The Twelve-Day War in Seven Signals

The June 2025 war between Israel, Iran and the United States lasted twelve days and was narrated in real time by more phones, cameras and correspondents than any previous conflict in the region. And yet the most reliable account of what actually happened was not written by any of them. It was recorded, continuously and without opinion, by machines that were never built to cover a war: the seismometers of the global earthquake network, the monitors watching internet traffic, the trackers logging every aircraft transponder, the transceivers on the world's oil tankers, the commodity tapes, the radiation sensors, and the satellites overhead.

These instruments share the one quality that makes them the closest thing to an impartial witness this war had. They do not take sides, they cannot spin, and they were already running before the first strike. A seismograph does not know there is a war on; it reports ground motion. An outage monitor does not know why a country went dark; it reports reachability. A flight tracker does not know a mission is secret; it reports whatever transmits. Read carefully, and cross-referenced against one another, these seven signals corroborate the public story in some places, complicate it in others, and flatly contradict it in a few. Where they contradict it, the machines are usually right, because a catalog entry is a measurement and a viral number is a rumor.

This briefing reads the Twelve-Day War through those seven signals. It is not a summary of the news. It is an audit of the news against the data, and every figure below is linked to the primary record it came from. Blackwing's founding principle is that accuracy is the product: where a number is uncertain we say so, where a claim is disputed we tag it, and where the instruments and the headlines diverge we show the divergence rather than smooth it away.

The seven instruments each saw a different face of the same twelve days. Seismographs were asked whether the bunker-busters shook the earth. Outage monitors watched a nation of ninety million switch itself off in stages. Flight trackers caught a deception operation and a force build-up in the same open feed. Ship transceivers logged a thousand vessels a day being electronically lied to in the world's most important oil chokepoint. The markets priced the war more coldly and more accurately than most of the commentary. Radiation networks measured the disaster that everyone feared and that never came. And satellites became the referee when governments disputed the damage. Taken together they do not just retell the war. They correct it.

A note on method

Reading a war through its instruments is a discipline before it is a story, and the discipline is what makes the story trustworthy. Four rules govern everything below.

The first is cross-reference over repetition. In an information war, a claim can be echoed by a hundred outlets and still trace back to a single unverified post. What matters is not how many times a number appears but how many independent instruments recorded it. When the American and European seismic catalogs, built by different institutions from different stations, return the same empty set at Fordow, that agreement carries more weight than any headline, because the two networks share no incentive and no source. Corroboration by independent measurement is the strongest evidence this field offers, and contradiction between independent measurements is the most important thing a briefing can surface.

The second is primary sources over aggregators. Every figure in this briefing links to the catalog, agency, exchange or research group that produced it, not to a summary of it. A USGS query, an IODA country signal and a FRED price series are things you can check yourself. A number repeated on a cable chyron is not. Where the only available source is a single claim, we say so in the text rather than laundering it into apparent fact.

The third is calibrated uncertainty. Not everything here is equally certain, and pretending otherwise is its own kind of lie. Figures are tagged the way an intelligence desk tags them: confirmed where the primary record is solid, reported where a credible source has stated it but it is not independently verified, and disputed where sources genuinely conflict. Casualty totals are given as ranges rather than false-precise single numbers, because the honest state of knowledge is a range. A clean number where the truth is a range is the most common way a war is misreported.

The fourth is scope discipline, and it is the least obvious. This briefing is strictly about the June 2025 Twelve-Day War. A separate, later US-Iran confrontation in 2026 is constantly conflated with it in search results and secondary coverage, dragging in a longer 2026 internet blackout, a 2026 shipping-insurance crisis and other events that did not happen in these twelve days. Keeping the two apart is not pedantry. It is the difference between describing what occurred and describing an average of two different wars. Where a source blurs them, we hold the line on the 2025 record.

None of these instruments is omniscient, and each section below is candid about what its signal cannot see. A seismograph cannot look inside a centrifuge hall. An outage monitor cannot read intent. A flight tracker goes blind over open ocean. The account that survives is the one built where the blind spots of one instrument are covered by the strengths of another. With that method stated, here is the war, signal by signal.

Seismic: the earthquake that never happened

Within hours of Operation Midnight Hammer, a single number went around the world: magnitude 4.3. Twelve GBU-57 Massive Ordnance Penetrators, each about 30,000 pounds, had, the story went, punched into the mountain above Fordow and rung the Earth like a bell, a "bunker-buster earthquake" so violent that global seismographs caught it. The framing was irresistible because it seemed to answer the only question that mattered. If the ground shook that hard, the argument ran, the underground enrichment hall must have been obliterated. Some accounts pushed further, floating a "magnitude 5.1 near Fordow" and hinting the shaking was so anomalous it might even mask an accidental nuclear detonation. The number felt like evidence. It was not.

What the machines actually recorded is a null. Blackwing's live query of the USGS earthquake catalog for the Fordow box (34.4 to 35.4 North, 50.4 to 51.5 East) across the entire war window, 13 to 24 June, returns zero cataloged events. Not a small one. None. The 22 June GBU-57 strikes did not register teleseismically at all. This is not the USGS being slow or selective. The independent European-Mediterranean network at EMSC returns the same null at Fordow. Its only nearby cataloged events in the window were natural M3.2 and M3.3 quakes 130 to 150 km away, ordinary background tectonics with nothing to do with the strikes. Two of the world's major seismic catalogs, built by different institutions on different continents from overlapping but distinct station networks, independently agree that no earthquake happened at Fordow. When two adversarial instruments return the same empty set, that agreement is the finding.

The strikes were real, and they did shake the ground. They simply stayed local. Mehdi Zare of Iran's Institute of Earthquake Engineering and Seismology (IIEES) reported that Iranian stations recorded more than 1,500 explosion-induced tremors across the twelve days. The blasts near Fordow, Natanz and Isfahan reached only about magnitude 2. That is the true seismic signature of this war: not one M4.3 shock, but roughly 1,500 tiny thumps near the noise floor, visible only to seismometers a short distance away and invisible to the global network that logs earthquakes for the entire planet. The viral figure and the recorded reality are not two readings of one event. They are a rumor and a measurement, and only one of them is in a catalog.

Trace the M4.3 itself and it evaporates. The claim appears in no primary seismic catalog and is attributed to no named agency, no USGS event ID, no EMSC record, no IIEES bulletin. It is a number with no institution behind it, which is the defining property of a viral statistic. The other figures in circulation are not invented, but they are misattributed, and untangling them is the core of the myth-bust. The magnitude-2.5 tremor near Qom that many outlets pinned on the American strike was in fact reported by Iran's Seismological Center on 16 June, coinciding with Israeli airstrikes near Fordow, six days before the US B-2s arrived. The "M5.1 near Fordow" was a genuine magnitude-5.1 earthquake on 20 June, a natural tectonic event near Semnan, about 200 km east of Fordow, unrelated to any strike. Iran sits on some of the most active fault systems on Earth; a real M5.1 during a twelve-day window is unremarkable geology. The viral narrative took a 16 June Israeli M2.5, a 20 June natural M5.1 two hundred kilometers away, and an uncataloged M4.3 that never existed, and fused all three into a single "bunker-buster earthquake" at Fordow on 22 June, a date and place where the catalogs record nothing at all.

The physics explains exactly why the global network stayed silent, and it is not a loophole. Seismic magnitude tracks energy release, and conventional explosives and nuclear devices are separated by orders of magnitude. Even fourteen GBU-57 penetrators, each carrying a few tons of high explosive, sum to a charge measured in tens of tons TNT-equivalent. The smallest nuclear tests that global monitoring is tuned to catch release energy equivalent to hundreds or thousands of tons. A bomb, even a very large one dropped in sequence, sits far below the teleseismic detection threshold at continental range, so its signal decays into background noise long before it reaches a distant station. The blasts also couple poorly: an air-delivered warhead detonating in or against rock transfers only a fraction of its energy into elastic seismic waves, whereas a tamped underground nuclear test is engineered to couple efficiently. That is why Fordow's roughly 80 to 90 metres of overburden did not amplify the strikes into a cataloged quake. The mountain that made Fordow hard to destroy is the same mountain that muffled its destruction from the world's seismographs. The absence of a teleseismic event is not suspicious. It is precisely what conventional weapons predict.

That same threshold logic rules out the darkest speculation. No seismic signal consistent with an underground nuclear explosion was recorded anywhere in Iran during June 2025. A nuclear detonation produces a large, near-isotropic source, an outward compression the whole planet's seismometers register and that can be distinguished from the double-couple slip of a natural quake using the same discriminant the CTBTO applies for test-ban monitoring. Its absence from both the USGS and EMSC catalogs closes the door on any "accidental nuclear test" theory. If a device had gone off, the world would have seen it, plainly and immediately, in exactly the datasets that here return empty. The null cuts both ways: it deflates the exaggeration of the strike and forecloses the conspiracy about it.

Why this matters is a warning about the limit of the instrument, and it is the discipline that separates intelligence from narrative. Seismology can time a blast to the second and estimate its size, but it cannot see through a mountain. It measures ground motion, not the fate of hardware. Even a perfectly recorded shock would say nothing about whether Fordow's roughly 2,700 IR-1 and IR-6 centrifuges were shattered, knocked out of balance, or merely rattled. Damage assessment is a separate problem for a separate instrument: the ISIS analysis of IAEA reporting, which estimated the campaign disabled on the order of tens of thousands of centrifuges across Fordow, Natanz and Isfahan, is a reasoned inference from imagery and inspection, not a seismic reading. Anyone who told you the shaking proved the mission destroyed Fordow was selling the number, not reading it. The ground did shake, about 1,500 times, up to about magnitude 2, and every reputable catalog on Earth agrees it was too small to log. The most important thing the seismographs recorded about the strike on Fordow was nothing, and knowing why nothing is exactly what should have appeared is the difference between a data point and a headline.

Connectivity: a country switched off in stages

The public story of Iran's internet during the Twelve-Day War was simple to the point of being wrong: the regime "pulled the plug," the country "went dark," and the world lost sight of it. That framing survives because it matches what happened in November 2019, when Tehran throttled its population off the global internet during the fuel-price protests by brute force. But the signals recorded in June 2025 describe something more deliberate and, for anyone who studies internet control, more alarming: not a plug pulled, but a country switched off in stages, with the lights left on in the window so the outside world would not immediately notice the house was sealed.

What the signal actually recorded was a phased, engineered descent measured in near-real-time by independent outage monitors. Georgia Tech's IODA (Internet Outage Detection and Analysis) tracked active internet reachability collapsing to roughly 3% of normal, a 97% drop, at the peak of the shutdown around 18 June. This is the exact divergence Blackwing wires as a live instrument: our /api/signals/connectivity endpoint proxies IODA server-side and pulls two independent measurement methods at once. The first, active probing (how many Iranian /24 networks answer probes), is real reachability, and it cratered to about 3%. The second, BGP routing visibility (whether Iran's routes are still announced to the world), stayed near 96% throughout. That gap between the two lines is not noise. The gap is the entire story.

The escalation was not a single switch but a staircase. Corroborated by Internet Society Pulse, NetBlocks and Cloudflare Radar, disruptions began the morning of 13 June as Israel's opening strikes landed, deepened by 16 June, dropped sharply again on the evening of 17 June, and reached a near-total nationwide shutdown by about 12:50 UTC on 18 June. Each step was a policy decision, not infrastructure failing under bombardment. The war gave cover, but the mechanism was administrative.

The most important finding, and the one that separates this event from every "Iran goes dark" headline, is how access was severed. The forensic account in the arXiv paper "Iran's Stealth Internet Blackout", reinforced by the Miaan Group's technical report, is unambiguous: Iran did not withdraw its BGP routes. To routing monitors, the country still looked reachable, which is precisely why a casual observer, or a lazy outage tracker watching only BGP, would have concluded nothing much had happened. Underneath that intact routing facade, access to the global internet was strangled at the border with three techniques working together. DNS poisoning redirected over 90% of blocked domains to the private black-hole address 10.10.34.34, where requests simply died. Protocol whitelisting allowed only DNS, HTTP and HTTPS traffic to cross the border, silently dropping VPNs, SSH and anything on a non-standard port. And deep packet inspection read the HTTP Host header, URL path and TLS SNI field of every connection, injecting an HTTP 403 page or a TCP reset when it saw something it disliked. Tellingly, the paper found all of this interference occurred at the same network hop across every tested ISP, meaning enforcement was unified at a single national gateway, not delegated to individual providers. This is a censorship architecture, engineered and centralized, not an emergency kill switch.

Crucially, the country did not actually go dark from the inside. Iran's National Information Network (NIN), its long-cultivated domestic intranet, kept government portals and internal platforms running, and the central bank stated that banking was "functioning normally" even as foreign websites were unreachable. Domestic messaging apps, state media and local services stayed up. What was severed was the outward-facing internet: the connection between Iranians and the world, and between the world and Iranians. Foreign apps were the deliberate casualties. WhatsApp, Instagram and the app stores were blocked, with state television urging citizens to delete WhatsApp outright on claims it was feeding user data to Israel. The population's response was measurable and immediate: VPN-related searches spiked about 707% as people scrambled for a way over the wall, a demand curve that both the arXiv researchers and NetBlocks captured. But protocol whitelisting was specifically designed to defeat exactly that: the VPNs people rushed to install were among the first things dropped at the border.

Into this sealed environment came a much-publicized lifeline that mostly was not one. On 14 June, Elon Musk announced that Starlink had been "activated" over Iran, and the phrase "the beams are on" traveled far faster than the reality. As Iran International and others documented, turning the satellites on does nothing for a person without a dish on the ground. Only the estimated 10,000 to 20,000 Iranians with existing, clandestine terminals could actually connect, a rounding error against a population of nearly 90 million. Worse, the regime turned the hope into a trap, floating a fake Starlink app to bait users into installing malware or exposing themselves. Satellite connectivity was real for the few and a lure for the many.

Recovery, like the shutdown, was a policy signal. Traffic began climbing back around 21 June and full service appeared restored by about 25 June, tracking the ceasefire rather than any repair of damaged infrastructure. Nothing physical had broken, so nothing physical needed fixing; the state simply reopened the gateway it had closed.

Why it matters runs in two directions. First, a scope correction that the sources themselves constantly blur: this June 2025 outage is not the far longer, far more severe January 2026 blackout (a roughly 88-day near-total shutdown belonging to a separate US-Iran confrontation) that gets folded into the same search results and the same Wikipedia neighborhood. The 2025 event lasted days and tracked a 12-day war; conflating it with the 2026 blackout inflates the duration and misattributes the cause. Blackwing's live connectivity instrument hardcodes the war window server-side (13 to 25 June 2025) precisely so a "now" query can never quietly return the wrong war.

Second, and larger: this was a field test of a new template for information control. The 2019 model, withdraw your routes and vanish, is loud, easy to detect, and diplomatically costly. The 2025 model keeps your routes announced so the world's dashboards read "reachable," preserves your domestic economy and banking, and severs only the outward connection your own citizens depend on, using DNS poisoning, protocol whitelisting and DPI at a single national chokepoint. It is quieter, more surgical, and far harder to prove without the kind of dual-method measurement IODA provides. The reason two independent instruments matter, active probing set against routing visibility, is that a state has now demonstrated it can make the second lie while the first tells the truth. A blackout you can only see if you are measuring reachability, not routing, is a blackout most of the world will miss.

Aviation: the war you could read from the sky

The public was told the June strikes came without warning. The White House called Operation Midnight Hammer a masterpiece of surprise, and Joint Chiefs chairman Gen. Dan Caine said the mission was known to only a handful of people in Washington. That is true of the target list and the timing. It is not true of the force. For anyone watching the right open data, the war was legible from above days before the first bomb fell, and the same open signal that gave the build-up away was also used, deliberately, to sell a lie. Aviation was the domain where deception and disclosure rode the same channel. The channel is ADS-B: the automatic position broadcasts most aircraft transmit and a global mesh of volunteer receivers logs, aggregated by services like Flightradar24 and open APIs. A transponder is a choice. What flies with it on can be read by anyone; what flies with it off has to be inferred, and the inference itself is a data point.

What the signal actually recorded, first, was the build-up. On the night of 15-16 June, more than thirty USAF KC-135 and KC-46 tankers were tracked on public flight software crossing the Atlantic toward Europe and onward, an armada large enough that The War Zone flagged it in real time as a visible force-posture shift. Tankers are the tell. Strike aircraft can run dark, but a bomber mission across an ocean needs a refueling bridge, and that bridge is a long train of aircraft whose entire job is to loiter at known points. When dozens of them surge east at once, the aircraft they are pre-positioning for has not been announced but has already been implied. The refueling architecture leaked the operation days before the operation. This is the recurring lesson of open flight data in 2025: you rarely see the sword, but you can count the scabbards being carried into position.

The second thing the signal recorded was a decoy, tracked live and believed. In the early hours of 21 June, two flights of B-2s squawking callsigns MYTEE 11 and MYTEE 21 departed Whiteman AFB in Missouri and flew west across the Pacific toward Guam, where open-source trackers followed them the whole way and The Aviationist reported the westbound movement as it happened. For hours the trackable story was Guam: a forward deployment, a signal to Iran, a build-up in the Pacific. It was a feint. The seven B-2s that actually struck Iran flew east from Whiteman with their transponders off, invisible on ADS-B, at the same time the world was watching the wrong ocean. The tell here did not come from the network at all. An aviation enthusiast in Missouri visually counted roughly nine B-2s heading east, on the ground, by eye, while the machines showed the fleet going the other way, a detail The Aviationist folded into its Midnight Hammer reconstruction. The decoy worked precisely because open tracking is trusted. A silent aircraft is not absent from the data; it is a hole in the data, and the westbound flights were staged to fill the hole with a story.

The third thing is the strike itself, and why so much of it went unseen. The Pentagon briefing put the scale at more than 125 aircraft, seven B-2s, 14 GBU-57 bunker-busters and roughly 75 precision weapons, a package supported by a B-2 leg of about 37 hours round-trip (roughly eighteen hours each way) with multiple air-to-air refuelings. The refuelings are the key to the invisibility. They happened over open ocean, outside the range of ground-based ADS-B receivers, so even the tanker rendezvous that could not fly dark had no listeners underneath them. ADS-B coverage is dense over land and sea lanes and thin to nonexistent in the middle of an ocean, which is exactly why the mid-Atlantic and mid-ocean segments of the mission were dark on the public feeds. The seven bombers then delivered the first-ever combat use of the GBU-57 MOP against Fordow and Natanz, and per Caine, Iran's fighters never flew and its surface-to-air missiles never engaged. The strike package was not so much hidden as flown through the gaps in the sensor map, a route chosen with the coverage of the open network in mind. The same enthusiasts who caught the tankers over the Atlantic and the decoys to Guam saw essentially nothing of the twelve hours over water, and that absence is not a failure of the data. It is the shape of where the data ends.

Why it matters extends past the strike, because the same open feeds recorded the war's largest and most legible effect: the emptying of the sky. The moment Israel opened Operation Rising Lion on 13 June, Iran, Iraq, Jordan, Syria and Israel closed their airspace, and Flightradar24 imagery that Friday showed the region wiped clean of commercial traffic. Ben Gurion shut, and Israeli carriers ferried their fleets out of the country empty to protect the aircraft. This is where open aviation data stops being an intelligence curiosity and becomes a civil-infrastructure map. With Iran and Iraq closed, traffic had to be funneled around the war, and the reroute is visible in the counts: daily overflights of Saudi Arabia jumped from about 700 in mid-May to about 1,400, while flights over Afghanistan spiked nearly 500 percent, per CNN Travel citing Flightradar24. The war did not just close corridors; it visibly rerouted the pattern of global aviation, and every displaced flight was legible in real time to anyone watching the same maps used to catch the bombers.

The closures returned, and told their own story, at the end. On 23 June, Qatar closed its airspace shortly before Iran's retaliatory missiles reached Al Udeid Air Base, and the UAE, Bahrain, Kuwait and Iraq followed before reopening about an hour later. The sequence is worth reading carefully. The airspace closed before the missiles arrived, which is itself a signal: a coordinated, pre-warned strike rather than a bolt from the blue. Satellite imagery separately showed Al Udeid pre-evacuated of aircraft, its ramps cleared in advance, consistent with a calibrated, casualty-free salvo that Iran could present at home as retaliation and Washington could absorb without escalation. Here the open aviation picture and the closure timeline align to correct a headline. The attack on the largest US base in the region was not a surprise assault that America failed to see coming. It was a telegraphed, hour-long event that both sides had effectively choreographed, and the empty ramps and the pre-emptive airspace closure recorded the choreography.

The through-line across every one of these signals is a single discipline: treat the open feed as adversarial. The tankers were real and gave the game away. The Guam flights were real and were a lie. The twelve hours over water were neither, just a gap that the operation was routed to exploit. The founding principle of open-source aviation tracking is that the sky is honest about what it shows and silent about what it hides, and the silence is as meaningful as the broadcast. Reading the June war from above meant knowing which planes were telling the truth, which were staged to be seen, and which mattered precisely because you could not see them at all.

Maritime: chaos in the world’s oil chokepoint

The headline the world absorbed was binary and alarming: Iran was going to close the Strait of Hormuz. After American B-2s hit Fordow, Natanz and Isfahan on 22 June, the story hardened into a countdown, because through this one channel moves roughly 20% of the world's oil, about 20 million barrels a day, plus a comparable share of its LNG, on 130 to 150 vessels a day. At its narrowest the Strait is only about 21 miles wide, with the shipping lanes narrower still, so the intuition that one determined state could plug it is not unreasonable. But the machines that actually watch the Gulf, the AIS receivers, the freight desks, the war-risk underwriters, recorded something more specific and more instructive than a closure. They recorded an electronic-warfare environment, a measurable but partial drop in traffic, a real collision, and a political decision that was voted for and never carried out. The gap between "Iran closed the Strait" and what the sensors logged is the whole story.

What the signal actually recorded first was interference at industrial scale. Maritime-intelligence firm Windward found that roughly 970 to 1,000 vessels a day were experiencing GPS jamming or spoofing across the Gulf and the Strait in mid-June, and the disruption was concentrated near Bandar Abbas, Iran's principal naval port on the northern shore ( Windward). Jamming and spoofing are not the same failure. Jamming drowns the satellite signal so a receiver goes blind; spoofing feeds it a convincing lie, a fabricated position the navigation system accepts as truth. The spoofing here was vivid enough to be absurd. On bridge screens and public tracking sites, supertankers appeared to circle over dry land, cargo ships crossed airports, and vessels drifted straight through a nuclear power plant ( Scientific American). These are not navigation errors a crew makes; they are the signature of a transmitter on shore overpowering the real satellites. University of Texas navigation expert Todd Humphreys, who studies GPS spoofing, judged the interference "almost certainly" Iranian, an attribution grounded in the geometry: the false positions clustered around, and appeared to emanate from, the Iranian coast near Bandar Abbas rather than the Arabian side.

The clearest physical consequence came on 17 June, and it is worth stating precisely because it is exactly the kind of claim this platform exists to get right. The crude supertanker Front Eagle, a VLCC carrying about 2 million barrels bound for China, collided with the tanker Adalynn in the Gulf of Oman near the eastern mouth of the Strait. Windward's reconstruction found that sustained GPS and AIS jamming had, in the days before, falsely placed Front Eagle inside Iranian territory, part of the same scrambled picture that put ships over land ( Windward). The outcome was a deck fire that was extinguished and 24 crew from the Adalynn evacuated by the UAE coast guard, with no casualties. The careful phrasing matters: the jamming did not "cause" the collision the way a hand steers a wheel, and no investigation has isolated it as the sole factor. What the record supports is that the collision happened inside a degraded-navigation environment the interference had created, which is why it became the war's most cited maritime incident. A confused position picture in one of the planet's busiest, tightest sea lanes is a standing collision risk, and on 17 June that risk paid out.

Traffic thinned, but it never stopped, and the numbers draw the line between disruption and closure. Windward's transit count fell from 147 vessels on 9 June to 111 on 15 June, a drop of roughly a quarter as owners held ships outside the Strait and waited ( Windward); tanker executives described around 20% less passage. That is a serious, insurer-moving contraction. It is also not a closure: three-quarters of normal traffic was still moving through a channel supposedly on the brink of being sealed. The money told the same two-sided story. Hull war-risk premiums for the Gulf roughly doubled to about 0.2% of a vessel's value, a large jump in a market that usually moves in basis points, and on 13 June VLCC freight on the Gulf-to-China route jumped 24% in a single day ( Marsh McLennan, via S&P Global). Underwriters were pricing a real, elevated probability of loss. But 0.2% of hull value is a war-zone premium, not a shut-passage premium, and freight that spikes on the fear of a cut is freight that is still being booked because ships are still sailing. Notably, oil prices ultimately fell by the ceasefire once traders concluded the Strait and its oil infrastructure had been spared, the tell that the market never truly believed the channel would go dark.

Which brings the central myth-bust. After the 22 June US strikes, Iran's parliament backed closing the Strait of Hormuz, and that vote became, in much of the coverage, the closure itself. It was not. The parliamentary move was advisory: under Iran's system the decision rested with the Supreme National Security Council and, ultimately, Khamenei, not the legislature ( Axios). The council never ordered it, and the Strait was never physically closed at any point before the ceasefire. The restraint was rational and the United States named the reason bluntly, calling a closure "economic suicide" for Iran ( CNBC, via the same reporting): Iran's own crude exports, overwhelmingly to China, transit the same water, and blocking it would invite direct confrontation with the US Fifth Fleet based across the Gulf in Bahrain. So the record shows a threat that was voted for, priced by insurers, absorbed by freight desks, and then declined. Iran leaned on the fear of closure, and on jamming as a cheaper form of pressure, without ever taking the step that would have been genuinely self-destructive.

Why this matters runs past a single strait. The June war was a live demonstration that you do not have to close a chokepoint to disrupt it; you can just corrupt the data that runs it. A handful of shore transmitters near Bandar Abbas degraded navigation for a thousand ships a day, contributed to the environment around a supertanker collision, pushed premiums and freight up, and thinned traffic by a fifth, all without a single mine laid or missile fired at a hull. That is a lower-cost, more deniable, and harder-to-attribute tool than a blockade, and it leaves the headline ("closure") and the reality ("degradation") pointing in different directions, which is precisely how a partial disruption gets miscounted as a total one. For anyone modeling the next Gulf crisis, the June record argues for watching the AIS integrity feeds, the war-risk premium, and the transit count, three independent instruments, rather than a parliamentary vote, because the vote said "closed" and every machine that actually measures the Strait said "open, and rattled."

Markets: what the money knew

The loudest claim of the war was priced in advance and printed on every front page: bomb Iran, and oil detonates with it. When Israel opened Operation Rising Lion on 13 June, the reflex forecast was a run to $100, then $120, then a global energy shock as Tehran closed the Strait of Hormuz, the chokepoint through which roughly a fifth of the world's seaborne crude moves. The public absorbed a simple equation: war on Iran equals an oil crisis. The tape says the equation was wrong, and it says so with unusual precision. The market read this war more coldly, and more accurately, than the commentary written about it.

Start with the single series that anchors everything, because we poll it live from the St. Louis Fed rather than quote it from memory. Brent closed at $70.84 on 12 June, the last session before the war. Israel struck overnight and crude gapped up to about $76 on the 13th, a roughly 7% jump, exactly the war-premium the models demanded. It kept climbing as the strikes widened, and on 19 June Brent peaked at $80.37, its highest of the episode. That is the number the "oil crisis" narrative captured and froze. What the narrative never explained is what happened next, because the peak was not a launch pad. It was the top. By the 24 June ceasefire Brent settled at $69.13, meaning crude finished the war below where it started. A twelve-day conflict that included the direct US bombing of three nuclear sites left oil cheaper than it was at peace. The war premium did not just deflate. It went negative.

The reversal has a date and a mechanism, and both matter. The break came on 23 June, the day of Iran's promised retaliation, when WTI fell about 7.2%, its largest single-day drop since April. Read that sequence slowly, because it inverts the intuition completely: oil crashed on the day Iran struck back at the United States, not on the day it stood down. The reason is that Iran's counterstrike was engineered to be survivable for everyone. Tehran hit Al Udeid Air Base in Qatar, the largest US installation in the region, but the strike was telegraphed in advance, the base was evacuated, and it produced zero US casualties even as Patriot batteries mounted what the Joint Chiefs chairman called the largest single Patriot engagement in US history. Crucially, Iran did not touch the Strait of Hormuz and did not touch a single barrel of oil infrastructure. Traders watched a retaliation designed as a face-saving exit rather than an escalation, and in real time they priced out the supply cut. The feared closure was not going to happen, so the premium built on it evaporated in a session. The money understood the choreography of the strike faster than the analysts narrating it.

Gas told a milder version of the same story. European TTF gas rose about 18% to its highest since February by 19 June, and Asian LNG hit a four-month high, as the market braced for disrupted flows. Then, as with crude, both fell roughly 20% back toward pre-crisis levels after the ceasefire. A spike that fully unwinds is not a shock. It is a scare, and the tape catalogued it as one. Even gold, the reflexive safe-haven trade, behaved with restraint. It climbed to a multi-week high near $3,451 as haven demand built, and this is where accuracy earns its keep: that was a multi-week high, not an all-time record. Gold did not exceed its April 2025 peak near $3,500, and it eased once the ceasefire held. The reader who absorbed "gold hits record on Iran war" absorbed a headline the price data does not support.

If commodities recorded fear that faded, equities recorded a colder judgment about who wins. On 13 June, the day the war opened and the S&P 500 fell, the defense complex rose against the tape: Northrop Grumman gained about 4%, Lockheed Martin 3.6% and RTX 3.3%, with Israel's Elbit Systems later running to record highs on war demand. That is the market pricing a restocking cycle, and the cycle is real: defending Israel and Al Udeid drew down scarce, expensive interceptor inventories, with estimates of the total interceptor spend running into the billions over twelve days and THAAD replenishment projected to take years. Every Patriot and SM-3 fired is a future order, and equity investors bought the order book while everyone else bought fear.

The single sharpest signal was not a commodity or a sector. It was a whole national market voting on the outcome. The Tel Aviv TA-35 and TA-125 gained about 3.2% over the war week and kept climbing, closing 2025 up roughly 52% at record highs, outpacing global markets. Israel's own stock exchange rose during Israel's own war. Markets do not celebrate being at war. They price the expected aftermath, and Tel Aviv priced a strategic win, then extended the rally as the ceasefire held. The mirror image sat 1,500 kilometres east: Iran's rial weakened from about 820,900 to the dollar toward roughly 880,000 on the open market, and the Tehran Stock Exchange halted trading for about nine sessions under emergency rules, then fell over 15% after reopening. We flag those rial figures as disputed: Iran's black-market rate has no clean authoritative feed, the reporting is contested, and the honest thing is to say so rather than launder a range into a fact. But the direction is not in doubt, and the divergence is the finding. One combatant's assets set records; the other's had to be switched off to stop the selling.

Why it matters is a lesson about where to look when a war is loud. The commentary treated the possible closure of Hormuz as a fact in waiting and priced the disaster into its language. The tape treated it as a probability and repriced it the instant Iran's own conduct on 23 June revealed the probability was near zero. Oil finishing below its pre-war level is not a footnote; it is the market's verdict that this was a contained, choreographed exchange, not the opening of a regional energy war. That is exactly the myth this platform exists to bust, and it is why we wire the Brent and WTI series live from FRED at /api/signals/oil rather than assert a number: you can watch the peak, the 23 June break, and the sub-$70 settle in the machine data yourself. The prices were on the record in real time, discounting the strike faster than the strike could be narrated. When the story and the tape disagree, the tape had already done the arithmetic.

Radiation: the catastrophe that did not come

The image was fixed in the public mind before the first bomb fell: bunker-busters cracking open Iran's nuclear sites, and a plume drifting over Tehran, the Gulf, maybe Europe. When twelve GBU-57 bunker-busters went into Fordow's ventilation shafts on 22 June and roughly thirty Tomahawks struck Isfahan, the expectation was a radiological event. The reporting was saturated with the question: how big are the radiation risks. It is the one question where the answer, from every instrument built to measure it, came back the same, and where the popular intuition was almost exactly wrong about which hazard to fear.

What the monitors actually recorded was nothing. In his 22 June statement to the UN Security Council, IAEA Director General Rafael Grossi reported that radioactivity outside Natanz remained "unchanged and at normal levels," and that Iranian regulators had registered no increase in off-site radiation at Fordow, Natanz or Isfahan after the US strikes ( IAEA to the UNSC, 22 June). This was not a single agency's optimism. It was a convergence: the international monitoring authority and the target state's own regulators, adversaries in almost every other respect, reporting identical readings. When the machines on both sides of a conflict agree, that agreement is the signal. There was no off-site release to detect, because there was none to release in the form the public imagined.

The reason lies in a physics distinction that headlines almost never make. An enrichment plant is not a reactor. In his 20 June statement, Grossi spelled out that the dominant hazard inside a struck enrichment facility is chemical, not radiological: the plants hold uranium hexafluoride (UF6), which on contact with moisture in the air reacts to form hydrogen fluoride, a highly toxic and corrosive gas, with contamination confined to the damaged buildings ( IAEA to the UNSC, 20 June). Enriched uranium itself is a weak alpha emitter: dangerous if inhaled or ingested, but its radiation does not travel, and it does not form the drifting, persistent plume that defines a nuclear catastrophe. The Bulletin of the Atomic Scientists laid the contrast out in plain terms: hitting an enrichment plant is a contained event, a local chemical hazard for anyone inside the building; hitting a power reactor is catastrophic, because a reactor holds thousands of kilograms of mobile, long-lived fission products, principally cesium-137 and iodine-131, the isotopes that seeded the Chernobyl and Fukushima exclusion zones. One is a spill. The other is a plume.

The site's own engineering reinforced the containment. The IAEA noted that contamination stayed limited to the damaged buildings, and analysts observed that the deep-underground design of Fordow (bored roughly 80 to 90 metres under a mountain) and Natanz did two things at once: it hardened the halls against the weapons, and it trapped dispersed material below ground where no off-site pathway existed ( Al Jazeera, quoting the IAEA). The same depth that made these targets so hard to destroy is the depth that kept whatever was destroyed from reaching the surface. At Isfahan, where roughly thirty Tomahawks struck the conversion complex, the IAEA counted eighteen buildings hit and still recorded no off-site increase. Damage was extensive. Release was not.

The genuine Chernobyl risk sat untouched, and that was deliberate. Iran's only operating power reactor, Bushehr, a Russian-built, IAEA-safeguarded VVER on the Gulf coast, was pointedly not on the target list. It holds thousands of kilograms of intensely radioactive material, and Grossi warned in the sharpest language of the war that a direct hit there "would result in a very high release of radioactivity" ( Al Jazeera). Russia's Rosatom, which operates the plant, called a strike on Bushehr a potential Chernobyl-scale catastrophe. This is the tell that runs beneath the entire radiological story: the targeting itself encoded the physics. The strikes went after enrichment and conversion, where the hazard was chemical and confined, and spared the one facility where the hazard was radiological and mobile. The absence of a catastrophe was not luck. It was a target list written by people who understood exactly the distinction the headlines blurred.

Two further points close the record. First, there was no nuclear-explosion signature to detect at all. The CTBTO's radionuclide network, the global system that sniffs the atmosphere for the fission products of a nuclear test, had nothing to register, because these were conventional munitions. There is no CTBTO detection of a release from these strikes on record, and there should not be: bunker-busters do not fission. Anyone who expected the test-ban monitors to light up misunderstood what those monitors are for. Second, and this is where the reassuring story turns uneasy, the danger that survived the war was not radiological but custodial. The IAEA had verified Iran's stockpile of 60 percent enriched uranium at 408.6 kg on 17 May, growing to 440.9 kg by 13 June, the day before the attacks ( IAEA report GOV/2025/50). After the strikes the Agency lost continuity of knowledge: over 400 kg of near-weapons-grade material fell outside IAEA monitoring, its location uncertain and likely buried in the tunnel complex at Isfahan or under Fordow, enough, if further enriched, for roughly ten weapons ( Bulletin of the Atomic Scientists). On 2 July, Iran suspended cooperation with the IAEA, cutting the monitors off entirely.

Why it matters: the radiological non-event is the clean case for reading the machines over the mood. Every authoritative monitor, IAEA and Iranian regulator alike, recorded no off-site release, and the reason was structural, not fortunate: enrichment plants hold a weakly radioactive, chemically toxic material that stays where it falls, deep burial trapped it, and the one reactor that could have produced a plume was deliberately spared. The catastrophe the public braced for was never physically available from these targets. But the same event that produced no radiation produced the more durable problem: a war meant to remove a proliferation risk ended with more than 400 kilograms of the most sensitive material on earth unaccounted for and the inspectors locked out. No plume crossed a border. The uranium simply went dark, which is the harder thing to monitor and the one that outlasts the smoke.

Satellite: the referee in orbit

For twelve days in June 2025, the damage to Iran's nuclear program was a claim before it was a fact. Washington called Operation Midnight Hammer a total success and said Fordow was "obliterated." Tehran insisted the enrichment halls were largely intact. An early, leaked US Defense Intelligence Agency assessment reportedly judged the setback at only months, not years. Three governments, three stories, and a public with no way to referee between them. The dispute was never settled at a podium. It was settled from orbit, by commercial and scientific satellites that answer to no government's talking points and record the same ground on every pass. Where the seismic catalogs stayed silent because a buried, tamped explosion couples almost no energy into the crust, the sky had an unobstructed line of sight, and it kept the receipts.

The most decisive receipt was a pattern of holes. On 22 June, Maxar imagery showed six penetration holes along the ridge over Fordow's underground hall, arranged in two clusters of three, according to Bloomberg's analysis of the satellite record and the Institute for Science and International Security. That geometry was not decorative. Four days later, in a briefing detailed by DefenseScoop, Joint Chiefs Chairman Gen. Dan Caine described exactly how the weapons were aimed: twelve GBU-57 bunker-busters were "walked" into two ventilation shafts. The first bomb blew off the concrete cap protecting a shaft, weapons two through five entered at more than 1,000 feet per second and detonated deep inside, and a sixth was held as a backup. The surface crater pattern the satellites recorded matched the two-shaft, six-per-side account the general described. This is the myth-busting inversion Blackwing exists to show: here the imagery did not contradict the official story, it corroborated it, down to the count and the clustering. The picture and the briefing were independent witnesses telling the same story.

The satellites had also been watching before the bombs fell, and that pre-strike record complicates any tidy verdict. Maxar frames from 19 and 20 June showed roughly 16 cargo trucks queued at Fordow's tunnel entrance, with bulldozers sealing the entrances, imagery The War Zone read as deliberate pre-strike backfilling by Iran, either to protect the halls or to move material out ahead of a strike Tehran clearly anticipated. That single observation became the spine of the enriched-uranium question that outlived the war: if trucks were loading before the first bomb, the 400-plus kilograms of 60 percent HEU may not have been under the mountain when the shafts collapsed. The orbital record cannot say what was in the trucks. It can say, with a timestamp, that the trucks were there, and that raises the question no ground assessment could raise on its own.

Then came the finding no optical camera could have produced, and the one that best captures why the referee has to sit in orbit. On the night of 21 June, hours before the American strike, NASA's VIIRS thermal sensor flagged an infrared anomaly about 3 km north of Fordow, work documented by the Payne Institute at the Colorado School of Mines. Two VIIRS passes, at 22:57 and 23:23 UTC, caught a heat source the institute's Mikhail Zhizhin estimated at about 1,200 K with roughly 2 MW of radiative output using the Nightfire algorithm. That is a pre-bombing event, invisible to anyone on the ground, invisible to daytime optical imagery, and visible only because a thermal instrument in polar orbit happened to be looking down at the right minutes. What it was (a vehicle fire, an industrial burn, activity tied to the truck movements) remains unconfirmed. That it happened is not in dispute, because the machine logged it. The lesson is the whole thesis of this platform in one detection: the sensor sees what the narrative does not know to mention.

The optical view degraded after the strikes, as dust, smoke, and Iranian control of access closed the window. Analysts changed sensors rather than accept the blackout. Bellingcat turned to Copernicus Sentinel-1, a radar satellite, to map damage across the nuclear sites, a technique the reporting on its Iran damage tool describes as working at night, through cloud, and precisely when optical imagery goes dark. Radar does not photograph light, it measures the phase of a reflected microwave pulse, so a collapsed hall or a shifted surface changes the return whether or not any camera can see it. In parallel, Orbital Sidekick's hyperspectral constellation detected ground subsidence over an area about 315 by 275 m at Fordow, with concrete pushed to the surface, a signature Asia Times reported as consistent with underground halls collapsing rather than merely being cratered. Four different sensing methods (visible light, thermal infrared, radar, and the fine spectral fingerprint of disturbed material) converging on the same site is what an independent referee looks like: no single method carries the verdict, and none of them takes a government's word for it.

The referee also kept watching after the ceasefire, which is where the "obliterated" framing meets its most sober correction. CSIS found virtually no rehabilitation at Fordow, Natanz, or Isfahan through 30 September, a finding from the CSIS iDeas Lab that cuts against Tehran's claim of quick recovery. But the same imagery cut the other way too: CSIS detected active tunneling at Pickaxe Mountain, about a mile south of Natanz, a possible new and deeper site. Both facts are true at once, and only sustained orbital monitoring can hold them together. The strikes did lasting damage that months of quiet did not undo, and Iran was digging anew somewhere the last bombs did not reach. A one-time snapshot would have delivered either the triumphant verdict or the alarming one. The continuous record delivers the honest, uncomfortable both.

Why it matters is a claim about who gets to be believed. In June 2025 the three governments with the most at stake each had an interest in a different answer, and none of them could produce evidence the others would accept. The satellites had no such interest. A crater count, a thermal timestamp, a radar phase shift, and a subsidence footprint are not opinions, and they were gathered by instruments that would have recorded the same ground whether the story that day was "obliterated" or "intact." That is precisely why Blackwing treats orbit as a first-class witness. We document NASA FIRMS, the free archival source behind the Fordow thermal detections, as a future live thermal feed on this platform, so that the next anomaly over a hardened site is caught by the same class of sensor that caught the last one, on the record, in near real time, before any government has decided what to call it.

What the seven signals settle

Stand back from the individual instruments and a single pattern emerges. On the questions that generated the most heat, the machines and the headlines diverged, and the machines were checkable. The story the data tells is quieter than the one most people remember, and it is more precise.

The corrections are specific, and each one is a link away from its source. There was no bunker-buster earthquake at Fordow: two independent global catalogs are empty and the real blasts topped out near magnitude 2. Iran did not simply pull the plug on the internet; it ran a stealth blackout that kept its routes announced while severing the border, a far more sophisticated and more repeatable act than a 2019-style shutdown. The American strike was not the total surprise it was sold as; a tanker bridge telegraphed it days early while a decoy flight to Guam drew the cameras away from the real package heading east in the dark. Iran did not close the Strait of Hormuz, whatever its parliament voted; it electronically scrambled it, and the tankers kept moving. Oil did not spiral; it fell below where it started because the money read the telegraphed, casualty-free retaliation for exactly what it was. There was no radiological release, because the physics of an enrichment plant is chemical, not nuclear, which is the same physics that spared Bushehr. And when governments argued over the damage, the satellites settled it, down to six holes over two ventilation shafts.

No single one of these signals is decisive on its own, and that is the point. A seismograph cannot see through a mountain. An outage monitor cannot explain a shutdown's intent. A flight tracker loses aircraft over open ocean. The method that produces a trustworthy account is not any one instrument but the discipline of fusing them: reading each independent stream against the others and against the primary record, keeping the uncertainty visible, and correcting the story when a measurement beats a rumor. That is what separates intelligence from narration, and it is the entire premise of this platform.

Some questions the instruments could not close, and honesty requires naming them. The seismographs cannot say how many of Fordow's centrifuges survived, only that the mountain muffled the blasts; that is a job for the inspectors and the imagery analysts, and even they disagree at the margins. The satellites can count impact holes but not read the state of the halls beneath them. And the fate of Iran's stockpile of near-weapons-grade uranium, more than four hundred kilograms of it that moved outside international monitoring in the chaos of the strikes, is a question no civilian sensor can answer. It is the most consequential open file the war left behind. A good record is as clear about the edges of its knowledge as it is about the center, and this is where the edge runs.

The war's human cost is the one figure that no instrument can soften, and it deserves to be stated plainly rather than rounded. Israel's twelve-day campaign left a reported toll on the order of a thousand or more dead inside Iran, a number that remains contested between Iran's official range and independent monitors. Iran's missiles killed 28 people in Israel. Those two figures, and the gap between them, are the ground truth under everything above.

Most of these instruments are still running. The war is over, but the seismometers are still pointed at the same ground, the outage monitors at the same network, and the market tapes at the same barrel of oil. That is why this record is not a museum. On the live intelligence page several of these signals are wired to their sources and refresh on their own, so the same discipline that reconstructed June 2025 keeps watching the present. The live map replays the twelve days strike by strike, and the full timeline lists every sourced event behind this briefing.

The larger lesson outlives the twelve days. The same public instruments that made this war legible, open seismic catalogs, keyless outage monitors, community flight and ship trackers, free market data and civilian satellites, are available for every conflict that follows, to anyone willing to read them with discipline. The barrier to a verifiable account of a war is no longer access to the data. It is the patience to cross-reference it, the honesty to publish the uncertainty, and the willingness to correct the record when a measurement overturns a headline. That is a method, not a scoop, and a method can be repeated.

It also changes who gets to hold the powerful to account. For most of the last century, the authoritative account of a war was written by the handful of institutions with the reach to gather the facts, and everyone else took it on trust. The seven signals in this briefing were gathered without a single classified source, a single embedded reporter, or a single official briefing. They came from instruments that publish to the open web. That does not make the resulting picture perfect, and this briefing has been careful about its limits. But it does mean the picture is auditable by anyone, which is a genuinely new condition. When the raw evidence is public, the monopoly on the first draft of history is broken, and the correction can come from anywhere.

If you can show, with a source, that any figure here is wrong, we will correct it. In an information environment where the first number is almost never the true one, that willingness to be corrected is not a weakness of the record. It is the whole point of keeping one.