[docxology]

Refresh the 2026 public snapshot from the current read-only program feeds and current affirmative participant-sharing answers. This is a draft 3.2.0 release candidate, awaiting organizer approval of the exact commit.

Candidate: 28689cf77719ba18861a12bf8c0de208084609e3.

Content and comparison

• 57 presentations, split 1 / 40 / 13 / 3 across the existing contribution labels.
• 18 UTC sessions: 10 on November 12 and 8 on November 13.
• 45 public participants: 15 additions, no removals or changed retained profiles.
• Presentation ID accounting: 31 additions / 6 removals = 25 new contributions plus six source-edit replacements. Session IDs: 7 additions / 3 removals = four new slots plus three replacements. No actual contribution removal is classified.
• All current titles, abstracts, attributions, labels and session fields retain the captured source values. Additional whitespace/title-formatting edits, Bradly's closing quotation mark, and Avel's HAL-link punctuation are called out in the private comparison.
• Session kind labels follow the current source: Live Keynote / Livestream Session / Workshop; the old snapshot used Livestream Keynote / Livestream / Workshop.
• Three abstracts remain pending: Karl Friston, Mike Levin, and the Holly Grimm joint contribution. No untitled sessions.
• 39 talks have source picture values; 38 portraits exported. Paulo Santiago's existing media omission remains recorded; no private-media substitution or consent expansion.

The explorer shows Program snapshot as of 2026-11-01 (UTC) and links the live program as revision authority, including its Sessions page. Timezone formatting is limited to scheduled sessions. Native books remain labeled as working drafts.

Implementation and validation

• Support finite typed Docs epoch values as UTC seconds; malformed epochs cannot fall back to display text. Plain session strings still need an offset; capture timestamps retain their explicit UTC requirement.
• Bind the optional snapshot date to source capture time; report optional event-field additions/removals correctly.
• Preserve native PDF glyphs using staged, hash-recorded font bytes from the existing Matplotlib dependency; reject missing-glyph compiler diagnostics.
• Full local suite: 648 passed, 9 skipped; 88.87% coverage. Locked sync, lint, formatting, wheel and sdist builds passed.
• Installed wheel tested outside the checkout with locked dependencies and external interpreter prefixes; its generated site matches source-rendered bytes.
• Public integrity, exact equality for all 57 presentation records / 18 sessions, private-data scan, and a deliberately rehashed email-leak negative control passed.
• Calendar: 18 UTC events, CRLF preserved, no attendees, organizers or invitations.
• Independent native review inspected 61 PDF pages and 58 Word pages; the Hawaiian ʻokina regression is fixed, and all 54 nonblank normalized abstracts are preserved.
• Browser checks cover search/reset, contribution filtering, keyboard disclosures, UTC/local times, light/dark layouts at desktop and 390/320 px, content/disclosures without application JavaScript, and downloads matching the reviewed hashes.
• Normal readiness verifies integrity and reports pending content. Strict readiness intentionally exits nonzero for the three pending abstracts.
• Independent source and frozen-candidate reviews found no blockers for a draft PR. Remote CI run 41364971062 passed all three Python jobs (3.12, 3.13, 3.14), including lint, formatting, tests/coverage, distribution builds, edition scaffold, and approved public-program/site rendering. Local Pages rendering and installed-wheel site byte parity also passed. Pages deployment remains approval-gated.

Bundle SHA-256: 4e677e70189dc54a6eb5e585d0e0a6fe3e983e4d87c55da493fef977c3ca0a33
Public manifest SHA-256: 40e62f423944a785aa591abeb732d00881f45879e2caa91d939b9cb6a23d1e84

Raw mappings, source IDs/responses, consent ledgers, detailed comparisons, render evidence and screenshots remain in private organizer storage. Earlier editions and archive content are preserved. These are source/integrity/rendering checks, not independent validation of research claims.

Do not merge, tag, release or deploy until an organizer approves the exact candidate commit. The private integration receipt is appended only after approved publication and deployed-byte verification.