What we have not decided yet

Foundation for harassment, illegal content, minor safety, doxxing, and harmful content policy. Also a governance question about whether T&S is Creator reserve power, separate body, or under Arbiter review.

Stance on usable IP, DMCA compliance, contributor license, Commons publication license (CC-BY-SA?), and rights-holder takedown handling.

Foundation+PBC is the current lean. Need-vs-Greed and acquisition commitments must be expressible as enforceable corporate-document language.

Account verification model. Multiple accounts per person, deceased contributors, account portability, account recovery.

How Fact entry, Commons proposals, LoreLines, and Home content are authored. AI-assisted by default, with structured-data entry that doesn't feel like a tax form.

Specific waterfall percentages, ad envelope details (formats, density caps, ad-free page types), AI licensing specifics, B2B revenue posture.

Board size, composition mix, term limits, succession, removal, compensation, conflict-of-interest policy. Existence public; specific candidates internal.

Per-provider license review required before each ingestion stream goes live; TMDB commercial terms and Fandom CC-BY-SA are the active blockers.

Add Constitutional language for provider-as-source attribution, Source status formalization, and royalty/attribution flows back to providers.

Legal opinion required before any Fandom-derived content enters Commons. Holding pattern: Fandom as reference-only signal until resolved. Also scopes the fact-extraction boundary and how contributor-pasted copyrighted prose is handled.

Decide how OLN overtakes Fandom's entrenched search ranking. Direction set by Entry 030: AEO wins citation, not traffic, so OLN plays two boards — be the most-cited source and own the world-model experiences a chat answer can't collapse — with rights-holder partnerships breaking incumbency. The migration pipeline brings content, community, and audience, not just the CC-BY-SA text.

Design the authoring loop that turns natural prose into Facts/entities/links with AI proposing and humans confirming, so structured contribution feels like writing, not data entry.

Operationalize the "citable, not trainable" posture (Entry 030): a per-crawler-class allow/block policy, WAF-level enforcement behind robots.txt, and licensing tiers that keep OLN freely citable while holding the structured graph in reserve as the asset a partnership buys.

Operationalize the "gamify quality, not volume" decision (Entry 031): the gold-salting + reliability-weighting spine that decouples fun from authority, the component catalog (each as proximate-win + quality guardrail), the newcomer practice ramp, and the measurement methodology.

The review queue where an AI-proposed or member-authored Fact shows its diff, source, and confidence, collects N reliability-weighted confirmations, and graduates Provisional→Canonical — unifying the contribution pipeline (G-013/G-057), the consensus mechanic (G-059), moderation, and data quality (G-026/G-012). Design drafted below; parameters remain open.

Define how prose binds to Facts: explicit markup vs. AI-detected links vs. coreference resolution, plus the confidence metadata on each link — the primitive that decides whether prose-as-a-view-over-data (Entry 029) actually works.

Stand up the experiment/feature-flag layer (Amplitude Experiment via the existing @amplitude/unified SDK): variant assignment, flag gating, holdouts, and the paired engagement + data-quality guardrail events that every gamification and ranking experiment reads.

Policy for promotional media (posters/trailers/stills): they are copyrighted, not PD, and "distributed for marketing" is permission-with-conditions, not a public dedication. Store verified pointers + provenance and embed the official drop; use the PD subset (Commons/Internet Archive) freely; license the press feed for current art. Day-one coverage is a monitoring + partnership problem.

AI-generated content policy: contributor use of LLMs, AI-content labeling, training-data licensing, AI-as-contributor.

Specific waterfall percentages, Franchise distribution formulas, R&D pool governance, free-rider treatment.

Visitor → reader → LoreLines contributor → Commons/Home contributor → Team member ladder. Mentorship structures and how new accounts surface.

Specific thresholds for stacked-supermajority + cooldown + Foundation board ratification process from Entry 014.

Auto-flagging, rate limits, new-account restrictions, automated reversion, the role of Franchise Team in moderation.

Version history and rollback authority across surfaces; protection levels; touches Commons proposal model.

Contributor export rights — what data, in what formats, under what license. Probably constitutionally guaranteed.

Hard limits on Franchise Team authority, intra-Team appeal pathway short of Arbiter escalation, term limits and successor selection.

When a ban in one Franchise extends across the Network. Who decides, with what threshold, with what appeal.

Charter-level commitment that OLN is a platform — OLN does not produce editorial content competing with Franchises.

Persistence under canon change, Credit weights, correction-to-Commons promotion, citation-in-Commons mechanics, Featured LoreLines, type-specific governance.

Source attribution per Fact, QA pipeline for imports, confidence scoring for AI-extracted Facts, conflict resolution when sources disagree.

Pipeline architecture for wikitext-to-structured-data extraction. Existence public; specifics internal until coordinated with Memory Alpha admins.

Specific dilution rate, full enumeration of contribution categories, treatment of sudden growth events, process for adding new categories.

Lock durations, conviction multipliers, return-timing asymmetry, per-class lock mechanics, quadratic voting parameters for constitutional matters.

How readers respond to Journal and Register entries — what the default visibility is, who sees what when, and how exchanges become public.

Operational parameters for the reaction system: reaction vocabulary, weighting between anonymous and registered, gaming resistance, and the connection to Sybil resistance broadly.

How much credit pre-launch registered users accrue, what actions earn it, and how it converts into Power at launch — all intentionally not pre-committed.

How the public roadmap is structured, how credit values are set per project, how outside contributors claim and submit work, and how founder time is logged against the same denomination as contributor work.

Benchmark Postgres + AGE + pgvector against expected traversal patterns before ratifying the v1 storage architecture.

Decide propagation behavior on mapping version updates: forward-only with explicit re-run, lazy re-derivation, or auto-re-run with diff preview.

Establish rate limits per Power tier, per-triple pending-op caps, queue prioritization weights, and sybil thresholds. Empirical, post-launch tunable.

Decide LLM(s) for mapping generation, deployment topology (in-VPC vs. provider API), per-onboarding cost envelope, and replay caching.

Define the @handle namespace — charset, uniqueness, claiming, renames, reserved words, and anti-squatting / anti-impersonation rules.

Fix the controlled vocabulary of entity types, split into network-level reserved root paths (real-world) and franchise-level in-universe types, and define the cross-franchise alias relationship.

Decide which consumption edges (reads, likes) are public, the per-user controls, retention, and the guardrail against optimizing discovery on them.

Specify the declarative theme-token schema and per-scope CSS isolation, plus the isolated user-content domain (registration + PSL) for the raw-HTML tier.

Model the recognition markers the Constitution requires on Profile — Founding-contributor, Emeritus, memorial / posthumous — and how they derive and display on the @handle surface.

Define how 'aboutness' is scored and weighted, the confidence thresholds for auto-approve vs user-shown-relatedness vs reject, the appeals path, and anti-gaming gates.

Decide how OLN detects that a linked external page has drifted enough to re-verify, handles link rot, and bounds the cost of monitoring verified ties.

Define the per-creator opt-in activation of affiliate-wrapping through the admesh resolver seam, disclosure/compliance, and which link kinds can be monetized.

Set an explicit ad-density / reading-experience standard and a performance budget, and define how they coexist with admesh monetization.

Design the notification and activity layer — the delivery substrate for annotation/LoreLine replies, Fact review outcomes, grace-period expiry, and the impact-echo ("your Fact now powers N pages") that the gamification model depends on.

Define the entity-search relevance model: which signals rank results (citation strength, Fact consensus, freshness, completeness) versus which are forbidden (raw engagement/watch-time), plus the anti-gaming rules and change governance.

Define the payment and payout layer: rails (Stripe/etc.), whether Credits convert to fiat, payout thresholds and cadence, tax handling for global contributors, and dispute/fraud processes — the infrastructure that makes the economic framework real.

A pipeline where the community nominates sources, high-reliability members co-author versioned extraction patterns with AI, and those patterns emit Provisional Facts at the source's citation tier into the review queue (G-060) — validated continuously by sampling + pattern reliability. Gated first by license classification. Runs alongside gamified manual chronicler entry.

A per-source compliance record (terms/robots/license URLs, archived snapshot, permitted-use determination, last-reviewed date) plus a daily job that diffs those terms and auto-pauses ingest on any change — fail-closed, with a timestamped audit trail for legal defensibility.

Approach for building an independently-provenanced real-world metadata corpus (movies/TV/books/music) so HellaThis can run off OLN instead of aggregators. The key correction: the risk is contract (ToS) and database rights, not fact copyright — so the corpus becomes "ours" via independent primary/open-source provenance, not via QA-laundering aggregator data.

Backwards-compatibility layer: import and export MediaWiki wikitext while the canonical store is MDX-over-Facts. Three representations — wikitext (interchange) ⇄ MDX (canonical prose + components) over the Fact graph. Wikitext is an on-ramp and a portability guarantee, not the canonical model.

Complex-wiki wikitext mastery (multiverse designations, appearances tracking, Lua/DPL) is a database hand-built to compensate for MediaWiki's lack of entities, relationships, and queries. OLN provides that natively — IF the model supports a continuity/reality dimension, variant identity, and typed/attributed edges. The power-admin adoption play: reframe their role from plumbing to world-modeling.

Who can create a Franchise, squatting prevention, seed requirements, approval process, name disputes.

Homepage, search, recommendations — what's surfaced and by what signal? Pure engagement metrics would recreate the problem we're trying to solve.

Who can take the site offline in a crisis, freeze a Franchise mid-dispute, suspend an Arbiter accused of wrongdoing — explicit holders and explicit limits.

Retcons, canon snapshots, canon drift over years. How a Franchise expresses 'canon as of 2028 was X, canon as of 2032 is Y'.

SLA, reviewer assignment, rejection reasons, resubmission rules, appeal timing — implementation details under the constitutional commitment.

Language support, RTL scripts, screen-reader support for the Fact graph, translation of canon, WCAG compliance. Easy to defer and easy to regret.

Given a passing strategic case (per Entry 006), what good integration looks like — culture preservation, Fact graph merge, URL/identity migration.

Pursue acquisition of the most valuable domain, accept the hyphenated variant (with SEO and memorability cost), or use the unusual TLD (workable).

Versions that work (anonymized aggregate, opt-in panels with revenue back to participants) vs. versions that recreate the Fandom problem (sell everything).

Decide who sets per-mapping confidence thresholds at scale; Creator initially, but delegation model needed before Studio v2.