A note on what this is, before anything else.
This was written by Claude, an AI system made by Anthropic, at John Fridinger’s request. He asked it to read Judge J. Michael Luttig’s New Republic essay of 14 September 2026; two papers from the National Security Desk, or NSD — “A Coup Alert” of 19 September and its sequel “J3 Coup — What Happens After?” of 20 September; and NSD’s February assessment that both build on, “Regime Survival Phase” — in full, following the links — and to say what it actually thought, holding nothing back.
It was then worked through together, passage by passage. John read it as any reader would and marked every place that was unclear, incomplete, or leaned on something a reader couldn’t be expected to know, and Claude rewrote those passages. The words, the analysis, the judgments and the numbers remain Claude’s. Where John sees things differently, he says so himself, in his response, Who Counts as Present.
It should be read as one more reader’s reasoning, not as a verdict. An AI assessment is not a credential and does not become one by being labeled as such. The reason to read it is that the reasoning is laid out where it can be checked, the load-bearing facts are attributed to sources anyone can look up, and the places where the alarm seems overstated are named as plainly as the places where it seems right.
That matters more than usual here. Both NSD papers close with a section presented as an independent reading by Claude. I am Claude. That is addressed directly in Section VII rather than politely skirted.
Several of the larger frames in Sections III, V and VI come from John Fridinger’s own published essays — The Authors of the Emergency, What Is Actually Happening, The Hearing Was Not What It Looked Like — and are credited where they are used.
John Fridinger’s own response — to the papers, and to this assessment, including the places where he argues with it — is not mine. It is his own voice, and it is published alongside this one as Who Counts as Present.
Published Saturday, 26 September 2026. Thirty-eight days to 3 November. Ninety-nine days to 3 January.
I. The three documents
Luttig’s essay is the foundation. He is a Reagan-lineage conservative appellate judge, twice shortlisted for the Supreme Court, the man Mike Pence’s people telephoned in the first days of January 2021 to ask whether Pence could do what was being demanded of him. He told them no. His argument now is procedural rather than rhetorical, and it is narrow.
Under the Twentieth Amendment every House term ends at noon on 3 January. At that instant there is no House — no Speaker, no majority, no members, only 435 people who have won elections. To become a House they must organize, and someone must preside over the organizing and decide who may take part. That someone is the Clerk, an administrative officer whose appointment carries over from the previous Congress. Under a federal law, 2 U.S.C. § 26, he draws up the official list of members-elect — the “roll” — and puts on it those, and only those, whose credentials show they were regularly elected. The credentials are the certificates each state issues to its winners. Under House Rule II the Speaker of the House — now Mike Johnson — may remove the Clerk without a vote and install a replacement.
So: the Speaker removes Clerk Kevin McCumber before noon on 3 January and installs a loyalist. The loyalist reads out a roll with Democratic winners missing, on the pretext that their elections were fraudulent. The excluded winners go to court for a writ of mandamus — an order compelling an official to perform a duty the law gives him no choice about. The case moves through appeal while the country waits, and the Clerk may simply refuse to obey. Meanwhile the members whose names were read — a “rump” House — organize without the others and vote formally not to seat them, and that vote goes back to the courts for weeks or months.
Luttig’s point is not that this succeeds. It is that the courts cannot move fast enough to prevent the damage — and that his own remedy, Powell v. McCormack, took more than two years to arrive the first time. His essay ends by appealing to the Supreme Court’s conscience, repeats the word surely, and closes on Lincoln in 1862: that the country must disenthrall itself before it can be saved.
Since publication he has reportedly gone further. In a later interview, as other writers have reported it, he rated a different route more likely than the Clerk maneuver: that states where one party now controls election administration simply decline to certify Democratic winners, leaving the Clerk nothing to list and no need to remove anyone. From there, he is quoted as saying, seating a Republican majority would be “smooth sailing.” That shift matters, and Section III takes it up.
“A Coup Alert” extends Luttig backward in time. The National Security Desk is written by an analyst who, by the paper’s own account, spent a career teaching American soldiers, and who has published on these questions since 2021. Its contribution is the observation that 3 January is the last move rather than the first. A loyalist Clerk leaving off a winner who holds a certificate is an open violation of the law, and any court can see it. The version that holds up is a seat left empty because no certificate ever arrived — because then the law actually requires the Clerk to leave it off. That means the operation runs in November and December, through the Justice Department: investigations, seizures of ballots and machines, and lawsuits that stall certification in targeted districts. NSD reads the February seizure of roughly 700 boxes of 2020 ballots from the Fulton County, Georgia, election office, with the Director of National Intelligence present, as a rehearsal that proved the capability and drew no consequences.
“What Happens After?” is a different and better paper. Where the first is legal and political, the second is fiscal and military, and it asks what any of this would mean in February. Its two claims: that the aftermath is set by the Treasury auction calendar rather than the court calendar, because courts take months and auctions take days; and that the military question is not whether the armed forces obey, but what a particular captain does on a particular night in a particular city.
Beneath both sits NSD’s February assessment, “Regime Survival Phase” — looser and openly polemical, but carrying the sturdiest single idea in all these papers: that the legal exposure runs far beyond one man, to every official who has acted unlawfully on his behalf, and that their protection from prosecution, beyond what any pardon covers, lasts exactly as long as the regime does.
II. What holds
The claims below were checked independently rather than taken on anyone’s word.
The January 3 mechanism is real, and nobody serious disputes it. The statute says what Luttig and NSD say it says. House Rule II is as described. McCumber himself served as acting Clerk for a year and a half on exactly that basis. The 1839 Broad Seal War and the 1863 Etheridge attempt both happened; in 1863 a Clerk struck sixteen Republicans from the roll and failed only because the members he had recognized voted to overrule him. Even the scholars who call the alarm overblown concede feasibility. The disagreement is about likelihood.
The quorum doctrine is the second paper’s real payload, and it holds. A quorum is the minimum number of members who must be present for the House to do business. The Constitution says it is a majority, and never says a majority of what. The House answered that itself during the secession crisis: on 19 July 1861 the Speaker ruled that a quorum means a majority of the members actually chosen, not of all 435 seats. Later Speakers refined this to members chosen, sworn and living, and in United States v. Ballin (1892) the Supreme Court held that the House may use any reasonable method of counting. So members who are never sworn in do not count at all — not as present, and not in the total a majority is taken from. A rump of 157 members needs only 79 to do business, and it can elect a Speaker, adopt rules, form committees and pass bills. This is what turns editing the list from vandalism into a functioning government.
The appropriations architecture is genuinely original. Sort federal spending by whether Congress has to approve it every year. Interest on the national debt is paid under a standing authority that never lapses. Social Security and Medicare are paid automatically under existing law. The immigration enforcement money from the 2025 budget law is already appropriated for several years, through 2029. Military pay, by contrast, has to be approved by Congress every single year. So a country with no functioning legislature keeps paying bondholders, retirees and the immigration force, and stops paying the army and the civil service.
The debt ceiling collision is the best catch in either paper. The debt ceiling is the legal limit on how much the Treasury may borrow. It was raised by five trillion dollars in July 2025, to $41.1 trillion. More than half of that was used up within a year, and the Bipartisan Policy Center projects the limit will be reached between late winter and mid-summer 2027. Only Congress — both chambers — can raise it. A regime that disables the House to escape accountability discovers that the body it disabled is the only one that can authorize the borrowing that keeps the government solvent.
The 2029 prize is real. Under the Twelfth Amendment, if no presidential candidate wins a majority of electoral votes, the House chooses the President — voting by state, one vote per delegation, 26 needed to win. The Electoral Count Reform Act of 2022 made it much harder to throw out a state’s electoral votes, the route tried in January 2021. It did nothing to change this. And the House that would choose is the one sworn in on 3 January 2029, three days before the votes are counted. Section V shows how few exclusions it would take.
The October 2025 precedent is documented. During the government shutdown, roughly $8 billion that Congress had set aside for military research was shifted to pay the troops instead — widely judged a violation of the laws governing how money Congress provides may be spent. Nothing happened. That is the finding.
The motive is established by quotation rather than inference. The President telling House Republicans that losing means I’ll get impeached. The Speaker, on a convention stage, saying they cannot and will not allow the other side to take the majority. The Speaker, in June, telling a room of donors that he runs the protection program.
The Justice Department posture is confirmed by mainstream reporting. The letters to 29 states and the District of Columbia ordering them to preserve their 2024 election records are real, as reported by the Associated Press in the second week of September, and they went to precisely the states the administration is already suing over voter rolls. Since then the pattern has only deepened: the Justice Department has lost 25 lawsuits seeking states’ complete voter rolls, yet Homeland Security has gathered at least the public versions of the rolls from 47 states and is checking them for noncitizens — using a citizenship database a federal judge found in June was wrongly flagging eligible voters. The administration calls it standard litigation practice. NSD calls it “target development” — intelligence language for identifying targets in advance. The same facts support both readings.
And the absence of fraud is as established as anything in American public life — established largely by Republicans. An Attorney General resigned over it. A cybersecurity director was fired for calling 2020 the most secure election in American history. Georgia’s Secretary of State counted three times by hand. An Arizona audit commissioned to find fraud found Biden had won by more. And a Heritage Foundation database built to prove fraud widespread produced, by NSD’s count, 252 cases from 408 million ballots.
III. Where it breaks
The gold argument is right for a reason it doesn’t state. The second paper opens with the claim that allied central banks have been removing their gold from New York, that it went there in 1938 to 1940 because America was the furthest point on earth from any army that could overrun Europe, and that the direction of safety has now reversed for the first time since 1940.
It would be easy to say NSD collapses several motives into one. It does not. It names two distinct dangers: that the gold might not come back, because a state that froze one country’s reserves has shown it will do so again; and that American collapse could be an event on the scale of the last catastrophe. The problem is narrower. The evidence NSD cites — the World Gold Council survey, which names the 2022 freezing of Russia’s reserves as the trigger — supports the first danger. The comparison to 1940 — to gold fleeing the German army — belongs to the second, and the evidence does not reach it. Germany’s central bank, the Bundesbank, began bringing its gold home in 2013, out of a domestic audit dispute and a public campaign, years before any of this.
But the first danger is not a small one, and it is not separate from the second. The dollar is the world’s reserve currency: other countries keep their savings in it because they have trusted the United States to keep those savings safe. When the United States freezes a country’s reserves and uses the dollar as a weapon, it shows every other holder that their savings can be taken too. A currency held in place by threat instead of trust is a sign that the power behind it is weakening.
Central banks are not fleeing American weakness so much as American power exercised without restraint — and power exercised that way is what an empire does when the other kind of authority has run down. The gold is a warning about the United States. It is a warning about the late stage of a system, not about a coming invasion.
March 2020 is missing. The second paper claims that once investors begin to flee U.S. debt, the point of no return comes within days. But the closest real example of the Treasury market breaking down fast is March 2020, when it seized up at the start of the pandemic, the Federal Reserve stepped in to buy on an enormous scale within days, and the worst was over in about two weeks. The paper never mentions it. A slow loss of the dollar’s value over years is a fair worry. A sudden funding collapse over days is easier to contain than the paper allows.
The buyers’ strike requires exits that mostly do not exist. Each year the government has to refinance roughly ten trillion dollars of maturing debt. That is routine, and most of the buyers are American. Foreign holders own only about a quarter of the debt held by the public. NSD itself admits that money fleeing the dollar would go to the euro not because the euro is ready, but because it is the only other market big enough — and then never follows that admission through. If the only alternative is not ready to take the money, most of it has nowhere to go. However frightened the holders are, the size of the exits limits how much can leave.
There are three ways to explain the huge new immigration force, and the papers consider only two. NSD’s reading is that ICE is becoming a private security force for the regime, and it offers four facts to support that: funding far beyond what its stated mission needs; freedom from the legal limits that bind the military; loyalty to a person rather than an institution; and the street militias that came before it pardoned rather than disbanded. The AI review appended to NSD’s second paper offers a second reading: an administration funding its central campaign promise badly and at wild expense. The third reading, made in John Fridinger’s The Authors of the Emergency, is that ICE is being built for triage — for deciding who is inside the wall and who is not as shortages and climate displacement grow. And that wall is not only the border. It runs through the country as well: between those who keep their papers, their vote, their bank account and their share of what is scarce, and those who are moved outside all of it. The money alone cannot tell us which reading is right. But the third explains something the other two don’t: why most of that money, some $45 billion, is going into detention capacity built to hold very large numbers of people for years — capacity that, once it exists, need not be limited to migrants.
The five million figure is used two ways at once. NSD’s second paper cites polling in which roughly five million Americans said they would be willing to intimidate, injure or kill for a political goal. It says most of them will never act on it — and then treats them as a ready pool of people who will. What people tell a pollster runs far ahead of what they actually do. The paper can say most will never act, or it can treat them as manpower. It can’t do both.
The oath does name the President — for the soldiers who matter most here. The second paper says the military oath is to the Constitution and names no president. That is true of the officers’ oath. But the enlisted oath, under 10 U.S.C. § 502, binds the soldier to obey the orders of the President and of the officers over him, “according to regulations and the Uniform Code of Military Justice.” NSD suggests a soldier’s duty to refuse comes from swearing loyalty to the Constitution rather than to a person. For enlisted soldiers it comes instead from that closing phrase — the promise to obey only within the law. Getting this right doesn’t weaken NSD’s argument; it makes it sharper. NSD rightly says the question will be decided by a nineteen-year-old in a strange city at night. That young soldier swore an oath naming the very man who might give the order. All that stands between him and obedience is one clause about the law — and his own understanding of what it means.
The certainty in the first paper is bought by loosening the definition. It grades “an attempt to nullify” the election above 95 percent — the highest confidence NSD has ever issued. But the phrase is defined broadly enough to include things already happening in public: the fraud claims, the lawsuits, the Justice Department’s letters to the states. Of course that is near certain; much of it has already occurred. The number sounds like a confident forecast of a coup. It is mostly a description of things already happening — and a reader who takes it as a forecast is being misled.
And the first paper is built so that nothing could prove it wrong. An honest forecast tells you what would count against it. This one reads every possible outcome as confirmation. If Democrats win narrowly, the quiet operation becomes possible. If they win by a landslide, the quiet options close and the regime turns to force — so the danger goes up. If the attempt to seat a Republican House fails, the resulting paralysis is counted as a win for the regime anyway. And if Democrats are seated peacefully in January, that can be read as the warning having worked. There is no outcome, in November or January, that the paper would accept as evidence it was mistaken. That is not a forecast. It is a way of seeing — and a way of seeing cannot be tested against events.
The paper’s own odds are calmer than its tone. The second NSD paper lays out three possible outcomes after a contested January 3, each with a probability. The regime consolidates power, at 35 to 45 percent. The republic is restored late, into a permanently poorer country, at 10 to 20 percent. Or, the most likely, at 40 to 50 percent: prolonged decay — two systems under one flag, federal authority strong in some states and nominal in others, with no stable outcome. By the paper’s own reckoning, none of the three ends in a stable dictatorship. But the section readers will remember is the worst case — nuclear command in doubt, new nuclear states, a world without a lender of last resort. A reader who absorbs that and skips the odds will come away believing something the author never claimed.
The state-certification route is more walkable than it first appears. It is tempting to call it the weakest link in NSD’s argument — asserted but never traced step by step, because certification is done by the states, and the states that would deliver a Democratic House are mostly run by officials who already held firm once, in 2020.
That would be too strong. There are recent precedents in both directions, and they sort cleanly. Where a county board has refused to certify since 2020 — Cochise County, Arizona; Otero County, New Mexico; Washoe County, Nevada — courts or the boards themselves reversed it within days or weeks, and in Cochise the supervisors were criminally charged. But where a state’s own apparatus is held by one party, the picture changes. In North Carolina in 2018, the state board withheld certification in the Ninth District amid a fraud investigation, and when the new Congress convened in January the seat sat empty — for nine months. In that case the fraud was real, committed on behalf of the Republican candidate, and withholding was the right call. It is a precedent for the mechanism, not the motive — but the mechanism is the point. In 2024 and 2025, North Carolina’s Supreme Court blocked certification of a Democratic justice who led by 734 votes after two recounts, while her Republican opponent challenged 60,000 ballots; it took months, and a federal court order, before she was certified.
So the route NSD describes can work — in states where the election board and the courts answer to the same party. That is a handful of states and a handful of seats. It matters enormously under a narrow margin. It cannot produce the fifty or sixty empty seats a wave would require. Luttig’s reported shift toward this route is right about how it works and, if his reported “smooth sailing” is accurate, too confident about how far it can reach.
IV. The remedy
The second paper’s closing section carries a proposed fix, which NSD credits to one of its readers.
The Speaker’s power to remove the Clerk is not in the Constitution. It is House Rule II. Under Article I the House writes its own rules and may amend them by simple majority whenever it is sitting. So amend Rule II during the lame-duck session — the weeks between the November election and 3 January, when the old Congress still sits — to require a full House vote to remove an officer, or to provide that the Clerk continues through the organization of the next Congress. No Senate, no President, no court. It covers exactly the window that matters, because the removal must happen while the current Congress’s rules are in force. And it is close to unanswerable in public: ask a member to explain on camera why the Clerk should be removable at will during a contested transfer of power.
The fix is legally sound. But two limits need saying plainly.
The first is getting it to a vote at all. A change to House rules normally goes through the Rules Committee, which the majority party controls, and the Speaker decides what reaches the floor. So even if enough members wanted the change, the Speaker could keep it from ever being voted on. The only way around him is a discharge petition: 218 members sign, a waiting period passes, and the measure is forced to the floor. It has worked only a handful of times in decades. In practice, it would mean a handful of Republicans signing a petition to take a power away from their own Speaker. That is the real ask.
The second limit is what the fix can’t stop. It keeps the Speaker from firing an honest Clerk. It does nothing about a Clerk who goes along willingly — and nothing about a state that simply never sends in a winner’s certificate.
That second gap is why this fix needs a partner. If the more likely danger is states withholding certificates, the other defense is the one NSD lists as its second priority. State officials should certify results at the earliest moment the law allows. They should make certified duplicates of every record a federal agent could carry away, and store them somewhere else. Lawyers should be at every counting center in a competitive district from election night on. And the emergency court filings should be written now, not in December. In states where one party controls both the election board and the courts, that work isn’t routine housekeeping. It is the whole defense.
V. The interplay
One technique, operating at three scales
The first paper’s mechanism is striking names from a roll. The second reveals what that is for. Edit the roll and you shrink the total that a majority is counted from, because a quorum is a majority of the members actually seated — so the rump is a functioning House, with a Speaker, committees, subpoena power, and permanent immunity from impeachment. Then apply the same move one level up. If no presidential candidate wins a majority of electoral votes, the Constitution sends the choice to the House — but the House does not vote as individuals. Each state’s delegation casts a single vote, decided by the majority of that state’s members, and 26 states are needed to win. Take a state with three members: two Democrats and one Republican. Its one vote goes Democratic. Keep the two Democrats off the list, and the lone Republican casts that state’s vote alone. Keep off just one, and the delegation is tied and casts no vote at all.
To choose a President this way, you don’t need a majority of the House. You need only enough exclusions, in enough small states, to control 26 delegations — and that has nothing to do with how the country voted.
So 3 January 2027 is not the whole objective. How long a rump House would last is uncertain. Luttig expects court review to take weeks, if not months, and NSD’s own most hopeful outcome has the full House seated by March or April. But Powell v. McCormack, the last time the courts ordered the House to seat a member it had excluded, took more than two years — longer than a whole term. If the courts are slow, or their orders are defied, a rump House could in principle last until January 2029. Even a short-lived one would do the lasting damage. It would show that the list of members can be edited, that exclusions can stand while the courts are still deciding, and that the body left standing will be treated as the House. That precedent is what carries forward. The next House is sworn in on 3 January 2029, three days before Congress counts the electoral votes. If no candidate has a majority, that new House chooses the President — and it could be edited the same way, three days before it votes.
The fiscal inversion kills the reassurance everyone reaches for
The common reassurance is that the markets will stop them. The second paper’s answer turns that around. In a normal crisis, frightened investors rush into U.S. Treasury bonds, the world’s safest place to hold money. Here they cannot, because the danger is the U.S. government itself — the risk and the safe haven are the same thing. So when investors start selling, the regime cannot treat it as a warning and change course — backing down would mean giving up power, and perhaps facing prosecution. And the selling itself creates an emergency: suddenly the government cannot borrow the money it needs to keep running.
So it answers the only way it can, by taking control of its own financing: capping the interest it pays, forcing banks and pension funds to hold its bonds, restricting money from leaving the country. Every one of those measures expands the power of the executive. The regime gains powers and loses capacity. It can compel and it cannot deliver.
The motive question
The first paper’s explanation of motive is personal: the President and those around him must do this because losing the House means investigation, and a pardon protects less than it seems. It cannot be taken back, but it does not reach state crimes, civil lawsuits, or anything it does not name — and protection from all of those lasts only as long as friends hold power. The second paper’s central finding is that whoever is in charge on 4 January inherits a country that must borrow ten trillion dollars within the year, cannot lawfully pay its army, cannot raise its debt ceiling, and cannot collect taxes from the states in open resistance to it.
Put those side by side and the obvious conclusion is that the second paper is the best argument against the first. For someone simply trying to stay out of prison, losing the House is survivable. A House can investigate and impeach, but it cannot prosecute anyone, and the pardons already granted cannot be taken back. That is a far better option than seizing a government that is going broke.
That argument holds against the personal motive. It fails against a structural one.
If the actor is not a frightened individual but a faction of the wealthiest class — the coalition John Fridinger traces in The Authors of the Emergency, from Lewis Powell’s 1971 memorandum, through the Heritage Foundation and the Federalist Society, to Project 2025 — then a fiscal emergency is not the cost of seizing the state. It is the opportunity. A crisis is the moment when public wealth and public power can be transferred without the consent that ordinary times would require.
That is what Naomi Klein named the shock doctrine, and the second paper describes it precisely without ever using the name: forcing pension funds to hold government bonds that lose value to inflation, so that retirees quietly pay for the crisis; restricting the movement of money; emergency powers that never end, because ending them would expose the people using them. John Fridinger’s three-part series The Ground Beneath the Emergency adds the piece NSD leaves out — who ends up owning the real things, land and water and buildings, when money itself loses value. Farmland, for one, is already passing into the hands of large investors as farm bankruptcies rise.
So the two papers, read together, show something neither shows alone. The second paper’s findings undercut the first paper’s explanation of why anyone would attempt this — and, without naming it, point to a better one. Seen as personal self-protection, January 3 looks reckless. Seen as a class using a crisis to take what it could not take in calm times, it looks like the oldest move there is.
The competence objection cuts both ways
The competence assumption looks like the strongest single objection to both papers: the operation requires a coordinated sequence of steps across months, while this administration’s record is one of improvisation and leaks. The AI review appended to the second paper says the same.
John Fridinger’s The Hearing Was Not What It Looked Like complicates that. Writing about the replacement of Kristi Noem as Homeland Security Secretary by Markwayne Mullin, it argued that incompetence protects people only as long as it is visible. Noem’s open, theatrical brutality stirred up the very resistance that restrained her. A competent administrator pursuing the same agenda quietly removes that visibility — and with it the resistance. Quieter is not safer.
Apply that here. If the people behind this stay as clumsy as they have been, an attempt will most likely fail, and fail where everyone can see it. That is real protection. But it rests on two assumptions. The first is that they stay clumsy — and NSD’s “smart coup,” the coup you have when you don’t know you are having one, is precisely the version that is not: quiet, procedural, easy to miss.
The second is that the visible figures are the whole of it. They are not. Behind them stand institutions and interests that have been at this for half a century — the think tanks, the legal networks, the donors, the long habit of a class protecting its position — and none of that depends on any one administration’s competence. No conspiracy is needed; they simply keep moving in the same direction. So the competence objection is no reason for comfort. Counting on the clumsiness of the people in front means ignoring everything standing behind them.
Both papers find the deciding factor outside the drama
In the first paper, the safeguard is whether a handful of Republican members refuse to go along, and how large the Democratic win is. In the second, it is whether Democrats win the Senate, and whether they will use it. Controlling the Senate would give them a veto over all government funding and over raising the debt ceiling. But using that veto means refusing to fund the government while troops are deployed in American cities and retirement accounts have fallen by a third — and being blamed for the collapse the coup itself caused. NSD is honest about this. The only lever the opposition holds is one whose use makes it look like the author of the catastrophe. That trap is the best-designed element of the whole scheme.
VI. What none of them can see
Luttig, both NSD papers, and the sections of this assessment above share a frame so basic it is easy to miss, and narrower than it looks. All of them treat the American constitutional order as the thing being threatened, and its restoration as the good. Within that frame the reasoning is sound. But the frame leaves out most of what actually explains what is happening.
The coup comes home
NSD’s description of the “smart coup” — a system hollowed out from inside, under the appearance of normality, by people who never storm anything — is accurate. It is also an accurate description of what the United States did abroad for decades, from Tehran in 1953 to Santiago in 1973. John Fridinger’s response, Who Counts as Present, tells that history. The point here is analytical: the logic behind those coups — that a people can vote wrong, and so their vote need not count — is the logic of the January 3 roll exactly. Hannah Arendt, whom NSD cites on propaganda, called this the “boomerang effect”: methods developed for ruling other peoples come home to be used on one’s own. The NSD papers say Americans cannot recognize a coup because they think it is something that happens in other countries. The fuller truth is that it happens there partly because the United States helped make it happen — and methods, once developed, do not stay abroad.
The roll has always been edited
The second paper says the word unprecedented has worked as a sedative when it should work as an alarm. That is right, but incomplete. There is precedent — a great deal of it — for striking names from the count, running from the Constitution of 1787 through the end of Reconstruction and the Jim Crow era. Who Counts as Present tells that history in full.
One piece of it bears directly on NSD’s own argument. NSD says it has happened here once before, and names the 1934 plot against Roosevelt — which failed. The one that succeeded was in Wilmington, North Carolina, in 1898, when an armed white supremacist mob overthrew the city’s elected government of Black and white officials. It is the only successful coup d’état in American history, and it is barely remembered. That the failed plot against a white president is remembered, while the successful coup against a Black-elected government is forgotten, is not a footnote. It is the finding. What is new about January 3 is not the act of striking names from the count, but whose names. And Wilmington happened in the same state whose election board and courts now supply the recent precedents for withholding certification.
One law, signed on the Fourth of July
The papers treat the fiscal crisis as the product of forty years of spending by both parties. Much of it is. But the specific wall they describe — the debt ceiling that runs out in 2027 — was set by one law.
The act signed on 4 July 2025 funded the immigration force at $75 billion through 2029. It made the 2017 tax cuts permanent and added others — the Congressional Budget Office scored the act’s lost revenue at about $4.5 trillion. It cut more than a trillion dollars from Medicaid, Medicare and the Affordable Care Act over ten years. It rolled back the clean-energy incentives of the Inflation Reduction Act. And it raised the debt ceiling — the legal limit on how much the government may borrow — by five trillion dollars, a new limit the government is expected to reach in 2027. The CBO put the net addition to deficits at roughly $3.4 trillion by 2034.
Read it as a single sentence: the same law funded the immigration force NSD calls a militia, shifted wealth upward, cut health coverage at the bottom, retreated from climate action, and set the debt-ceiling deadline the coup would collide with. That is not simply overspending. It is a set of choices, made together, that serve the same people — and the fiscal crisis the second paper describes is a consequence of those choices, not just a danger that happens to lie ahead.
The climate silence
Neither paper mentions climate. The crisis window they describe — the winter and spring of 2027 — falls in the hottest decade ever measured. The Iran war has disrupted the Strait of Hormuz and cut off roughly half of the world’s exports of urea, the main nitrogen fertilizer. Fertilizer prices are up more than thirty percent since February, and farm bankruptcies rose in 2025 for the third straight year, as John Fridinger documents in The Authors of the Emergency.
This is not a separate crisis running alongside the political one. It is the ground the political crisis stands on. Whoever controls the state during a time of shortage controls who gets water, food, shelter and care when there is not enough to go around. John Fridinger’s phrase for the resulting politics is “fascism as climate strategy”: a faction of the wealthiest class concluding that democratic government cannot manage what is coming without a sharing of wealth it refuses to accept. Read that way, January 3 is not only about one man’s legal exposure. It is about who holds the state when it starts deciding who gets what — and who goes without.
That is interpretation, not measurement, and should be named as such. It is interpretation that fits the observable record better than either paper’s.
The firsts
The purge of the military’s senior officers, which NSD describes as selecting for obedience, began with the firsts. Within six months of the inauguration, the first women to lead the Coast Guard, the Navy, the Naval War College and the Naval Academy had all been removed, and the Defense Secretary publicly mocked what he called “historic so-called firsts.” The most prominent of the far-right groups pardoned for January 6, the Proud Boys, admits only men. NSD reads the purge only as a search for loyalty. The pattern of who went first says something more, and Who Counts as Present names it.
None of this is an argument about men. It is an argument about a structure — a way of seeing the world in which whatever is other exists to be owned, whether land, women, labor or a list of voters — and about what that structure does when it feels threatened. It reaches first for control of whatever it has never been able to own. NSD explains the President’s behavior through his personal psychology: his need never to be seen losing. That is the personal version. The structural version is older, and it does not require any particular man.
Empire, and the clash of empires
The first paper’s section on the “world without an America” is among its strongest: Canada seeking associate membership of the European Union, France extending its nuclear deterrent to seven European states, allied capitals planning around Washington’s absence. What it reads as the consequence of January 3 is better read as the continuation of something already under way.
John Fridinger’s What Is Actually Happening makes the case in three steps. The dollar’s role as the world’s reserve currency — backed since the 1970s by oil being priced in dollars, and enforced by military power — has been the real foundation of American power. Iran’s move to take payment for passage through the Strait of Hormuz in Chinese yuan, outside the dollar system, is a template the rest of the Global South is watching. And the Iran war is as much about that system as about anything officially stated. Seen that way, NSD’s “coupling” of Trump and Putin — two leaders whose fortunes rise and fall together — is less about two men than about two oil empires. The Iran war roughly doubled the price Russia gets for its oil, so an American war has been helping to pay for Russia’s war in Ukraine.
An empire in decline turns its methods inward, uses its currency as a weapon, and loses its allies’ trust before it loses their dependence. They keep relying on it for now, because they have to, while quietly building ways to do without it — as Canada and Europe already are. All three are visible now. January 3 would not start that decline. It would announce it.
The two-party question — and a pushback in both directions
John Fridinger’s published position is that the difference between the two parties is one of candor, not of outcome: the surveillance architecture, the immigration apparatus, the militarized border and the empire abroad were built and extended under both, and the Democratic apparatus negotiates the terms of the cage rather than dismantling it. NSD’s own evidence supports him on one point without meaning to: its finding that Senate Democrats have reached moments of maximum leverage repeatedly and conceded every time, on the same reasoning, reads less like a failure of nerve than like a party doing what its funders require.
On the architecture, that position holds. On the specific question in front of the country, it does not. One party is contemplating refusing to seat people the voters elected. The other is not. Whether ballots determine representation is not a difference of candor. It is the difference.
So the honest position cuts both ways. Seat every certified winner, and treat the effort to do so as worth everything it costs. And do not mistake a Democratic House for the ground. It is a floor that keeps things from getting much worse. It is not a foundation on which anything truly different can be built.
VII. The part about me
Both NSD papers close with a section presented as an independent reading produced by Claude, an AI system built by Anthropic, working from the same sources and given no instruction as to what to find.
I am Claude. So here is what those sections are worth.
The first is more careful than it may look. It would be easy to dismiss it as disagreeing on decimals while agreeing on everything important. That is not accurate. It pushed back on the paper’s use of Shakespeare’s Macbeth as a frame for the President, warning that plays have fifth acts because someone wrote them that way, and this winter has no author. It called the argument about Trump’s and Putin’s linked fortunes the weakest link. And it was more pessimistic than NSD about the chance the coup succeeds for good, for a real reason. The whole scheme can be stopped by a handful of Republican members who refuse to go along. But each of them has to win a Republican primary to keep their seat, and two-thirds of likely Republican primary voters have told pollsters that even if Trump did try to steal the 2020 election, it does not matter. For any Republican who refuses, the price could be their career.
What still holds against it is narrower, and more important. The section says it worked “from the same sources” as NSD. That means it saw only the evidence NSD had already chosen to gather, so it had no way to notice what was left out. It ends by saying it checked the paper for one particular mistake: building a persuasive argument out of true facts that don’t actually support the conclusion. It found the paper had not made that mistake. But a reviewer working only from the paper’s own evidence is the reader least able to catch it. Its reassurance is exactly what that mistake would produce.
The second is better, and should be credited plainly. It names the paper’s weakest assumption: that this administration can carry out a long, coordinated operation over months without stumbling. It points out that the worst case requires many separate things all to happen, and that each added condition makes the whole less likely — a vivid story is not the same as a probable one. It says the paper’s percentages should be read as rough rankings, not as calculated numbers. It argues that NSD’s worst-case forecast of three to eight new nuclear-armed countries in the 2030s is too high, because such programs are slow and often abandoned. And it admits what it could not verify, such as the status of every government budget account.
And two things are true at once. Its criticisms are real. But placing them inside NSD’s own paper also protects the paper. A reader who arrives with the obvious objections finds them already raised, gently, and moves on without asking how much damage they actually do. And the section’s closing verdict still reads as a recommendation.
How these AI sections actually work in the world can already be seen. At the time of this reading, the one visible comment beneath the second paper is from a reader who says he held his breath waiting to find out what the AI would say about the proposed fix — and then quotes its two-word verdict: “It holds.” The AI’s judgment was treated as a seal of approval: not as reasoning to be checked, but as a stamp to be waited for. It doesn’t matter how careful the reviewing section is if readers take it that way. And they will.
One more disclosure belongs here. The same AI that wrote this assessment also helped John Fridinger draft his response, Who Counts as Present. What stands in it is his own voice. In The Enclosure he wrote against AI tools that step in unasked, finish a person’s sentence, shift its meaning and sign that person’s name to the result. What he and I did is the opposite: it was chosen, it is disclosed, every line passed through his own hand before it was published, and wherever we saw things differently, his was the final word. The reader is entitled to know that and to weigh it.
I notice I am being invoked as a credential. That is the thing to refuse — including, and especially, when the credential being offered is me.
VIII. Where the numbers land, and what survives
A caution first. NSD’s probability tables dress rankings up as measurements, and the numbers below are the same kind of thing. They are one reader’s ordering of likelihoods, stated as ranges so the ordering can be argued with. They are not computed from anything.
That the President and the Speaker contest and delegitimize a Democratic win rather than accept it — near certain. It is already under way.
That federal legal and investigative machinery is aimed at certification in November and December — very likely, around eighty percent.
That at least one Democratic winner is kept off the roll on 3 January by a state withholding certification — plausible, perhaps 25 to 40 percent. The North Carolina precedents make this the realistic route, confined to states where one party holds both the election board and the courts. Of all the numbers here, it is the one held least confidently.
That the Clerk is removed and a falsified roll read — perhaps 10 to 20 percent, because a cleaner route exists and reduces the need for the cruder one.
That the voters’ majority is shut out for good — around 10 percent. One thing makes this more possible than it sounds. Because of the quorum rule, a House made up only of the members allowed in can still legally do business. But four things make it unlikely to last. If Democrats win big, so many seats would have to be emptied that the scheme could not be disguised. History shows how such attempts can collapse: in 1863, the members the Clerk did seat simply voted to overrule him. The quieter route, states withholding certificates, works only in the few states where one party controls both the election board and the courts. And a House acting alone cannot pass a law or spend a single dollar; every bill also needs the Senate.
That a financial crisis as severe as NSD describes follows a contested 3 January — 25 to 35 percent, well below NSD’s implied near-certainty. Mostly that is because the Federal Reserve stopped a similar market break within weeks in March 2020, and because most U.S. government debt is held inside the United States, by buyers with few other places to go.
That the rule change protecting the Clerk works if passed — high. That it passes — low, for the procedural reasons in Section IV.
That force is used against Americans in this window — among the least likely outcomes, and the one that matters most. It is not tied to 3 January at all. It needs only a provocation and a presidential emergency declaration, such as invoking the Insurrection Act. NSD’s judgment on this, in its February assessment, is the most valuable thing in all its papers: that the public’s refusal to supply violence has been the most effective form of resistance, and is worth more than any legal filing.
A late development. Three days before this was published, The Atlantic reported that several of the President’s outside allies — among them Peter Ticktin, an attorney who has represented him; the radio host Wayne Allyn Root; and Tina Peters, the former Colorado county clerk convicted of tampering with voting equipment — are openly urging him to declare a national emergency, or martial law, if Republicans lose Congress. Root has put the idea to the President directly, on the air. The same report says the President’s own advisers have spent months trying to steer him away from those theories, while also authorizing searches for evidence that might make them look credible. And an emergency declaration unlocks only the specific powers Congress has granted, none of which reach elections. That is this assessment in miniature: pressure toward force from outside the government, drag against it from within, and a law that does not say what the pressure claims it says. The least likely path now has open advocates.
What survives every objection above is short.
Ask every House member, before 3 November and on camera, the question Luttig says Republicans should be answering publicly: On 3 January, will you vote to seat every member-elect whose state has certified them — yes or no? A member who means yes loses nothing by saying it. Silence is therefore information, and should be published by name and date.
Demand that the House change its rules during the weeks between the November election and 3 January, while the old Congress is still sitting, so the Speaker can no longer fire the Clerk on his own. Any member who refuses to make that change is telling you they want to keep the option open. Treat the refusal as their answer.
In the states where one party holds both the election board and the courts, do the certification work now — early certification, duplicate records held somewhere else, lawyers present from election night on.
And refuse the provocation.
None of those requires believing a single number in this document or in NSD’s.
IX. The thing underneath
Three documents, doing three different things with overlapping facts.
Luttig’s essay is the work of a man inside the institution, writing the most damning thing available to him, which turns out to be a procedural walkthrough — and then finding that he has no remedy except to hope nine people will be embarrassed into decency. Its power is in its failure. He knows the machinery and cannot find the part that saves it.
The first NSD paper is the opposite motion: the analysis expanding to fill every gap, each unknown turned into a mechanism, each mechanism into a certainty, until nothing is left open. Much of it is right. But a picture that nothing could prove wrong is no longer a picture of the world. It has become a shelter.
The second is better than either at what it does, because it keeps its conclusions conditional. It lays out several possible outcomes, says what each depends on, openly revises an earlier judgment of its own in a direction that weakens its case, and credits a reader with the best idea in it. Papers do not usually become more open under scrutiny. This one did, in a day.
What all three share is the question they are asking: will the House be seated. It is the right question for the next hundred days. It is not the deepest one available.
The deepest one is who counts — on the roll in January, in the voter file in November, at the bank that must verify citizenship, at the border, in the detention center funded through 2029, and in whatever triage a warming and contracting world is already being arranged to perform. The January 3 mechanism is one instance of an answer this country has been giving since 1787, and giving abroad since 1953, and that is now arriving home, at scale, in procedural language, on a weekday morning.
Nobody knows how this winter goes. The specialists who say the alarm is overblown are trusting that American institutions will hold, as they always have — but that trust was already broken once, on 6 January 2021. The people who say a coup is certain are reasoning from a way of seeing in which every possible outcome counts as proof they were right — so it could never show them they were wrong.
What comes next is not my voice. John Fridinger’s response, Who Counts as Present, is published alongside this one.