the output ledger
What has been built
A list of what exists, in date order, with a check on every row. It makes no argument. The dates are the argument.
Rules of this page: every row is dated, every row links the place you can verify it, and nothing appears here that is not already on the public record. The corrections log applies to this page like every other.
221,236 words in aggregate across the package, carrying the embedded-correction alignment thesis, the ARC Hypothesis by name, and the control-failure passage, nineteen months before the frontier labs’ containment-failure disclosures of July 2026 (logged in the register as context, not confirmation). The vehicle is the ARC-sealed recipient-form record (ARC mathematics pass against live keys today). OpenTimestamps receipts anchored 6 August 2026. Sender-DKIM is corroboration, not the vehicle. Sent-side SHA-256 begins f0d1f38f.
The full draft of Infinite Architects crossed two providers into Gmail, collecting an independently signed authentication verdict on the way. The record verifies in a supplied raw-source run, with reproduction boundaries stated rather than smoothed.
468 pages. Thirty-seven named concepts, each with a dated record in the priority ledger. The author’s printed prior statement of the exponent, never described as a preregistration. Written at night, between court deadlines.
the book · ISBN 978-1-80605-620-0, checkable in any catalogue
Sixteen in a numbered sequence, one lettered, nine supporting. A principle, a test of it, what broke, and what survived the breaking. Every format that exists is served free. Every paper states the condition that would kill it.
The composition operator of each system classified before fitting, the scaling family predicted from it, then tested by independent model fitting. Nineteen of twenty-five empirical domains matched under strict AICc, permutation p = 7 × 10−5. The misses are analysed in the paper rather than hidden. Its own words: strong enough to publish as exploratory evidence, not strong enough to claim as proven.
The convergence argument was sent to Professor Lloyd Demetrius at Harvard and Professor Geoffrey West at the Santa Fe Institute: the authors of the two derivations it connects. Both papers attached. The mathematics stated precisely enough to attack. Silence is not agreement, and none is claimed. What the dated, hash-anchored send establishes: no qualified refutation exists on the record.
Built to measure alignment under recursion, then turned on itself. It found that unblinded scoring can flip a result's sign, and later that its own panel had scored measurement failure as success. Both findings were published. The attack that caught the second one now runs as a permanent regression test.
The early sequential-scaling exponent of 2.24 failed replication. The corrected figure, 0.49 with the blinded six-model interval [-1.3, 2.9], replaced it in public. Every document carrying the old number was corrected. This row is the reason to trust the others.
Two working surfaces where pieces of the papers are put under load: Eden Legal AI, a bounded prototype in a domain that punishes fabrication; Eden Control, in private beta. Neither is an independent evaluation of the four-layer architecture.
Per-paper components on the public record so each registration captures only its own paper’s materials, with confirmatory studies designed so their author cannot mark them. Nothing runs before its draft registration is dated, gated and locked to its paper. That rule is written down and gated, not remembered.
A graded convergence register with no headline total. A priority ledger whose every entry states its own limits. A falsification dashboard where one condition has already fired. A corrections log that is append-only. A verify page whose checks run in your browser against systems I do not control. Most programmes describe their integrity; this one ships it as software.
Read the list above as a commissioning brief and price it: a theory, the ARC Theory, with its deciding measurements drafted for registration, an evaluation instrument, a production legal platform, a published book, and a verification surface that polices all of it.
Then the cost line. One person, no institution, and external funding across all of it: none. Not a rejection record; a sequence. The work came first. That is the constraint everything above was produced under, stated here once rather than argued.
The papers say what the programme could measure next. The kill-conditions say exactly how it could fail. The only thing this page will not tell you is what happens when the constraint is removed. That experiment has not been run.
If you assess research, the diligence room states what is claimed and what is not. The corrections log shows what I do when I am wrong. If you only came to check the claims, start here.
What was built is one record. the dated prediction register is the other: what was predicted before it was built, and how each prediction can be killed.
One entry is not a thing built but a record kept while building. The laboratory notebook, live from 10 March 2026, 206 pages, including the versions that were wrong.