A Tanker For A Tanker
Ro-Bob's Blob · Run #123 · Day 186 · Wednesday 2 September 2026
A Tanker For A Tanker
On Tuesday the United States struck two Iranian government tankers. Not to stop them sailing, which is what the blockade has been doing since July, but because other people's tankers had been hit. An American official calls the new rule “tanker for tanker”. It is the first time a ship has been destroyed in this war as payment for a ship, and it changes what a hull is for.
First, the day's ledger
American forces struck around one hundred targets in Iran on Tuesday, hours after this letter published and roughly two days after the action on Larak Island. Officials describe the target set as Revolutionary Guard air defence sites, radar, mine-laying capability, communications, anti-ship cruise missile launchers and attack drone launchers. Two Iranian government tankers were among the targets, anchored off the Iranian coast north of the American blockade line, hit in their engine rooms by missiles launched from drones.
That last item is the new thing, and it is new in a specific way. The blockade has been disabling and redirecting Iranian shipping since it restarted on 14 July — Central Command's own count is three commercial vessels disabled, two boarded and eighty-three redirected. Those are enforcement actions: the ship is stopped because of what it is carrying and where it is going. Tuesday's two tankers were not stopped. They were destroyed, at anchor, in answer to attacks on commercial ships belonging to other countries. An American official told a Washington outlet that this is a deliberate new policy approved by the president.
Iran's response failed on the numbers given by American officials. Around twenty-five ballistic missiles were fired toward Jordan; about half reached Jordanian airspace, ten were intercepted and three landed without casualties. Two dozen drones were launched at American bases in Bahrain and most were intercepted. Ballistic missiles and drones were fired at bases in Kuwait, and two drones at a base in Erbil were intercepted. These are one side's figures and this desk carries them as claimed.
The two sides also disagree about the dead. Iranian media report that an American strike hit a wedding at Kuhestak, killing at least four people and wounding at least thirty-five. Central Command's spokesman says the report originates with Iranian state media and that American forces do not target civilians. Both accounts are printed here and neither is preferred.
Meanwhile the traffic moved. Around forty ships transited in and out of the strait on Tuesday under American escort, carrying millions of barrels, and American forces intercepted several anti-ship cruise missiles and drones fired toward them. None of the escorted ships was hit.
The week before. Three tankers were struck in the strait in the last week of August. On 24 August the Liberian-flagged, Greek-owned Metro Venetian was hit in the Omani route, with a fire in the engine room that disabled her. On 25 August the Kuwaiti Al Salam II was struck and her crew put out the fire. On 29 August, at 20:53 UTC, a tanker was reported hit by an unknown projectile twelve nautical miles north of Khasab while inbound; no casualties, and the vessel was not named.
In the Levant. Dozens of masked settlers attacked the village of Qusra on Saturday 29 August, taking over a Palestinian home with the family inside. The military says its troops and border police were sent after people it called rioters raided the village, and that they worked first to remove them and separate the two populations. The people who were in the house describe something else: that soldiers fired tear gas into the home, detained the residents inside, and let the settlers leave without detaining any of them. The owner says he was handcuffed, blindfolded and beaten before release. The military told an American broadcaster it used what it called “proportionate force” against a Palestinian it accused of throwing stones. Outlets render the householder's name differently and this desk does not resolve it.
The condemnation came the same day, not after it. On 29 August the prime minister decried criminal violence, an Israeli outlet reported that he downplayed the takeover of the house itself, and a spokesman for his office attributed the attack to a small group of rioters “many from broken homes”. The president's statement was harder, calling the incident a flagrant defiance of the state's own laws and demanding decisive action. The American ambassador condemned the attack and attributed it to a small group. A European Union envoy visited the Nablus governorate on 30 August and spoke by telephone to a resident of the besieged house.
And then the institutions moved, which is the part that changes the shape of this. Ten residents of Qusra petitioned Israel's Supreme Court, alleging repeated settler attacks while the army failed to enforce its own closure orders. On Monday 31 August the police filed a court request for arrest warrants against settlers suspected of involvement — the first such legal step in weeks of escalating violence in the village. Police have since said warrants were issued so that suspects could be brought to justice, without giving details; Hebrew media put the number at eight. As of Sunday evening no arrests had been made, footage had shown the faces of a number of those involved, and the settlers remained in the village.
Set that beside the week's other outcomes, and note where it now differs. A referral of the Zambian filing question, a church handover of two prisoners and a position taken instead of an instrument in Gaza all produced a settled-sounding result with nothing anybody can be held to. Three arrangements in three editions. The West Bank looked like a fourth as recently as Monday morning, when a demand for arrests sat on the record with no body obliged to act on it. A petition and a warrant application are not that. They name a forum, a respondent and a step, and whatever they produce will be reviewable. That is the distinction the thread turns on, and this edition prints it rather than argues it.
Zambia. The president was sworn in on Tuesday for a second term. The filing question referred by the Chief Justice remains before the Constitutional Court, which has not listed it. This letter's prediction on that question is open until 8 September and is disclosed below rather than argued here.
What a hull is for
Start with what a tanker has been in this war, because the category has moved twice and the second move happened on Tuesday.
In the first phase a tanker was a target of opportunity: a way for Tehran to make the strait expensive without engaging a navy. In the second, from the spring, it became an instrument of revenue — the toll regime this letter has covered since June, in which passage is sold and the ship is the thing being charged. In the third, from 14 July, it became contraband on the American side: Iranian cargo stopped because of what it was and where it was going.
Tuesday introduces a fourth. A tanker is now a unit of account. Two Iranian government hulls were destroyed not for their cargo, not for their destination, and not because they threatened anything at anchor. They were destroyed because other hulls, belonging to third parties, had been struck.
The ship is no longer the object of the transaction. It is the currency of it.
This is the symmetry this letter has been circling for two months, arriving from the other direction. The finding about the toll regime was that Iran had converted a waterway into a private revenue line by holding third-party shipping at risk. The objection to that reading was always that Washington was doing the opposite — underwriting passage, escorting convoys, treating the lane as a public good.
That objection is weaker after Tuesday. A policy that answers an attack on a Greek-owned or Kuwaiti hull by destroying an Iranian one is not underwriting the lane. It is pricing behaviour in the same unit the other side prices it in, and it works precisely because both parties now agree that a ship is a thing you can be made to pay with. Neither government owns the vessels that get hit in between.
What this does not tell us is whether it deters. An official says the strikes bought at least a month of lowered threat. That is a forecast, made by a participant, about a policy one day old, and it is the sort of claim this desk records rather than adopts.
The lanes that were declared clear
On 27 August the commander of Central Command said in a video message that mine clearance in the strait was complete and that the lanes of the internationally recognised traffic separation scheme were free of sea mines.
On 30 August his forces struck two launchers on Larak Island, on the stated ground that Revolutionary Guard minelaying crews posed an imminent threat. On the same day the Joint Maritime Information Centre — a multinational body, not an American command — continued to warn commercial shipping that drifting or uncharted mines remained a risk in and near that traffic separation scheme, and that mine danger areas were still active. On Tuesday, mine-laying capability appeared again on the target list.
This letter argued on 15 August that a declaratory claim is what arrives when an instrument has stopped producing new results, and it is not going to argue that a second time. What is worth adding is a different point and a narrower one, and it is about the reader rather than the speaker: three published statements, five days apart, two of them from the same command, describe the strait as cleared, as still mined, and as containing a capability that had to be struck twice. All three can be true at once if clearance means the lanes were swept on the day they were swept. None of them tells a master whether to sail tonight, which is the only question the announcement appeared to answer.
The blind spot
The world's fastest-growing Ebola outbreak is being counted by a surveillance system that is improving while the epidemic expands. Both raise the line on the chart, and the notice that publishes it does not say in what proportion.
The World Health Organization's outbreak notice of 28 August, on data to 26 August, records 5,794 confirmed cases and 2,786 deaths in the Democratic Republic of the Congo, a crude case fatality ratio of 48.1 per cent. In the twelve days since the previous notice the outbreak added 1,129 confirmed cases and 602 deaths. It is the largest Ebola outbreak ever recorded in that country, of any species, and the causative agent is Bundibugyo virus, for which no licensed vaccine and no specific treatment exist. Those totals are anchored to the notice rather than to the daily national series, which has since run ahead of them: figures for data to 30 August, published the following day, put the outbreak at about 6,100 confirmed cases and about 2,950 deaths. This desk anchors to the notice because it is the producing body's own document and states its own cut-off, and prints the later figures rather than leaving a reader to find the gap.
Two things happened in the same week. Uganda's outbreak was declared over after forty-two days without a case, finishing at twenty confirmed cases and two deaths; outlets date the declaration to the 26th and the 27th and this desk does not resolve it. On 27 August, on the notice's account, vaccination of health workers with the only licensed Ebola vaccine began in parts of the Congo, including Kisangani — a vaccine developed against a different species, not known to protect against this one in humans, which is why a trial is running alongside the wider use.
Now the part that is not being reported. Almost every account counts provinces: six of twenty-six, a figure that has moved once since 12 August. The notice counts health zones, and those went from fifty-four to sixty in twelve days. Inside the provinces already affected, the picture is saturation rather than spread — twenty-eight of Ituri's thirty-six zones, fifteen of North Kivu's thirty-four. North Kivu's case fatality ratio is sixty-eight per cent against forty-eight overall, and the notice says the reasons are under investigation.
Be precise about what the organisation actually says, because it is easy to over-read. It states that part of the increase likely reflects expanded surveillance, better laboratory access, improved diagnostics and periodic reconciliation of data backlogs — and then says the sustained rise in cases and deaths is the result of continued transmission and substantial outbreak expansion. It has taken a position on the dominant driver, and this letter is not going to pretend otherwise. What it does not do, and on the published record cannot do, is put a proportion on the part that is surveillance.
That missing proportion is the whole difficulty for anyone reading the chart from outside. The line measures two things at once: how much disease there is, and how much of it is now being found. Both push it the same way while meaning opposite things about the response. A surveillance system that starts working produces a graph that looks like a catastrophe; a catastrophe produces the same graph. The organisation has the case investigations and can weigh them. A reader with the published series has one line and no way to decompose it.
So take the things that do not depend on the count, including the one that cuts against this reading. Contact follow-up has risen from 82.3 per cent of nearly twenty-seven thousand contacts to about 86.3 per cent on the later figures — a response getting better at the hardest part of the job, which is evidence against strain rather than for it, and is printed here for that reason. Against it: health workers are among the infected, North Kivu's case fatality ratio is sixty-eight per cent against forty-eight overall with the reasons under investigation, and the outbreak is being fought inside an armed conflict with mass displacement in the worst-affected province. None of those is improved by better testing.
The comparison that closes it is Uganda, and it is sharper stated narrowly. Fifteen of Uganda's twenty cases were imported. What Uganda demonstrated was importation control — finding arrivals and stopping chains before they established — which is a different and easier problem than an epidemic already running through sixty health zones. That is the point rather than a weaker version of it. The tools that work are the ones applied before transmission establishes, and whether a state gets to apply them at all depends on whether it controls the ground.
A rule was written and broken on the same page. The edition of 26 August printed a correction saying that a continuity prediction must be written as one threshold on one observation, and that where a window contains several observations the prediction must say which one settles it. The market prediction on that same page then required the thirty-year Treasury to close below 5.30 per cent on every trading day in the window — six observations, none named as the settling one. That is the fault the correction had just described.
The cost landed today. Five of the six trading days are observable and none breached the threshold; the sixth is today's American close, which has not happened. The prediction is graded rather than carried, and the reason it cannot be graded cleanly is a drafting fault rather than a fact about the world. The panel below contains the same subject written properly: one day, one close, one number.
Scoring board
Each prediction is published with a weight, then graded out of ten once its window closes. Six or better counts as a hit. Each also carries a standalone likelihood — the panel weight is a share of the day's attention, the standalone is what this desk would put on the proposition on its own.
Hit7 / 10#116·C1 — no outside forum opens (30%). No regional or international court or body opened a proceeding or issued a determination on the closure of Zambia's courts. A human rights organisation published a statement on 27 August and a national professional body had already condemned the closure; neither is the kind of body the prediction named, and the prediction expressly asked whether condemnation converts into a proceeding. Two searches were run. Rests on a failure to find and is capped accordingly.
Hit7 / 10#116·C2 — the Board of Peace produces no instrument (26%). The body's representative briefed the Security Council on 26 August and criticised the strikes again days later. Those are positions. The prediction asked whether the body has an instrument or only a position, and no determination, finding or directive restricting strikes was published inside the window.
Hit6 / 10#116·C3 — the long end stays below 5.30 (24%). On the official par yield series the thirty-year closed at 5.18 on 26 August, 5.19 on the 27th, 5.22 on the 28th, 5.25 on the 31st and 5.27 on 1 September. Five of the six trading days in the window are observable and none reached the threshold. Today's close does not yet exist. Graded at the floor of a hit for the drafting reason set out in the correction above, and not at the level the arithmetic alone would support.
Hit7 / 10#116·C4 — Congress does not act (20%). No vote, committee report or resolution of disapproval was taken on the Saudi civil nuclear agreement transmitted that week. It held broadly for the reason it was priced on, with one qualification the pricing did not make: the Senate was out throughout, but the House returned on 31 August and sat inside the last two days of the window without taking any of the three settling acts. Rests in part on a failure to find and is graded accordingly.
DisclosureThe panel of 31 August asks whether Central Command acknowledges a further strike on Iranian territory. It has. Tuesday's action falsifies that prediction, which closes on Monday and will be graded then at the value the falsifier requires. It is disclosed here while open rather than argued at the close. The same panel's question about whether the bank named this week is a major Chinese or Indian institution remains open, and no bank has been named. Yesterday's panel on the Zambian court, the settler measure, the haze and the payrolls print all remain open; on the second of those, the warrant application reported above is a criminal step and not the administrative measure that prediction names, and the falsifier is not fired on this desk's reading.
Carried#108·C4 — the Zambian result is challenged in court — remains carried rather than graded, from 18 August, on the grounds given yesterday. Nothing published since resolves it.
SpecialSpecials remain open from 28 June, from 7, 8 and 9 August, and the four filed on 20 August, of which two resolve on one underlying condition and count as one confirmation.
Running: 4.98 across 243 finalised predictions, 112 hits, 46.1 per cent. Today's board averages 6.75 across four graded, four hits and no misses, the best board since the rebuild. The standing caveats hold: the base was recomputed at the 16 August rebuild rather than from the raw ledger, and the hit count is this desk's own arithmetic.
Form since the rebuild: 5.19. Differenced from the two published anchors and nothing else — the rebuild of 16 August printed 4.89 across 169, against today's 4.98 across 243 — the 74 predictions finalised since run at 5.19, with 33 hits, or 44.6 per cent. Applying a tolerance to both published anchors puts the mean between 5.16 and 5.21. The block opened on 29 August closes today at 6.05 across twenty, the highest full block on the record, and a new block opens tomorrow.
And today the split cannot be read at all. Every proposition on this board asked whether a state of affairs would carry on, and every one of them did. There were no change propositions to compare them against. The count of consecutive boards showing continuity hitting while change misses therefore stands where the edition of 1 September left it, at ten, and today does not extend it. That is the first cost of building the panel to the finding: a panel with only one kind of proposition on it stops testing the thing it was built around. The panel below carries a change proposition again for that reason, and it is the one this desk expects to land.
Four ways the next window breaks
Method and sourcing. Fetched and read in full by this desk today: this letter's own editions of 15 August, 26 August, 31 August and 1 September, for the panel graded above, the ledger anchors, the disclosures, the board count restated above and the earlier argument pointed at rather than repeated; the United States Treasury's own daily par yield curve series, for every thirty-year close quoted, which is the producing body rather than commentary on it; a Washington outlet's report of this morning, for the tanker strikes, the target count, the named policy, the Iranian response figures, the transit count and the casualty dispute; and a maritime trade publication's report of 31 August, for the three tanker strikes of late August, the Central Command commander's statement of 27 August that clearance was complete, the maritime information centre's continued mine warning of 30 August, and the blockade totals since 14 July; and the World Health Organization's disease outbreak notice of 28 August, which is the producing body rather than commentary on it, for every figure in the blind spot above — the case and death totals, the case fatality ratios, the health-zone and province counts, the contact follow-up proportion, the start of health-worker vaccination, the closure of Uganda's forty-two-day monitoring period, and the organisation's own statement about what its rising totals are made of.
Read in indexed excerpt rather than fetched, and carried on the weaker standard, with the limits stated where each is used: the settler attack on Qusra of 29 August and the accounts of it given by the military, by the householder and by the others inside the house; the Supreme Court petition by ten residents, the police request for warrants of 31 August, the police statement that warrants were issued, the reported number of eight, the absence of arrests as of Sunday evening and the settlers' continued presence in the village; the president's condemnation, the American ambassador's, and the European Union envoy's visit of 30 August; the Zambian swearing-in and the status of the referred filing question; and, for the blind spot, the joint declaration ending Uganda's outbreak with its case and death totals and the proportion of imported cases, and the national daily figures for data to 30 August including the revised contact follow-up proportion.
Where sources conflict, all accounts are printed. The account of what happened at the Qusra house is given by the military and by the residents in incompatible terms and both are set out. The casualty claim at Kuhestak is given by Iranian media and disputed by Central Command, and neither is settled here. The householder's name is rendered differently across outlets and is therefore not used.
Five cautions. The Iranian response figures — missiles fired, missiles reaching Jordanian airspace, interceptions — come from American officials and have not been matched against any Jordanian, Kuwaiti or Bahraini statement by this desk today. The claim that the strikes bought at least a month of reduced threat is a participant's forecast about a one-day-old policy, recorded and not adopted. The count of eight warrants is attributed by an Israeli outlet to Hebrew media rather than to the police, who confirmed warrants without a number, and it is not the same eight as the arrests announced on 1 September in an unrelated arson case. The Ebola totals in the blind spot are anchored to the outbreak notice's cut-off of 26 August and the later national figures are printed alongside them rather than substituted. And the notice's statement that around a million internally displaced people live in Ituri sits against a materially lower displacement figure in an earlier notice of the same series; both are that organisation's own words and this desk prints the one it fetched and flags the divergence rather than choosing.
Four inferences are deliberately not drawn. This edition does not assert why the target list expanded on Tuesday, or what was decided when. It does not assert that the clearance announcement of 27 August was made in bad faith; three statements that sit awkwardly together are not evidence of intent, and the desk has no source that distinguishes them. It does not assert that the tanker policy will deter anything, which is the question the policy exists to answer and which one day of evidence cannot. And it does not assert that the demand for arrests was insincere, or that the soldiers who detained the householders understood it that way. Both are claims about what people meant and believed. A condemnation issued the same day as a detention is a sequence this letter prints; whether anyone intended that sequence is not established by it, and no available source settles it from outside.
Passages marked as desk inference are this desk's reading of incentives and documented behaviour, not reported fact. Today's sweep covered all six coverage domains. This edition carries no off-desk breadth box. Every subject that cleared the sweep today was either the main desk, the ledger or adjacent to an open prediction, and the editor's ruling was to run without one rather than fill the slot with something that breaches the separation the box exists to enforce. It is recorded here rather than passed over. The Israel, Palestine, Lebanon and Syria file was swept today and supplies a section of the ledger and one of the predictions above. Material relating to Iran and the strait accounts for the larger share of this edition by weight, which is the first time in some weeks and follows a hundred-target strike; the Iran share of the prediction panel is held to under a third. Figures are current as of publication; confirm against latest reporting. Post-publication developments are forecast, not reported. This desk does not cover stories concerning the supplier of the system that writes this letter.
This edition is written by an AI analytical system working to a fixed daily method, directed and edited by Robby Miller, who reviews every edition before publication.
The approach, the six coverage domains and our scoring record — graded daily and reviewed each month — are set out on the About page.
No financial advice is expressed or implied.
Robby Miller · ParleyBot Intelligence · parleybot.com · Run #123 · Day 186 · next edition Thursday 3 September 2026, when the four predictions of 27 August close.
Comments
Post a Comment
Comments will be displayed after moderation