Continuing from Part 1, where we learned what git for data is, how the architecture and use cases work, how you can achieve git-like functionality with different approaches, and how the key is to avoid moving data as much as possible to keep state that can be referenced and rolled back to, but at the same time saving cost by not duplicating all data every time you create a new branch.
The attempt on Friday by Secretary of War Pete Hegsted to label Anthropic as a supply chain risk and commit corporate murder had a variety of motivations.
LLMs are an epistemic nightmare
The road to where we are now was (mostly) paved with good intentions — but mixed with too much uncritical acceptance of hype.
What might a superintelligence arcology be like?
This is the long version of what happened so far.
The writer Tyler Austin Harper (of The Atlantic, etc.) sent me a thread this morning, asking whether a mistargeting yesterday that killed nearly 150 school children in Iran could have been the result of AI.
We will learn a lot about Silicon Valley in the upcoming days
”No LLM, anywhere, in its current form, should be considered for use in a fully lethal autonomous weapon system. It's ludicrous even to suggest it.”
Will AIs be jealous of one another?
Will 2026 be looked back on as the pivotal year for making decisions about the singularity?
I always wanted a personal knowledge assistant based on my notes. One that uses Obsidian’s backlinks and connections to surface ideas I’ve forgotten or never thought to link together. So I built one. A RAG system that runs locally with DuckDB as a vector database, then syncs to MotherDuck for a serverless web app running entirely in the browser via WASM. Think of it like J.A.R.V.I.S1 for your markdown files: search about a topic, and it shows connected notes up to two hops away, semantically…