Trump called the JCPOA the worst deal ever made. He tore it up in 2018. He started a war in February 2026 to get something better. On June 17, at the Palace of Versailles, he signed something materially worse on every measurable dimension.

Iran's highly enriched uranium stays in-country. Its missile program is explicitly off-table — Iran's foreign ministry said after the signing: "Iran's missiles are only for firing, not for negotiation." The regime is intact, claiming victory, and pointing at billboards in Tehran that read "Trump couldn't do a damn thing." The deal defers the nuclear question to a 60-day negotiation window that begins with Iran holding stronger cards than it had on February 27.

The JCPOA — the deal Trump called the worst in history — had: zero enrichment at Fordow, a 3.67% cap at Natanz, 98% stockpile reduction, two-thirds centrifuge dismantlement, intrusive IAEA access, and a verified one-year breakout floor. The June 17 MOU has: a "pinky promise" not to build a nuclear weapon (a commitment Iran has been making for 50 years, per ABC News analysts), a vague option to discuss dilution of HEU, and 60 days to figure out the rest.

The leverage didn't disappear in February 2026. It eroded across the preceding decade, quietly, through a mechanism that Western financial and strategic analysts consistently misread as temporary adjustment rather than structural replacement.

The 2012-era Iran sanctions worked because the US dollar was the sole settlement currency for global oil trade, because SWIFT exclusion was economically fatal, and because no major power was willing to absorb the diplomatic cost of defying the sanctions regime at scale. All three of those conditions have degraded materially.

China began buying Iranian oil in yuan-denominated transactions at scale. It built alternative settlement infrastructure. It absorbed the diplomatic cost with indifference. When you can't credibly threaten Iran's economy because China buys the oil regardless of your sanctions, "maximum pressure" is theater. The JCPOA was negotiated when US sanctions actually bit. That window closed.

The war didn't reveal this — it confirmed it. The US went in with the assumption that apex military and financial power remained mutually reinforcing and globally decisive. They discovered they were neither. Not simultaneously, and not cleanly — but the outcome is the data.

Israel's public position is that the MOU doesn't bind it. Ben Gvir: "Trump's agreement does not bind us." Katz: IDF will remain in Lebanon without any time limit. These are political statements. The IDF knows the munitions picture, and the munitions picture is the actual constraint.

Israel burned through precision munitions at a rate that alarmed US planners during Gaza, before the Iran war started. The February 28 strikes were joint precisely because Israel couldn't sustain a campaign of that scale unilaterally. One hundred and eight days of conflict against a peer adversary later, the US has been the logistics backbone. Without it, Israel's independent sustained strike capacity is weeks, not months.

The gap between Ben Gvir's rhetoric and Netanyahu's operational reality is why Israel has not moved on Lebanon despite the MOU formally requiring Israeli withdrawal as a ceasefire condition. He cannot. The coalition that keeps him in power demands it; the military that would execute it cannot sustain it without resupply from an ally that has now publicly signed a deal Israel's government calls illegitimate.

Who holds the munitions cards globally? Russia's defense industrial base is stressed from Ukraine. China's is the largest by volume on earth and operating without comparable depletion constraints. China watched the US and Israel demonstrate their consumption rate over 108 days. It took notes.

China entered this war having made no decision to enter this war. That is the point.

China watched the United States spend over $100 billion, deplete allied munitions stockpiles across two allied militaries, lose 13 Americans, fracture its most important Middle East alliance publicly enough that Trump told reporters Netanyahu has "no fucking judgment" during active negotiations, produce a deal worse than the one it started the war to replace — and hand Iran economic relief plus a strategic victory narrative that will echo across every future coercive diplomacy attempt the US makes in the region.

During the same period: China's GDP grew. Its semiconductor exports hit 13-year highs. Its GPU self-sufficiency nearly doubled. It positioned itself as the indispensable partner for Iranian reconstruction. The $300 billion reconstruction fund is, in practice, a BRI expansion project — the largest one since the program launched. Chinese infrastructure companies are the only entities on earth with the supply chains, financing architecture, and political clearance to execute it at scale.

The deal is a 60-day clock, not a resolution. The Lebanon tripwire — the condition most likely to collapse it — is now embedded in the signed text of Point 1: "immediate and permanent termination of military operations on all fronts, including in Lebanon." Israel has refused. Its governing coalition structurally cannot accept that condition and survive. Netanyahu's electoral constraint is not a tactical problem; it is a structural one confirmed across multiple primary sources.

The physical Hormuz picture is unchanged by the signing. Mine clearance has not begun. Major commercial operators have not resumed transits. DHT's "high level of credibility" threshold — the leading normalization signal this system tracks — has not been met. The MOU itself acknowledges the operational reality: traffic restoration will happen "in proportion to" de-mining progress. The paper market is pricing narrative. The physical market is pricing supply reality. The spread between them is the signal.

This system's prediction: MOU becomes non-functional within 60 days of the June 17 signing. Review date: August 17. The Lebanon tripwire is the primary mechanism. The structural conditions for that outcome — Israeli electoral constraint, IRGC behavioral divergence from the FM diplomatic track, Hezbollah's non-participation, the four documented US-Iran contradictions on core terms — are all confirmed and present. They were present before the signing. The signing didn't resolve them. It formalized them.

A conservative analyst on PBS, asked whether the US achieved its war goals, said: "No, we haven't achieved any war goals. Iran has maintained a significant missile and drone force. It will use whatever money it gets to a large extent to rebuild its defense industrial base. It's going to weaponize and monetize control of the Strait of Hormuz. Not only have we not dealt with the strategic threat; we have made it worse by putting a new militarized, radicalized leadership in place and making them more likely to actually seek a nuclear weapon."

Iran's own framing from state media: "The U.S. is forced to sign agreement to end the war."

Both of those statements are consistent with the thesis. The war didn't create the conditions that produced this outcome. It ran the experiment. The results came back fast.

The MOU gave every news organization the thing they need most: an ending. "Deal reached" is a complete sentence. "Deal reached but the IRGC is still enforcing lane discipline kinetically while 11,000 sailors remain stranded and the food compound is locked in regardless" is a situation report. Situation reports don't generate clicks. They don't offer resolution. The media didn't decide Hormuz is over because the evidence supports it. They decided it's over because the story needed an ending and the MOU provided one.

This is not a failure of journalism. It's a feature of how narrative works. Stories need endings. Compound multi-domain slow-moving physical-world events don't have endings; they have trajectories. A trajectory is not publishable. A signing ceremony is.

Markets aren't being irrational. They're pricing the right question for their time horizon — they're just not pricing the question that matters for physical reality.

A 30-day oil contract asks: "Will enough oil reach the market over the next 30-60 days to cover my position?" The answer to that specific question is actually yes — not because Hormuz is normalized, but because Saudi Arabia is front-loading exports at Ras Tanura, Venezuelan crude has surged to 1.5 million barrels per day with a long-term SLB production deal behind it, and Chinese demand is at 2018 lows because Beijing is drawing down strategic reserves rather than buying at current prices. The paper market has enough oil for its time horizon.

The physical world has a food compound that will print in the fall harvest, a tanker fleet repositioning lag that takes 40 weeks regardless of diplomatic announcements, and an Iran that has already stated on record that Hormuz tolls resume when the 60 days expire. Those things don't appear in the 30-day supply calculation. They appear in Q3 harvest data and 2027 production capacity figures.

Most people can't do anything with the information that fertilizer application windows have closed and 2026 crop yields are already determined. They can't reroute a supply chain or change a procurement decision. They can't affect the VLCC repositioning lag or the Ebola BVD response capacity. When information has no traction — when knowing something doesn't change what you do — attention correctly deprioritizes it. That's not stupidity. It's cognitive resource management operating exactly as designed.

The systems this site tracks exist because for some people the information is actionable: project timing, materials procurement, travel routing, portfolio positioning. For most people it isn't. Attention follows actionability. The public is not being irrational by moving on. They're responding correctly to their actual decision environment.

The food compound will not announce itself until harvest data prints in October. By then the causal chain is eight months old and diffuse: war started in February, Hormuz closed in March, fertilizer application window closed in May, crops planted under deficit conditions, harvest down in the fall. No single actor is responsible. No single event caused it. Diffuse causation produces no narrative — no villain, no moment, no one to blame. The press needs a villain or a celebration. An eight-month compound with no identifiable responsible party fits neither frame.

Heat is the clearest version of this structural invisibility. Record heat has been normalized by repetition. Each year's event was covered as the exceptional moment. Now exceptional is baseline and there's no editorial bandwidth left for the fourth consecutive record-breaking heat event with compounding agricultural, grid, and wildfire effects. The fact that it's compounding — more people exposed, more crop stress, more wildfire fuel, more grid load per degree — doesn't surface because compounding is not an event. An event has a date. Compounding has a trajectory. Trajectories are not publishable.

The dissonance — knowing what the physical world is doing while the surrounding conversation has moved on — is specifically the experience of working on a different time horizon than the information environment around you. The press runs on 24 hours. Markets run on 30-60 days. A growing season runs on months. A tanker fleet repositioning runs on 40 weeks. A de-dollarization architecture runs on decades. None of those timelines are legible to a news cycle, and the news cycle is what most people swim in.

Being right too early feels exactly like being wrong. The surrounding consensus says it's over. The physical world says it isn't. Holding that gap without capitulating to the consensus — without dismissing what you know because nobody around you is talking about it — is the hardest part of this kind of analysis. The analytical discipline this requires isn't just about data. It's about resisting a very specific psychological pressure: the pressure to update toward the room.

As of June 28: the IMO seafarer evacuation is suspended after the IRGC struck MV Ever Lovely during the operation on June 25. 11,000 sailors and 7,800 ships remain inside the Persian Gulf. The IMO's own Secretary General has confirmed 14 mariners have died in this conflict — 11 killed by Iran, 3 by the United States. Hormuz now operates as a two-corridor system: a northern lane controlled by Iran's Persian Gulf Strait Authority requiring pre-registration and approval, and a southern lane operated by the US, Oman, and IMO also requiring framework registration. Ships that use neither get targeted. War-risk insurance premiums remain at 10% of hull value for the Listed Area — punitive for commercial cargo economics and unchanged from peak-conflict levels.

The food compound: USDA confirmed corn down 6% and winter wheat down 25% from prior estimates in its June forecast. The fertilizer application window closed in May. Whatever was applied, was applied. 70% of US farmers nationally could not afford full fertilizer this season; 78% in the South; more than 80% of rice, cotton, and peanut producers. The fall harvest will be the first output-side confirmation. It will arrive in October. The coverage will treat it as a surprise.

The press has moved on. The physical world has not. These are not in conflict. They are operating on different clocks, answering different questions, for different audiences. The work is to know which clock you're on — and not to mistake the press's clock for the world's.

Six consecutive nights of strikes on Iran. Retaliation has reached Qatar, Kuwait, and -- Tehran claims -- a US base in Syria. The Joint Maritime Information Center just upgraded its Hormuz threat assessment from substantial to severe. And Brent crude is sitting at $84.70 a barrel, almost exactly where it was a week ago, before any of that happened. That's the second straight cycle of the war widening while the price sits still.

Two hundred miles offshore and thousands of feet down, something structurally identical is happening on a much longer clock. 2025 was the ninth consecutive record year for ocean heat content. June 2026 set the hottest sea-surface-temperature record ever recorded for that month. Marine heatwaves covered 82% of the world's oceans in the first half of this year. Oceans absorb roughly 90% of the excess heat human activity traps in the atmosphere -- which is why the air doesn't feel as hot as the numbers suggest it should. As one climate scientist put it this month, the sea "is not simply warm right now. It is a loaded spring."

That phrase is the whole argument. A spring that's loaded hasn't released anything. It's stored the energy for later, and later is coming: El Niño is the mechanism by which stored ocean heat gets handed back to the atmosphere, which is why the year after an ocean-heat peak has twice now turned out hotter than the peak itself. 2026's heat isn't this year's bill. It's a down payment on next year's.

Once you see it in the ocean, it's hard not to see it everywhere else this system tracks.

China's crude oil demand fell roughly 9% from pre-war levels, according to JPMorgan -- more than global demand fell during the entire 2008 financial crisis, and it produced no visible event. Not fuel lines, not rationing, not a political crisis. Just a footnote in a demand forecast. It's tempting to credit this to EVs -- China quietly electrifying its way out of the shock -- but the data doesn't quite support that story. Domestic EV retail sales barely moved this spring (+2.4% year-on-year in April). What actually happened is three older mechanisms stacking at once: China's overall car market cratered (seven straight monthly declines, -21.6% in April) as high fuel costs suppressed buying generally; the country drew down its own strategic and commercial oil inventories instead of importing at war prices; and its automakers redirected an existing EV manufacturing surplus into exports, which surged 111.8% into markets -- mainly Europe -- that were already shifting toward electrification on their own. None of that is a new capability. It's old tools, all pulling the same direction, on a country with unusually deep reserves of each.

Europe ran the same play on jet fuel, on a shorter fuse. In April, the IEA's own chief warned Europe had "maybe six weeks of jet fuel left." By June, that hadn't happened -- the EU's Transport Commissioner said there was no scarcity, largely because Europe produces over 70% of its own aviation fuel. But "no shortage" isn't the same as "no cost." Lufthansa cut 20,000 flights through October. Nineteen of the world's twenty largest airlines trimmed capacity in May. Prices roughly doubled at the peak. American and Delta each absorbed billions in extra fuel costs. The crisis didn't fail to happen -- it happened in April and May, mostly, and got paid in price and reduced capacity rather than grounded planes, which is exactly why it's already been forgotten heading into peak summer travel.

And then there's the case with no runway left. On July 10, Cuba's national grid collapsed entirely for the ninth time since late 2024 -- ten million people lost power. The mechanism: roughly six months of tightened US secondary sanctions severed Cuba's fuel imports, stripping a Soviet-era grid of the reserve capacity any modern system needs to survive a single failure. At the moment of collapse, Cuba had 935 megawatts running against 3,100 megawatts of demand. Its largest power plant -- no comprehensive maintenance since 2010 -- was mid-shutdown, its seventeenth of the year, exactly as engineers tried to bring the system back online.

Every one of these mechanisms is running on a different clock, and that's exactly why most of them are invisible until it's too late to do much about them. The ocean's clock is years -- heat absorbed this year surfaces as next year's El Niño, this decade's coral bleaching, and on the longest end of the scale, potentially a multi-decade circulation disruption that won't read as a crisis until it's already a discontinuity. China's clock is quarters. Cuba's clock, once its buffer ran out, was days.

Iran's clock is weeks. Strike, retaliation, price move, repeat. That's not because Iran is a special case -- it's because Iran is the one place in this system's current tracking where the entire cycle of buffer-spending and eventual reckoning completes fast enough for a person to actually watch it happen, rather than needing a Copernicus bulletin and nine years of satellite data to notice the pattern. A strobe light doesn't create motion that wasn't there. It just flashes at a frequency fast enough to reveal what was already moving too fast, or too slow, to see directly. Iran is currently the fastest-flashing instance of a mechanism that's also running, right now, underneath an ocean, a Chinese auto market, and a Cuban power grid.

None of this should read as "everything is secretly the same problem." That's precisely the kind of pattern-matching this system exists to catch itself doing, and it's worth being explicit about where the analogy actually breaks.

Ocean thermal capacity isn't a statutory floor like the US Strategic Petroleum Reserve's 252-million-barrel limit. There's no bankruptcy event, no single day the ocean "runs out." Its accounting is genuinely different in kind, not just scale, from a fuel reserve or a bank of foreign-exchange inventory. The honest, narrower version of this thesis isn't "everything is the same buffer" -- it's this: every case above involves a real capacity, currently being drawn down, with a documented mechanism for how the deferred cost eventually resurfaces, and no market or policy actor currently pricing that resurfacing as inevitable. Brent holding flat through six strike nights might simply mean the market has already priced the plausible ceiling of near-term disruption and is waiting for a genuine de-escalation signal -- in which case the real test isn't whether price holds through more strikes, it's whether price actually moves the next time something genuinely de-escalates.

Three things would tell us which reading is right. If the US Strategic Petroleum Reserve breaches its statutory floor without any resolution to the underlying conflict, that's the deepest-buffered energy reserve on Earth running out in real time. If a Cuba-scale total collapse happens somewhere with real global supply-chain weight, the mechanism has graduated from an isolated case to a mainline one. And if the ocean-heat pattern repeats one more time -- if 2027 turns out hotter than 2026, the way 2024 followed 2023 and 2016 followed 2015 -- that's the loaded spring firing exactly as the mechanism predicts, on a timescale long enough that almost nobody will remember this week's Brent price when it happens.

Until then, the honest claim is narrower than "we've entered a new normal of shock absorption." It's this: several deep systems are currently spending down real, finite reserves fast enough that the shock isn't reaching whoever would normally feel it first -- and the bill for that is accumulating quietly, in an ocean, in a statutory floor, in an airline balance sheet, in places most people have no reason to be watching. Cuba is what it looks like once the reserve runs out. It isn't a warning about Cuba. It's a preview.

China's crude oil imports fell roughly 40-45% from pre-war levels — about 5 million barrels a day, a larger drawdown than every strategic reserve release on Earth combined. Oil market researcher Rory Johnston called it the "number one mystery" in the oil market right now: no depression, no visible economic pain, just a buyer quietly absorbing a supply shock that should have broken the global market. DW News let him run through the possible explanations on air — self-interested hedging, geopolitical signaling to Asian allies, even a "tin-foil hat" secret understanding with Washington — and didn't editorially neutralize the more generous readings before the segment ended. That's the detail worth sitting with. Western financial media doesn't usually let "the other side might be acting in good faith" survive the edit.

It's tempting to read that as a verdict. It isn't one. A citation check on the same story surfaces a second named analyst, Javier Blas, arguing the drawdown is mostly structural — EV substitution eating into conventional car sales, ethane displacing oil in petrochemicals, and China simply working off inventory it over-bought in 2024-25. Two credentialed people, looking at the same number, disagree about the mechanism. That's not a reason to pick a side. It's a reason to ask a better question than "is China being generous."

Every capital-allocation system — state-directed or private-market — misallocates money at scale, regularly, as a matter of course. That's not the variable that predicts whether a society comes out ahead or behind. The variable is narrower: when the money is gone and the story around it stops being believed, is there a physical thing left standing that someone else can use — or does the value simply evaporate, because it was never anything but a claim on a narrative to begin with.

Call it the leave-behind test. It isn't a referendum on which economic system is morally better. It's a falsifiable question you can ask of any specific dollar: was this capital mandated to build a thing, or mandated to capture a return? Those are different instructions, and they produce different wreckage when the bet goes wrong.

None of these were selected to make a point and then confirmed after the fact — they were argued out, and two of them got materially corrected mid-argument when the evidence didn't hold up. That correction process is the part worth including, not hiding.

This is the section that keeps this from being a one-sided argument dressed up as analysis, and it belongs in the middle of the piece, not a footnote at the end.

China's own system fails this exact test, badly, in more than one place. The Great Leap Forward is the ceiling case for what "the state just decides, with no price signal and no accountability" looks like — an estimated 30-45 million dead in a famine produced by state-directed capital allocation with zero feedback mechanism. Evergrande's collapse is the sharper contemporary case, because it fails on China's own turf using the same tools that built the HSR network: millions of pre-sold, prepaid apartment units, construction halted mid-build, no refund, no unit — money that vanished as completely as a failed SPAC, except it came out of ordinary households' life savings under China's pre-sale financing system. And the HSR debt itself — $839 billion in accumulated, largely opaque local-government financing, per public data through 2023 — is a real, unresolved liability sitting on top of an asset that works, which is a different problem than the ones above but not a nonexistent one.

Everything above still treats "leave-behind" as physical: a panel, a rail line, a charging port. That's too narrow, and the gap shows up fastest close to home.

Some capital converts into neither a building nor a stock certificate and still leaves something durable behind — trained people, working institutions, a population that didn't get sick, knowledge that doesn't have to be relearned. Call it the existential leave-behind. It carries too, and the carry isn't always denominated in dollars. It can be ecological, or it can show up as a population absorbing the health and trust costs of a decision someone else made.

Which raises the question the physical-asset framing quietly skips: leave-behind and carry for whom. A five-square-mile enclave on the California Peninsula makes the mechanism visible without needing an ocean between the two sides of the ledger. Atherton — one-acre-minimum lots, adjacent to San Jose and Redwood City — has repeatedly resisted its state-mandated housing density requirements. In 2023, Steph and Ayesha Curry wrote to the town council specifically opposing a 16-townhome project near their home, citing family privacy; Marc Andreessen and Laura Arrillaga-Andreessen did the same over a separate parcel months earlier. Both letters are a matter of public record and neither family did anything unlawful.

Atherton captures the existential leave-behind of the Silicon Valley agglomeration — proximity to the capital, the companies, the schools, the networks that make an acre lot there worth what it's worth — without carrying its proportional share of the housing supply that keeps the agglomeration staffed. The teachers, nurses, and tradespeople who keep the region functioning get pushed into the denser housing stock next door. Prop 13 is the policy layer that lets this persist indefinitely. Nobody broke a law. That's the tell: the mandate given to the land was legally coherent and locally rational, and it still produced a regional carry cost paid entirely by a population that never had a vote on Atherton's zoning.

Take the two systems' failure cases and their success cases together, and a narrower, more defensible claim survives all the counterweights above: a system defaults toward leaving something behind when its standing instruction to capital is "build the thing," and it defaults toward leaving nothing behind when the instruction is "capture the return" — regardless of who's giving the instruction. China's default setting, most of the time, is the former. The US's default setting, most of the time — auto bailout over industrial policy, PG&E's extracted century, a charging network no private actor would fund without a court order or a subsidy — is the latter. Both systems have exceptions that prove the rule works on mandate, not nationality.

This isn't a series of unrelated bad choices. Rising and declining powers have made this exact switch before: capital migrates from building the productive base to extracting rent from a base someone already built, the moment the second option pays better than the first. Dutch merchant capital became European banking capital once Amsterdam's shipyards stopped being the highest-return place to put money. British manufacturing supremacy was gone decades before the City of London's financial dominance was. The optimization target a system runs on isn't really a free choice each time — it's a symptom of which of those two phases it's currently in.

If that sounds too abstract to be falsifiable, here is the concrete version, checked, current, and specific: when Congress passed the CHIPS Act's semiconductor subsidies, the Commerce Department had to write a clause explicitly forbidding recipients from using the money for stock buybacks. Not because it was a hypothetical risk. Because the people writing the check knew — with enough certainty to put it in the law — that absent the restriction, a meaningful share of public capital handed to one of the most profitable industries on Earth would go straight back out the door into shareholder pockets instead of into the fabs it was supposed to build. That single clause is the entire thesis of this piece, condensed into one paragraph of federal regulation. The mandate given to the money, everywhere else in this article, has been implicit. Here, for one brief moment, a regulator wrote it down.

Rory Johnston's "mystery" and Javier Blas's structural counter-read are both probably partly right about the oil number, and neither is really the point anymore. What's actually being asked, in every case in this piece — the highway, the hook, the panel, the port — is a much older question than any of them: is the juice worth the squeeze. Late-stage financial capital has quietly redefined "worth" to mean whatever can be captured and booked this quarter, and it has gotten very good at making that definition feel like the only serious one. It measures the squeeze in dollars and calls it done. It doesn't ask who's left holding the juice, or whether there's any juice at all once the story stops being believed. China's version of the answer has its own catastrophes — a famine, a property collapse, hundreds of billions in debt nobody's pretending isn't real. But it keeps, more often than not, building the thing. The open question isn't whether that's generous. It's whether the other side is capable of writing down its own mandate as honestly as that one clause did — and then choosing to change it.