The Weight of Absence

Zero is a number that tells a story. In August 1964, zero was the number of torpedoes that struck the USS Maddox on the second night in the Gulf of Tonkin. That zero was classified for decades. By the time it surfaced, the war it justified had ended and fifty-eight thousand names had been carved into black granite.

This is a story about a different zero, and a different way of finding it. Not through leaked cables or declassified archives, but through a formal system that treats official claims as testable hypotheses. The method is new. Whether the zero it finds means what the Tonkin zero meant is a question the method cannot answer. That judgment belongs to readers, not frameworks.

Since September 2025, the United States has conducted more than thirty military strikes on vessels in the Caribbean, killing over one hundred people. The stated justification is drug interdiction. The claims are specific: these boats carry fentanyl bound for American shores; each one destroyed saves twenty-five thousand lives. These claims are not rhetoric. They are assertions with logical consequences. And logical consequences can be tested.

Claims as Computable Objects

The traditional approach to accountability journalism is narrative: gather quotes, identify contradictions, present them to readers, let them draw conclusions. This works, but it does not scale, and it leaves the epistemics implicit. Why should we believe the contradiction matters? How unlikely is it under innocent explanations?

A different approach: treat claims as first-class objects (things the system can store, query, and reason about) in a formal ontology. Not as strings of text to be quoted, but as structured data with properties, relationships, and inferential consequences.

@prefix acc: <http://accountability.org/ontology#> .
@prefix strike: <http://accountability.org/data/venezuela/> .

strike:claim_trump_destination a acc:FactualClaim ;
    acc:claimant strike:donald_trump ;
    acc:claimDate "2025-09-03" ;
    acc:claimContent "Boat was headed to the United States" ;
    acc:claimAbout strike:strike_2025_09_02 .

strike:claim_rubio_destination a acc:FactualClaim ;
    acc:claimant strike:marco_rubio ;
    acc:claimDate "2025-09-04" ;
    acc:claimContent "Probably headed to Trinidad" ;
    acc:claimAbout strike:strike_2025_09_02 .

strike:claim_bradley_destination a acc:FactualClaim ;
    acc:claimant strike:admiral_bradley ;
    acc:claimDate "2025-09-05" ;
    acc:claimContent "Headed for Suriname" ;
    acc:claimAbout strike:strike_2025_09_02 .

strike:contradiction_destinations a acc:Contradiction ;
    acc:contradictionType acc:MutuallyExclusive ;
    acc:involvesClaimSet ( 
        strike:claim_trump_destination 
        strike:claim_rubio_destination 
        strike:claim_bradley_destination 
    ) .

When claims are modeled this way, contradictions are not rhetorical observations. They are structural relationships that can be queried, counted, and compared against baselines. The question shifts from "does this seem inconsistent?" to "what is the probability of observing this level of inconsistency if officials are accurately describing events?"

More importantly, the ontology can reason. SHACL (Shapes Constraint Language) defines validation rules that fire automatically when data violates constraints. The following shape detects when multiple claims about the same event assert mutually exclusive values:

@prefix sh: <http://www.w3.org/ns/shacl#> .
@prefix acc: <http://accountability.org/ontology#> .

acc:ClaimConsistencyShape a sh:NodeShape ;
    sh:targetClass acc:Strike ;
    sh:sparql [
        sh:message "Contradictory destination claims detected: {?claim1} vs {?claim2}" ;
        sh:select """
            SELECT $this ?claim1 ?claim2 ?dest1 ?dest2
            WHERE {
                ?claim1 acc:claimAbout $this ;
                        acc:claimContent ?dest1 ;
                        acc:claimType acc:DestinationClaim .
                ?claim2 acc:claimAbout $this ;
                        acc:claimContent ?dest2 ;
                        acc:claimType acc:DestinationClaim .
                FILTER (?claim1 != ?claim2 && ?dest1 != ?dest2)
            }
        """
    ] .

When this shape validates against the Venezuela strikes data, it fires on strike_2025_09_02, detecting the Trump/Rubio/Bradley contradiction automatically. The ontology does not merely store claims; it reasons about their consistency.

# Validation Report
sh:conforms false ;
sh:result [
    sh:focusNode strike:strike_2025_09_02 ;
    sh:resultMessage "Contradictory destination claims detected: 
        claim_trump_destination ('United States') vs 
        claim_rubio_destination ('Trinidad') vs 
        claim_bradley_destination ('Suriname')" ;
    sh:resultSeverity sh:Violation
] .

What the Ontology Discovers

Detecting contradictions is useful, but a database could do that. The ontology does something a database cannot: it reasons about what claims imply and catches contradictions humans might miss.

Consider Admiral Bradley's statement that the boat was "headed for Suriname." A reader might note this differs from Trump's "United States" and Rubio's "Trinidad." Three destinations, at least two wrong. But there is a deeper problem, one that requires geographic knowledge to see.

Suriname is a transit node on the cocaine corridor to Europe. Drugs passing through Suriname go to West Africa, then to the Netherlands and Spain, then to European markets. This is documented in UNODC reports and DEA assessments. If Bradley is right about the destination, the boat was Europe-bound, not America-bound.

The ontology encodes this geographic knowledge. When Bradley's claim enters the system, the reasoner automatically infers: Suriname → Europe corridor → European markets → not the population the operation claims to protect. No human analysis required.

GOAL-ROUTE INCOMPATIBILITY DETECTED:

Admiral Alvin Bradley claimed the boat was headed for Suriname.

The ontology knows:
  • Suriname is a transit node on the Suriname-Europe Corridor
  • The Suriname-Europe Corridor leads to Europe
  • The operation claims to protect United States

INFERENCE: This claim implies the boat was Europe-bound,
           not United States-bound.

CONCLUSION: The destination claim contradicts the stated
            justification ("protecting Americans").

The same inference fires on Rubio's "Trinidad" claim. Trinidad feeds Caribbean markets, not the United States. Two of three destination claims, when traced through their trafficking routes, point away from America entirely.

A journalist covering this story would need to know Caribbean drug geography, remember which transit nodes feed which markets, and manually connect each claim to the stated justification. The ontology does this automatically, across any number of claims, in milliseconds.

This is what formal ontology provides that simpler approaches cannot: not just storage, but inference. The system reasons about implications and surfaces contradictions that require domain knowledge to detect. When the next hundred claims enter the system, the same reasoning applies without additional human effort.

Five Tests for Truth

The framework rests on a simple premise: legitimate operations produce characteristic signatures. When an operation is what it claims to be, certain patterns hold. When those patterns break, it tells you the official story isn't the real reason.

Each invariant becomes a test. Each test asks a simple question: if the official story were true, would we expect to see what we're seeing? If the answer is "almost certainly not," the test fails.

This is not journalism with numbers attached. It is a formal framework that anyone can run, audit, and extend.

Test I: Where Are the Drugs?

The first test is the simplest. If you are interdicting drug shipments, you should occasionally recover drugs. The Coast Guard's FY2024 report to the Department of Homeland Security recorded drug disruptions in 91 out of 125 vessel interdictions, a rate of 73 percent. This is what successful interdiction looks like: you stop the vessel, you find the contraband, you record it.

SELECT (COUNT(?strike) AS ?total)
       (SUM(IF(?hasEvidence, 1, 0)) AS ?withEvidence)
WHERE {
    ?strike a acc:Strike ;
            acc:operation strike:venezuela_2025 .
    OPTIONAL { ?strike acc:hasDocumentedDrugEvidence ?hasEvidence }
}

Zero is not merely low. If these were real interdictions, the chance of seeing this outcome is less than one in ten thousand. The first pattern breaks.

But the 73% baseline matters. What if it's wrong?

The Coast Guard's own report gives 73%. But suppose the true rate for Caribbean operations is lower. Suppose intelligence is worse, vessels are faster, conditions are harder. At 50%, the probability of zero in thirteen is still less than one in eight thousand. At 30%, it is one in thirty. At 20%, it is one in fourteen. The baseline would need to fall below 15% before zero in thirteen becomes unsurprising. No published interdiction data supports a rate that low.

The test is robust to reasonable baseline uncertainty. Even pessimistic assumptions produce significant results.

Test II: Three Officials, Three Stories

The ontology models claims as nodes and contradictions as edges. When three officials describe the same event with mutually exclusive facts, the graph registers a contradiction cluster. The question becomes quantitative: what contradiction rate should we expect from officials accurately describing real events?

These destinations are mutually exclusive. At least two of the three statements must be false. When officials describe real events, they occasionally miscommunicate, maybe five percent of the time. A hundred percent contradiction rate on the first strike's destination is not miscommunication.

The Suriname detail matters substantively, not just statistically. Suriname is a transit corridor for cocaine destined for Europe, not the United States. If Admiral Bradley's account is accurate, the "protecting Americans" justification does not apply to this strike.

Test III: The Impossible Number

Some tests are not statistical. They are arithmetical.

President Trump has claimed that each boat destroyed saves twenty-five thousand American lives. With thirty boats struck, the implied total is 750,000 lives saved. Approximately 77,000 Americans die from drug overdoses annually, from all drugs, from all sources worldwide.

The claim exceeds the number of deaths that exist to be prevented by nearly ten times. This is not a statistical argument. It is arithmetic. The numbers are impossible.

Test IV: The Shifting Story

Legitimate operations have stable justifications because the stated goal is the actual goal. When rationales shift, it suggests the original framing was instrumental rather than authentic. The ontology tracks justification statements over time and flags semantic drift.

"He wants to keep on blowing boats up until Maduro cries uncle." Susie Wiles, White House Chief of Staff, November 2025

This statement does not describe drug policy. It describes a coercion campaign against a foreign government. The goal is not interdiction but regime pressure.

"His days are numbered." President Trump, referring to Nicolás Maduro, December 2025

The ontology registers two documented shifts from "stopping drugs" to explicit regime-change rhetoric. When the real goal is the stated goal, you don't need to change the story. The justification drifted because the original framing was cover, not cause.

Test V: The Wrong Country

The stated goal is stopping fentanyl from reaching Americans. This raises a question the other tests do not ask: where does fentanyl actually come from?

The answer is not Venezuela. According to the State Department's 2025 International Narcotics Control Strategy Report, Mexico is "the most significant source of illicit fentanyl and fentanyl analogues significantly affecting the United States." The DEA's 2025 National Drug Threat Assessment identifies two Mexican cartels, Sinaloa and Jalisco New Generation, as the primary suppliers. Chinese companies provide precursor chemicals. Mexican labs synthesize the drug. Mexican networks smuggle it across the southern border.

Venezuela does not appear in the fentanyl supply chain. Not as a source. Not as a manufacturing hub. Not as a significant transit point. The country is mentioned in U.S. drug reports primarily in connection with cocaine transshipment to Europe, not fentanyl trafficking to America.

If you want to stop fentanyl from killing Americans, you interdict the supply chain that kills Americans. That supply chain runs through Mexico. The strikes target Venezuela. The pattern does not match the goal.

This is not a statistical argument. It is a geographic one. The fentanyl crisis and the Venezuela strikes exist in different supply chains, different trafficking networks, different parts of the hemisphere. Striking Venezuelan boats to stop fentanyl is like patrolling the Atlantic to stop Pacific typhoons. The intervention and the problem do not occupy the same space.

The Control Case: What Real Interdiction Looks Like

A framework that only produces one answer is not a framework. It is an argument dressed in methodology. To demonstrate that these tests distinguish real operations from pretextual ones, we need cases where the tests pass.

We have two such cases, running simultaneously with the Venezuela strikes, in the same waters, under the same administration.

Operation Pacific Viper launched in August 2025 as a Coast Guard surge in the Eastern Pacific. Same stated goal as the Venezuela strikes: stopping drug traffickers. By December, it had conducted 34+ interdictions, seized over 150,000 pounds of cocaine, and detained 86 suspects. Evidence rate: approximately 100%. Every interdiction produced documented seizures, photographed, weighed, and published.

Caribbean Coast Guard interdictions continued throughout 2025, operating in the same waters as the missile strikes. CBP Air and Marine Operations and Coast Guard cutters based in Puerto Rico seized thousands of pounds of cocaine in dozens of operations: 300 pounds near Dorado in January, 1,155 pounds south of Cabo Rojo in June, 3,175 pounds near the Virgin Islands in an international operation, nearly 4,000 pounds near Cabo Rojo in July. In every case: drugs seized, suspects arrested, evidence documented.

The 73% baseline comes from the Coast Guard's own FY2024 fiscal year report, which recorded drug disruptions in 91 out of 125 vessel interdictions. This is the official statistic, cited by Senator Rand Paul in congressional testimony and verified by PolitiFact. The 27% without drugs includes cases where suspects jettisoned cargo overboard, intelligence was incorrect, or vessels were intercepted for other violations.

Run the tests on these operations and watch them pass.

Test I: Evidence. Pacific Viper: 150,000+ pounds seized across 34 interdictions. Caribbean operations: thousands of pounds in documented seizures throughout 2025. The Hamilton offloaded 76,000 pounds in a single port call. The Munro seized 20,000 pounds in one interdiction. Evidence rate: near 100%.

Test II: Consistency. When Pacific Viper officials describe an interdiction, they agree on the facts. The cocaine came from a go-fast vessel southeast of the Galápagos. The suspects were Ecuadorian nationals. No contradictions. No three officials giving three different destinations for the same boat.

Test III: Plausibility. Pacific Viper does not claim to save 25,000 lives per boat. It claims to seize cocaine. The amounts are large but verifiable: 150,000 pounds across 34 interdictions is roughly 4,400 pounds per interdiction, consistent with known go-fast vessel capacities. The Coast Guard's own estimate: 193 million potentially lethal doses removed. The math works.

Test IV: Stability. The justification for Pacific Viper has remained consistent: drug interdiction. Admiral Kevin Lunday calls it "taking the fight to criminal networks." Rear Admiral Jeffrey Novak describes it as "dismantling narco-terrorist networks." No official has suggested the real purpose is to pressure a foreign government or make anyone "cry uncle."

This is what drug interdiction looks like when drug interdiction is the actual goal. You stop the boat. You find the drugs. You arrest the crew. You document everything. You do not blow the boat out of the water and claim, without evidence, that it was carrying fentanyl.

The contrast could not be starker. Three operations, same administration, same time period, same stated goal. Two produce hundreds of thousands of pounds of seized cocaine and hundreds of arrests. One produces zero documented seizures and over 100 dead. Two pass every test. One fails every test.

As Secretary of State Marco Rubio said after the first strike: the President "could have ordered the alleged drug boat be intercepted. Instead of interdicting it, on the president's orders, we blew it up."

Why blow it up? The framework cannot answer that question. But political science offers context for understanding what we are seeing.

Why States Kill

Scholars have studied state violence for decades. Three findings matter here.

First: targeted killing does not work. Audrey Kurth Cronin studied over 400 terrorist and insurgent groups to answer a simple question: what actually ends these organizations? Her answer: almost never targeted killing. Capturing or killing leaders sometimes accelerates collapse, but only when the organization is already weak and isolated. Against networked groups with popular support, decapitation fails. It often backfires, creating martyrs and increasing recruitment. Cronin is direct: the belief that killing leaders will destroy organizations is "tinged with emotion, not dispassionate analysis."

This matters because drug trafficking networks are exactly the kind of decentralized, adaptive organizations that survive decapitation. Kill a boat crew and another takes their place. The Coast Guard knows this. That is why interdiction focuses on seizing drugs, not killing smugglers. Seized cocaine cannot be replaced by the next crew. Dead smugglers can.

Second: democracies kill civilians when it serves their interests. Alexander Downes studied every major war of the twentieth century to understand why governments target civilians. His conclusion upends a comfortable assumption: democracies are not more restrained than dictatorships. When democracies face pressure to win, when wars drag on, when leaders want results without costly ground campaigns, they bomb cities, starve populations, and kill noncombatants. The decision depends on strategic incentives, not regime type.

The implication is uncomfortable. The question is not whether the United States would kill civilians under cover of counternarcotics. The question is whether strategic incentives exist to do so. If the actual goal is pressuring Maduro, and if dead Venezuelan boats send that message more clearly than seized cocaine, then Downes's research predicts exactly what we observe: violence that serves the real goal while official statements invoke the stated one.

Third: legal justifications follow power, not the reverse. The Finkelstein volume on targeted killing traces how the war on terror transformed extrajudicial execution into policy. Before September 11, killing named individuals without trial was assassination, prohibited by executive order. After September 11, it became "targeted killing," authorized by legal memoranda that redefined battlefields to include any country where a target might be found. The law did not constrain the killing. The killing reshaped the law.

The Venezuela strikes fit this pattern. There is no declared war. There is no congressional authorization for military strikes against drug traffickers. There is no judicial process. There are only assertions: these boats carried drugs, these people were traffickers, these deaths protect Americans. The assertions arrive after the missiles. The legal framework, such as it is, consists of whatever justification the administration offers this week.

None of this proves the Venezuela strikes are illegitimate. Political science describes patterns, not individual cases. But the patterns are clear: targeted killing rarely achieves counterterrorism goals, democracies kill civilians when strategically useful, and legal constraints bend to accommodate state violence. The framework in this analysis does not discover these patterns. It tests whether a specific operation fits them.

The Aggregate

The individual tests combine, but combining them honestly requires acknowledging what they do and do not establish.

Each test asks a narrow question: how likely is this observation if the stated explanation is true? Low probability does not mean the stated explanation is false. It means the stated explanation does not predict what we observe. There may be other explanations, ones I have not modeled, that predict the data better while remaining consistent with official good faith.

There is also a dependence problem. The tests assume independence, but they may not be independent. If one failure causes another (contradictions arise because there is no evidence, which arose because the goal was never interdiction), then multiplying p-values overstates the combined significance. I do not have a clean solution to this. The Fisher combination below is illustrative, not definitive.

What the tests do establish: the observed patterns are highly unlikely under the specific hypothesis that this operation is a conventional drug interdiction mission. That hypothesis predicts evidence, consistency, stability, and plausible claims. We observe none of these. The mismatch is severe.

All five tests fail. Four are statistical: the patterns observed are highly improbable under the stated explanation. One is geographic: the intervention targets the wrong supply chain entirely. The failures are individually significant and collectively striking. But "striking" is not "proof." The tests falsify a specific hypothesis: that this operation functions like a conventional drug interdiction mission targeting fentanyl. They do not prove what the operation actually is. That would require evidence I do not have.

Predictions

A framework that cannot be wrong is not a framework. It is advocacy. The legitimacy signature theory makes predictions that can be tested against future evidence.

If the official story is true (these are drug interdictions), we should expect:

• Future strikes will begin producing documented drug evidence as operational procedures improve. Evidence rate will converge toward the 73% baseline.
• Officials will align on facts. The contradiction rate will decrease.
• Justifications will stabilize around drug interdiction, not regime change.
• Quantitative claims will become plausible, or officials will stop making them.

If the official story is not true (the real purpose is regime pressure), we should expect:

• Evidence rate will remain near zero. No documented seizures will emerge.
• Contradictions will continue or increase as officials improvise justifications.
• Justifications will continue shifting toward explicit regime-change rhetoric.
• Strikes will concentrate on Venezuelan-flagged or Venezuelan-origin vessels regardless of drug intelligence.
• The operation will end or change character if Maduro's government falls or makes concessions, even if drug flows continue.

These predictions are falsifiable. If documented drug seizures begin appearing in future strikes at rates approaching the baseline, the framework's conclusion weakens. If the evidence rate remains at zero across 30, 50, 100 strikes, the conclusion strengthens. The data will tell us.

What This Is and Isn't

This framework tests a hypothesis. The hypothesis is: "This operation functions like documented drug interdiction missions." The tests check whether observed patterns match that hypothesis. They do not.

What the framework does not do: prove intent, establish what the operation actually is, or adjudicate legality. When I say the patterns are inconsistent with the stated goal, I mean exactly that. I do not mean officials are lying (they may believe what they say). I do not mean the operation is criminal (that is a legal question I cannot answer). I do not mean I know the real purpose (I am inferring from pattern mismatch, not from evidence of alternative goals).

A hostile reader might object: perhaps intelligence is classified, perhaps evidence exists but is not public, perhaps the operation has features that explain the anomalies. These objections are valid. I cannot rule them out. What I can say is that the public record, as it stands, does not support the stated explanation. If the administration has evidence that would change this assessment, they have not produced it.

The framework's value is not certainty. It is structure. The ontology makes claims queryable. The tests make significance calculable. The baselines make assumptions explicit. Anyone can audit the reasoning, challenge the assumptions, extend the data, or apply the framework elsewhere. That is different from journalism, which presents conclusions. This presents a machine for reaching them.

• • •

In 1964, zero torpedoes struck the Maddox on the second night. That fact took fifty years to establish through declassification and archival research. The war happened anyway.

In 2025, zero boats produced documented drug evidence. That fact is knowable now because the logic is explicit and the tests are public. What the fact means is less certain. Perhaps evidence exists but is classified. Perhaps the operation has features that explain the anomalies. Perhaps I am wrong.

What the framework provides is not certainty but visibility. The patterns are documented. The tests are reproducible. The assumptions are stated. If the patterns change, if evidence emerges, if the baselines are wrong, the framework will register it. That is a different kind of accountability than waiting fifty years for archives to open.

Whether it matters depends on what we do with it.