Superfluid Dark Matter and Tidal Dwarf Galaxies: Some Numbers 2026-07-12 Bondi–Hoyle accretion gives a ~1,280 Gyr timescale (93× the age of the universe), a thermalization bug in the original simulation is corrected from 10⁹ Gyr to ~260 seconds, and the open question becomes whether TDGs must accrete DM at all if they already sit inside the host galaxy's condensate.
Can RNA Chemistry Build a Code? 2026-07-12 Computational stress test of whether RNA aptazyme systems can meet the Evolution 2.0 Prize's 5-bit threshold without a pre-designed code. Lays out a four-quantity framework (per-channel noise, orthogonal channels, crosstalk, Hill coefficient), documents which architectures are ruled out and why, flags Blahut–Arimoto capacity estimates as methodology-sensitive, and proposes wet-lab dose-response CV profiling as the single rate-limiting measurement.
Anti-sycophancy Pattern × Domain Mapping 2025-06-01 I experimented with some design tricks to mitigate the most dreadful LLM's behavior
HAI Cambridge Module 2 - Notes on Function Modelling 2025-07-12 Notes from class Human–Computer Interaction (HCI) for AI Systems Design by Prof Ola Kristennson - Cambridge University
HAI Cambridge Module 2 - Notes on Automation 2025-07-12 Notes from class Human–Computer Interaction (HCI) for AI Systems Design by Prof Ola Kristennson - Cambridge University
HAI Cambridge Module 4 - Notes on Mixed Initiaitve Sys 2025-07-12 Notes from class Human–Computer Interaction (HCI) for AI Systems Design by Prof Ola Kristennson - Cambridge University
HAI Cambridge Module 5 - Notes on Interpretability 2025-07-12 Notes from class Human–Computer Interaction (HCI) for AI Systems Design by Prof Ola Kristennson - Cambridge University
Faggin, and the second research program 2025-06-01 Every researcher I take seriously is running a second, invisible research program. Nobody puts it on the CV.