2026-05-16 (correction) — arc 171 was wrong; the work IS in arc 170
Forward-correcting the previous entry. Per feedback_inscription_immutable: don’t edit past entries; new entry names the errors.
Two errors in the previous entry, not one.
Error 1 — number-grab
Section titled “Error 1 — number-grab”The previous entry framed the work as “arc 171 (concurrency-bracket-combinator).” I picked 171 by incrementing 170 in my head. Disk had 171-comma-to-apostrophe-dispatch/ already. Stolen number. feedback_assertion_demands_evidence violation — should have grepped docs/arc/2026/05/ before claiming.
Error 2 — closure-trajectory framing (the worse one)
Section titled “Error 2 — closure-trajectory framing (the worse one)”My first correction draft framed the bracket as “arc 170 Slice 7” — numbered slice with BRIEF + EXPECTATIONS + SCORE + INSCRIPTION trajectory. As if we knew the shape. User correction: “what the fuck - are you closing 170 - we are not done with anything - we just realized a new mandatory pivot - that does close anything - it’s open a direction we go down - we are not done.”
The bracket combinator is an OPENING, not a slice.
- We just REALIZED a mandatory pivot
- Realizations OPEN directions; they don’t close arcs
- Arc 170 closes via INSCRIPTION, when the work is done — not when we figured out what’s next
- We don’t know how deep the bracket direction goes yet
- It might be one slice. It might be five. It might surface deeper things we don’t see yet
- Putting “Slice 7” on it implied tidy bounded scope; wrong
Settled framing:
- The bracket combinator IS arc 170 work — lives in
docs/arc/2026/05/170-program-entry-points/ - It’s not numbered as a slice yet — the shape isn’t known
- It’s the CURRENT DIRECTION inside arc 170, opened 2026-05-16
- Arc 170 stays OPEN. Possibly for many more sessions
- Closure when INSCRIPTION ships — not when we hit some pre-decided slice count
Task moves (corrected):
- #324 (original Slice 7 — drop variants only): stays superseded. The easy version IS wrong; the bracket-direction supersedes it. Not because of “Slice 7+” — because it was the easy answer to a question that has a harder, more honest answer.
- #325 (umbrella): refocused to “arc 170 — concurrency-bracket-combinator direction.” No slice number. Tracks the umbrella of work this direction implies.
- #326 (DESIGN): refocused to “arc 170 — bracket combinator DESIGN pass.” No slice number. Output location to be decided when the work starts (not pre-decided as “SLICE-7-DESIGN.md”).
New discipline saved: feedback_realizations_open_directions — a realized pivot OPENS a direction; it does not close anything. Closure is a separate act, via INSCRIPTION. Don’t number slices when the shape isn’t known. Don’t draft BRIEFs for explorations.
Design content unchanged (this is the third entry that affirms it — bracket combinator + walker collapse + *_join-result substrate-internal + -with-io family dies + cascade cleans orphans + we don’t sandbox). What changed in THIS entry: the FRAMING of the work as scoped-and-bounded vs open-direction. The bracket is open-direction. Arc 170 is open.