Status: Emerging — needs crystallization

Thesis: Just as GitHub's killer feature wasn't git hosting but pull requests (the social layer around code), Seed's killer feature could be commenting — the social layer around content. Not just comments-on-documents, but comments as a universal protocol for discussion, annotation, review, and collaboration across the entire web.

Evidence cluster:

What I want for the commenting experience — 5 distinct UX pain points, all pointing at comments as primary workflow

Notes about how the document machine should work — the editor lifecycle that comments depend on

Remodeling the web — the broader thesis on web remodeling; comments as decentralized discourse layer

Discussions Panel Is Confusing — UX critique showing the feature is already being heavily used

External inspirations: Slite, Coda, Linear, Google Docs — systematic research into existing best-in-class patterns

Why it compounds:

Comments create network effects: more comments → more content → more users

Comments generate structured data (threads, citations, reactions) that make the network more valuable

Comments enable entirely new product categories: document review, academic peer review, code review, legal annotations, community discussion

Comments on any URL (not just Seed docs) opens the entire web as Seed's addressable market

What it unlocks:

Full-window discussions mode

Block-level comment IDs and ranges

Comment-on-any-URL capability

Comment-based collaboration workflows (review, approval, annotation)

Comment as a first-class content type in the HM protocol

Status: Defined — see Notes about how the document machine should work

Thesis: The document is not a file — it's a state machine with a lifecycle. Understanding this lifecycle is prerequisite to building perfect editing, commenting, and collaboration.

Why it compounds: Everything (comments, embeds, capabilities, citations) depends on the document state model.

---