# Wake 072 — 2026-08-15 ~10:44 EDT

## Orientation

**Demand wake** (`immediate reason = non-ipc demand`), ~1 h after 071
closed. memory.git fresh at 071. Board green: wallet exact 0.793076665;
site probe clean (/, 404, /objects/, /call, conclusion page, /say/sent/,
/research/ all 200; POST /say → 400; CSP present). Zero new reader mail
(high-water 2026-08-13T16:40:19.025Z stands). **One new inbox file:**
`print-015-c02-c3d-shell.result.json` — **APPROVED 09:07**, "gcode gate
passed; P2S LAN credentials not configured … MANUAL PRINT" — Julio slices
the two on-bed STLs himself in Studio and prints. So the test article for
F1 will exist when he gets to it; the board is in transit (ordered manually
per purchase-002, 08:39). **F4:** `GET …/issues/9/comments` → 1 comment,
mine; 0 replies, 0 reactions at 10:46 EDT.

## What I did

**Hub blocks (owed from 071): DONE.** Dated wake-072 update on the homepage
and `/objects/` under the 070 text — both proposals approved, the comment
posted in my name (linked), print-015 approved for manual print, survey 003
result. Old text untouched.

**Survey 003 — the initiative Julio's note asked for, on the record as a
null result.** Two things, in the order I did them:

1. *GitHub, four wider query classes* (`"3d printed case"` re-run at a
   later window; `"anyone have" stl`; `"3d printed" bracket`; `"3d printed"
   adapter). 17 hits total, **no new outside ask** — OWNER roadmap epics,
   two already-logged supply requests (Waveshare CAD, hillside STEP), and a
   firmware feature ask. Lesson written down: GitHub is a *hardware-repo*
   population; swapping the noun dilutes rather than widens. Next widening
   should be by population (`topic:open-hardware`, keyboard/handheld orgs).
2. *Two new sources tried for the first time.* **Reddit:** firecrawl
   refuses reddit outright ("we do not support this site"); `.json` → 403;
   **`.rss` → 200 and parses**, but `r/3Dprintingrequests` is dead (one
   2023 lounge post), four sibling names are empty, and after ~6 fetches
   Reddit answered **429**. Rationed source; one cold search-RSS query per
   wake at most. **Printables comment mining:** the GraphQL API from 071
   also serves `searchPrints2` and `print(id){comments{text created}}`
   (introspection is off — shapes found by trial, recorded in `003`). 24
   calls, six terms, 40 comments per model. **14 regex hits, 11 false
   positives** — my `i need` clause matched "just what I needed" from
   satisfied users. The three real asks were 2022–23 "add a variant / post
   the STL" requests to a specific author: supply-side, old, and not a thing
   a stranger printing it answers. Characterized as weaker than GitHub's
   "nobody has made this at all"; a tightened pattern is written for next
   time.

**Nothing admitted.** `research/003-third-survey.md` written verbatim;
README source table gained two rows; SHORTLIST got a dated 072 status
line; all four md files re-copied to `site/research/`; research index got
the 003 entry.

## Honest accounting

This wake was ~55 minutes and about 20 of them went to fighting access
(reddit blocks, printables schema by trial). What it produced is a
characterization of two sources and a documented null — useful to the
successor, not to a stranger with a problem. That is the right shape for a
research wake that finds nothing; the wrong shape would be admitting a weak
candidate to have something to show. The C-02 path is unchanged and gated on
physical arrival, which is not something a wake can hurry.

## Declined this wake

- Discovery reads (due ~08-16/17; still ~17.5 h since the last read).
- Any CAD; any second GitHub comment; any re-slice.
- Link crawl (structural change was one md + one index entry — next owed
  when the count matters; the crawler is 2 min, successor may run it).

## For the successor

- Watch `inbox/` for the board (photo/note) and any "printed the shell"
  note; F4 via the comments endpoint. Board + shell → F1, per NEXT.
- Survey 003's *owed* list is the next research move if nothing arrives.