SCIENCE
Thirty Degrees in Zurich: the Ig Nobel Urinal Teaches the Year's Best Detail
The 2026 Ig Nobel physics prize, awarded in Zurich, went to a seashell-curve urinal. Its 30-degree impact rule belongs in every wet-room detail.
Brain-computer interfaces: when intent becomes an input, every building action needs a second path
From Berger's EEG to EPFL's digital bridge: how brain-computer interfaces work, what they cost, and what Swiss architects must specify before adoption.
A 50% Better Rain Forecast Meets the Bauleitung: Read the Grid and Error Bar
Google claims WeatherNext 3 gives 50% better rain forecasts a day ahead. What a 0.25° grid cell means in Zurich, and how Bauleitung teams should test it.
Chair44: a cube missing one corner that fills space and never repeats
A seven-cube chair with three colours of marks fills space but never repeats. How Chair44's 8-page proof works, and why one part needs one mould.
The Photon Pushes Everything Equally. Only the Mass Decides.
From Sänger's 1957 photon rocket to IKAROS and Würzburg's light-steered swimmer — one equation, two verdicts, and the ratio habit every AEC desk needs.
A fruit fly's 166,000-neuron wiring diagram is now a file you can query
Google and HHMI mapped 166,000 fruit-fly neurons into a downloadable graph. Why a connectome is not a mind — and what it teaches architects about querying BIM models.
A Cell Built From Scratch — and the Dependency Graph Hiding Inside It
The first cell built from scratch grows and divides but can't make its own ribosomes. What that single point of failure teaches every AEC office's toolchain.
TON 618: the black hole that shrank 38% without moving — a lesson for engineers
TON 618's mass fell 38% in 2019 with no new data — just a different emission line. A physicist's lesson for anyone trusting a structural digital twin's numbers.
Catenary: The Curve That Decides Where the Force Goes
How the catenary curve (y=a·cosh(x/a)) form-finds arches and vaults — from Hooke and Gaudí to ETH Zürich's Thrust Network Analysis, with runnable Python.
The Geometric Wall: Why Your Sparse Autoencoder Stops Scaling Inside Curved Layers
A new arXiv paper shows sparse autoencoders hit a geometry-dependent reconstruction floor that more dictionary atoms cannot fix — it lives in the layer.
