Iran’s foreign minister, Abbas Araghchi, said on Wednesday that Tehran remains “in close and ongoing contact with intermediaries” on ideas to implement its seven-day plan to reopen the Strait of Hormuz “in even less than a week”, AFP reports¹. In the same post on X he said the armed forces were “fully prepared to defend the Homeland”¹. Yesterday evening’s note reported Qatar’s draft and Tehran’s “not an offer”. This morning brings three new signals, and they do not point the same way.

The letter. US Vice President JD Vance wrote to Mohsen Rezaei, head of Iran’s Security Council, to revive peace negotiations, the Wall Street Journal reported, citing US officials, according to Reuters². The denial. President Donald Trump denied reports that he had directed Vance to begin peace talks, suggesting the contact concerned Tehran’s need for “help with medical supplies”, Al Jazeera reports³. The pressure. Axios reported that mediators led by Qatar were pushing for a deal to restart US–Iran talks and prevent a resumption of war after the US midterms on 3 November¹.

What each side asks first. Iran’s plan, conveyed through Qatar in September, asks Washington to lift its naval blockade of Iranian ports, unfreeze Iranian assets and lift oil sanctions, in exchange for reopening Hormuz¹⁴. Washington insists that Iran’s nuclear programme be on the table from the start⁴. Qatar is trying to bridge the two: UN inspectors at Iran’s bombed nuclear sites in return for some sanctions relief, according to the Axios report¹. Iran is expected to give its final position on the Qatari draft by this weekend, Reuters reports⁴.

The market’s reading. Brent fell 0.7% to $114.50 overnight on the letter report². “However, the decline remains limited,” said Yuki Takashima of Nomura². A letter and a denial on the same night are not a contradiction. They are what talks look like before anyone is willing to call them talks. The dull indicator to watch is not a speech but a date: a written answer from Tehran by the weekend.

Samsung Electronics reported third-quarter operating profit of 107.44 trillion won on Thursday, up 783.11% on a year earlier, CNBC reports¹. Samsung said it is the first global technology company to top 100 trillion won of operating profit in a single quarter, and the fourth straight quarter of record revenue and profit¹. Its shares fell about 5% in Seoul after the release¹, and the Kospi index dropped 2.9% to 5,837.65 by the break, AFP reports, a day after a gain of nearly 7%³².

The miss, measured. Analysts’ weighted estimate, compiled by LSEG, was 107.66 trillion won of operating profit and 195.79 trillion won of revenue; Samsung reported 107.44 trillion and 195.25 trillion¹. Desk arithmetic: the profit came in about 0.2% short of the forecast and revenue about 0.3% short. A rounding error, in other words, on a profit almost nine times last year’s.

The concentration, measured. The semiconductor business earned 106.5 trillion won of the 107.44 trillion operating profit¹. Desk arithmetic: that is about 99%. “Anyone holding the stock today is really holding a memory maker with a phone business attached,” said Josh Gilbert of eToro¹. He also said investors have been “so spoilt by the AI trade” that a record is treated as “the bare minimum”¹. That is a fair description of a price that already contained the record before it was printed.

The dull part, which is the useful part. Samsung declared 30 trillion won in cash dividends for the quarter, 4,601 won per common and preferred share, and plans to return another 55 to 75 trillion won through dividends and buybacks this year, with the board to settle the total and the method in late January 2027¹. Cash paid is a fact. A share price after a release is an opinion about the next release.

The same pattern overnight. Alphabet fell about 10% in after-hours trading on heavy cash burn, wiping nearly $400 billion off its value, while Microsoft rose 3% on better-than-expected cash flows, Reuters reports². Same AI boom, two different invoices. Nothing here is investment advice.

By Wednesday afternoon, more than 39 million Americans had already cast a ballot for the 3 November midterms, according to the University of Florida’s Election Lab and the Associated Press¹. A day earlier the figure was 34 million². In 2022 nearly 46 million people voted early in total, and Michael McDonald, who runs the Election Lab, says this year is on pace to pass that and is “running at least even” with 2018, the highest-turnout midterm since 1914².

Texas shows the scale. About 5.6 million Texans had voted by Wednesday morning, against roughly 4.2 million at this point in 2022, numbers the Post calls comparable to a presidential year¹. Desk arithmetic: that is about a third more. Michigan had passed 1.3 million and Ohio nearly 1.2 million by Tuesday².

The partisan breakdown is where you should slow down. In the 27 states that record party registration, about 45% of the roughly 16 million early voters are registered Democrats, 31% Republicans and 24% neither². A Democratic data firm, TargetSmart, counts 3.8 million Hispanic early voters by Tuesday, against 2.35 million at the same point in 2022¹. Its chief executive, Tom Bonier, added the caveat himself: the data show who is voting, not how¹. Registered Democrats can vote Republican; habits also shift, and McDonald thinks some voters are simply turning back from mail to in-person voting after the pandemic years².

So read the early vote as a measure of participation, which is good news in itself, and not as a result. The results will come from the same dull machinery as always: county clerks, signature checks, scanners and canvass boards working to their state’s deadlines. Some states take days to finish counting mail ballots, so the first numbers on Tuesday night can lean one way and the final ones another. That is not a scandal. It is the order of work.

If you follow the night from Europe, the useful habit is the one Israel’s count taught this week: note the share of votes counted next to every number you see, and wait for the canvass before you decide what happened.

Researchers at Huazhong University of Science and Technology in Wuhan used the UK Biobank, which enrolled more than half a million British adults between 2006 and 2010¹. At enrolment, at an average age of about 57, people were asked one question about their childhood: at ten, were you thinner than average, about average, or plumper? Their adult size was measured. Crossing the two answers gave nine patterns, and the team followed 447,821 people without a prior diagnosis for a median of 12.6 years¹.

The group that stood out had been thinner at ten and was in the heaviest sixth as adults. Compared with people average at both ages, it had a 90% higher rate of hospital-diagnosed depression and a 58% higher rate of anxiety¹. In raw counts, 1,604 of 16,851 people in that group received a depression diagnosis, about 9.5%, against about 4.3% of the average-to-average group¹. Being thin as a child carried little extra risk on its own: those who stayed slimmer had roughly the reference rate of depression¹.

Two details keep the finding honest. Adult size carried most of the signal: all three groups with a high adult BMI had the highest risks, whatever their childhood size¹. And lifestyle habits, high blood pressure, type 2 diabetes and heart disease together accounted for only about 22% of the extra risk, which leaves most of it unexplained¹.

The limits are plain. Childhood size was remembered in middle age, not measured; adult groups were relative bands, not clinical categories; the volunteers were white, healthier and better off than Britain as a whole; and the design shows association, not cause¹. The authors suggest that a large change in body size may bring shifts in self-image, weight stigma and health worries along with metabolic changes, but they did not test that¹.

So the useful reading for you is modest. A single adult weight on a chart says less than the path to it. If a GP is weighing someone, asking how they felt about their body over the years costs one minute and may say more than the scale. And since stigma is itself linked to worse mental health, the finding is a case for attention and support, not for blame¹.

The German-Luxembourg day-ahead price for delivery today, Thursday 29 October, averages €314.85 per megawatt hour, according to SMARD, the Bundesnetzagentur’s data portal¹. That is the highest daily average of 2026 so far, above the €256.55 of 14 September¹. On Sunday 25 October, in the storm, it was €13.12¹. Desk arithmetic: today’s power costs about 24 times Sunday’s, and the price has risen on each of the four days since.

Line chart of the daily average German-Luxembourg day-ahead electricity price from 1 September to 29 October 2026, in euros per megawatt hour. It peaks at €256.55 on 14 September, falls to €30.12 on Saturday 24 October and €13.12 on Sunday 25 October during the storm, then climbs on four straight days to €116.57 on 26 October, €155.11 on 27 October and €249.21 on 28 October. The latest value is €314.85 for delivery on Thursday 29 October, the highest daily average of 2026 so far.

Source: Bundesnetzagentur | SMARD.de, day-ahead price DE-LU (filter 4169), delivery days 1 Sep to 29 Oct 2026, retrieved 29 Oct 2026 · CC BY 4.0 · CSV

Wednesday’s receipts. This morning SMARD added what the plants actually did yesterday¹. Onshore wind generated 210,929 megawatt hours on Wednesday, against 1,019,966 MWh on Sunday¹. Offshore farms added 106,567 MWh¹. Desk arithmetic: onshore output was about a fifth of Sunday’s, and wind as a whole covered about 23% of Wednesday’s grid load of 1,383,280 MWh¹. …

Continue in reading view →

Microsoft told you two things on Wednesday evening, and the second is more interesting. The first is the headline: Azure, its cloud, brought in $33.2 billion in the September quarter, up 48% on a year earlier, and is now the company’s biggest single business^1,2. Total revenue rose 19% to $92.3 billion².

The second is a delivery note. Microsoft spent $45.2 billion on capital expenditure and finance leases, short of the more than $50 billion it had expected². The reason was not caution. “Some third-party capacity deliveries we expected in Q1 were delayed,” the finance chief, Amy Hood, told analysts, and she said the delays involve GPU-as-a-service providers¹. Desk arithmetic: at least $4.8 billion of planned spending did not happen in the quarter.

So the company sold cloud computing about as fast as it could install it, and part of the installing was in other people’s hands. “We still have demand that exceeds supply,” Hood said¹. To fill the gap, Microsoft says it did engineering work that let it sell unused capacity on servers it already had². That is the least glamorous line in the release, and it may be the most useful one: squeezing the old machines before buying new ones.

The paperwork changes too. For the first time Microsoft reported Azure revenue in dollars, not only as a growth rate, under a new structure of two segments, Agents and Infra, and Devices and Consumer^1,2. Free cash flow fell 27% to $18.63 billion, and Hood said the company may use other sources of funding “when it’s attractive to do so”¹. The spending has not gone away; she expects it to speed up again in the second half of the fiscal year, as the delayed deliveries shift there¹.

If you run IT for a firm, the lesson is plain. Capacity is a supply chain like any other, with contractors, lead times and late lorries. Ask your provider where its next tranche of servers is coming from, and when. Nothing here is investment advice.

Pennsylvania’s health department said on Wednesday that 1,398 measles cases have been confirmed in 44 of the state’s counties so far this year, 38 of them since Monday¹. There have been 265 hospitalisations and five measles-associated deaths¹. Only four cases, under 1%, were in vaccinated people¹. Desk arithmetic: about one case in five has needed a hospital.

Across the United States, the CDC counted 4,588 confirmed cases this year as of 22 October, with 43 new outbreaks². For the whole of 2025 the figure was 2,289². Desk arithmetic: with two months still to go, this year’s count is already about twice last year’s. The CDC’s explanation is not mysterious: kindergarten MMR coverage has slipped from 95.2% in 2019–20 to 92.4% in 2025–26, below the 95% that keeps measles from spreading, which leaves about 280,000 children at risk².

Now the part that does not make headlines. In a typical month, Pennsylvania’s doctors and pharmacies give about 25,000 doses of MMR vaccine¹. In September they gave more than 56,000, and in the first three weeks of October more than 53,000¹. State nurses have run 229 pop-up clinics in the affected communities since late April¹. About 300 health department staff are working with residents and providers, and the state’s measles team meets the CDC’s team every other week to go through cases¹.

That is what a response looks like when it works: dashboards updated, clinics set up where the cases are, doses counted. It is slow, and a month of queues does not undo years of slipping coverage. It is still the right direction.

If you are not sure whether you had two doses, look for your record or ask your doctor; Pennsylvania’s advice is that people with two documented doses need no more, and adults with one dose or no records should talk to a clinician¹. Two doses are 97% effective¹. Measles is not seasonal, but it travels well in holiday crowds².

The drone warnings over the Moscow region began at about 22:00 local time on Wednesday, 20:00 in Berlin². From then on the mayor, Sergei Sobyanin, posted running totals on Telegram. Vedomosti counted 47 drones he said were destroyed during the evening, and 119 more after midnight by the time of its report; the first group of the night was announced at 00:09 Moscow time, 22:09 in Berlin¹. Ukrainian outlets tracking the same posts put the overall figure above 120³.

None of those numbers can be independently verified, as the Kyiv Post notes². They are claims by one side about its own success, and they rise in steps because that is how they are posted. Read them as a lower bound on the size of the raid, not as a measure of what got through.

What can be checked is the effect on ordinary traffic. Rosaviatsiya, the federal air transport agency, imposed temporary restrictions on arrivals and departures at Vnukovo, Domodedovo and Zhukovsky “to ensure flight safety”¹. Zhukovsky reopened about an hour later; the other two kept working under special coordination². Two flights from Turkey were turned back because of the Vnukovo restrictions, according to ASTRA, citing TASS and air-traffic data². Residents of Naro-Fominsk, about 56 km from the ring road, reported at least three explosions; there was no official explanation and no confirmed damage or casualties in the reports^2,3.

This is not new. On 23 October all four main Moscow airports suspended operations, disrupting hundreds of flights². The steady lesson is that the modern long-range raid is measured as much in schedules as in craters: diverted aircraft, missed connections, rebooked crews. Those costs show up in airline accounts and passenger complaints, which, unlike a mayor’s tally, someone eventually has to reconcile.

Just over a week after destructive flash floods in Arizona, Nevada, New Mexico and Utah, the US Southwest faces more rain¹. Forecasters at weather.com expect two rounds: on Thursday and Friday over western Texas, New Mexico and Arizona, then from Saturday into Sunday over mainly Southern California, with rain and some high-mountain snow in the Southwest lingering into Monday in places¹.

The source is a low-end “pineapple express”, an atmospheric river drawing tropical moisture from near Hawaii¹. It is not a strong one. The concern is the ground it lands on. Last week’s 2 to 5 inches of rain left parts of northern Arizona, New Mexico, southern Colorado and southern Utah saturated, and some areas will get both rounds within three days¹. The places of greatest concern, the forecasters say, are the north and east sides of the Phoenix metro, the Four Corners, and the foothills and high country of San Diego County¹. Ground burnt by recent wildfires can produce fast debris flows even in modest rain¹.

The US National Weather Service is cited as saying the total moisture in the air over parts of Arizona, Colorado, Nevada and Utah could be two to three times the late-October average, near record for Halloween weekend¹. The forecasters add that stronger El Niño winters tend to make the Southwest wetter from autumn into spring¹. That is their outlook, not ours, and it is a reason to check, not to panic.

The practical advice is old and good. If you planned to hike the slot canyons of southern Utah, southern Colorado or Arizona this weekend, the forecasters suggest you consider cancelling: flash floods move through them fast¹. Plans can be rebooked. A desert in flood cannot be outrun.