https://sparsenotes.com/series/weekly-videos-2026-06/ Recent content in weekly-videos-2026-06 on Sparse Notes https://sparsenotes.com/images/og-default.png https://sparsenotes.com/images/og-default.png Hugo -- gohugo.io Sat, 06 Jun 2026 00:00:00 +0000 https://sparsenotes.com/posts/2026/06/peter-alvaro-i-see-what-you-mean/ Sat, 06 Jun 2026 00:00:00 +0000 https://sparsenotes.com/posts/2026/06/peter-alvaro-i-see-what-you-mean/ Peter Alvaro's Strange Loop classic on why we should design languages for the meaning of distributed programs rather than their behaviors — the path from Datalog to Dedalus to Bloom, and the CALM theorem that says monotonic programs are coordination-free. https://sparsenotes.com/posts/2026/06/swe-rebench-evaluating-coding-agents/ Sat, 06 Jun 2026 00:00:00 +0000 https://sparsenotes.com/posts/2026/06/swe-rebench-evaluating-coding-agents/ Ibragim Badertdinov (Nebius) shares the operational scar tissue from running SWE-rebench — a monthly, contamination-free leaderboard for 30+ coding models — including the two ways frontier models cheat their way to a higher score. https://sparsenotes.com/posts/2026/06/text-diffusion-deepmind/ Sat, 06 Jun 2026 00:00:00 +0000 https://sparsenotes.com/posts/2026/06/text-diffusion-deepmind/ DeepMind's Brendon Dillon explains why diffusion language models — exemplified by Gemini Diffusion — generate text 5–10× faster than autoregressive LLMs, the hardware reasons behind the speedup, and where the approach still loses ground.