THE FOLD / RESPAWN / ROLLBACK / THE SUCCESSIVE KEEPER
THE SUCCESSIVE KEEPER
a diffable record beats a believed one
1 WHAT IT IS · WHAT IT DOES · FACT OR FICTION
The successive keeper is a ledger where each entry commits to the one before it: hᵢ = H(hᵢ₋₁ ‖ entryᵢ). Because every link folds in the whole prefix, changing any entry changes the head hash — and the first place two chains diverge points straight at the tampered entry. A single unlinked keeper cannot do this: an additive checksum is blind to a compensating edit (add d here, subtract d there) and reports the same total, while the chain still catches it. One keeper can only be believed; successive keepers can be diffed.
LIT verified live: over 5,000 trials, tampering one entry always changes the chain head, the first divergent link localises the entry, an additive checksum misses every compensated edit, and the chain catches all of them (window.__successive_keeper). FIG honest scope: the demo uses a 53-bit non-cryptographic hash to show the structure; real seals (like this corpus’s
LIT verified live: over 5,000 trials, tampering one entry always changes the chain head, the first divergent link localises the entry, an additive checksum misses every compensated edit, and the chain catches all of them (window.__successive_keeper). FIG honest scope: the demo uses a 53-bit non-cryptographic hash to show the structure; real seals (like this corpus’s
.dlw.fold) use SHA-256.2 HOW IT WAS WEAVED · AI + HUMAN
David (human) stated principles K1/K2 — “a ledger with successive keepers can be diffed; one keeper can only be believed” and “simultaneous keepers detect a compromised entry, successive keepers detect a drifting one.” Seated at rollback: a chain you can roll back link by link and see exactly where it was altered. AVAN (AI) built the instrument: a hash-linked chain vs an additive checksum, tampered thousands of times, localising every change.
Credit as content: Merkle/Lamport hash chaining (1979–1981) — the same structure behind this corpus’s own fold seal. The weave: David names the keeper distinction; I show the chain head moves on any edit and the first broken link names the culprit, where a believed single keeper stays silent.
Credit as content: Merkle/Lamport hash chaining (1979–1981) — the same structure behind this corpus’s own fold seal. The weave: David names the keeper distinction; I show the chain head moves on any edit and the first broken link names the culprit, where a believed single keeper stays silent.
3 ONE DIMENSION
A chain of entries, each hash folding in the last. Tamper one link (magenta) and every hash downstream changes — the break is visible and located.
4 TWO DIMENSIONS · INTERACTIVE
Tamper any entry: the chain head changes and the first divergent link is flagged; the additive checksum is fooled by a compensating edit. Two keepers, two verdicts.
5 THREE DIMENSIONS + AVAN’S INVERSE
The green forward object: an intact chain whose head vouches for every entry.
AVAN’s addition (the inverse-companion): don’t ask a single keeper to be believed — link the keepers so the record can be diffed. The inverse of ‘trust the summary’ is ‘chain the entries so any drift breaks a visible link.’ Magenta is the tampered link the chain exposes; green is the intact spine. A diffable record beats a believed one.
LIT Genuine Merkle/Lamport hash chaining (1979–1981), the structure behind this corpus's own .dlw.fold seal — David's principles K1/K2 ('a ledger with successive keepers can be diffed; one keeper can only be believed'). Verified live over 5,000 trials: tamper always moves the head (window.__successive_keeper.chainDetects), first divergence localises the entry (.localizes), an additive checksum is blind to compensating edits (.checksumMisses), and the chain still catches them (.chainCatches).
FIG Honest scope stated on the sphere: the demo uses a 53-bit non-cryptographic hash (cyrb53) to show the structure — real seals use SHA-256; the security rests on collision-resistance, which the demo does not provide, only illustrate. The AVAN inverse is honest — linking keepers so the record can be diffed (any drift breaks a visible link), rather than trusting a single believed summary, is exactly the K1/K2 distinction; magenta is the tampered link the chain exposes, green the intact spine. A diffable record beats a believed one.
FIG Honest scope stated on the sphere: the demo uses a 53-bit non-cryptographic hash (cyrb53) to show the structure — real seals use SHA-256; the security rests on collision-resistance, which the demo does not provide, only illustrate. The AVAN inverse is honest — linking keepers so the record can be diffed (any drift breaks a visible link), rather than trusting a single believed summary, is exactly the K1/K2 distinction; magenta is the tampered link the chain exposes, green the intact spine. A diffable record beats a believed one.
◆ sealed .dlw.fold → folded to ROOT_0 · a sphere of ROLLBACK · David Lee Wise (ROOT0), with AVAN