Record both redesign lanes, and remove three confusions

.okf/log.md

@@ -51,6 +51,45 @@ make it green: restructure same-day entries under one heading, and add
 `timestamp` to the 23 concepts missing it (anchored to each file's last commit
 time, which is verifiable - never invented).
 
+## 2026-08-21 - two lanes on the redesign, and three confusions removed
+
+Paul opened a from-scratch design exploration ALONGSIDE the extraction, not
+instead of it. Recorded everywhere a cold session might read one as the other:
+`docs/projects/2608-site-design-system/README.md`, the rollout plan, and
+[workflows/site-redesign-rollout.md](workflows/site-redesign-rollout.md).
+
+**Extraction governs; discovery has no authority.** ADR-0003 states plainly that
+this is "extraction and propagation, not a redesign" - the system comes from the
+course page and `/services/vibe-code-rescue/`. The discovery lane produces
+candidate designs and never a PR; it wins only via a superseding ADR carrying
+evidence, because 2.1/2.2 already shipped the extracted system to the blog,
+which is 77% of the site's Google traffic. Brand invariants (logo, ruby, display
+face, covers) are fixed even in discovery - "from scratch" means layout,
+structure and components, not a rebrand.
+
+The discovery brief deliberately never mentions the course page. Anchoring it
+there would reproduce extraction wearing a different hat and prove nothing.
+First artifact: Stitch project `4487418121858436329`, which produced a system it
+called "Artifact Editorial" and, unprompted, rejected the logo wall, the
+testimonial carousel and the repeated "Book a call".
+
+Three confusions removed while writing this up:
+
+1. **"Held until the read is in" hid a date.** The 2.4 gate needs a read due
+   approximately 2026-09-17 - 28 days after the `e1fa5409d` DEPLOY, which was
+   never confirmed. It reads as imminent and is not. Anything depending on 2.4
+   is parked until mid-September, and the first action on picking it up is
+   confirming that live date.
+2. **Phase 2.5 has no spec.** 2.1 and 2.2 each got one; 2.5 is nine words in a
+   table cell. Its first deliverable is `20.05`, not code - and the binding
+   constraint is canon-sensitive copy across 12 pages, not the code-line cap.
+   Only 2 of 12 service pages carry screenshot baselines.
+3. **The sequencing decision was unrecorded.** Next unit is 2.3, not 2.5,
+   decided by a three-lens panel that SPLIT - the ICP lens ranked service pages
+   first, the risk lens won on facts the ICP lens lacked (no spec, 2/12
+   baselines, and 2.3 rehearses the gate chain that failed in #540 at near-zero
+   blast radius).
+
 ## 2026-08-21 - what 15 review findings on one document taught, and one rule reversed
 
 Closing the OKF gap on PRs #541/#543: the delivery prompt shipped and was then

.okf/workflows/index.md

@@ -7,7 +7,7 @@
 * [Blog Post Pipeline](blog-pipeline.md) - mandatory end-to-end workflow for writing/publishing blog posts, plus the execute-don't-read claims gate, the frontmatter claims check, and the `## Sources` citation convention
 * [LinkedIn Post Pipeline](linkedin-post-pipeline.md) - Paul Keen voice rules and posting workflow
 * [CSS Maintainability Redesign](css-maintainability-plan.md) - approved plan for hand-editable CSS + FL-Builder retirement
-* [Rescue Room rollout sequence](site-redesign-rollout.md) - why phases are ordered by whether layout moves, why blog runs before site-wide chrome, why phase status must be checked against git rather than the plan table, and the gates that block regardless
+* [Rescue Room rollout sequence](site-redesign-rollout.md) - why phases are ordered by whether layout moves, why blog runs before site-wide chrome, why phase status must be checked against git rather than the plan table, and the gates that block regardless; plus the extraction-vs-discovery lane split and the 2.4 gate date
 * [Visual Scroll Gate](visual-scroll-gate.md) - blocking pre-handback visual walk for content/visual changes
 * Test suites, rake tasks, and the visual-regression gate live in [test-gates](/build/test-gates.md) (`workflows/testing.md` was squashed into it 2026-08-20 - it duplicated the gate rules and still carried the superseded "run both suites per commit" instruction)
 * [Analytics Access](analytics-access.md) - live GA4 + Search Console MCP servers, correct property IDs, credential split, data-lag traps, and why GA4 sessions must be reconciled against GSC clicks before being quoted

.okf/workflows/site-redesign-rollout.md

@@ -7,8 +7,8 @@ tags: [design, rollout, sequencing, decision, adr]
 status: stable
 generated:
   by: claude/opus-5
-  at: 2026-08-20T23:11:35Z
-timestamp: 2026-08-20T23:44:03Z
+  at: 2026-08-21T10:57:01Z
+timestamp: 2026-08-21T10:57:01Z
 ---
 
 # Why this concept exists
@@ -119,6 +119,38 @@ learning whether the design engages anyone before the whole site commits to it.
   moves PDFs too. Scope tripwire: `macos/{desktop,mobile}/course/chapter.png`
   must come out byte-identical.
 
+# Two lanes: extraction governs, discovery explores
+
+Since 2026-08-21 the redesign runs an **extraction** lane and a **discovery**
+lane at once, and conflating them is the live confusion risk.
+
+* **Extraction governs.** ADR-0003 is the decision of record and says plainly
+  that it is "extraction and propagation, not a redesign" - the system is taken
+  from the two pages already implementing it (course page,
+  `/services/vibe-code-rescue/`). 2.1/2.2 shipped it to the blog.
+* **Discovery has no authority.** A from-scratch exploration (Paul, 2026-08-21)
+  runs ALONGSIDE extraction, not instead of it. Its brief deliberately never
+  mentions the course page - anchoring it there reproduces extraction wearing a
+  different hat and proves nothing. Output is candidate designs, never a PR.
+* **Brand invariants are fixed even in discovery**: logo, ruby `#cc342d`,
+  display face, cover system. "From scratch" means layout, structure and
+  components.
+* **Switching lanes costs shipped work.** Discovery wins only via a superseding
+  ADR carrying evidence, because 2.1/2.2 already shipped the extracted system to
+  the blog - 77% of the site's Google traffic.
+
+A discovery lane does not consume WIP: it produces nothing mergeable, so it
+cannot collide with the delivery lane. That is continuous discovery, not a third
+parallelism exception.
+
+# The 2.4 gate is time-bound, and "held until the read is in" hides the date
+
+The homepage phase is gated on a blog engagement read due **28 days after the
+`e1fa5409d` deploy, approximately 2026-09-17** - and that clock starts at the
+first production DEPLOY, which was never confirmed. Writing the read up early
+does not unblock anything; it pre-writes a conclusion the data cannot support.
+Confirm the live date first, because the window is measured from it.
+
 # Citations
 
 [1] `docs/projects/2608-site-design-system/20-29-strategy/20.01-rollout-plan.md`

docs/projects/2608-site-design-system/20-29-strategy/20.01-rollout-plan.md

@@ -238,6 +238,38 @@ bot-filtered): blog pages **25.2% avg scroll depth, 26.3s avg engagement,
 - Watch Clarity recordings of ≥20 post sessions in week one after ship — the
   qualitative gate stays primary at this traffic.
 
+**The 2.4 gate is time-bound, and the date is NOT soon (recorded 2026-08-21).**
+"Held until the read is in" reads as imminent; it is not. Per
+[40.01](../40-49-measurement/40.01-blog-engagement-baseline.md) the read is due
+**28 days after the `e1fa5409d` deploy — ≈2026-09-17**, and that clock starts at
+the first production DEPLOY, which was never confirmed (deploys were managed by
+another session at merge time). So:
+
+* Writing the read up before the window closes does not unblock 2.4 — it
+  pre-writes a conclusion the data cannot support.
+* Anything that depends on 2.4 is parked until mid-September. Plan around it
+  rather than waiting on it.
+* **First action when picking this up:** confirm the live date of `e1fa5409d`,
+  because the whole 28-day window is measured from it and it is currently
+  unknown.
+
+**Sequencing decided 2026-08-21 (three-lens panel, recorded here because it
+deviates from "biggest ICP impact first").** Next unit is **2.3 course-single**,
+not 2.5. The panel split — the ICP lens ranked service pages first (they are the
+worst ICP betrayal per unit of traffic, at the highest purchase intent), the
+risk lens ranked 2.3 first and won on facts the ICP lens did not have:
+
+* **2.5 has no spec.** 2.1 and 2.2 each got one (20.03, 20.04); 2.5 exists as
+  nine words in a table cell. Its first deliverable is `20.05`, not code.
+* **Baseline coverage on service pages is 2 of 12** — ten money pages would ship
+  visually blind.
+* 2.3 is ~47 lines, baseline-covered, and rehearses the exact gate chain that
+  failed in #540 (merge-base run, CI baseline record, browser-computed seam/AA)
+  at near-zero blast radius, before that chain is trusted across 12 money pages.
+* The plan already sequences 2.3 "immediately after 2.2", and 2.2 shipped
+  2026-08-20 — so 2.3 is overdue, not a detour.
+
+
 ### 2.4 (homepage) carries an extra, blocking gate
 
 The homepage cut removes the twelve service tiles and the framework logo grid.

docs/projects/2608-site-design-system/README.md

@@ -40,6 +40,48 @@ mobile; and a lead conversion event exists so the next change can be measured.
   page, and `/services/vibe-code-rescue/` as a dark-variant campaign page.
   Phase 1a is unblocked — see [20.02](20-29-strategy/20.02-phase-1a-plan.md).
 
+## Two lanes — read this before changing any design artifact
+
+Since 2026-08-21 this project runs **two lanes at once**, and confusing them is
+the failure this section exists to prevent.
+
+| Lane | Authority | Produces | Status |
+|---|---|---|---|
+| **Extraction** (governing) | **ADR-0003** — the decision of record | merged CSS, shipped phases | active; 2.1/2.2 shipped, 2.3 next |
+| **Discovery** (exploration) | **none** — informs a future decision | candidate designs only, never a PR | opened 2026-08-21 (Paul) |
+
+**Extraction is still the decision.** ADR-0003 is explicit: *"this ADR is not
+proposing a new language… That makes this extraction and propagation, not a
+redesign."* The system is taken from the two pages that already implement it —
+the course page and `/services/vibe-code-rescue/`. Nothing below changes that.
+
+**Discovery answers a different question:** what would a designer produce for
+this ICP with no anchor at all? Paul asked for it 2026-08-21 *alongside*
+extraction, explicitly not instead of it. Its brief deliberately never mentions
+the course page — anchoring it there would reproduce extraction wearing a
+different hat and prove nothing.
+
+**Fixed even in discovery** (ADR-0003 brand invariants, not up for exploration):
+logo, ruby `#cc342d`, display face, cover system. "From scratch" means layout,
+structure and components — a rebrand was not asked for.
+
+**Discovery artifacts so far** — Stitch project `4487418121858436329`
+("JetThoughts — From-Scratch Exploration"), generated 2026-08-21. It produced a
+system it named *"Artifact Editorial"*: warm paper `#fbf9f8`, Source Serif 4
+headlines with Hanken Grotesk body, sharp 0px corners, ruby reserved for
+critical actions only, evidence-over-assertion structure. Unprompted, it
+rejected the logo wall, the testimonial carousel and the repeated
+"Book a call". A Claude Design project (*JetThoughts — Rescue Room*, 14
+previews) predates this and renders the EXTRACTED system — do not confuse the
+two.
+
+**What it would take for discovery to win:** a superseding ADR, with evidence
+that it beats the extracted system for the ICP. Until that exists, discovery
+output has **no authority** and must not be implemented. The cost of switching
+is not theoretical: 2.1 and 2.2 already shipped the extracted system to the
+blog, which carries 77% of the site's Google traffic.
+
+
 ## Current state
 
 **Re-sequenced 2026-08-20 (Paul): BLOG FIRST** — confirm engagement on the
@@ -79,6 +121,21 @@ estimated) the cheapest viable test needs 192 days.
 
 ## Outstanding
 
+**Phase 2.5 has no spec — write `20.05` before any 2.5 code.** 2.1 and 2.2 each
+got a written spec (`20.03`, `20.04`); 2.5 is nine words in the Phase 2 table:
+"stock photo → month-one card, decision table, 4-step timeline". Found
+2026-08-21 while sequencing. Two things the spec must settle, because they are
+not CSS decisions:
+
+* The real bulk of 2.5 is **per-page frontmatter copy ×12**, not stylesheet —
+  and that copy is canon-sensitive. `content/services/fractional-cto/index.md`
+  already carries "15+ years of industry experience" in its FAQs, the exact
+  figure class corrected on 2026-08-14. Copy review, not the ~500-line code cap,
+  is the binding constraint.
+* **Only 2 of 12 service pages have screenshot baselines** (`fractional_cto`,
+  `app_web_development`). Ten ship visually blind unless coverage is added
+  first — the plan's own 0.3 coverage rule applies.
+
 **Regenerate root `DESIGN.md` against ADR-0003.** It still frames dark
 JetVelocity as *"normative for new brand/conversion surfaces"*, calls LIGHT
 page chrome *"legacy/incumbent"*, and specifies `#1a8cff` primary buttons —