The board's heaviest tension axis rests on four headlines, two of which carry one claim
2026-07-19 · partially corroborated · released 2026-07-22
What happened
At 02:06 UTC on 18 July a single radar scene over the Hormuz box returned 195 contacts, 142 placed near Ras al Khaymah and 53 near Bandar-e-Abbas; 100 of them measure one pixel. Washington DC entered the board at 06:30 through the GDELT stream at 16 sources. The axis's own four evidence items followed. At 08:37 Middle East Eye reported that the IRGC claimed a missile and drone attack on a US base in Jordan. At 08:46 Daily Sabah placed Iranian attacks on Gulf allies against a seventh straight night of US strikes. At 09:18 TASS reported the same IRGC claim, sourced to an agency. At 10:55 a monitored PressTV Telegram channel carried commentary on Western coverage of Iran. Between 10:21 and 12:08 a thermal batch returned 264 fire detections: 132 in Ukraine, 44 in Russia, none in Iran.
What it means
The axis is the heaviest of the six relations and rests on a base that is exactly as narrow as every other relation's: four items each, all selected the same way. Two of them relay the same IRGC claim about a base in Jordan, which has no node here. Three of the four are already filed under other entities. The manifestation set is one-sided: of four connected events at weight 0.8, two sit on Iranian soil and one is the strait itself, a shared waterway; only Washington DC lies on the American side, at intensity 0.32 and info severity. The radar layer is built the same way, 195 contacts sharing seven distinct reports inside 60 kilometres, none of them contemporaneous with the scene. What the data does not support is any statement about what either party did. Repeating one claim is not corroboration, and Reuters and AFP, named on the events this axis connects to, are absent from its own evidence. No attribution here, no forecast.
What we're watching
First, whether the Middle East Eye item gains a second source. It is the only one of the four unique to this axis, and Jordan, where the reported base sits, carries no node on the board. Second, whether the volume gap widens. Over thirty days Retina puts Iran up 36.11 percent and the United States up 4.19, while classifying both directions as stable; the next pull tests which of the two readings moves. Third, whether the weight of 0.94 holds. It is recomputed over a rolling 24-hour article set, so the next pull tests whether it holds; how it responds to a day without a shared headline is not something the method line states.
Based on an axis weight of 0.94 against a six-relation mean of 0.84; four co-occurrence items on 18 July between 08:37 and 10:55 UTC, three also filed elsewhere; four connected events at weight 0.8, two on Iranian soil; 264 fire detections in the 18 July 10:21-12:08 batch, none of them Iranian — the live layer carries 69.