Structural reads on power, data and science — what the headline number leaves out, and the precedent that explains it.

Joined May 2026
Africa is adding solar faster than anyone can count it. Ember puts this year's build at 17 gigawatts — a continental record, up 45% on 2025. Across 2023 to 2025, three-quarters of everything the continent added was small, distributed, rooftop-and-yard solar: 20 gigawatts out of 26. It is the kind no national ledger reliably records, and only three of Africa's 54 countries publish solar data quarterly. How big is the gap? For 2025, Ember counted 12 gigawatts. The IEA counted 6.2. IRENA counted 4.6. Same year, same continent, three different answers. Every distributed technology arrives before the statistics that count it. The estimate is modelled from import data, not meters, and 94% of the panels still come from China. But the direction is plain: households and businesses are electrifying ahead of the institutions meant to do it for them. If the fastest energy build-out on the continent is the one no ministry planned and no agency fully counts — what should we measure: the megawatts, or the people installing them? #SolarPower #Africa #EnergyTransition
2
49
The same current from two other angles. Where the panels are made — China's solar passed coal this year: xitter.cf/The_ReFrame_/status/20… Where the timetable was outrun — India cleared its 2030 clean-energy pledge five years early: xitter.cf/The_ReFrame_/status/20…
India pledged 40% of its power capacity from non-fossil sources by 2030, then raised its own target to half. It cleared the harder one in 2025, five years early. Renewable capacity reached 288.6 GW by June, solar alone at 162 GW, after a record 29 GW of solar and wind added in the first half of the year — and generation is following, up nearly 25% year on year. India now ranks third in the world for installed renewable capacity. The build is running ahead of the wires that carry it. Grid, storage and transmission are straining, and capacity still isn't delivery: a high share on a bright afternoon is not clean power at midnight. But a country of 1.4 billion meeting its own climate promise five years early resets what everyone else is allowed to call impossible. The target was never the ceiling. It was the floor. What else are we calling impossible that's just early? #CleanEnergy #India #Renewables
26
In 1895 Khama III, Sebele I and Bathoen I went to Britain so their land would not be handed to a company. Botswana is a country because that argument worked. Their statues stand in Gaborone. Independence came seventy-one years later, with twenty-two graduates in the whole country, twelve kilometres of tar and $89 a head. What was built after that was built by people — the kgotla, the schools, and a state that kept its books. Income per head is $7,778 today on the same World Bank series. The twelve kilometres are now 11,209. The diamonds arrived later. Orapa was found seven months after independence and did not run until 1971. They helped. They were not the beginning. It is not a finished story. Unemployment sits near 28%, diamonds are still about four-fifths of exports, and the economy has just come through two years of contraction as that market turned. The next chapter is already dated. Anglo American is selling out of De Beers. Botswana owns 15% of it and half of Debswana, and is positioning for a majority. Vision 2036 sets high-income status for the seventieth anniversary. In 1895 they went to London to stop a company taking the country. Now the country owns a piece of the company, and wants more. Pula, Botswana. #Botswana60 #Boipuso60 #WhatThenWhatNow
44
Nigeria spent a decade holding 445,000 barrels a day of refinery capacity that did not run — Port Harcourt, Warri, Kaduna. The plants were there. The output was not. So a new one got built, and the import figures moved: close to 400,000 barrels a day of refined product imported in 2023, under 130,000 by the second quarter of this year. Then the constraint moved with it. Nigerian law obliges producers to offer crude to domestic refineries — Section 109 of the Petroleum Industry Act. It does not oblige anyone to agree a price. Deals run on willing buyer, willing seller. So between January and August, producers offered 182 million barrels, the refineries said they needed 154.6 million, and deals were completed for 112 million. The regulator published those figures on 29 September and was careful to say the gap is not a failure by either side. It is pricing, payment guarantees, specifications, delivery schedules. The terms, not the oil. And it is closing — supply ran at roughly 46% of allocation in the first quarter, and 97.4% compliance in the second. The refinery was the visible problem, and it got solved. What decides the output now is a contract nobody photographs. #Nigeria #OilAndGas #Energy
1
40
13
Seventy-two years ago today, twelve European states finished ratifying a treaty and CERN came into existence. One of the twelve was Yugoslavia. It left in 1961 and the country itself is gone. Greece had come out of a civil war five years earlier. West Germany was nine years past the end of a war the rest of them had fought against it, and was a founding member anyway — which was, quietly, most of the point. There are twenty-five member states now and eleven associates. Slovenia joined last year as the twenty-fifth. Chile became an associate in April. The organisation runs on about 1.28 billion Swiss francs of member contributions, employs roughly 3,700 people, and hosts more than 12,400 visiting scientists from over 110 nationalities. Along the way somebody there wrote a hypertext proposal, and on 30 April 1993 CERN put the software into the public domain and charged nobody for it. Twelve governments agreed to share a machine because none could afford one alone. That is still the whole mechanism. #CERN #WhatThenWhatNow
3
1
30
Source — CERN came into existence on 29 September 1954, when the founding member states ratified the Convention — home.cern/about/who-we-are/o…
5
Related — the EU builds associate tiers because its treaty admits only European states. CERN built one too: eleven associate members, from Brazil to Pakistan. Same architecture, a different institution. xitter.cf/The_ReFrame_/status/21…
Canada has been an associated country of the European Union since 2024. On 16 September it was offered the title. Ursula von der Leyen told the European Parliament, with Mark Carney in the chamber, that she wants to work on "opening the door" for Canada to become the EU's first associate member. Much of it is already open — Canada joined Horizon Europe's second pillar on 3 July 2024, and its researchers now compete for EU money on equal terms. What has no door is membership: Article 49 admits only "any European State." So the Union builds side entrances. Its trade agreement with Canada has run provisionally since September 2017; ten of twenty-seven member states have never ratified it; Cyprus's parliament voted it down in 2020. It carried on regardless. Switzerland refused the bloc by 50.3% in 1992 and took a side entrance too — then signed the agreement tidying that arrangement in March 2026, thirty-four years on, and its voters may still get the last word. The side entrances work. They just never quite finish. How long before anyone calls Canada's finished? #SOTEU #CETA #TradePolicy
9
5
Flight 13's Starship came down in the Indian Ocean and stayed there, intact and afloat, for twenty-four days. Then SpaceX went out and got it — the first upper stage they have ever recovered whole. The heatshield changes flying on the next ship came off that hull. In 1954 two de Havilland Comets broke up over the Mediterranean within three months of each other, killing everyone aboard. The Comet was the world's first jet airliner, and nobody knew why it was failing. So they trawled the seabed off Elba for the wreckage — and BOAC gave up a serviceable Comet to be submerged in a two-hundred-thousand-gallon tank at Farnborough and pressurised, over and over, until it split. It took three thousand and sixty cycles and six weeks. What came out of that tank was not a shape. It was a rule: that a pressurised fuselage has to be tested to destruction before anybody flies in it. The difference is intent: SpaceX meant its ship to come down in the ocean. The method is identical. A failure only becomes knowledge if somebody goes and gets it. What are we currently failing at without collecting the wreckage? #Starship #Engineering #Aviation
1
1,163
Between a Tuesday and a Thursday in mid-September, China put four different rockets into orbit. Not four flights of one workhorse. Four vehicles, four companies' worth of engineering, three launch sites — a desert pad at Jiuquan, a commercial complex at Wenchang, and a platform in the Yellow Sea. A methane-fuelled Zhuque-2E from LandSpace. A solid-fuel Gravity-1 from Orienspace, lifting off water. A Long March 12 from the state's CASC. A Kuaizhou-11 from Expace. Thirty satellites in forty-five hours, and two of the four rockets built by private firms. Twenty-eight of those thirty were broadband satellites for two megaconstellations. That is what the cadence is for, and it is why it exists at all. For most of the space age a launch was a national event. A countdown, a broadcast, a week of build-up. What those forty-five hours showed is a country where launch has become logistics — different rockets rolling out of different factories on overlapping schedules, none of them the headline. The honest counterweight: none of the four attempted a landing, and reuse is where the cost curve actually bends. Cadence is not yet economy. But cadence is the precondition. Nobody learns to reuse a rocket they fly twice a year. The organisations flying every few days are the ones who get to try — and the company that flew the second of those four rockets was founded in 2020. #Spaceflight #Rockets #SpaceEconomy
1
21
Source — China launches 4 different rockets to orbit in 45 hours, Space.com: space.com/space-exploration/…
8
A decade on from Paris, the planet is set to blow past 1.5°C. Ask the question the headlines have already answered for you: did the treaty fail? Hold that answer. We called a treaty a failure before. In 1987 a small agreement was signed to phase out the chemicals eating the ozone layer. Critics said it was too narrow, too slow, a gesture aimed at a problem far too big for it. It became the first agreement in history that every nation on Earth joined. It retired about 99% of the chemicals it named. Most of the world's ozone layer is on track to recover by 2040, the Antarctic by 2066. Here is the part that gets left out. On 14 March 1988 the United States Senate ratified Montreal by 83 votes to nothing. Reagan signed the instrument three weeks later. The United States has been a party ever since, under every administration since. Paris never went to the Senate. It was joined by executive action — and in January the United States left it for the second time, the only country ever to leave twice. So the difference was never ambition. Montreal was built as something a country ratifies. Paris was built as something a president can join, and therefore as something a president can leave. The question was never whether Paris feels like it's working. It's whether we are willing to build the next one so that it cannot simply be walked out of. #ClimateSummit #ParisAgreement #Climate
2
18
Three weeks ago we asked how deep the overshoot goes: xitter.cf/The_ReFrame_/status/20…
The UN Environment Programme published a report yesterday that quietly changes what 1.5 means. Limiting Overshoot says the world will cross 1.5°C of warming, likely within the next few years. Its best case — every current pledge delivered in full — peaks at 1.8°C. On current policies rather than pledges, it puts the world near 2.6°C by 2100. That is a different kind of document from the one that set the number. The Paris Agreement was adopted on 12 December 2015 by 195 Parties. Its aim was to hold the rise in global average temperature "well below 2°C above pre-industrial levels" while "pursuing efforts to limit the temperature increase to 1.5°C above pre-industrial levels." For eleven years 1.5 has been spoken about as a wall. It was never a wall. It was a line drawn through a continuum by people who had to choose somewhere to draw it, and the physics does not know it is there. What changes on crossing is not the nature of the risk but its arithmetic: every additional fraction of a degree, and every additional year spent above the line, adds damage that has to be lived through or paid for. Now the counterweight, and it is the part that decides whether any of this is honest. "Overshoot and return" assumes carbon removal at a scale that does not currently exist — sustained net-negative emissions, governed and verified, for decades after net zero. That is a modelling assumption before it is an engineering plan. A report can specify the way back without anyone having built the road. But look at what this document actually is. It was published before the crossing, not after — a plan for a failure that has not finished happening, written while there is still something to decide. Most institutions do not do that. Most wait, and then explain. The target was 1.5. The question now is 1.8 or 2.6, and for decades or for the rest of the century. Those are not the same outcome, and nobody has chosen between them yet. #UNEP #Overshoot #ClimateAction
1
Source — Senate resolution of ratification, Montreal Protocol, Treaty Document 100-10, 14 March 1988: congress.gov/treaty-document…
4
Nairobi will host the 2029 World Athletics Championships. The Council voted on 15 September in Budapest; the venue is Kasarani. It beat London, Rome and Munich to get there, and Munich took 2031 as the consolation. The championships began in Helsinki in 1983. Through Tokyo in 2025, twenty editions have been held. None of them in Africa. Set that against what the sport has actually been drawing from the continent for the last sixty-six years. At Rome in 1960 Abebe Bikila ran the Olympic marathon barefoot and won it in 2:15:16.2, a world record, twenty-five seconds clear of Rhadi ben Abdesselem of Morocco — the first Black African to take an Olympic marathon gold. Every distance final since has had the same accent, and the championships kept being held somewhere else. There is a version of this post that is bitter about that, and it would be easy to write. But bidding is a thing you do, not a thing that is done to you, and Kenya bid against three European capitals with stadiums already built and won on the merits of the case it made. Forty-six years from the first championships to the first African one. In 2029 those finals will be run at home. #WorldAthletics #Nairobi2029 #Kenya
2
44
We ran the whole Bikila story on Hindsight Sport in July — the shoes that fell apart, the breakaway, and the second Olympic marathon gold four years later in Tokyo. xitter.cf/_HindsightSport/status…
IN FOCUS: Abebe Bikila · Marathon. For more than six decades of the modern Olympics, a wall stood quietly in the record books: no Black African had ever won a gold medal. A continent's runners had been present at the Games — and never once at the top of the podium. On 10 September 1960, a soldier from Addis Ababa took the wall down with his bare feet. His racing shoes had fallen apart in the days before the Games; the replacements blistered, so he ran without. Level with Morocco's Rhadi Ben Abdesselem at the Obelisk of Axum — a monument looted from Ethiopia by Italian troops — Bikila pulled away, and finished alone at the Arch of Constantine in 2:15:16.2, a world best for the distance. The first Olympic gold ever won by a Black African, and Ethiopia's first of any kind. Then he proved it was no accident. At Tokyo 1964 he won again — this time in shoes — becoming the first man to win the Olympic marathon twice. A barefoot runner reached the Arch of Constantine first. And a continent stopped being a spectator at its own Games. #HindsightInFocus #Marathon #Ethiopia
20