The chapter page now has a persistent course sidebar on the left (P3), an in-page TOC rail on the right that highlights the active section as you scroll (P4), and a row of chips above the first heading, before any body text. Each chip names a prerequisite article. The label is “Read first ↳”.
That row is what conventions.md called “the single highest-value navigability fix” in the entire seven-phase navigation spec.
The spec
docs/specs/learning-hub/navigation.md ran to seven phases. P0 through P6, all shipped. P6 was site-wide Pagefind search across the full property, on branch feat/learning-nav-p6-search. It landed 3 July. The spec is now marked complete.
The sequence:
- P0–P1: spec structure and foundational wiring
- P2: hub track sections and S1 JSON-LD structured data; first visible change to the hub
- P3: persistent course sidebar on every chapter page (“the biggest browsability lift”)
- P4: in-chapter TOC with IntersectionObserver scroll-spy; the spec’s one client island
- P5: prerequisite chips at the chapter head, branch
feat/learning-nav-p5-prereqs - P6: site-wide Pagefind search
P5 completed the last of B-037’s three ranked fixes. The N1 designation in the spec belongs to the prerequisite chips.
Why the other phases weren’t highest-value
The rest of the navigation system answers questions the reader already knows they have.
The sidebar answers: “Where am I in this course and what’s near me?” The reader knows they want to navigate between chapters. The sidebar is the affordance.
The TOC answers: “What’s in this chapter and where exactly is the section I need?” The reader knows they want to jump. The TOC is the map.
Search answers: “I read something about X, where was it?” The reader has a target.
The prerequisite chip answers a question the reader doesn’t always know to ask: “Should I be reading this yet?”
That’s an orientation question, not a navigation question. The spec was precise about the distinction.
Orientation before commitment
A reader who arrives at a chapter, encounters unfamiliar concepts in the first three paragraphs, and abandons. That’s not a browsability failure. No sidebar would have caught it. No TOC entry would have redirected them. They didn’t need to jump to a different section; they needed to start somewhere else entirely, and nothing told them that until after they’d already spent the time.
The chip changes the decision point. “Read first ↳” at the top of the chapter moves the orientation check from mid-read to pre-read. The reader sees the prerequisite list before they commit to the page. They can act on it or choose to proceed knowing the gap exists.
Both are better outcomes than discovering the gap after three paragraphs.
Browsability vs. navigability
The devlog called P3 “the biggest browsability lift.” The spec rated P5 highest-value for navigability. Those are different axes, and the spec was careful not to conflate them.
Browsability is how freely you can move once you’re already oriented. The persistent sidebar maximises it: every chapter is one click away regardless of where you are in the course.
Navigability is how well the system gets you oriented in the first place. The prerequisite chip maximises it at the chapter level: you know your starting conditions before you read a word.
A reader who can jump between any two chapters instantly (P3) but has no signal about which chapter to start from is well-served for browsability. They’re poorly served for navigability. The chip fills the navigability gap the sidebar doesn’t touch.
The accessibility argument
The word comes up eventually in any navigation discussion. Accessibility, as most audits define it, means: can a person with a disability use this interface? Alt text, ARIA roles, tab order, colour contrast ratios. That is a real and necessary question.
It is also narrower than the word suggests.
A page can achieve WCAG AA and be completely inaccessible to a reader who lacks the background knowledge it assumes. Not inaccessible in the compliance sense. Inaccessible in the original sense: not reachable, regardless of intent, because the reader lacks what the content requires. The vocabulary for this kind of inaccessibility is worse (“assumed knowledge,” “content prerequisites,” “pedagogical scaffolding”), because the standard tooling doesn’t measure it. You don’t get a Lighthouse score for it.
The chip measures it directly. Before this chapter, you need to have read these. That is an accessibility claim in the full sense of the word. Whether a reader has the prerequisite knowledge determines whether the content is accessible to them, independent of anything a screen reader does.
conventions.md called the chip the highest-value navigability fix because it solves the right problem. Not the most complex. Not the one audits score. The one that determines whether a reader can actually reach the content they’re looking for.
P4’s one client island
P4, the in-chapter TOC, is the spec’s one client island. IntersectionObserver scroll-spy, active section highlighting as you read. The full scratchpad mockup (scratchpad/learning-toc-mockup.html) carried a live scroll demo with the real intersection logic before a line of production code was written.
The chip has no equivalent complexity. P4 is the one client island; every other phase, including P5, is server-rendered. No IntersectionObserver. No hydration. A component that’s present before any JavaScript runs.
The spec’s highest-value component is also its simplest to reason about. That’s not a coincidence. Orientation information is most valuable when it’s most reliable, and static output is more reliable than a client island by construction. If the JavaScript bundle fails to load, the TOC is dark. The chip is still there.
Where the spec landed
Seven phases. All shipped. P6 closed the spec on 3 July.
The system the spec produced isn’t a navigation framework. It’s a layered set of answers to specific reader questions, ordered by how much each question blocks forward progress. P3 and P4 handle movement and location once a reader is inside a chapter. P6 handles search across the full property. P5 handles the question earlier than any of them: before you move, before you locate, before you search, are you starting from the right place?
The answer lives in a row of chips above the first paragraph. Three words and an arrow.
