D0 vs D1
Both received 62 observations and the same budget. The record reports no executable second candidate from D0; D1 selected a previously validated artifact to construct an intermediate feature and synthesize D2.
Recorded internal experiment · 7 September 2026
D0 → D1 → D2
Did a previously acquired capability causally enable acquisition of another under the same evidence budget?
Availability: internal case summary, not an independently reproducible public package. The hashes below identify reported artifacts; the corresponding private files are not downloads. The public continuity test and its signed proofs do not rerun this succession experiment.
Remove and restore
Removing D1 made D2 unusable; restoring D1 recovered it. Removing D2 removed its measured gain; exact restore recovered it. The candidate references D1 by hash.
Restart and transfer
The internal record reports 64/64 post-freeze holdout cases, a real Resident process restart, four longer-sequence transfer cases and a repeated parent ablation after restart.
Frozen identities
| Campaign | CUMULATIVE-3926197fd1f6 |
|---|---|
| D1 artifact SHA-256 | fe5026386f6244de2701f1809f4ea66d5812d738c33f904390d7881bff6143e3 |
| D2 artifact SHA-256 | 1d57ed77990fab34fa0f505444d2856c4f3e332697f17e285694d2244ebc79b4 |
| Bundle ID | bf017875c79d369cddbe4b7699e9431978c063f6da84b3c6ceb536a3a5109112 |
| Evaluator | Internal adversarial reviewer in the recorded experiment; no independent external replication is claimed, and no criterion relaxation was recorded. |
| Origin of candidate | The recorded lineage attributes D1 selection and D2 synthesis to DRENDA/Ω; the generic succession bridge and assessment infrastructure were externally engineered. |
Evidence trail
The internal freeze protocol, pre-restart result and final result were hashed as separate files. Their SHA-256 digests are listed below so a supervised evaluator can compare them with the original records:
| FROZEN_PROTOCOL.json | e2d03ad40c73521f43c6af2c24f17fe78c297dec8d84449a0882a226310a13f8 |
|---|---|
| PRE_RESTART_RESULT.json | 283c57d725dd07582bfb6913f37ebf556479576c5192ad16f767f0837bd8882d |
| FINAL_RESULT.json | 7e6200153d2dffb92f957a1743943045695310f9b4d4176a9db667094f6c1116 |
The published summary is not a substitute for access to the frozen implementation, raw inputs and independent repetition. The 64-case holdout remains private. An independent reviewer can request a supervised replay with a newly chosen post-freeze task, identical budgets, ablation, restore and restart; the outcome should be reported whether positive or negative.
What a new independent trial requires
Download the preregistration and reproduction protocol. It specifies a new evaluator-controlled task family and concealed holdout after the freeze, D0/D1 equal-budget comparisons, D1/D2 ablation and exact restore, fresh-process restart, transfer, provenance and failure reporting.
The protocol is a plan, not a claimed run. A sanitized executable package, permitted initial state, artifact bytes, independent Judge rules and evaluator-controlled execution environment must first be provided under an explicit agreement. Publishing those currently private materials requires a separate review and authorization; it must not expose the Resident, memory, credentials or original hidden test answers.
Existing public proof downloads · Capabilities versus available tests