// the find
danielmiessler/Substrate
An Open-source Framework for Human Understanding, Meaning, and Progress.
Substrate is Daniel Miessler's attempt at a structured knowledge base for tracking problems, solutions, claims, and evidence, paired with a growing collection of economic, health, and social datasets (GDP, CDC mortality, EPA air quality, etc.) pulled via small TypeScript/Bun scripts. It's aimed at OSINT researchers, policy wonks, or anyone who wants government data pre-fetched into CSV/JSON with documented provenance rather than a general-purpose dev tool.
The data sourcing is genuinely solid — FRED, CDC WONDER, Census ACS, BLS JOLTS, EPA AQS are all wired to real APIs with update scripts, logs, and per-source README documentation of methodology and update frequency. Each dataset ships as plain CSV/Markdown, so you can use the data without touching any code. The fetch scripts are short, readable TypeScript files you can audit in a few minutes rather than trusting a black-box pipeline.
The 'framework' part is mostly a directory naming convention (Problems/, Solutions/, Claims/, Arguments/) with no code that actually links, queries, or validates entities against each other — it's markdown files with IDs in the filename, not a graph or database. There's no schema enforcement across the ~30+ per-dataset update.ts scripts, so consistency depends entirely on each contributor copying the pattern correctly. Test coverage is sparse (a handful of checks.test.ts files scattered in a couple of dataset folders) for a repo that claims to be infrastructure. The README is almost entirely landing-page copy about 'human progress' and 'collective intelligence' with very little said about how the pieces actually connect at the code level.