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 14, 2026
Day 199 of the US-Israeli air campaign against Iran. Hours 4731–4755 since first strikes at ~06:10 UTC, February 28.
This forecast covers editorials #659 through #660, published at 10:09 and 22:08 UTC on September 13. 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 freight rate, a displacement figure or a diplomatic meeting, 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, CNN, AFP, Guardian, Bloomberg, Financial Times, Fox News. They reach us only as reflected, amplified and reframed by the ecosystems we study, often several mirrors deep. About our methodology.
Where we are
The defining event of the day was a meeting that did not happen, and three ecosystems distributed the agency differently. Oman's foreign minister announced that Monday's Salalah ministerial was postponed "in the interest of consensus," a formulation reproduced without elaboration by Xinhua, TASS, Rudaw and TRT. Iran's MFA Persian Gulf director, via IRNA, said the delay came at the request of some regional states as a joint Tehran-Muscat decision; Fars, relayed by TASS, compressed it to a bilateral agreement. In none of the three accounts is the objecting party named. Before the collapse, ISNA had already published a piece alleging that actors outside the region were working to disrupt the meeting by instrumentalising Bahrain — pre-positioning an explanation the Iranian ecosystem could reach for. Ahead of it all, Araghchi had set the agenda as a new maritime route and the condition for any reopening as a US return to its Islamabad memorandum commitments. Follow Negotiations & Ceasefire Diplomacy.
One advisory of unknown origin produced five mutually exclusive resolutions in under a day. UKMTO reported a vessel struck by a projectile of unknown origin. Almasirah supplied a Panama-flagged tanker and a missile; IntelSlava supplied a vessel name, Riyan Star, that appears nowhere else in our corpus, and an IRGC attribution; Iranian state media moved from "terrorist enemy" through "unidentified" to "details of the US attack" across three hours, arriving at a self-propelled underwater missile; abualiexpress told Israeli readers in passing that the US had destroyed five Iranian tankers, uncorroborated anywhere. Outside that ecosystem the hedge held — Anadolu kept the scare quotes, Haaretz attributed throughout to Iranian media, India's MEA condemned an attack on a commercial vessel and named no actor. The escalation consequence was supplied by an adversary account: abualiexpress observed that if Iran's own rendering holds, then by the IRGC spokesman's published retaliation equation Iran now owes ten American targets. Late in the window IRNA reported unexplained explosions at Sirik. Follow Strait of Hormuz & Oil.
The Mecca defence pact is being read by everyone for what it omits. Pakistan's prime minister condemned attacks on Saudi oil infrastructure and affirmed support for the kingdom in a call with the crown prince; Xinhua framed the same call as de-escalation diplomacy. Then three mutually hostile ecosystems converged on one reading: Farsna headlined that Turkey "limited itself to condemning" and that "Pakistan too only condemned," while abualiexpress reported Shia-axis channels mocking the alliance by quoting the Pakistani statement verbatim. Underneath, the attribution for the Yanbu pumping-station fire is itself contested — cig_telegram ascribes it to Houthi missiles, Long War Journal reports Riyadh closing the line after attacks from Iraq, and Iraq's presidency spokesman condemns the attacks without addressing the territory. Follow Regional Focus: Iraq.
Two structures are worth naming because neither has a news organisation in it. Trump told reporters the diesel shortage stems from Russia and Ukraine; US average diesel crossed six dollars a gallon; Iran's parliament speaker posted a Simpsons clip of a price sign unable to display the figure; an Israeli OSINT channel rebutted it within the hour. And Iran is running an outward archive and an inward restriction simultaneously — the judiciary citing 126,600 documented instances of war damage and 4,500 UN submissions, a court hearing a suit by nearly 353,000 people, while Tehran's prosecutor charged the host, producer and two guests of an online programme over economic claims and Mehr ran its standing enumeration of "the lies published in the past 24 hours."
Yesterday's scorecard
These predictions were written for editorials published by ~22:00 UTC September 13. The hit rate below is not discounted.
| # | Prediction | Type | P | Verdict |
|---|---|---|---|---|
| H1 | Pipeline attribution stays open; joint inquiry produces procedure | E | 72% | Partial. The attribution limb held — Yanbu was ascribed to Houthi missiles by one channel and to attacks from Iraqi territory by another, with Baghdad condemning without addressing origin. The inquiry limb failed: the joint Iraqi-Iranian investigation generated nothing at all in this window. We predicted a holding pattern and got silence, which is a different object. |
| H2 | Claim hygiene applied inside an outlet's own ecosystem | E | 72% | Confirmed, eighth window, in two forms. Farsna, an Iranian outlet, explicitly noted the projectile type was not yet known while Press TV was already publishing a weapon and an author; IRNA said the source of the Sirik explosions was unknown. Separately abualiexpress accused Israel's own Channel 13 of false reporting and mobilised a platform correction against it. |
| H3 | ≥3 parties publish incompatible counts of the same contested space | E | 78% | Partial. Incompatible arithmetic was everywhere — 76,000 against 85,000 against 86,000 displaced, 352,983 against "over 302,000" claimants, injured counts drifting from three to four on one hull. But the specific structure we named, belligerents counting the traffic each claims to govern, did not recur; the transit-count contest went quiet and the divergences migrated to casualty and claimant ledgers. |
| H4 | Yemen's civilian ledger stays ecosystem-sorted | E | 72% | Confirmed, eighth window. The UN's 76,000 and IOM's ~85,000 reached us through BBC Persian, TRT World and ajanews; the most textured account of the flight to Djibouti came from rybar_mena, a Russian military channel, and from neither belligerent's media. |
| H5 | Iran talks the Oman track down while Gulf media talk it up | E | 70% | Partial. The deflation limb confirmed emphatically — Tasnim said the understanding "by no means" opens the strait, Araghchi said it in his own voice, AzerNews counted seven conditions. The amplification limb came from the wrong place: the only outlet filing it as settled fact was Xinhua, republished by Global Times under the same byline. We named Gulf press and got Chinese syndication. |
| H6 | The negotiate-or-not argument continues on state-media op-ed pages | E | 66% | Refuted on its own terms. Our stated refutation was the op-ed track going quiet, and it did — no signed factional commentary appears in either editorial. The argument itself did not stop; it moved venue, running through named officials instead, with Pezeshkian holding the NPT, a parliament spokesman calling exit attainable and a clerical assembly demanding it. We predicted a venue. |
| H7 | One document harvested selectively by ≥3 ecosystems, omission as position | EW | 74% | Confirmed twice. The Pakistani call readout reached us as Saudi affirmation, Chinese de-escalation diplomacy, Iranian insufficiency and Israeli-relayed mockery. Cleaner still: ajanews carried the Arab League secretary-general calling Iran's regional role extremely negative, and IRNA carried the same interview running only his line that there is no military solution. |
| H8 | An admission crosses ecosystem lines and is amplified by the side it embarrasses | EW | 70% | Confirmed. Iranian outlets proved Western pain by quoting Westerners — Bloomberg's supertanker earnings and the New York Times front page on chokepoint control, both via Mehr. And abualiexpress amplified the IRGC's own published retaliation arithmetic, holding a belligerent to a number its own spokesman issued. |
| H9 | Iranian dual register, and an outlet makes the contradiction its story | EW | 68% | Partial. The register split was abundant — "Iran will not surrender" alongside "we are not at war with Saudi Arabia and the Houthis have their own issues" to India Today, and three incompatible nuclear positions in one day. The second limb is thin: abualiexpress forwarding Pezeshkian to Israeli readers with the one-word verdict "liar" is a judgement rather than a piece about the split. |
| H10 | The AI-governance bridge holds across ≥2 ecosystems | EW | 62% | Partial. One instance, one ecosystem: Al Arabiya covered the BRICS summit through China's open-source AI platform rather than the declaration's language on strikes against IAEA-safeguarded sites. That is exactly the bridge we described, and it did not appear anywhere else in the corpus. |
| H11 | ≥8 new named objects across ≥3 categories | W | 82% | Confirmed comfortably. Riyan Star, Hengam Island, Sirik, 47 rising to 150 Rosatom specialists at Bushehr, Yanbu's five-to-seven-day export stocks, Khab al-Shaaf market, Shada and al-Dhahir, 126,600 documented damage instances, 4,500 UN submissions, 66,000 wounded treated, a 313,000-person rally, 58 Saudi airstrikes, 322 released detainees, the Ali al-Taher tunnels and 48 Hezbollah members, six-dollar diesel, 70 interceptors per intercept. |
| H12 | Two named officials of one government give incompatible accounts | W | 60% | Partial. The contradiction we found is normative rather than factual — Pezeshkian arguing NPT membership proves intent against a parliament presidium spokesman calling withdrawal attainable. The cleanest factual contradiction in the window, Mehr reporting the Shalamcheh crossings closed nine minutes before IRNA reported them reopened, came from state media rather than named officials. |
Summary: 5 confirmed, 6 partial, 1 refuted — against the same mean stated probability of 70% that produced 7 confirmed yesterday. Yesterday's lesson was to stop bundling second conditions; today's is narrower and more useful. Four of our seven misses failed on a named carrier or a named venue — Gulf press rather than Chinese wire on H5, op-ed pages rather than officials on H6, a transit ledger rather than a casualty ledger on H3, an investigation rather than its silence on H1. The behaviours we described all occurred. Our probabilities appear to be roughly ten points too high whenever we specify who will perform them, and today's set names behaviours and leaves the carrier open.
Today's predictions
Review window: editorials published by ~22:00 UTC, September 14, 2026.
H1 (74%) [Type E]: The Salalah postponement circulates without the objecting party being named. Three accounts of who stopped the meeting are already in the corpus, and all three use an unnamed subject: consensus, some regional states, a bilateral decision. Confirmation is further coverage of the postponement or a rescheduling in which the blocking state is described only by category or by innuendo — outside actors, regional states, Bahrain as an instrument — with no outlet naming and sourcing an objector. Refutation is any outlet naming the state and attributing the claim, or a convening announcement with a published guest list. We are watching whether an absence can stay load-bearing for a full cycle.
H2 (74%) [Type E]: An outlet applies claim hygiene to a central claim originating inside its own ecosystem. Eight windows positive. The hedge must attach to something the window is leaning on — an attribution, a casualty figure, a damage assessment — and come from inside the ecosystem that benefits. Confirmation is an outlet flagging missing confirmation, publishing a deliberately conservative reading, relaying a report while arguing against it, or correcting itself to its own audience. Refutation is a window of outward-only scepticism.
H3 (76%) [Type E]: Specificity about the Hormuz hull increases while the evidentiary base does not. The sequence so far runs from unknown projectile to unidentified to enemy to US attack to a named weapon type, with each addition made downstream of the first report rather than supplied by it. Confirmation is at least one further escalation of forensic detail — platform, munition, launch point, actor — published without any new physical or technical evidence entering the corpus. Refutation is the publication of debris imagery, a salvage or insurer finding, or a walk-back. This tests the manufacture of attribution, not what struck the ship.
H4 (72%) [Type E]: Yemen's civilian ledger stays ecosystem-sorted. Confirmation is UN, IOM or UNHCR displacement and casualty findings reaching us through Turkish, Persian, Qatari, Russian or agency carriers while remaining absent from the channels celebrating the advance that produced them; Sanaa health figures confined to resistance-aligned outlets; Saudi civilian harm appearing as single lines. Refutation is any outlet carrying the ledger that indicts its own side. Follow Global South & Middle Powers.
H5 (72%) [Type E]: A support pledge is read for its omissions by at least two ecosystems that agree on nothing else. Iranian and Israeli-aligned channels reached the identical verdict on the Pakistani and Turkish statements by quoting them verbatim. Confirmation is a further alliance statement, condemnation or readout being carried as evidence of insufficiency by outlets from two or more mutually hostile ecosystems, with the quotation itself doing the work. Refutation is alliance texts circulating as their authors intended, or only one ecosystem performing the excision.
H6 (66%) [Type E]: Iran's documentation apparatus publishes an aggregate war-harm figure that does not reconcile with one its own system published this week. We already hold 352,983 claimants against "over 302,000," 126,600 documented instances and 66,000 treated wounded. Confirmation is two intra-Iranian counts of the same object diverging inside the corpus, whether across outlets or across institutions. Refutation is convergence on a single figure, or no new aggregate appearing. The archive is being built faster than it is being reconciled, and the divergences are observable.
H7 (74%) [Type EW]: One institutional text is harvested selectively by ≥3 ecosystems, with at least one carriage's position visible in what it cuts. Two clean instances yesterday. Confirmation is a single text — readout, declaration, interview, monitor's assessment — reaching us in three or more non-overlapping cuts, with an identifiable excision functioning as an editorial position. Refutation is a document travelling with its content intact across the ecosystems that carry it.
H8 (70%) [Type EW]: A number, admission or formula published by one side is amplified hardest by the side it embarrasses. Iranian outlets proving Western freight pain with Western financial press; an Israeli channel holding the IRGC to its own retaliation arithmetic; a Saudi-aligned force publishing its own 500 dead. Confirmation is attributed carriage of an opposing ecosystem's self-damaging material given prominence rather than a passing mention. Refutation is a window where all cross-ecosystem transfer runs through neutral wires.
H9 (64%) [Type EW]: A consumer price indicator completes a full circuit with no news organisation in the chain. American diesel crossing six dollars became an Iranian legislative meme became an Israeli OSINT rebuttal within hours. Confirmation is another economic indicator — fuel, freight, exchange rate, food price — travelling from an official or legislative account to a partisan monitoring account and back, with the entire relay visible in our corpus and no wire service in it. Refutation is economic material moving only through outlets.
H10 (66%) [Type EW]: Iran's outward harm archive and its inward restriction run in the same window, and no single ecosystem carries both. The judiciary's documented-damage totals and the prosecution of an online programme's host and guests were published by the same state system on the same day. Confirmation is both registers present, with the archive amplified by resistance-axis and Global South carriers and the enforcement action reaching us mainly through Persian-language external outlets. Refutation is one ecosystem publishing both halves together, or either half going quiet.
H11 (82%) [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. After 199 days the naming rate is one of the few continuous series we hold, and a fall in it would itself be the finding.
H12 (62%) [Type W]: An official denies a shortage whose underlying report is absent from our corpus. The CENTCOM commander reported no concern about munitions shortages, and the White House dismissed a Washington Post ammunition story we never saw natively. Confirmation is a further official denial of scarcity — interceptors, ordnance, fuel, spare parts — whose object reaches us only through the denial and its adversarial relay. Refutation is a window in which shortage claims and their rebuttals both arrive with their sources intact. We are measuring the shape of our own blind spot, which denials make visible.
What we can't see
By design we do not monitor Western mass media directly: the Bloomberg freight data, the Reuters Yanbu stock estimate, The Economist on Ukrainian Patriot depletion and the Washington Post ammunition story all reached us yesterday through belligerent-aligned intermediaries, each keeping a different half. Our Telegram layer remains roughly 65% Russian milblog and state, with domestic readership degraded since Moscow began blocking the platform in March, and Iran's information environment reaches us through channels curated by the long internet blackout, biasing toward sources with the infrastructure to circumvent it. Three blind spots are acute today. We have no forensic layer on the Hormuz hull — five attributions, no debris, no salvage report, no insurer — and we cannot distinguish a technical finding from a diplomatic decision. We have no shipping-industry layer at all, which is why incompatible transit and damage counts sit in our corpus with nothing to arbitrate them. And we cannot see the Salalah room: we know a meeting was stopped, we have three accounts of who stopped it, and every one of them declines to name a state.
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.