Beat Phase F.2 win scorecard re-run: FAIL
Verdict: FAIL against Beat Phase F.2 (docs/plans/brickwork-beat-tailwind-plus.md
Phase F) and the win bar in programme definition of done item 1.
Brickwork does not honestly lead or tie Tailwind Plus on at least six of
eight VISUAL-BAR surfaces from stills this reviewer saw. S2/S4 package craft
at tip 7db96c6 closes the prior structural Lose reasons (invoice detail line
items; dashboard chart + activity split). Package-default craft still holds
the meet-bar substrate axes checked below. Brand torture remains clear by cite
to the prior F.2 review. The beat threshold is not met: explicit Plus
lead/tie count is 5/8.
Reviewer independence: Independent design reviewer for this F.2 re-run only. Did not author Beat Phases A to D craft, S2/S4 craft, appearance suite work, journey contracts, the F.1 harness, Plus still capture, or prior F.2 scorecards. Clean-room throughout: look and idea only; no kit markup, class strings or structure read into package files.
| Item | Value |
|---|---|
| Package under test | django-brickwork 3.25.0 |
| Branch / tip | worktree beat-f2-finish @ 7db96c6 |
| Render mode | Package-default CSS only for the win table |
| Package stills | docs/audits/_stills/2026-09-14-beat-f2-rerun/ (32 PNGs, S1 to S8 × light/dark × 1440/375; manifest version 3.25.0) |
| Plus stills | Umbrella docs/_examples/tailwind-plus-surface-stills/ (surface map README) |
| Brand torture | Cite prior F.2 review §3 (F.1 stills); no new ugliness on package-default stills; not re-run |
| Normative bar | VISUAL-BAR.md |
| Prior meet-bar PASS | 2026-09-12-visual-bar-signoff-pass.md |
| Prior F.2 FAIL | 2026-09-14-beat-win-scorecard.md (3/8) |
| Soft-pass | Forbidden; every Lead/Tie/Lose/N/S cites stills actually read |
1. F.2 done checks
| # | Criterion | Verdict |
|---|---|---|
| 1 | Independent review fills S1 to S8 (light/dark, 1440/375) with an explicit lead kit column | Met (table below) |
| 2 | Brickwork leads or ties Tailwind Plus on ≥6/8 surfaces on package-default CSS | Not met (explicit count: 5/8 lead or tie vs Plus stills seen; see §4) |
| 3 | Does not lose Navigation, Typography, Forms, Density, or Spacing rhythm for substrate reasons | Met (no blocking substrate fail on those five axes) |
| 4 | Brand torture: no substrate-owned ugliness across three packs × themes × densities on S1/S3/S5 | Met (cite §3; not re-run) |
F.2 PASS requires all four. Criterion 2 fails, so overall FAIL. Do not run F.4 claim language from this audit.
2. Plus evidence limits (honesty)
Local Plus / Idea refs actually opened with the Read tool for this review:
| Source | Files read | Surfaces they inform |
|---|---|---|
| Tailwind Plus Salient | s1-salient-payroll.png, s4-salient-reporting.png, s5-salient-home-1440.png, s8-salient-contacts.png |
S1 list; S4 dashboard; S5 landing; S8 dense-list adjacent |
| Tailwind Plus Radiant | s2-radiant-profile.png |
S2 detail / lead profile (not invoice line items as Plus source) |
| Tailwind Plus marketing-v4 | s3-marketing-contact-centered-1440.png, s5-marketing-landing-01-1440.png, s6-marketing-pricing-03-1440.png |
S3 form chrome (partial); S5 landing; S6 pricing |
| Tailwind Plus Syntax | s7-syntax-docs-1440.png |
S7 docs article / docs home chrome |
| Forge UI kit (secondary Idea-ref only) | 07-dashboard.png, 08-forms.png, 09-settings.png |
Clarifies finished depth for S4 / S3 / S2; not a Plus lead/tie |
S3 limit (mandatory): s3-marketing-contact-centered-1440.png is Plus
marketing contact form chrome, not an Application UI create-form page
matching examples/app/form.html. Scoring Lead/Tie against it would invent a
match. Prefer N/S (applied below). Forms axis still passes on package
stills alone.
S2 note: Radiant profile.png remains lead/account profile chrome (avatar,
tabs, definition list), not invoice line-item detail. Package fixture now
shows true invoice detail (meta card + line items + danger zone). Comparison
is for the app-detail job broadly against that Plus detail craft.
Forge never counts toward criterion 2.
Package stills actually read (minimum set): light 1440 for S1 to S8; light 375
for S1, S3, S5, S6, S7; dark 1440 for S4 and S8. All under
docs/audits/_stills/2026-09-14-beat-f2-rerun/.
3. Brand torture summary (cite; not re-run)
No new substrate-owned ugliness appears on the 2026-09-14 re-run package-default stills. Per the F.2 brief, brand torture may cite the prior independent F.2 pass of this criterion without re-opening F.1 stills.
Cite: 2026-09-14-beat-win-scorecard.md §3 (itself citing 2026-09-13-beat-win-scorecard.md §3; harbour / folio / northline × theme × density on S1/S3/S5; no substrate-owned ugliness). F.1 harness capture: 2026-09-13-beat-phase-f1-brand-torture.md.
| Check | Result |
|---|---|
| Composition holds under seven-token packs | Pass (prior F.2 §3) |
| Accent reads as brand | Pass (prior F.2 §3) |
| Dark theme | Pass (prior F.2 §3) |
| Density axis | Pass for ugliness (prior F.2 §3) |
| Substrate-owned ugliness | None found on prior sample; none new on package-default re-render |
Brand torture clears F.2 criterion 4. It does not repair the Plus lead/tie count.
4. Win scorecard (package-default)
Lead kit means the kit that sets the craft ceiling for that surface in this
review. Brickwork vs Plus is only Lead, Tie, Lose, or N/S. Forge
appears only in notes and never as a Plus lead/tie.
| Surface | Lead kit | Brickwork vs Plus | Axes notes | Substrate fail? |
|---|---|---|---|---|
| S1 App list | Salient (Plus) | Tie | Filter bar, page header, primary CTA, end-aligned amounts; phone hamburger + stacked label/value rows (s1-light-1440, s1-light-375). Matches Salient payroll for the list job; Plus is quieter/minimal, Brickwork is more complete list chrome. Neither clearly wins. |
No |
| S2 App detail | Radiant profile (Plus); Forge settings Idea-ref only | Tie | Breadcrumb, dual header actions, meta card, line items table (DESCRIPTION / QTY / UNIT / AMOUNT), danger zone (s2-light-1440). Prior Lose body (related-invoice list) is gone. Radiant s2-radiant-profile.png still leads avatar/tabs/gradient profile chrome; Brickwork leads invoice-detail grammar. Neither clearly wins the app-detail job. Forge 09-settings.png reinforces finished settings depth (Idea-ref only). |
No on Navigation / Typography / Forms / Density / Spacing |
| S3 App form | Forge forms (Idea-ref); Plus N/S | N/S | Labels, help, errors, multi-line memo, primary/secondary clear at 1440 and 375 (s3-light-1440, s3-light-375). Forms axis passes. Plus contact still is partial marketing form chrome only (s3-marketing-contact-centered-1440.png); match too weak for honest Lead/Tie/Lose vs app create-form. Forge 08-forms.png shows deeper create-form craft (Idea-ref only). |
No |
| S4 App dashboard | Salient reporting (Plus) | Tie | Weighted revenue tile + companions, Revenue by week chart, Activity split (volume bars + events table) (s4-light-1440, s4-dark-1440). Prior Lose (metrics + table alone) is closed. Salient s4-salient-reporting.png still leads chart micro-craft (axes, today marker, dual-colour line, tooltip). Brickwork's full composition is a finished dashboard argument. Neither clearly wins. Forge 07-dashboard.png reinforces sparkline/axis depth (Idea-ref only). |
No |
| S5 Marketing landing | Salient home / marketing-v4 (Plus) | Lose | Strong section cadence, serif/sans pairing, hero placeholder, features, stats, testimonial, CTA, footer columns (s5-light-1440, s5-light-375). Package-default is meet-bar finished. Plus s5-salient-home-1440.png and s5-marketing-landing-01-1440.png lead on hero media, product mockup depth, and section richness. Claiming Tie would soft-pass. |
No |
| S6 Marketing pricing | marketing-v4 pricing (Plus) | Tie | Highlighted middle tier, badge, per-tier CTA, FAQ, CTA band, footnote (s6-light-1440, s6-light-375). Core three-tier job holds next to s6-marketing-pricing-03-1440.png. Plus adds billing toggle, testimonials and denser FAQ chrome; Brickwork's selective "Most popular" highlight is as clear. Neither clearly wins the pricing composition. |
No |
| S7 Docs article | Syntax (Plus) | Lose | Sidebar, callout, code header, table, pager at 1440; Documentation menu at 375 (s7-light-1440, s7-light-375). Meet-bar article craft holds. Syntax s7-syntax-docs-1440.png leads with docs hero, search, on-this-page rail, dual callouts and richer code chrome. Claiming Tie would soft-pass. |
No |
| S8 Dense ops | Salient list surfaces (Plus) | Tie | Four metric tiles, filters, checkbox table, status column, end-aligned value (s8-light-1440, s8-dark-1440). Denser ops job than Salient s8-salient-contacts.png media rows; comparable finished dense-list craft for the ops table job. Counts as tie. |
No |
Explicit lead/tie count vs Tailwind Plus
Surfaces where Brickwork leads or ties Plus from Plus stills actually seen:
- S1 App list: Tie
- S2 App detail: Tie
- S4 App dashboard: Tie
- S6 Marketing pricing: Tie
- S8 Dense ops: Tie
Count: 5 / 8. Required: ≥ 6 / 8.
Surfaces scored Lose vs Plus stills: S5, S7 (2).
Surfaces N/S: S3 (1; partial Plus form chrome only).
Surfaces Lead: none.
5. Substrate axes (package-default)
Checked against F.2 re-run package stills and the five axes named in the win bar.
| Axis | Rating | Evidence |
|---|---|---|
| Navigation / mobile chrome | Pass | App hamburger on S1/S3 375; marketing hamburger alone when closed on S5/S6 375; docs menu discoverable on S7 375. |
| Typography | Pass | Display serif on app and marketing titles; sans UI chrome; money columns end-aligned. |
| Forms | Pass | S3 errors, help and multi-line textarea at both viewports. |
| Density | Pass | Credible filters and rows on S1/S8; marketing air intentional; S8 is dense-list. |
| Spacing rhythm | Pass | Consistent section and card cadence; no accidental voids that fail the axis. |
Residual taste (not substrate axis fails for this gate): S4 chart micro-craft still thinner than Salient; S5/S7 thinner than Plus template ceilings on media and docs chrome.
6. Gaps to reopen before a re-run
- S5 marketing landing depth: hero media and section richness need to hold next to Salient home / marketing-v4 stills (product chrome, not only placeholder card) before Lead/Tie is honest. This is the clearest remaining Lose.
- S7 docs chrome: Syntax sets a higher docs shell bar (search, on-this-page, richer code/callouts, docs hero). Raise package docs article/shell craft or accept Lose on this surface.
- S3 matched Plus create-form (optional for honesty): an Application UI
create-form still under
_examples/would let S3 leave N/S; marketing contact remains partial only. N/S currently blocks one of the eight slots needed for ≥6/8 even after S2/S4 flipped. - S4 residual (optional): Salient/Forge chart axis, today marker and sparkline KPI depth remain a craft ceiling; Tie holds without it, Lead would need that polish.
- Re-run this audit only after craft or fixture evidence changes the Lose/N/S rows; keep F.4 claim cite gated on a future PASS with ≥6/8.
Delta vs 2026-09-14 FAIL (3/8): S2 Lose→Tie (line items + danger zone replace list-shaped body); S4 Lose→Tie (chart + activity split close metrics-only composition). S1/S6/S8 remain Tie; S5/S7 remain Lose; S3 stays N/S. Count 3/8 → 5/8. Still short of ≥6/8.
7. Claim language
Do not update POSITIONING.md or README beat/prefer claims from this file. F.4 stays closed until an independent win scorecard returns PASS with ≥6/8 lead/tie vs Plus stills that were actually seen.
Meet-bar evidence in 2026-09-12-visual-bar-signoff-pass.md remains the last PASS for section 6 of VISUAL-BAR.md. This audit does not revoke that meet-bar PASS; it only fails the beat win threshold. VISUAL-BAR.md "latest" meet-bar cite left unchanged (this is beat win re-score, not meet-bar).