title: "Constitutional Arbitrage: How Losing in Court Became a Governing Strategy" subtitle: "The Supreme Court struck down the president's tariffs. Four hours later, they were back. The problem isn't defiance — it's a system designed to be outrun." author: Rebecca Rowan publication: For the Republic date: 2026-02-21

Constitutional Arbitrage: How Losing in Court Became a Governing Strategy

Friday morning, February 20, 2026. The Supreme Court strikes down President Trump's IEEPA tariffs, 6-3. Chief Justice Roberts writes that two words buried in a 1977 statute -- "regulate" and "importation" -- "cannot bear such weight." Gorsuch and Barrett join the majority. Cable news calls it a landmark.

By Friday afternoon, the president has signed a new order reimposing tariffs under Section 122 of the Trade Act of 1974. They take effect Monday. He calls the justices "a disgrace." Treasury Secretary Bessent announces the administration will "leverage Section 232 and Section 301 tariff authorities that have been validated through thousands of legal challenges."

The most powerful court in the world said no -- and it mattered for about four hours.

This is not an aberration. It's a strategy. And it has a name.

The Menu

Arbitrage, in finance, is a simple trick: exploit the gap between what something costs in one market and what it's worth in another. Constitutional arbitrage is the executive-power version -- when one law says no, hop to another that says yes, and move faster than the courts can keep up.

The U.S. Code offers a very long menu. Multiple statutory provisions hand the president tariff authority: IEEPA, Section 122, Section 201, Section 232, Section 301, and others besides. A court can say "not under this law." There is always another law. The goal never changes. Only the legal vehicle rotates.

Even Justice Kavanaugh -- writing in dissent -- conceded the ruling was "not likely to greatly restrict Presidential tariff authority going forward" because "numerous other federal statutes authorize the President to impose tariffs." Think about that. The dissent endorsed the thesis. The menu is deep, losing one item barely narrows it.

Tariffs are the cleanest example, but the pattern has at least three other variants -- each uglier than the last. Semantic compliance: rescind the specific document a court blocked, then claim the identical policy continues under different authority. Retaliatory escalation: lose a case, then go after the person who won it. Procedural undermining: rewrite the rules of litigation itself so courts lose their sharpest tools. Not all of these are equally alarming. But they all do the same thing -- they make judicial review structurally irrelevant without ever crossing the bright line of outright defiance.

But here is the uncomfortable part: the tariff case, taken alone, actually looks like the system working.

The Steelman

I want to be honest about this.

The Supreme Court's ruling did constrain the president. Materially. The maximum tariff rate dropped from 145% under IEEPA to 15% under Section 122 -- a roughly 90% reduction at the top end. Duration: indefinite to 150 days. Scope: country-specific punitive rates to a uniform global floor. Businesses and trading partners immediately started operating under the new, lower parameters.

Grant this fully. Real constraints. Thirty-nine judges appointed by five different presidents -- including Trump's own appointees -- have ruled against the administration. The Competitive Enterprise Institute is right that "separation of powers is a liberty-protecting principle, not a technicality." On tariffs specifically, the steelman is strong. The court caught the pitch.

Congress bears real responsibility here, too -- the libertarian critique is right about this much. The overlapping statutory delegations that make arbitrage possible weren't invented by this president. Multiple tariff authorities. Broad emergency powers enacted decades ago. Congress built the menu. But Congress leaving the toolkit unlocked doesn't make it acceptable for a president to grab every tool at once and gut the spirit of a court ruling. The arsonist isn't absolved because someone else left the matches out.

But here's the question the steelman can't answer: if the president can impose 15% tariffs under Section 122 today, launch Section 232 national-security investigations next week, and pursue Section 301 unfair-trade actions next month -- does stacking properly delegated authorities still count as "constraint" when the cumulative effect approximates the original policy the court just killed? The system caught one pitch. There are five more arms in the rotation.

Meanwhile, 274 cases sit awaiting ruling. The judiciary is structurally behind -- processing challenges slower than the administration generates them.

Outside tariffs, the pattern gets ugly fast.

The Rotation

Start with immigration. The administration invoked the 1798 Alien Enemies Act -- a wartime statute never used outside a declared war -- to fast-track deportations of Venezuelans. The Fifth Circuit blocked it. So the administration pivoted: the Supreme Court allowed Trump to terminate Temporary Protected Status for nearly 350,000 Venezuelans, stripping their legal status and making them deportable through ordinary proceedings. Same deportations. Different legal pathway. The American Immigration Council called it "de-documenting." The term deserves to stick.

Federal spending is where the mask slipped entirely. A court blocked the OMB's funding-freeze memo. The administration rescinded the memo -- then declared the freeze continued under different executive orders. White House spokesperson Karoline Leavitt said it out loud: "This is NOT a rescission of the federal funding freeze. It is simply a rescission of the memo." Read that again. The court had explicitly ordered the administration not to reinstate the freeze "under a different name." They did it anyway. In the same breath as the rescission. That's not arbitrage. That's contempt with a press release.

And then the courts kneecapped themselves. In Trump v. CASA, the Supreme Court effectively barred nationwide injunctions, 6-3. Before CASA, one federal judge could freeze a policy for the whole country. After CASA, you need class actions -- slower, more expensive, and by the time they grind through the system, the policy has been running for months. The judiciary's best real-time weapon against executive overreach? Gutted by the judiciary itself.

These cases sit on a spectrum, and the spectrum matters. Authority-switching on tariffs is aggressive but arguably how delegated power was designed to work. Claiming a freeze continues while rescinding the document that authorized it is lawyerly bad faith. And the Washington Post's finding that the administration defied, delayed, or manipulated rulings in a full third of all adverse decisions -- or the Cato Institute's documentation that immigration accounted for 57% of all cases where judges found the administration misled the court -- that's something else entirely. You don't need to lie to judges if your authority-switching is legitimate. A DOJ official says he was fired for refusing to do exactly that.

Not every variant here is equally alarming. But the spectrum, taken together, reveals a strategy. Not a series of coincidences.

Outrun, Not Defied

Here is the thing about defiance: everyone can see it. When Andrew Jackson (apocryphally) said "John Marshall has made his decision; now let him enforce it," that was a crisis you could point to and name. Defiance triggers alarm bells.

Arbitrage is invisible.

Constitutional arbitrage is more dangerous than defiance because it avoids the confrontation that would force a response. This is where the "presidents have always clashed with courts" argument falls apart. Yes, Lincoln defied the judiciary on habeas corpus -- during an actual civil war. FDR tried to pack the Court after a series of adverse rulings within one policy domain during the Depression. But neither of them turned court-dodging into a system -- an all-purpose governance tool deployed across every major policy area at once. And neither of them did it in peacetime. That's the difference, and it's not a small one.

I'm going to editorialize here: what we're watching is the emergence of a decorative judiciary. The courts keep issuing rulings. Cable news keeps calling them "major blows." And the next morning the policy is back under a different statute. The republic still has courts, but their rulings are increasingly cosmetic -- a constitutional Potemkin village. All the forms of judicial review. None of the substance. 655 lawsuits filed. 217 plaintiff wins. And an agenda that continues largely unimpeded. The losses aren't defeats. They're overhead -- the cost of doing business in a system where another statute is always available and reimposition is always faster than adjudication.

The Diagnosis

Return to where we started. Friday morning, February 20. The Supreme Court says no. By Friday afternoon, the president shrugs.

81% of Americans believe the administration must follow court orders. The problem isn't public will. The problem is that our constitutional system assumed good-faith compliance with the spirit of rulings, not just their letter. Nobody bothered to write that down, because who imagines a president treating legal authority like interchangeable inventory?

Naming the mechanism is the first step. You can't fix a system failure you can't describe. "Trump defies courts" is a headline. "The administration exploits gaps between overlapping statutory authorities faster than the judiciary can respond" is a diagnosis -- and a diagnosis points toward treatment. Statutory consolidation. Mandatory compliance windows after adverse rulings. Congressional reclamation of the delegated powers that were never meant to be stacked as a judicial workaround.

Constitutional arbitrage has a name now. The question is whether the public, Congress, and the courts themselves act on the diagnosis before the patient stops responding to treatment.

Revision Log

Fact-Check Corrections

• Cato 57% statistic (RED FLAG): Rewrote from "judges found officials misled the courts in 57% of immigration cases" to "immigration accounted for 57% of all cases where judges found the administration misled the court." The original inverted the denominator; the actual finding is that 57% of all misleading-courts instances involved immigration, not that 57% of immigration cases involved deception.
• Lincoln/FDR historical comparison (RED FLAG): Removed the factually vulnerable "one court / one dispute" framing. Lincoln's habeas suspension involved thousands of detentions across multiple jurisdictions; FDR's court-packing responded to multiple SCOTUS losses across several policy areas. Revised to "one constitutional question" (Lincoln) and "a series of adverse rulings within one policy domain" (FDR), which preserves the distinction -- that Trump's approach is qualitatively different in scope and peacetime context -- without minimizing what Lincoln and FDR actually did.
• CRS vs. CFR attribution (YELLOW): Removed the claim that "The Congressional Research Service identifies six separate statutory provisions" since the link pointed to a CFR article. Simplified to "Multiple statutory provisions" with the CFR link, which supports the general claim. Also corrected the statute list: removed "the Trade Expansion Act" (which IS Section 232) and added Section 201 per the actual CRS enumeration.
• CASA "eliminated" overstatement (YELLOW): Changed "eliminated nationwide injunctions" to "effectively barred nationwide injunctions." The Court left class-action pathways open, and courts began using them within days.
• 81% source attribution (YELLOW): Changed the link from Fox News (which contains no polling data) to the actual source: NBC News Decision Desk Poll.
• CEI quote link (YELLOW): Changed the link from a SCOTUSblog year-in-review article to the actual CEI blog post where the quote appears.
• "90% reduction in tariff power" (YELLOW): Qualified to "a roughly 90% reduction at the top end" -- clarifying this refers to the maximum rate, not to tariff "power" broadly, since Section 232 and 301 tariffs with no rate caps remain available.

Structural Changes

• Relocated Congress paragraph: Moved from "Outrun, Not Defied" (where it broke momentum between the decorative judiciary insight and the close) into the steelman section, where it strengthens the concession and flows naturally into the pivot. Compressed slightly. The "arsonist/matches" line survives.
• Renamed steelman section header: Changed "The System Worked -- Until It Didn't" to "The Steelman" -- shorter, more direct, and the content already conveys the concede-then-pivot move without the header needing to narrate it.
• Tightened historical precedent paragraph: Compressed the Jackson/Lincoln/FDR comparison. Cut "that is as old as the republic" and the compare-and-contrast framing. The peacetime distinction now lands as the punch rather than being buried in a subordinate clause.
• Removed "the most significant judicial check on executive power in years" from the hook per editorial note -- the specifics (6-3, Roberts, Gorsuch and Barrett) already convey significance without the cable-news summary register.

Voice Adjustments

• Framework introduction rewritten: Replaced "In finance, arbitrage means profiting from price differences across markets -- buying cheap in one place and selling dear in another" with a more conversational construction per editorial guidance. "Selling dear" was archaic/Economist register.
• Added register variation throughout: Inserted em dashes as pivots (the voice's signature punctuation), fragments as punches ("Arbitrage is invisible." as a standalone paragraph), and a parenthetical aside style. Increased rhythmic variation between long analytical sentences and short declarative ones.
• Rewrote zoom-out thesis: Replaced the single-sentence thesis statement with the editor's suggested split: "Here is the thing about defiance: everyone can see it." / "Arbitrage is invisible." -- then expands into the structural argument. This gives the article's most important claim the compressed, declarative delivery the corpus uses for its biggest moments.
• Rewrote CASA paragraph: Shed the legal-brief quality per editorial note. "Before CASA, one federal judge could freeze a policy nationwide. After CASA, you need class actions -- slower, more expensive, and by the time they grind through the system, the policy has been running for months."
• Rewrote "decorative judiciary" closer: Compressed "where the forms of judicial review persist while the substance drains away" to "all the forms of judicial review, none of the substance" per editorial suggestion.
• Rewrote framers line: Changed "That assumption was never codified because the framers could not imagine a president who would treat legal authority as interchangeable inventory" to "Nobody bothered to write that down, because who imagines a president treating legal authority like interchangeable inventory?" -- more conversational, less constitutional-commentary boilerplate.
• Replaced baseball metaphor extension: Changed "The batter stepped back up to the plate with a different bat" to "There are five more arms in the rotation" -- connects to the next section header and is more compressed per editorial note.
• Dropped "and saying so matters" after "Not all of these are equally alarming" -- the tiering throughout the article already does this work without the self-congratulatory metacommentary.
• Simplified honesty flag: Changed "I want to be honest about this, because the credibility of everything that follows depends on it" to "I want to be honest about this." Full stop. Let the concession speak for itself.
• Transition fix: Changed "Then there are the courts themselves" to "And then the courts did it to themselves" -- more voice, more punch, foreshadows the irony.

Unresolved Notes

• Just Security tracker numbers (655 lawsuits, 217 wins, 274 awaiting): These come from a living tracker that updates daily. The fact-checker flagged that current numbers may differ. The author should verify against the tracker on publication day and update if needed.
• Kavanaugh dissent quote: Widely reported in identical phrasing across multiple outlets but not verified against the full 63-page PDF of the opinion. The author should check the exact wording before publication.
• Word count: Final article runs approximately 1,500 words (body text, excluding revision log), in line with the target. The combination of relocating the Congress paragraph, compressing the historical comparison, and tightening the CASA section offset the slight overrun in the original draft.
• Abrego Garcia retaliatory prosecution: Not included per the original draft writer's judgment call -- at 1,500 words, a fourth case study risks turning The Rotation into a list. The three examples cover the spectrum from legitimate-but-aggressive to clear bad faith. Could be added if word count expands.
• DOJ whistleblower link: The CNN link in the draft (https://www.cnn.com/2026/01/29/us/federal-judges-trump-immigration-credibility) may not be the best source per the fact-checker. CBS News and NPR have more detailed reporting on Erez Reuveni. Author may want to swap the link.

Humanizer Notes

Patterns Found

This article had already been through extensive editorial revision, so the AI fingerprint was faint compared to raw AI output -- but still present at the sentence and paragraph level. The primary tells: (1) Sentence-level uniformity -- several passages, particularly in The Rotation and The Steelman, had sentences clustering in the 20-30 word range with minimal variation, creating that metronomic quality; (2) Symmetrical case-study structure -- the three examples in The Rotation followed an identical pattern (action, court blocks, pivot, result) with near-identical paragraph lengths, reading like a formatted list in prose clothing; (3) Residual formal register -- scattered phrases reading more like a legal brief than FTR voice ("authorize presidential tariff action," "the qualitative difference," "constitute 'constraint'," "standard proceedings"); (4) Missing contractions -- "does not," "were not," "is not" where the voice would naturally contract; (5) Uniform paragraph length -- most paragraphs sat at 3-4 sentences, missing the one-sentence punches and longer discursive passages that characterize the corpus; (6) Overly clean transitions -- "And this is where" and "And yet --" doing structural work that could be done with more variety.

Key Changes

• Broke the case-study symmetry in The Rotation: Changed "In immigration" opener to "Start with immigration" -- a direct, voice-appropriate instruction to the reader. Gave the funding-freeze example a different shape ("Read that again" as a two-word interjection). Changed "And then the courts did it to themselves" to "And then the courts kneecapped themselves" -- more aggressive, more in voice. Ended the CASA paragraph with a fragment/question construction instead of a declarative.
• Varied sentence length aggressively: Split comma-joined sentences into fragments ("655 lawsuits filed. 217 plaintiff wins."). Used colon-compressed constructions in the Steelman ("Duration: indefinite to 150 days. Scope: country-specific punitive rates to a uniform global floor.") to create staccato against the longer analytical sentences. Added "Think about that." and "Read that again." as two/three-word sentences for rhythmic punch.
• Injected contractions throughout: "were not" to "weren't," "does not" to "doesn't," "is not" to "isn't," "can not" to "can't," "we are" to "we're." The original had a formal-prose aversion to contractions that read as AI-generated even though the article was otherwise conversational.
• Strengthened editorial commitment: Added "That's not arbitrage. That's contempt with a press release" in the funding-freeze section. Changed "Legitimate authority-switching does not require lying to judges" to "You don't need to lie to judges if your authority-switching is legitimate." Added "and it's not a small one" after the peacetime distinction -- the original just dropped the point and moved on, where the voice would underscore it.
• Tightened formal register: "authorize presidential tariff action" became "hand the president tariff authority." "Constitute 'constraint'" became "count as 'constraint'." "The qualitative difference" became a construction that avoids the phrase entirely. "Approximately" became "about." "Simultaneously" stayed (it's natural here) but other Latinate formalities were swapped for Anglo-Saxon equivalents.

Confidence

High. The article was already strong from its editorial passes, and the humanization work here was surgical -- tightening rhythm, adding contractions, breaking symmetry, injecting the kind of brief editorial interjections ("Read that again." "Think about that.") that the corpus deploys at key moments. The voice now sits comfortably alongside the corpus. The Steelman's data-heavy tariff comparison still reads denser than typical FTR prose, but that's the content -- you can't make rate comparisons breezy without losing the argument, and the colon-compressed construction at least gives it rhythmic texture. The Rotation section improved the most: breaking the case-study template and adding reactive editorial voice ("That's contempt with a press release") makes it read like a mind working through examples rather than cataloging them at arm's length.