the failure mode changes shape entirely. Danielle Citron's "Technological Due Process" (2008) warns that automated systems collapse two functions administrative law has always kept separate — writing the general rule and applying it to one person's case — into a single, undocumented act, evading the notice, hearing, and reasoned explanation either function alone would owe a citizen (Citron, "Technological Due Process," Washington University Law Review 85, 2008). Michigan's unemployment agency ran exactly this experiment between 2013 and 2015: its MiDAS system flagged fraud automatically, ran start-to-finish without human review in most cases, and falsely accused more than 34,000 people, at an error rate later estimated at 85 percent, triggering wage garnishments and bankruptcies (IEEE Spectrum, "Michigan's MiDAS Unemployment System"). No rulemaking notice, no individual hearing, no appeal built in — due process was not violated by any one decision; it had been designed out of the system before the first case was ever flagged.

Benjamin Bratton's The Stack (2015) offers the frame that holds Cloudflare, Facebook, Zuboff's harvesters, and Michigan's benefits algorithm as one phenomenon rather than four unrelated scandals: planetary computation — cloud, city, address, interface, user, layered as a single accidental megastructure — now performs sovereign functions (deciding who may speak, who receives a public benefit, who is reachable at all) continuously and at a scale no state apparatus can match, without ever standing for election or answering to a constitution (Bratton, The Stack, MIT Press, 2015). The state has not been abolished by the stack. It has been joined by a second, unelected one, and most days it is not obvious which one actually decided.

Connections

Chapter 06 owns the responsibility gap this chapter's automated systems keep opening, asking who is accountable rather than who governs. Chapter 11 explains the bracket that lets "behavioral surplus" be priced as a quantity while what it is of goes unasked. Chapter 14 picks up Ostrom's commons again through consensus protocols — cooperation among strangers with no sovereign to enforce it.

Open questions

Is "code is law" a warning about what has already happened, or a program for what should happen next — and does confusing the two let more power slide into architecture than anyone actually voted for? Can a private platform hold due-process-like authority legitimately, short of becoming a state in miniature complete with the checks that make states bearable? And does Ostrom's commons model generalize to physical infrastructure — cables, satellites, spectrum — the way it generalized from irrigation to code, or does infrastructure's need for capital always pull governance back toward a state or a firm?

Go deeper

• Lawrence Lessig, Code: Version 2.0, Basic Books, 2006 — the revised edition of the argument this chapter's opening case runs on rails built by.
• Kate Klonick, "The New Governors: The People, Rules, and Processes Governing Online Speech," Harvard Law Review 131 (2018), read against Elinor Ostrom & Charlotte Hess, eds., Understanding Knowledge as a Commons: From Theory to Practice, MIT Press, 2007 — private jurisprudence against the commons it displaced.
• Stanford Encyclopedia of Philosophy, "Political Legitimacy" — the standing account of what a governing power owes the governed, against which every case in this chapter can be measured.