There are two ways to destroy a republic using nothing but its own election machinery, and the men who wrote the Constitution could see both.
Here’s the first. Suppose the states run congressional elections entirely, with no federal involvement. A handful of states decide they’ve had enough of the national government. They don’t rebel, and they don’t secede. They simply decline to hold the elections. No dates set, no polling places, no ballots printed. Within a cycle, Congress can’t seat a quorum. Nobody overthrows the national government. It simply starves.
Hamilton names that danger by hand in Federalist 59. Of the state governments, he writes that they “could at any moment annihilate it, by neglecting to provide for the choice of persons to administer its affairs.”
Here’s the second. Suppose Washington runs congressional elections entirely, with no state involvement. Now the people currently holding federal office write the rules determining whether they keep holding it. Registration, identification, timing, counting, all set by the body whose membership is at stake. Nobody has to cheat. They only have to draft.
Two dangers, both structural, both real, and they don’t feel symmetrical at all. The first sounds like a civics-class hypothetical that could never happen. The second sounds like ordinary legislating. Which is roughly backward from how dangerous each one is in a given decade.
There’s a bill in the Senate right now that makes you pick which one frightens you more, and most Americans are picking based on who currently holds power, which is the exact method the Constitution was designed to make unnecessary.
My conclusion up front, so you can argue with me instead of waiting for it. The core principle behind the SAVE Act is sound, and the constitutional authority for it is real. The bill as written goes considerably farther toward nationalizing election administration than a Madisonian should want, and the people cheering hardest for it haven’t run the test the Founders would have run first. Still, I support it, and I expect after reading this you will, too.
The clause is itself a mixed government provision
Article One, Section Four:
“The Times, Places and Manner of holding Elections for Senators and Representatives, shall be prescribed in each State by the Legislature thereof; but the Congress may at any time by Law make or alter such Regulations.”
Notice what it doesn’t say. It doesn’t say Congress shall administer congressional elections; it doesn’t establish uniform national procedures, and it creates no federal elections agency. There was no shortage of men at that Convention who knew how to draft an agency into existence when they wanted one.
What it establishes is a presumption followed by an override. States prescribe. Congress may alter. Default and backstop.
We don’t have to guess about the reasoning, because Hamilton explained the choice plainly. He begins from a premise he treats as obvious, that “every government ought to contain in itself the means of its own preservation,” and then lays out the menu. The power over elections, he writes, “must either have been lodged wholly in the national legislature, or wholly in the State legislatures, or primarily in the latter and ultimately in the former.” Wholly federal, wholly state, or primarily state with a federal reserve. The Convention took the third, and he says so directly:
“they have submitted the regulation of elections for the federal government, in the first instance, to the local administrations; which, in ordinary cases, and when no improper views prevail, may be both more convenient and more satisfactory; but they have reserved to the national authority a right to interpose, whenever extraordinary circumstances might render that interposition necessary to its safety”
In the first instance, to the local administrations. In ordinary cases. A reserved right to interpose when extraordinary circumstances make it necessary to the safety of the national government. Four moving parts, all of them working.
And notice who Hamilton is arguing with, because it changes how you read him. Federalist 59 is a defense. Anti-Federalists had looked at that clause and said, correctly, that it handed the national legislature a weapon. Hamilton’s answer concedes the weapon and argues that a government without it can be killed by inaction. So the man defending the federal power is the man describing it as a reserve. If the strongest advocate for the clause calls it an emergency instrument, be suspicious of anybody who calls it a general grant of administrative authority two hundred and forty years later.
Put that next to Madison in Federalist 51, because the two essays describe one machine from two angles:
“In the compound republic of America, the power surrendered by the people, is first divided between two distinct governments, and then the portion allotted to each subdivided among distinct and separate departments. Hence a double security arises to the rights of the people.”
And the line everybody knows: “Ambition must be made to counteract ambition.”
The Elections Clause is that theory applied to the most sensitive piece of hardware in the system. Every other contest for power in this country runs through elections. Control the input and you eventually control everything downstream without ever winning an argument, which is the joint where republics die, and the Founders knew that history better than we do because for them it was recent.
One modern footnote. Moore v. Harper rejected the independent state legislature doctrine in 2023, confirming that legislatures exercising Elections Clause power remain bound by their own state constitutions. So the state side of this arrangement is itself internally checked, which is Madison’s subdivision one level down.
What the courts have actually held
Now the part my side doesn’t enjoy. Federal involvement in congressional elections is settled law, it’s broad, and it’s been settled a long time.
Smiley v. Holm in 1932 asked how far “Times, Places and Manner” reaches, and the answer came back as a list. The clause covers “a complete code for congressional elections, not only as to times and places, but in relation to notices, registration, supervision of voting, protection of voters, prevention of fraud and corrupt practices, counting of votes, duties of inspectors and canvassers, and making and publication of election returns.”
Registration is on that list. Fraud prevention is on that list. Counting is on that list.
And when Congress acts, its rules win. Federal election regulations are paramount to those made by a state legislature, the Court has said, and where they conflict the state rule ceases to be operative. That language comes from Ex parte Siebold in 1880, which upheld federal prosecution of Baltimore election judges under the Enforcement Act of 1870, and the Court quoted it again in Foster v. Love in 1997. So retire one bad argument permanently. “The Constitution leaves elections to the states” is half of an accurate statement of American law, and leaving off the other half is how people end up surprised in court.
Now the case at the center of this debate. Arizona v. Inter Tribal Council of Arizona, 2013. Scalia wrote it, seven to two.
Arizona’s Proposition 200 required documentary proof of citizenship with a registration application. The federal form created under the National Voter Registration Act requires an applicant to attest to citizenship under penalty of perjury, without documents. Arizona wanted to reject federal forms that arrived without them. The Court said no. The NVRA requires states to “accept and use” the federal form, and Scalia read that at face value: a form a state must accept and use has to be sufficient for the thing it’s meant to accomplish.
That irony runs against the way the case usually gets cited. Inter Tribal Council is a loss for a state that wanted documentary proof of citizenship. It’s also the strongest available argument that Congress may require it. The case rests on Congress’s Elections Clause authority over federal registration procedure, and if that authority is broad enough to tell Arizona it may not demand documents, it’s broad enough for Congress to amend the statute and tell every state it must.
But the same opinion draws the line that matters. Scalia wrote that the Elections Clause “empowers Congress to regulate how federal elections are held, but not who may vote in them.” That distinction has a constitutional home. Article One, Section Two provides that electors for the House “shall have the Qualifications requisite for Electors of the most numerous Branch of the State Legislature,” and the Seventeenth Amendment says the same for the Senate. Congress regulating the manner of an election stands on firm ground. Congress inventing substantive voter qualifications doesn’t.
One detail complicates the partisan read of all this. Justice Thomas and Justice Alito each dissented separately, and Thomas argued that because states set voter qualifications, they must also determine whether an applicant satisfies them. The strongest conservative voice on that Court, in the case everybody cites, was arguing for state control of exactly the thing this bill nationalizes.
Verification, qualification, and what the bill actually says
Here’s the distinction as cleanly as I can put it. A qualification rule says only persons possessing characteristic X may vote. An administration rule says here is how officials determine whether a person possesses X. The first is constitutionally awkward for Congress. The second sits comfortably inside the Elections Clause, because registration and fraud prevention are on the Smiley list by name.
Citizenship is already a qualification. Federal law has made it a crime for a noncitizen to vote in a federal election since 1996, and a crime to falsely claim citizenship in order to register. Every state requires citizenship. The federal form requires attestation under penalty of perjury. Anyone telling you this bill would newly ban noncitizens from voting is confused or selling something.
Federal identification requirements aren’t unprecedented either. The Help America Vote Act of 2002 requires first-time voters who registered by mail to present identification, and it required every state to build a computerized statewide voter list matched against motor vehicle and Social Security records, with a 2004 deadline and waivers running to 2006. Washington has been in the voter list business for more than twenty years.
So the proposition is defensible. When you register for a federal election, demonstrate that you’re legally eligible to vote in it or pound sand.
So we should pay attention to where the line blurs. A federal rule where the only acceptable document is a United States passport is formally a verification rule that in practice disqualifies roughly half the country, and no court would treat it as mere procedure. The same rule with a long document list, free issuance, and an affidavit fallback has the identical stated purpose and a radically different effect. If the cost of cleaning up our elections is taxpayer funded ID, I think that’s a compromise most would be willing to make. Nothing in the clause tells you where between those two the line falls, by the way, and the right answer depends on what records people actually have where they live, which argues for fifty decisions and not one.
Now let’s look at the bill, which isn’t the same thing as the core idea.
The version people remember is H.R. 22, introduced by Rep. Chip Roy in January 2025, passed by the House that April, 220 to 208, and stalled in the Senate. It required documentary proof of citizenship for federal registration.
What’s in front of the Senate now is larger, and the paperwork trips people up. The text is the SAVE America Act, H.R. 7296, but the House didn’t pass that bill as such. It took a Senate bill already sitting on its calendar, S. 1383, substituted the SAVE America Act text under a closed rule, and passed that on February 11, 2026, by 218 to 213, with Rep. Henry Cuellar the only Democrat in favor. The number the Senate holds is S. 1383.
It requires documentary proof of citizenship for federal registration. It requires photo identification for in-person voting, and for mail voting either a copy of photo ID or the last four digits of a Social Security number with an affidavit. It requires states to submit complete, unredacted voter registration lists to the Department of Homeland Security for checking against its SAVE system, with removal after notice and a chance to establish citizenship, plus a separate pathway letting a state remove a registrant at any time upon receiving documentation or verified information of noncitizenship, where the notice obligation is much less clear. It creates criminal penalties tied to registering people who can’t produce the required proof, and private rights of action. It appropriates nothing.
The Senate has rejected it five times this year. It agreed to take the bill up in March, 51 to 48, then failed cloture twice, on March 21 and again on March 26 at 53 to 47. In late April Sen. Kennedy forced a vote on attaching it to a reconciliation package, and that failed 48 to 50. On June 4 a Graham amendment to an immigration funding package failed 48 to 50, with Collins, Murkowski, McConnell and Tillis voting no.
Then the strategy changed. In late July, the House attached the text to the National Defense Authorization Act, to a budget reconciliation framework, and to a third bill. Reconciliation is the one that matters, because reconciliation isn’t subject to the filibuster. Whether these provisions carry enough budgetary justification to survive the Byrd rule is live and unresolved. So the bill isn’t dead, at least not yet.
A few footnotes before we move on. The DHS system is called SAVE, for Systematic Alien Verification for Entitlements. It was built in the 1980s so benefits agencies could check immigration status, and it has never been a voter database. Pointing tens of millions of voter records at a system designed for something else is an engineering decision, and engineering decisions have error rates.
The second footnote is constitutional, and almost nobody raises it. The Elections Clause covers elections for Senators and Representatives. It doesn’t reach presidential elections. Congress’s authority there comes from Article Two, Section One, and it amounts to determining the time of choosing electors and the day they vote. The NVRA, which this bill amends, covers elections for federal office including the presidency, and Congress has legislated there for decades on other theories. But the constitutional footing for the presidential half of a sweeping federal registration and identification mandate is materially thinner than for the congressional half. If you’re arguing this bill is constitutionally conservative, that gap is the part you need an answer for.
Then there’s the enforcement architecture: criminal exposure for officials, private lawsuits, mandatory database checks, no money. Ask what a rational county clerk does under that combination. In every ambiguous case, the safe move is to reject and let the citizen sort it out. Call that incentive design. It doesn’t require anybody to be conspiring, and it’s a policy choice about error rates that was made without ever being debated as one.
The question mixed government teaches you to ask
So Congress probably has the power, it appears as though they do. Set that aside, because it’s the less interesting question. Ask instead why Congress should exercise all of the power it possesses.
Hamilton’s own language tells you the ethos even where it can’t give you a rule. State administration suits “ordinary cases.” The federal power exists so Congress may interpose when “extraordinary circumstances” make it necessary to the safety of the national government. That isn’t judicially enforceable. The text says Congress may act at any time, and no court is going to strike down a valid election statute for insufficiently extraordinary circumstances. So this isn’t an argument about litigation, but about what the provision is for, and that has to be honored voluntarily or not at all. It seems an invasion of foreign nationals attempting to access the franchise, either way, would substantiate extraordinary circumstances.
Here’s the strongest thing in this argument. The people currently holding federal office should not control the machinery determining whether they continue holding federal office. That’s precisely the alignment of power and self-interest that mixed government exists to pull apart. You don’t let the man being measured hold the ruler.
The symmetry in the original design is genuinely elegant. A state might manipulate federal elections to weaken the Union, so Congress gets an override. Congress might manipulate federal election law to entrench the incumbent coalition, so states remain the primary administrators. State administration checks federal power, federal override checks state obstruction. Ambition counteracting ambition, applied to the mechanism that decides who holds ambition’s levers, as Madison would have demanded.
Decentralization also buys you contained failure. If Georgia adopts a bad procedure, Georgia has a problem, and forty-nine states can watch it, litigate it, refuse to copy it, and run alternatives that generate comparative evidence. We saw this in Florida in the early 2000s, and they corrected themselves immediately (and now have the best electoral process in the nation). If Washington adopts a bad procedure, every federal election has the identical problem in the same cycle, with no control group and no jurisdiction far enough outside the damage to notice.
Which brings us to the part that should make everybody uncomfortable. Progressives sometimes want comprehensive national voting standards because they distrust Republican state governments. Conservatives sometimes want comprehensive national integrity standards because they distrust Democratic state governments. Those are the same argument with the labels swapped, and the constitutionalist answer to both is identical: that’s exactly why neither faction should want Washington holding this pen permanently.
So run the test. Don’t just nod at it. Take the election statute you most want passed. Now imagine it administered for twelve consecutive years by the coalition you trust least, with the enforcement discretion, the database access, the criminal referral authority, and the rulemaking power all in their hands. If your answer is that this would be fine because your side would never lose that badly, you’re doing prediction, not constitutional reasoning, and you’re doing it badly, because every faction that has ever held power in America has eventually stopped holding it.
You don’t design institutions assuming your people will run them. You design them assuming your enemies eventually will. This is, of course, about restrictive voting policies we sometimes discuss, not about securing elections from illegals.
The best case against everything above
Start with the hardest one. Federalism in voting has a history, and it's ugly.
My argument assumes decentralization is basically protective. For a long stretch of American history, it was the opposite. State control of election administration was the instrument of mass disenfranchisement: literacy tests, poll taxes, grandfather clauses, white primaries, registrars available Tuesdays between two and three except when they weren’t. All of it administered locally, all of it defended in the language of state authority over elections, and all of it working for roughly seventy years.
The Fifteenth Amendment exists because that happened, and the Voting Rights Act exists because the amendment alone didn’t work, and even that gave the federal government power that was abused. So when I say failure is contained under decentralization, a reasonable person answers: contained for whom? If your state is the one doing the excluding, you’re trapped inside the containment, and it invites correction from the next admin.
Anybody making a federalism argument about voting without reckoning with 1890 through 1965 isn’t serious. The design, left alone, didn’t protect those voters. Federal intervention did. Where it stops working is against this particular bill, because it argues for a federal floor protecting access to the franchise, which is what the Fifteenth Amendment enforcement power was built for and what the Voting Rights Act is in principle. It doesn’t argue for general federal control of administration, and it doesn’t touch the incumbency objection at all.
And one detail in that history cuts in an unexpected direction. The Voting Rights Act established prohibitions and a preclearance mechanism aimed at specific jurisdictions with specific records, and it left the running of elections where it was. The most successful federal intervention in the history of American election law was structured as a backstop, which is what Hamilton described the power as being.
The second objection targets me rather than the bill.
My safety valve reading has no enforceable limiting principle. Hamilton’s “ordinary cases” language isn’t law. The text says Congress may act at any time. A restraint that’s purely prudential holds exactly as long as it’s convenient to the majority of the moment, and a constitutional argument that reduces to “Congress ought to show restraint” has stopped being an argument and become a wish.
That’s correct about the law, and I’m not claiming a court should strike this bill down on structural grounds. It would lose, and it should lose because this bill is clearly constitutional. What it misses is that an enormous amount of American constitutional practice has always been prudential. Presidents left office after two terms for nearly a century and a half with no amendment requiring it. A limit unenforceable in court is still a limit. It’s a political one, and it holds only if the public holds it. A political check has in fact been operating here, five times this year, twice at 48 to 50, with Republican senators supplying the margin against their own leadership. I say that to say, just like every other situation, some of this isn’t structural but political, and it depends on us to enforce a standard.
And briefly, the scale objection. Someone will say the framing is obsolete, that the Founders designed for four million people with three-week mail. Partly right, I guess. Scale creates real coordination problems, and interstate data sharing is a legitimate federal contribution. But an officeholder’s incentive to shape the rules of his own reelection isn’t a function of population size. And note the conflation buried in it. Scale argues for federal capability. It doesn’t argue for federal control.
What I would actually favor
Criticism without an alternative is complaining, so here’s the alternative in a few parts.
Start with the floor: only legally eligible voters participate in federal elections. This is the most obvious thing in the world, so let’s state it plainly as a federal principle. It’s true, and it’s already the law.
Then require that states maintain a reliable mechanism for verifying eligibility and identity, specifying the outcome and saying nothing about the method. Congress establishes a narrow minimum standard only where necessary to protect the integrity or continued operation of federal elections, which is Hamilton’s extraordinary circumstances converted into a drafting discipline Congress imposes on itself. And then states choose how to satisfy it, with every excuse answered. Some subset can’t get the ID? We provide it, no excuses.
Next, federal databases are made available to states for verification on request. Available to states, requested by states, and emphatically not a federally controlled master voter roll (you don’t want the other side controlling this, either).
That last point is where I part company with the current bill most sharply. A single national voter list is a single point of capture, a single point of failure, and a standing temptation for whoever runs the executive branch. The difference is who holds the initiative.
Under a verification service, a state official has a name and a question and decides what to do with the answer. Under a master list, the federal system holds the roster, runs its own matching on its own criteria, and produces outputs that states have to act on. A false positive there is a letter to a citizen demanding she prove she belongs on a roll she’s been on since she was eighteen, with removal at the end of it. And it should be on the citizen and on the locale to ensure those rolls are up to date.
Finally, ordinary questions stay decentralized unless Congress can demonstrate compelling federal necessity. Registration procedure, early voting, ballot handling, polling locations, identification mechanisms, recount procedures. Those are the ordinary cases. Protecting the country from an illegal invasion or vote tampering is clearly extraordinary and invites oversight.
And if Congress imposes a mandate, it should pay for it. A federal requirement with no appropriation takes credit in Washington and sends the bill to a county clerk in a building with four employees and a copier.
So is the SAVE Act a good idea?
The principle is certainly better than the institutional architecture.
Requiring citizenship verification for federal voting is entirely compatible with the idea that elections must be restricted to lawful electors, and Congress has a strong constitutional argument that registration and identification requirements fall inside its Elections Clause authority. Inter Tribal Council hands that argument to them.
The 2026 version goes well beyond telling states to ensure federal voters are citizens. It builds a nationally standardized election administration apparatus: federal identification rules for in-person and mail voting, mandatory interaction between state rolls and a Homeland Security database, criminal exposure for local officials, private rights of action, and no money to fund any of it.
From the standpoint of mixed government, federalism, and the Founders’ documented suspicion of concentrated power, I’d rather Congress establish the constitutional floor and leave states broad discretion about how to meet it. Impose and enforce severe penalties for those who are derelict, similar to restrictions imposed in the VRA.
The principle in one sentence: the federal government should possess enough authority over federal elections to prevent the states from destroying or corrupting the national government, and not so much authority that the national government can easily control the process by which it perpetuates itself.
That comes remarkably close to what Hamilton was defending in Federalist 59.
And it leads somewhere more interesting than either position currently on offer. States’ rights absolutism was never the Founders’ position on elections. It’s just not true that “states control the vote.” They rejected it explicitly, in writing, and explained why. National election administration wasn’t their position either, and they rejected that in the same paragraph.
The constitutional design deliberately puts the election machinery between the two, in an uncomfortable place where neither institution settles the question alone. That discomfort is load-bearing. Treating it as a defect awaiting a fix by whichever party finally reaches sixty votes gets the whole thing backward.
If you find yourself agreeing with any of this only when your party is in or out of power, that’s worth knowing about yourself. Our system was designed to withstand bad leaders, and to never create a weapon powerful enough to erode the entire thing at once.