Social LibertySplitting the Atom of Sovereignty
No ruler in history ever answered to this many bosses at once. That wasn't an accident. It was the design.
Most of what a government is, historically, is one thing: a single will at the top, however that will gets chosen or replaced. A king, a parliament, a politburo — the label changes, but sovereignty itself stays whole, a single indivisible fact about who ultimately decides. Justice Anthony Kennedy, writing in U.S. Term Limits, Inc. v. Thornton (1995), named the thing the American design did instead, in a phrase worth sitting with: "federalism was our Nation's own discovery. The Framers split the atom of sovereignty."
That's not a metaphor for a compromise. It's a description of an engineering choice, and it's worth treating it as one. This site more often asks what a given layer got wrong in a given fight than what the whole architecture got right in making that fight possible to have at all — so it's worth pausing, for once, on the structure itself, because it is a genuine, still-underappreciated feat: no people in history has ever held this many simultaneously real, separately enforceable claims on its own governance at once.
Four layers, one root
The design starts from a single premise and then deliberately refuses to let that premise collapse into one office. Sovereignty rests, first and only, with the people themselves — "We the People," not as a rhetorical flourish but as the actual legal foundation every other layer has to trace its authority back to. Neither the federal government nor any state has any power of its own; each holds only what the people specifically delegated to it, a point the Tenth Amendment states as bluntly as constitutional text ever states anything.
From that one root, the design splits. The federal layer gets supreme authority over a defined, enumerated list — interstate commerce, foreign policy, currency, national defense — matters that by their nature can't be handled by fifty separate answers. The state layer keeps the residual police power over health, safety, and welfare that the federal list didn't claim, operating as a genuinely independent sphere rather than a federal subsidiary.
And running alongside both, older than either, is a third layer most retellings of this story leave out entirely: tribal sovereignty. Native nations were recognized by Chief Justice Marshall in Cherokee Nation v. Georgia (1831) as "domestic dependent nations" — a status that constrains them in ways full sovereigns aren't constrained, but that never pretends their sovereignty was created by the Constitution rather than predating it.
The framers and the courts that followed them chose, at least in law, not to absorb that layer into the other two. That choice has been honored far more consistently on paper than in practice, and a fully honest accounting has to say so — but the legal architecture itself never stopped treating tribal governments as something other than subject populations, which was not the only available choice and wasn't the easy one.
Three constitutionally real sovereigns, all answering, ultimately, to the same people — that's the atom Kennedy meant, split rather than fused, on purpose.
A design built to bend without breaking
The relationship between the federal and state layers wasn't fixed once and left alone, either, and that turns out to be the more remarkable design decision. Political scientists describe two working models for how the layers interact: dual federalism, the "layer-cake" version, where each sphere stays strictly separate and neither interferes with the other; and cooperative federalism, the "marble-cake" version, where the layers intermingle, share resources, and collaborate across the same policy areas. American federalism has moved between these modes more than once over two and a half centuries, without anyone needing to rewrite the Constitution to make the shift. The boundary between the layers flexes. The fact of the layering doesn't.
That's a rare property in institutional design. Most systems either lock a relationship in place so rigidly that it eventually cracks under pressure it was never built to absorb, or they leave the relationship so undefined that it dissolves the first time two levels of government actually disagree about something that matters. The American split was built to do neither — to hold as separate sovereigns through a fight, rather than requiring one side to simply yield because the other happened to hold more power that year.
The zoning fight is the design working, not failing
Beacon Hill's Interval Training covered a live instance of exactly that kind of fight this year: the Massachusetts Attorney General suing nine towns for refusing to comply with the MBTA Communities Act's transit-zoning mandate, several of those towns suing back on an unfunded-mandate theory, and a judge throwing those countersuits out.
It's worth being precise about which layer that argument actually took place inside, because the honest answer sharpens the point rather than softening it. A Massachusetts town is not a sovereign in Kennedy's sense. Under the Commonwealth's 1966 Home Rule Amendment, municipalities can govern themselves in ways not inconsistent with state law — but the state legislature keeps ultimate supremacy over that grant, and can narrow or override it by statute. Marblehead's zoning authority was never a fourth sovereign layer standing beside the state's. It was the state's own reserved power, devolved downward as a matter of policy, and therefore always a matter the state could choose to devolve less of.
That's not a smaller question than a fight between sovereigns. It's a different one, and no less real for being internal to a single layer: how far down should the state's authority actually be pushed, and on whose judgment. Growing a Movement as a Design Problem asked what the right altitude is for a given decision — not because there's one correct answer, but because the question is worth asking on purpose rather than defaulting to whichever altitude happened to inherit the authority first.
The MBTA Communities litigation is that question being fought out in an actual courtroom, with an actual ruling, rather than settled by whichever side had more votes on a given Tuesday. A town gets to make its case for local control. The state gets to make its case that a housing shortage spread across dozens of towns was never any single one of their businesses alone to fix. Both cases get heard. That's the design doing exactly what it was built to do — not eliminating the disagreement about where authority belongs, but giving it somewhere real to go.
What happens to the intention on the way down
That framing still leaves the harder question sitting underneath it, and it deserves a straight answer rather than a tidy one. If a municipality was never a sovereign to begin with — only a state's own reserved power, devolved downward as policy rather than owed as right — then nothing stops a state from tightening that devolution whenever it wants to, by exactly as much as it wants to.
California's housing statutes of the last several years show what that tightening looks like when a state stops litigating the question town by town and starts overriding it wholesale. SB 9 requires cities to allow duplexes and lot splits on parcels that were zoned exclusively single-family. SB 35 and AB 2011 grant developers streamlined, by-right approval for qualifying housing, bypassing local discretionary review entirely. And the "builder's remedy" is closer to a switch than a negotiation: once the state's housing agency finds a city's housing plan out of compliance, that city can lose its ability to deny a qualifying project at all, regardless of its own zoning code, until compliance is restored.
Massachusetts sues a town and lets a judge decide. California, increasingly, just removes the lever — and that difference in kind is worth calling by its name. A lawsuit is an argument the town still gets to show up and make. The builder's remedy is not an argument; it's the state deciding, in advance and by category, that a noncompliant city's own judgment no longer counts, for anything, until the state says otherwise. That's an aggressive act against the municipality as a decision-maker, not a neutral policy adjustment, and dressing it up as mere devolution-in-reverse understates what it actually does.
Legally, none of it violates anything Kennedy's phrase protects, because the phrase was never describing municipalities in the first place. But the legal answer doesn't dissolve the intention that got lost in the process, and pretending it does would be its own kind of dishonesty. Devolving a decision down to the smallest workable scale was never only a constitutional nicety — it was supposed to buy something real: a decision-maker who lives with the consequences of a zoning change, who a resident can find at a town meeting rather than track down in a capital three hundred miles away, whose accountability runs through an electorate small enough that one household's objection is still a meaningful fraction of it. Growing a Movement as a Design Problem called that the right altitude for a decision, and the case for it doesn't evaporate just because the layer holding it turns out to be a delegation rather than a sovereignty.
What tightens the state's grip, honestly, is that the delegation produced a result its own defenders wouldn't have chosen: town after town, each individually rational, declining to add its share of regional housing supply and counting on its neighbors to absorb the shortfall instead — a collective-action failure no single town could have fixed by acting alone, however sincerely devolved its authority was. That's a real argument for pulling the decision back up a layer, not a manufactured one.
But pulling it back up doesn't restore what proximity was buying; it trades the failure mode of dispersed authority for the failure mode of distant authority, and a state legislature setting one housing target for thousands of towns with wildly different land, water, and infrastructure realities is making a version of the same category error this site keeps finding in state-level daylighting rules — uniform judgment substituted for local knowledge, at exactly the moment local knowledge might have mattered most.
The intention behind devolution doesn't survive the trip up the chain intact. It survives as something the state now owes an explanation for spending, the way any layer that absorbs another layer's discretion owes an account of what it did with it.
The sharp edge that proves the design is honest
The clearest evidence that the layers are treated as genuinely, not just administratively, separate is a doctrine that costs people something real: the Supreme Court has held, most recently reaffirmed in Gamble v. United States (2019), that a single act violating both federal and state law can be prosecuted by each sovereign in turn without offending the Fifth Amendment's ban on double jeopardy. From the federal government's vantage, a state's prosecution was never its own jeopardy to begin with, and vice versa — because they aren't legally the same body making the same accusation twice. That's a genuine cost the design imposes on real people, not a footnote.
It's also proof the abstraction isn't decorative. A government that only pretended its layers were separate would have folded that doctrine the first time it became inconvenient, and called the two prosecutions one government acting through two masks. The Court didn't. Holding the line on separate sovereigns costs something precisely because the courts have taken the premise as seriously as Kennedy's phrase suggests they should — an engineering choice defended even where it's expensive, which is usually the only way to tell a design commitment from a slogan.
What the design is actually for
None of this makes any single layer correct in any given fight. It's not an argument that Massachusetts was right to sue Marblehead, or that Marblehead was right to resist. It's an argument for what the aspiration underneath this whole architecture was actually built to deliver: a people that never has to accept "that's just how authority works here" as the end of an argument, because the argument about which layer should hold which decision was never supposed to be settled once, permanently, by whoever happened to be holding power when the question first came up. It was built to stay open — arguable in court, revisable by statute, answerable, ultimately, back to the one root every layer traces to.
That's the feat worth naming aspirationally rather than only critically, for once: not that the American design gets every allocation of authority right, because it plainly doesn't, but that it built the machinery for getting the allocation argued about, region by region, fight by fight, rather than frozen. Every dispatch on this beat so far — California's insurance fight, Texas's redistricting, Florida's property tax trade, Massachusetts's zoning suits — is the same underlying question recurring in a different courtroom: who actually holds this piece of authority, and on what continuing permission. A design that keeps that question askable, indefinitely, for a people willing to keep asking it, split more than an atom. It built a habit of not letting sovereignty go quiet.
Sources: Justice Anthony Kennedy, concurring, U.S. Term Limits, Inc. v. Thornton, 514 U.S. 779 (1995) ("the Framers split the atom of sovereignty"); Chief Justice John Marshall, Cherokee Nation v. Georgia, 30 U.S. (5 Pet.) 1 (1831) (tribes as "domestic dependent nations"); U.S. Const. amend. X; Gamble v. United States, 587 U.S. 678 (2019) (reaffirming the separate sovereigns doctrine); Massachusetts Home Rule Amendment, Mass. Const. amend. art. LXXXIX (1966); California SB 9 (2021), SB 35 (2017), and AB 2011 (2022) (statewide streamlined-approval and by-right housing statutes); California Government Code § 65589.5 (the "builder's remedy" provision of the Housing Accountability Act).