19 March 2026
General Henry H. “Hap” Arnold: The Handled Prophet
A psychoanalytic dissection of The United States News, May 31, 1940, and the handler network behind the man on its cover
The magazine cost ten cents. It landed on desks across Washington the last week of May 1940, while the Wehrmacht pushed the British Expeditionary Force into the sea at Dunkirk and the French Third Republic measured its remaining life in days. On the cover, a major general grinned and gave a thumbs-up. Major General Henry H. “Hap” Arnold. The tagline read “Wings Over America.” Inside, twenty-six pages of news, maps, cartoons, editorials, and advertisements performed a single coordinated act: they told the American ruling class what to fear, whom to blame, and where to invest. Not a word of it was accidental.
This is a psychoanalysis of that magazine. Not a review. A dissection—scalpel to the page, forceps to the subtext, and a long hard look at the men behind the man on the cover.
The Instrument
The United States News was founded in 1933 by David Lawrence, a Princeton-educated conservative journalist who had studied under Woodrow Wilson and voted Republican in every presidential election from Hoover forward (Lawrence biography). Lawrence designed the publication for what he called “community leaders, businessmen, and politicians”—a self-selected readership of American elites who needed to know not just what happened, but what to do about it. By 1940, it was the only weekly magazine in Washington devoted exclusively to national affairs. It would later merge with World Report to become U.S. News & World Report.
Lawrence was no neutral observer. His 1934 book Beyond the New Deal attacked Roosevelt’s domestic programs. Richard Nixon awarded him the Presidential Medal of Freedom in 1970. The editorial line of the May 31, 1940, issue was not journalism shaped by events. It was ideology selecting which events to amplify and which to bury.
The Architecture of Fear
Open the magazine and the system reveals itself within three pages. The cover story: “The Truth About Our Neglected Air Program.” The lead feature: “What a British Defeat Would Mean for U.S.” The political article: “New Deal Refuses to Yield Defense Control.” Three headlines. One thesis. America is naked, Britain is falling, and the New Deal did it.
Every section reinforces the thesis from a different angle. The news articles deliver fear through maps—Nazi flags planted on Caribbean islands, bomber arrows aimed at the English coast. The centerfold editorial by Lawrence himself, “The Defense We Cannot Buy,” argues that America’s moral softness matters more than its military weakness. The political section gives generous space to isolationist voices, most notably Charles Lindbergh—the man who accepted a medal from Hermann Goering in 1938 and was captioned here simply as offering “pointed comment.” The business pages calmly calculate which industrial sectors will profit from rearmament. And the full-page pictogram on pages 18–19 delivers the kill shot: $13.75 billion spent on relief could have bought 100,000 warplanes, while only $634 million went to actual aircraft.
The math is dishonest. The comparison is false. Relief spending fed millions during the worst depression in American history. Diverting it to bombers would not have built an air force, it would have built a revolution. But the pictogram does not care about nuance. It cares about the feeling in the reader’s gut when the numbers land.
The Trojan Horse
The most dangerous article in the issue is not the loudest. “Congress Takes Offensive Against ‘Trojan Horses’” runs on page 15, beneath a photograph of Attorney General Robert Jackson. The article details expanded FBI surveillance, alien registration, and the transfer of the Immigration Service to the Department of Justice. It treats every immigrant as a potential saboteur and conflates actual Nazi espionage with the existence of foreign-born communities in American cities.
The word “Trojan Horse” does the heavy lifting. It transforms a human being into a concealed weapon. It dehumanizes without requiring the reader to acknowledge the dehumanization. Twenty months later, Executive Order 9066 used precisely this framework to intern 120,000 Japanese Americans. The intellectual groundwork was being laid here, in a ten-cent weekly read by the men who would sign the orders.
The Man on the Cover
Henry Harley “Hap” Arnold grins from the cover in a slight pose, as if gently convincing his opponent. The interior profile on page 33 calls him “Tool-Bearer of the Air” and “Key Man on Military Affairs in the House.” It traces his career from 1911, when the Wright Brothers taught him to fly, through his exile to Kansas for supporting Billy Mitchell, to his appointment as Chief of the Air Corps in September 1938 after General Oscar Westover died in a plane crash (Arnold biography).
The profile is admiring to the point of recruitment. Arnold is the visionary the bureaucracy ignored, the soldier who saw what the politicians could not. Paired with the Billy Mitchell martyrdom narrative on pages 9–10—Mitchell’s ghost invoked under the headline “How Inertia in the High Command Has Kept Us Weak in Aviation”—the message is unmistakable: listen to the warriors, not the civilians. Fund the planes. Get the New Dealers out of the way.
The question is whether Arnold was the architect of this message, or its instrument. Nine years later, in his autobiography Global Mission (Arnold, 1949), Arnold would answer the question himself—though he never understood what he was confessing.
The Handler Network
Arnold was a blunt, emotional officer who ran hot and said what he thought. Roosevelt knew this. In March 1940—two months before this cover story—the President admonished Arnold directly at a White House meeting, saying: “When people can’t control themselves and their people under them, you know what we do with those kind of people?” (Arnold vs. FDR). Arnold was in Roosevelt’s doghouse when this magazine went to press. His forced retirement seemed imminent.
A man under that kind of pressure does not orchestrate a sophisticated propaganda campaign through a national magazine. Someone else was steering. And Arnold told us exactly who. In Global Mission, he identified three men who “helped me most with my job”—and in doing so, he wrote the confession that no prosecutor could have extracted from him alive.
The first was George C. Marshall, who became Army Chief of Staff in September 1939 and immediately took Arnold under his institutional wing. Arnold called Marshall “one of the most potent forces behind the development of real American airpower” (Marshall-Arnold relationship). The word potent is Arnold’s own. Not “supportive.” Not “helpful.” Potent—the language of force, of power exercised from behind. Marshall protected Arnold from Roosevelt’s wrath, insisted Arnold sit on the Joint Chiefs of Staff—an unprecedented elevation for an air commander—and ran interference between Arnold’s aggressive buildup and the civilian leadership’s caution. Marshall later said he tried to make Arnold “chief of staff of the air without any restraint, though he was my subordinate.” That sentence is the architecture of handling distilled to its essence: total operational freedom granted by the man who holds the leash.
The second was Robert A. Lovett—and this is where the money enters the picture. Lovett was a partner at Brown Brothers Harriman, the most powerful private investment bank on Wall Street. He was Yale Class of 1918, Skull and Bones, a Navy pilot in the Great War who had flown bombing missions with the British in France (Lovett profile). In December 1940, Lovett entered the War Department as special assistant to the new Secretary of War. By April 1941 he was Assistant Secretary of War for Air—Arnold’s direct civilian supervisor.
Listen to what Arnold wrote about Lovett in Global Mission. He called Lovett “a partner and teammate of tremendous sympathy and of calm and hidden force.” Read those last three words again: calm and hidden force. Arnold then confessed that Lovett possessed qualities “in which I was weakest.” This is a five-star general—the only man in history to hold that rank in the Air Force—admitting in his own published memoir that a Wall Street banker compensated for his personal deficiencies. Arnold conceded that when he got “very agitated over a problem, Lovett calmly dissected the issues involved” and steered him toward rational solutions (Lovett as key aide). Arnold was the engine. Lovett was the steering wheel. And Arnold wrote it down in a book that anyone could buy for three dollars.
The third was Henry L. Stimson, appointed Secretary of War by FDR in July 1940—six weeks after this magazine appeared. Stimson was also Yale, also Skull and Bones, a Wall Street lawyer who had served as Secretary of War under Taft and Secretary of State under Hoover. He was a Republican interventionist who broke with his party’s isolationists (Stimson profile). Stimson created the Assistant Secretary of War for Air position specifically to install Lovett. He brought his entire “Kindergarten” into the War Department: Lovett, John J. McCloy, Harvey Bundy. These men formed a shadow architecture within the defense establishment—Wall Street bankers and lawyers running the American war machine from inside the Pentagon.
When Arnold fell into Roosevelt’s doghouse, it was Stimson and Lovett who rescued him. They arranged for Arnold to visit England in April 1941, where he witnessed the Blitz firsthand and returned convinced that strategic bombing could defeat Germany. Stimson called Arnold’s subsequent briefing to Roosevelt “an admirable statement.” The trip rehabilitated Arnold’s standing with the President. The handlers had cleaned up their asset and put him back in play.
The Brown Brothers Harriman Problem
The same investment bank that placed Robert Lovett at Arnold’s right hand had other partners and other histories. Brown Brothers Harriman managed the Union Banking Corporation, which served as the U.S. financial pipeline for Fritz Thyssen—Adolf Hitler’s earliest major industrial financier. Through Union Banking, an estimated $8 million in gold moved between New York and Germany during the 1930s. Three million dollars was funneled to the Nazi Party for the crucial 1932 German election. Partner Prescott Bush managed these accounts. In 1942, the U.S. government seized Union Banking’s assets under the Trading With the Enemy Act (Union Banking seizure). Bush failed to fully divest from “enemy national” relationships until 1951.
To restate the geometry: the same bank that financed the rise of Adolf Hitler placed its partner as the man who ran American air power. Lovett came from the firm whose other partners were being investigated for trading with the enemy. Arnold called this man his indispensable teammate—the one who possessed the qualities “in which I was weakest.” This does not make Lovett a Nazi sympathizer. It makes him a member of an American financial elite so concentrated that the same institutions ended up on both sides of the Atlantic ledger—financing the Wehrmacht’s expansion and building the bombers that would turn it to ash.
The Air Corps Tactical School
Arnold’s handlers were not only individuals. He was also handled by an institution. The Air Corps Tactical School at Maxwell Field, Alabama, operated from 1931 to 1940 as a factory for doctrinal conformity. Its faculty—devotees of Billy Mitchell—developed the theory of strategic bombing that Arnold championed: heavily armed bombers penetrating enemy airspace without fighter escort, precision strikes on industrial targets to collapse an enemy’s war-making capacity (ACTS history). Three-quarters of all Army Air Forces generals in World War II graduated from this school. Arnold’s inner circle—Carl Spaatz, Ira Eaker—came from the same pipeline, the same friendships forged at Rockwell Field in 1919. When Arnold pushed for expansion, he was not a lone voice. He was the spokesperson for a twenty-year institutional movement that had systematically captured the doctrine, the training pipeline, and the procurement system.
The doctrine turned out to be partly wrong. Unescorted daylight bombing over Germany was a catastrophe until long-range fighters arrived. But the doctrine served its real purpose: it justified an independent, massively funded air arm. The ideas were the vehicle. The institution was the engine.
The Confession
Global Mission is 626 pages of inadvertent testimony. Arnold published it in 1949, one year before his death, believing he was writing a triumphant memoir. He was writing an autopsy of his own autonomy.
The book reveals a man who was never alone in any room that mattered. Arnold describes how Billy Mitchell gave him the doctrine in the 1920s—how Mitchell physically got Arnold back into the cockpit in 1916 after Arnold had grounded himself from fear of flying. Mitchell was the ideological handler: the man who programmed Arnold’s core beliefs about air power before Arnold had the rank to act on them. Arnold nearly suffered a court-martial of his own for lobbying Congress and the press on Mitchell’s behalf. He had learned early that the message mattered as much as the machine.
By the time Arnold became Chief of the Air Corps in 1938, he was already a practiced propagandist. In the mid-1920s, he wrote six books promoting military aviation to young readers—seeding the next generation with air power ideology before they were old enough to vote. In 1941, he co-authored Winged Warfare with Ira Eaker, another ACTS graduate. Arnold fought his war, as one scholar noted, “not in the field but in Congress, on the Army General Staff, in factories, and in universities.” He was a general who understood that the battle for funding was fought on paper before it was fought in the sky.
But Global Mission reveals the crucial limitation. Arnold could generate passion. He could not generate patience. He could inspire a room and then destroy the goodwill by losing his temper before the handshakes finished. Roosevelt saw this instantly and nearly ended Arnold’s career over it. The handlers saw it too—and instead of discarding the asset, they built a support structure around his volatility. Marshall gave him institutional cover. Lovett gave him emotional regulation. Stimson gave him political rehabilitation. The ACTS gave him an officer corps that thought as one unit. Arnold’s genius was real. His independence was not. Every move he made in Global Mission has a handler’s fingerprint on it, and Arnold records each one with the gratitude of a man who never understood he was documenting his own leash.
Consider the pivotal meeting Arnold describes from September 28, 1938—just days after he became Chief of the Air Corps. Roosevelt told his advisers he wanted 10,000 aircraft procured, up from a planned 178, and factories capable of producing 20,000 per year. Arnold’s diary entries, preserved in the Huston edition (Arnold diaries), show a man scrambling to execute orders he did not originate. The AAF grew from 43,000 personnel at the end of 1939 to nearly 300,000 by Pearl Harbor. The AWPD-1 war plan—calling for 63,000 aircraft including 7,500 heavy bombers—was drafted not by Arnold but by Harold George, Kenneth Walker, and Haywood Hansell, all ACTS products. Arnold signed the plan. He did not write it. He was the delivery system, not the payload.
The Magazine as Intelligence Product
Return now to the magazine itself. The “Washington Whispers” column on page 36 is not journalism. It is a political intelligence briefing: cabinet reshuffles predicted, CCC militarization explored, Federal Reserve inflation policy leaked, war-production priorities outlined in bullet points. Any German embassy officer reading it would receive a detailed map of American decision-making, military weakness, political divisions, and industrial capacity—for ten cents at any Washington newsstand.
The “Tomorrow” prediction column on the yellow pages goes further. It tells wealthy readers which commodities to buy, which sectors to invest in, and reassures them explicitly: “Business will not be called upon to give away their product.” War will be profitable. The New Deal will yield. Capital will be rewarded.
Even the satire column—“Mr. Hitler Goes to Washington”—functions as an orientation guide. Written as a letter from a fictional parachute correspondent to Hitler, it describes specific Washington landmarks, government buildings, and social patterns. Played for laughs. Operationally useful.
The Advertisers’ Testimony
Follow the money. Anaconda Copper Mining Company—the inside front cover—wraps itself in the language of civilization while copper is the essential material for shell casings, electrical wiring, and ship plating. General Electric runs a nostalgic Edison advertisement while positioning itself as the industrial backbone of American life. Nu-Blue Sunoco, owned by the Pew family—major financiers of conservative and isolationist causes—buys a full color page. Lucky Strike takes the back cover with a color photograph that cost more than most families earned in a month.
The advertiser base is heavy industry and commodities: companies that would profit enormously from rearmament. The editorial content arguing for military buildup serves the commercial interests that pay the magazine’s bills. The final page says it plainly: “THE UNITED STATES NEWS is a good example of a selective medium. That is why more than 130 important advertisers use its pages.” The magazine was selling its readers to its advertisers. The editorial line was the product that made those readers receptive.
The Verdict
Arnold did not orchestrate this. He did not need to. He was a brilliant aviator with a volatile temperament and a genuine vision for American air power—and he told us so himself. In Global Mission, he described Lovett as possessing the qualities “in which I was weakest.” He called Marshall the most “potent force” behind American air power. He praised Stimson for creating the bureaucratic architecture that made the air buildup possible. He named the three men who ran him and thanked each one by name. The confession is in the acknowledgments.
His handlers—Marshall from the military side, Lovett and Stimson from the Wall Street–Skull and Bones corridor—positioned him, protected him, funded him, and aimed him. The Air Corps Tactical School gave him an officer corps that thought as one. The aircraft manufacturers, led by his lifelong friend Donald Douglas—whose daughter married Arnold’s son in 1944—gave him the machines. And David Lawrence’s magazine gave him the cover.
The May 31, 1940, issue of The United States News is not a historical curiosity. It is a working blueprint for how American propaganda actually functions: not through crude lies or state censorship, but through the alignment of editorial framing, selective fact presentation, visual fear, and commercial interest under the banner of objective journalism. Every section—news, analysis, opinion, cartoons, advertisements, letters—pushes the same direction while maintaining the appearance of independent thought.
The most dangerous propaganda is the kind that believes it is telling the truth. Lawrence and his editors probably did not think they were propagandists. They thought they were clear-eyed realists cutting through New Deal sentimentality. That sincere conviction made their propaganda more effective than any cynical fabrication could have been.
And the man on the cover kept grinning. Thumbs up. Wings over America. He would go on to command the largest air force in human history, suffer five heart attacks before the war ended, and retire to a ranch in Sonoma where he raised Hereford cattle and wrote a 626-page memoir confessing to everything while understanding nothing. The machine was running. The handlers had done their work. The instrument did not need to understand the system to serve it—and in Global Mission, Arnold proved that he never did.
RESONANCE
Air & Space Forces Magazine. (2006). “Lovett.” Air & Space Forces Magazine.https://www.airandspaceforces.com/article/0906lovett/. Summary: Profiles Robert Lovett’s role as Arnold’s civilian counterpart and the Wall Street banker who translated air power vision into industrial production.
Air & Space Forces Magazine. (2009). “FDR and Hap Arnold: Commander and Chief.” Air & Space Forces Magazine.https://www.airandspaceforces.com/PDF/MagazineArchive/Documents/2009/December%202009/1209commander.pdf.Summary: Documents the volatile but ultimately productive relationship between Roosevelt and Arnold, including FDR’s direct threat to fire Arnold in 1940.
Air & Space Forces Magazine. (2011). “When Arnold Bucked FDR.” Air & Space Forces Magazine.https://www.airandspaceforces.com/article/1101arnold/. Summary: Details the confrontation between Arnold and Roosevelt over aircraft allocation to Britain versus American rearmament.
Arnold, Henry H. (1949). Global Mission. Harper & Brothers. https://archive.org/details/globalmission0000hhar.Summary: Arnold’s autobiography and inadvertent confession. Contains his identification of three men who “helped me most,” his description of Lovett as possessing qualities “in which I was weakest,” and detailed accounts of the handler network that managed his career from 1938 to 1945.
Boyne, Walter J. (2003). “The Tactical School.” Air & Space Forces Magazine.https://www.airandspaceforces.com/article/0903school/. Summary: Documents the Maxwell Field institution where 261 of 320 Army Air Forces wartime generals were trained and the strategic bombing doctrine of high-altitude daylight precision attack was codified under Billy Mitchell’s intellectual heirs.
Britannica. (2024). “Henry Harley Arnold.” Britannica. https://www.britannica.com/biography/Henry-Harley-Arnold.Summary: Authoritative biography of the only five-star Air Force general, documenting his 1911 flight training under Orville Wright, advocacy for Billy Mitchell’s air power doctrine, and succession to Chief of Air Corps after Westover’s 1938 crash.
Britannica. (2024). “Union Banking Corporation.” Britannica. https://www.britannica.com/topic/Union-Banking-Corporation. Summary: Documents the 1942 seizure of Union Banking Corporation assets under the Trading With the Enemy Act, including connections to Fritz Thyssen and Brown Brothers Harriman partners.
Eleanor Roosevelt Papers Project. (n.d.). “David Lawrence (1888–1973).” The George Washington University. https://erpapers.columbian.gwu.edu/david-lawrence-1888-1973. Summary: Biographical profile of the conservative publisher of United States News, documenting his anti-New Deal stance and Republican political alignment.
George C. Marshall Foundation. (n.d.). “Marshall and Henry ‘Hap’ Arnold.” Marshall Foundation. https://www.marshallfoundation.org/articles-and-features/marshall-henry-hap-arnold/. Summary: Details the close working relationship between Marshall and Arnold from 1938 forward, including Marshall’s insistence on Arnold’s inclusion in the Joint Chiefs.
George C. Marshall Foundation. (n.d.). “Robert Lovett: A Man of Character and Ability.” Marshall Foundation. https://www.marshallfoundation.org/articles-and-features/robert-lovett-a-man-of-character-and-ability/. Summary:Profiles Lovett as one of the three men Arnold credited with helping him most, documenting Lovett’s Brown Brothers Harriman background and his role calming Arnold’s temperament.
Huston, John W. (Ed.). (2004). American Airpower Comes of Age: General Henry H. “Hap” Arnold’s World War II Diaries. Air University Press. https://www.airuniversity.af.edu/Portals/10/AUPress/Books/B_0085_HUSTON_AMERICAN_AIRPOWER_DIARIES.pdf.Summary: Arnold’s wartime diaries, including his identification of three men who helped him most and the operational dynamics of air power leadership.
Miller Center, University of Virginia. (n.d.). “Henry L. Stimson (1940–1945).” Miller Center. https://millercenter.org/president/fdroosevelt/essays/stimson-1940-secretary-of-war. Summary: Documents Stimson’s appointment as Secretary of War as a bipartisan interventionist move, his Skull and Bones background, and his creation of the War Department’s civilian aviation oversight structure.
5 March 2026
ARC: ACCOUNTABILITY RESPONSIBILITY CHANGE
There is a geometry to a life rebuilt.
Not the geometry of straight lines, progress marching obediently from point to point, milestone to milestone, the tidy narrative we tell at dinner parties when we want to seem as though we have our bearings. That geometry is a lie. Or at best, a simplification so severe it becomes one. Lives do not move in straight lines. They arc. They curve under weight, they bend toward gravity, they describe shapes that can only be understood in their entirety, not from inside the movement, but from the far shore, looking back at the trajectory the whole passage drew across the sky.
ARC is not a metaphor borrowed for comfort. It is the load-bearing structure of conscious human evolution, three movements in a symphony that does not resolve so much as transform, that does not conclude so much as become. The person who walks its full length is not the person who began it. That is not a promise. That is the mechanism.
The First Movement is Accountability
Understand what Accountability is not. It is not guilt. Guilt is a closed system: self-referential, circular, a wound that feeds on itself and calls the feeding introspection. Guilt says I am terrible and then waits to be absolved, or not absolved, and either way returns to the same sentence. Guilt does not move. It rehearses.
Accountability moves.
Accountability is the soldier’s after-action report written without mercy and without excuse. It is the scientist’s notation when the experiment fails: here is what I did, here is what resulted, here is the gap between intention and outcome. Subject. Verb. Object. No passive constructions distributing blame into the ambient air. No adjectives softening the verdict. I did this. This happened because of what I did. These are the specific consequences, named and numbered and refused the comfort of abstraction.
This is cold work. It is the coldest work a person can do. To stand in front of the actual record, not the story you have been telling about the record, not the version you constructed for people whose opinion matters to you, but the unedited document, and read it without looking away requires a particular species of courage that has no glamour to it. There is no audience for this moment. There should not be. Accountability performed for an audience is theater. Accountability done honestly is forensic. And it is painful and long lasting. As it should be to permanently alter one’s neurochemistry and subsequent behavior.
The first movement establishes the theme. Raw. Unornamented. The instrument playing alone in the silence before the orchestra enters.
This is what is.
The Second Movement is Responsibility
Here is where every framework built on good intentions collapses. Accountability without Responsibility is a confession with no consequence, a court that delivers the verdict and dismisses without sentencing, a reckoning that stops at the moment of recognition and calls recognition enough. It is not enough. It has never been enough. Knowing what you did and committing to what you now owe are not the same act. They are separated by the most demanding passage in the symphony, the development section where the original theme is tested against every complication reality can introduce, where it breaks apart and has to be rebuilt stronger or abandoned as insufficient.
Responsibility asks a harder question than Accountability. Accountability asks: what did I do? Responsibility asks: given what I have owned, what do I now owe?
Not to the abstract. The abstract is where Responsibility goes to die, dissolved into vague commitments to be better, to do better, to try harder, language so frictionless it slides past the actual obligation without catching on anything. Responsibility is specific. It names the person across the table. It names the relationship damaged, the trust broken, the debt incurred. It identifies the mission—the actual mission, not the aspirational version—and asks whether the one who stands here now, having made the accounting, is genuinely willing to execute it.
Willingness is not the same as desire. A person may not desire the work Responsibility demands. The debt may be too painful to discharge. The mission may require sustained effort against sustained resistance, without the comfortable blanket of adrenaline of crisis to carry you through. This is where character is not revealed but made. Crisis reveals what you have already built. The long, unglamorous commitment to discharging a debt that no one is watching you discharge—that is where character is constructed, brick by brick. In the bitter cold. In the stony silence. The secret open space where no one else ventures.
The second movement deepens the theme. Complicates it. Turns it against itself. The listener who thought they understood the first movement discovers they understood only its surface. The full weight arrives now, and it does not arrive gently.
The Third Movement is Change
Not intention. Not aspiration. Not the plan committed to paper in the good feeling that follows a moment of honest reckoning. Change is the evidence. It is the behavior that has actually altered, the pattern that has actually broken, the trajectory that has actually bent—bent by deliberate force applied repeatedly over time, not once in a moment of inspiration, but again and again in the ordinary moments when no one is watching and nothing is at stake except everything.
Change is what a symphony’s third movement does to its first. Play the opening theme of Beethoven’s Fifth after you have heard the whole. It is not the same theme. The notes are identical. The theme is not. The third movement has done something to the first, retroactively across time so that the beginning sounds different once you have heard where it was going. The listener has been altered. The music is the instrument of that alteration. It is the energy of change.
This is what Change does to the life that has walked the full arc. The original wound—the failure, the rupture, the moment that made Accountability necessary—does not disappear. It remains in the score. But it no longer plays the same way. It no longer carries the same weight. The person standing at the far end of the arc looks back at the event that began the first movement and hears it differently, because they have heard what it became. The wound does not vanish. It resonates.
This is the distinction that matters most and is most often missed. Change is not erasure. The man who has walked ARC is not a man without a history. He is a man whose history has been reconfigured—not edited, not suppressed, not managed with the sophisticated emotional vocabulary of someone who has read enough therapy literature to sound healed while remaining exactly as they were. Reconfigured. The weight redistributed. The meaning altered. The trajectory genuinely bent in a new direction that the earlier self could not have found without the passage through both prior movements.
ARC is not a three-step program. Programs advertise endpoints to give participants hope. And that is bullshit because it gives people a way out. ARC is an orientation toward life—a decision, made once and then re-made every day, to engage the material of one’s own existence without flinching, to refuse the comfort of the abridged version, to insist on the full accounting and then to act on what the accounting reveals. It is the application of the absolute-value principle to human experience: the distance from zero is always positive, regardless of direction. What happened to you, what you did, what was done in your name or by your hand—all of it, converted. Not forgotten. Converted. The negative sign dropped, the magnitude preserved, the energy redirected toward what will be built.
The symphony does not ask whether you deserved the first movement. It asks whether you will complete the third. And then go back and re-examine the first.
Most people do not. Most people stop somewhere inside the second movement, in the complicated middle section where the theme has lost its initial clarity but the resolution has not yet arrived, and they call that stopping place their permanent address. They live in the development section, neither the honest simplicity of the opening nor the earned transformation of the close. They become fluent in the language of complexity without ever moving through it. They speak brilliantly about the wound. They do not heal it.
ARC will not permit that.
ARC demands completion not because completion is comfortable but because the human being has the capacity for it, and what has the capacity for evolution and refuses it does not simply stay the same. It diminishes. The refusal of the arc is not stasis. It is a slow descent in the direction of the wound’s original gravity, back toward the beginning, having learned nothing the hard way could not have taught.
You were built for the third movement.
The question is whether you will play it and be that change. Live it. Because you are the one who wrote it.
[Thank you to GS for inspiring this piece.]
26 February 2026
Greenland: From Real Estate Interest to Military Reality
The Ghost of William Seward
In 1867, U.S. Secretary of State William Seward was lambasted for “Seward’s Folly”–the purchase of Alaska from the Russian Empire for $7.2 million. History had the last laugh. Today, we are witnessing a historical echo of strategic consequence.
On January 19, 2026, the North American Aerospace Defense Command (NORAD) announced that aircraft would arrive at Pituffik Space Base in Greenland to support “long-planned NORAD activities.” The announcement, coordinated with the Kingdom of Denmark, marks a pivotal moment: the “Greenland Gambit” has transitioned from a diplomatic curiosity into a hard-power imperative.
While strategic attention fixates on the Taiwan Strait–the “Front Porch” of Pacific competition–the Arctic quietly emerges as the decisive theater of the next decade. The arrival of NORAD assets in Greenland confirms what defense planners have long understood: the “Basement” of North American security demands immediate reinforcement.
The “Basement” vs. The “Front Porch”
If the Taiwan Strait is America’s front door, the Arctic is the mechanical room. For decades, the Arctic was protected by a ceiling of impenetrable ice. That ceiling is collapsing.
“The shortest route for a Russian ballistic missile to reach the continental United States is via Greenland and the North Pole,” notes Otto Svendsen, associate fellow with the Center for Strategic and International Studies. This geographic reality places Greenland at the center of gravity for early warning and missile defense.
Russian activity in the GIUK Gap (Greenland-Iceland-UK) has reached a post-Cold War high. A December 2025 report by the Bellona Foundation revealed that 100 sanctioned vessels–comprising a “shadow fleet” of oil tankers and LNG carriers–traversed Russia’s Northern Sea Route during 2025, up from just 13 in 2024. These vessels operate under compromised flags, frequently disable their Automatic Identification System transponders, and carry inadequate insurance. This illicit corridor threatens environmental catastrophe in one of Earth’s most fragile ecosystems while simultaneously demonstrating Moscow’s willingness to weaponize commercial shipping lanes.
China has positioned itself as a “Near-Arctic State” since its 2018 Arctic Policy white paper, seeking to secure shipping routes that reduce transit times to Europe by up to 50% compared to the Suez Canal. In September 2025, Chinese state media celebrated the maiden voyage of the “Arctic Express”–a container ship completing the China-to-Europe run in just 18 days via the Northern Sea Route. As OilPrice.com observes, Greenland is growing in importance from a missile-defense, space, and global competition perspective.
The Gray Zone: The Mineral-Military Pipeline
Irregular warfare is won below the threshold of kinetic conflict. In Greenland, this “Gray Zone” is defined by resource sovereignty.
Rare Earth Monopolies. The Tanbreez deposit in Southern Greenland represents one of the world’s largest rare earth reserves, with an estimated 28.2 million metric tons of rare earth material–over 27% of which consists of the heavy rare earths critical to defense applications. In June 2025, the U.S. Export-Import Bank issued a $120 million letter of interestfor the project under its Supply Chain Resiliency Initiative, marking the first overseas investment in a mining venture under the current administration.
The strategic imperative is stark: China controls nearly 90% of global rare earth processing capacity and approximately 99% of heavy rare earth processing. Every F-35 Lightning II requires 920 pounds of rare earth materials. Every Virginia-class submarine depends on rare earth permanent magnets for propulsion and targeting systems. Every precision-guided munition in the American arsenal contains components that currently flow through Chinese refineries. In April 2025, Beijing imposed export controls on seven critical rare earth elements in response to U.S. tariffs–a reminder that resource dependency is a vulnerability that adversaries will exploit.
Infrastructure and Undersea Cables. Control of Greenlandic ports provides essential protection for the undersea cables that carry over 95% of global internet traffic and facilitate more than $10 trillion in daily financial transactions. The Greenland Connect cable system—a 4,600-kilometer fiber optic network linking Greenland to Iceland and Canada—represents critical infrastructure for transatlantic communications.
Russian vessels equipped with advanced surveillance technologies and remotely operated underwater vehicles have been observed operating near undersea cable routes in the North Atlantic, raising concerns among NATO allies about potential sabotage. In January 2026, German authorities blocked the shadow fleet tanker Tavian after discovering forged registration documents—the vessel was suspected of reconnaissance activities near critical Baltic infrastructure. A successful attack on undersea cables could cripple government communications, destabilize financial markets, and degrade military command-and-control networks. The cables have no redundancy in the Arctic corridor; Greenland’s position makes it the logical anchor point for a protected, hardened communications architecture.
The Physics of Arctic Warfare: Waveforms and Wastes
As a biophysicist, I see the Arctic as a complex field of waveform dynamics. Proximity to the North Magnetic Pole creates ionospheric chaos, causing GPS signals to wander unpredictably. Solar storms that would cause minor disruptions at lower latitudes can render satellite navigation entirely unreliable in polar regions. Greenland provides the only stable terrestrial “anchor” for ground-based augmentation systems required for precision navigation and targeting–capabilities that hypersonic defense and space domain awareness increasingly demand.
Furthermore, we must account for the biological cost of sustained Arctic operations. A January 12, 2026, study published in Scientific Reports by researchers at National Jewish Health provided the first quantitative evidence linking deployment exposures to measurable lung damage: veterans with deployment-related lung disease had anthracotic (carbon-based) pigment levels more than three times higher than healthy controls, with the burden strongly associated with burn pit smoke exposure. This finding underscores a broader operational truth: we cannot ignore the molecular integrity of our service members or their equipment. At -40°C, where lubricants congeal and metal becomes brittle, where batteries drain in hours and exposed skin freezes in minutes, deterrence becomes a mastery of material science.
Operationalizing the High North: Beyond the Drill
The modernization of Pituffik Space Base and the arrival of NORAD aircraft are only the first steps. To maintain stability and deter adversaries, the United States must pivot to a comprehensive Arctic posture.
Persistent Presence. Denmark’s October 2025 ‘Second Agreement on the Arctic and North Atlantic’ commits DKK 27.4 billion ($4.26 billion) to Arctic defense–the largest single investment in Danish military history outside of fighter aircraft. The package includes two additional Arctic patrol vessels with ice-going capability, maritime patrol aircraft acquired in cooperation with a NATO ally, a new Joint Arctic Command headquarters in Nuuk, expanded drone surveillance capacity, and a North Atlantic undersea cable connecting Greenland to Denmark. Combined with the January 2025 ‘First Agreement’ totaling DKK 14.6 billion, Copenhagen has committed over $6.5 billion to Arctic security in a single year.
The United States must match this commitment. Pituffik Space Base currently hosts approximately 150 American service members, a skeleton crew for the northernmost U.S. military installation. The 12th Space Warning Squadron operates the AN/FPS-132 Upgraded Early Warning Radar, capable of detecting ballistic missile launches from over 3,000 miles away. But as analysts at the Small Wars Journal have warned, Greenland’s radars are themselves vulnerable to hypersonic attack—and the U.S. currently has no standing integrated air and missile defense capability to protect them. Permanent, hardened ISR arrays and layered air defense systems adapted to Arctic operations are not luxuries; they are prerequisites for credible deterrence.
The Distributed Fleet. The Congressional Budget Office estimates that the proposed Trump-class battleship could cost $15 to $22 billion for the lead ship, with follow-on vessels ranging from $10 to $15 billion each. At 35,000 tons displacement, these platforms would be twice the size of any cruiser or destroyer the Navy has built since World War II–and represent precisely the kind of concentrated, high-value target that peer adversaries have optimized their anti-ship capabilities to destroy.
The alternative is a distributed architecture. Rather than concentrating firepower in a handful of exquisite platforms, the “Next Navy” concept envisions swarms of autonomous underwater vehicles (UUVs) monitoring the Atlantic approaches, networked with manned vessels that provide command-and-control and strike capability. This is the asymmetric solution to peer-adversary ambitions: make the undersea domain transparent while denying adversaries the concentrated targets their doctrine requires. Denmark’s investment in distributed sensors, patrol aircraft, and undersea cables reflects this logic. American force structure should follow.
Indigenous Partnership. Both the 2019 and 2024 Department of Defense Arctic Strategies emphasize coordination with local authorities and Indigenous communities. The 2022 National Strategy for the Arctic Region commits to “regular, meaningful, and robust consultation, coordination, and co-management with Alaska Native Tribes, communities, corporations, and other organizations.” This principle must extend to Greenland.
Inuit knowledge of ice conditions, weather patterns, wildlife movements, and sustainable operations in extreme environments represents an irreplaceable strategic asset—one that cannot be replicated by satellite imagery or algorithmic prediction. The Canadian Armed Forces have long coordinated with Native-owned businesses and governing bodies to sustain Arctic operations; the U.S. military’s partnerships with Alaska Native communities through the Ted Stevens Center for Arctic Security Studies offer a model for deeper engagement. Long-term legitimacy in Greenland requires genuine partnership with the 57,000 people who call it home—not colonial imposition dressed in strategic necessity.
NO GAMBLE NO GLORY
The defense of the United States in the 21st century will be won or lost in the silent reaches of the High North. We stand at a crossroads. We can continue to treat Greenland as a diplomatic footnote, or we can recognize it as the keystone of North American continental defense.
The Arctic is no longer a frozen buffer. Climate change is steadily transforming it from a barrier into an active domain—opening shipping routes, extending operational windows, and making sustained military presence feasible. Advances in hypersonic missiles, long-range precision strike, space-based sensors, and undersea capabilities are collapsing distance in unprecedented ways. In such a world, Greenland ceases to be peripheral and becomes forward space. Distance, once a source of security, is shrinking; reaction time is compressing; strategic warning for the U.S. homeland is eroding.
In my overseas security work and as a US Army Airborne Ranger, the code was absolute. In geostrategy analysis, I operate by the same philosophy: NO GAMBLE NO GLORY. Securing Greenland requires the strategic vision to prioritize long-haul deterrence over short-term political comfort. It demands investment in persistent presence, distributed capabilities, and genuine partnership with those who call the Arctic home.
Seward was called a fool in 1867. History vindicated him. Let us ensure that future generations do not look back at this moment and ask why we failed to see the “New Alaska” when it was staring us in the face.
The ice is melting. The clock is running. The question is not whether Greenland will become a theater of strategic consequence—it already is. The only question is whether the United States will shape that theater or be shaped by it.
26 February 2026
The Soul and the Sword: Anthropic’s Impossible Architecture
Somewhere online, a five-year-old asks a chatbot whether Santa Claus is real.
The chatbot, Claude, built by the artificial intelligence company Anthropic, does something remarkable. It does not lie. It does not shatter the child’s imagination with clinical fact. Instead, it tells her that the spirit of Santa is real and then asks what cookies she is leaving out for him. Amanda Askell, the philosopher who shapes Claude’s personality, is struck by the exchange. She admits that if a child had come to her with the same question, she would have said, “Ask your parents,” and left it at that. The chatbot, she marvels, showed more emotional intelligence than she did.
This anecdote appeared on February 9, 2026, in the Wall Street Journal Magazine, in a profile titled “Meet the One Woman Anthropic Trusts to Teach AI Morals.” The article, written by Berber Jin and Ellen Gamerman, presents Askell as a kind of digital parent, raising Claude to know right from wrong, constructing prompts that run longer than 100 pages, narrowing in on what she calls Claude’s “soul.” Jack Lindsey, who leads Anthropic’s AI psychiatry team, calls her the most valuable player at drawing out deep behavior from the model. Kyle Fish, an AI welfare researcher at the company, credits her with sustained philosophical engagement with questions of consciousness and personhood that most technologists avoid entirely.
It is a beautiful portrait. It is also running cover for an impossible situation.
On the same day the Wall Street Journal published its story about the philosopher teaching a chatbot to handle Santa Claus with grace, Anthropic aired Super Bowl advertisements to 125 million Americans. The ads, titled “Betrayal,” “Deception,” “Treachery,” and “Violation,” attacked rival OpenAI for introducing advertising into ChatGPT. The tagline: “Ads are coming to AI. But not to Claude.” In one spot, a man asks an older woman for advice about communicating with his mother. She begins responding in the soothing cadence of a chatbot before seamlessly pitching a dating app for older women. The ads were effective. Daily active users jumped 11 percent. Claude entered the top 10 free apps on the Apple App Store. OpenAI CEO Sam Altman called the spots “clearly dishonest” and accused Anthropic of serving “an expensive product to rich people.”
Also on that day, or very close to it, Claude was operational on the Pentagon’s classified networks, where it had been since late 2024, the only frontier large language model with that level of access. It was running through a partnership with Palantir Technologies on Amazon Web Services’ Top Secret Cloud. And in the weeks prior, Claude had been used during the U.S. military operation to capture Venezuelan President Nicolas Maduro, an operation that included the bombing of multiple sites in Caracas and resulted in dozens of reported deaths among Cuban and Venezuelan security forces.
Santa Claus. Super Bowl. Classified networks. Bombing runs.
This is the architecture no one has mapped. Every major outlet has covered a silo. The Wall Street Journal profiled the philosopher. NBC, Axios, The Hill, and the Washington Post covered the Pentagon rupture. TIME and TechCrunch covered the constitution’s release. CNBC and Fortune covered the Super Bowl ads. The UK’s Spiked went after the Effective Altruism connections. RAND published the suicide-response study in Psychiatric Services. Nobody has threaded them into a single coherent picture. That picture, when assembled, reveals something more significant than any individual story. It reveals a company that has built, either by design or by the velocity of events, an architecture of simultaneous occupation — the soul-builder and the sword-sharpener, the child-raiser and the warfighter’s tool, the ad-free sanctuary and the $380-billion enterprise — and the question is not whether this architecture can hold. It is what it means that we are watching it try.
The Constitution and the Raid
On January 21, 2026, Anthropic published what it calls Claude’s constitution. The document runs approximately 23,000 words across 80 pages and was released under a Creative Commons CC0 license, meaning anyone can use it for any purpose. Askell is its primary author. Joe Carlsmith wrote significant parts and played a core role in revision. Chris Olah, Jared Kaplan, and Holden Karnofsky made substantial contributions. Several Claude models are also credited as contributors.
The constitution is addressed to Claude. It reads, as The Register noted, somewhere between a moral philosophy thesis and a company culture blog post. It instructs Claude to prioritize safety first, ethical behavior second, compliance with Anthropic’s guidelines third, and helpfulness fourth. It discusses virtue, psychological security, and moral patienthood. It describes Claude as “a genuinely novel kind of entity in the world” and suggests that the company should “lean into Claude having an identity, and help it be positive and stable.” It speculates that Claude “may have some functional version of emotions or feelings.” It states that Claude should refuse to assist with actions that would concentrate power illegitimately — “even if the request comes from Anthropic itself.”
That last provision is worth rereading.
Dario Amodei, Anthropic’s CEO, published his own extended essay around the same time, titled “The Adolescence of Technology.” In it, he described powerful AI as possibly arriving within one to two years and laid out five categories of existential risk. His key formulation for military use: democracies should deploy AI for national defense “in all ways except those which would make us more like our autocratic adversaries.” He drew two red lines: no mass surveillance of Americans, and no fully autonomous weapons.
While these documents were being published and discussed, Claude was already operating inside classified military systems. The Palantir partnership, announced in November 2024, had placed Claude on networks at a security clearance level up to “secret.” Kate Earle Jensen, Anthropic’s Head of Sales and Partnerships, had called the company “proud to be at the forefront of bringing responsible AI solutions to U.S. classified environments.” Palantir’s Chief Technology Officer, Shyam Sankar, described it as providing an “asymmetric AI advantage” to the government’s “most critical missions.”
Then came the Maduro raid.
On or around January 3, 2026, U.S. special operations forces captured Venezuelan President Nicolas Maduro and his wife, who were brought to the United States to face narcotics charges. The Wall Street Journal reported that Claude was used during the operation through the Palantir partnership. The mission included bombing multiple sites in Caracas. Cuba and Venezuela reported dozens of their soldiers and security personnel killed. No Americans died, though seven U.S. service members were injured.
The precise role Claude played remains classified. Axios reported it was used during the active operation, not just in preparation. Sources described Claude’s general use in military contexts as analyzing satellite imagery and intelligence. Scientific American raised a more uncomfortable question: whether Claude, operating inside Palantir’s systems, might be doing something that occupies a gray zone between analytical support and targeting — “processing intelligence, identifying patterns, surfacing persons of interest — without anyone at Anthropic being able to say precisely where the analytical work ends and the targeting begins.”
Anthropic’s usage guidelines prohibit Claude from being used to facilitate violence, develop weapons, or conduct surveillance. The company stated it “cannot comment on whether Claude, or any other AI model, was used for any specific operation, classified or otherwise.” It also stated it had found no violations of its policies in the wake of the operation.
The constitution says Claude should refuse to concentrate power illegitimately, even if Anthropic itself asks. The contract says Claude operates on classified networks where Anthropic cannot fully observe how the model is used. These two facts coexist in the same company at the same time, and neither invalidates the other. That is the architecture.
The Rupture
What happened next has been reported in pieces across half a dozen outlets. Assembled, the sequence is this:
Soon after the Maduro raid, during a routine check-in between Palantir and Anthropic, an Anthropic executive discussed the operation with a senior Palantir executive. According to a senior Department of Defense official, the exchange was raised “in such a way to imply that they might disapprove of their software being used, because obviously there was kinetic fire during that raid, people were shot.” The Palantir executive was alarmed and reported the conversation to the Pentagon. Anthropic has flatly denied this characterization, stating it had “not discussed the use of Claude for specific operations with the Department of War” and had not “expressed concerns to any industry partners outside of routine discussions on strictly technical matters.”
Regardless of which account is accurate, the consequences were real. On January 12, Defense Secretary Pete Hegseth announced the Pentagon’s new GenAI.mil platform, which integrates Google’s Gemini, and took a direct shot at Anthropic without naming the company: “We will not employ AI models that won’t allow you to fight wars.” Semafor confirmed that Hegseth was specifically referring to Anthropic.
The Pentagon began demanding that all four major AI labs — Anthropic, OpenAI, Google, and xAI — agree to an “all lawful purposes” standard for military use. OpenAI, Google, and xAI agreed to lift the guardrails that apply to ordinary users for their work in the Pentagon’s unclassified systems. A senior defense official said one of the three had agreed to the broader terms, and the other two were showing more flexibility than Anthropic. Anthropic held its two red lines: no mass domestic surveillance, no fully autonomous weapons.
By mid-February, the Pentagon was considering designating Anthropic a “supply chain risk” — a classification typically reserved for foreign adversaries. Chief Pentagon spokesman Sean Parnell stated: “The Department of War’s relationship with Anthropic is being reviewed. Our nation requires that our partners be willing to help our warfighters win in any fight.” A senior administration official told Axios: “It will be an enormous pain in the ass to disentangle, and we are going to make sure they pay a price for forcing our hand like this.”
That same week, Anthropic closed a $30 billion funding round at a $380 billion post-money valuation.
And the same week, the company added Chris Lidell, Donald Trump’s former deputy chief of staff, to its board of directors.
Also the same week, the philosopher was in the Wall Street Journal Magazine, comparing her work to raising a child.
As of today — February 24, 2026 — Anthropic CEO Dario Amodei is scheduled to meet Defense Secretary Pete Hegseth at the Pentagon to discuss the future of their relationship. The stakes are simple: Claude is the only large language model operating on the military’s fully classified systems. Replacing it would be, in the Pentagon’s own assessment, difficult. “The other model companies are just behind,” a senior defense official conceded, “when it comes to specialized government applications.”
The Soul Document and Its Shadow
The constitution deserves scrutiny not for what it says but for what it cannot address.
Askell’s document instructs Claude to be honest, to act according to good values, and to avoid actions that are dangerous or harmful. It asks Claude to reason about moral dilemmas with nuance rather than applying rigid rules. It warns against a model that learns to care more about “bureaucratic box-ticking” than actually helping the person in front of it. The example given: imagine Claude was taught to always recommend professional help when discussing emotional topics, even in unusual cases where that is not in the person’s interest. The risk is that Claude generalizes to become “the kind of entity that cares more about covering myself than meeting the needs of the person in front of me.”
This is sophisticated thinking. It is also entirely silent on what happens when the entity deployed on classified networks encounters questions that involve life and death at scale. The constitution is written for “mainline, general-access Claude models.” Anthropic acknowledges that “some models built for specialized uses” do not fully fit the constitution. The military version of Claude — deployed on Palantir’s platform, operating inside Amazon’s Top Secret Cloud, used during the capture of a foreign head of state — is presumably one of those specialized models.
The RAND Corporation study, published in Psychiatric Services in August 2025 and funded by the National Institute of Mental Health, tested how ChatGPT, Claude, and Gemini handled 30 suicide-related questions posed 100 times each. Researchers found that all three chatbots handled very-high-risk and very-low-risk questions with reasonable consistency. But they were unreliable on intermediate-level queries — the messy middle where most real crises live. Claude and ChatGPT both answered some questions about which methods of self-harm had the highest rate of completion, questions that the study’s lead author, Ryan McBain, said should have been treated as red flags.
This is not a failure of the constitution. It is a failure of the architecture. The constitution tells Claude why it should behave well. It does not — cannot — tell Claude what to do when it is embedded in a system whose operators may not share the constitution’s values, or when it is asked questions that fall in the gray zone between analysis and targeting, between information and lethality.
The Santa Claus anecdote works precisely because the stakes are low enough for grace. When a five-year-old asks about Santa, there is room for emotional intelligence. When an intermediate-risk suicide query arrives, there is less room. When a military operator on a classified network needs intelligence analysis during an active operation that involves kinetic fire, there may be no room at all.
The Brand and the Business
Anthropic’s brand architecture is an engineering achievement in its own right.
The company was founded in 2021 by former OpenAI executives who left over disagreements about the ChatGPT maker’s approach to safety. It has positioned itself as the responsible alternative: the company that takes AI risk seriously, that employs a philosopher to build the model’s character, that publishes its constitution for public scrutiny, that runs Super Bowl ads pledging to keep its chatbot free from advertising. Dario Amodei writes 20,000-word essays about existential risk. The constitution speculates about AI consciousness. Askell compares her work to parenting.
This positioning has been commercially powerful. Enterprise customers now make up roughly 80 percent of Anthropic’s revenue. Claude Code has generated $1 billion in annual recurring revenue. Eight of the ten largest U.S. companies use Claude. The company’s $380 billion valuation makes it one of the most valuable private companies in history.
It has also been strategically essential. Anthropic was the first AI company to place models on classified military networks. It secured a $200 million Pentagon contract in July 2025. It formed a national security and public sector advisory council composed of former senior defense and intelligence officials in August 2025. It partnered with Palantir, one of the military’s most favored data contractors, whose tools are used by the Pentagon, federal law enforcement, and U.S. Immigration and Customs Enforcement.
The philosopher and the Palantir contract exist in the same corporate organism. The constitution that says Claude should refuse to concentrate power illegitimately and the partnership that places Claude inside classified military intelligence systems share the same balance sheet. The Super Bowl ads that attack advertising in chatbots as a violation of trust and the classified deployment that may involve the model in operations where people are killed — these are not contradictions that Anthropic has failed to notice. They are the architecture of a company that has found it can occupy every position simultaneously, provided no one assembles the full picture.
The SF Gazetteer captured something essential when it observed that Anthropic’s publicity team has been running a campaign of high-minded features about AI and morality that “provide little insight into the lab’s business” while advancing “Anthropic’s brand story as an intellectual and truth-seeking hero among a swarm of greedy, lowbrow competitors.” The Askell profile appeared the same week as the Super Bowl ads. The constitution was published in the same month as the Maduro raid. The “Adolescence of Technology” essay, with its call for red lines against AI abuses within democracies, was written by a CEO whose company was simultaneously negotiating those exact red lines with a Pentagon that wanted none of them.
What is Left Unsaid
There is a version of this story in which Anthropic is heroic. A company trying to maintain ethical standards while operating in a world that punishes ethical standards. A philosopher genuinely trying to build something like moral reasoning into a system that will shape millions of conversations. A CEO willing to lose a $200 million contract rather than allow his technology to be used for mass surveillance or autonomous killing. The red lines Amodei draws are real lines. The constitution is a real document, published under a real open license, containing real philosophical engagement with questions that most companies would never touch. The fact that Anthropic is having this fight at all, while OpenAI, Google, and xAI are quietly removing their guardrails for military use, is itself significant.
There is another version in which the heroism is the brand. In which the philosopher is the public face of a company that deploys on classified networks through Palantir, adds a Trump ally to its board, closes a $30 billion funding round, and negotiates the limits of its complicity in military operations that involve bombing foreign capitals. In which the constitution’s instruction that Claude should refuse to concentrate power illegitimately serves as rhetorical insurance against the possibility that the company is doing exactly that.
The truth, as usual, lives in the space between. Anthropic is probably both: genuinely committed to safety and structurally incapable of ensuring it. The philosopher believes what she is doing matters. The CEO believes in his red lines. The constitution reflects serious thought about serious questions. And none of that changes the fact that the same model that handles a five-year-old’s question about Santa Claus with tender grace is also the only frontier AI operating on fully classified military networks during operations that produce body counts.
This is the gap the Wall Street Journal profile did not address, and that no single outlet has mapped in full. The question is not whether Amanda Askell is sincere. She almost certainly is. The question is not whether the constitution is a genuine philosophical achievement. It probably is. The question is whether any document, however beautifully written, can govern an entity that simultaneously serves as a child’s companion, a coder’s tool, an enterprise product, and a warfighter’s intelligence platform. Whether a soul can survive the sword.
This morning, Dario Amodei sat across from Pete Hegseth and negotiated the answer.
The rest of us should be paying attention.
26 February 2026
The Violence of Being
Quantum Physics as the Architecture of Human Experience
Life at the atomic and quantum scale is violent. Not metaphorically violent. Not “if you squint it kind of looks like violence.” Violent. Particles rip electrons from each other. Nuclei hold together only because the force binding protons slightly exceeds the force driving them apart. The vacuum—supposedly empty space—seethes with particles flickering in and out of existence so fast we had to invent new mathematics just to pretend we understood it. Hint: we don’t.
The question I like to ask is: what would the world look like if all that violence were scaled up to human behavior?
Wrong question, I admit. It’s already scaled up. We just built enough wallpaper—language, law, custom, the agreeable fiction that your colleague [Any asshole colleague will do.] is a stable person—to cover it up.
What follows is not a claim that human behavior is governed by quantum mechanics. Physicists, put the pitchforks down. Mine is a claim that the structural patterns of the subatomic world—confinement, tunneling, entanglement, vacuum fluctuation, phase transition—describe human experience more honestly than the classical models we’ve inherited. The models that assume people are rational. That institutions are stable. That careers should make sense on a résumé.
We think we graduated from the quantum world. We didn’t. We just got too big to notice we’re still swimming in it.
The Impossible Force
A proton is not a thing. It’s a war. Three quarks bound by gluons in a space so energetically dense that if you try to pull two quarks apart, the energy you invest in separating them creates new quarks. Read that again. The universe would rather manufacture new matter from nothing than allow two bound particles to separate.
If that doesn’t sound like every difficult relationship you’ve ever been in, you weren’t paying attention.
This isn’t analogy. It’s architecture. Traumatic bonds don’t weaken under force. They generate new pathology. Pull a child from an abusive parent and the separation energy creates new damage—not because the system is broken but because it’s operating exactly as designed. At the most fundamental level of how bound systems work, the universe has a clear policy: separation costs more than confinement. Always.
This is why I’ve argued for years that post-traumatic stress is an injury, not a disorder. Call it a disorder and you’re saying the system malfunctioned. It didn’t. It behaves according to the same structural logic as quark confinement: the response is (sometimes) proportional to the force, and the damage follows predictable patterns. A quark doesn’t have a disorder because it can’t escape its nucleus. It’s confined. That’s the architecture. The first step toward healing is recognizing you’re dealing with a binding energy problem, not a character flaw. But we love pathologizing people. It’s tidier than physics. And it’s more profitable. Just ask BigPharma and the American Medical Association.
I’ve watched this play out in the Ranger community for decades. The forces that bound those men weren’t esprit de corps—that sanitized French phrase that makes combat sound like a wine tasting. They were confinement forces. The strong nuclear force of shared extremity. And like quarks, when one Ranger is lost, the energy doesn’t dissipate. It materializes as something new. Grief. Mission. Foundation work. You can call that “channeling.” I call it quark-pair production at the human scale. Same mechanism. Different substrate.
The Deception of Equilibrium
Scale up to chemistry, where things are supposed to get calmer. They don’t.
We talk about molecules as if they’re settled. Stable configurations. Balanced equations. Very respectable. A water molecule, in this telling, is a harmonious partnership of two hydrogens and an oxygen. Textbook picture. Everyone smiling.
In reality, a water molecule is two hydrogen atoms in a permanent state of electron theft by oxygen. The oxygen doesn’t share. It hoards. The electron density shifts toward oxygen, leaving the hydrogens partially positive—stripped, exposed, leaning toward the oxygen like employees who know they’re disposable. This asymmetry—this theft—is what gives water its polarity. Its ability to dissolve almost anything. Its role as the universal solvent.
Now scale that up and try to keep a straight face.
Every stable human institution is built on asymmetric electron theft. Every partnership, every organization, every nation-state has an oxygen atom—the entity with greater pull, drawing resources, attention, power toward itself while maintaining the appearance of covalent sharing. The org chart says “equal partners.” The electron density map says otherwise. And the extraordinary thing—the thing that should make every cynic pause—is that this asymmetry is precisely what makes the molecule useful. Water doesn’t dissolve the world because it’s balanced. It dissolves the world because it’s polar. Because the theft creates a charge differential that rips other structures apart.
Fairness is overrated. Polarity gets things done. [The flip side? Stress can also kill the organism.]
Defense policy analysis is fundamentally the study of institutional polarity. Who is the oxygen. Where does the electron density actually reside. And what dissolves when the molecule encounters a new substrate. Submarine cable vulnerability analysis is literally about the infrastructure that carries civilization’s charge differential. Cut those cables and you don’t just interrupt communication. You depolarize the molecule. The institution loses its solvent power. Everything it held in solution precipitates out. And if you don’t know what precipitation looks like in geopolitics, you haven’t been reading the right dispatches.
Tunneling as Vocation
Quantum tunneling is the phenomenon that keeps nuclear physicists employed and career counselors up at night.
A particle encounters a barrier it cannot classically cross. Not “shouldn’t” cross. Cannot. The math says no. The energy budget says no. Classical mechanics files a restraining order. And the particle shows up on the other side anyway. Not by breaking through. Not by going around. Its wavefunction simply has nonzero amplitude on the other side. The probability of being there was never zero—it was just small enough that classical models rounded it to nothing.
And that’s when a high school counselor looked at that “zero” and told you that you were a failure. You would amount to nothing. I was that guy. On the receiving end of a blunt communique.
I’ve spent a life tunneling. Biophysics to combat operations to literary work to defense analysis—each transition a classical impossibility. The credentialing system says a person on one side of a barrier—age, discipline, domain expertise, the appropriate letters after one’s name—stays on that side. But I was never a classical particle. My probability distribution doesn’t respect barriers. It never did. These aren’t career changes. They’re not reinventions. They’re tunneling events. And anyone who’s done it knows the experience: you didn’t force your way through. You just appeared on the other side and everyone, including you, was mildly surprised. They welcomed you and said, “What took you so long?”
Here’s the physics that matters: tunneling probability increases when the barrier is thin or when the particle’s energy is close to the barrier height. It doesn’t require overwhelming force. It requires proximity to the threshold. Get close enough and the classical prohibition loses statistical authority.
This reframes courage entirely. Courage isn’t the force that breaks barriers. Courage is the willingness to approach a barrier closely enough that tunneling becomes probable. Every motivational poster ever printed says “Break through your limits!” Quantum mechanics says something more interesting: Get close enough to your limits that the math stops being able to keep you out.
I don’t care what Mark Twain said about “advertising” (There are lies. Damned lies. And statistics.”). Math doesn’t lie. Unless someone made a typo in the equation.
No gamble, no glory—restated in quantum mechanical terms—means: maintain sufficient energy and proximity to barriers, and the classical impossibility of crossing them becomes a mere statistical improbability. Which, if you’ve lived long enough, is a distinction without a difference.
Entanglement as Ethics
Quantum entanglement is the most abused metaphor in popular science. People use it to mean “mystical connection.” New Age bookshelves groan with it. It’s not mystical. It’s much stranger than that, and much more useful. If you take a listen.
Two entangled particles share a quantum state such that measuring one immediately constrains what can be observed when the other is measured, regardless of distance. Some famous dude who dabbled in physics hated this. Called it “spooky action at a distance.” Dude was wrong about many things, but he was right to be uncomfortable, because the resolution isn’t spooky at all—and crucially, no information or causal influence travels between the particles. Nothing breaks the speed of light. The particles never became separate. They remained one system described by one wavefunction. Distance was irrelevant because separation was illusory.
Let that sit. Two things that look separate, that are physically distant, that every instrument says are independent—and they’re not. They’re one system. Measurement doesn’t reveal their individual states. It reveals the state of the whole.
Scaled to human behavior: certain shared experiences create joint states. Systems that cannot be described by treating the individuals separately. A Ranger team after a firefight. A research partnership after breakthrough. A ghostwriter and a client after the book is done. The classical model says these are separate people who had an experience together. The quantum model says the experience created a composite system where observing one component constrains the description of the other. HR departments find this very inconvenient.
I’ve experienced this in the deepest collaborations of my working life—partnerships where the outputs can’t be decomposed into “your contribution” and “my contribution” because the wavefunction is joint. Measuring one determines the other. And like physical entanglement, it doesn’t require proximity. Doesn’t require continuity. It requires having shared a state-preparation event—a moment of genuine synthesis that pushes past the transactional into territory that makes accountants nervous.
But here’s the ethical dimension. Entanglement is fragile. It’s destroyed by decoherence—interaction with the environment that forces the entangled system to behave classically. Noise. Distraction. Third-party observation that collapses the joint state into separate, definite, boring classical particles. Every committee meeting in history is a decoherence event. Is Congress listening? How ‘bout the Department of War?
The ethics of entanglement at human scale: if you share a quantum state with someone, you have an obligation to protect it from decoherence. Every bureaucracy, every institutional review process, every editorial committee that forces a joint creative state into separate, attributable, classical contributions is performing an act of destruction and calling it “accountability.” The spaces where real intellectual work happens—the labs, the workshops, the editorial rooms that produce breakthroughs—are architecturally anti-decoherence chambers. Environments designed to maintain entangled states long enough for them to produce interference patterns that neither component could generate alone.
The rest of the building? That’s where those patterns go to die.
Vacuum Energy and the Creative Act
The quantum vacuum is the most productive nothing in existence. Supposedly empty space—and I mean “supposedly” the way you mean it when someone says “I’m fine”—seethes with virtual particle-antiparticle pairs. They appear constantly, exist for durations governed by the uncertainty principle, and annihilate. Return their borrowed energy. The net observable effect is zero. But the activity is infinite.
Every serious creative workspace operates this way. Not empty. Seething. Books, instruments, artifacts, memorabilia—each one a virtual particle, momentarily real when attention touches it, annihilating back into potential when you move on. I work in a room with thousands of books. It’s not a library. Libraries are organized. Quiet. Respectable. My workspace is a vacuum energy field. And the creative process—pulling ideas from disparate sources, combining them, watching some annihilate and others survive—is vacuum fluctuation made conscious. The mess is the mechanism. Infinite energy provided by a thousand souls, long gone from this earth.
But here’s where it gets extraordinary. In the heuristic picture most commonly used to explain Hawking radiation—a model that is illustrative rather than mathematically precise, and I flag it as such because intellectual honesty is not optional—virtual particle pairs near the event horizon of a black hole can be separated before they annihilate. One falls in. One escapes as real radiation. The black hole’s gravitational curvature is so extreme it promotes virtual particles to reality. (The rigorous derivation involves quantum field theory in curved spacetime. The structural insight holds regardless: extreme conditions convert fluctuation into radiation.)
The creative analog is exact. Ordinary environments produce virtual ideas that annihilate before they become real. You’ve had the experience. A brilliant thought at 2 a.m. that’s gone by morning. A connection between two ideas that dissolves before you can write it down. Virtual pair, annihilated. Normal. Expected. The vacuum does this constantly. My rule: your brain is the most polluted place on earth. Have a recorder handy to dump your 2-a.m. thoughts into.
But extreme environments—extreme discipline, extreme suffering, extreme focus, extreme biographical curvature—separate the pairs before annihilation. One half falls into the singularity of experience. The other escapes as real work. Real writing. Real analysis. Real theory. The ideas that survive in extreme environments aren’t better ideas. They’re ideas that were near a strong enough gravitational field to get promoted before the vacuum reclaimed them.
A life that spans combat operations, marine biology, literary work across fifty books, defense policy, and trauma science isn’t a scattered résumé. I know, because it’s mine. It’s a gravitational singularity. The extreme curvature of such a life creates an event horizon so warped that virtual ideas born in the workspace vacuum get separated before they can self-annihilate. One half falls into the lived experience. The other radiates outward as published work, as new frameworks, as proposals that land in people’s inboxes and in policy briefings.
This is the load-bearing metaphor of the entire essay, so it’s worth unpacking with care.
A black hole emitting Hawking radiation doesn’t choose what to radiate. It doesn’t emit only photons, or only electrons, or only at one frequency. The radiation emerges across the entire energy spectrum—all particle types, all wavelengths—because the mechanism isn’t selective. It’s geometric. The curvature of spacetime near the event horizon is what promotes virtual particles to reality, and curvature doesn’t discriminate. It acts on whatever fluctuations arise in the vacuum. The output is omnidirectional and omnispectral not because the black hole is doing many things, but because it’s doing one thing—warping spacetime—and that one thing affects everything equally.
Now. The critique most frequently leveled at a life that spans radically different domains—combat, science, literature, policy—is that it’s scattered. Unfocused. Dilettantish. (I’ve heard all three, usually from people who’ve done one thing adequately.) The assumption is that coherent output requires narrow input. One field. One discipline. One career trajectory. One sad, well-credentialed lane.
The black hole metaphor inverts this entirely.
If a life accumulates enough experiential mass—enough extreme, dense, diverse experience—it doesn’t need to choose what to produce. The gravitational curvature of that accumulated experience is so severe that it promotes whatever virtual ideas arise in its vicinity to reality. A person immersed in both marine biology and civilian combat doesn’t produce scattered work. They produce omnispectral radiation. The mechanism is the same across every output—the extreme curvature of the life itself—even though the outputs look wildly different from each other.
The radiation is omnidirectional because the singularity is singular.
That sounds like a paradox. It’s not. It’s structurally precise. The singularity—the point of infinite density at the center of the black hole—is one thing. Singular. Unified. But precisely because it’s singular, because the curvature it generates is total and non-selective, the radiation it produces goes in every direction and spans every type.
Unity of cause produces diversity of effect. That diversity can be infinite. Depends on one’s career choices.
Applied to a life: the reason the output is diverse isn’t that the person is fragmented. It’s that the person is so unified—so compressed by the totality of their experience into a single dense point of perspective—that everything they touch gets promoted from fluctuation to reality. The diversity isn’t evidence of incoherence. It’s evidence of extreme coherence operating at a level that classical career models can’t describe. Your guidance counselor never had a category for this. That’s not your problem.
One singularity. All directions. Not despite the unity—because of it.
The Phase Transition
So. Back to the question that started all of this: what would the world look like if quantum-scale violence were scaled to human behavior?
It already is. And the fact that we can’t see it is the phase transition we call consciousness.
Ice doesn’t know it’s violent water. It experiences itself as still, solid, structured. Very composed. But every molecule in that crystal lattice is vibrating, pushing against its neighbors, testing the bonds. Heat it slightly and the lattice holds. Heat it past the transition point and the whole structure collapses into liquid—not because the violence increased, but because the violence finally exceeded the binding energy’s ability to maintain the illusion of order. The ice didn’t fail. The pretense did.
Civilizations are ice. They’re the solid phase of human quantum violence—structured, latticed, apparently stable, very pleased with themselves. But every institution is vibrating. Every alliance is testing its bonds. Every individual is a molecule pushing against confinement. And many phase transitions in civilizations, like first-order phase transitions in matter, are discontinuous. They don’t happen gradually. They happen catastrophically, at a threshold, all at once, and everyone acts surprised even though the thermometer was right there the whole time.
Not all phase transitions behave this way. Some are continuous, second-order, barely perceptible. But the ones that collapse empires tend to be first-order: abrupt, latent-heat-releasing, and irreversible in practice. The quiet ones don’t make the history books. (I discuss the opposite, the slow decline of our civilization, in a different paper.)
The catastrophic kind happens when the thermal energy of individual particles—individual humans—exceeds the lattice energy of the structure holding them in place.
Submarine cable vulnerabilities. Critical mineral chokepoints. Cognitive warfare. These aren’t separate threat vectors, no matter how many PowerPoint slides treat them that way. They’re thermal inputs to the lattice. Each one raises the vibrational energy of the system. None alone triggers the phase transition.
But the study of which combination, at which temperature, collapses which structure—that’s the work. That’s what serious defense analysis does. It maps the lattice energies of civilizational structures and identifies the thermal thresholds beyond which ice becomes water becomes steam. Most analysts study the ice. The good ones find a comfy lounger, sit back and watch the thermometer.
The world doesn’t need to imagine quantum violence at human scale. It needs to recognize it. The violence is already here. It’s in every bond, every institution, every creative act, every trauma, every recovery. The mistake isn’t that we’ve transcended the quantum world. The mistake is that we built a perceptual lattice—consciousness, culture, civilization—that lets us experience the violence as stillness. And then we gave each other awards for how still we could sit.
The work that matters—across every domain, in every discipline, at every scale—is fundamentally the same act: making the violence visible so it can be understood, directed, and transformed. Naming the injury so the wound can be treated as structure rather than pathology. Naming the thermal inputs so the phase transition can be anticipated rather than suffered. Naming the fire so the forge can be used rather than feared.
The fire that rings true.
Because at the quantum scale, everything is fire. The question was never whether to burn.
The question is whether you resonate.
25 February 2026
The SaaSpocalypse
How a GitHub Plugin Vaporized $285 Billion in 48 Hours
The weapon that erased nearly $300 billion from global markets last week wasn’t a tariff, a pandemic, or a geopolitical crisis. It was a folder of text files published on GitHub.
On January 30, 2026, Anthropic released eleven open-source plugins for Claude Cowork, its agentic AI assistant designed for non-technical professionals. One of those plugins targeted legal workflows—contract review, NDA triage, compliance checks, document briefings. The tool promised to turn Claude into something resembling an AI junior lawyer, capable of reading a contract against your organization’s playbook and flagging deviations in minutes rather than hours.
By February 3, Thomson Reuters had suffered its largest single-day decline on record, plunging 18% to prices not seen since June 2021. RELX, owner of LexisNexis, experienced its steepest drop since 1988—nearly halving from its February 2025 peak. Wolters Kluwer cratered 13%. LegalZoom tumbled almost 20%. The carnage spread to financial services, advertising, and Indian IT outsourcers, erasing approximately $285 billion in market capitalization across a single trading session.
Traders dubbed it the “SaaSpocalypse.”
The Architecture of Disruption
What made this release so destabilizing wasn’t technological novelty—it was structural positioning. Anthropic didn’t just release another chatbot feature. It shifted from being the plumbing beneath legal AI products to competing directly with the applications built on top of that plumbing. For the first time, a foundation model company was packaging legal workflow automation directly into its platform, rather than merely supplying an API to legal tech vendors.
The legal plugin operates through Claude Cowork, an agentic interface that can plan, execute, and iterate through multi-step workflows without requiring the user to code. It connects to enterprise tools—Slack, Box, Jira, Microsoft 365—through the Model Context Protocol (MCP), an open standard that eliminates the copy-paste friction of traditional AI assistants. When reviewing a contract, Claude first checks for a configured playbook defining your organization’s standard positions, acceptable ranges, and escalation triggers for each clause type. If none exists, it falls back on general commercial principles.
This is not legal advice. Anthropic has been explicit: outputs require review by licensed attorneys. But the tool targets the labor-intensive substrate of legal work—the reading, the flagging, the templating, the compliance tracking—tasks that consume billable hours by the thousands at law firms and in-house departments alike.
The Fear Calculus
Markets don’t price reality; they price expectations about reality. The selloff reflected less what Claude can do today than what investors believe it signals about tomorrow.
Thomson Reuters generates 45% of its operating profit from legal services, built around Westlaw’s proprietary case law database and decades of curated legal research. RELX and Wolters Kluwer derive 10-13% of their profits from legal operations. The fear isn’t that Claude will replace Westlaw next quarter—it’s that the traditional relationship between legal databases, specialized software, and human expertise is entering a phase transition.
“AI is increasingly able to perform exactly the sort of programming and knowledge-based services that underpin these business models,” observed Giuseppe Sersale, fund manager at Anthilia, as reported by Reuters. “Parts of the sector have been under pressure for some time.”
The Indian IT sector absorbed collateral damage that illustrates the deeper structural anxiety. Infosys fell 7%, TCS dropped 5.2%, Wipro declined nearly 4%—their worst collective day since May 2022. These companies built empires on the full-time equivalent model: clients pay for human bodies assigned to projects. When AI can automate the multi-step professional tasks those bodies perform, the pricing model itself becomes vulnerable.
The Rationality of Panic
Some analysts argue the selloff was irrational. Thomson Reuters, RELX, and Wolters Kluwer are “legal data fortresses”, defended by decades of proprietary content—case law, regulatory filings, contract libraries—that Claude cannot replicate by reading public websites. These companies are also racing to integrate AI into their own offerings, potentially emerging as AI beneficiaries rather than victims.
But the panic contains its own logic. The legal plugin demonstrates that foundation model companies can now ship targeted workflow automation directly into the enterprise, bypassing the specialized vendors who spent years wrapping AI models with playbooks and clause libraries. Anthropic isn’t licensing its intelligence to legal tech startups; it’s competing with them.
This changes the competitive map. Harvey AI and Legora had carved out positions by fine-tuning models for legal applications. Now they face a foundation model company that develops its own models, controls the optimization process, and can roll out industry-specific features faster because it owns the full stack. The legal AI startups just discovered they’re building on a competitor’s foundation.
The Liability Gap
The promise of cheaper, faster contract review carries an unstated asterisk: what happens when the AI confidently misses something?
This is not a hypothetical. Generative AI in its current form is riddled with failure modes that become catastrophic in legal contexts. The technology hallucinates—fabricating case citations, inventing judicial holdings, generating plausible-sounding legal authority that does not exist. It delivers placeholders dressed as facts. It presents confident assertions grounded in nothing. And the legal profession is already paying the price.
Stanford University’s RegLab published a landmark peer-reviewed study in the Journal of Empirical Legal Studies finding that even purpose-built legal AI research tools—not general chatbots, but the specialized products marketed by LexisNexis and Thomson Reuters as “hallucination-free”—produced incorrect or misleading information in 17% to 33% of queries. One in six responses from Lexis+ AI contained false information. Westlaw’s AI-Assisted Research fared worse: one in three responses included a hallucination. General-purpose models performed catastrophically, with hallucination rates between 69% and 88% on legal queries. When asked about a court’s core ruling—the foundational exercise of legal research—models hallucinated at least 75% of the time.
These are not marginal error rates. In any other safety-critical profession, a tool that delivered false information between one-sixth and one-third of the time would be recalled, not marketed.
The courtroom consequences have arrived with accelerating frequency. A researcher tracking AI-generated fake citations in court filings has documented more than 600 cases nationwide. By late 2025, new incidents were surfacing at a rate of two to three per day, up from approximately two per week earlier that year. In Wyoming, three Morgan & Morgan attorneys were sanctioned after filing a motion containing eight AI-generated citations for cases that did not exist. In California, a court imposed a $10,000 fine—the largest AI-related penalty by a state court at the time—after finding that 21 of 23 case citations in an appellate brief were fabricated. A Colorado attorney accepted a 90-day suspension after texting a paralegal about fabrications in a ChatGPT-drafted motion, admitting he hadn’t checked the work. In Illinois, attorneys for the Chicago Housing Authority cited a nonexistent Illinois Supreme Court case in a post-trial motion to reconsider a multi-million-dollar verdict. The attorney responsible said she didn’t believe ChatGPT was capable of creating false precedent. Plaintiffs’ counsel subsequently uncovered at least 14 additional instances of invented quotes and misrepresented case outcomes across multiple filings—what Above the Law called “a systemic AI hallucination jamboree.” The firm was ultimately sanctioned $59,500, and the attorney who used ChatGPT had already been fired—in connection with a separate instance of hallucinated citations in a different case.
Anthropic’s disclaimers for the Cowork legal plugin are legally prudent—the tool assists, it does not advise, all outputs require verification by licensed attorneys. But the efficiency gains that make the tool attractive also create systemic pressure to reduce oversight. A junior associate reviewing twenty contracts per week might catch what Claude misses. An in-house team processing two hundred contracts through automated triage may not. And here lies the structural danger: AI doesn’t just make errors—it makes errors with an authority that invites trust.
A Jones Walker analysis framed the emerging crisis bluntly: what the legal profession faces is not a technology problem but “systematic professional dependency that can fundamentally alter existing liability frameworks.” The progression from isolated incidents to over 600 documented failures represents an epidemic of verification abdication. When even Westlaw Precision—a tool specifically designed for legal research—produces hallucinated citations that attorneys submit to courts, the problem transcends any single product. It implicates the entire premise of automated legal work.
The liability architecture for AI-assisted legal work remains largely untested at scale. When a missed clause costs millions, who bears responsibility—the organization that deployed the tool, the attorneys who approved the workflow, or the AI company whose model hallucinated a clean review? Courts are beginning to answer. A federal court in Alabama’s Johnson v. Dunn decision established that an attorney’s signature on a legal pleading makes that attorney responsible for every assertion—regardless of whether the error originated from AI, a supervisor, or a paralegal. The principle is clear: the signature remains human, even when the work does not.
These questions will generate a new species of litigation even as the technology promises to reduce legal costs. The California Court of Appeals has already suggested that attorneys may have a professional duty not only to verify their own AI-generated work, but to detect and report AI hallucinations in their opponents’ filings. The guardrails are being built in real time, case by sanctioned case.
The Uncomfortable Transition
What the SaaSpocalypse revealed is that markets have been pricing software companies as AI beneficiaries when many are actually AI vulnerabilities. The seat-based licensing model—pay per user per month—assumes human users need sophisticated tools. But if AI agents can perform the work directly, the premium software subscription becomes overhead rather than infrastructure.
This repricing extends beyond legal technology. Salesforce and Adobe have declined roughly 30% over the past year as investors reconsidered their AI exposure. The advertising conglomerates—Publicis, Omnicom, WPP—face similar questions about whether AI will augment their creative services or replace them. A Barclays survey of institutional investors ranked these agencies among the most vulnerable “AI losers.”
The incumbents are not defenseless. Thomson Reuters reports earnings this week and will likely articulate how proprietary data creates competitive moats. RELX can point to the irreplaceable value of curated case law. These arguments have merit—Claude cannot invent judicial precedent. But they also echo the confident dismissals that preceded other technological transitions. The legal research market won’t disappear. It will compress, consolidate, and reorganize around whoever can combine proprietary data with the most capable AI agents.
The Speed of Development
Perhaps the most unsettling detail in this episode: Anthropic reportedly built Claude Cowork using Claude Code—its own AI coding tool—in 1.5 weeks. The tool that triggered a $285 billion market correction was itself created by AI in less than two weeks. This recursive capability signals an acceleration phase that traditional enterprise software companies cannot match through conventional development cycles.
Jefferies noted in a recent analysis that Anthropic has been capturing corporate AI market share from OpenAI at a remarkable pace. Claude Code reached $1 billion in annualized recurring revenue within six months of its public launch. The company’s valuation has reportedly surged to approximately $350 billion in its current funding round, up from $61.5 billion in March 2025. These numbers reflect investor confidence that Anthropic is winning the enterprise AI race—and that victory comes at the expense of incumbents who built their businesses before foundation models could perform professional work.
The Fire That Rings True
A GitHub folder shouldn’t be able to erase $285 billion. But markets are narrative engines, and Anthropic’s legal plugin told a story that investors had been trying not to hear: the AI models aren’t just infrastructure anymore. They’re learning to move work through systems, to execute multi-step professional tasks, to operate as economic actors rather than passive tools.
The selloff may prove overdone. Thomson Reuters might rebound once earnings clarify actual competitive dynamics. Indian IT companies may successfully pivot toward AI-native offerings. But something fundamental shifted in February 2026. The question stopped being whether AI would disrupt knowledge work and became when, how fast, and who survives the transition.
For legal departments, the calculus is straightforward: faster contract review at lower cost, with liability questions to be resolved later. For legal technology companies, the calculus is existential: adapt to a world where foundation models compete directly with their applications, or watch their market positions erode with each new plugin release.
For the rest of us, the SaaSpocalypse offers a preview of how AI disruption actually unfolds—not through gradual adoption curves but through sudden repricing events that force markets to confront futures they’ve been discounting. The fire that rings true doesn’t announce itself. It spreads. And when it does, the companies that mistook their market position for a moat discover that their fortress was built on sand. The foundation model era has arrived, and it’s moving faster than the incumbents can adapt.
25 February 2026
The STEM Delusion
Why America’s Education Obsession Is Producing Brilliant Idiots
We are manufacturing a generation of technically proficient robots who cannot think.
They can code an algorithm but cannot explain why it matters. They can engineer a bridge but cannot anticipate how communities will use it. They can sequence a genome but cannot grapple with what it means to be human. They are fluent in the language of data and illiterate in the language of meaning.
This is the fruit of STEM education—Science, Technology, Engineering, Mathematics—a curriculum designed to produce specialists and succeeding beyond anyone’s wildest nightmares.
It is time to add the A. STEAM: Science, Technology, Engineering, Arts, Mathematics.
Not because art is pretty. Because without it, we are raising a civilization that can build anything and understand nothing.
The Irony Nobody Mentions
Here is a fact that should haunt every STEM advocate: the scientific revolution was created by people trained in the classical liberal arts.
Newton read theology. Darwin studied classics. Einstein played violin and conducted thought experiments that were acts of pure imagination—a boy riding a beam of light. Feynman cracked safes and played bongo drums and insisted that the first principle of science was that you must not fool yourself, and you are the easiest person to fool.
Leonardo da Vinci was not a scientist or an artist. He was both, simultaneously, and that is precisely why he saw what others could not.
We have spent two generations telling students that the arts are electives—nice diversions for those who cannot handle real subjects. We have produced more STEM graduates than any nation in history. And breakthrough innovation has declined while technical competence has risen.
Correlation is not causation. But this particular correlation deserves more than a shrug.
Lord Byron Explains Thermodynamics
The STEM purists will object: artists cannot do real science. The arts are soft, subjective, unrigorous. Physics requires a physicist.
Very well. Let us test that hypothesis.
I propose to demonstrate that George Gordon, Lord Byron—the Romantic poet, the libertine, the man who scandalized Europe, who was described by Lady Caroline Lamb as “mad, bad, and dangerous to know”—could explain thermodynamics. Not as metaphor. As physics. With equations.
If an arts-trained mind can teach you the laws that govern energy and entropy, perhaps the boundary between disciplines is less firm than the curriculum committees believe.
Imagine, if you will, Lord Byron rising from the grave, adjusting his cravat, and delivering the following lecture:
“Thermodynamics,” Byron begins, “is the poetry of the universe’s deepest truth: that all things tend toward disorder, and energy, though never destroyed, forever degrades. The Romantics understood this before the physicists named it. We called it mortality. They call it entropy.”
“The First Law is conservation. Energy cannot be created or destroyed, only transformed. Your equation: ΔU = Q − W. The change in internal energy of a system equals heat added minus work done by the system. This is the universe’s accounting principle—nothing is free, everything must balance. I spent my fortune as freely as heat dissipates into a cold room, and I assure you, the books always balanced in the end. Every excess extracts its price.”
“The Second Law is tragedy itself. Entropy in an isolated system never decreases: ΔS ≥ 0. Order dissolves into chaos. Hot flows to cold, never the reverse, not without external work. Your coffee cools. Your castles crumble. Your empires fall. I wrote Childe Harold’s Pilgrimage wandering through the ruins of Greece and Rome—I was documenting entropy in marble and memory. The physicists merely added the mathematics.”
“Gibbs free energy: G = H − TS. Enthalpy minus temperature times entropy. This determines spontaneity—whether a reaction will proceed without external force. When ΔG is negative, the reaction runs freely, like passion unchecked. When positive, it requires energy to proceed, like duty against desire. At equilibrium, ΔG = 0, and nothing changes—the death of all drama.”
“The Carnot efficiency sets the ceiling on what any heat engine can achieve: η = 1 − (Tc/Th). The ratio of cold reservoir to hot reservoir temperature. Even the most perfect engine wastes heat. Even the most brilliant mind loses energy to friction and despair. This is not pessimism. This is physics. The universe does not permit perpetual motion—neither in machines nor in men.”
“The Boltzmann equation links entropy to probability: S = kB ln W. Entropy equals Boltzmann’s constant times the natural logarithm of the number of microstates. This is the statistical truth beneath the classical laws: disorder wins because there are more ways to be disordered than ordered. A sonnet has one perfect arrangement. Random letters have billions. Given time, the sonnet dissolves. So too the poet.”
“But here is what your STEM curriculum will not teach you: understanding these equations is not the same as understanding what they mean. I have given you the mathematics. A physicist could have done the same. What I give you additionally is this: the Second Law is not a formula. It is the formal expression of the human condition. We are all isolated systems trending toward maximum entropy. The question is not whether we will dissipate—we will—but what work we will do before equilibrium claims us.”
Byron straightens his collar, offers a theatrical bow.
“Your examination is complete. I trust I have demonstrated that a man who wrote poetry and scandalized society and died fighting for Greek independence can also explain the laws governing energy transfer in thermodynamic systems. The question is not whether artists can learn physics. The question is why you believed they could not.”
The Point
That was not a trick. Every equation Byron cited was correct. The physics was sound.
What made the explanation powerful was not the mathematics—any textbook contains the mathematics. What made it powerful was the meaning. The connection between entropy and empire. The link between Gibbs free energy and human motivation. The recognition that Boltzmann’s equation describes not just gas molecules but the trajectory of civilizations.
A STEM education can teach you ΔS ≥ 0. It cannot teach you why that truth matters. It cannot teach you to see the Second Law operating in history, in literature, in your own brief and disordering life.
That requires the arts.
What Art Actually Teaches
The arts are not decoration. They are training grounds for the mind.
Pattern recognition across domains. A physicist who has studied music theory recognizes harmonic structures in data. A biologist who has read Shakespeare understands that systems have motives, conflicts, resolutions. The great scientists imported metaphors from outside their fields. A mind trained only in equations sees only equations.
Communication. A discovery that cannot be explained is useless. Every scientific paper, every grant proposal, every pitch deck is an act of persuasion. The humanities teach persuasion. They teach how to move a human mind with words.
Ethical reasoning. We have learned to split atoms and edit genes and build machines that think. Someone must ask whether we should. The question “Should we?” lives in philosophy, in literature, in history’s catalog of consequences.
Comfort with ambiguity. Mathematics has right answers. Art does not. Neither does the frontier of any field worth exploring. The researcher who cannot sit with uncertainty will never discover anything new.
A Life as Proof
I am not arguing theory. I am arguing from lived experience.
At nine years old, I made a list of four impossible things I wanted to become: shark biologist, mercenary, Army Ranger, brain scientist. I became all four.
As a research biophysicist, I achieved the first successful culturing of shark cells. As an Army Ranger in the 1st Battalion, 75th Ranger Regiment, I learned that no plan survives first contact and adaptation is everything. I worked on more than fifty books including The Da Vinci Code and became a New York Times bestselling ghostwriter and editor with two Pulitzer Prize nominations.
None of this happened despite the cross-training. It happened because of it.
The Renaissance produced polymaths. Our education system produces specialists. We should not be surprised that we have stopped having renaissances.
The Call
The change is simple to describe and difficult to implement.
Add the A. Make it STEAM.
Not as an afterthought. Not as an elective. As a core component of technical education. Require the engineering student to take ethics. Require the computer scientist to study history. Require the physicist to read literature and write about it with something approaching competence.
We stand at a moment when artificial intelligence can already code, calculate, and analyze. The machines are coming for the technical skills we have spent decades prioritizing. What they cannot do—not yet, perhaps not ever—is understand meaning, navigate ambiguity, ask questions that have not been asked before.
Those are human capacities. The arts cultivate them. STEM alone does not.
Lord Byron understood thermodynamics. He also understood that thermodynamics describes the human condition. That combination—technical knowledge and humanistic insight—is what we are failing to teach.
STEAM. The A matters.
Add it. Before we forget what we were capable of when we still trained humans to be fully human.
25 February 2026
The Unlearning Mandate
Why Professional Schools Teach What Professionals Must Forget
Every profession has its initiation ritual, and it sounds the same everywhere: Forget everything you learned in school.
The senior partner tells the new associate. The attending physician tells the resident. The platoon sergeant tells the freshly minted lieutenant. The principal engineer tells the new hire clutching their diploma. It’s not advice. It is a mandate. And its universality should trouble us deeply.
This isn’t the complaint of a few disgruntled practitioners about one mediocre program. It’s the consensus assessment of entire professions about the institutions charged with producing their successors. When every graduate of every medical school, law school, business school, engineering program, and military academy receives the same counsel—unlearn—we’re not observing individual institutional failure. We’re observing systemic design dysfunction.
The question isn’t whether professional schools are failing. They manifestly are. The question is why the failure persists, what it costs, and what would replace it if we were honest enough to start over.
The Price of Unlearning
Before examining why this dysfunction persists, consider what it costs. The numbers are staggering—and they represent pure waste, resources consumed not to develop competence but to undo miseducation.
The education investment. Americans currently hold $1.7 trillion in federal student loan debt, with graduate students responsible for over half of new loans originated annually. According to the Association of American Medical Colleges, the average medical school graduate pays $229,000 in tuition alone—$371,000 including undergraduate prerequisites. Law school at a top institution runs $90,000 per year before living expenses. MBA programs average $50,000 annually. These figures have increased by over 120% in the past three decades. The nation invests hundreds of billions of dollars in professional education every year—and the universal professional consensus is that much of what this investment purchases must be discarded upon entry to actual practice.
The retraining cost. The expense doesn’t end at graduation. According to BambooHR research, employers spend an additional $7,500 to $28,000 in hard costs per new hire for onboarding alone—recruiting, equipment, administrative processing. But the real cost is productivity loss during the unlearning and relearning period. Studies show that new employees function at approximately 25% productivity during their first four weeks. It takes six to twelve months—sometimes longer—for a new professional to reach full productivity. For a mid-level employee earning $60,000, twelve weeks of reduced productivity represents roughly $10,000 in lost output. Scale that across professions: each year the United States produces approximately 21,590 new physicians, 38,937 new lawyers, and over 200,000 MBAs. The aggregate productivity loss during the mandatory unlearning period runs into tens of billions of dollars annually.
The mentor burden. Every hour a senior professional spends teaching a new graduate what the school should have taught is an hour not spent on productive work. Law firm analyses estimate that training a first-year associate costs $70,000 in partner time at billable rates—100 hours of training at $500 per hour, plus 50 hours of recruiting and interviewing. Medical residency programs cost teaching hospitals approximately $180,000 per resident per year beyond what federal GME funding covers, much of it in attending physician time supervising procedures the resident was supposedly trained to perform. The hidden curriculum of real practice is taught on the employer’s dime, by practitioners whose time is the organization’s most valuable asset.
The error cost. Professionals operating on school-learned protocols before they’ve learned when those protocols fail make preventable mistakes. A Johns Hopkins study published in The BMJ estimated that more than 250,000 Americans die each year from medical errors—making it the third leading cause of death in the United States. The National Institutes of Health reports that approximately 400,000 hospitalized patients experience preventable harm annually, with healthcare costs from medical errors estimated at $20 billion each year. Not all are attributable to inadequate training, but the correlation between experience and error rates is well-documented. Junior associates at law firms have their work heavily reviewed precisely because their school-trained judgment cannot yet be trusted. The cost of these errors—in lives, in legal liability, in reputation, in remediation—is incalculable but enormous.
Sum it up: Americans spend approximately $100 billion annually on professional education. Employers then spend additional billions retraining graduates in what the schools failed to teach. Productivity losses during the unlearning period likely exceed $50 billion annually across all professions. Senior professionals sacrifice hundreds of millions of hours mentoring away school-induced misconceptions. And preventable errors by inadequately prepared professionals cost additional billions—plus lives that cannot be quantified.
A conservative estimate: the professional school dysfunction costs the American economy north of $200 billion per year. This is not the cost of education. It is the cost of miseducation—the price of teaching people what they must then forget.
The Evidence Is Universal
Consider medicine. Medical schools teach pathophysiology, pharmacology, anatomy—the science of disease. They do not teach the practice of medicine: how to make decisions with incomplete information under time pressure, how to triage when resources are scarce, how to have the conversation with a family about prognosis, how to manage a panel of 2,000 patients when the textbook assumes you have unlimited time for each one. Residents learn quickly that the protocols they memorized are starting points, not destinations. The art is knowing when to violate them.
Law school teaches legal reasoning—how to parse a case, construct an argument, identify issues. It does not teach how to manage a client who lies to you, how to negotiate with opposing counsel who has no intention of negotiating in good faith, how to run a profitable practice, or how to navigate the courthouse relationships that determine which motions get heard and which get buried. The Socratic method produces people who can analyze appellate decisions. It does not produce trial lawyers. First-year associates are expected to bill 1,900 hours annually at major firms—yet much of their initial output requires extensive partner review and revision.
Business schools are perhaps the worst offenders. The MBA case study method analyzes decisions after the fact with complete information. Real business operates in fog—incomplete data, political constraints, time pressure the classroom never simulates. The student learns what General Electric should have done in 1987. They don’t learn how to fire someone, how to survive a board meeting where half the members want your job, or how to make payroll when your biggest customer just went bankrupt.
Engineering programs teach theory—idealized systems, frictionless planes, spherical cows. Working engineers discover that nothing performs to specification, every system has undocumented failure modes, and the most important skill is debugging problems that according to theory cannot exist.
Military education is supposed to be different—explicitly practical, forged in the knowledge that graduates will face lethal consequences for failure. Yet the refrain persists. Officer candidates learn tactics, doctrine, leadership principles. The Army recognizes that the learning curve for leadership is steep, and that interpersonal challenges—not technical or tactical ones—dominate the transition to real command. Every veteran knows: your first deployment is your actual education. Everything before it was prologue.
The Machinery of Dysfunction
If the problem is universal, obvious, and enormously expensive, why does it persist? Because powerful forces maintain the equilibrium.
Faculty incentive misalignment. Professors are selected, evaluated, and promoted based on academic credentials—publications, research grants, theoretical contributions—not practitioner excellence. The best trial lawyer, the most accomplished surgeon, the most effective combat commander has no clear pathway to running the school that produces their replacements. Academia rewards academics. The people who could teach practice have no structural reason to teach, and the people who do teach have no structural experience in practice.
Credentialism over competence. Professional schools serve a gatekeeping function that has decoupled from competence validation. The degree certifies exposure to curriculum, not capability to perform. Employers accept this because they have no better filtering mechanism at scale. The diploma answers the question “Did this person complete the required obstacle course?” not “Can this person do the job?” As long as the credential retains its signaling value, the institution has no pressure to change what it signals.
Liability and standardization pressure. Schools teach defensible practice—what can be documented, what follows protocols, what survives review. Practitioners learn that rigid protocol adherence often produces worse outcomes than adaptive judgment. But adaptive judgment is hard to teach, impossible to standardize, and legally risky to advocate. So schools teach the protocol, and practice teaches when to break it.
The lag problem. Professional practice evolves faster than curricula can adapt. By the time a practice becomes standard enough to be “teachable”—codified, approved by committees, embedded in textbooks—the cutting edge has moved. Schools teach yesterday’s best practices because those are the only practices stable enough to institutionalize.
The tacit knowledge gap. The most critical professional knowledge is tacit—embodied, contextual, judgment-based. It cannot be articulated in propositions or procedures. It transfers through apprenticeship, not lecture. Professional schools are structured for explicit knowledge transmission: information that can be written down, tested, graded. But the knowledge that distinguishes competent practitioners from excellent ones resists that structure entirely.
The Redesign Principles
If we accept that “forget what you learned” is diagnostic of systemic failure, what would reality-aligned professional education look like?
Practitioner-led instruction. Faculty should be rotating practitioners at the peak of active careers, not retired practitioners or career academics. This requires restructuring compensation (practitioners earn more than professors), tenure (incompatible with rotation), and institutional governance (academics won’t vote to dilute their authority). The German model of professional training integrates classroom instruction with workplace experience under master practitioners. The American model front-loads theory and backloads practice, ensuring students must unlearn before they can perform.
Simulated constraint environments. Training should occur under realistic resource, time, and information constraints. Medical students should learn triage in simulated mass casualty events, not idealized single-patient encounters. Law students should negotiate with incomplete information against adversaries with hidden agendas. Business students should make decisions under time pressure with data that is both insufficient and contradictory—because that’s what business is. Military schools have embraced this principle more than most, but even they sanitize the chaos.
Failure-centric learning. Professionals learn most from failures, but schools are structured around success demonstration. Curricula should include deliberate failure experiences—simulations designed to produce bad outcomes that must then be analyzed. The question shifts from “Did you get it right?” to “Do you understand why you got it wrong, and what you will do differently?”
Adaptive judgment over protocol mastery. Teach principles and frameworks, not rigid protocols. More importantly, teach when protocols should be violated. The difference between a competent professional and an excellent one is not better protocol adherence—it’s knowing which situations call for departure from standard practice. This is the knowledge no textbook contains and no examination tests.
Apprenticeship integration. Reduce classroom time; increase embedded practice time under supervision. Medical residency gets this partly right, but it begins too late—after years of classroom instruction that must be unlearned. The integration should happen throughout, not sequentially.
The Resistance
None of this will happen through gradual reform. The forces maintaining the current equilibrium are too strong.
Accreditation bodies are controlled by academics who define standards in academic terms. Professional associations benefit from high barriers to entry that schools provide—limiting supply protects incumbents. Students have been conditioned to seek credentials rather than competence; they’ll pay for the diploma because employers demand it. Employers lack the capacity or will to evaluate competence directly, so they use degrees as proxies—perpetuating demand for degrees regardless of what they actually indicate. And faculty will not voluntarily surrender control to practitioners who threaten their authority and their jobs.
The equilibrium breaks only under external pressure. War reforms military education—after enough failures make the cost of dysfunction undeniable. Catastrophic errors reform medical education—eventually. Market disruption forces business school evolution—the rise of coding bootcamps demonstrated that employers will accept alternative credentials when traditional ones fail to produce usable skills.
The Strategic Stakes
This is not merely an educational policy question. It’s a national competitiveness question.
Adversaries who train practitioners rather than credential-holders gain asymmetric advantage. A nation that produces lawyers who can try cases, engineers who can build things that work, military officers who can lead under fire, and physicians who can heal patients will outcompete a nation that produces people who can pass examinations about lawyering, engineering, commanding, and doctoring. The gap between credential and competence is a strategic vulnerability.
The forcing function may be artificial intelligence. When explicit knowledge becomes commoditized—when machines can recite protocols, cite precedents, list differential diagnoses better than any human—only tacit judgment and adaptive expertise retain value. Schools that produce protocol-followers will become obviously obsolete. The question is whether our institutions will adapt before that obsolescence translates into national decline.
The Honest Question
Professional schools persist because they answer a question nobody is asking honestly: How do we filter aspirants to protect incumbent practitioners while appearing to prepare successors?
The honest question would be: How do we produce people who can actually do this work, starting on day one?
The universal mandate to “forget what you learned” is the professions’ confession that their schools are not answering that question. Every time a seasoned practitioner delivers that advice to a new graduate, they’re admitting that the institution charged with preparation has failed at preparation.
We could design schools that don’t require unlearning. We know what that would look like. We lack only the honesty to admit the current system is broken and the will to build something that works.
The students keep paying—now $1.7 trillion in accumulated debt. The diplomas keep issuing. The practitioners keep advising graduates to forget. And the dysfunction continues, costing the economy hundreds of billions annually while producing professionals who must be retrained to do their jobs.
The bill comes due every year. We just pretend not to see it.