Literary map and fixed-key probe

Date: 19 September 2026.

Result: no new plaintext. The comparison supports a blend of literary and esoteric motifs. It does not establish a unique external book, source order, coordinate system, or cipher key. A separate, explicitly exploratory repeating-key probe found no coherent plaintext in inspected candidate excerpts.

1. Source map

Locations below use identifiers in liber-primus__translation.txt. They identify passages, not original printed page numbers. The English translation is a community aid: original runes remain authoritative for cipher work. Similarity grades are qualitative judgments, not measured likelihoods.

Primary sources and stable textual locators:

• Emerson, Self-Reliance, essay body only. Editorial notes elsewhere in this collection are not treated as Emerson's text.
• Blake, The Marriage of Heaven and Hell, The Voice of the Devil; priesthood passage; Isaiah dialogue; final chorus.
• Crowley, The Book of the Law, I:3, I:54–56, II:54–55. The retrieved webpage repeats the work; those repetitions are webpage duplication, not authorial structure.
• Hesse, Demian, chapter five, consulted in the preceding comparison. Translation wording is edition dependent; its 2024 ebook release is not the novel's historical publication date.
• LP welcome transcription and the preserved local master translation for LP passage order.

Historical source use is a separate claim from any proposed influence on LP wording. Community walkthroughs document Blake in 2012, Crowley in 2013, and Emerson in 2014. These historical book ciphers were not independently reconstructed or PGP-authenticated in this session. Earlier descriptions treating Blake and Crowley as only hypothetical Cicada references were too weak and are superseded by this distinction.

2. Sequence and instruction audit

The solved opening proceeds through warning, pilgrimage, individual authority, mathematical wisdom, an identity koan, loss of divinity, a second koan, and instructions. The solved final parable returns to emergence, circumference, and inward divinity. That recurrence is an observable framing device, but 56 intervening LP2 pages remain unsolved in our working corpus, so this is not a complete narrative reconstruction.

Two simple forward-source predictions fail on the selected correspondences:

1. Blake: LP's holiness statement precedes its circumference discussion. In Blake, circumference occurs early, while the holiness statement closes the work. Adding the later LP inward-divinity motif produces source order late, early, middle. It does not rescue a simple forward or reverse traversal of the book.
2. Emerson: LP discusses preservation/possessions before adherence/conformity. Emerson's society/conformity passage comes before his closing property discussion. The selected correspondences therefore do not preserve forward order. Earlier inward/outward and personal-law parallels also prevent treating the entire mapping as a simple reversed traversal.

These are qualitative counterexamples to a simple traversal hypothesis, not statistical tests. Source assignments are partly interpretive, and isolated topics may recur elsewhere in a work. No significance level or claim of exhaustive exclusion is warranted.

The four explicit instruction blocks are attached to: self-command, four unreasonable daily acts, programming mind/reality, and questioning/following truth without imposing. They do not provide a consistent one-author-per-block assignment. Some motifs occur in multiple sources, and the explicit mathematical wisdom is a separate block. No fixed author cycle, repeated source-distance rule, or unambiguous source coordinates emerged.

The strongest interpretation is a thematically coherent selection or adaptation of ideas. That is our reading of the evidence, not knowledge of the authors' intent. Initiation and transformation fit a hero's-journey comparison; a return to society with a boon remains unestablished.

3. Bounded cipher experiment

Because the audit supplies no unique book-coordinate rule, no supposedly definitive literary book cipher was invented. Instead, a smaller exploratory question was tested: could a short source-related phrase serve as a repeating rune key, a mechanism with precedent in solved LP material?

The following ten candidate strings were fixed in code before inspecting any unsolved results:

The selection is exploratory and not exhaustive. Crowley's symbolic-alphabet passage was examined more closely after this candidate set was fixed; no extra keys were added in response to results.

Implementation: zero-based 29-rune indices; deterministic greedy GP transliteration; C/Q to K, V to U, Z to S. Test every phase of each key under three forms: c minus k, c plus k, and k minus c, modulo 29. Test section-continuous or page-reset counters, with no skips or with all ciphertext F symbols preserved and excluded from key advancement. Punctuation consumes no key positions. No reversed keys, added constant shifts, learned keys, selective exception subsets, or subsequent layers.

Each of eight substantial unsolved sections gets 1,452 settings, for 11,616 settings total. The nine-rune section is excluded. Counts include duplicate transforms and are not independent evidence.

Selection uses the first 60 percent, rounded down to complete words. The best phase from each key/reset/skip/form family competes in a common prefix ranking. The ten leading family winners per section are retained with complete decoded text. This guarantees inclusion of the global prefix winner, but is not necessarily the global top ten individual settings. Suffix scores never choose candidates. These suffixes have been inspected in earlier sessions and are not historically untouched data.

Six artificial controls use the real LP2 page-57 plaintext twice, with a page boundary, a SELFRELIANCE key and phase three. They cover all three algebraic forms and both counter-reset policies, without skipped symbols. Every control recovered the exact full 190-rune plaintext at rank one. This validates those implemented cases, not unknown real encryption or selective-F handling. The control plaintext is absent from model training.

Results, with normalized language-model log scores (not probabilities):

Opening and ending excerpts of all 80 retained candidates were inspected. None provided a coherent passage or a reason to promote it as plaintext. Whole candidate texts are saved for audit, but a full human review of every character of every candidate was not performed. The small English ranking model can miss a solution; this is a negative result only for the documented probe and inspection. It does not eliminate literary keys, book ciphers, non-English text, different transliteration, or more complicated encipherment.

Reproduce: run literary_key_pass.py with Python and NumPy. It reads the existing master transcription and the local lp_core/word_core modules, makes no network calls, and writes literary_key_results.json.

Decision

The exercise strengthens the case for examining Blake, Emerson, and Crowley as part of Cicada's intellectual vocabulary. It does not improve our confidence in any unsolved plaintext. Hesse and Campbell remain interpretive comparisons, not demonstrated keys or source attributions.

The most useful next literary work would identify an exact, distinctive source passage or a demonstrable link between a passage and original page layout/numerals. Without that selector, expanding the phrase list would mainly add arbitrary choices. Known mathematical instructions and independently reproducible cipher behavior retain priority.