Today / Archive
The Daily Edition
Thursday, April 9, 2026
The Index
◈ Science Experiments Ring the ‘Death Knell’ for Sterile Neutrinos 8 min
⬡ AI Eight Years of Wanting, Three Months of Building with AI 10 min
◉ Wildcard Are the Mysteries of Quantum Mechanics Beginning to Dissolve? 12 min
◈ Science
Experiments Ring the ‘Death Knell’ for Sterile Neutrinos
Quanta Magazine · Charlie Wood · ~8 min read

Thirty years of puzzling anomalies — reactor baselines, gallium experiments, accelerator signals — pointed toward a phantom fourth neutrino that would solve multiple mysteries at once. Now KATRIN and MicroBooNE have closed the door with high precision, ruling out the single sterile neutrino hypothesis. The anomalies remain, unexplained, and physicists are now forced to confront something considerably weirder.

Read on Quanta Magazine →

⬡ AI & Product
Eight Years of Wanting, Three Months of Building with AI
Simon Willison's Weblog · Simon Willison · ~10 min read

Developer Lalit Maganti spent eight years hesitating over a complex SQLite tooling project — then shipped it in three months with AI assistance. The resulting postmortem surfaces a sharp insight: AI makes poor design decisions feel consequence-free, quietly eroding architectural thinking. Required reading for anyone building non-trivial software with AI in the loop.

Read on Simon Willison's Weblog →

◉ Quantum Foundations
Are the Mysteries of Quantum Mechanics Beginning to Dissolve?
Quanta Magazine · Philip Ball · ~12 min read

Physicist Wojciech Zurek has spent decades developing decoherence and quantum Darwinism — a framework for how the classical world emerges from quantum mechanics through environment-induced selection of stable states. Philip Ball’s assessment: Zurek has “almost tied up the loose ends confounding physics for 100 years.” A provocative claim, carefully argued, and a useful counterpoint to the standard view that quantum foundations are permanently unresolvable.

Read on Quanta Magazine →

The Rabbit Hole
GLM-5.1: The First Open-Weight Model to Top SWE-Bench Pro

Z.ai’s GLM-5.1 — 754 billion parameters, MIT-licensed, fully open weights — just became the first open model to top SWE-Bench Pro, edging out Claude Opus 4.6 and GPT-5.4. The ‘open source is always behind’ narrative officially expired this week. What changes when frontier-level coding capability is also fine-tunable by anyone with the compute?

The W Boson Mystery, Settled

The 2022 CDF result measured the W boson at 7σ above Standard Model predictions — the biggest anomaly in years, briefly tantalizing. Now CMS at CERN has measured it at 80,360.2 ± 9.9 MeV using 100 million collision events: fully consistent with theory. Another ‘new physics’ window, firmly closed. The Standard Model keeps winning.

Curated for one. Published for practice.
The Daily Edition · groundstate.ink