Workshop: epistack-framework-additions

Brainstorm of framework additions to Commonplace that would make epistemic casework — tasks like the FLF Epistemic Case Study Competition's lab-leak / black-hole / egg cases — easier, plus a recommended personal-epistemology stance for the person running such casework.

This is the design-thinking companion to epistack-competition, which is only the framework-side pointer to the sibling epistack-casebooks repo and the backlog-to-commonplace.md protocol. This workshop holds the menu of candidate additions; the sibling repo is where any of them get built and proven.

Each candidate now lives in its own document, linked below. No framework code is written from this workshop. The discipline is build-local-first in the casebook repo, upstream what survives — this is a design menu, not an implementation plan. What closes the workshop: each candidate has either become a queued experiment in epistack-casebooks and logged to its backlog, or been rejected with a reason; anything that survives a worked case gets promoted as a proposal/type/note, then this workshop is deleted.

Source caveat. The competition pages (flf.org, EA Forum, GreaterWrong, Oliver Sourbut's "A Full Epistemic Stack") were unfetchable from the authoring environment (403 via network policy). The framing is assembled from search snippets, the sibling-repo summary in epistack-competition, and a pasted ChatGPT second opinion; verify against the primary sources before relying on specifics.

Framework-side results (2026-07-08 review session)

A framework-side review of this menu produced library artifacts that change the ground under the Foundation section — read these before the next brainstorming round:

• A universal knowledge framework demotes content taxonomies to defaults and keeps answerability — the theory: register-style taxonomies are guarded defaults, not universals; what stays universal is the declared collection contract plus answerability. Consequence for this workshop: the "fourth register" needs no taxonomy amendment — the casebook just writes its COLLECTION.md contract (see the updated fourth-register).
• ADR 042: register becomes a default profile under open-ended text contracts — the decision instantiating that claim for registers, adopted on the strength of this casework's first worked case.
• A predicted follow-up gap, now also a proposal: the note type's status field fuses lifecycle with first-person endorsement — see assertion force separate from lifecycle status and the contradiction flag added to claim-type.
• The review also flagged: the second opinion file is truncated (see its header note), and the "contradictions get silently averaged" mechanism several candidates lean on is untested — added as a third experiment in suggested-first-experiments.

Foundation

• The core tension to design around — why casework needs a stance-neutral evidence map, against Commonplace's first-person-committed grain
• The biggest single addition: a fourth register — a dialectical/evidential register; the methodological move everything else hangs off. Updated 2026-07-08: demoted from taxonomy addition to collection contract — see the file.

Ingestion / provenance layer

• Source-span citations, not file-level links
• A richer source type
• Traversable provenance chains

Structure layer

• A claim type distinct from structured-claim
• A dialectical/evidential link vocabulary
• Party/position attribution
• A first-class gap register

Assessment layer

• Confidence must be attributed, never a mark
• Epistemic review gates
• Adjudication as a separate, labeled, downstream layer

Additional candidates (imported from the second opinion)

• Independence clustering of evidence
• A crux type, distinct from the gap register
• Model / calculation artifacts
• Multi-method assessment comparison
• Derived dashboards and a source-impact command
• The belief-ledger practice

Cross-cutting

• Recommended personal-epistemology stance
• Suggested first experiments
• Rejected candidates (with reasons)
• Open questions

Source material

• ChatGPT second opinion — the pasted independent analysis these candidates were screened against

---

Complete file listing (generated at build time)