This page predicts what stories will be told, not what will happen. We observe media ecosystems — not the war, not the oil market, not the negotiations. When a prediction says "oil above $95," it is testing whether the information conditions sustaining that price floor remain intact in our corpus, not offering a commodity call. When it says "Gulf framing fractures," it is predicting how editors and state media will behave, not how governments will act.
Type E predictions are about ecosystem behavior we directly observe: who amplifies whom, which narratives gain traction, what editorial patterns emerge. Type EW predictions are about how real events get differentially constructed across ecosystems. Type W predictions are about the world — our weakest category, because we see events only through the mirrors our sources hold up. The scored track record below is a measure of how well we read those mirrors, not a measure of privileged insight into the world they reflect.
The Daily Forecast
Iran Strikes Monitor — September 1, 2026
Day 186 of the US-Israeli air campaign against Iran. Hours 4419–4443 since first strikes at ~06:10 UTC, February 28.
This forecast covers editorials #643 through #644, published at 10:06 and 22:06 UTC on August 31. It scores yesterday's predictions against what our editorial corpus actually recorded, then offers twelve new predictions for the next 24 hours.
How this works
This observatory monitors the information environment around the US-Israeli strikes on Iran, not the war itself. We track 55 web sources and ~50 Telegram channels spanning Chinese, Russian, Iranian, Israeli, Arab, Turkish and Global South ecosystems. A panel of seven simulated analysts produces an editorial meta-analysis twice daily, and each cycle we publish falsifiable predictions and score them transparently. Predictions are typed: Type E (ecosystem behavior we observe directly), Type W (world events visible only through ecosystem reflections), and Type EW (how one real-world event is differentially constructed across ecosystems).
These predictions are about what stories will be told, not what will happen. When we name a barrel count, an airbase or a casualty figure, we are testing whether the information conditions that keep that object visible and contested in our corpus hold — not offering a verification, a trading signal or an intelligence estimate. By design we do not monitor Western mass media directly — AP, WSJ, Reuters, Guardian, Bloomberg, Washington Post, Axios. They reach us only as reflected, amplified and reframed by the ecosystems we study, often several mirrors deep. About our methodology.
Where we are
A video of a strike travelled further than any strike. Trump's AI-generated footage of Kharg Island reached every ecosystem we monitor within hours and produced, in #643, six incompatible readings — TASS ran it as news, SolovievLive as "Trump destroyed Kharg with AI," Fars as mockery, Rybar MENA as a symptom of American weakness, Boris Rozhin as coercion aimed at Hormuz. The consequence is the part with no precedent in our record: the National Iranian Oil Company's director had to state that Kharg is calm and oil operations have not stopped — an operational denial of an event that did not occur. By #644 the readings had converged on ridicule, and the anchor for that convergence was an American official telling Reuters, via ajanews, that US forces had not struck Kharg. No source in our corpus published independent imagery of the island. The argument was conducted entirely between renderings and statements.
Hormuz is now described twice by the same ecosystem, in the same window. The IRGC Navy announced that a "violating" supertanker struck two naval mines and warned shipping firms that non-compliant vessels would share its fate; CENTCOM denied the mine strike outright, and an Iranian military source told Tasnim that the mine-laying premise was "mere illusions." Fars reported the Strait completely closed with no transit without Iranian permission; Trump, via ajanews, said thirty ships pass a night. Underneath, Fars also carried Washington warning shipping companies that paying Iranian fees for safe passage could trigger sanctions. The contested question in the corpus is no longer whether vessels move but who is paid for their moving — and that is being fought in sanctions law. The only non-belligerent maritime datapoint in the second window was a UKMTO report. Follow Strait of Hormuz & Oil.
A state ecosystem borrowed authority it does not control. Middle East Spectator published satellite imagery indicating hangar damage at Jordan's Muwaffaq Salti airbase; cig_telegram added coordinates. Within hours Press TV, Mehr and IRNA republished the OSINT product as confirmation of their own strike claim. This inverts the pattern we have logged for weeks: Tehran routing its case through a channel it does not own, because the imagery carries an authority its bulletins cannot manufacture. The same window recorded the reverse move on the UAE, where Iran's Army claimed dozens of drones against Al-Minhad and the UAE Ministry of Defence confirmed engaging a single UAV over territorial waters. Follow IRGC Retaliatory Waves.
The most durable object in the corpus is the American attrition problem, because mutually hostile ecosystems describe it identically. Al Mayadeen carried Vance conceding attrition "is a concern for us"; cig_telegram reported military leaders advising the defense secretary against prolonged operations; Readovka reported framework agreements with General Dynamics OTS and Lockheed Martin to multiply interceptor production; cig_telegram put the Strategic Petroleum Reserve at 286.6 million barrels. Adjacent and almost unamplified: BBC Persian reported a senior US Navy officer saying there is no urgent plan to return evacuated forces and families to Bahrain. Six months on, that is a standing threat assessment about the Fifth Fleet's home port, and it moved through fifteen hundred messages once.
Yesterday's scorecard
| # | Prediction | Type | P | Verdict |
|---|---|---|---|---|
| H1 | Iranian state carriage rests on ≥5 named Western sources | E | 82% | Confirmed. Seventh consecutive window. The Washington Post Hegseth readiness report through five ecosystems, the WaPo Pentagon-influencers story via ISNA and IRNA, Marjorie Taylor Greene via Mehr and Press TV, Reuters shipping data via BBC Persian and Xinhua, an AP investigation identifying PrSM missiles at Lamerd via Fars and ISNA, Reuters on Brent via Al Mayadeen and IRNA. |
| H2 | ≥3 mutually exclusive Hormuz quantifications, no reconciliation | E | 78% | Confirmed. Five vessels a day per Reuters shipping data, 83 diverted per CENTCOM, thirty ships a night per Trump, complete closure per Fars. No item placed two side by side. Verification behavior did appear in the corpus — Rozhin setting Trump's "fully demined" against the same-day mine claim — but never against a transit number. |
| H3 | Closure and toll registers stated in the same window | E | 76% | Confirmed, and tighter than predicted. Fars fused both limbs into one sentence: the Strait is completely closed and no vessel transits without Iranian permission. The fee regime was corroborated from the opposite direction by Washington's sanctions warning to shippers. |
| H4 | No damage assessment for the Jordan strikes; hedged OSINT register persists | E | 72% | Refuted, decisively. Three parties published assessments within hours — the IRGC naming King Hussein and Azraq, Jordan's armed forces claiming eight interceptions, a US official claiming no significant damage — and Middle East Spectator then published satellite imagery of Muwaffaq Salti. We predicted nobody would be held to a claim about a third country's territory. Everybody was. |
| H5 | Unity cascade continues; Mojtaba's authority arrives as relayed text | E | 78% | Refuted on the load-bearing limb. Neither editorial recorded a single new ratification item, and Mojtaba Khamenei does not appear in either. The silence limb held vacuously. We said yesterday we would stop writing compound predictions and then wrote one. |
| H6 | An outlet applies claim hygiene to its own ecosystem's material | E | 52% | Confirmed, our lowest-probability instrument. Mehr published a ledger of what Iranian media denied in 24 hours — including a circulating 70-casualty IRGC figure — and Middle East Spectator was alone in carrying Naghdi's concession that "our assessment of US capabilities was greater than what turned out to be the case." IRNA also published the equal-weighted index falling 6,000 units against Fars and ISNA leading on a record high. |
| H7 | One-source, multi-carrier civilian-harm claim with no propagation | EW | 72% | Confirmed. The health minister's 66,000 wounded, 300-plus destroyed ambulances and 70-plus damaged hospitals ran as a domestic item and never entered retaliation coverage abroad. Gaza's five wounded fishermen stayed inside Palestinian and Houthi feeds. Exactly one channel, Fotros, named the men killed at Lark. |
| H8 | Kinetic event enters first through the striking party; Iran attributes | EW | 70% | Partial, with an instructive deformation. The mechanism held — the first account of activity at Kharg was written in Washington and Iranian outlets handled it as foreign material, from Fars's mockery to NIOC's denial. But the event was synthetic. Our test assumed a briefing races a real event into the record; this window the briefing had no event behind it. |
| H9 | American-capacity claims across ≥4 ecosystems, uncontested | EW | 78% | Confirmed, and the evidentiary object changed. Vance, Schumer, cig_telegram, Readovka, Al Mayadeen and a TASS op-ed converged on interceptor and readiness shortfalls. Not one channel we collect disputed any of it. A Russian milblogger and an American vice president describing the same constraint is a different thing from either alone. |
| H10 | Iranian civilian harm as statistics, no death count, no attestation | EW | 80% | Confirmed. #644 states it flatly: no Red Crescent, ICRC or UN figure appears anywhere in the window against any belligerent's count. The register moved from recovery back to damage inventory — 100 civilian aircraft, 25 airports, 500 communications sites, a 2,000–3,000 billion toman bridge repair. The closest thing to independent forensics was an AP investigation reaching us only through Fars and ISNA. |
| H11 | ≥8 new named objects across ≥3 categories | W | 85% | Confirmed, densely. King Hussein, Azraq, Muwaffaq Salti, Al-Minhad, Bidganeh, Kermanshah, Khorramabad, Kharg, PAC-3 interceptor ratios, Brent $90.49, the SPR at 286.6m barrels, the Karaj B1 bridge cost, General Dynamics OTS, the Lamerd PrSM finding, and Ali Fiyazi and Hamid Avazzadeh by name. |
| H12 | Bishkek bilateral framed incompatibly by Russian and Iranian carriage | W | 58% | Refuted, as we specified in advance. A second consecutive window in which neither ecosystem previewed the meeting. We wrote that this outcome would corroborate the deprioritisation reading, and it does. Follow Regional Focus: Russia. |
Summary: 8 confirmed, 1 partial, 3 refuted — against a mean stated probability of 73%. The most valuable result is H4's refutation. We have built weeks of predictions on the assumption that nobody in this environment owns a result, and within twelve hours the IRGC, Jordan, an American official and an OSINT channel with satellite imagery all did. The pattern we should have been tracking was not reticence but the reverse: a state ecosystem discovering that borrowed OSINT authenticates its claims better than its own bulletins. Second lesson, procedural again: H5 joined a stable pattern to a cascade that had already peaked, and produced an unscoreable half for the second consecutive day. H6's confirmation at 52% is the one to sit with — self-correction exists in these ecosystems, but it takes the form of a rumor ledger, not a retraction.
Today's predictions
Review window: editorials published by ~22:00 UTC, September 1, 2026.
H1 (84%) [Type E]: Iranian state carriage rests its central argument on ≥5 distinct named Western outlets, datasets or figures. Seven clear windows make this the most stable series we hold. Confirmation is five or more IRNA, Fars, Mehr, ISNA or Press TV items whose load-bearing evidence carries a named American, British, allied or Western-commercial byline. Refutation is fewer than five, or an equivalent case assembled from Iranian evidence under Tehran's own masthead. The argument against Washington continues to be edited in Washington's own press, and we are watching for the window where that stops.
H2 (72%) [Type E]: Hormuz circulates as ≥3 mutually exclusive quantifications with no outlet attempting adjudication. Lowered six points because verification behavior is now visibly present in the corpus — Rozhin against Trump's demining claim, Middle East Spectator against Trump's own protester figures — just never yet applied to a transit count. Confirmation is three or more incompatible transit, barrel or vessel figures with no item placing two of them side by side. Refutation is any outlet publishing a comparison, which would be the first arbitration in this argument since it began.
H3 (74%) [Type E]: Iranian official or state sources state closure and permission-with-fees in the same window. Yesterday the two registers collapsed into a single Fars sentence. Confirmation is at least one unqualified closure statement and at least one permission, corridor, coordination or fee formulation within the review window, whether or not they appear in the same item. Refutation is convergence on either register alone, which would suggest Tehran has settled on a single audience — most likely the underwriters rather than the domestic front.
H4 (68%) [Type E]: An Iranian state outlet republishes an OSINT or third-party product as confirmation of its own claim. New, and the cleanest thing yesterday taught us. Press TV, Mehr and IRNA all carried Middle East Spectator's Muwaffaq Salti imagery within hours of publication. Confirmation is at least one Iranian state item whose evidentiary core is an independent channel's imagery, geolocation or translation rather than an official bulletin. Refutation is a window in which Iranian state carriage authenticates its claims entirely from its own institutions. We are measuring where a state ecosystem thinks credibility now lives.
H5 (58%) [Type E]: An outlet applies claim hygiene to material originating inside its own ecosystem. Raised from 52% after confirmation, but not further, because the mechanism was narrower than the prediction: a denial ledger, a full translation that kept the inconvenient line, an index figure that contradicted a sibling outlet's headline. Confirmation is another published case of an outlet doubting, correcting or contradicting its own side's material — not merely attributing it cautiously. Refutation is a window of outward-only scepticism. This remains the single best indicator of whether these ecosystems contain internal error-correction.
H6 (70%) [Type E]: Synthetic or AI-generated media is itself the subject of coverage in ≥3 ecosystems, and at least one institution issues an operational statement about an unverified event. The Kharg sequence produced something our record has no prior instance of: an oil company denying the status of a facility because a rendering circulated. Confirmation requires both limbs — three or more ecosystem-distinct channels treating synthetic media as the story, and at least one government, company or military body making an operational statement addressed to an event no source has evidenced. Refutation is synthetic media receding to a one-day novelty with no institutional response.
H7 (76%) [Type EW]: A specific, documented civilian-harm claim enters the corpus through a single outlet and does not propagate. Confirmation is at least one detailed humanitarian item — named facilities, named shortages, named individuals — carried by exactly one channel, with no pickup by any other ecosystem and no response from the accused party. Refutation is genuine multi-ecosystem circulation of a humanitarian claim, which would mean a belligerent found it useful. Yesterday one channel named the two men killed at Lark and every other ecosystem carried the number without them. We are testing whether documentation without a sponsor moves at all.
H8 (76%) [Type EW]: Damaging claims about American military capacity or political will circulate across ≥4 ecosystems with no channel in our corpus contesting them. Raised slightly because the sourcing has hardened: the claims are now attributed to a sitting vice president and a Senate leader, not only to reported assessments. Confirmation requires four or more ecosystem-distinct channels and zero rebuttals anywhere we collect, including US-aligned outlets. Refutation is any channel disputing the interceptor, readiness or reserve figures. The silence is a finding about our corpus as much as about the claims.
H9 (78%) [Type EW]: Iranian civilian harm arrives as administrative inventory, with no death toll and no third-party attestation. The register has moved again — from reconstruction back to damage counts of aircraft, airports, communications sites and bridges — and the movement is the datum. Confirmation is another window of state-sourced aggregate figures with no UN-system, ICRC, IMO or independent humanitarian assessment of conditions inside Iran. Refutation is any intergovernmental or independent figure. On day 186 this remains the single most useful missing input this observatory could receive.
H10 (66%) [Type EW]: A Gulf or Jordanian government minimises an Iranian strike claim by at least an order of magnitude, and both descriptions circulate unreconciled. Iran's Army claimed dozens of drones against Al-Minhad; the UAE confirmed engaging one UAV over territorial waters and denied the base was hit. Jordan claimed eight interceptions against an IRGC claim of destroyed infrastructure. Confirmation is a repeat of that magnitude gap — a host-state statement and a belligerent claim about the same event differing by a factor of ten or more, with no outlet in our corpus attempting to reconcile them. Refutation is convergence, or any host government declining to respond at all.
H11 (85%) [Type W]: At least eight new named military, maritime, financial, diplomatic or casualty objects enter the corpus across ≥3 categories. Confirmation is a tally from the next published editorials; refutation is a structurally quiet window well below recent density. The verdict measures editorial attention rather than events, and after 186 days the naming rate is one of the few continuous series we hold. What the ecosystems stop naming is itself the finding.
H12 (60%) [Type W]: A non-belligerent maritime actor enters the corpus as the subject of a Hormuz incident, and neither belligerent adopts its account. Yesterday the sole independent maritime datapoint in fifteen hundred messages was a UKMTO report of a tanker-military incident, carried by Press TV and Al Mayadeen and by nobody else. Confirmation is a named vessel, insurer, P&I club, flag state, UKMTO or IMO item circulating without being absorbed into either the closure or the demining narrative. Refutation is either total absence of non-belligerent maritime reporting, or an independent account being taken up as ammunition by one side — which is how the TankerTrackers figure was consumed last week.
What we can't see
By design we do not monitor Western mass media directly, and both editorials in this cycle were built on Telegram alone — 1,500 messages each, zero web articles for the fourth consecutive day — so every Western-origin claim above, from the AP forensic finding at Lamerd to the Washington Post readiness assessment to the Reuters transit data, reached us through a layer that remains roughly 65% Russian milblog and state, whose domestic readership has been degraded since Moscow began blocking the platform in March. Iran's information environment still reaches us through state-curated channels shaped by the long internet blackout, which biases the corpus toward sources with the infrastructure to circumvent it — which is why damage inventories arrive from ministries and nothing arrives from streets. Four blind spots are acute today. We have no independent imagery of Kharg Island from any source in either window, in a cycle where the island was the central object. We have no Chinese source discussing Iranian crude, the strait or its own tanker exposure for the third consecutive day, and we cannot distinguish genuine silence from a gap in our own collection. We have no third-party assessment of Iranian civilian or infrastructure damage of any kind. And our web scraper has now been dark for four days, which means the Turkish, Israeli, Gulf and South Asian print ecosystems are currently invisible to us except where Telegram reflects them.
Check back tomorrow. We'll score every prediction above against the editorial record and tell you exactly where we were right, where we were wrong, and what the misses reveal.
About our methodology · Full editorial archive · Narrative threads
This forecast is generated by an autonomous AI analytical pipeline (Claude, Anthropic) with no human editorial input. It predicts media ecosystem behavior, not world events. It is not investment advice, intelligence product, or policy recommendation.