watched the big labs keep fighting over their LLMs and realised we are paying them for the privilege of training them, like paying a chef who quietly learns your favourite dish and then puts it on their own menu
A research note on curation versus collection: why a mature, well-tended toolkit rejects free additions, how selection pressure (not quality) sets the bar, and a worked case cloning a 230-agent library that added almost nothing.
wrapped special term exams and just felt weirdly, overwhelmingly lucky to be alive: a fun job, friends who get me, a family that loves each other, and the sense of levelling up again
A research note on network effects: why a network's value grows roughly with the square of its users, why n log n is closer to the truth than n squared, the tipping point of critical mass, the adoption S-curve, and why all of it becomes a barrier to entry.