Overlapping Jurisdictions Are a Feature
Page 10 spends a lot of its length worrying about one state’s rules binding another state’s people. I want to put that worry in a frame that makes it look less exotic, and then be careful about how much the frame actually justifies.
I already live like this
I am a dual EU/US citizen resident in Finland. Finnish law governs most of my daily life. American law follows me regardless, because the United States taxes its citizens wherever they live. EU law sits above the Finnish layer. NATO commits a set of other countries to my defence and commits Finland to theirs. Four overlapping authorities, none of which I negotiated with individually, all of which bind me simultaneously.
On net this is enormously to my benefit. It is also not costless, and page 10 has the specific bill : the interaction of EU consumer-protection rules with American tax law means I cannot own an ordinary index fund. That is the overlap malfunctioning, and it malfunctions in a way nobody intended and no treaty anticipated.
Both of those are the same structural feature. Overlapping jurisdiction produces NATO and it produces the PRIIPs vice, and you do not get to keep one without risking the other.
This has a literature and it is not the one you would guess
The idea that legal order can come from several overlapping providers rather than one territorial monopoly is old and reasonably well developed. Edward Stringham’s Anarchy and the Law collects the serious version of the debate, including contributions from Douglass North, Barry Weingast, Tyler Cowen and Robert Ellickson, which should indicate it is not purely a fringe enthusiasm. Elinor Ostrom’s work on polycentric governance won a Nobel prize. And at the playful end there is Polystate , Zach Weinersmith’s thought experiment about non-geographic states you can switch between annually — which is Weinersmith’s, not Bryan Caplan’s, though the adjacency of the crowd makes that an easy slip.
The strand I find most useful is not the liberty argument, though. It is the one about commitment.
The part that actually transfers
There is an intuition that the equilibrium for well-functioning government shifted at some point from monarchy to democracy — that something changed which made the second arrangement outcompete the first. The most defensible version of that story is North and Weingast’s account of seventeenth-century England, and its mechanism is narrower and more interesting than “representation is nicer.”
Their argument is that the constitutional settlement after 1688 worked by making the Crown less able to act. Binding the monarch to Parliament made royal promises credible, which meant creditors would lend at rates an absolutist sovereign could not obtain, which meant the state could finance wars its rivals could not. The institution that tied its own hands beat the institution that kept them free.
That is the lesson worth carrying over, and it is not about democracy at all: an institution that can credibly bind itself outcompetes one that can only promise. A frontier lab issuing a safety framework is making a royal promise. A frontier lab whose researchers carry priced liability has done the 1688 thing — accepted a binding it cannot quietly revoke, and become able to be believed as a result. The premium is the bond, and being believed is worth more than the freedom that was given up. Page 08 is that argument in its own terms.
There is a historical rhyme worth noting too. The borrowed-forum machinery page 10 leans on — arbitration, neutral venues, awards enforceable across borders — is descended from the medieval law merchant, which existed because merchants trading across jurisdictions needed adjudication that no single sovereign could supply. Private cross-border dispute resolution is not a novelty invented for this proposal. It is one of the oldest working examples of the thing.
Where I part company
I should be careful, because there is a version of this note that borrows conclusions it has not earned.
The attractive version is voluntary and mine is not. What makes polycentric legal order appealing in the anarcho-capitalist literature is choice — you select your protection agency, and competition disciplines it. Nothing about my situation is chosen. I did not opt into citizenship-based taxation and could only escape it by renouncing a citizenship. A researcher subject to a bounty statute has not opted into anything either. This proposal is polycentric in the descriptive sense and not in the consent sense, and the consent is where most of the theory’s moral force lives. I am not entitled to the warm feelings.
Michael Huemer’s position is stronger than mine and points elsewhere. His work denies that states have the legitimate authority they claim. This site asks a state to create a new statutory penalty and enforce it internationally. I cite him on the vote because he is good on when inaction is justified, not because I can claim his conclusions.
My overlaps are between liberal polities. Finnish, EU, American and NATO authority overlap tolerably because all four operate under recognisable rule of law. The frame is far less reassuring when one of the overlapping authorities is not, which is exactly the reservation page 10 records about what “lawful reach” is doing in a sentence. An argument that overlapping jurisdiction is fine because mine is fine is an argument from an unusually comfortable position.
What the frame is actually for
Not justification. Reframing the size of the ask.
“International AI governance” sounds like it requires a treaty organisation, a world regulator, and a level of agreement that has never existed for anything. What this proposal actually requires is one more overlapping layer in a world that is already thoroughly layered — of a kind that hundreds of millions of people live under without remarking on it, and which mostly works. That is a much smaller thing than the phrase suggests, and noticing it is the difference between an impossible demand and a difficult one.