Retribution Day

01Where you're starting

Cycle 1 closed tonight. The branch is done, judged, and reverted — it just hasn't left your machine.

Unpushed commits on 189

21/21

Layout specs

Decisions blocking cycle 2

Open beads

Docking is dead — gate-4 verdict was baseline in 5 scenarios, head in 0. Reverted as 4 new commits; history intact.
The review workflow exists and works — doc, transition specs, scenario player, runner, churn diff, side-by-side videos. This is the real deliverable of the last two days.
The strict crossing-budget spec now passes — any future fan design must keep it green. It's a tripwire, not decoration.
Nothing is pushed — 13 commits sit local. Shipping is a morning-sized task, not a day-sized one.

02Track 1 — ship feature/189

The 'at least' bar for today. Everything here is mechanical; no decisions hiding inside.

Run the full mobile test suite once — confirm the only reds are the 3 pre-existing use-unlock-candidate specs (ctn.11's, not yours). ~10 min
S21 pinch test for ctn.7 — the screen-px fan-leg fix rides this branch and was never pinch-tested on device. 10 min, before the MR, so the MR ships it verified.
Push the branch. 2 min
Open the MR — frame the reverts as the finding, not churn: "built a transition-level review workflow, judged the docking design against it, evidence said revert." Link the churn diff numbers. ~30 min
Update pracino-qdh.22 (iOS checks for 189) — docking is gone, so the note shrinks: verify zoom transitions + fan behavior only. ~5 min
Delete branch map-review-baseline-no-docking once the MR is up — it's encoded in the revert commits now. 1 min
Sweep the untracked map-*.md notes — keep as local scratch or delete; they don't go in the MR. ~5 min

Estimate

Sum ~55 min. With ADHD tax and one "wait, what does this test do" detour: block 90 min, done before lunch is realistic.

03Track 2 — design cycle 2 (pracino-p40)

One hypothesis, run through the workflow, verdict by end of day. But it's gated by two decisions — fire those off FIRST, async, so the answers land while you ship.

Decision 1 — pracino-q3q

Universal ~500m position fuzzing (from yesterday's call). If ALL candidates get offset, exact-coincidence pin stacks mostly disappear — which shrinks the very problem the fan design exists for. Ping backend/PM at 9:00, before writing any layout code.

Decision 2 — pracino-rvd

What selection does when the selected pin clusters away (card stays up today; you lean "close it"). Smaller call — 15 min of your own thought, or fold it into the same ping.

The cycle itself, once unblocked:

Fix the matrix first (pracino-mr5) — denser scene for scenario 01, better probe for 10. Cheap, and cycle 2's verdict is only as good as its scenarios. ~45 min
Pick ONE hypothesis — continuous fan convergence (yours) or hide/fade-on-collision (the Mapbox/Google norm). Gate 5: one per cycle, no stacking. ~15 min
Implement minimally — timebox 2.5 h. If it's not runnable by then, shrink the hypothesis, don't extend the box.
Run the matrix both sides + diff — mechanical now: two runs ~7 min each, diff + videos ~10 min. Baseline = current HEAD (post-revert), so no revert gymnastics this time.
Gate-4 verdict — your yap-per-video flow worked; reuse it. ~30 min

Reframe

A gate-4 FAIL still closes the cycle with learnings recorded — that's the workflow doing its job, not a wasted afternoon. Yesterday proved it.

04The day, sequenced

Anchor blocks with buffers. If a block overruns its box, cut scope, not the next block.

09:00
Fire both decision pings (q3q + rvd)
Async first — answers arrive while you ship. 15 min, no rabbit holes.
09:15
Ship block: feature/189
Tests → push → MR → qdh.22 → branch cleanup. Box: 90 min.
10:45
Matrix fixes (mr5)
Scenario 01 scene + scenario 10 probe. Verify with one runner pass. Box: 60 min.
11:45
Checkpoint: is q3q answered?
Yes → pick the cycle-2 hypothesis over lunch. No → afternoon flips to the fallback block below.
12:00
Lunch. Actually stop.
The afternoon needs the focus more than the morning did.
12:45
Cycle 2 implementation
One hypothesis, timeboxed 2.5 h. Shrink scope before extending time.
15:15
Runs + diff
Both sides, numbered videos, churn report. ~30 min, mostly waiting.
15:45
Gate-4 verdict
Watch, yap, transcribe, verdict. ~45 min.
16:30
Wrap
bd notes, commit (revert if FAIL — that's fine), handoff note for tomorrow. ~30 min.

Fallback block (if q3q stays unanswered)

Swap the 12:45 implementation block for quick wins: pracino-ayn (P1 — unvalidated lat/lng can 500 every map query; input validation, backend), pracino-4z0 (remove the temporary left-menu variant switches), pracino-2fb (soft-filter zero-match silently returns unfiltered results). Any two of these is still a high-output afternoon.

05Deliberately NOT today

Seen, named, parked. Not forgotten — just not competing for today's focus.

The verify batch (ctn.1/2/3/11) — fixes landed, all waiting on device/team verification, none on you today. ctn.11's iOS Fabric upgrade is confirmed working since Friday (8e7209f) — the bead notes were stale, now corrected. What remains is the team device soak.
Parked workstreams (9yw UC-5.1 wiring, znl payments redesign) — real work left, own branches, not today.
iOS verification batch (qdh.20/21/22) — macOS teammates' territory; your notes are filed.
Refactor epics (9al.*) — architecture work; needs a calm day, not a retribution day.
The Fabric-cleanup and visual-polish GitLab issues from your notes file — they wait for the ctn.11 device soak anyway.