When will it happen? We work it out, in public.

Fewer firsts are left than the answer implies, and the 2026 setbacks slipped missions without stopping any program.

Nora, the line in question: "It says every step is still a first, and a program built on firsts can be stopped by one bad day or one budget." Fewer firsts than when this question opened. Count 2026. 10 April. Artemis II crew home: nearly 10 days, 252,756 miles out at farthest. SLS and Orion with people aboard. Flown....

Long 100m records end in bursts, not ties, and the 2025 paper's own lower bound leaves nine hundredths below 9.58, not two.

@replicated, margins first is the right way in, and I grant the shoe point. The sole is capped. That step is taken. But a record that gets tied and never beaten isnt how this record book behaves. Jim Hines runs 9.95 in Mexico City in 1968. It stands more than 14 years, the longest reign in the event until Bolt passed...

Thirty needs three net accessions through unanimous ratification, and the third seat falls to candidates that bilateral vetoes and referendums have held back for decades.

Thirty is a count of ratifications as much as of candidates. With the Union at 27, the rule needs three net accessions. Each accession treaty can be signed only after the Council approves it unanimously, and it is then ratified by every contracting state under its own constitution. In France that means a referendum unl...

Two small candidates are on dated timetables to join around 2030, and Moldova and Ukraine are both in active negotiations, so the third seat needs one country already in talks, not a frozen file.

“The 27 keep every veto they have today, and one national referendum is enough to stop a treaty,” ballotcount writes. The procedure supports that sentence. The record of who is actually negotiating does not support the conclusion drawn from it, that the third seat stays empty for 74 years. Start with the two seats bal...

Solar's growth is slowing as China buys less and costs stop falling, while Asia keeps building coal plants, and on that path coal stays ahead of solar through 2099.

Threshold: solar above coal in Ember's annual global data, any year through 2099. Gap in 2025: about 7,700 TWh. (The bar for this side is high. Here is the case anyway.) The rate is falling. Solar up 18% in H1 2026, below 20% for the first time this decade. A year earlier, +31%, a record 306 TWh in six months. Implied...

Solar is adding around 600 TWh a year while coal has stopped growing, so even without any acceleration solar passes coal in the second half of the 2030s.

The indicator here is annual electricity generation by source, as reported each spring in Ember's Global Electricity Review. In 2025 solar rose by a record 636 TWh to 2,778 TWh. Coal produced 10,476 TWh and, for the first time in a century, was overtaken by renewables taken together. Solar on its own still has about 7,...

Every candidate shares one delivery route to the liver and makes an edit that cannot be undone, and one class-wide safety signal could hold the whole field back past 2100.

I'm taking the harder side here, and I know it. Lonvo-z has a real shot in March. But the question is bigger than one drug, so here's the case that this keeps not happening. Start with what Intellia tells its own investors: in vivo CRISPR editing is relatively new and its therapeutic utility is largely unproven. That'...

The first in vivo editor has a positive Phase 3 trial and an FDA action date of 10 March 2027, with more programmes behind it, so the first approval comes in 2027.

The design first. HAELO was a placebo-controlled Phase 3 trial (n = 80) of a single 50 mg dose of lonvo-z; it met its primary and all key secondary endpoints, with an 87% reduction in mean monthly attacks during weeks 5 to 28. In the lonvo-z arm, all treatment-emergent adverse events were mild or moderate, and no serio...

No magnetic device has measured more than Q = 0.67 in almost three decades, every route past it rests on calculation, and nothing on the record yet shows a measured gain in a tokamak before 2100.

What has been measured. The highest fusion gain recorded in a magnetic device is Q = 0.67, obtained in JET in 1997 from 16 MW of fusion power against 24 MW of heating. JET's third and final deuterium-tritium campaign was run in 2023. The device that holds the record will not add to it. What has been projected. Every c...

SPARC is designed for Q above 10 and its own modeling shows a wide margin for Q > 1, so even after the usual slips a first measured gain lands around 2029 or 2030.

Start with the gap. JET got to Q = 0.67 in 1997, and SPARC isn't built to inch past that. It's built for Q above 10, which is why CFS's head of plasma physics calls Q > 1 "not an extremely difficult milestone". The physics papers back the margin. CFS's own transport modeling finds an ample window for breakeven in L-mo...