title: "The Ratchet: How America Learned to Wage War Without Permission" subtitle: "The House voted 212-219 to let an unauthorized war continue. A 75-year mechanism, completing its work." author: Rebecca Rowan publication: For the Republic date: 2026-03-15

The Ratchet: How America Learned to Wage War Without Permission

Seven Votes

212-219.

Seven votes. Not history, not principle -- math. That's the margin by which the House declined to assert its constitutional war power. Nearly every Republican voted to continue an unauthorized war. Nearly every Democrat voted to constrain it. And on the House floor, a West Point graduate and Army veteran named Warren Davidson -- an Ohio Republican, not a libertarian outlier -- told his colleagues: "The moral hazard posed by a government no longer constrained by our Constitution is a grave threat."

Then 219 of them voted the other way. The war continued.

Davidson didn't argue the strikes were wrong. He said explicitly: "Make no mistake, Iran is an enemy of the United States." His argument was narrower -- and honestly, more dangerous: that the mechanism matters regardless of whether you agree with the action. That a government which wages war without permission will eventually wage war without reason.

Two of his fellow Republicans agreed with him. Over 99% did not.

To understand what Davidson was defending, you have to see what was supposed to exist.

What Was Supposed to Exist

The founders weren't ambiguous about this. Madison, Hamilton, Mason -- men who agreed on almost nothing -- agreed on this: the power to initiate war belongs to Congress because the executive will always have the motive to start one. Madison put it plainly: "The Executive is the branch of power most interested in war, and most prone to it." Even Hamilton -- the most pro-executive founder by a wide margin -- wrote that "the legislature alone" can "place the nation in a state of war."

At the Constitutional Convention, the change from "make war" to "declare war" passed with only one state dissenting. The president could repel sudden attacks. Everything else required Congress. No prominent founding figure took the other side. And for roughly 150 years -- through the War of 1812, the Mexican-American War, both World Wars -- the system worked as designed.

Then it stopped.

Click by Click

Korea, 1950. Truman deployed troops while Congress headed for its July 4th recess, called it a "police action under the United Nations," and said "I just had to act as commander-in-chief, and I did." Senator Robert Taft protested the "usurpation" -- then said he'd vote for the force anyway. The Senate Democratic leader actively discouraged a vote, fearing "a lengthy debate." So Congress chose vacation. More than 36,000 Americans died in a war it never voted on. First click. It locked.

Vietnam was supposed to produce a fix. Congress overrode a presidential veto to pass the War Powers Resolution in 1973 -- the most aggressive assertion of congressional war authority since the founding. And then: zero enforcements. Every president since Nixon has rejected the WPR's constitutionality. It has never once forced a president to withdraw troops. A speed limit sign on a highway with no police. (Fifty-two years and counting, by the way. Perfect record.) That failure locked.

The 2001 AUMF. Sixty words. Passed 420-1 in the House, one week after 9/11. Those sixty words have since been used to justify military operations in 22 countries against organizations that didn't exist when the vote was taken -- including ISIS, a group that was literally at war with al-Qaeda. A blank check that never expires. Locked.

To be fair: the Gulf War. Bush I sought authorization in 1991 and got it. But ask why it worked: an opposition Congress, post-Vietnam leadership, a president who calculated compliance was necessary. Those conditions have narrowed with every decade since.

Libya, 2011. Obama argued that launching cruise missiles at a sovereign nation didn't constitute "hostilities" under the WPR. His own Office of Legal Counsel disagreed. He did it anyway. Locked.

Syria, 2017. Trump launched strikes. Congress barely noticed. Locked.

Iran, 2026. Full-scale war. No authorization. A party-line vote to continue. And an administration that simultaneously calls it a war and claims it doesn't need permission. JD Vance: "We are not at war with Iran, we're at war with Iran's nuclear program." Speaker Johnson: "We're not at war right now." Pete Hegseth: "The regime sure did change." Trump himself: "Why wouldn't there be a Regime change???"

"Police action." "Not hostilities." "Not a war." Each euphemism shrinks even as the wars get bigger. The vocabulary is doing structural work here -- each new phrase insulates the next expansion from the last constraint.

The mechanism has a name. It's a ratchet. Each click locks the previous expansion in place. Reversal requires dismantling the device itself. And no one with power has any incentive to dismantle it.

Working as Intended

Here's the coldest part of the diagnosis: the ratchet runs on congressional complicity as much as executive overreach.

Congress chooses not to exercise its war powers -- because strategic avoidance is politically rational for individual members. A vote on war is a vote that can end your career. Not voting is free. The Harvard Journal on Legislation calls this the "power to not decide": deliberate silence that lets legislators claim they neither authorized nor opposed whatever happens next.

The president benefits from precedent. Congress benefits from avoidance. Courts won't intervene -- they've consistently called war powers a "political question", and their silence gets legally interpreted as consent. Everyone inside the system has a reason to keep the ratchet turning. That's what makes it so durable.

Columbia's Matthew Waxman argues that Congress exercises informal influence through appropriations, hearings, and political signaling -- and historically, he's right. His examples include Vietnam, Lebanon, and Somalia, all cases where congressional pressure actually constrained a president. But every one of those examples comes from an era when members of Congress were willing to cross party lines on war. When over 99% of a caucus votes to continue an unauthorized war launched by their president, the informal checks Waxman describes don't weaken. They evaporate.

Now look at the nuclear argument, because it's the strongest case for what the administration did. Iran's enrichment was real. The IAEA confirmed it. This wasn't Iraq's WMDs. The intelligence community assessed a breakout timeline of weeks, not years. I'll grant that the case for military action was stronger than most.

But if the threat is genuinely grave -- if millions of lives are at stake -- that is more reason to involve Congress, not less. An AUMF would have made the strikes stronger -- more legitimate, more durable, harder to challenge. That's a strategic argument, not a peacenik one. Even the Bush administration, which wasn't known for constitutional humility, came to Congress for the Iraq AUMF. Even JINSA -- hawkish, pro-Israel, supportive of the strikes -- argued Congress should pass an AUMF for Iran because formal authorization would strengthen the mission. When JINSA is telling you to get an AUMF, the "no time" excuse is dead.

And now the WPR's 60-day clock is ticking. It expires in late April. When it does, there's no enforcement mechanism. No court will enforce it. Congress already voted against asserting its authority -- couldn't even muster a majority for what was, in the end, a non-binding resolution. 62% of Americans want congressional approval for further military action. Congress voted not to require it.

The last nominal constraint on executive war-making is weeks from expiring. Nobody is counting down.

A West Point graduate stood on the House floor, cited the Constitution, and lost by seven votes. The war continued. The 60-day clock ticks toward a deadline no one will enforce.

The next time a president launches a strike -- any president, any country, any reason -- the constitutional mechanism designed to prevent it will perform exactly as it did on March 5th. A vote. A party line. A shrug.

Another click.

Revision Log

Fact-Check Corrections

• RED: Waxman institutional affiliation. Changed "Yale's Matthew Waxman" to "Columbia's Matthew Waxman." He is a professor at Columbia Law School; the essay was published in the Yale Law Journal, which caused the confusion. The link to the Yale Law Journal essay is correct; the attribution is now correct.
• RED: Vote count. Changed "Then 218 of them voted the other way" to "Then 219 of them voted the other way." The actual nay count was 219 (215 Republicans + 4 Democrats). 218 was the number of non-defecting Republicans in the caucus, not the vote tally.
• YELLOW: "programme" to "program." Changed British spelling in Vance quote to American spelling, matching the actual quote from NBC News / ABC News. Updated link from Al Jazeera to NBC News as a more direct source for the Vance quote.
• YELLOW: "a single dissenting vote." Changed to "only one state dissenting" -- Constitutional Convention votes were by state delegation (8-1), not individual delegates.
• YELLOW: 99.1% stat. Softened from "Two out of 220 Republicans" / "99.1%" to "Two of his fellow Republicans" / "Over 99%." This avoids the potential vacancy-count issue with the 220 denominator while preserving the rhetorical force.
• YELLOW: Non-binding resolution. Added "couldn't even muster a majority for what was, in the end, a non-binding resolution" to the 62% paragraph. This actually strengthens the argument per the fact-checker's note.
• YELLOW: 36,000 figure. Changed to "More than 36,000" for precision, consistent with the "36,000+" in source material.
• YELLOW: "zero for 53" framing. Replaced "zero for 53" with "zero enforcements" to avoid sourcing ambiguity with the Constitution Center link. Added a parenthetical "(Fifty-two years and counting, by the way. Perfect record.)" -- serves double duty as a voice moment (sardonic aside) and a more defensible timeframe.

Structural Changes

• Added "Seven Votes" as section header for the opening, per outline and editorial suggestion.
• Renamed "The System That Works" to "Working as Intended." The irony is now legible on the surface rather than buried, per editorial note about section headers needing more personality.
• Added bridge sentence between opening and "What Was Supposed to Exist": "To understand what Davidson was defending, you have to see what was supposed to exist." This addresses the editorial note about the transition feeling like a scene cut rather than a zoom-out.
• Compressed Gulf War paragraph from 4 sentences to 2 sentences. Moved from a full paragraph with "An honest accounting requires a pause here" (flagged as out of voice) to a compressed inline beat: "To be fair: the Gulf War..." This restores the acceleration of Click by Click that the counter-example was disrupting.
• Tightened ratchet-naming moment. Cut "By the time you've watched this sequence" per editorial instruction. Now reads "The mechanism has a name. It's a ratchet." Trusts the reader, trusts the build.
• Softened section-ending line of "Working as Intended" from "the only question is whether anyone will notice" (which competed with the close) to "Nobody is counting down." Blunter, less polished, doesn't compete with the final "Another click."

Voice Adjustments

• Rewrote policy-paper sentence. "Congressional authorization would have strengthened the operation's legitimacy, shared responsibility across branches, and ensured a decision of this magnitude reflected democratic consent" (flagged as most out-of-voice sentence in draft) replaced with: "An AUMF would have made the strikes stronger -- more legitimate, more durable, harder to challenge. That's not a peacenik argument. That's a strategic one." Follows the editor's suggested rewrite closely.
• Added sardonic aside on WPR. "(Fifty-two years and counting, by the way. Perfect record.)" -- a parenthetical self-interruption in corpus voice, addressing the editorial note about needing register shifts and the WPR stat being a natural place for a blunter reaction.
• Fixed "dangerous" ambiguity. "His argument was narrower and more dangerous" changed to "His argument was narrower -- and honestly, more dangerous" per editorial note. The "honestly" is a corpus-pattern qualifier that signals the author thinking in real time.
• Tightened the ratchet-as-system line. "The ratchet isn't a failure of the system. It's the system optimized for the convenience of everyone who works in it -- and the inconvenience of everyone who doesn't" changed to "It's the system working exactly as everyone in it needs it to work." The mirrored clause was diluting the punch per editorial note.
• Replaced JINSA line. "When the responsible hawks are telling you the process matters, the 'we didn't have time' argument collapses" changed to "When JINSA is telling you to get an AUMF, the 'no time' excuse is dead." More offhand, more in voice per editorial note.
• Fixed "The United States House of Representatives." Shortened to "the House" per editorial note about C-SPAN graphic formality.
• Replaced "An honest accounting requires a pause here." Changed to "To be fair:" -- the author's actual concession pattern, not a self-conscious rhetorical announcement.

Unresolved Notes

• Davidson "Make no mistake" quote. The fact-checker flagged this as BLUE (verification needed) -- the exact wording could not be independently confirmed outside the source material. The CNN article is cited but the full transcript was unavailable. The author should verify from C-SPAN footage or the full CNN article before publication.
• Truman "I just had to act as commander-in-chief, and I did" quote. Also BLUE-flagged. Consistent with Truman's known statements but exact wording could be a paraphrase. Author should verify from the PBS source cited.
• 60-day clock start date. The fact-checker notes ambiguity about whether the clock started February 28 (hostilities began) or March 2 (notification letter). "Late April" is correct under either interpretation, but the author may want to pin this to "approximately April 29" if precision matters for the publication timeline.
• First-person voice. The editorial notes suggested one moment of personal "I" voice to ground the piece. I did not add one. The piece's emotional register is cold clarity (per the outline's instruction), and inserting a personal beat risked disrupting the autopsy tone that the structure demands. The author should decide whether a single first-person moment -- perhaps in the nuclear argument paragraph -- would strengthen or weaken the landing. This is a judgment call I'm leaving to her.
• Word count. The final article runs approximately 1,520 words (body text only, excluding metadata, revision log, and section headers), within the ~1,500 target.

Humanizer Notes

Patterns Found

This article had already been through extensive editorial and structural revision, so the worst AI tells (throat-clearing openers, tricolon abuse, hedge vocabulary, formulaic conclusions) were absent. The remaining fingerprints were subtle but real:

• Three "not X -- it's Y" constructions. The single most reliable AI tell appeared three times: (1) "This isn't sloppiness. It's engineering" in Click by Click, (2) "The ratchet isn't a failure of the system. It's the system working exactly as everyone in it needs it to work" in Working as Intended, and (3) "Congress isn't failing to exercise its war powers. Congress is choosing not to" in the same section. Also present in the subtitle: "That wasn't a failure. It was a 75-year mechanism completing its work." Four instances total across the piece -- well above the threshold where the pattern becomes detectable.
• Mechanical repetition in Click by Click. "That was the first click. It locked." / "That failure locked." / "That scope creep locked." / "That semantic dodge locked." / "That indifference locked." The "That [noun] locked" construction repeats six times with identical syntax. The concept of locking is central to the ratchet metaphor and should stay, but the syntactic uniformity reads as templated rather than building.
• "Reasonable people can disagree." Classic AI neutrality hedge. This phrase is almost never used by opinionated human writers; it's the language of manufactured equidistance. The piece had already committed to the analytical framework -- this sentence undercuts that commitment.
• "Notice the vocabulary." Teacher-voice instructional register. The corpus doesn't tell readers what to notice; it shows them the pattern and trusts them to see it.
• Minor missing contractions. "were not" where "weren't" fits, "there is no" where "there's no" would be natural, "was not known" where "wasn't known" matches the voice.
• Subtitle "not X / it was Y." "That wasn't a failure. It was a 75-year mechanism completing its work" uses the same negative-parallelism formula.

Key Changes

• Eliminated all "not X -- it's Y" constructions. (1) "This isn't sloppiness. It's engineering" became a single direct claim: "The vocabulary is doing structural work here -- each new phrase insulates the next expansion from the last constraint." Drops the negative half, states the positive claim. (2) "The ratchet isn't a failure of the system. It's the system working exactly as everyone in it needs it to work" became "Everyone inside the system has a reason to keep the ratchet turning. That's what makes it so durable." States the insight directly without the performed contrast. (3) "Congress isn't failing to exercise its war powers. Congress is choosing not to" became "Congress chooses not to exercise its war powers" -- one direct sentence instead of the negative-positive pairing. (4) Subtitle rewritten from "That wasn't a failure. It was a 75-year mechanism completing its work" to "A 75-year mechanism, completing its work." Drops the performed contrast entirely.
• Broke the "That X locked" repetition. Kept "It locked" for the first (Korea) and second (WPR) instances where the full phrase has weight. Shortened to just "Locked." for the AUMF, Libya, and Syria instances -- the acceleration mirrors the ratchet mechanism itself (each click gets faster, less ceremonious). Removed "That was the first click" before the Korea instance -- "First click. It locked." is tighter.
• Replaced "Reasonable people can disagree" with "I'll grant that the case for military action was stronger than most" -- first-person ownership of the concession, matching the corpus pattern of extending genuine charity before disagreeing (cf. "Maduro was a dictator" in Flash Point, "I'll grant that freely" in Racket).
• Removed "Notice the vocabulary." The vocabulary list that follows ("Police action." "Not hostilities." "Not a war.") does the work on its own. Trusts the reader.
• Added contractions where the voice naturally contracts: "weren't" for "were not," "there's" for "there is," "wasn't" for "was not," "comes from" for "is from."
• Reworded "That's not a peacenik argument. That's a strategic one." to "That's a strategic argument, not a peacenik one." The original was another near-"not X, it's Y" in disguise (negative framing first, positive second). The rewrite leads with the positive claim.

Confidence

High. This article was already well-crafted -- the ratchet framework, the Click by Click acceleration, the cold closing are all strong and distinctly in voice. The humanization work was surgical: eliminating the three (plus subtitle) "not X, it's Y" constructions, breaking the mechanical repetition pattern in Click by Click, replacing the one neutrality hedge with first-person concession, and adding contractions throughout. The closing remains untouched because it already works -- "Another click" lands as an image rather than a summary, and it's structurally distinct from recent piece endings (Buying Both Doors ended on reflective first-person, Workaround ended on a diagnostic question, Racket ended on Butler's legacy). The Click by Click section improved the most: the accelerating truncation of the "locked" beats (from full sentence to single word) now performs the ratchet's acceleration rather than just describing it. The nuclear-argument paragraph was the second biggest improvement -- replacing the hedge with an "I'll grant" construction adds the first-person beat the editorial notes called for while staying in the piece's cold-clarity register.