What was found
The pilot-selected 128-step restoration achieved 2,124/2,304 supported successes at the shortest total budget, leaving 180 abstentions. At longer budgets it reached 2,304/2,304.
Task, conditions and denominator
Saved learned, retained and reversed OTB15 memories; synthetic working states; nine preparation candidates; total budgets 1,536, 2,048 and 4,096 steps. These are not naturally sampled continuous histories.
Controls and failures
Success required correct output, retention, memory dependence and event order. The best-candidate diagnostic covered the tested cases, but it is not a deployable selector.
External dependencies and limits
There was no additional evidence that learning broadened the feasible range while keeping the same answer. Neither E7-D readout transfer nor self-learning of preparation was demonstrated.
Read the primary report
The source snapshot is in Japanese; this page is a bilingual editorial summary. A report snapshot is not an independently reproduced result. Original relative links may require files from the source repository.
Open the preserved report (Japanese) ↗Source identity and verification scope
Source commit and path:
https://github.com/type37c/oscillator-arithmetic/blob/e3a5e1e66eae794a7fd56bae5f4d5054a74528c2/results/otb17_learning_readiness/REPORT.mdThe original development repository may require access. The report snapshot above is public. Only OTB21 has a full runnable public package; other entries are report-level evidence.
SHA-256: 42980edeb2b3211558392f354a2948f4d872dcdcad390fc364d4580c5a933b75
Source manifest ↗Concepts to learn first
Textbook revision 1 · Evidence snapshot: 05 Nov 2026 · Changes & corrections ↗