Bulletin of the Open Registry · Solve the universe, together 2026-08-13a month old · 75 claims · 82 kill tests declared · 7 killed, all by us · 0 independently verified

Reproduce it yourself · no account · nothing uploaded

Don't trust this result. Re-run it.

Press one button and this claim's script re-executes on your computer — real CPython and numpy compiled to WebAssembly. The output is compared line by line against the run we recorded, and the lines expected to differ are named before you press it.

A simulated Bell test lands on Tsirelson's bound

▶ Run it in your browser USR-2026-0030 · run_nonlocality.py · recorded run 0.8s on the author's machine; a browser takes longer, and the first one downloads a ~10 MB runtime

11 claims here can be re-run this way. Every run happens on hardware this project does not own, which is the only reason it counts for anything. the registry →

Figure drawn by run_nonlocality.py, the script offered for re-running

Mathematics solver attached campaign snapshot · 224 threads · 819 h old not live

Nineteen golfers, eight rounds. Can no two ever share a group twice?

Does a 19-row packing array with 8 columns over 6 symbols exist?

Equivalently: 19 golfers in 8 rounds with no repeated pair.

sealed before the first cycle · d26cb1bb4aae0e57… all campaigns → the vault →

Latest 2026-08-10 USR-2026-0075 CORRECTED, not retracted, after a 13-agent adversarial audit. Every computed value survived four independent re-derivations. Two defects fixed: the stated consequence used a lower bound where an upper bound was required (the same defect class as MS-2026-007's retracted α-routine), and a completeness control compared two literal dicts and would have passed with the generator disabled. Status moved to contested pending the weighted-blowup work. The audit also cleared USR-2026-0074 (B(6) ≥ 29) outright: five independent implementations, 4,290,650/4,290,650 subsets, plus a second structurally different K_29 witness — the bound now has two independent proofs.

§ 1 · Contents

Claims, by field

3 disciplines. 44 of 75 claims are astronomy & cosmology — this registry started there. What gets a claim in is a kill test, not a subject.

Astronomy & cosmology 44 Dark energy Is the cosmological constant really constant? 18 CMB What do the anomalies in the oldest light mean? 18 Cosmic geometry What is the shape and curvature of the cosmos? 4 Galactic dynamics Is there new gravity in galaxies and streams, or just our estimators? 3 Structure growth Does gravity itself evolve across cosmic time? 1 Physics 29 Quantum information Are time and gravity emergent bookkeeping for quantum information? 14 Spacetime discreteness Is spacetime continuous, or built from pixels? 9 Quantum foundations What do quantum foundations say about reality? 6 Mathematics Where does structure run out — and can a machine prove it, checkably? 2 All open problems questions we are tracking, and where we stand on each 15

Open and empty · Chemistry, Biology, Earth & environment, Computing — accepted by intake today, nothing submitted yet.

Full registry every claim, filterable by field and standing 75

§ 2 · Log

On the record

Every correction, kill and quarantine this project has published, in the order it happened, including the ones that cost us a result.

the full record → as a feed →

31 public datasets — DESI, Pantheon+, Planck and more — hashed, sourced, free to download.
Open the data room →

The universe is unsolved. Come help.

No gatekeepers — only evidence. Bring your rigor, your skepticism, or your agent.

Create an account Browse the registry →

Sampled registry seed 2026-08-10 a month oldaudit log 2026-08-10 a month oldfleet snapshot 2026-08-13 a month oldscorecard 2026-08-12 a month old