The things we call “technologies” are ways of building order in our world.
Langdon Winner, “Do Artifacts Have Politics?”, 1980
In the autumn of 2024 I sat with a draft rule from one European jurisdiction that would have required the non-custodial rail I was building to hold a license it structurally cannot hold. The rule did not name us. It named a category, and the category assumed a custodian somewhere inside the flow: a party that could be licensed, audited, sanctioned, served. The rail has no such party. If finalized, the rule would change not the rail but whether a merchant in that jurisdiction could plug into it. I worked through the available moves. We could write comments, sign a coalition letter, fund an amicus brief at the court that would eventually hear it. But the vote that mattered had happened at a committee meeting two years earlier, in a room where the category had been defined. By the time the draft reached public comment, the architectural assumption was already in the language. The ballot was still on the wall. It had never reached the room where the category got its shape.
That is the asymmetry. A regulator is a person sitting in an office inside a country. The country has borders, the office has a phone, and the regulator will be replaced after the next election. A protocol is a set of rules running on machines that do not know where they are or what year it is. Bitcoin does not know it is in El Salvador or in France or in North Korea. Email does not know whose country it is leaving. TCP does not file taxes.
Democracy is anchored to place and to the four-year cycle; the software runs everywhere, on a timescale that outlasts any administration. That much is old. What is new is the scale at which it matters. Who can transact, what gets remembered, what the AI says, what the money does when you try to spend it: these used to be political questions, and they are now protocol questions, which the ballot cannot structurally reach. Even uncaptured and sharp, the corrective institutions from The Capture of the Corrective Institutions would still be bounded by jurisdiction on both axes. Lessig named the four modalities in 1999, law, norms, market, and code, and warned in the 2006 revision that code would become the dominant one. Code does not pass through a legislature. Architecture is eating jurisdiction.
The Spatial Axis
Jurisdiction is the legal concept that matches a rule to a place. A court in California cannot jail a person standing in Tokyo. It works when there is a corporation to subpoena and fails when there is only a protocol. Google, Apple, and TikTok have a headquarters, a legal entity, an officer to name. Bitcoin, BitTorrent, and Tor have been reached for by the same regulators for a decade, and there is nothing to grip. The extraterritorial mechanisms the modern state has built, FATCA, GDPR, the CLOUD Act, all require a target that has a place, and a protocol does not have one.
Pressure can still reach the shoulders where the protocol meets the world. The Tornado Cash sanctions reached developers and code repositories; the Storm prosecution reached a person. The lesson is not that protocols are untouchable. It is that the touch lands on services and on people, the reachable surfaces, and passes straight through the protocol itself. The money kept moving. The chain kept confirming. Nor should the protocol side be romanticized into something outside politics: protocols have forks, maintainers, economic consensus, long fights over what the rules should be. “Cannot be erased” is not the same as “outside political reach.” What the gap describes is narrower and harder. The rules that win those fights run everywhere at once, and no single jurisdiction gets a veto.
The Temporal Axis
Jurisdiction has a second axis: time as well as place. Democratic cycles are short by design, on the assumption that the decisions that matter happen inside the cycle. When the decisions that matter compound over decades, the cycle stops being a correction and becomes a blindfold.
Monetary policy is the clearest example. A central bank that expands its balance sheet by trillions in a crisis has not made a four-year decision. It has made a twenty-year one, because the effects (asset-price inflation, generational wealth transfer, debt service on the next sovereign) compound on a timescale no election touches. The administration that chose is retired before the consequences arrive, and the correction, when it comes, comes not as a democratic act but as a collapse, blamed on whoever holds the lever when the structure gives way. The ballot reaches four years out. The decision reaches forty. The consequences reach the grandchildren. A protocol built for permanence is out of jurisdiction across place and across time: Bitcoin’s monetary schedule is fixed for roughly 130 years, and no election can shorten it.
Two Lenses
Two books saw both axes coming. The Sovereign Individual (Davidson and Rees-Mogg, 1997) predicted that information technology would shrink what the state could reach even as its ambitions grew; Bitcoin and end-to-end encryption have aged the prediction into the present. Principles for Dealing with the Changing World Order (Dalio, 2021) traces five centuries of reserve-currency cycles that no administration ever sees whole, because each generation votes on the politics visible to it and never on the cycle. Neither is prophecy. What they share is a diagnosis: the systems we live inside run on scales the ballot does not reach. The Ring, in Tolkien’s grammar, is older than the king who happens to be holding it, and the response is not to put it on a better finger. It is to build something the Ring does not fit inside of.
Democracy is not failing because voters are voting wrong. It is running into the wall of its own scope on two axes at once. You cannot vote in a protocol, only fork one, and you cannot vote against a cycle that takes eighty years to complete from inside a four-year term.
This is an architectural diagnosis, not an indictment of democracy. The lever has a reach, and the reach stops at the border and at the end of the term. The software does not stop at either. The cycle does not stop at either. What happens in that gap is already happening, and no one voted for it.