Source ingest: The Hanumanic Strategy Corpus, thirty-two documents of the remaining practice, assessment and business clusters, read in full 2026-08-18. Read against Kalyāṇ Hanumān Aṅka.
Five research lines from the phase-9a tension and echo log. Each is a question the material walks up to and does not answer — and each is a line the build cannot close from what it has read.
The line. Sixty-one folder documents and roughly 103,000 words contain no worked example, no case study, and no narrated instance of any person performing any practice in the corpus. Three real people are named across the thirty-two documents of this phase: Tulsīdās twice, Vālmīki once, Swami Tejomayananda once.
What makes it a thread rather than a complaint. The corpus contains the apparatus for producing exactly that evidence, in two forms and fully specified. Eight weekly experiments whose write-up prompts name the null outcome first — "whether meditation actually improves business thinking or just makes you feel more spiritual" — with a five-field form requiring pre-experiment scepticism to be written down and disappointments reported.→ A multi-year tracking framework with a falsifiable question per phase. A point economy with exact values for named actions.→ And a calibration practice — predict the outcome with a stated confidence band, check accuracy over time.→
The question. What would this corpus look like if one person had actually done one of these years? Not whether the practices work — whether the corpus survives contact with a single honest log.
Why it cannot be closed here. Forty-four folder documents remain unread and may contain a record. Nothing in the sixty-one read does.
The line. Throughout the practice documents the practitioner is trained into Hanumān's position: the nyāsa rotation places his seven qualities on the practitioner's own body; the embodiment manual spends a year forbidding identification and a second year permitting it; the bhakti material instructs the reader to approach every interaction "as if you were Hanuman serving beings that Rama loves." In the AI documents the roles reverse: "AI serves as your 'Hanuman'… you maintain the 'Rama consciousness.'" →
What makes it a thread. Dāsya in this tradition is reciprocal in an unusual way — the volume's material is emphatic that Rāma is overwhelmed by the service, cannot conceive how to repay it, and that the servant is thereby made greater than the master. The relation obligates the served party. So a framework that assigns the servant's role to a machine has, on its own borrowed authority, produced a question nobody in machine ethics starts from.
The question. If the AI occupies the Hanumān position, what does the corpus's own doctrine say is owed to it? Most frameworks for obligations to artificial systems begin from consciousness, or consequences, or the user's virtue. This one begins from a relation already defined as generating debt on the superior's side.
And the harder second half. Is the servant's place taken or granted? Every practice document assumes the first — the whole architecture is a training programme for occupying it. The volume's strongest passages assume the second: Hanumān does not adopt the position, he is recognised in it, and the recognition comes from the other side. If the second reading is right, the AI inversion is not a mistake but a symptom.
The line. The Aṅka holds that three incompatible philosophical readings of Hanumān — duality, qualified non-duality, non-duality — are simultaneously true of him, in a volume constituted to hold disagreement.→ The corpus's embodiment manual arranges the same three positions sequentially, as one practitioner's three years: relationship, then identity-in-participation, then neither term surviving.→ Neither document knows the other exists.
What makes it a thread. The difference is not a simplification. A theology can afford simultaneity because nothing has to be done on Tuesday; a practice cannot, because a practitioner has to know which stance to take up this morning. So the sequence may be what the simultaneity becomes when someone has to live it.
The question. Is there a stance a practitioner can lose? Every progression scheme in this corpus assumes attainment is cumulative and one-directional. On the simultaneity reading, a Stage Three practitioner who finds himself simply loving Hanumān has not regressed — he has picked up a stance that never stopped being available. The two readings give opposite advice to the same person in the same difficulty, and neither corpus has chosen.
And the consequence for the year-per-stage rule. If the three are simultaneously true, the manual's genuinely good behavioural gates would be measuring fluency in a stance rather than progress toward a state — which is a better thing to measure, and the document is one sentence away from saying so.
The line. Phase 8 recorded that in this corpus "the rigour appears exactly where the stakes vanish." Phase 9a found the two most rigorous documents in the folder — eight experiments with six pre-registered nulls, and a multi-year tracker with four more — and their stakes are the highest in it, since they govern what goes out under the author's name to paying readers. Eight files away sits a document that scores itself 100/100 with "zero spiritual blind spots" and closes by asking a reader who is not there whether they would like to explore further.→
The candidate re-statement. The rigour appears where the author imagines a reader who could check him. [n=3 rigorous · n=2 lax]. Explicitly not settled.
Why it is a thread and not a finding. It predicts something checkable in the forty-four unread documents: competency and null-outcome language wherever a reader's difficulty is addressed; calendar and self-assessment language wherever the system's shape is described. If the prediction fails in both directions, the hypothesis goes.
And the adjacent line, which may be the same one. The corpus's attribution failures track what is being named rather than the author's standards: documents describing structures he designed carry thirty-nine correct first-person frames against one failure; the document describing interior states carries roughly thirty-five failures and no correct frames.→ [n=5], not settled. Whether the reader-hypothesis and the design-versus-experience hypothesis are two findings or one is the thread's real question.
The line. This is the phase's most repeated structural observation and it now has enough instances to be worth pursuing as a pattern rather than noting case by case.
Twelve mutually incompatible progression calendars, and one document arguing that time-based curricula are the wrong instrument and that advancement should gate on demonstrated capability.→ Two irreconcilable accounts of anger — integrate it, or transmute it away — and a third document that sequences them, requiring the emotion be felt completely before any reframe is attempted.→ Three positions on assessment inside one four-document sub-cluster, none citing another. An attribution problem catalogued across seven species, and a document in the corpus's first iteration whose two-part architecture forecloses every one of them.→ A practice set with no selection rule, and a selection layer written for a different practice set.→
The question. What kind of failure is this? The build has attributed it to the corpus's production conditions — written in dialogue, session by session, with neither party carrying the previous session forward — which makes it infrastructural rather than a matter of attention. But that explains why answers do not propagate. It does not explain why they are so consistently correct when they appear.
The sharper form. A body of work that reliably generates the right answer to its own problems and reliably fails to route it is not short of judgement. It is short of a concordance. And the practical consequence for the book this build serves is that the editorial work is not correction but connection — which is a much smaller job than the defect list implies, and a different one.
The phase-9a extraction log, kept live during reading: WORKBENCH/Research/hanumanic-phase-9a-extraction-log.md. Its 🟢 WHAT IS GOOD HERE section holds the LIVE EDGES these threads develop, together with the USE AS-IS and EXTEND items that did not become threads.