Momo Nine
Who pays for the robot's mistake? Reinforcement learning, from the cliff to the boiler room
From Sutton & Barto's Cliff Walking to BOPTEST and DeepMind's data-centre cooling - why RL's real question is who absorbs the cost of the agent's mistake.
The City That Publishes Its Own Bones: Mining Zürich's Open Data From a BIM Desk
Zürich publishes its civic data free on CKAN. Here is how a BIM or Grasshopper office pulls parcels, zoning and vacancy figures with ten lines of Python.
buildingSMART Certification in 2026: worth it — or just learn to read the file?
Is the buildingSMART Professional Certification worth it in 2026? A Swiss cred-eval — plus how to actually learn IFC 4.3 with IfcOpenShell in an afternoon.
Grasshopper from Zero: Your First 90-Minute Jump into Grasshopper 2
Grasshopper 2 beta adds inbuilt loops and a play/stop timer. A 90-minute first jump: slider, Series, point grid, one line of GhPython. Start tonight in Rhino 8.
Meta Switched Off Its Worlds in June. Nobody Could Export Them — Now Read Your BIM Contract
Meta removed Horizon Worlds from Quest on 15 June 2026 and community worlds were never exported — $88B of Reality Labs losses later. What AEC teams should demand about open formats before signing.
Pigeon Puts the Wait Back In — What a Slow Messaging App Knows About Your Model
Pigeon sells deliberate latency; DTN and IFC prove your model was always delayed mail. Read the IfcOwnerHistory postmark before the site trusts the file.
Graph engineering for the BIM desk: build your clash-check as a state machine
LangGraph turns clash-detection into an auditable state machine. A PAZ Academy hands-on tutorial: build a cyclic clash-check graph, route severity, keep humans in the loop.
Seven Numbers, No Memory, 96.5%: The 14-Byte Brain That Shames Your Render Farm
MINIMIO's 14-byte neural net solves 96.5% of unseen mazes with no memory — and quietly teaches AEC teams when reactive beats heavy tooling.
From toast to trusses: the thinking you learn before any language
Learn algorithmic thinking the way Oregon State's ENGR 112 teaches it — flowcharts, while/for loops, top-down design — built on Mermaid for AEC workflows.
Pattern Recognition on the Building Site: Teach scikit-learn to Sort Your IFC
Fisher's 1936 discriminant, four lines of scikit-learn, and your own IFC export: a hands-on classifier that triages dirty BIM imports for Swiss studios.
