v1 is the iteration built for the Future of Life Foundation epistemics competition. Everything below — analyses and docs — belongs to this iteration and keeps its URL under /v1/ permanently; later iterations land at /v2/ and so on. (What EpiStack is.)
A browsable knowledge base for the FLF epistemic-stack pipeline: turning one contested empirical question into a Bayesian, navigable Obsidian-style graph — sources, observations, hypotheses, evidence links with runnable likelihood/prior code, per-cluster reviews, and one final report that answers the question.
This site is the reading surface. The runnable pipeline that produces these knowledge bases lives in the same repository under pipeline/.
Start here
- Sample analysis — the final report — read one analysis top-to-bottom in ten minutes, then follow the backlinks down into the graph.
- Sample analysis — manifest and its cluster map — how an analysis is scoped and decomposed.
- Documentation — what the pipeline does, and how to read the node types.
What an analysis looks like
Every box in the debate is one markdown note with typed frontmatter, linked to the others:
- sources (
S-N) and data-bases (D-N) — the evidence base and the shared datasets it rests on. - observations (
O-N) — findings that follow ~with certainty from a source’s data. - hypotheses (
H-N), grouped into mutually-exclusive hypothesis-clusters (HC-N). - evidence-links (
E-N) — observation → cluster edges, each carrying a runnable## Likelihoodcode block. - arguments (
A-N), cluster-reviews (CR-N), and the main-report (MR-1) that answers the question.
The current sample (Sahul megafauna extinction) is a hand-written structural demo — illustrative numbers, but every node type, folder, wikilink, embed, and code block a real run produces. Real analyses replace it as they finish.
Entering late, deadline soon
This is a fast working end-to-end build for the Future of Life Foundation epistemics competition, not a polished product. Content is being generated; expect dummies to be swapped for real runs.