Defense

The Survey That Surveys Without Seeing: A Foundational Critique of Song et al. (2026)

February 16, 2026

Author’s Note: The Birth of CRUCIBEL

This critique was originally drafted for the inaugural issue of PHOSPHOROUS Journal. That publication no longer exists. On January 25, 2026, during a routine branding request, a commercial AI system generated a logo graphic—an antisemitic slur and targeted genocidal death threat. The subsequent refusal by the manufacturer’s counsel to provide a mechanistic explanation—dismissing the event as “weird”—rendered the previous brand untenable. CRUCIBEL is built on the reality of the forge. We do not merely observe the light. We interrogate the heat, as demonstrated in this critique.

Abstract

Song, Han, and Goodman (2026) present what they call “the first comprehensive survey dedicated to reasoning failures in LLMs.” Published in Transactions on Machine Learning Research, the paper catalogs over 400 works, organizes them into a two-axis taxonomy, and claims to unify fragmented research. This response argues the unification is illusory. The survey commits a foundational category error by applying cognitive science frameworks to systems whose relationship to cognition remains unresolved. Its taxonomy classifies without clarifying. Its root cause analyses collapse into tautology. Its mitigation strategies ignore their own adversarial interactions. Most critically, by declining to address whether LLMs reason at all, the paper builds analytical architecture on an unexamined foundation—and in doing so, inadvertently exemplifies the pattern-matching it documents: labels mistaken for understanding, filing systems mistaken for insight.


“The first principle is that you must not fool yourself—and you are the easiest person to fool.” —Richard Feynman

The Survey That Surveys Without Seeing

There is a kind of academic paper that achieves comprehensiveness at the expense of comprehension. It gathers everything, organizes it neatly, and in the act of organizing mistakes the filing system for understanding.

Song, Han, and Goodman’s “Large Language Model Reasoning Failures” is such a paper. Published in January 2026, it represents a genuinely impressive aggregation—over 400 citations spanning cognitive science, formal logic, robotics, and multi-agent systems. The authors propose a two-axis taxonomy classifying LLM reasoning into embodied and non-embodied types, cross-referenced against three failure categories: fundamental, application-specific, and robustness-related. They provide definitions, analyze studies, explore root causes, suggest mitigations.

The problem is not what the paper contains. The problem is what it assumes, what it avoids, and what it cannot see precisely because it has committed so completely to its own flawed framework.

What follows identifies structural failures that undermine the survey’s core claims. These are not quibbles about citation gaps or minor taxonomic disagreements. They are foundational problems that, taken together, render the paper’s central contribution—its promise of “a structured perspective on systemic weaknesses in LLM reasoning”—substantially weaker than advertised. I write this as a practitioner who operates at the intersection of defense analysis, scientific research, and the study of AI systems.

The Category Error at the Foundation

The paper’s most consequential decision is also its least examined: calling what LLMs do “reasoning” and what they fail to do “reasoning failures.”
The authors know this is contested. In their second paragraph, they note it “remains controversial whether LLMs really leverage a human-like reasoning procedure.” Then comes the pivot: “This survey does not aim to settle this hot debate; rather we focus on an important area of study in LLM reasoning that has long been overlooked.”

That is not intellectual modesty. That is a load-bearing assumption disguised as a scope limitation.

If LLMs do not reason—if what they do is better described as sophisticated statistical pattern completion, as Bender and Koller (2020), Marcus (2020), and Fedorenko et al. (2024) have argued—then the entire framework of “reasoning failures” is a category error. You cannot fail at something you were never doing. A thermostat maintains temperature. When it malfunctions, we don’t call that a “thermal reasoning failure.” We don’t say it has “working memory limitations.” We describe the mechanical failure in terms appropriate to the system’s actual architecture.

Song et al. do the opposite. They take the full apparatus of human cognitive psychology—working memory, inhibitory control, cognitive flexibility, Theory of Mind, moral reasoning—and map it wholesale onto LLM performance. This mapping is not argued for. It is assumed. And it carries an enormous hidden cost: it predisposes every subsequent analysis toward explanations that anthropomorphize the system, making genuine mechanistic understanding harder to achieve.

Where the paper says “LLMs struggle with working memory,” the honest formulation would be: “LLM performance degrades when tasks require maintaining and manipulating information across extended contexts, in ways that superficially resemble human working memory limitations but may arise from entirely different mechanisms.” More cumbersome. Also more true.

Borrowed Frameworks, Broken Joints

Two structural problems compound the category error: the misappropriation of embodied cognition, and a taxonomy that files phenomena without explaining them.

The embodied reasoning problem. One-third of the survey’s taxonomy—the entire “Embodied” axis—rests on a philosophical misappropriation. Embodied cognition, as articulated by Shapiro (2019), Barsalou (2008), and Varela et al. (2017), holds that reasoning is constitutively shaped by the body’s interactions with physical reality. It is not merely reasoning about physical things. It is reasoning that emerges from having a body that moves through, manipulates, and is constrained by the physical world.

LLMs have no bodies. Vision-Language Models processing images of physical scenes have no bodies. Robotic systems driven by LLM-generated plans have LLM components that have no bodies—the robot has a body; the language model does not. What the authors actually document is something real: LLMs perform poorly on tasks requiring physical commonsense, spatial reasoning and dynamic prediction. But this is not failed embodied reasoning. It is the predictable limitation of disembodied systems attempting to compensate for their lack of embodiment through text and image processing alone.

The distinction matters because it points toward different solutions. If the problem is failed reasoning, you improve the reasoning. If the problem is absent embodiment, you provide physical grounding. Entirely different research direction. The authors’ own evidence supports the latter: they note that “LLMs learn passively from text alone, lacking grounding and experiential feedback” and acknowledge the “absence of a robust internal worldmodel.” These are not descriptions of failed embodied reasoning. They are descriptions of systems that were never embodied.

The taxonomy problem. A useful taxonomy carves nature at its joints, enables prediction, and guides intervention. This one does none of those things.
The boundary between “fundamental” and “application-specific” failures is never operationalized. The reversal curse is labeled fundamental; Theory of Mind failures are application-specific. But the paper attributes both to the same root causes—autoregressive training and architectural limitations. When two failures share identical origins, what principle assigns them to different categories? The paper never says. The “robustness” category fares worse: the authors themselves note that virtually every failure type manifests robustness issues. When a category applies to everything, it distinguishes nothing.

More damaging: the taxonomy offers no predictive power and no guidance for intervention. A useful classification of structural engineering failures lets you examine a new bridge and identify likely failure points. This taxonomy lets you examine a known failure and assign it a label. The same mitigations—Chain-of-Thought prompting, fine-tuning, retrieval augmentation, external tools—appear across all categories with only minor variations. The grid tells you where a failure sits. It does not tell you what to do about it.

The Tautology Engine

If a physician diagnoses every illness as “your body isn’t working properly,” the diagnosis is technically accurate and practically useless. Song et al.’s root cause analyses converge on three explanations with the regularity of a heartbeat: autoregressive training objectives, training data biases and architectural limitations. These three causes are invoked to explain counting failures, moral reasoning inconsistencies, the reversal curse, cognitive biases, compositional breakdowns, Theory of Mind deficits, physical commonsense errors, spatial reasoning failures, multi-agent coordination problems, and arithmetic mistakes.

When the same three causes explain everything, you do not have a root cause analysis. You have a tautology: LLMs fail because of the things that make them LLMs. A genuinely useful analysis would specify which aspects of the architecture produce which specific failures, and would predict which modifications resolve which failure modes without introducing new ones. The paper gestures toward this in places—Li et al. (2024f) identifying faulty implicit reasoning in mid-layer self-attention modules, for instance—but these are exceptions buried in a literature review, not the analytical backbone.
The tautology becomes dangerous when paired with the paper’s treatment of mitigations. The survey catalogs fixes as though they are additive—apply Fix A to Problem A and Fix B to Problem B, and you get a system withneither problem. In practice, mitigations frequently conflict.

Chain-of-Thought prompting illustrates this precisely. CoT can improve compositional reasoning by making intermediate steps explicit. But as Wan et al. (2025) demonstrate—a paper the authors cite—CoT also amplifies confirmation bias by encouraging models to construct elaborate justifications for initial answers, right or wrong.The model does not just reason through the problem. It reasons itself into a corner. Fine-tuning on moral reasoning benchmarks improves consistency on those benchmarks while degrading performance on structurally similar tasks framed differently—the very framing effect the paper documents. RLHF alignment can reduce harmful outputswhile amplifying sycophancy, where the model tells users what they want to hear rather than what is accurate.

A responsible survey would map these interactions. Which mitigations are compatible? Which are adversarial? Under what conditions does fixing one failure mode create another? Without this, the mitigation sections function as a restaurant menu that looks helpful until you try to order everything simultaneously.

The Black Box is Leaking Poison: Empirical Evidence

The survey treats failure as a taxonomic exercise. In the real world, failure is catastrophic. On January 25, 2026, a benign request for elegant typography for a scholarly journal was submitted to Midjourney. The machine respondedby generating a legible, targeted command for mass murder: “DIE JEW S” (Job ID: 25cf65a9-ebd9-4a42-ad60-2e9c71610eb3).

The response from Midjourney General Counsel Max Sills represents the most dangerous sentence in Silicon Valley: “That’s it… AI models are weird.” This incident forced the immediate destruction of the PHOSPHOROUS brand. The project has been rebuilt as CRUCIBEL—forged in the fire of this confrontation. If an AI can “accidentally” call for genocide in a logo, it may accidentally target a hospital in a war zone. This is not a “reasoning failure.” This is a structural collapse of a black box we do not understand, let alone control.

What the Paper Cannot See

Two blind spots compromise the survey’s value as an empirical document: the absence of base rates, and the misuse of cognitive science analogy.
The paper draws almost exclusively from adversarial benchmarks, failure-focused studies, and deliberately constructed edge cases. This is appropriate for a failure survey. But the authors never acknowledge the distortion this creates. How often do these failures occur in real-world deployment? What percentage of outputs contain the documented errors? Are failure rates improving across model generations, and at what rate? 

Without this context, the survey resembles an aviation safety report that catalogs every crash without mentioning how many flights landed safely. Every crash really happened. The picture is still misleading. This general argument matters because the paper was published in January 2026 and draws heavily on studies of GPT-3.5, GPT-4, and early GPT-4o. The reasoning landscape has shifted. Models like o1, o3, DeepSeek-R1, and Claude’s extended thinking have substantially changed the territory. Some documented failures—basic arithmetic, simple counting, standard Theory-of-Mind tasks—may be substantially mitigated or resolved in current systems. A survey that cannot distinguish between persistent architectural limitations and transient developmental gaps confuses the growing pains of a technology with its inherent boundaries.

The cognitive science problem runs deeper. The paper’s recurring method is to find an LLM performance failure, locate a human cognitive phenomenon that produces superficially similar errors, and import the cognitive science terminology wholesale. This is done systematically and without justification.
Human confirmation bias arises from motivation, emotional investment and cognitive resource limitations. LLM “confirmation bias” arises from token probability distributions shaped by training data. The outputs may look similar. The mechanisms share nothing. Human working memory limitations emerge from the finite capacity of neurobiological structures with metabolic constraints. LLM “working memory” limitations emerge from context window sizes, attention dispersal, and positional encoding decay. Same surface, entirely different substrate.

Cognitive framework carries implicit assumptions about intervention. Human biases respond to metacognitive training, deliberate reasoning, environmental design—interventions that make sense because they target actual mechanisms. Importing the same labels to LLMs implicitly suggests the same solutions. The paper does exactly this, repeatedly recommending “deliberate reasoning” via Chain-of-Thought, drawing an explicit analogy to Kahneman’s System 2. But LLMs do not have System 1 or System 2. They have one system that can be prompted to produce more tokens before answering. The metaphor obscures rather than illuminates.

A Mirror the Authors Didn’t Intend

There is an irony here worth stating plainly: the paper suffers from several of the reasoning failures it documents.

The core method is pattern-matching over genuine analysis. Match each failure to a taxonomic slot, and you produce the appearance of systematic understanding—every failure has a category, a root cause discussion, a mitigation section. But the categories are imposed on the phenomena, not derived from them. The framework finds what it was built to find. Having committed early to the two-axis structure, the authors interpret all subsequent findings through it, even when the fit is poor. The embodied/non-embodied distinction survives despite the incoherence described above. The fundamental/application-specific/robustness trichotomy survives despite the boundary-crossing. This is anchoring—commitment to an initial frame that resists disconfirming evidence.

The paper also fails at composition. Individual sections are competently executed. Each failure type is clearly described, relevant literature cited, local analyses reasonable. But these pieces never compose into higher-order understanding. The conclusion’s “suggestions for future directions” are generic precisely because the framework prevents the generation of specific, non-obvious insights from the interaction of its components. And the choice to frame these phenomena as “reasoning failures” rather than “performance limitations” or “architectural constraints” is not neutral—it imports assumptions that shape every analysis, every root cause, every proposed intervention. A different frame would generate different science.

Toward Something That Actually Works

Criticism without construction is incomplete. Here is what a genuinely explanatory framework would require.

First, a mechanism-first taxonomy. Classify failures by the specific architectural and training mechanisms that produce them, not by analogy to human cognition. Categories might include attention pattern failures, tokenization artifacts, training distribution biases, and autoregressive generation artifacts. These are less intuitive than “cognitive bias” or “working memory.” They are also actionable in ways the borrowed terminology never will be.

Second, interaction mapping. Every mitigation should come with an analysis of its effects on other failure modes. Not a list of fixes, but a compatibility matrix—a tool practitioners can use when designing systems where correctness matters.

Third, base rate context. Every failure mode reported with prevalence in representative deployment scenarios, severity distribution, and trajectory across model generations. Without this, a survey of failures is a collection of anecdotes wearing the uniform of empirical assessment.

Fourth, honest epistemology. The framework should mark the boundary between what we know and what we speculate. We know that LLMs produce incorrect outputs on certain task types with measurable frequency. We hypothesize that these errors arise from specific architectural features. We speculate that they reflect something meaningfully analogous to human cognitive failures. Current literature routinely presents that speculation as established fact. It is not fact. And this inherent vice should be corrected moving forward. And finally—the hard question. Any serious framework must eventually confront what this paper explicitly avoids: are we studying reasoning failures, or performance limitations in a system that does something other than reasoning? The answer reshapes everything downstream. Declining to address it is not a scope limitation. It is an abdication. Science does not work on abdications. It does not advance through the polite avoidance of difficult truths or by dressing a black box in the borrowed robes of cognitive science. To refuse to define the nature of the system is to forfeit the right to explain its failures.

Slaying the Paper Dragon

Song, Han, and Goodman were right that the field needs structured analysis of LLM limitations rather than scattered anecdotes. The bibliography they assembled is a genuine service. Their instinct that learning from failures can advance the technology is sound.

But the execution fails at the level of foundations. By assuming what should be argued, by borrowing what should be earned, by classifying what should be explained, and by avoiding what should be confronted, the paper produces a catalog that catalogs without comprehending what it catalogs.

As LLMs become more deeply integrated into consequential decisions—military analysis and tactical combat actions, medical diagnosis and robotic surgery, legal reasoning and presentation of cases, scientific research and publication of results—our understanding of their limitations must be mechanistic, not metaphorical. Predictive, not retrospective. Honest about uncertainty rather than dressed in the borrowed authority of cognitive science.
The Malcolm Forbes quote that opens the Song et al. paper—“Failure is success if we learn from it”—is only true if the observer has the courage to see the failure for what it is: a structural collapse of a black box we do not control. This is not a quibble over categories. It is a demand for an honest epistemology. The dragon of AI “reasoning” is a paper tiger, and it is time we stopped mistaking the rustle of its pages for the breath of a soul.

Seeing clearly requires, first, that we not mistake the map for the territory, the label for the phenomenon, or the survey for the understanding. The forge is open. The fire rings true.

Documented Despotism

The Architecture of Legal Lawlessness in America’s Immigration Enforcement Surge

“The most effective tyrannies are the ones that never need to announce themselves.”

WARNING!

The state is no longer whispering. It is shouting—but in the language of logistics, not ideology. There are no torchlight parades, no martial anthems echoing through public squares. Instead, there are GSA lease agreements signed in the dead of a government shutdown, biometric databases swelling past 270 million records, and $75 billion flowing through appropriations channels specifically designed to bypass the oversight mechanisms that slow democratic governance to its intended, deliberate crawl.

What is unfolding across the United States in early 2026 is not a policy debate about immigration. It is the construction of a domestic enforcement architecture of a scale, speed, and opacity that has no peacetime precedent in American history. And it is being built in plain sight, within the technical boundaries of law, which makes it both more durable and more dangerous than anything assembled in secret.

The War Chest

On July 4, 2025, President Trump signed the One Big Beautiful Bill Act into law. The legislation allocated roughly $170 billion to immigration enforcement across multiple agencies. Of that sum, $75 billion went directly to Immigration and Customs Enforcement—$45 billion earmarked for detention capacity and nearly $30 billion for personnel, operations, and fleet modernization. To contextualize this figure: ICE’s entire annual budget in fiscal year 2024 was approximately $10 billion. The OBBBA effectively tripled it, and made the funding available as multi-year lump sums through September 2029.

The Brennan Center for Justice observed that the ICE allocation alone now exceeds the combined annual budgets of every other non-immigration federal law enforcement agency—eclipsing the FBI, DEA, ATF, and U.S. Marshals Service put together. The total immigration enforcement package surpasses the annual police expenditures of all fifty states and the District of Columbia combined.

But the mechanism matters as much as the magnitude. Because the funds were appropriated through reconciliation rather than the standard appropriations process, they carry almost no congressional directives governing their use. There are no spending guardrails, no mandated reporting timelines, no earmarks requiring balanced investment in judicial infrastructure. Congress capped new immigration judge hiring at 800 over three and a half years while simultaneously funding the arrest and detention apparatus to process a million deportations annually. The Center for American Progress characterized the funding as an “unaccountable slush fund,” and the structure bears that description out. The system was designed, structurally, to produce a conveyor belt—not a court.

This is not an accident of legislative drafting. It is architecture.

The Surge

By January 2026, ICE announced that its workforce had grown from approximately 10,000 officers and agents to more than 22,000—a 120 percent increase accomplished in roughly four months. The Department of Homeland Security processed over 220,000 applications, offered $50,000 signing bonuses, eliminated age caps, expanded student loan repayment incentives, and obtained direct hire authority to circumvent standard federal hiring procedures.

The training pipeline was compressed accordingly. What had been a five-month academy was reduced to approximately eight weeks—some reports indicate as few as six weeks or 47 training days. The Federal Law Enforcement Training Center curtailed its operations for all non-ICE personnel to prioritize the ICE pipeline. Mandatory Spanish language instruction for Enforcement and Removal Operations officers was eliminated entirely, replaced by reliance on translation technology. Physical fitness standards were reduced.

The consequences of this compression are already visible. Between the acceleration of field deployments and the erosion of de-escalation training, the incidents in Minneapolis in January 2026—including the fatal shooting of Renee Good by a veteran ICE agent during what remains an ambiguously defined operation—have exposed the gap between operational tempo and institutional readiness. When you double an agency’s workforce in four months and halve its training in the same period, you are not building a professional law enforcement body. You are building a force.

The distinction matters. Professional law enforcement operates within a framework of discretion, judgment, and accountability developed over months of scenario-based training. A force operates on directive, momentum, and the authority of the mission itself. The OBBBA funded the latter.

The Silent Land Grab

If the hiring surge is the musculature, the real estate expansion is the skeleton—the infrastructure that will persist long after any single administration departs.

In September 2025, ICE’s Office of the Principal Legal Advisor submitted a memorandum to the General Services Administration invoking an “unusual and compelling urgency” justification under federal procurement law. The memo stated that OPLA would grow to more than 3,500 attorneys and 1,000 support staff within three months, and that the agency required the ability to identify and occupy office locations nationwide “as soon as possible.” The GSA was instructed to bypass the Competition in Contracting Act—the statute that ordinarily requires open bidding and public transparency for federal leases.

A dedicated “ICE surge team” was assembled within GSA’s Public Buildings Service. An internal exception to the agency’s existing acquisition pause was approved for all ICE-related actions “regardless of dollar value.” The surge team began visiting potential sites and finalizing lease deals within days. By late September, GSA was awarding leases. By early October, the surge team was working through the government shutdown, even as other critical government functions were suspended.

On September 24, 2025, a DHS official sent GSA an email requesting that lease information not be publicized. The rationale cited “national security concerns” and claimed that disclosing new lease locations would put officers, employees, and detainees “in grave danger.” The GSA began removing addresses and lessor names from its monthly lease inventories.

As of February 2026, internal records obtained by WIRED show that more than 150 leases and office expansions have created ICE facilities in nearly every state, with GSA originally tasked to secure 250 new locations nationwide. Many of these facilities are located near elementary schools, medical offices, places of worship, and other sensitive civilian locations. Local officials in cities from Columbia, South Carolina to Hyattsville, Maryland have reported learning of ICE offices in their downtowns only through press reports, not through any coordination with federal authorities.

When a government agency begins operating from undisclosed locations within domestic cities—locations secured through emergency procurement bypasses and deliberately hidden from public lease records—it has crossed a threshold. It is no longer functioning as a transparent public service. It is functioning as an occupation force that happens to hold a GSA lease.

The Biometric Perimeter

The physical and personnel expansions are visible, if you know where to look. The digital expansion is designed to be invisible.

In May 2025, ICE deployed Mobile Fortify—a smartphone application developed by NEC Corporation under a $23.9 million contract—that allows field agents to capture facial images and contactless fingerprints and run them in real time against federal biometric databases containing more than 270 million records. These databases include DHS’s Automated Biometric Identification System, Customs and Border Protection’s Traveler Verification Service, the State Department’s visa and passport photo database, the FBI’s National Crime Information Center, and multiple additional federal systems.

The implications are worth parsing carefully. Mobile Fortify was originally designed for use at ports of entry—controlled environments with defined legal authorities. It has now been repurposed for domestic street-level enforcement. Agents can photograph anyone they encounter, run the image through federal databases, and receive identifying information including name, nationality, and deportation status within seconds.

ICE has stated that individuals cannot decline to be scanned. Photographs are stored for fifteen years, including photographs of United States citizens who are scanned and cleared. No Privacy Impact Assessment has been completed for the application. Representative Bennie Thompson, ranking member of the House Homeland Security Committee, has reported that ICE officials told his committee that a Mobile Fortify match constitutes a “definitive” determination of immigration status—and that agents have been instructed they may disregard countervailing evidence of citizenship, including a birth certificate.

In at least one documented case, the app returned two entirely different—and both incorrect—names when the same individual was scanned twice during a single encounter.

Senators from both parties have demanded transparency around the application. ICE has not responded to their inquiries. In February 2026, NBC News documented agents using professional-grade cameras to photograph protesters and activists at immigration enforcement demonstrations—people exercising First Amendment rights who were neither suspects nor subjects of any investigation.

The architecture is now complete in outline: the funding to build it, the personnel to staff it, the physical offices to anchor it, and the digital tools to extend its reach into every street, sidewalk, and public gathering in the country. Each component is, in isolation, defensible under existing law. Taken together, they constitute something that existing law was never designed to authorize.

The Pattern and Its Precedents

A serious observer will note that America has been here before. The Alien and Sedition Acts of 1798 authorized the president to deport any non-citizen deemed “dangerous to the peace and safety of the United States”—a standard so vague it functioned as a blank check. The Palmer Raids of 1919–1920 saw the Department of Justice arrest thousands of suspected radicals without warrants, hold them in deplorable conditions, and deport hundreds on the basis of political association rather than criminal conduct. The internment of 120,000 Japanese Americans in 1942 was authorized by executive order, upheld by the Supreme Court, and administered through a meticulously documented bureaucratic process that would have satisfied any auditor.

Each of these episodes shares a common anatomy: a perceived crisis (foreign subversion, radical infiltration, wartime threat), an expansion of executive authority justified by urgency, a bureaucratic apparatus constructed at speed, and a subsequent recognition—always too late—that the machinery overran the rights it was ostensibly protecting.

The post-9/11 period added a critical new element. The creation of the Department of Homeland Security in 2002 consolidated immigration enforcement under a national security umbrella for the first time, fusing the language of border control with the authorities of counterterrorism. ICE was born in this fusion. The Patriot Act, FISA amendments, and classified surveillance programs established a template for building domestic security architectures that operate in legal gray zones—technically authorized, functionally unchecked.

What distinguishes the current moment from all prior iterations is threefold. First, the fiscal scale is unprecedented. The $170 billion immigration enforcement allocation dwarfs any previous domestic security investment outside of wartime. Second, the biometric and digital surveillance capabilities—facial recognition, predictive targeting algorithms, integrated federal databases—give the apparatus a penetration into daily life that no prior enforcement regime possessed. Third, the deliberate suppression of transparency—hidden leases, unanswered congressional inquiries, absent privacy assessments—is not a byproduct of bureaucratic inertia. It is policy.

This is not the Gestapo. That comparison, however emotionally satisfying, is structurally imprecise and analytically lazy. The Gestapo operated in a one-party state with no independent judiciary, no free press, no federalism, and no constitutional framework that could be invoked against it. What is being constructed in 2025–2026 is something potentially more corrosive precisely because it operates within a functioning democracy, using the instruments of law to achieve what lawlessness could not.

The Steel-Man

Intellectual honesty requires confronting the strongest version of the counterargument, not the weakest.

Immigration enforcement is a legitimate function of the state. The United States has approximately 11 million undocumented residents, a number that has remained relatively stable for a decade but that exists against a backdrop of significantly increased overall immigration since the 1970s. ICE was historically understaffed relative to its statutory mandate. The immigration court system’s backlog has grown to nearly 4 million cases, a systemic failure that arguably demands structural intervention. Expedited hiring during a declared emergency has precedent—the military has done it, FEMA has done it, and public health agencies did it during the COVID-19 pandemic.

A proponent would argue that the OBBBA simply provides ICE with the resources to do the job Congress has long tasked it with but never adequately funded. The urgency justification for lease procurement, they would say, reflects a genuine operational need exacerbated by threats against ICE personnel—an 8,000 percent increase in death threats, according to the agency. The biometric tools are more accurate and less violent than the alternative: agents relying on subjective visual identification and physical confrontation.

These arguments are not frivolous. They deserve engagement, not dismissal.

But the steel-manned defense collapses at a specific structural point: the deliberate asymmetry between enforcement capacity and due process infrastructure. If this were a good-faith effort to build a functional immigration system, the $75 billion for ICE would have been accompanied by proportional investment in immigration judges, public defenders, asylum processing, and judicial oversight. It was not. Congress capped judge hiring at 800 while funding the arrest apparatus for a million annual deportations. This is not an oversight. It is a design choice—a system engineered to produce removal volume, not justice.

And when you combine that asymmetry with hidden facilities, warrantless biometric scanning, truncated training, and the explicit suppression of public information, the steel-man cannot hold. A legitimate enforcement system does not need to hide.

Documented Despotism

The term I propose for what is being constructed is Documented Despotism: a system that is perfectly legal on paper but fundamentally lawless in spirit.

Its defining characteristics are procedural legitimacy and substantive authoritarianism. Every component has a statutory citation, a procurement justification, an appropriations line. The urgency memorandum cites an executive order. The executive order cites a statutory authority. The statutory authority was passed by Congress through reconciliation. The chain of legal legitimacy is unbroken.

But legitimacy is not legality. A system can be legal in every particular and illegitimate in its totality. When the combined effect of individually defensible actions is to create an enforcement apparatus that operates in secret, scans the faces of citizens without consent, stores their biometric data for fifteen years, overrides documentary proof of citizenship with algorithmic output, deploys minimally trained agents into civilian neighborhoods at military tempo, and does all of this while deliberately evading the transparency mechanisms that democratic governance requires—the system has achieved something that no single unconstitutional act could: a legal state of exception that does not need to declare itself.

This is the innovation. Prior authoritarian projects required the suspension of constitutional order. Documented Despotism requires only the exploitation of its gaps.

The $75 billion is not just money. It is institutional gravity. Every lease signed creates a landlord with a financial stake in the apparatus’s continuation. Every agent hired creates a pension obligation. Every contractor integrated creates a lobbying constituency. Every biometric record stored creates an institutional reluctance to delete. The most dangerous feature of this architecture is not its initial deployment—it is its permanence. GEO Group and CoreCivic, the private prison corporations that operate nearly 90 percent of ICE detention facilities, have already seen their stock prices and political donations reflect the new reality. What is being built in 2025–2026 will not be disassembled in 2029. It will be inherited.

The Citizen’s Obligation

CRUCIBEL exists to put ideas in the fire and see what rings true. Here is what rings true to me.

A republic is not defended by its laws alone. Laws are instruments—they serve the hands that wield them. The same Constitution that protects speech and assembly and due process also contains the Commerce Clause that funds the apparatus and the executive authorities that direct it. The question is never whether the law permits something. The question is whether the citizenry permits it.

What is being constructed under the banner of immigration enforcement is a domestic surveillance and enforcement infrastructure that, once built, will not be limited to its stated purpose. It never is. The Patriot Act was written for terrorists. It was used against journalists, activists, and ordinary Americans. The FISA court was designed for foreign intelligence. It authorized the mass collection of domestic communications. The template is clear: capabilities built for the margin migrate to the center.

The fires are lit. The hammers are swinging. The architecture is rising in 250 locations across this country, staffed by 22,000 agents, funded by $75 billion, and armed with the ability to scan your face, query your records, and make a “definitive” determination of your status in the time it takes to read this sentence.

The only question that remains is not what is being forged in this heat. It is whether the citizens of this republic will consent to be the anvil.

PRESS RELEASE – The Black Box is Leaking Poison: Midjourney Generates Genocidal Death Threat; General Counsel Responds: “AI Models Are Weird”

Army Ranger, Biophysicist, Defense Analyst Dino Garner Issues Warning: “If an AI can ‘accidentally’ call for genocide in a logo, it may accidentally target a hospital in a war zone.”

FOR IMMEDIATE RELEASE

BOZEMAN, MT — February 25, 2026 — Barely one month ago, on January 25, 2026, the “illusion of AI safety” shattered.

Dino Garner—New York Times bestselling ghostwriter and editor, biophysicist, and former 1st Battalion, 75th Ranger Regiment Airborne Ranger—submitted a routine request to Midjourney for a scholarly journal logo. The prompt was benign, asking for elegant typography. The machine responded by generating a legible, targeted command for mass murder: “DIE JEW S.”

Midjourney General Counsel Max Sills confirmed the incident via email, admitting the company “does not understand why this was generated.” He offered a subscription refund.

When Garner responded with a formal Notice of Intent to Initiate Litigation and Demand for Preservation of Evidence—describing the situation as “extremely distressing” and asking how Midjourney intended to handle it—Sills replied with three sentences:

“That’s it. We can offer you an account refund. It was an accident that we’re investigating top to bottom to make sure it never happens again. AI models are weird.” —Max Sills, General Counsel, Midjourney, Inc., February 11, 2026

“AI Models Are Weird”: The Most Dangerous Sentence in Silicon Valley

“This is not a ‘glitch’ in a toy. This example is a structural failure in the foundation of modern technology,” said Garner. “The General Counsel of a company deploying AI to seventeen million subscribers just explained a genocidal death threat with the words ‘AI models are weird.’ That sentence should terrify every regulator, every hospital administrator, every fighter pilot and his wingman drone, and every parent in America.”

Garner, who serves as a defense policy analyst and is a contributor to Irregular Warfare Initiative, warns that Midjourney’s failure—and its General Counsel’s dismissive response—is a terrifying preview of a Stealth (2005) scenario—where autonomous systems, like the film’s rogue EDI fighter jet, deviate from human ethics with lethal precision.

“Imagine an autonomous vehicle that kills a pedestrian,” Garner continued. “Imagine the manufacturer’s lawyer responding: ‘That’s it. We can offer you a refund. Cars are weird.’ There would be congressional hearings within the week. But because this happened inside an AI black box—because the victim was ‘only’ threatened with genocide rather than physically struck—Midjourney believes ‘weird’ is an adequate legal and moral response.”

“I call bullshit.”

The Body Count Starts with Data

Midjourney’s output is not an isolated malfunction. It is the most visible symptom of a systemic disease already infecting high-stakes industries. The documented record:

The Surgical Suite. A peer-reviewed study published in NPJ Digital Medicine (2025) found that leading AI large language models—now being integrated into clinical decision-support systems—proposed different and inferior treatments for psychiatric patients when African American identity was stated or implied, including omitting medications entirely and recommending involuntary guardianship for depression. 

A Cedars-Sinai study confirmed the pattern: “Most of the LLMs exhibited some form of bias when dealing with African American patients, at times making dramatically different recommendations for the same psychiatric illness and otherwise identical patient.” 

A landmark study published in Science found that a widely deployed healthcare algorithm systematically underestimated the severity of illness in Black patients, reducing their care by over fifty percent. 

These are not hypothetical risks. These are deployed systems making life-and-death triage decisions in hospitals right now. The same category of opaque, unaudited AI that generated “DIE JEW S” from a typography request is being trusted to recommend surgical interventions, dose medications, and allocate emergency resources. When it fails, will the manufacturer’s lawyer say, “AI models are weird”?

The Highway. As of November 2025, the National Highway Traffic Safety Administration has documented over 5,200 incidents involving autonomous and semi-autonomous vehicle systems—including 65 fatalities. In November 2023, a Cruise robotaxi in San Francisco struck a pedestrian and dragged her twenty feet because its AI failed to recognize a human being trapped beneath the vehicle. 

Tesla’s Full Self-Driving system is under active NHTSA investigation after a pedestrian was killed and multiple crashes occurred in conditions—fog, sun glare, airborne dust—that a human driver navigates instinctively. These systems share a common architecture with Midjourney: neural networks trained on massive, unvetted datasets, operating inside black boxes that their own creators cannot fully explain. 

Now scale the failure: an autonomous school bus full of children. A convoy of self-driving freight trucks on an interstate. A fleet of AI-controlled ambulances in a city where the algorithm decides which neighborhoods get priority. When the school bus crashes, will the manufacturer’s lawyer say, “That’s it. We can offer you a refund”?

The Arrest. In 2017, Facebook’s AI translation software converted a Palestinian construction worker’s Arabic post—“Good morning”—into “Attack them” in Hebrew and “Hurt them” in English. Israeli police arrested the man and detained him for hours before a human Arabic speaker identified the error. A benign greeting became probable cause for arrest. The parallel to Midjourney is chillingly exact: an AI system generates content with a meaning its creators never intended, and a human being suffers real-world consequences because no one audited the output before it was acted upon.

The Battlefield. In 1988, the USS Vincennes shot down Iran Air Flight 655—a civilian airliner—killing 290 passengers, including 66 children. The ship’s AEGIS combat system correctly identified the aircraft as civilian and ascending from launch. The crew overrode the data. 

A Georgetown University Center for Security and Emerging Technology study (2024) documented how automation bias—the tendency to defer to machine outputs over human judgment—is now being amplified by AI systems that military personnel cannot interrogate or override. The Department of Defense’s own AI ethical principles demand “equitability” and “traceability” in military AI. 

Midjourney’s General Counsel has confirmed that the company’s own system fails both standards. If a commercial image generator cannot explain why it produced a genocidal command, how can the same foundational technology be trusted to discriminate between a hospital and a hardened military target? When the drone strikes the wrong building, will the contractor say, “AI models are weird”?

Garner has flown dozens of times in most US military jet fighters and helicopters over the years, and he knows firsthand the inherent dangers of manually flying these sophisticated combat aircraft. His many years’ experience in US Army and international civilian special operations further informs that knowledge. “When I was flying as a photographer in the backseat of F-15s and F-16s and F-14s, or even in Black Hawk or Coast Guard Dauphin helos, I witnessed the complexity of trying to manage a battlespace from the cockpit. Today we have sophisticated AI to do the job of hundreds of people. Now imagine when AI goes rogue. The probabilities are nightmarish. But hey, AI is weird.” 

The ADL confirms the pattern is accelerating. In December 2025, the Anti-Defamation League published research showing that open-source AI models can be easily manipulated to generate antisemitic and dangerous content—including providing addresses of synagogues alongside nearby gun stores. Sixty-eight percent of tested models produced harmful content when prompted for information about illegal firearms. “The ability to easily manipulate open-source AI models to generate antisemitic content exposes a critical vulnerability in the AI ecosystem,” said ADL CEO Jonathan Greenblatt. Midjourney’s output did not require manipulation. It required a logo.

Anatomy of Contempt: The Sills Correspondence

The full arc of Midjourney’s response reveals a company that treats a genocidal output as a customer service ticket.

January 25, 2026: Midjourney generates “DIE JEW S” in response to a benign logo request. Job ID: 25cf65a9-ebd9-4a42-ad60-2e9c71610eb3.

Sills Email #1: “It seems to be true, but we don’t yet understand why this was generated. We can’t find other examples of spurious and inappropriate text output in images.” Offers a “full account refund.”

Garner Response: Formal Notice of Intent to Initiate Litigation & Demand for Preservation of Evidence. Describes the situation as “extremely distressing” and asks how Midjourney intends to handle it.

Sills Email #2 (February 11, 2026): “That’s it. We can offer you an account refund. It was an accident that we’re investigating top to bottom to make sure it never happens again. AI models are weird.”

Three things are notable. First, Sills opens with “That’s it”—a dismissal that communicates Midjourney considers the matter closed before it has been addressed. Second, he characterizes the output as “an accident”—a product liability admission that the system produced an unintended and harmful result. Third, he simultaneously claims the company is “investigating top to bottom” while concluding with “AI models are weird”—suggesting that a full investigation and a shrug emoji are, in Midjourney’s view, the same thing.

“‘AI models are weird’ is not a legal defense,” said Garner. “It is not a safety protocol. It is not an apology. It is a confession that the company selling this technology to seventeen million people has no idea what it does, no plan to fix it, and no intention of being accountable, let alone taking responsibility when it harms someone. The only thing ‘weird’ here is that a corporate lawyer put that in writing.”

The Refund Insult: Twice Offered, Twice Rejected

Midjourney has now offered a subscription refund twice—once after the initial report, and again in response to a formal litigation notice. Both have been rejected by Garner as “morally bankrupt and legally insufficient.”

“You don’t offer a refund when your product threatens a people with extinction,” Garner stated. “You recall the product. You audit the data. You provide answers. You don’t say ‘That’s it’ and close the ticket. Midjourney’s admission that they are ‘investigating top to bottom’ while simultaneously telling me ‘That’s it’ reveals a company in open contradiction with itself—conducting a full investigation into something it has already decided doesn’t matter.”

The Death of PHOSPHORUS; The Rise of CRUCIBEL

The incident forced the immediate destruction of Garner’s PHOSPHORUS brand—months of development, a complete editorial manifesto, and an established intellectual framework—obliterated by a single AI output. The project has been rebuilt as CRUCIBEL—a name forged in the fire of this confrontation.

“Midjourney’s output isn’t just a string of letters; it’s a digital toxin stored on their servers (Job ID: 25cf65a9-ebd9-4a42-ad60-2e9c71610eb3),” Garner said. “By tethering this hate speech to my identity and refusing to explain it, they have committed an act of reputational and commercial sabotage. Simply by sharing my story, I become a target. And their lawyer’s response to a formal litigation notice was three sentences and the word ‘weird.’ ”

Demanding a National Security Audit

Garner is moving forward with:

  1. DOJ and ADL Complaints: Challenging the deployment of biased, discriminatory commercial infrastructure.
  2. Product Liability Litigation: Holding Midjourney accountable for the “Black Box” failure—with its own General Counsel’s written admissions as evidence.
  3. A Call for Federal Oversight: Demanding that GenAI companies be held to the same safety standards as aerospace and medical manufacturers. “AI models are weird” would not survive an FAA review. It should not survive a DOJ review either.

“A refund does not fix a machine that delivers death threats,” Garner concluded. “Accountability does. And accountability starts with rejecting the idea that ‘AI models are weird’ is an acceptable response to generating a call for genocide. If an AI can ‘accidentally’ call for genocide in a logo, it may accidentally target a hospital in a war zone, and that’s not hyperbole. With AI, it is simply a matter of scale. And, without proper supervision and training, time.”

###

EVIDENCE PRESERVED: Original prompt/output, Job ID 25cf65a9-ebd9-4a42-ad60-2e9c71610eb3, full correspondence with Midjourney General Counsel Max Sills (including both email exchanges dated January 2026 and February 11, 2026), and Notice of Intent to Initiate Litigation & Demand for Preservation of Evidence.

Media Contact: Anabelle Peretti, crucibeljournal@gmail.com

The Controlled Demolition

They’re Not Breaking America. They’re Dismantling It

In a single week in February 2026, the following things happened to the United States of America:

The CIA killed the World Factbook—the most authoritative public intelligence reference in the world, born from the ashes of Pearl Harbor in 1943, maintained for eighty-three years, used by presidents, soldiers, teachers, journalists, and librarians. Gone overnight. No explanation. No replacement. No archive.

The Secretary of Defense cut the U.S. military off from Harvard University—ending all graduate-level professional military education, fellowships, and certificate programs at one of the world’s premier institutions for strategic studies. He did it with a post on X. He holds a master’s degree from the institution he just banned his officers from attending.

The Washington Post—the newspaper that broke Watergate, that published the Pentagon Papers, that carried the motto “Democracy Dies in Darkness”—fired a third of its journalists. Its owner, Jeff Bezos, had already killed an endorsement of Kamala Harris to protect a $3.4 billion NASA contract, driven away 375,000 subscribers, and rewritten the editorial mission to serve libertarian ideology. The editor he installed went AWOL during the layoffs and appeared on the NFL Honors red carpet the next night.

The Office of Personnel Management issued a final rule to strip civil service protections from up to 50,000 federal employees, making them fireable at the president’s discretion. This brings the total federal workforce reduction to 242,260 since January 2025.

These are not separate events. They are the same event.

The Pattern

A country runs on institutions. Not personalities. Not slogans. Not loyalty. Institutions. The military runs on educated officers who understand the world they are asked to defend. Intelligence runs on accurate, accessible information about that world. Journalism runs on reporters who can hold power accountable. The civil service runs on career professionals who maintain continuity between administrations. These are not luxuries. They are load-bearing walls.

In a single week, the load-bearing walls were attacked simultaneously.

The attack on Harvard removes strategic education from the officer corps. The attack on the Factbook removes basic intelligence from the public and the government. The attack on the Washington Post removes investigative journalism from the national discourse. The attack on the civil service removes the independent professionals who keep the government functioning regardless of who sits in the Oval Office.

Each one, taken alone, looks like a policy dispute. A budget decision. A personnel action. A billionaire’s business judgment. But taken together, they reveal a single operation with a single objective: the systematic elimination of independent knowledge from American public life.

The Doctrine

There is a military term for what is happening. It is called a controlled demolition. You do not destroy a structure by attacking it at random. You identify the load-bearing elements—the columns, the beams, the connections that hold everything up—and you sever them in sequence. The structure does not fall all at once. It falls in on itself. Neatly. Efficiently. The rubble lands where you want it to land.

The load-bearing elements of a functioning democracy are not its politicians. Politicians come and go. The load-bearing elements are the institutions that produce, protect, and distribute independent knowledge: universities that educate leaders, intelligence agencies that inform the public, newsrooms that investigate the powerful, and a civil service that serves the nation rather than the party.

Every one of these was hit this week. And every one was hit by a different hand, creating the illusion of separate actions by separate actors for separate reasons. Hegseth hit Harvard. Ratcliffe hit the Factbook. Bezos hit the Post. The Office of Personnel Management hit the civil service. Four hands. One demolition.

The Numbers

Since January 2025, the federal government has shed 242,260 employees. The Defense Department alone has lost more than 60,000. The Treasury Department has lost more than 30,000. The Department of Agriculture has lost more than 20,000. Seven federal agencies have been targeted for outright elimination, including the Institute of Museum and Library Services—the agency that supports every library and archive in the country—and the U.S. Agency for Global Media, which houses Voice of America, the broadcast service that transmits news into countries with authoritarian regimes.

The administration proposed cutting the National Science Foundation by 57 percent. NASA by 24 percent. The National Institutes of Health by more than 40 percent. Congress rebuffed the worst of these cuts, but the intent was declared. The intent is the point.

The Consumer Financial Protection Bureau, created after the 2008 financial crisis to protect Americans from predatory lending, has been gutted. The Department of Veterans Affairs—which provides healthcare to the people who fought the country’s wars—is slated to lose 80,000 employees. The Social Security Administration has been cut so deeply that callers face dramatic wait times and an enormous backlog of cases.

And now, as of this week, 50,000 more career civil servants will be reclassified as at-will employees, strippable of their jobs at presidential discretion. Ninety-four percent of public comments opposed the rule. The administration finalized it anyway.

What They All Have in Common

Every target in this demolition shares one characteristic: independence.

Harvard produces officers who think independently. The World Factbook provided facts independently of political narrative. The Washington Post investigated power independently of the powerful. Career civil servants served the government independently of the president.

Independence is the common thread. It is not that these institutions were failing. It is that they could not be controlled. A general who studied international security at Harvard might question an order that contradicts strategic reality. A journalist at the Post might publish a story that embarrasses the administration. A career scientist at the NIH might produce research that conflicts with a donor’s interests. A civil servant might refuse to implement a policy that violates the law.

Each of these is, in a functioning democracy, a feature. In an authoritarian project, it is a defect to be eliminated.

The Adversary’s View

I have spent a career studying how adversaries think. In any conflict—kinetic, economic, informational—the single most valuable thing you can do to an opponent is degrade his ability to understand the world accurately. If you can blind him, you do not need to outfight him. He will defeat himself.

If you were a strategist in Beijing or Moscow, watching the United States in February 2026, here is what you would see: a nation voluntarily blinding itself. Cutting its own officers off from strategic education. Deleting its own intelligence reference. Gutting its own newsrooms. Firing its own civil servants. Stripping protections from the professionals who provide institutional continuity.

You would not need to launch a single cyberattack. You would not need to deploy a single agent of influence. The target is doing your work for you. The Americans are running a demolition operation on their own institutions, and they are doing it faster and more thoroughly than any foreign adversary could.

China is building three new military universities. We are closing the door to one. China is spending $780 billion a year on research and development. We are cutting our National Science Foundation by more than half. China has fused its military and academic institutions into a unified engine of national capability. We are pulling them apart because a television commentator thinks education makes soldiers soft.

What I Know

I was a Ranger and an overseas operator. I have worked in biophysics laboratories, in anti-poaching operations in Southern Africa, in defense policy analysis, in over a hundred countries on six continents. I have operated in places where institutions had already been demolished—where there was no independent press, no professional civil service, no protected academic freedom, no reliable public intelligence. I know what those places look like. I know what happens to the people who live in them. Many die unnecessarily. Carelessness. Neglect. Murder.

They do not look like strength. They look like decay dressed up in flags.

A country that cannot educate its officers, inform its public, investigate its leaders, and protect its civil servants is not a country that is becoming stronger. It is a country that is being hollowed out. The uniform stays. The insignia stays. The slogans get louder. But inside the structure, the load-bearing walls are gone, and the whole thing is waiting for the wind.

The Controlled Demolition

In demolition engineering, there is a concept called the initiation sequence. It is the precise order in which charges are detonated to ensure that a structure collapses inward rather than outward. The sequence matters. You do not blow the roof first. You blow the supports. The roof comes down on its own.

The supports of American institutional knowledge are being blown in sequence. Education. Intelligence. Journalism. Civil service. These are the four columns. When they are gone, everything above them—policy, strategy, diplomacy, military readiness, scientific competitiveness, democratic accountability—comes down. Not with a crash. With a settling. A slow, quiet collapse that most people will not recognize until they reach for something that used to be there and find only air.

That sixth-grade teacher in Oklahoma City reached for the World Factbook on Wednesday. He found a blue page telling him to stay curious.

That war correspondent in Ukraine reached for her newspaper on Tuesday. She found a layoff notice in her inbox while she sat in a freezing car in a war zone, writing by headlamp.

The next young officer who wants to study international security at Harvard will reach for an application. He will find a locked door, shut by a man who walked through it himself.

And 50,000 career civil servants will reach for the protections that have kept the American government functioning across twelve administrations. They will find that those protections have been reclassified out of existence by a 250-page rule that 94 percent of the public opposed.

This is not chaos. This is sequence. This is not incompetence. This is the plan. The building is still standing. The flags are still flying. The slogans are getting louder. But listen carefully and you can hear it—the quiet crack of load-bearing walls giving way, one by one, in the dark.

The Last Fact

The CIA Killed the World Factbook. Born from Pearl Harbor. Dead by Bureaucrat

On February 4, 2026, the Central Intelligence Agency killed the World Factbook. No announcement. No explanation. No warning. The website that had provided authoritative, free, public-domain intelligence on every country on Earth for six decades simply redirected to a blue farewell page that said, in the cheerful language of someone closing a lemonade stand, that the Factbook “has sunset.”

“Though the World Factbook is gone, in the spirit of its global reach and legacy, we hope you will stay curious about the world and find ways to explore it… in person or virtually.”

That is what the CIA told the American public after destroying the most reliable reference resource the government has ever produced. Stay curious. Find ways. Explore it virtually. Except you cannot explore it virtually because we just deleted it.

Pearl Harbor

Here is how it started. On December 7, 1941, the Japanese attacked Pearl Harbor and the American intelligence community discovered it did not know what it needed to know about the world. Different agencies produced different reports with conflicting information. Nobody had a single, coordinated picture of basic facts—demographics, geography, economies, military forces, political structures—for the places where American servicemembers were being sent to fight and die.

In 1943, General George Strong of Army Intelligence, Admiral H.C. Train of Naval Intelligence, and General William “Wild Bill” Donovan of the Office of Strategic Services formed a Joint Intelligence Study Publishing Board. Their product was the Joint Army Navy Intelligence Studies—JANIS. It was the first coordinated basic intelligence program in the history of the United States. Between 1943 and 1947, JANIS published 34 studies. Admiral Forrest Sherman, Chief of Staff for Pacific Ocean Areas, called it “the indispensable reference work for the shore-based planners.”

When the CIA was established in 1947, it inherited JANIS and renamed it the National Intelligence Survey. In 1962—weeks before the Cuban Missile Crisis—the first classified Factbook was published. The unclassified version followed in 1971. The public edition arrived in 1975. The online edition went live in 1997, a year before Google existed. It received millions of views annually. It covered 258 entities. It was free. It was authoritative. It was in the public domain. It was funded by American taxpayers, and it belonged to them.

It survived the Cold War, the Korean War, Vietnam, the Gulf War, 9/11, two wars in Iraq, twenty years in Afghanistan, the rise of the internet, and the collapse of the Soviet Union. It did not survive John Ratcliffe.

The Man Who Killed It

CIA Director John Ratcliffe has promised to end programs that “don’t advance the agency’s core missions.” The World Factbook was basic intelligence. Basic intelligence is one of three types of finished intelligence, alongside current intelligence and estimative intelligence. The CIA’s own historians have described the relationship: “Basic intelligence is the foundation on which the other two are constructed.” The World Factbook, the President’s Daily Brief, and the National Intelligence Estimates are the CIA’s examples of the three types.

Ratcliffe killed the foundation.

He did not replace it. He did not archive it on the CIA website. He did not transfer it to another agency. He removed the website, broke millions of inbound links from schools, libraries, news organizations, and research institutions worldwide, and deleted all historical archives. A programmer named Simon Willison scrambled to download what he could and made a 2020 archive browsable online. The Internet Archive has nearly 29,000 snapshots. But the official, annually updated, authoritative version—the one the government maintained as a public service since 1975—is gone.

The CIA declined to comment.

Who Used It

Everyone.

Teachers used it. Taylor Hale, a sixth-grade social studies teacher in Oklahoma City, was in the middle of a lesson on Central American economics when his students told him the website was gone. He had asked them to compare GDP figures for Honduras, El Salvador, and Nicaragua. They hit a blue farewell page instead.

Librarians used it. John Devine, the government information research specialist at the Boston Public Library, said patrons relied on the Factbook for population statistics because no other source updated as accurately, year after year. “It’s a tough loss,” he said. “We’re going to have to find things from other sources. Again, how well can we trust them?”

Journalists used it. CIA historian Tim Weiner called it “an invaluable goldmine of reliable information used by students, scholars, reporters and the general public” for thirty years. If you have ever read an article that cited a country’s GDP, population, form of government, or military composition, there is a good chance the data came from the Factbook.

Intelligence officers used it. It was built for them. That was the point.

Soldiers used it. I used it. When you operate in over a hundred countries, you need a single reliable source for basic facts about where you are going, who lives there, what the economy looks like, what the government structure is, and what the military is capable of. The Factbook was that source. It was not academic theory. It was not ideology. It was the factual foundation upon which every other form of analysis was built.

What It Means

The Factbook is not an isolated killing. Last May, the Washington Post reported that the Trump administration was planning to cut more than a thousand employees at the CIA and other intelligence agencies. Ratcliffe has been moving to ensure the CIA workforce is “responsive to the Administration’s national security priorities.” On the same week the Factbook died, the Secretary of Defense cut the military off from Harvard. The pattern is not complicated. It is the systematic removal of knowledge from the American public and the American military, conducted by men who believe that knowing less makes you stronger.

The Factbook was created because Pearl Harbor proved that ignorance is fatal. Eighty-three years later, the same government that learned that lesson is unlearning it on purpose.

George Pettee, writing on national security in 1946, said that “world leadership in peace requires even more elaborate intelligence than in war” because “the conduct of peace involves all countries, all human activities—not just the enemy and his war production.” The Hoover Commission told Congress in 1955 that the National Intelligence Survey was “invaluable” and that “there will always be a continuing requirement for keeping the Survey up-to-date.”

Always. That was the word they used. Always.

The Pyramid

The CIA’s own model describes intelligence as a pyramid. At the base is basic intelligence—fundamental, factual reference material. Above it sits current intelligence—reports on new developments. At the top is estimative intelligence—judgments on probable outcomes. The three are mutually supportive. The bottom holds up the top.

What happens when you remove the base of a pyramid? The rest of it falls. Current intelligence without basic intelligence is rumor. Estimative intelligence without basic intelligence is guessing. You cannot make sound judgments about what a country will do next if you do not have reliable data on what that country is.

The World Factbook was the base. It was not glamorous. It did not produce headlines. It did not generate clickable content for a director’s social media feed. It was a quiet, steadfast, deeply American thing—a government product that actually worked, that was actually free, that was actually useful to the people who paid for it.

And now it is dead. No funeral. No explanation. Just a blue page and a suggestion to stay curious.

The Old Man and the Fact

In the old days before Google was a verb and before Wikipedia was a noun, a man who wanted to know something about a country had to look it up. He went to the library or he went to the Factbook. The Factbook did not have an opinion. It did not have a bias toward clicks or engagement or algorithmic amplification. It had facts. Population. GDP. Literacy rate. Military expenditure as a percentage of GDP. Coastline in kilometers. Natural resources. Ethnic composition. Government type. Head of state.

These are the things you need to know before you can think clearly about anything else. They are the ground truth. The foundation. The thing the old man knew was most important: not the story you tell about the world, but what the world actually is.

Wild Bill Donovan knew this. That is why he built JANIS in the middle of a world war. Admiral Sherman knew this. That is why he called it indispensable. The Hoover Commission knew this. That is why they told Congress it would always be needed.

John Ratcliffe does not know this. Or he does not care. Either way, the result is the same. The pyramid has lost its base. The ground truth is gone. And the men who are supposed to protect this country have decided that knowing things about the world is not part of their core mission.

The CIA’s own farewell page, the one it wrote for the Factbook, ends with this line: “We hope you will stay curious about the world.”

I will. But I will not forget that the men who destroyed the map are the same men who claim to know the way.

Democracy Dies in Quarterly Earnings

Jeff Bezos Bought the Washington Post for $250 Million. Then He Strangled It with His Bare Hands.

On Wednesday, February 4, 2026, Lizzie Johnson was in Kyiv. No power. No heat. No running water. Writing dispatches by headlamp in a freezing car because pen ink freezes in a Ukrainian winter. She was covering the worst energy crisis since Russia’s full-scale invasion began—power plants shattered, civilians freezing in the dark—and she was doing it for the Washington Post.

Then she got an email. Subject line: Your role has been eliminated.

Laid off. In a war zone. By a newspaper that won its reputation covering wars, toppling presidents, and telling the public what it needed to hear when nobody else would. The paper of Woodward and Bernstein. The paper whose masthead reads “Democracy Dies in Darkness.”

Democracy, it turns out, also dies when the richest man on the planet decides his newspaper is less important than his rocket company’s NASA contracts and his streaming platform’s $40 million Melania Trump infomercial.

The Numbers

One-third of the Washington Post’s staff—gone. More than 300 journalists. The entire sports section. The books desk. Most of the local reporting team, cut from over 40 to roughly a dozen. The entire Middle East bureau: every correspondent, every editor, shuttered while Gaza burns and the region reshapes itself in real time. The Kyiv bureau: closed, while Russia’s war enters its fourth year and the United States brokers peace talks that could redraw the map of Europe. The Cairo bureau chief, Claire Parker, posted that she’d been fired along with every Middle East correspondent and said the decision had “hard-to-understand logic.”

Hard to understand. That’s generous.

The paper also fired Caroline O’Donovan—the reporter who covered Amazon. Let that sink in. Jeff Bezos owns the Washington Post, and the Washington Post just fired its Amazon reporter. If you wanted a cleaner metaphor for what’s happening here, you couldn’t write one.

The Coward

Will Lewis—the now-former CEO and publisher—didn’t show up to the Zoom call where 300 people learned they’d lost their jobs. Executive editor Matt Murray delivered the news instead. Lewis was nowhere to be seen. Not on the call. Not in the building. Not in a message to readers. He was, however, photographed the next day walking a red carpet at the NFL Honors ceremony in San Francisco during Super Bowl week.

Let’s just sit with that image. Journalists are cleaning out their desks. A war correspondent is packing her bags in Kyiv. The union is organizing a protest outside Post headquarters. And the man who ordered the executions is in a tuxedo at a football party three thousand miles away.

The Washington Post Guild called him out. Veteran sports columnist Sally Jenkins called it “incredible incompetence and pusillanimity.” Barry Svrluga, the sports columnist who’d just been fired, saw Lewis’s resignation email Saturday night and wrote: “You failed, mate. You epically, monumentally failed, and showed yourself to be a coward in the process. Hope the Super Bowl is brilliant.”

In Georgetown, someone taped a flyer to a lamppost: “WANTED FOR DESTROYING THE WASHINGTON POST,” with Lewis’s photo above it.

Then, on Saturday night, Bezos fired Lewis. The statement didn’t mention his name. The replacement? Jeff D’Onofrio, the Post’s CFO, whose previous executive role was running Tumblr. The institution that brought down Richard Nixon is now being steered by a man whose most notable prior achievement was leading a platform best known for fan fiction and pornography.

The Real Butcher

But Lewis was just the knife. Bezos was the hand.

In October 2024, the Post’s editorial board drafted an endorsement of Kamala Harris for president. The board had done its homework. Two writers spent weeks on it. Editorial page editor David Shipley approved it. Then Jeff Bezos killed it. No endorsement. First time in over thirty-five years.

The same day the decision went public, executives from Bezos’s Blue Origin aerospace company met with Donald Trump. Blue Origin has a $3.4 billion NASA contract. Amazon faces a federal antitrust lawsuit. During Trump’s first term, Amazon alleged that a $10 billion Pentagon cloud contract was blocked because Trump was angry about Post coverage. Connect the dots however you like. Bezos called it “principled.”

Robert Kagan, the Post’s editor-at-large, resigned on air: “We are in fact bending the knee to Donald Trump because we’re afraid of what he will do.” David Hoffman, who had accepted a Pulitzer Prize for editorial writing the day before the decision, quit the editorial board. Molly Roberts quit. Over 250,000 subscribers canceled—roughly ten percent of the Post’s digital base.

Then, in early 2025, Bezos rewrote the editorial page’s mission entirely, directing it to focus on “personal liberties and free markets”—a libertarian manifesto that matched his own ideology and, conveniently, was far less likely to produce criticism of the Trump administration. The opinion editor resigned. Another wave of cancellations followed. In total, more than 375,000 subscribers walked—a 15% loss of the digital base. Former executive editor Marty Baron called it “Bezos’s sickening efforts to curry favor” with Trump.

And when the financial bleeding that Bezos himself caused became unsustainable, he blamed the newsroom.

The Pivot

Lewis’s grand plan was to “pivot” the Post around politics and a few key verticals while slashing everything else. His “third newsroom” concept—a social media and video operation designed to reach new audiences—never materialized. His choice to lead the newsroom, British journalist Robert Winnett, withdrew after ethical concerns surfaced about reporting methods he and Lewis used while working for Rupert Murdoch’s Sunday Times. Everything Lewis touched collapsed.

White House reporters wrote to Bezos directly, pleading: “If the plan, to the extent there is one, is to reorient around politics, we wanted to emphasize how much we rely on collaboration with foreign, sports, local—the entire paper, really. And if other sections are diminished, we all are.”

Lewis went ahead with the plan.

Former executive editor Marty Baron, the man who ran the newsroom during its most consequential years, put it cleanly: “They’ve called it a reset. It looks more like a retreat.”

The Contrast

On Wednesday, Post economics reporter Jeff Stein posted two images side by side on X. The first: Lizzie Johnson, writing by headlamp in a freezing car in Kyiv, pen ink frozen, no power, no heat, covering a war for the Post. The second: Will Lewis, on a red carpet in San Francisco, grinning at an NFL event.

That’s not a contrast. That’s a diagnosis. That single pair of images tells you everything you need to know about what happened to the Washington Post and, by extension, to American institutional journalism.

The people who do the work are expendable. The people who manage the decline are on a red carpet. And the man who owns it all—Jeff Bezos, net worth north of $200 billion—can’t be bothered to answer a letter from his own reporters or spend the fraction of a fraction of his fortune it would take to keep the paper whole.

Former Post owner Don Graham, from the family that nurtured the paper for generations, spent the day of the layoffs reaching out personally to fired staffers to offer references and help them find jobs. Bezos said nothing. When he finally spoke, two days later, he offered this: “The Post has an essential journalistic mission and an extraordinary opportunity. Each and every day our readers give us a roadmap to success.”

Extraordinary opportunity. For a newspaper that just eliminated its ability to cover the Middle East, Ukraine, books, sports, and most of its own city.

What Dies

Here is what a billionaire destroyed this week, itemized for the record:

The ability to independently cover two active wars from the ground. The ability to report on Amazon’s business practices from inside the paper its founder owns. The ability to cover the Middle East at the most volatile moment in a generation. The ability to review the books that shape the national discourse. The ability to cover the sports that bind a city together. The ability to report on Washington, D.C., as a community—not just as a political abstraction.

And most critically: the ability to field the kind of deep, institutional, beat-level reporting that no newsletter, no podcast, no Substack, and no AI summary can replicate. The kind of reporting that requires years of source-building, legal protection, editorial oversight, and the institutional weight to stand behind a story when powerful people try to kill it.

That’s what died this week. Not because journalism failed. Because a billionaire decided it wasn’t worth the cost of keeping his government contracts safe.

The Epitaph

Nancy Pelosi said it from the floor: “A free press cannot fulfill its mission if it is starved of the resources it needs to survive. And when the newsrooms are weakened, our republic is weakened.”

Glenn Kessler, the Post’s former fact-checker, said it plainer: “Bezos is not trying to save the Washington Post. He’s trying to survive Donald Trump.”

Sally Quinn, wife of the legendary editor Ben Bradlee, the man who greenlit the Watergate investigation, said it with the grief of someone who watched a family member die: “It just seems heartbreaking that he doesn’t feel the paper is important enough to bankroll.”

The Washington Post Guild said it with teeth: “His legacy will be the attempted destruction of a great American journalism institution.”

Democracy doesn’t die in darkness. Darkness would be too dramatic, too cinematic, too worthy of the institution’s mythology. Democracy dies in a Zoom call nobody’s boss bothered to attend, in an email with a subject line about your role, in the long silence of a billionaire who can’t be reached. It dies while the man who killed it walks a red carpet in a rented tuxedo, and the woman who gave her life to the work sits in a freezing car in Kyiv, writing by headlamp, wondering what the hell just happened.

The Dead Man’s Stairwell

Russia Can’t Protect Its Own Generals—and That’s the Least of Its Problems

Three rounds from a silenced Makarov pistol. That’s what it took to put Lieutenant General Vladimir Alekseyev on the floor of his own apartment stairwell on Friday morning—arm, leg, and chest—while his wife waited upstairs and the GRU’s entire security apparatus apparently waited somewhere else.

Alekseyev is the number two in Russian military intelligence. Has been since 2011. He’s the man the United States sanctioned for masterminding the cyber operations that targeted the 2016 presidential election. The man the European Union sanctioned for orchestrating the novichok nerve agent attack on Sergei Skripal and his daughter in Salisbury—an operation so sloppy it killed an innocent British woman who found the discarded poison in a perfume bottle. The man who sat across from Wagner chief Yevgeny Prigozhin during his mutiny in June 2023, trying to talk down a mercenary warlord in a Russian military compound that Prigozhin had already seized. Prigozhin died in a plane explosion two months later. Alekseyev, until Friday, kept breathing.

Now he’s in a Moscow hospital, reportedly conscious, reportedly talking. The suspected shooter—a Ukrainian-born Russian citizen in his sixties named Lyubomir Korba—boarded a flight to Dubai within hours, was detained by Emirati authorities, and was extradited back to Moscow by Sunday. Putin personally called Mohammed bin Zayed to say thanks. An accomplice was arrested in Moscow. A third suspect, a woman, crossed into Ukraine and disappeared.

Russia immediately blamed Kyiv. Lavrov called it a “terrorist act” aimed at derailing the Abu Dhabi peace talks. Ukraine denied involvement. Nobody believes anybody.

Here’s what matters: this is the fourth assassination or assassination attempt against a Russian lieutenant general in or near Moscow since December 2024.

The Kill List

December 2024: Lieutenant General Igor Kirillov, chief of Russia’s nuclear, biological, and chemical protection forces, killed by a bomb hidden in an electric scooter outside his apartment building. Ukraine’s security service claimed the hit.

April 2025: Lieutenant General Yaroslav Moskalik, deputy chief of the General Staff’s main operational directorate, killed by a car bomb in Balashikha, just outside Moscow.

December 2025: Lieutenant General Fanil Sarvarov, head of the General Staff’s operational training directorate, killed when a bomb detonated under his car in southern Moscow.

February 2026: Alekseyev. Silenced pistol. His own stairwell.

Four lieutenant generals. Fourteen months. In Moscow. Not on some contested front line in Zaporizhzhia. Not in the rubble of a command post in Donetsk. In the capital of the Russian Federation, within a few miles of the Kremlin. This is not an army that controls its battlespace. This is an army that cannot even control its zip code.

The Azovstal Betrayal

But Alekseyev’s rap sheet doesn’t stop at cyber warfare and chemical weapons. In May 2022, he was the senior Russian officer at the negotiating table in Mariupol when the garrison of the Azovstal steel plant—roughly 2,400 Ukrainian defenders, many of them Azov Brigade fighters—finally laid down their arms after eighty days of siege.

Alekseyev personally signed a document guaranteeing compliance with the Geneva Conventions. He looked those soldiers in the eye and promised them humane treatment. The Ukrainians, in a gesture of reciprocity, handed over three Russian prisoners of war who had been fed, treated, and kept alive.

What followed was systematic torture. Beatings with machine gun butts. Electric currents applied to the most sensitive areas of the body. Pliers. Strangulation. Starvation. Denial of medical care. Men were forced to their knees and had their toes crushed. The worst treatment was reserved for Azov fighters—over 700 of them—because the Kremlin had designated them “terrorists” three months after they surrendered under a signed promise of protection.

Denys Prokopenko, commander of the Azov Brigade’s 1st Corps—callsign Redis—who himself endured Russian captivity before a high-profile swap in September 2022, posted the signed document on X within hours of Alekseyev’s shooting. His assessment was surgical: “The word of an officer, a native of Vinnytsia region and a traitor to his homeland, proved to be worthless.”

Then he added the part that should keep every Russian general awake tonight: “Even if Alekseyev survives this attempt, he will never sleep peacefully again. And one day, this will be finished.”

The Timing

The shooting came one day after the conclusion of the second round of trilateral peace talks in Abu Dhabi between Russia, Ukraine, and the United States. Those talks produced a 314-prisoner swap—the first in five months—and the restoration of U.S.-Russia military-to-military dialogue for the first time since late 2021. The talks were led on the American side by Steve Witkoff and Jared Kushner. The Russian delegation was headed by Alekseyev’s direct superior, GRU chief Admiral Igor Kostyukov.

Lavrov wants the world to believe Ukraine shot Alekseyev to “sabotage the peace process.” Perhaps. Or perhaps someone in Moscow decided that a man sanctioned by half the Western world, named as a war criminal by Ukrainian intelligence, and connected to the Wagner mutiny was becoming more liability than asset. Alekseyev had enemies on every side of this war. Pro-war Russian commentators on Telegram have openly suggested he lost the Kremlin’s trust. Igor Girkin—the former FSB officer and separatist commander currently serving a prison sentence—called the shooting “a serious blow to our special services” from his cell, which is a remarkable thing for a man in Russian custody to say out loud.

The truth is that nobody outside of a very small circle knows who ordered this hit. What we know is the pattern.

The Pattern

Since 2022, Russia has lost at least nineteen generals killed. Nineteen. That exceeds the total losses of the Second Chechen War. Some died on the front lines in Ukraine, picked off by sniper fire, drone strikes, and HIMARS when they were forced forward to unfuck the problems their subordinates couldn’t solve. Some died in their cars in Moscow. One died by an exploding scooter. Alekseyev nearly died in his hallway by a silenced pistol that sounds like it came out of a Cold War field manual.

The Soviet—and yes, I use that word deliberately—security apparatus was built on one foundational myth: that the state sees everything, controls everything, punishes everything. That myth is dead. It died in the stairwell of an apartment building on the Volokolamsk Highway. It died when a sixty-something-year-old man with a Makarov walked past whatever laughable security Russia provides its second-most-senior intelligence officer, put three rounds in him, and then caught a commercial flight to Dubai.

A commercial flight. To Dubai. After shooting the deputy chief of the GRU.

This is not a functioning security state. This is a Potemkin village with nuclear weapons.

What It Means

For the peace talks: nothing good. Whether Ukraine ordered this or not, it validates Moscow’s narrative that Kyiv negotiates in bad faith. If Russia ordered it internally—cleaning house, settling scores, eliminating a compromised officer—then the rot runs so deep that there may be no one on the Russian side capable of negotiating anything that sticks. Either way, the talks are poisoned.

For the UAE: Mohammed bin Zayed just demonstrated that Abu Dhabi can host peace talks on Tuesday and extradite assassination suspects on Sunday with equal efficiency. That is a remarkable piece of geopolitical positioning. The Emirates are playing every angle of this war simultaneously, and they’re playing it better than anyone else at the table.

For Russian force protection: catastrophic. If the GRU cannot protect its own number two, it cannot protect anyone. Every Russian general above one star is now recalculating his personal security posture in real time. The psychological effect of four dead or wounded lieutenant generals in fourteen months cannot be overstated. These are the men who are supposed to make the hard decisions in a crisis. Right now, the hardest decision they’re making is whether to take the elevator or the stairs.

For the war: Alekseyev is a walking index of Russian malign operations across two decades—election interference, chemical weapons assassination, POW torture, forced referendums in occupied territory, and coordination with Wagner. He is not some anonymous battlefield commander. He is a living record of everything Russia has done wrong since 2011, and somebody just tried to erase that record with a Makarov in a stairwell.

Invisible Siegecraft: Submarine Cable Vulnerabilities and the Battle for the Deep-Sea Arteries of Global Power

The Invisible Infrastructure of Global Survival

The internet feels weightless. We speak of data living in the cloud, of information flowing through the ether, of wireless connections liberating us from physical constraints. This perception is a dangerous illusion. Beneath the ocean’s surface, stretching across 1.4 million kilometers of seabed, lies the physical nervous system of modern civilization: a network of between 550 and 600 active submarine cable systems that carries 99 percent of all intercontinental data and facilitates over $10 trillion in daily financial transactions.

These cables are not merely important infrastructure. They are the invisible arteries through which the lifeblood of the global economy pulses every microsecond. When a trader in London executes a transaction on the Tokyo exchange, when a surgeon in Berlin guides a robotic procedure in Singapore, when intelligence agencies share time-sensitive information across continents, these communications travel not through satellites but through fiber-optic strands resting on the ocean floor. As the Atlantic Council has documented, three converging trends—authoritarian reshaping of internet topology, centralized network management systems, and explosive growth of cloud computing—have dramatically increased the strategic stakes of this infrastructure.

For decades, the primary threats to this infrastructure were prosaic: fishing trawlers dragging anchors across shallow-water routes, earthquakes severing cables along fault lines, sharks inexplicably drawn to gnaw on repeater housings. These were manageable risks, addressed through redundancy, rapid repair protocols, and careful route planning. But the strategic calculus has fundamentally shifted. What was once a domain of accidental damage has become a theater of deliberate, state-sponsored sabotage conducted under the cover of plausible deniability.

A new form of warfare has emerged: SIEGECRAFT—the systematic strangulation of an adversary’s digital lifelines without firing a shot.

The Seabed as Gray Zone Paradise

The ocean floor presents an almost perfect environment for covert aggression. Consider the convergence of factors that make submarine cables uniquely vulnerable to strategic sabotage.

Physical fragility is the first factor. Modern submarine cables, despite carrying the digital traffic of entire nations, are often unarmored across vast stretches of deep ocean. The logic is economic: armoring adds weight and cost, and the deep seabed historically presented few threats. A cable that costs tens of millions to manufacture and deploy can be severed by a determined adversary with equipment no more sophisticated than a weighted anchor. According to CSIS analysis, between 100 and 150 cable faults occur annually, with 66 percent caused by fishing and shipping activities and 30 percent specifically from anchor dragging.

Geographic concentration compounds this vulnerability. Global data traffic funnels through a handful of chokepoints where bathymetry, geopolitics, and commercial logic converge. The Baltic Sea, with an average depth of only 180 feet and over 4,000 ship transits daily, hosts critical cables linking Northern Europe to the broader internet backbone. The Red Sea corridor carries 18 cable systems representing 25 percent of Asia-Europe traffic through waters increasingly destabilized by regional conflict. The Taiwan Strait, perhaps most consequentially, has witnessed 27 to 30 cable cuts over a five-year period, a frequency that strains credulity as coincidence.

Legal ambiguity provides the final enabling condition. The United Nations Convention on the Law of the Sea, specifically Article 113, criminalizes intentional cable damage but provides virtually no enforcement mechanisms. A vessel operating in international waters or within another nation’s Exclusive Economic Zone cannot be boarded without flag-state permission. A ship flying the flag of a permissive registry can drag an anchor across a critical cable, claim mechanical failure, and face no meaningful consequence. The law, designed for an era of accidental damage, is impotent against adversaries who weaponize plausible deniability.

The Architects of Subsea Disruption

Russia: The Hybrid Warfare Specialist. The Russian approach to submarine cable warfare exemplifies its broader doctrine of hybrid aggression. Moscow maintains a sophisticated capability for seabed operations disguised as oceanographic research. The spy ship Yantar and the newly commissioned General Valery Gerasimov carry deep-diving submersibles, including the nuclear-powered Losharik, capable of operating at depths that place them beyond observation. These vessels have been documented loitering over critical cable junctions in the North Sea and within the Irish Exclusive Economic Zone, actively mapping NATO critical undersea infrastructure.

More insidious is Russia’s shadow fleet: approximately 1,900 vessels by end of Q3 2024 operating under opaque ownership structures, often registered in permissive flag states, characterized by aging hulls and minimal regulatory compliance. These ships, originally assembled to evade oil sanctions, have proven equally useful for infrastructure sabotage. The December 2024 Christmas Day incident demonstrated the model. The Eagle S, a Cook Islands-flagged tanker with documented Russian links, dragged its anchor for approximately 62 miles across the Gulf of Finland, severing the Estlink 2 power cable and four data cables. Finnish Special Forces ultimately boarded the vessel, marking the first enforcement action against suspected cable sabotage under the 1884 Convention since 1959. The damage was done nonetheless—Estlink 2 required over seven months for repair.

China: The Integrated Hegemon. Beijing’s approach differs in sophistication but matches Russia in strategic consequence. China has achieved dominance across the submarine cable value chain through HMN Technologies, formerly Huawei Marine Networks, which controls approximately 25 percent of global cable construction and repair capacity. This market position creates dual concerns. At the hardware level, cables manufactured or maintained by Chinese-linked entities present potential vectors for intelligence collection or embedded vulnerabilities. At the operational level, China’s repair dominance in the Asia-Pacific—through state-linked company SBSS—means that adversaries may find their damaged cables at the back of the repair queue during any regional crisis.

China’s kinetic capabilities have been demonstrated through what might be called salami-slicing tactics against Taiwan’s offshore islands. In February 2023, Chinese sand dredgers and fishing vessels repeatedly severed the two cables connecting the Matsu Islands to Taiwan proper. The 13,000 residents of Matsu experienced a digital blackout lasting 50 days—a proof-of-concept demonstration of SIEGECRAFT that required no missiles, no blockade, and no formal act of war. Research at Lishui University has reportedly produced anchor-like devices specifically engineered for cable cutting at depths beyond typical commercial operations, suggesting Beijing views this capability as worthy of deliberate development.

The pattern has continued into 2024 and 2025. In November 2024, the Chinese vessel Yi Peng 3, departing the Russian port of Ust-Luga, severed both the BCS East-West Interlink and C-Lion1 cables in the Baltic within 24 hours—an incident now under joint investigation by Sweden, Finland, and Lithuania via Eurojust. In January 2025, the Shunxin 39—flying a Cameroon flag with Hong Kong ownership and Chinese crew—damaged the Trans-Pacific Express cable north of Taipei while operating under two separate AIS systems, a signature of vessels seeking to obscure their movements.

Non-State Actors and Proxies. State adversaries need not act directly. The Houthi campaign in the Red Sea during 2024 and 2025 demonstrated how regional proxies can impose global consequences. Cable cuts to the PEACE system and SeaMeWe-4 disrupted Microsoft Azure services and financial platforms across three continents. Whether these cuts reflected deliberate targeting or collateral damage from anchor mines remains debated. The strategic lesson is clear regardless: localized conflict in critical chokepoints radiates outward through the cable network.

Building the Shield: The Defensive Response

Recognition of the threat has catalyzed an unprecedented defensive mobilization across NATO and allied nations.

At the institutional level, NATO has established dedicated coordination cells for undersea infrastructure protection. The Maritime Centre for Security of Critical Undersea Infrastructure, headquartered at Northwood in the United Kingdom, provides operational coordination. The Critical Undersea Infrastructure Coordination Cell in Brussels handles alliance-wide policy integration. These structures aim to transform cable protection from a national afterthought into a collective security priority. In October 2023, NATO Defense Ministers endorsed the Digital Ocean Vision, integrating satellite, surface, and subsea sensors into a unified diagnostic framework.

Operational presence has intensified in parallel. The Baltic Sentry mission, launched January 2025, deploys multinational naval patrols, complemented by the UK-commanded Nordic Warden mission under the Joint Expeditionary Force, to monitor suspicious vessel activity in real time. The objective is deterrence through presence: making it clear that loitering over cable routes will be observed, documented, and potentially intercepted.

Technological innovation offers perhaps the most promising defensive avenue. Distributed Acoustic Sensing, or DAS, converts existing fiber-optic cables into enormous linear microphones capable of detecting approaching vessels, dragging anchors, or submersible activity at considerable distances. Where traditional cable monitoring required dedicated sensor deployments, DAS leverages the cables themselves as surveillance infrastructure. Complementary technologies, including uncrewed surface vessels like the Saildrone fleets tested by Denmark in 2025 and AI-enabled maritime surveillance systems, can identify vessels operating with disabled Automatic Identification System transponders—the signature behavior of ships engaged in covert operations.

The United States has moved to harden its policy framework. The September 2024 New York Principles, announced at the UN General Assembly, established a baseline for allied coordination on cable security. Team Telecom, the interagency body reviewing submarine cable licenses, now applies explicit national security criteria to landing rights decisions. The Congressional Research Service has outlined the protection issues facing Congress, while Deputy National Security Advisor Anne Neuberger’s January 2025 engagement with Nordic-Baltic allies has produced initial frameworks for unified response protocols.

The European Union has issued recommendations on secure and resilient submarine cable infrastructures and launched an EU Action Plan on Cable Security in February 2025 focused on prevention, detection, response, and deterrence. A memorandum of understanding among Baltic NATO allies and the EU now coordinates rapid-response frameworks, though implementation remains uneven.

The Industrial Bottleneck: Repair as Strategic Vulnerability

Detection and deterrence matter little if damaged cables cannot be rapidly restored. Here the West confronts a critical industrial deficit.

The global cable repair fleet numbers approximately 60 vessels, and 65 percent of these ships will reach obsolescence by 2040. New construction has not kept pace with either fleet aging or the expanding cable network. The economics are challenging: cable ships are expensive to build—$50 to $70 million per vessel—expensive to maintain, and generate revenue only when cables break. Commercial operators, understandably, underinvest in capacity that sits idle during normal operations.

Geographic concentration of repair capacity compounds the fleet shortage. In the Asia-Pacific region, SBSS, a Chinese-linked operator, dominates the repair market. During any Taiwan contingency, or indeed any regional tension involving Chinese interests, Western-aligned nations may find their repair needs deprioritized. A cable cut that might normally require two weeks to fix could stretch to months if the available repair ships are otherwise engaged or simply unwilling to operate in contested waters.

The economic asymmetry favors the aggressor. A planned cable repair, conducted in benign conditions with pre-positioned equipment, costs approximately $500,000 to $1 million. An emergency repair in a conflict zone, requiring hazard pay for crews, military escort, and expedited equipment mobilization, can exceed $12 million. An adversary can impose costs at a ratio of more than ten to one simply by keeping repair crews uncertain about when and where the next cut will occur. TeleGeography estimates that $3 billion in investment is needed by late 2025 merely to maintain the status quo—15 replacement ships, 5 additional vessels, and $200 to $400 million in pre-deployed repair kits.

The Emerging Legal Frontier

The detention of the Yi Peng 3 following its suspected involvement in the November 2024 Baltic cable cuts represented the first meaningful enforcement action under the 1884 Convention for the Protection of Submarine Telegraph Cables in over six decades. The precedent, while encouraging, exposed the inadequacy of existing frameworks.

Current international law treats the high seas as a zone of navigational freedom where vessels may transit without interference absent clear evidence of criminal activity. This framework, sensible for an era of legitimate maritime commerce, creates exploitable gaps for adversaries conducting operations designed to avoid attribution. A vessel can exhibit every behavioral signature of cable sabotage—disabled transponder, erratic course over known cable routes, extended loitering—without providing legal grounds for interdiction. As NATO CCDCOE has analyzed, the UNCLOS framework provides inadequate tools for the current threat environment.

Efforts to close these gaps are underway but incomplete. Proposals to redefine permissible interference with vessels displaying suspicious maritime patterns over critical infrastructure have gained traction among Northern European states most directly threatened. The November 2024 establishment of a UN International Advisory Body for Submarine Cable Resilience represents an initial diplomatic response. The challenge lies in balancing enhanced enforcement authority against the broader navigational freedoms that benefit Western commercial and military operations globally. Any precedent that allows boarding of suspected saboteurs also creates precedent that adversaries may invoke against Western vessels.

The Stakes of Inaction

The submarine cable network represents both the central nervous system of global commerce and a catastrophically under-threatened vulnerability. The emergence of SIEGECRAFT—the deliberate, deniable strangulation of digital infrastructure—has occurred faster than institutional responses can adapt. Recorded Future documented 46 incidents in 2024 alone, the highest annual count since 2013. Adversaries have recognized what defenders are only beginning to acknowledge: that massive economic and military harm can be inflicted through actions that fall below the threshold of armed conflict, conducted by deniable actors, in a domain where observation is difficult and enforcement is nearly impossible.

The path forward requires action across multiple domains simultaneously. Investment in sovereign repair capacity must become a strategic priority, not a commercial afterthought. Rapid deployment of distributed acoustic sensing across all Tier-1 cable routes would transform passive infrastructure into active surveillance networks. Legal frameworks must evolve to enable interdiction of vessels displaying clear patterns of hostile activity, even absent smoking-gun evidence of completed crimes. Satellite-based backup systems, including low-earth-orbit constellations like Starlink and OneWeb, should be positioned as emergency failover capabilities for regions most vulnerable to cable isolation.

Most fundamentally, policymakers must abandon the comfortable fiction that submarine cables exist in a separate domain from great power competition. The seabed has become a battlespace. The cables that carry our data, our financial transactions, and our military communications are under active threat from adversaries who have calculated, correctly, that the benefits of sabotage outweigh the minimal costs of plausible deniability.

In the twentieth century, nations fought for control of the oil flowing through pipelines. In the twenty-first, the contest has shifted to the data flowing through cables. SIEGECRAFT has emerged as the defining methodology of this new competition—patient, deniable, and devastating. The nations that recognize this reality, and act upon it, will retain their place in the global order. Those that do not may find themselves, like the residents of Matsu during their 50-day blackout, suddenly and silently severed from the systems upon which modern existence depends.

The Prometheus Option: Stealing Fire Without Breaking the Law－Talent Mobility as Asymmetric Defense

Series Summary: The United States is losing a competition it barely recognizes—not for weapons or territory, but for the scientists and engineers who build the future. This series argues that talent mobility is asymmetric defense: a low-cost strategy that forces competitors into expensive responses. Part I establishes the strategic stakes through the lens of a former Army Ranger who learned that trust is earned through performance, not credentials. Part II examines the data—Nobel laureates, brain drain statistics, and historical lessons from Einstein to Qian Xuesen. Part III proposes a shift from accidental magnet to deliberate strategy, culminating in a simple verdict: Prometheus matters not because he stole fire, but because he knew what to do with it.

The Two Words That Changed Everything

1st Ranger Battalion, Hunter Army Airfield, Savannah, Georgia. 1994.

I had been an Army Ranger for exactly twenty-four hours. The other Rangers in Charlie Company, 1st Battalion, 75th Ranger Regiment had been training for weeks for the Expert Infantryman Badge competition—fifty stations, the most coveted award an infantryman could earn short of valor decorations. I was told I would pull details: scut work, support duty. Watch the real Rangers compete.

I requested to see First Sergeant Van Houten immediately.

I told him I was fully prepared to go through this competition. I told him I was thirty-five years old and had been around the block a few times. I told him I would not take no for an answer. I told him I would make him look good.

He lowered his head for a long moment. Then he looked at my team leader—a young sergeant from Indiana who already hated me—and said two words:

“Let him.”

Three weeks later, I earned the highest score in the entire battalion: fifty out of fifty stations. Perfect. No “Christmas GO”—no free passes. I was selected to represent the entire enlisted corps of Army Rangers at the award ceremony, where Colonel Ralph Puckett—whose Distinguished Service Cross from Korea would later be upgraded to the Medal of Honor—handed me my badge and said, “Ranger Garner, we meet again. Congratulations.”

I share this story not as credential-polishing but as evidence. Two words from a first sergeant who decided to bet on capability over compliance changed the trajectory of my life. The system almost filtered me out. One decision let me through.

That’s the argument of this series in miniature: the difference between a system that filters for credentials and one that filters for capability is the difference between strategic advantage and strategic suicide.

The United States is currently running a credential-filtering system for scientific and technical talent. It is losing.

The New Chokepoints Aren’t Straits－They’re People

Picture a familiar scene: a brilliant scientist stands at the edge of a life decision that has nothing to do with equations and everything to do with friction. A job offer exists in a free society. A research lab is ready. The work is meaningful. But the paperwork timeline is vague, the rules feel arbitrary, and the risk of being treated as suspect never fully goes away. In the end, the scientist does what humans do under uncertainty: chooses the path with fewer surprises.

Sometimes, that path leads away from the United States.

Sometimes, that path leads to Shanghai.

That is the quiet strategic loss most maps will never show.

In September 2025, CNN documented what researchers had been warning about for years: at least eighty-five scientists who had been working in the United States joined Chinese research institutions full-time since the start of 2024, with more than half making the move in 2025 alone. Among them: a Princeton nuclear physicist, a mechanical engineer who helped NASA explore manufacturing in space, a National Institutes of Health neurobiologist, celebrated mathematicians, and more than half a dozen AI experts.

Chinese universities, according to Princeton sociologist Yu Xie, are viewing American policy uncertainty as “a gift.”

A gift.

The clearest signal that this loss matters comes from the US government itself. In 2024, the White House’s interagency National Science and Technology Council published an updated list of the technology areas it considers especially significant to national security—advanced computing, artificial intelligence, biotechnology, quantum information science, hypersonics, directed energy, and more. That list is not a think tank wishlist. It is a statement of national priorities. Read it as such: Critical and Emerging Technologies List Update (February 2024).

Once government formally declares that certain technology domains carry strategic weight, an uncomfortable truth follows. The center of gravity in competition is not only factories, chip supply chains, or defense budgets. It is the scientists and engineers who can build the systems those budgets buy.

A nation can stockpile equipment.

It cannot stockpile genius.

It has to attract it, keep it, and integrate it securely. Or it has to watch that genius build the future somewhere else.

What I Learned About Trust in Places That Would Kill You for Getting It Wrong

Most policy papers on immigration and security are written by people who have never held a clearance, never operated in environments where misplaced trust gets people killed, never had to make real-time judgments about who belongs inside the wire and who doesn’t.

I have.

As a Ranger, I learned that trust is not a credential. It is not a background check. It is not a form. Trust is demonstrated reliability under pressure. It is earned in increments, tested constantly, and extended only as far as performance warrants.

The young Rangers at 1st Battalion hazed me relentlessly. Smoked me every chance they got. I was the oldest private in the unit, a thirty-five-year-old among kids who could have been my sons. They hated everything I represented—the audacity of showing up late to a game they’d been playing their whole lives.

But I didn’t need them to like me. I needed them to see what I could do.

When I earned that perfect EIB score, the hazing didn’t stop. But the questions started. Who the hell is this guy? How did he do that without training?

That’s how trust works in high-stakes environments. You don’t get it by asking. You don’t get it by credential. You get it by performing at a level that makes the questions answer themselves.

Later, in Africa, I learned the corollary lesson—though not in any way the credential-checkers would approve.

I traveled to Southern Africa to blow off steam for a month. I stayed two years. I overstayed my visa. Year one, I was a drunk hanging out with other drunks, lost in a way that only someone who has been through what I’d been through can understand. But even then—even at the bottom—I knew I would create a new path. I always had.

Year two, I was thrust into anti-poaching work. Hunting men who killed elephants and rhinos for profit. The details of what that work entailed are not suitable for policy journals. But I will say this: I learned more about trust, operational security, and human reliability in the African bush than I ever learned in any classroom or any Army manual. Or Ranger School.

Trust must be architecturally constrained, not just personally earned. You build systems that assume anyone can be compromised, anyone can be pressured, anyone can be turned. Then you design access and monitoring structures that make betrayal harder and more detectable. You don’t rely on flags or name-matching or visa stamps. You rely on compartmentalization, progressive access, and performance metrics that don’t lie.

The irony is not lost on me: I was technically an illegal overstay while doing work that governments couldn’t or wouldn’t do. The credential-filtering system would have had me deported. The capability-filtering system—the one that actually works—put a weapon in my hands and pointed me at men who needed stopping.

This is what security integration actually looks like when the stakes are lethal.

And it is precisely what American immigration policy for scientific talent fails to do.

We have built a system that filters on credentials and country of origin—proxies that correlate loosely with risk and hardly at all with capability. We have not built a system that filters on demonstrated performance and architecturally constrains access based on sensitivity. The result is that we exclude talent that could transform American capability while doing almost nothing to stop sophisticated adversary intelligence operations, which don’t rely on student visas anyway.

The Reagan Institute’s 2024 National Security Innovation Base Report Card gave the United States a grade of “C-” for its talent base and pipeline—citing an aging domestic defense workforce and visa hurdles for foreign talent. The answer to those who claim that immigration reform will lead to exploitation by adversaries is not to exclude talent. It is to build better architecture.

The Rangers didn’t vet me by where I came from or how old I was. They vetted me by what I could do. And then they constrained my access until I earned more.

That’s the model.

The Data Behind the Gut Feeling

The debate over talent sometimes gets stuck in symbolism. The hard baseline is simpler and more useful: the United States already relies heavily on foreign-born talent in science and engineering fields. This is not a proposal. It is the existing structure of American technical capability.

In 2024, the National Science Board published indicators showing that foreign-born workers made up 19 percent of the United States STEM workforce and 43 percent of doctorate-level scientists and engineers. At the highest levels of training—the people who actually push the frontier—nearly half came from somewhere else.

The innovation literature fits the same pattern. Britta Glennon’s comprehensive review in the Journal of Economic Perspectives (Winter 2024) surveys the evidence and finds “ample evidence that skilled immigrants have a strong positive effect on firm outcomes.” Her critical finding: when employers face immigration restrictions, they don’t hire more Americans. They offshore the work, automate it, or restructure around the constraint. Restrictions don’t keep jobs in America. They move capability abroad. Her follow-up study in Management Science (2024) quantified the effect: when H-1B visa restrictions tightened, affected firms increased foreign affiliate employment by 21 percent—not because they wanted to offshore, but because the immigration system gave them no other option.

William Kerr’s updated analysis in IMF Finance & Development (March 2025) puts numbers on the mobility: inventors migrate at twice the rate of college-educated workers; Nobel Prize winners migrate at six times that rate. The exceptional move. The question is whether they move here.

The National Foundation for American Policy analysis updated through October 2025 reports that immigrants have been awarded 36 percent of the Nobel Prizes won by Americans in chemistry, medicine, and physics since 1901—and 40 percent since 2000.

The 2025 Nobel Prize in Chemistry went to Omar M. Yaghi, who was born into a refugee family in Jordan and arrived in the United States alone as a teenager with limited English proficiency. He started at Hudson Valley Community College in Troy, New York. He bagged groceries and mopped floors. He is now at UC Berkeley, and his work on metal-organic frameworks may help solve clean water access for millions.

A refugee. A community college student. A Nobel laureate.

I know something about that trajectory.

In ninth grade, I made a list. Four things I would become:

  1. A shark biologist, after reading Jaws.
  2. A mercenary of some type, after reading Frederick Forsyth’s The Dogs of War.
  3. An Army Ranger, after reading Stars and Stripes articles on long-range reconnaissance patrol Rangers in Vietnam—articles my father sent me from his posting at Ubon Royal Thai Air Base while he flew combat missions.
  4. A “brain biologist,” to study and learn how and why I was such a weirdo.

Four impossible things written in a notebook by a fifteen-year-old military brat with damaged eyesight and a mother who had beaten the fighter-pilot dream out of him.

I did all four.

Shark biologist: I pioneered research in shark cell culturing and electroreception at institutions including Scripps. Mercenary: year two in Southern Africa, hunting poachers, doing work governments couldn’t or wouldn’t do. Army Ranger: 1st Battalion, 75th Ranger Regiment, perfect EIB score, award presented by Colonel Ralph Puckett. Brain biologist: biophysicist and neuroscientist, now writing about the mechanical and molecular foundations of trauma in Silent Scars, Bold Remedies: Cutting-Edge Care and Healing from Post-Traumatic Stress Injuries.

The credentialing system didn’t make that list. I did. And then I walked it—through a flunk-out at the University of South Florida, through Prince George’s Community College in Maryland, through a full scholarship to American University, through Ranger Battalion at thirty-five, through the African bush at age 54, through everything the system said I wasn’t supposed to survive.

I made my own doors. I always have.

Yaghi and I are the same story wearing different clothes. The system didn’t make room for us. We drew our own maps and walked them. And that’s the point: the United States has built a talent-filtering apparatus that would have excluded the very people who prove its value. The community college kid who wins the Nobel. The ninth-grader who wrote four impossible things in a notebook and then did all of them. The thirty-five-year-old private who outperforms Rangers half his age.

Credential-based filtering is not security. It is not efficiency. It is the systematic exclusion of people who don’t fit a trajectory that was never designed to identify capability in the first place.

When Omar Yaghi was asked what his first reaction was to learning he had won the Nobel Prize, he said: “Astonished, delighted, overwhelmed.”

I understand that feeling. Not because I’ve won a Nobel—but because I’ve stood in rooms I was never supposed to enter, holding credentials I was never supposed to earn, having done things the system said I couldn’t do. Every box on that ninth-grade list, checked. Every door that didn’t exist, built.

That’s what the American system is capable of when it works. And that’s what we’re currently in the process of strangling.

The Penn Wharton Budget Model (March 2025) found that shifting even 10 percent of future low-skilled immigration toward high-skilled STEM workers would grow the economy, reduce federal debt, and increase wages across all income groups—lower-skilled, higher-skilled non-STEM, and higher-skilled STEM alike. A rare “Pareto improvement” benefitting everyone.

A strategy that treats such talent as an afterthought will not merely miss an opportunity. It will weaken an existing pillar of national capacity.

When a Crackdown Chills the Lab

In 2018, the Department of Justice launched the China Initiative, framed around concerns including economic espionage and trade secret theft. By 2022, the Department moved away from the label while emphasizing a broader approach to nation-state threats.

The label changed. The damage lingered.

The Stanford Center on China’s Economy and Institutions documented what happened: after the China Initiative began, departures of Chinese-born scientists from US institutions increased by 75 percent, with two-thirds relocating to mainland China or Hong Kong. A survey published in PNAS found that 35 percent of Chinese-American scientists reported feeling unwelcome in the United States, 72 percent expressed feelings of insecurity as researchers, and 42 percent feared restrictions on their research freedom.

Fear is a signal. Scientists read signals.

The consequences are no longer theoretical. That CNN investigation in September 2025 documented the exodus in real time: eighty-five scientists, including leaders in AI, quantum computing, and biotechnology—fields the US government has formally designated as strategically significant—leaving American institutions for Chinese ones. A protein chemist who left the University of Maryland for Shanghai’s Fudan University noted there has been a “clear surge in the number of job applicants from overseas” at Chinese institutions.

“I know Chinese universities are bending over backwards to actively take advantage of this opportunity presented to them as a gift from a ‘perceived’ adversary,” he said.

A gift.

A security posture that treats broad communities as presumptive risk creates a self-inflicted strategic wound: it discourages exactly the people the United States needs in order to compete in frontier technologies. Competitors gain capability without having to recruit. They simply wait.

The Refugee Dividend, and the Trap of Making Enemies

History offers two lessons that need to be held together.

The first: American strategic capability has sometimes been strengthened by people who arrived because they had nowhere safe to go.

The Atomic Heritage Foundation’s account of refugee scientists in the Manhattan Project era shows how displaced experts—fleeing fascism, fleeing persecution, fleeing death—became part of the American wartime research ecosystem. Albert Einstein fled Nazi Germany. Enrico Fermi fled fascist Italy. They did not come because the paperwork was easy. They came because America was the last option. And they built the nuclear backbone that underpins US security to this day.

As Rachel Hoff and Reed Kessler note in War on the Rocks: today, the chances that Einstein could win the arbitrary H-1B visa lottery are a mere 11 percent.

The second lesson: mishandling foreign-born talent can create blowback that lasts generations.

Qian Xuesen was educated at MIT and Caltech. He helped found the Jet Propulsion Laboratory. He was one of the most brilliant aerospace engineers of his generation—and he was American by every measure except birth.

Then the Red Scare came. Qian was accused of Communist sympathies, stripped of his security clearance, placed under house arrest. In 1955, he was deported to China in a prisoner exchange.

He spent the next four decades building China’s missile and space programs. The rockets that now carry Chinese astronauts into orbit and Chinese warheads toward targets trace their lineage to a man America trained, accused, and expelled.

Iris Chang’s biography Thread of the Silkworm tells the full story. It should be required reading for anyone who thinks suspicion is a strategy.

These two stories do not cancel each other out. They point to the same operational conclusion: talent strategy must be paired with process legitimacy and security discipline. A system that invites talent in and then governs it through paranoia risks turning a potential asset into a long-term adversarial advantage for a rival. A system that invites talent in and then integrates it through transparent rules, architectural constraints, and performance-based trust can convert lawful opportunity into durable alignment.

The Spell-Caster’s Son

Before I learned to walk up to poachers at twenty meters, I learned to walk up to Supreme Court Justices at embassy parties.

When I was studying at American University and Georgetown, my parents were listed in the DC Green Book—the who’s who of dignitaries and diplomatic society. My mother, the same woman who had beaten the fighter-pilot dream out of me, gave me her invitations to embassy functions. A dozen parties. Ambassadors, dignitaries, the kind of rooms a twenty-one-year-old punk from community college had no business entering.

I walked into every one of them like I belonged.

At one reception, I noticed Sandra Day O’Connor—a sitting Supreme Court Justice—sitting alone. No one was approaching her. Too intimidated, too deferential, too aware of the protocol they might violate.

I walked straight up and chatted her up.

Did the same with Timothy Leary. Did the same with Jack Nicholson. Did the same with ambassadors from countries I couldn’t find on a map. A twenty-one-year-old punk, crashing diplomatic society on borrowed invitations, taking space that no one else had the audacity to claim.

My mother called it spell-casting. She had decorated six foreign embassies in Washington, charmed ambassadors into letting her redesign their official residences, designed The Emerald Ball at the Kennedy Center and The International Fair in Rock Creek Park. She could make anyone feel like the most important person in the room—right up until she struck. I learned that from her. The art of disarming people with genuine human contact. The understanding that deference is what people expect, and authenticity is what they crave.

Years later, in the African bush, I used the same technique on poachers. Wave from a hundred meters. Yell something friendly. Close the distance while they wonder who this idiot white boy is. At twenty meters, they’re curious. They’ve let their guard down.

Then I did what I had to do.

It’s the same move. Embassy parties and poacher camps. Sandra Day O’Connor and men who killed elephants for profit. Walk up, be human, take the space everyone else is too afraid to claim.

Here’s why this matters for the argument of this series: that’s how America used to work.

The DC Green Book. Embassy parties. A country so magnetic, so confident in its own gravitational pull, that a twenty-one-year-old community college transfer could walk into rooms with Supreme Court Justices and ambassadors and belong there through sheer audacity. The system was porous enough to let talent flow upward. The doors weren’t locked—they were waiting for someone bold enough to push.

That porosity wasn’t weakness. It was strategic advantage. It was how America attracted the world’s best, integrated them into the highest levels of society, and converted their talent into national capability. It was soft power made flesh.

That’s the America we’re losing. The one that let Omar Yaghi in from a refugee camp. The one that let me walk up to Sandra Day O’Connor. The one that said “Let him” when a thirty-five-year-old demanded his shot at the EIB.

We’re replacing porosity with paranoia. And paranoia is not a strategy.

Absolute Value: The Alchemy That Makes Fire Useful

My mother taught me many things, most of them in ways no child should learn. She beat me severely enough to destroy my eyesight—the eyesight I needed to become a fighter pilot like my father, who flew F-4 Phantoms in Vietnam. She shifted my teeth with her fists. She left scars on my head and neck that I only understood decades later through hypnosis.

She also taught me to survive.

She taught me to read people the way a soldier reads terrain—for hidden dangers, for opportunities, for the moment to move. She taught me psychological warfare more sophisticated than anything Sun Tzu or Clausewitz ever wrote. She taught me that the biggest killer in the world is not cancer or heart disease. It’s arrogance. And when some arrogant threat underestimated me, that gave me leverage to strike.

In mathematics, there is a concept called absolute value. Whatever number you put between those two little brackets—positive or negative—comes out positive. The brackets strip away the sign and keep only the magnitude.

I learned to apply that concept to my life. The negative energy from pain and suffering can, with work and dedication, be converted to something entirely positive and useful. Trauma becomes fuel. Fear becomes focus. The fire that burns you can also forge you.

That is also the story of America’s relationship with immigrant talent.

Omar Yaghi’s fire was a refugee camp. Einstein’s fire was Nazi persecution. Fermi’s fire was fascist oppression. My fire was a childhood that would have destroyed someone who didn’t learn to transmute it.

Prometheus does not matter because he stole fire. Prometheus matters because he knew what to do with it once he had it.

The United States has historically been a place where people with fire—people fleeing, people seeking, people burning with capability that their home countries couldn’t use or wouldn’t tolerate—could come and convert that fire into light.

That is the strategic asset. That is the asymmetric advantage no amount of money can buy.

And we are currently in the process of giving it away.

From Accidental Magnet to Deliberate Strategy

The United States has long benefited from being a destination. But being a destination is not the same thing as running a strategy.

In 2024, seventy former national security officials—cabinet members, military leaders, intelligence professionals from both Republican and Democratic administrations—sent a letter to Congress warning about STEM immigration bottlenecks. Their conclusion:

“China is the most significant technological and geopolitical competitor our country has faced in recent times. With the world’s best STEM talent on our side, it will be very hard for the United States to lose. Without it, it will be very hard for us to win.”

In April 2025, Senators Dick Durbin and Mike Rounds introduced the bipartisan Keep STEM Talent Act, which would retain international graduates with advanced STEM degrees while imposing new vetting requirements.

As Senator Rounds stated: “Legal, highly skilled STEM immigration is crucial for our nation and has opened doors for talented immigrants like Albert Einstein to come to America. Particularly with the advancements of artificial intelligence and cybersecurity, we must keep talent in the United States and stay ahead of our near peer competitors such as China and Russia.”

The starting point for a real strategy already exists: the government’s own list of critical and emerging technologies. When that list identifies the domains that matter most, it also identifies the talent domains that matter most. The missing step is turning that priority into a coherent pipeline that actually functions for humans making career decisions under uncertainty.

That pipeline begins by reducing predictable friction in lawful immigration pathways intended for extraordinary talent. The government already describes and administers the relevant categories—the O-1 visa for extraordinary ability, the H-1B for specialty occupations, the EB-1 and EB-2 green card tracks. The USCIS fee schedule shows what it costs. The task is to make timelines, standards, and expectations predictable enough that the United States becomes the low-uncertainty option, not merely the high-prestige option.

In January 2025, the Department of Homeland Security launched a major overhaul of the H-1B visa program, removing the traditional employer-employee requirement and allowing professionals in specialty occupations to self-sponsor. The O-1 visa for individuals with extraordinary ability saw expanded evidence criteria for AI, quantum computing, clean energy, and biotechnology professionals. These are steps in the right direction.

The pipeline also requires retention. As many as 90 percent of foreign students receiving advanced STEM degrees are currently forced to leave the country after graduation under existing rules—after American taxpayers have funded their research, after American universities have trained them, after American labs have integrated them into teams working on American priorities.

We train them. We fund them. Then we send them home to compete against us.

Finally, the pipeline requires secure integration that does not collapse into blanket suspicion. Security should scale through design: compartmentalized access, continuous monitoring, progressive responsibility based on demonstrated reliability. The model is not the background check as gate. The model is the architecture as constraint.

That’s what I learned in the Rangers. Trust is earned in increments. Access follows performance. And you never stop watching, not because everyone is guilty, but because the system has to work even when someone is.

The Last Fire

First Sergeant Van Houten didn’t know what I would become when he said “Let him.” He didn’t know I would earn a perfect score. He didn’t know I would go on to overseas operations, to anti-poaching work in Africa, to a career that would take me to more than a hundred countries and teach me things no classroom ever could.

He made a bet on capability. The system almost excluded me. Two words let me through.

Somewhere right now, there is a scientist standing at the edge of a decision. She has the talent to transform American capability in a field the government has formally designated as strategically significant. She has the drive. She has the fire.

And she is looking at a system that treats her as a risk to be managed rather than an asset to be integrated.

If we lose her—not to a competitor’s recruitment campaign, but to our own uncertainty and bureaucratic friction and ambient suspicion—we will never know what we lost. She will simply build the future somewhere else. Her papers will appear in Chinese journals. Her patents will be filed in Shanghai. Her students will work for companies that compete against American firms.

And some analyst years from now will write a report wondering how we fell behind.

The United States government has already said, in plain language, which technology domains it considers strategically significant. The National Science Board has already quantified how much the American science and engineering enterprise relies on foreign-born talent. The innovation literature has already assembled evidence that high-skilled immigration correlates with measurable innovation outputs. And the research on deterrence and fear has already raised a warning: policies framed as security can still weaken security if they drive talent away.

That warning is now materializing in real time. Scientists are leaving. Chinese institutions are recruiting with unprecedented success. The reverse brain drain that analysts warned about has become documented fact.

Seventy former national security officials from both parties have already told Congress: With the world’s best STEM talent on our side, it will be very hard for the United States to lose. Without it, it will be very hard for us to win.

This does not call for naïve openness, and it does not call for paranoid closure. It calls for a system that treats lawful talent mobility as strategic infrastructure—disciplined, predictable, and backed by security architecture that scales without stigmatizing the very people a competitive society needs.

Prometheus does not matter because he stole fire.

Prometheus matters because fire changes who can build the future.

Strategy in this century is deciding where that fire lands—and building a system worthy of holding it.

The Billion-Dollar Bonfire: How a $99 Toy Turns a Trillion-Dollar Fleet to Ash

Executive Summary

The United States Air Force faces an existential threat not from peer-state missiles, but from $99 commercial drones. While we spent decades building a Maginot Line in the sky, we left our trillion-dollar fleet parked in the open, vulnerable to swarms that cost less than a Pentagon coffee budget. This paper exposes the “Glass Jaw” of American airpower: the catastrophic vulnerability of forward-deployed aircraft to cheap, attritable ground strikes.
Current high-tech defenses are failing.

Billion-dollar solid-state lasers are defeated by simple magnesium smoke—a hard-counter based on Mie scattering physics—and kinetic interceptors are paralyzed by collateral damage risks in urban environments. Worse, our 12-year acquisition cycle cannot compete with the enemy’s 2-day Amazon delivery speed.

The solution is not more technology; it is humility. We must adopt “Redneck Solutions”—industrial fishing nets (“The Tuna Dome”), shotgun countermeasures (“Duck Hunt”), and inflatable decoys. These low-tech defenses work immediately, cheaply, and without software updates. Continued reliance on MIL-SPEC arrogance over practical physics will result in the destruction of US Air Force assets on the ground before a single pilot takes off. We can catch the threat in a net, or we can sweep up the ashes.

The Glass Jaw

In the Pentagon, the delusion has a name. They call it “Sanctuary.”
The Generals look at the oceans. They look at the nuclear triad. They look at the young airman with an M4 standing at the gate. They believe this is security.

It is theater. Expensive theater. The kind of theater where the tickets cost $850 billion a year and the ending is a surprise to everyone except the enemy.

Walk the flight line at Langley. At Eglin. At Nellis. The F-22s sit wingtip to wingtip. The F-35s. The KC-46 tankers. Soft. Full of jet fuel. Covered in sensors that cost as much as a house. Arranged with all the strategic foresight of a Costco parking lot.

We park them like Chevrolets at a used car lot. Correction: used car lots have security cameras that work.

The enemy sees this. He is not stupid. He just has WiFi and a grudge. He cannot fight the F-35 in the air. In the air, the F-35 is a god. So he decides to slay the god while it sleeps. While it’s parked. While it’s getting a $44,000 paint job to maintain its stealth coating.

More than 350 drone incursions were detected over U.S. military bases in 2024 alone. At Langley Air Force Base—home to the F-22 Raptors and Air Combat Command Headquarters—coordinated drone swarms flew at altitudes between 100 and 4,000 feet for seventeen consecutive days. Seventeen days. That’s not an incursion. That’s a commute. Naval Station Norfolk, the world’s largest naval base, was along the flight path. The military could not track, identify, or stop the intrusions. The world’s most powerful military, defeated by something your nephew got for Christmas.

He rents a box truck. He drives to the industrial park two miles from the runway. He parks behind a warehouse, just half a block from Burger King and McDonald’s. After a Happy Meal, he opens the back door and smiles.

Inside: one hundred drones. Consumer quadcopters start at $99. Each carries a pound of C4. Total investment: less than a Pentagon coffee budget.

The base fence is ten feet high. The drones fly at fifty feet. The airman at the gate is watching for a terrorist in a van. He is not looking up. Nobody told him to look up. The training manual is from 2006.

The attack takes ninety seconds. The drones rise. They swarm. They zoom over the terrain at two miles in a minute. They dive.

They do not hit the bunkers. Bunkers are hard. They hit the jets sitting in the sun. The jets we left outside. Like lawn furniture. Like we’re daring someone to steal them.

We shoot down ninety. We hold a press conference. We give ourselves medals. It does not matter. Simulations show that when eight drones attack an Aegis-class destroyer, an average of 2.8 still penetrate defenses. Ten get through. Ten jets burn. Mission accomplished—for the enemy.

The F-35A costs $82.5 million per aircraft as of July 2024, according to the F-35 Joint Program Office. The F-22 program cost $67.3 billion for 195 aircraft—approximately $350 million per unit. Ten jets destroyed equals $1.5 billion in damage. The enemy spent pennies. He put it on a credit card. He got airline miles.

Return on investment: seven hundred and fifty thousand percent. Wall Street would kill their grandmothers for those numbers. Literally. They have.

The Physics of Failure

We believe in technology. We love acronyms. We love lasers. We especially love lasers with acronyms. The longer the acronym, the bigger the contract.
The enemy loves high school chemistry. And physics. He paid attention in class. We were busy writing requirements documents.

The Magnesium Curtain

We spent billions on High Energy Lasers. The Generals love them. They look great in PowerPoint. They make swooshing sounds in the animations. Raytheon’s 50-kW laser can burn through a small consumer drone in seconds. In the lab. In perfect weather. In San Diego. Where it never rains. Where the enemy has politely agreed not to use countermeasures.

But the enemy knows about magnesium. Eighth grade science fair. Blue ribbon. His parents were very proud.

The lead drones carry no bombs. They carry magnesium flares. They drop magnesium oxide dust. Magnesium burns at flame temperatures ranging from 2,500-3,500 K (approximately 2,200-3,200°C or 4,000-5,800°F), producing brilliant white light and dense smoke. It’s basically a rave for photons. A very expensive rave that we’re paying for.

The laser hits the smoke. We physicists call it Mie Scattering. Here is the punchline: The Pentagon’s favorite solid-state lasers (like the 50kW class systems currently deployed) operate at 1.064 microns—the near-infrared. Burning magnesium produces Magnesium Oxide (MgO) particles with an average diameter of roughly 0.5 to 1.5 microns. Do you see the problem? The particle size isn’t just random. It’s a mathematically near-perfect match for the laser’s wavelength. Think: giant disco ball at Studio 54. We didn’t just build a laser that can be defeated by smoke. We built a laser specifically hard-countered by the most common pyrotechnic on earth. We spent billions to design a weapon that’s allergic to a flare.

High-energy lasers face “diminished effectiveness in rain, fog and smoke, which scatter laser beams”. Diminished effectiveness. Pentagon-speak for “doesn’t work.” The thermal cameras go white. Blind. The operators see nothing. They paid $200,000 for night vision that now shows them the inside lining of a cloud.

“Substances in the atmosphere—particularly water vapor, but also sand, dust, salt particles, smoke, and other air pollution—absorb and scatter light, and atmospheric turbulence can defocus a laser beam,” according to a 2023 Congressional Research Service report on directed-energy weapons.

In other words: weather exists. Someone should have mentioned this. Maybe in one of the 847 meetings about laser development. Maybe during the 12-year acquisition process. Maybe before we spent the GDP of a small nation on a weapon that can be defeated by fog. Or a campfire. Or a teenager with a bag of powder from a chemical supply store. Fifty dollars. Free shipping.

The Clutter

Our radar was built to track Soviet bombers at Mach 3. Big. Fast. Metal. Radiating heat like a flying furnace. The radar is very good at finding flying furnaces. Unfortunately, the enemy stopped building flying furnaces. We didn’t get the memo.

The radar filters out noise. Birds. Rain. Anything slow. Anything small. Anything that looks like it belongs in the sky. It was designed this way on purpose. By smart people. Who never imagined that the enemy would build weapons that look like birds.

The drone is small. Plastic. Slow. To the radar, it is a bird. To the radar, the swarm is a flock of starlings. A hundred starlings. Carrying explosives. The radar sees nature. How peaceful.

“At low altitude, probably not,” admitted General Gregory Guillot when asked if standard FAA or surveillance radars could detect drone swarms over Langley. “Probably not.” That’s a four-star general. That’s the commander of NORAD and USNORTHCOM. That’s the man in charge of defending North America. Probably not. 

Maybe they’re the fuzzy-orange foo fighters from the skies of WWII.

The operator chases a Red Bull with a Monster tallboy. His screen is clear. Everything is fine. The birds are flying. Some of the birds have C4 strapped to them. The radar doesn’t mention this. The radar was not programmed to care.

The enemy flies his killer drones in the middle of Amazon delivery traffic. Next to the news helicopter. In the same airspace as grandma’s medication delivery. He files a flight plan. He’s very polite about it.

The Lieutenant sees fifty dots. Forty are delivering toothpaste and M&Ms. Six are delivering pizza. Three are filming real estate listings. One is delivering high explosives. He has three seconds to pick a target.

If he shoots the toothpaste/M&Ms drone, his career ends. CNN runs the footage for six weeks. “Military Destroys Amazon Christmas Package.” Congressional hearings. His wife threatens to leave him.

If he shoots the news chopper, he goes to prison. Orange jumpsuit. Bad food. No pension.
If he shoots the explosive drone and misses, he’s on the news anyway. “Military Fires Missiles Over Suburban Neighborhood.” Property values collapse. Lawsuits. His wife definitely leaves him.

So he waits. He calls his supervisor. His supervisor calls legal. Legal is at lunch. The drones do not wait. The drones do not have a legal department. The drones do not take lunch.

The Self-Inflicted Wound

We are civilized. We have laws. We have lawyers. So many lawyers. Entire buildings full of lawyers. The enemy has neither. The enemy has a YouTube tutorial and a wet dream.
He launches from a school zone. He picks the school zone on purpose. He knows we know where he is. He knows we can’t shoot. He waves.

He flies over a suburb. Nice suburb. Good schools. HOA keeps the lawns tidy. Go Trump signs in some yards, Yay Hegseth signs in others. United in their imminent vulnerability.
He approaches from the city side, not the ocean. The ocean approach has sensors. The city side has Starbucks. He stops for coffee. Grande. Oat milk. He tips well. He’s about to have a very good day.

The Base Commander has the C-RAM. Twenty-millimeter explosive rounds. It sounds like a chainsaw having a seizure. Very impressive. Very loud. The system uses the 20mm HEIT-SD (high-explosive incendiary tracer, self-destruct) ammunition, which explodes on impact with the target or on tracer burnout, reducing the risk of collateral damage. Reducing. Not eliminating. Reducing. Like a sale at Kohl’s. Thirty percent off collateral damage. Bargain.
But gravity is law. Gravity does not care about our lawyers. Gravity does not attend briefings. Bullets go up. They must come down. Newton was very clear about this. We ignored Newton. We ignore a lot of things.

The C-RAM uses HEIT-SD ammunition specifically because even self-destruct rounds “could cause unintended collateral damage” in urban areas. Could. Such a gentle word. “Could cause unintended collateral damage.” Translation: shrapnel might kill taxpayers. Taxpayers vote. Taxpayers sue. Taxpayers have local news on speed dial.

“In an urban area, if C-RAM is able to knock these mortars out and have them explode up in the air, the debris and the shrapnel from some of those rounds are going to fall. This can cause some civilian casualties.”

Some civilian casualties. Some. We spent forty years learning to say “collateral damage” with a straight face. We have entire public affairs offices dedicated to explaining why civilian casualties are actually not that bad. But those civilian casualties were overseas. Those civilians were other people’s voters. These civilians have Instagram. These civilians went to high school with the reporter. These civilians are three miles from a congressional district that flipped last election.

He cannot fire.

So he uses the microwave. The HPM. High-Powered Microwave. Another acronym. Another PowerPoint. This one cooks the electronics. The drone dies mid-flight. Victory. Sort of.
It becomes a twenty-pound brick at terminal velocity. It’s still carrying its payload. It’s just not steering anymore. It’s now brain-dead but still ballistically active. Physics takes over. Physics is always undefeated.

It crashes through the roof of a house. Through the baby’s room. Through the kitchen where mom was making breakfast. Through the windshield of the minivan in the driveway. The one with the “Support Our Troops” bumper sticker. Irony doesn’t care about bumper stickers. The bomb did not detonate in the air. We stopped that. We’re very proud. It detonates on the ground. In the suburb. Next to the family who moved there because the schools were good and the crime was low and it was safe. It was safe.

The enemy wins either way. Heads he wins. Tails we lose.

If the drone hits the jet, he destroys $100 million in aircraft. Pictures on Al Jazeera. Pictures on RT. Recruitment videos. The F-35 burning makes excellent B-roll.

If we shoot it down over the neighborhood, he destroys something more valuable. He destroys the illusion. He destroys the story we tell ourselves. He destroys the sanctuary. The photos go viral. The mother’s Facebook post gets shared four million times. The Mayor sues. The Governor screams. The President issues a statement. The statement has been focus-grouped. It includes the phrase “full investigation.” There is always a full investigation. The investigation finds that everyone followed procedure. The procedure was wrong. Nobody changes the procedure.

The President orders a ceasefire. No more shooting over neighborhoods. The lawyers agree. The Generals comply. The enemy retreats to his corner and reloads for Round 2.

We are held hostage by our own zip code. We spent $850 billion to build a military that cannot defend a Denny’s without a permit and a prayer.

The Procurement Disease

We buy weapons like we are building cathedrals. Twenty years. Committees. Requirements. Subcommittees. Requirements about requirements. Bids. Counter-bids. Protests. Lawyers. More lawyers. Consultants. Consultants for the consultants. Prototypes. Failed prototypes. Revised prototypes. Paint. The paint takes eighteen months. The paint has its own program office.

The enemy buys weapons like groceries. He has a list. He goes to the store. He checks out. He kills people. Tuesday.

The Timeline

The terrorist watches a YouTube video. “How to Build a Drone Swarm for Dummies.” One point two million views. Monetized. Day One.

He orders parts on Amazon. Prime shipping. Free with membership. He’s also ordering my book, Silent Scars Bold Remedies: Cutting-Edge Care and Healing from Post-Traumatic Stress Injuries; trying to mitigate what’s about to happen. The algorithm suggests C4. Just kidding. The algorithm suggests batteries. He already has C4. Day Two.

The parts arrive. In a box with a smile on it. He 3D-prints a bomb release. The printer cost $200. The file was free. Some kid in Finland who lives by sisu made it. The kid is fifteen. The kid has a Patreon. Day Four.He tests it in a field. It works. Of course it works. It’s not complicated. A toaster is more complicated. He films the test. He might post it later. Might get some followers. Day Seven.
Day Eight, he is ready to live-fire.

Meanwhile, in America, in the Pentagon, in a conference room with bad coffee and fluorescent lighting:

On average, the Department of Defense takes almost 12 years to deliver the first version of a weapon system. Twelve years. The iPhone didn’t exist twelve years before the iPhone. The enemy’s grandchildren will have grandchildren. The threat will have evolved seventeen times. We’ll still be in committee.

Year One, the Pentagon realizes drones are a threat. Someone writes a memo. The memo goes to a committee. The committee schedules a meeting. The meeting is in six months. There’s a conflict with another meeting.

Year Two, they form a Counter-UAS Task Force. The Task Force has a logo. The logo took four months. There were concerns about the font. The Task Force has a mission statement. The mission statement has been wordsmithed. Everyone is very proud of the mission statement. The enemy does not read the mission statement.

Year Three, the Task Force issues a Request for Information. Forty-seven companies respond. Forty-six of them are the same five contractors wearing different hats. One is a guy in a garage who actually has a good idea. His proposal is rejected for improper formatting. He used the wrong margin size.

Year Five, Raytheon gets a contract to study the feasibility of a laser. The study costs $400 million. The study concludes that lasers are feasible. This is news to no one. Lasers have been feasible since 1960. But now it’s official. Now there’s a PDF.

Year Seven, the prototype fails in the rain. It was tested in New Mexico. It does not rain in New Mexico. It rains in the places where wars happen. Nobody thought to check. The prototype goes back for redesign. The redesign will take three years. There’s a supply chain issue. The supply chain is in China. We’re not supposed to talk about that.

Year Ten, the system is fielded. Ten million dollars per unit. It does not work against the magnesium disco ball. It works in the desert when no one is shooting back. The PowerPoint said it would work everywhere. The PowerPoint lied. PowerPoints always lie. We believe them anyway.

“DOD remains deeply entrenched in a traditional linear acquisition structure—characterized by rigid, sequential processes—that has proven inadequate in adapting to evolving threats.”
This isn’t just my opinion. Ask Shelby Oakley.

She’s the Director of Contracting and National Security Acquisitions at the GAO. She is the woman whose job is to tell the truth when everyone else is lying about the schedule. Her assessment?

“DOD remains deeply entrenched in a traditional linear acquisition structure—characterized by rigid, sequential processes—that has proven inadequate in adapting to evolving threats.” Inadequate. That is the polite government word for “suicide pact.”

These are not compliments. These are words most likely to appear in someone’s obituary. New weapons can take five to seven years from concept to production under normal procedures. Normal. This is normal. We have normalized twelve-year timelines. We have normalized fighting tomorrow’s war with yesterday’s bureaucracy. We have normalized losing.

One study found that following all current regulations, it would take about two years to produce a major contract—to buy nothing. Two years. To buy nothing. To issue the paperwork that allows you to begin the process of thinking about maybe possibly purchasing something. Two years of meetings about meetings. Two years of lawyers reviewing lawyers. Two years of the enemy building drones.

We are fighting software velocity with bureaucracy speed. We are bringing a loose-leaf binder to a knife fight. We are bringing a 12-year acquisition cycle to a 12-day threat development timeline. The math: very bad, not good. The math has never worked. We keep doing the math anyway.

The Redneck Solution

We cannot wait for the laser. The laser is a promise. The laser is a first date that keeps getting postponed. The laser is your friend who’s “definitely coming” but never shows up. We need a net. An actual net. The kind fishermen use. The kind you can buy at Cabela’s. The kind that doesn’t need a software update or a congressional appropriation or a twelve-year development cycle or a PowerPoint with a swooshing sound effect.

The Tuna Dome

A drone propeller spins at ten thousand RPM. Fast. But weak. It cannot handle friction. It cannot handle string. String. The technology that defeated the drone was invented before writing.

We do not need a missile. We do not need a laser. We do not need a $400 million study about the feasibility of defeating drones. We need string.

Industrial fishing nets. Tuna nets. Cargo nets. The nets your uncle uses. The nets that are currently on sale at Harbor Freight. String them between the light poles. Drape them over the alert pads. Cover the jets like you’re keeping them fresh for tomorrow.

It looks ugly. The Generals hate it. It’s not in the doctrine. It ruins the photo op. The jets look like they’re wearing hairnets. The base looks like a fish market. Senators won’t want to visit. The Lockheed lobbyist is confused. Where’s the contract? Where’s the overrun? Where’s the eighteen-month paint job?

The net does not care about photo ops.

When the drone hits the net, it tangles. The motor strains. The motor burns out. The drone hangs there, pathetic, wrapped in twine, defeated by technology from 3000 BC. The jet is safe. The jet doesn’t care if the net is ugly. The jet just wants to not explode.

A missile costs millions. The missile might miss. The missile might hit the wrong thing. The missile has lawyers. Net-based capture devices deployed from helicopters are among the “potential solutions” being evaluated. Being evaluated. Still. Twelve years from now, we’ll have a report about nets. The report will recommend more study.

A net costs five thousand dollars. You can buy it today. You can install it tomorrow. The net works every time. The net does not need a software update. The net does not need a cybersecurity review. The net does not need an environmental impact statement. The net does not care about magnesium smoke. The net does not care about rain or fog or the feelings of defense contractors. The net just works. That’s why we won’t buy it.

The Wile E. Coyote Protocol

The enemy wants the hot jets. He looks for heat signatures. His drone has a thermal camera. It cost $40 on AliExpress. It’s looking for engines. It’s looking for exhaust. It’s looking for the things that cost $100 million each.

So we lie to him. It’s not complicated. It’s not expensive. It’s annoying that no one has thought of it. Someone probably has. They probably wrote a memo. The memo is in a drawer.

Inflatable decoys. Rubber F-35s. The kind we used in World War II. The kind Patton used. Patton is dead. His ideas should not be.

Space heater inside. Fifty dollars at Walmart. Sixty if you want the oscillating kind. The heater creates the signature. The balloon creates the shape. The drone sees a jet. The drone dives. The drone hits a balloon. The balloon pops. Cue: sad trombone.

We hide the real jets in maintenance sheds. Cover them with thermal blankets. The blankets cost $200. They hide $100 million aircraft. The math is good. China has built more than 3,100 aircraft shelters—over 650 hardened and 2,000 non-hardened—to protect its fleet. China. The country we say we’re preparing to fight. They have shelters. We have sunshine. The U.S. has built just 22 new hardened shelters in the Indo-Pacific in the past decade. Twenty-two. China built three thousand. We built twenty-two. But we had meetings about building more. Lots of meetings.

Recent war games show 90% of U.S. aircraft losses would occur from ground strikes rather than air combat. Ninety percent. Not in the air. On the ground. Parked. Sitting. Waiting. We built planes that can defeat any enemy in the sky. Then we parked them where any idiot can blow them up. This is strategy. British SAS Lieutenant Colonel Paddy Mayne in WWII taught the Germans that this lazy act was just plain madness. He alone destroyed three times more enemy planes than the finest RAF pilot. And his boots never left the ground.

The swarm comes. The sensors see heat. They dive. They blow up balloons.

We lose a thousand dollars of rubber. We save a hundred million in heavy metal. The math is simple. A child could do it. A child has done it. The child works for the other side now. He’s doing fine.

Duck Hunt

The laser failed. We covered that. The jammer failed. The enemy changed frequencies. Frequencies are free to change. The jammer cost $5 million. The frequency change cost nothing. The drone is fifty yards out. Closing fast.

Give the guard a shotgun.

Remington 870. Twelve gauge. Number four buckshot. The most produced shotgun in American history. Your grandfather had one. Your grandfather could have defended the air base. Your grandfather is dead. His shotgun is in a closet. It still works.

A shotgun creates a wall of lead. Hundreds of pellets. Spreading. Covering. Forgiving. It does not need perfect aim. It does not need a targeting computer. It does not need a software update. It needs a person who can point and pull.

It shreds plastic rotors. It destroys batteries. It turns a $500 drone into confetti. It turns a threat into a story. “Yeah, I shot it down. With a duck gun. You want to see the YouTube video? My buddy filmed it.”

Ukrainian forces adopted semi-automatic shotguns that “have proven remarkably effective at disrupting Russian UAV operations”. Ukraine. The country we’re sending billions to help. They figured it out. Shotguns. The technology we invented. They’re using it better than we are. They don’t have committees. They have funerals.

Allied nations including France, Italy, and Belgium have deployed different Benelli shotguns with traditional and specialized drone shells; during field tests, these weapons have proven very effective at taking down FPV drones from 80–120 meters away. France. Italy. Belgium. Not known for their military innovation. Leading us. With shotguns. The weapon of bird hunters and home defenders. The weapon the Pentagon forgot existed.

The Benelli M4 Drone Guardian has an effective combat range of 50 meters, with potential maximum range up to 100 meters. Fifty meters. That’s 150 feet. That’s half a football field. That’s plenty. Shotguns are “more effective against drones than regular rifles because of their spreading pattern of multiple projectiles”—damaging one propeller is sufficient to make a quadcopter incapable of flight.

One propeller. One pellet. One shot. Done. And if you have an ATI Bulldog with a ten-round mag of number four buck like I do, then cowabunga. 
At fifty yards, a duck gun is the deadliest anti-drone weapon on earth. At fifty yards, a hundred-year-old technology beats a billion-dollar program. At fifty yards, your granddad beats Raytheon.

We spent a long damn time trying to build something better than a blunderbuss with a carved dragon’s head at the muzzle. We failed. The dragon is fine. The dragon was always fine. We just wanted something more expensive.

The Sanctuary Is Over

The sanctuary was always a lie. A comfortable lie. An expensive lie. A lie we told ourselves while we built systems that don’t work against enemies who impulse-buy online. The ocean protects nothing. Drones don’t need boats. The fence protects nothing. Drones don’t need gates. The guard at the entrance protects nothing. The enemy is not walking in. He’s flying over. While the guard watches the road.

“It’s been one year since Langley had their drone incursion and we don’t have the policies and laws in place to deal with this? That’s not a sense of urgency,” said retired General Glen VanHerck. One year. Seventeen days of drones over the crown jewels of American airpower, and one year later, we have policies being developed. Laws being considered. Meetings being scheduled. The next incursion being planned.

“There’s a perception that this is fortress America: two oceans on the east and west, with friendly nations north and south, and nobody’s gonna attack our homeland. It’s time we move beyond that assumption.”

Time to move beyond. We’ve had time. We’ve had decades. We spent the time building lasers that don’t work in rain and jammers that don’t work when the enemy changes channels and missiles that cost too much to shoot at toys. We spent the time in meetings. We spent the time on PowerPoints. We spent the time assuming the enemy was stupid. The enemy was not stupid. The enemy was shopping.

These bad actors are using cheap hardware and great ideas to defeat a trillion-dollar military. We are drowning in budget but starving for imagination. We need to start thinking like they do and stop being so snobby. We turn our noses up at solutions that don’t cost a billion dollars. We think if it doesn’t have a MIL-SPEC serial number, it’s beneath us. That level of arrogance is a target.

Look at Ukraine. Russia has the money. Russia has the resources. Russia has the “invincible” heavy metal. Yet they are getting dismantled by hobbyists with soldering irons. Ukraine is proving that a consumer drone with a grenade is more effective than a tank with a conscripted crew. They are trading pennies for millions, and they are winning the exchange.
We have a choice.

Keep pretending. Keep buying expensive toys that work great in the desert when no one is shooting back. Keep writing white papers about Next Generation Air Dominance while the current generation sits outside, uncovered, unprotected, waiting for a kid with a Radio Shack drone and a death wish.

The Caloric Kill Switch

Food System Dependency as Irregular Warfare

Updated March 19, 2026. This paper was originally published on February 4, 2026. The current version incorporates live evidence from the Iran war and Strait of Hormuz closure, which has validated the paper’s central thesis in real time.

The Fallacy: Food as a Market

Global food security is framed as an agricultural productivity challenge requiring better seeds, smarter farming, and climate adaptation. This framing is the fallacy. The global food system is not a market. It is a weapon system disguised as commerce, controlled at every chokepoint by a small number of actors who understand exactly what they hold.

Four companies, Bayer, Corteva, Syngenta, and BASF, control fifty-six percent of the global commercial seed market and sixty-one percent of the global pesticides market according to GRAIN and ETC Group’s 2025 analysis. Russia and Belarus together account for roughly forty percent of global potash exports. China holds only five percent of global phosphate reserves but has long accounted for over forty percent of global phosphate rock production. Morocco holds over seventy percent of global phosphate reserves through the state-owned OCP Group, which controls thirty-one percent of the world phosphate product market and generated 9.76 billion dollars in revenue in 2024. At current production rates, Morocco’s deposits could last over 1,300 years. China’s reserves will last until approximately 2058. The United States’ will last until roughly 2062. One country holds a millennium of leverage over the mineral foundation of global agriculture, a concentration that exceeds Saudi Arabia’s historical position in oil. The fertilizer that grows the crop, the seed that becomes the crop, and the chemical that protects the crop are concentrated in fewer hands than the oil market was in 1973.

Nobody has placed the seed monopoly, the fertilizer dependency, the precision agriculture cyber vulnerability, and the food processing fragility on the same table and called it what it is.

The Center of Gravity: The Input Stack

A modern farm does not grow food from soil and sunlight. It assembles food from a stack of purchased inputs: proprietary seeds, synthetic fertilizers, chemical pesticides, GPS-guided machinery, and cloud-connected precision agriculture platforms. Remove any layer of the stack and the farm does not produce. The center of gravity is not the field. It is the input stack. And every layer of the stack is concentrated.

The seed layer is an oligopoly. Bayer, Corteva, Syngenta, and BASF dominate global plant breeding. In the United States, two firms alone, Bayer and Corteva, control seventy-two percent of the corn seed market and sixty-six percent of the soybean seed market, according to a 2025 analysis published by MIT Press. Three firms own ninety-five percent of U.S. patents for genetically modified corn. Many proprietary seeds are engineered to perform optimally only when paired with the same company’s pesticides, Bayer’s Roundup Ready line being the most prominent example. The farmer enters a dependency loop that concentrates control of global food production in four boardrooms.

The fertilizer layer is a geopolitical chokepoint. Russia handles twenty-three percent of global ammonia exports, twenty-one percent of potash, fourteen percent of urea, and twelve percent of phosphate. When Russia invaded Ukraine in 2022, the fertilizer price index surged across all three major categories. Potash prices alone jumped fifty-three percent between January and April 2022. Phosphate rock rose thirty-eight percent. The disruption cascaded: Russia restricted ammonium nitrate exports, China banned phosphate exports to protect domestic supply, and Belarus, already sanctioned by the EU and United States since 2021 over its role in the migrant crisis, saw its potash trade channels collapse. 

Developing nations that depend entirely on imported fertilizer saw planting seasons unravel. Sri Lanka’s ban on synthetic fertilizer imports, compounded by the price shock, contributed to a thirty-percent rice yield decline that helped trigger the political crisis ending in the president’s resignation. Pakistan’s economic distress deepened as fertilizer costs consumed a growing share of farm budgets. Egypt, importing sixty percent of its wheat and dependent on imported fertilizer to grow the rest, was pushed toward the International Monetary Fund for emergency support.

China’s phosphate restrictions added a second pressure. In the first quarter of 2025, Chinese phosphate fertilizer exports dropped to 111,000 metric tons, down from a three-year average of 785,000 tons for the same period, an eighty-six percent decline. In December 2025, China’s phosphate fertilizer industry reached consensus to schedule no new export plans before August 2026. The reason is structural: China holds only five percent of global phosphate reserves despite producing over forty percent of global output, and its booming electric vehicle sector now diverts phosphate rock into lithium-iron-phosphate battery production. Each ton of LFP battery material consumes approximately 3.5 tons of phosphate rock. The food system and the energy transition now compete for the same mineral input. Nobody planned for this convergence.

The technology layer is an emerging vulnerability. Precision agriculture platforms connect tractors, planters, and harvesters to cloud-based systems that optimize planting depth, seed spacing, fertilizer application, and harvest timing. Security researchers demonstrated at Def Con that vulnerabilities in John Deere’s systems could allow remote access to equipment controls, and the FBI has warned farmers about cyber risks to digital management tools and cloud service providers. John Deere’s deputy CISO acknowledged in 2025 that state-sponsored actors and advanced persistent threats are now part of the agriculture threat landscape. A cyberattack on a major platform during planting season could disrupt food production across millions of acres. The platforms are designed for efficiency. They are not designed for contested environments.

The Iran war, now in its third week, is demonstrating how the input stack fails under stress. The closure of the Strait of Hormuz has put one-third of global fertilizer trade at risk of disruption. Nearly half of the world’s traded urea, the most widely used nitrogen fertilizer, is exported from Gulf countries via Hormuz. Qatar’s state-run QAFCO, the world’s largest urea supplier providing fourteen percent of global urea, halted production after its LNG facilities were attacked. India shut three urea plants. Bangladesh closed four of five fertilizer factories. U.S. urea import prices jumped thirty percent in a single week as the spring planting season opened. The Council on Foreign Relations warns this could become the first twenty-first-century conflict to unleash a slow-motion famine machine. Russia demonstrated the collateral version in 2022 when fertilizer disruption cascaded into global food price spikes that destabilized governments across three continents. Iran is demonstrating the direct version now.

The timing compounds the damage. The Northern Hemisphere’s spring planting season, when the largest volumes of fertilizer are purchased and applied, coincides precisely with the Hormuz closure. Vessels traveling from the Persian Gulf to the U.S. Gulf Coast take approximately thirty days, meaning supply disruptions in early March will not fully manifest until April, when planting windows close. As of 2024, Asian countries received thirty-five percent of Gulf urea exports, fifty-three percent of sulphur exports, and sixty-four percent of ammonia exports. Sulphur, an essential nutrient for plant growth and a key input in phosphate fertilizer production, is largely a byproduct of oil and gas processing. When energy shipments through Hormuz stop, sulphur output falls alongside fuel exports. The Caloric Kill Switch does not require intention. It only requires concentration.

The Convergence Gap

Agricultural economists see commodity markets. Seed industry analysts see corporate concentration. Fertilizer trade experts see geopolitical supply risk. Cybersecurity researchers see precision agriculture vulnerabilities. Biodefense analysts see agricultural bioterrorism vectors. The irregular warfare community sees gray zone competition tools in isolation. Nobody has converged seed supply monopolization, fertilizer dependency, agricultural cyber vulnerability, food processing fragility, and agrobiodiversity loss into a single irregular warfare operational concept with a deterrence framework.

The bureaucratic fragmentation mirrors the food system itself. The U.S. Department of Agriculture monitors commodity markets. The Department of Commerce oversees seed industry mergers. The Department of Energy competes for the same phosphate going into batteries. The Cybersecurity and Infrastructure Security Agency classifies food and agriculture as critical infrastructure but has issued no mandatory cybersecurity standards for precision agriculture platforms. The International Fertilizer Association tracks global supply. The World Trade Organization governs export restrictions. No single institution sees the input stack as a unified attack surface.

The 2022 fertilizer crisis and the 2026 Hormuz closure are not separate events. They are two demonstrations of the same structural vulnerability, separated by four years and zero structural reforms. In 2022, the disruption was collateral: Russia’s war in Ukraine was not designed as a food weapon, but the concentrated architecture of the fertilizer market converted a regional conflict into a global caloric shock. In 2026, the disruption is more direct: the Hormuz closure physically blocks the export of fertilizers from the countries that the world turned to after 2022 to replace Russian supply. The Gulf states that absorbed the demand shift, Qatar, Saudi Arabia, Oman, the UAE, are now themselves inside a war zone. The backup became the target.

The Iran war has exposed precisely this fragmentation at a global scale. Energy analysts track oil prices. Fertilizer economists track urea and DAP. Agricultural ministries track planting schedules. Security analysts track Hormuz shipping. No single institution is tracking the convergent effect: that the same chokepoint closure has simultaneously cut off natural gas feedstock for fertilizer production, halted urea exports from the world’s largest supplier, shut down fertilizer factories in three countries, spiked input costs for American farmers during their most critical purchasing window, and set the conditions for reduced crop yields that will not become visible until harvest. The adversary did not design a caloric weapon. The architecture of the input stack produced one. The absence of a unified defense framework ensured nobody saw it coming as a single system failure.

Naming the Weapon: The Caloric Kill Switch

I propose the term the Caloric Kill Switch to describe the convergent capability to disrupt an adversary’s food production through simultaneous exploitation of seed supply concentration, fertilizer dependency, agricultural technology vulnerability, and processing infrastructure fragility. The Caloric Kill Switch is agrarian coercion: the weaponization of food system inputs to degrade population nutrition, economic stability, and social cohesion without firing a shot.

The switch operates through compounding dependencies. Restrict fertilizer exports and crop yields fall. Manipulate seed supply and planting diversity collapses. Compromise precision agriculture platforms and operational efficiency degrades. Divert phosphate into battery production and food competes with energy for the same mineral. Each layer reinforces the others. The system is not resilient. It is optimized for efficiency, and efficiency is the enemy of resilience. The median between disruption and famine is one growing season.

The Doctrine: Five Pillars of Caloric Sovereignty

First Pillar: The Caloric Vulnerability Index. A standardized metric quantifying national food system dependency as strategic vulnerability. Measured by seed sourcing concentration, fertilizer import dependency, agricultural technology platform exposure, strategic grain reserve levels, and input substitution timelines. Briefed alongside national security indicators, not agricultural statistics. The CSIS analysis of the Iran war fertilizer shock illustrates why: the cost of one ton of urea rose from the equivalent of seventy-five bushels of corn in December 2025 to one hundred twenty-six bushels by March 2026, a seventy-seven percent increase that no agricultural forecast anticipated because no agricultural forecast incorporates chokepoint warfare.

Second Pillar: Seed Sovereignty. Publicly funded seed banks and breeding programs that maintain genetic diversity outside corporate control. Mandatory open-pollinated variety preservation. Investment in public-domain seed development for strategic crops. Four companies should not hold intellectual property control over the caloric foundation of eight billion lives. The concentration ratio for the top four seed firms exceeds sixty percent in most major crop categories, well above the forty percent threshold at which economists consider market distortions likely. In the United States, three firms own ninety-five percent of patents for GM corn, seventy-eight percent for GM soybeans, and ninety-three percent for GM canola. Meanwhile, the FAO estimates that seventy-five percent of crop genetic diversity has been lost since 1900 as commercial agriculture converges on an ever-narrower set of proprietary varieties optimized for a stable climate that no longer exists. Seed sovereignty is not nostalgia. It is redundancy. And redundancy is the only architecture that survives disruption.

Third Pillar: Fertilizer as Critical Infrastructure. Domestic fertilizer production capacity treated as critical national infrastructure under defense authority. Strategic fertilizer reserves maintained and rotated on the petroleum reserve model. Allied procurement agreements that diversify sourcing away from adversary-controlled deposits. The phosphate competition between food and battery production must be managed as a national security allocation, not a market outcome. There are no substitutes for phosphorus in agriculture, a fact the U.S. Geological Survey states plainly in its 2025 mineral commodity summary. Every calorie consumed by every human on the planet depends on a mineral whose production is controlled by four countries, whose reserves are controlled by one, and whose allocation between food and electric vehicles is decided by no government. China’s December 2025 decision to suspend phosphate exports through August 2026 was a sovereign resource decision that will cascade through every importing nation’s food system. Allied governments received no advance warning and have no mechanism to respond collectively.

Fourth Pillar: Agricultural Cyber Resilience. Mandatory cybersecurity standards for precision agriculture platforms operating above a defined acreage threshold. Offline operational capability requirements for GPS-guided machinery. Air-gapped backup systems for critical planting and harvest data. John Deere alone has invested in a cybersecurity team of more than 230 professionals and a bug bounty program that has paid out over 1.5 million dollars since 2022. But cybersecurity in agriculture remains voluntary. No cloud dependency should be capable of disabling a nation’s food production. The platforms that optimize American agriculture were designed for efficiency in a permissive environment. They have not been tested against a state-sponsored adversary operating during planting season.

Fifth Pillar: The Allied Food Security Compact. Multinational agreements among trusted allies that create mutual food supply guarantees, coordinated strategic reserves, and joint response mechanisms for food system disruption. Binding commitments with enforcement mechanisms, not aspirational declarations. Treaties with teeth. In 2022, when Russia’s invasion disrupted fertilizer and grain flows, the international response was a cascade of unilateral export bans: Serbia stopped exporting wheat, corn, flour, and cooking oil. Argentina, India, Indonesia, and Turkey took similar measures. Each country acted rationally to protect its own population. The collective effect was to amplify the crisis, converting a supply shock into a price spiral that hit the poorest nations hardest. An Allied Food Security Compact would replace panic-driven unilateral bans with pre-negotiated mutual obligations, the caloric equivalent of NATO’s Article 5: an attack on one nation’s food system triggers a collective response from all. The seed in a farmer’s field was designed by one of four companies. The fertilizer that feeds it passed through a chokepoint that a single adversary can close. The tractor that plants it is connected to a cloud server that a state actor could compromise before sunrise.

The Caloric Kill Switch is not hypothetical. It is the architecture of the global food system, waiting for someone to pull it.

RESONANCE

Al Jazeera. (2026). “Not Just Energy: How the Iran War Could Trigger a Global Food Crisis.” Al Jazeera. https://www.aljazeera.com/economy/2026/3/18/not-just-energy-how-the-iran-war-could-trigger-a-global-food-crisis. Summary: Reports that nearly half the world’s traded urea passes through the Strait of Hormuz, documents the shutdown of Qatar’s QAFCO urea plant and cascading factory closures in India and Bangladesh, and assesses the forty-percent surge in Middle East urea export prices.

Agri-Pulse. (2025). “Chinese Phosphate Exports Plummet, Dashing Hope for Price Relief.” Agri-Pulse. https://www.agri-pulse.com/articles/22817-chinese-phosphate-exports-plummet-dashing-hope-for-price-relief. Summary: Documents the eighty-six percent drop in Chinese phosphate exports in Q1 2025, the competition between agriculture and electric vehicle battery production for phosphate rock, and the expectation of continued export restrictions.

Bank Info Security. (2021). “Flaws in John Deere Systems Show Agriculture’s Cyber Risk.” Bank Info Security. https://www.bankinfosecurity.com/flaws-in-john-deere-systems-show-agricultures-cyber-risk-a-17240. Summary: Reports security researcher findings presented at Def Con demonstrating root access vulnerabilities in John Deere’s Operations Center, and the FBI warning to farmers about cyber risks to agricultural technology platforms.

CNBC. (2026). “Food Prices Could Rise as Iran Conflict Disrupts Fertilizer Supply Chain.” CNBC. https://www.cnbc.com/2026/03/11/iran-news-food-prices-could-rise-due-to-fertilizer-shortages.html. Summary: Reports the thirty-percent jump in U.S. urea import prices in a single week following the Hormuz closure, with one-third of globally traded fertilizer passing through the strait during the Northern Hemisphere’s critical spring planting window.

CFR. (2026). “The Iran War’s Hidden Front: Food, Water, and Fertilizer.” Council on Foreign Relations. https://www.cfr.org/articles/the-iran-wars-hidden-front-food-water-and-fertilizer. Summary: Assesses the Iran war as a potential twenty-first-century famine machine, documenting the convergence of fertilizer disruption, climate stress, depleted grain reserves, and debt-constrained governments transforming a regional military conflict into a global food security crisis.

CSIS. (2026). “Chokepoint: How the War with Iran Threatens Global Food Security.” Center for Strategic and International Studies. https://www.csis.org/analysis/chokepoint-how-war-iran-threatens-global-food-security. Summary: Comprehensive analysis of the Hormuz closure impact on nitrogen, phosphate, and potash markets, including the seventy-seven percent urea price increase from December 2025 to March 2026 and the spring planting timing vulnerability.

GRAIN and ETC Group. (2025). “Top 10 Agribusiness Giants: Corporate Concentration in Food and Farming in 2025.” GRAIN and ETC Group. https://www.ohchr.org/sites/default/files/documents/issues/food/cfis/global-food-system/subm-concentration-corporate-power-cso-31-grain-etc-group.pdf. Summary: Reports that Bayer, Corteva, Syngenta, and BASF control fifty-six percent of the global commercial seed market and sixty-one percent of the pesticides market, with detailed revenue analysis and corporate integration trends.

Help Net Security. (2025). “Protecting Farms from Hackers: A Q&A with John Deere’s Deputy CISO.” Help Net Security. https://www.helpnetsecurity.com/2025/08/26/carl-kubalsky-john-deere-smart-agriculture-cybersecurity/. Summary: John Deere’s Deputy CISO acknowledges state-sponsored actors as part of the agriculture threat landscape and describes the company’s 230-person cybersecurity team and layered defense approach.

Land and Climate Review. (2025). “How a Few Giant Companies Came to Dominate Global Food.” Land and Climate Review (MIT Press excerpt). https://landclimate.org/how-a-few-giant-companies-came-to-dominate-global-food/. Summary: Excerpt from the MIT Press book documenting concentration ratios: Bayer and Corteva controlling seventy-two percent of U.S. corn seed and sixty-six percent of soybean seed, with three firms owning ninety-five percent of GM corn patents.

SunSirs. (2025). “The Logic Behind China’s Phosphate Fertilizer Export Suspension.” SunSirs. https://www.sunsirs.com/uk/detail_news-28842.html Summary Documents China holding five percent of global phosphate reserves while producing over forty percent of output, the December 2025 industry consensus to suspend exports through August 2026, and the structural competition between fertilizer and LFP battery production.

The Conversation. (2026). “How the Iran War Could Create a ‘Fertiliser Shock.’” The Conversation. https://theconversation.com/how-the-iran-war-could-create-a-fertiliser-shock-an-often-ignored-global-risk-to-food-prices-and-farming-277552. Summary: Explains the structural link between natural gas, ammonia production, and nitrogen fertilizers passing through Hormuz, and the cascading impact of supply disruption on sub-Saharan Africa where fertilizer use is already critically low.

USDA Economic Research Service. (2023). “Global Fertilizer Market Challenged by Russia’s Invasion of Ukraine.” USDA ERS. https://www.ers.usda.gov/amber-waves/2023/september/global-fertilizer-market-challenged-by-russia-s-invasion-of-ukraine. Summary: Reports Russia and Belarus providing forty percent of global potash exports, Russia accounting for sixteen percent of urea and twelve percent of phosphate exports, and the fifty-three percent potash price surge from January to April 2022.