At this level
app.matakite · 6- app.matakite.audit record An append-only attestation that a policy-selected auditor observed an app.matakite.* record create, update, or delete, or intentionally voided an unrepresentable event. The auditor repository establishes observation chronology for clients that trust its DID; source-record createdAt and Jetstream time_us remain provenance rather than trusted time.
- app.matakite.dispute record A public counter-attestation against a resolution. With no bonds available (no money), the deterrent is reputational: disputes are signed records attached to the disputer's DID, so frivolous disputes cost the disputer. Disputes feed resolver track records ('N resolutions, M disputed, K upheld') that appviews surface when applying trust policies.
- app.matakite.flag record A public report that another app.matakite.* record violates community norms — spam, unresolvable criteria, private-individual target, harassment, duplicate, etc. Flags are signed records in the flagger's own PDS: public, timestamped, and attached to the flagger's DID. Frivolous flagging feeds the flagger's own quality record; the design mirrors app.matakite.dispute — the deterrent is reputational, not economic. Flags are inputs to appview indexing, display policy, and labelers; they are not cens
- app.matakite.forecast record A forecast on a binary, multi, numeric, or date question. Revisions are new records with the same subject uri; the full trail is retained and used for time-weighted scoring. Appviews select the payload matching the cid-pinned question type and reject records with conflicting typed payloads.
- app.matakite.question record A resolvable forecasting question. Immutability contract: editing this record after forecasts exist changes its CID; forecasts reference the question by strongRef (uri+cid), so appviews detect and flag post-forecast edits.
- app.matakite.resolution record An outcome determination for a question. Resolution is data, not authority: any DID can write one, and appviews choose which resolvers to trust and display (default: the question's designated resolver). Outcome semantics follow Metaculus: 'ambiguous' = reality could not be determined under the criteria; 'annulled' = the question itself was defective (underspecified, assumptions violated). Both are excluded from forecaster Brier scoring; 'annulled' counts against the question author's quality rec