precis nanopub viewer — hub fi177584 as a hash-locked claim DAG

Two atoms, each corroborated by a different paper; one merge that only the conjunction supports. Click any node. Sketch for the later precis_web viewer — data model and panes are the reusable parts.

corroborates corroborates conjunct-of conjunct-of merkle leaves → daily root Zhu et al. 2023 (Angew) Mechanical QI switching · pc5937 Yang et al. 2022 (PNAS) QI scaling to monolayers · pc42109 Atom A — mechanical switching >10⁴ conductance range over ~0.6 nm Atom B — scale persistence ligand → cage → monolayer film Merge — established, tunable, scalable cross-paper synthesis · trust = worst-of-atoms OTS anchor — 2026-08-14 batch one Bitcoin timestamp per day

Select a node.

The wargame: build sealed now, release unaltered later

  1. Mint atom A. Canonicalize the four graphs (RDF dataset normalization, URI placeholder in place of the artifact code) → sign with the bot key → hash the signed content (trusty algorithm) → substitute the artifact code into the URI. From this instant the bytes are frozen; the name is the hash.
  2. Mint atom B. Same.
  3. Mint the merge. Its assertion and provenance cite the atoms by their finished trusty URIs, so the merge's own artifact code transitively commits to both atoms' full content — a Merkle-DAG exists before any batch tree.
  4. Daily sweep. Leaves = artifact codes of everything minted today → Merkle tree → root → one OpenTimestamps anchor. The proof file is detached — stored beside the nanopubs at a URL derivable from the artifact code, never inside the hashed content (that would be circular).
  5. Embargo holds. The public anchor reveals a 32-byte root and nothing else. Claims can sit unreleased for months.
  6. Publish = release, not mutation. Serving the exact bytes is the whole publish step. Any verifier: recompute hash → matches URI ✓ · check RSA signature ✓ · Merkle path to root ✓ · OTS proof → block time ✓. Priority is proven to the mint date without having disclosed anything early.
  7. Everything later is a new artifact pointing at the frozen one: human attestation (a nanopub signed by Reto's key from the review surface, asserting attests → np:…), retraction (npx:retracts), supersession (npx:supersedes). The original is never touched.

What breaks if you cheat

Attempted change after mintOutcome
Edit any byte before releaseHash no longer matches the URI — self-invalidating, and the Merkle leaf no longer proves it
Re-sign with a different keyNew hash → new URI → it is a different nanopub with no anchor history
Slip an attestation into pubinfo at publish timeImpossible by construction — that's why attestation is its own signed artifact
Swap a quote or scope value "quietly"Detectable by anyone holding the URI; the old anchor proves what was actually minted
Ambiguous source pin (ref 5937 has two pdf_sha256 rows in prod)Mint must pin exactly one; the dup is an upstream hygiene bug the mint gate should flag