Decision
Atlas profiles include a linktree: a curated set of external links, rendered
as a safe-tier OLN widget on the indexed /@handle profile (Entry 027) —
structured and declarative, never the raw-HTML sandbox.
- Two kinds, one list (with a
kindflag): affinity / endorsement links (others' content the member likes) and own-content links (the creator's presence on other platforms). - Entity tie is optional but rewarded. A link may be tied to a franchise, canon entity, or Fact (v1 tie targets); a tie earns graph placement and entity-page surfacing. Untied links (e.g. "my Twitter") are allowed.
- Any tie must be proven about its target. Verification is AI-first relatedness scoring at publish; on low confidence the member is asked to show how it relates, then it is decided; humans handle edge cases and appeals (Entry 022 role split). Relatedness is weighted — a strength tier, reusing Entry 020 — not binary.
- Own-content links need aboutness proof only when tied to an entity or object; untied own-content links are safety-scanned only.
- Lifecycle: propose → verify → re-evaluate when the external target changes (content drift) → quarantine if the proof no longer holds. Quarantine hides a tie from public/entity surfacing but retains the record, reversibly.
- Verified, weighted endorsements surface publicly on entity pages — the "entity as reference hub" goal (visuals deferred). OLN learns about content it does not host, enriching the entity graph.
- SEO posture: external links render
rel="ugc nofollow"(sponsoredwhen monetized). OLN's value is the internal association data, not outbound link equity. - Monetization is admesh-ready, activation deferred. Links route through an OLN resolver/redirect that admesh's subID bridge can hook. The linktree ships unmonetized; affiliate-wrapping activates per creator, opt-in (contributor ad sovereignty, Entry 005). The seam exists from day one so nothing is retrofitted.
Reasoning
A linktree is usually a vanity list. Tying links to the graph and proving aboutness turns it into crowd-sourced, verified reference-gathering: it enriches what OLN knows about an entity and feeds AEO, because entity pages dense with verified references are exactly what answer engines reward (G-026).
It reuses machinery OLN already has rather than inventing new primitives: citation strength tiers (Entry 020) for weighting, the Fact lifecycle (Layer 10) for the verify→quarantine states, the AI/ML role split (Entry 022) for the proving, anti-flood / spam handling (Entry 024, G-017) for quarantine, and contributor ad sovereignty (Entry 005) + admesh for monetization.
And it fits the Entry 027 customization ladder cleanly: structured, declarative, safe, indexable — a first-class OLN widget on the apex profile, not user code on the isolated sandbox.