# Cicada 3301 research checkpoint

Latest status: sixteen sessions completed; no verified new page solved.

- SESSION_1.md: initial arithmetic and affine searches, known-solution reproduction, and period probe.
- SESSION_2.md: section-marker resets, adaptive F-skip search with shuffled controls, and verified base-60 byte-key tests.
- SESSION_3.md: verified number-square reconstruction and bounded prime/Fibonacci combined-key tests, with controls and shuffled comparisons.
- SESSION_4.md: held-out scoring checks, word/phrase mapping, adjacent-repeat deficit, and bounded word/rune feedback tests.
- SESSION_5.md: community-chart corroboration, adjacent-pair-preserving phrase comparison, and changing-alphabet/exclusion tests.
- SESSION_6.md: phrase-constrained unknown periodic-key fitting, three controls and unchanged cross-section validation.
- SESSION_7.md: exact joint 34-rune decoding, failed held-out recovery controls, and two randomized comparisons.
- SESSION_8.md: historical rune comparison, Old English/Latin frequency benchmarks, and rune-name key tests.
- SESSION_9.md: cuneiform numeral reading, ancient-script evidence audit, illustration inspection, and bounded numerical-key tests.
- SESSION_10.md: documented Gibson connection, cypherpunk comparison, and limited early Bitcoin code inspection; research only.
- SESSION_11.md: validated Agrippa decoder and direct three-alignment byte-block test; no readable ASCII result.
- SESSION_12.md: exact reproduction of the 2012 Agrippa book cipher from a 2011 web snapshot; limited transfer checks failed.
- SESSION_13.md: source/instruction audit, Blake parallels, and 16 direct interior-extraction candidates; no new plaintext.
- SESSION_14.md: six nonlinear word-indexed schedules with recovery controls and a boundary-gap rotation comparison; no positive result.
- SESSION_15.md: literary source map and order audit; ten fixed literary keys, six recovery controls, 11,616 settings, no positive result.
- SESSION_16.md: corrected four-marker layout; 15 fixed reading-order tests with exact constructed recovery controls.
- Scripts and JSON results: reproducible experiments and failed candidates.
- provenance.json: sources, checksums, and validation limits.

The candidate outputs in JSON are retained for audit. They are not solutions. All work is bounded to the documented methods; no background task is running.