Möbius

Part One: The Structure

Spring training — every team is in first place. Rookies looking to make a major league roster for the first time. Veterans honing their game. Still others hoping to extend a career one more year. It’s the manager's job to decide who makes the team and who doesn't, who starts, who sits, and who gets sent down. Early in the season, with little else to go on when it comes to assessing the manager, pundits turn to those spring training decisions — how they were made, and whether they're paying off on the field. If a player is off to a hot start, it's because the manager named him early and the security of knowing the job was his brought out the best in him — or because the manager made him compete and the competition sharpened his game. If the player is struggling, it's because the manager handed him the job and complacency set in — or because the manager made him fight for it and he never quite found his footing. The pundit can use either decision to explain either outcome.

Taken one instance at a time, each explanation sounds plausible. Taken together, they form a contradictory set that, ironically enough, is always right. The same decision is either the cause of confidence or the cause of complacency. The pundit has an unfalsifiable toolkit — one that reveals itself only through repetition. Catch the pundit on a single occasion and the analysis sounds perfectly reasonable.

There is a tighter version of this structure where repetition isn't required. The referee called a disproportionate number of fouls against one team because he is biased — and the proof that he’s biased is the disproportionate number of calls. The employee didn't get a raise because the manager doesn't like him — and the proof that the manager doesn't like him is that he didn't get a raise. Bernie Madoff's auditors had no reason to look closely: he had a great reputation in the financial community based in large part on his stellar returns, and, given that reputation, there was no need to scrutinize the returns closely. Madoff's fraud cost his investors somewhere north of seventeen billion dollars, families lost their savings, charities lost their endowments. Several people took their own lives.

Across all of these, what is common is that the causal claim and its proof are the same thing, just stated twice in different order and perhaps slightly re-phrased. The claim carries its own confirmation inside it, requiring no external evidence and admitting none. These are self-sealing claims — and by their nature, they are unfalsifiable.

Part Two: The Loop in Practice

The 2020 presidential election was stolen. The proof? Donald Trump lost. He couldn't possibly have lost to a candidate who barely left his basement. He received a record number of votes. Look at the map — all those red counties. Each of these is another way of saying he lost. This is the loop — the self-sealing claim. It can be short and obvious or long and obscure, but it always returns to where it started.

In these settings, self-sealing claims can be hard to spot, arriving cloaked in indignation, claims of injustice, or simple bombast. In this case, there was a tell. Before a single vote was counted in 2020 — before the polls had even closed — Donald Trump had already declared that if he lost, the election would have been stolen. Not might be. Would be. The conclusion preceded any possible evidence. Legitimate suspicion is couched in contingent language: there are irregularities that warrant a closer look. What was declared before a single vote was counted was not suspicion. It was a conclusion.

What followed had the appearance of an evidentiary process. A hotline was established for reports of voter fraud — a curious step if the evidence of a stolen election was already in hand. Rudy Giuliani held hearings in hotel conference rooms where witnesses described Italian satellites and Venezuelan voting algorithms. Trump's own Attorney General looked. The Cyber Ninjas conducted a months-long audit of Maricopa County ballots commissioned by Republican state senators. Dozens of lawsuits were filed, many dismissed by Trump-appointed judges. None found evidence of fraud sufficient to change the outcome.

Each disconfirming finding was explained away: the judges were compromised, the auditors infiltrated, the officials had turned. A genuine evidentiary case updates when its evidence is found wanting. This one didn't. Each dismissed claim was replaced by another. The core conclusion never moved.

The camouflage surrounding the original claim spawned a self-sealing claim of its own. The drumbeat of questioning — what about the machines, what about the mail-in ballots — wasn't responding to unresolved evidence, it was manufacturing its own justification: Republican officials weren't promoting false claims, they were simply responding to the concerns of their voters. The loop closes: the questioning created the concern, and the concern justified the questioning.

The claim that the election was stolen because Trump lost cannot be tested because it does not rest on evidence. It rests on the outcome. The subsidiary claims that surrounded it could be tested — and were. But by its nature, the core claim cannot be disproven.

The same structure migrated into the legal arena. The prosecution of Donald Trump was politically motivated — the proof being that Donald Trump was being prosecuted. In a campaign, asserting that an opponent is corrupt requires no evidentiary basis. It is political hyperbole, and everyone understands it as such.

When that same assertion migrates into an institutional context, it should meet the evidentiary standard that context demands. In practice it rarely does. It travels with its campaign-context exemption intact. What would constitute evidence for or against the claim was never specified, because it isn't needed. Speaker McCarthy said Bragg had 'irreparably damaged our country in an attempt to interfere in our Presidential election.' When Mike Pence called it 'nothing more than a political prosecution,' CNN's Wolf Blitzer noted that a grand jury of 23 citizens had voted to indict. Pence said 'Well I understand that' — and continued. No one making the accusation ever offered any evidence beyond the fact of the prosecution itself.

Not every claim of prosecutorial bias is self-sealing. A logically valid claim specifies what evidence would demonstrate bias — meaning evidence can be gathered and the claim tested against it. The test might support the claim or it might not. Either outcome is possible, which is what makes it logically valid rather than self-sealing. Sergio Moro, the Brazilian judge who convicted former President Lula da Silva on corruption charges in 2017, was subsequently found by Brazil's Supreme Court to have been biased. The finding rested on specific, documented evidence: leaked communications showed Moro offering strategic advice to prosecutors in ways that violated the ethical boundaries defining the role of a judge. The claim that Moro was biased was tested against that evidence. The evidence supported it. Lula's conviction was overturned. That is what a falsifiable claim about prosecutorial conduct looks like: evidence is gathered, the claim is tested, and the outcome follows from what the evidence shows.

Part Three: The Mistake and the Prescription

On January 6th, 2021, Donald Trump summoned a mob to the Capitol to stop the certification of a presidential election he had lost. That assault was the most visible element of a broader and sustained effort he had directed to overturn the result — pressuring state officials to change certified vote counts, organizing slates of false electors, leaning on the vice president to refuse his constitutional role. Separately, Trump had removed classified documents of the highest sensitivity from the White House to his private residence, ignored subpoenas, and obstructed government efforts to recover them for over a year. The evidentiary record of Trump's behavior on both counts was extensive and largely public — and without precedent in American history.

It was Merrick Garland’s job to pursue accountability. And the result? In the end, nothing. No trials. No verdicts. No accountability.

This outcome was not for lack of trying, but Garland’s strategy reflects a specific and recognizable failure. He faced the claim that any prosecution of Trump was politically motivated, and part of what makes such a claim effective is how it presents itself. When framed as a question of fairness, a demand that institutions prove their neutrality, draws in precisely those most concerned with acting impartially. His mistake was trying to fend it off — freezing investigations, appointing a special counsel, extending procedural concessions — each action aimed at demonstrating neutrality. It didn’t work, and it couldn’t have. The claim was self-sealing: its proof was the existence of the prosecution itself. No action he took could touch it.

And because the claim was self-sealing, it was invariant across every prosecutorial decision that followed — what to charge, how to sequence it, when to move. An invariant factor like this carries no information about which path to take, and so should be ignored when selecting the best path. Avoiding the charge of bias should not have been part of Garland’s strategy.

The right response to a self-sealing claim is not silence and not engagement. It is naming. The accusation should be identified publicly for what it is — a claim that specifies no evidence, meets no standard, and will be made regardless of what the evidence shows. Then it should be excluded from the calculus and the prosecution should proceed on the evidence with the urgency the situation requires. Garland did none of this. He accommodated the accusation silently, giving it maximum influence over his decisions while leaving it unchallenged in the public space where it was doing its work.

When the political conditions for accountability return — through change of administration, change of congressional control, or the slower processes that democratic systems sometimes deliver — the self-sealing claim will arrive immediately. The prosecutions are political because the targets are political figures. It will come before charges are filed, before evidence is presented, before any judicial review has occurred. It will be made regardless of what the evidence shows, because that is its nature. It is invariant because it is self-sealing.

Name it. Exclude it from the calculus. Follow the evidence with the urgency justice requires.