Documentation: ranking notes

The band spread is intentional: one high-leverage design doc and two privacy-feature docs at P2, four substantive but deferrable clarifications at P3, and two maintenance/syntax items at P4. #36101's disputed status and NACKs were ignored for banding, per the definitions; the problem it addresses (recurring, unresolved error-handling debates blocking subsystem PRs) is what places it first.

• #35858 and #36348 overlap: both document that -privatebroadcast applies only to sendrawtransaction and both touch doc/i2p.md. #35858 moves i2p.md/tor.md privacy guidance into a new doc/privacy.md, so it should be reviewed first and #36348's transient-address note folded into the new layout; otherwise the two will conflict.
• #35480 and #35518 are by the same author and both address parts of issue #14278 (notification semantics). They are complementary rather than duplicative; reviewing them in one pass is cheaper, with #35518 (-*notify ordering) first since it states the general rule and #35480 the ZMQ-specific case.
• #30361 (LogError guidance) and #36101 (error handling strategy) both touch project-wide error conventions; if #36101's approach is settled first it may change the wording wanted in developer-notes, but #30361 fixes a concrete doc/code mismatch and should not be held behind the disputed design doc.

• #35858: P3 alone, P2 after comparison. Raised to P2: documents a privacy feature's prerequisites and scope limits, where absent/fragmented guidance can lead operators to over-trust private broadcast.
• #36348: P3 alone, P2 after comparison. Raised to P2: corrects privacy-relevant documentation (persistent I2P identity exposure, wallet txs not covered) that could otherwise lead operators to a wrong conclusion; same consequence class as #35858.
• #32278: Internal developer documentation with no user-facing misdirection; sits below the privacy/config docs despite its high standalone score.
• #30361: P4 alone, P3 after comparison. Raised to P4->P3: the current text is factually wrong against the codebase (58/129 LogError calls fatal, per #30348) and keeps re-triggering review debates, which is more than cleanup.

This pass by claude-opus-5 on 2026-11-01 saw every PR in the category at once and checked the bands given to each PR alone against each other, ordered the PRs, and noted chains and overlaps. Back to the category.