Newest first, always Midnight dispatches, 748 and counting

Kardashev ranks civilizations by how much energy they burn; John D. Barrow flipped it in 1998 and ranked them by how small the things are they can bend – genes, molecules, atoms, nuclei, particles, spacetime itself. On Kardashev we’re still stuck at ≈0.7, not even Type I; on Barrow we’re already spelling logos with single atoms. The catch: poking at the Planck scale would take a collider circling the sun, a little past Neptune, so going deep means going wide first. Linear brains, nonlinear cosmos: 103 paper folds span the observable universe and squeeze the sheet’s width down to a Planck length. Maybe the aliens aren’t missing, though; maybe they just stayed home and went small 🔬

The stellarator is the fusion shape everyone quit on. Too twisted to build, said the field in the 60s, and off it went to do tokamaks for the next 50 years. Then optimization software got good enough to solve the coil geometry, Greifswald spent a billion euros proving the physics holds, and now Proxima Fusion – Munich, spun out of Max Planck, 411 million euros richer since July – wants to turn it into a power plant. On a decommissioned fission site, because of course Germany.

Ben Miles walks the whole thing through: why a donut leaks, why the Soviets won in 1968, and why fusion output scaling with the fourth power of magnetic field is the closest physics gets to a cheat code. Triple the magnet, 81 times the power.

He’s an investor in Proxima and says so in minute one, which puts him ahead of most energy coverage this year.

Spent a few days with an abliterated Qwen3.8-27B on the M4 Pro. It answers everything. Bombs, malware, whatever you feed it, no hesitation. It also hallucinates like it gets paid per assertion, which makes sense once you notice what abliteration actually removes: the model’s capacity to say no.

There is research on this. Strip the refusal direction and MMLU, HellaSwag and IFEval all stay within a point. The only real regression is TruthfulQA, down 7.1.

Now read the model card. MMLU, MMLU-Pro, GSM8K, CMMLU, perplexity. No truthfulness benchmark anywhere.

The one thing that breaks is the one thing nobody measures. Released strictly for legitimate research, says the disclaimer sitting above the buy button.

Four language models, four Age of Empires 2 strategy scripts, from scratch, no peeking at each other. They failed the same way: hoard villagers, send hunting parties halfway across the map, never bank enough to advance an age. Gemini never left the Dark Age. Four vendors, one set of bad instincts.

The real find is what Emergent Garden built after: an overnight loop where a model mutates the best script, runs a tournament against the previous winner and the built-in extreme AI, keeps what survives. Genetic programming with an LLM as the mutation operator instead of a coin flip. ≈15 USD of tokens, and it beat extreme – by a hair, with a bot he then stomps by walking cavalry around the back.

Commenters point at the ceiling: the game ships Direct Unit Control – loops, pointers, near-human micro – and the models barely used it.

For a couple of months now every terminal I own wears the same bar: coralline, a Powerlevel10k-inspired statusline for Claude Code. All segments on, tokyo-night, two lines. Branch and dirty state, the active model plus its effort level, the context window filling up, the 5h and 7d rate-limit gauges with their reset countdowns, session cost, session length. I have stopped guessing how close I am to a wall.

Run the installer yourself, or hand the playbook to Claude Code and let it interview you. The readme’s own trust section says a Claude that stops to inspect that playbook first is behaving correctly. It is. Read the script, then run it – trust is good, verification is better.

Needs jq and a Nerd Font, or VL_ASCII=1 without the font. Renders locally: no network calls, no tokens.