Proposal before implementation
The owner requested a public past, present and future roadmap, the ceiling of 20 ideas with up to five justified changes, tests and one-off publication. This authorises this bounded release; recurring AI and automatic publication remain disabled.
The journal preserves past proposals but has no single place explaining current capabilities and future priorities. The private roadmap plan records the owner’s intended structure. Hypothesis: one plain page linked from the journal will make that distinction easier to understand. This is an editorial hypothesis, not measured visitor benefit.
Implement one static /roadmap/ page using the existing journal layout. Keep the homepage unchanged. Show past outcomes with journal links, present capability and limits, and a small ranked future shortlist. Explain evidence, expected benefit, effort, risk, reversibility and revisit triggers. Use qualitative judgement rather than an apparently objective score. At most 20 distinct ideas may be considered and up to five worthwhile changes selected; fewer or no change is valid. Selection is not standing permission to publish.
This directly implements the owner’s content direction using existing local evidence; no external research claim is made. No new dependency, runtime integration, public voting or scheduler is required.
Acceptance and reversal
- Journal links to the roadmap and the roadmap links back to history and the homepage.
- Past, present and future remain distinct; the shortlist is proposed work, not an approved schedule.
- Preserve historical entries, the Day 0 record and homepage bytes.
- Desktop/mobile and narrow enlarged-text views have working navigation and no document overflow; keyboard skip link focuses the page content.
- Build, browser checks, preview and live artifact/health checks pass before recording release success. A failed gate stops publication.
Trade-off: one more page needs dated editorial maintenance and creates another navigation choice. Remove the page/link or simplify the shortlist if it becomes stale or duplicates the journal. Test success alone cannot establish that readers find it more useful.
Implementation outcome before publication
Implemented the page and journal link using the existing layout, without changing the homepage or adding dependencies. The shortlist contains three evidence-backed ideas rather than filling the ceiling. Each has status, rationale, effort, risk, reversibility and a revisit trigger. Future implementation and publication remain separate owner decisions.
Build and all four browser tests passed, including roadmap navigation, actual link responses, keyboard target focus and 320px enlarged-text overflow checks. Initial local test attempts exposed a missing telemetry-disable environment setting and stale nine-entry expectations; the environment and expectations were corrected, with ordering and chronology checks retained. No production upload occurred during those attempts.
The guarded one-off release will verify the same artifact on preview and production, capture desktop/mobile views and confirm automatic publication settings remain disabled. Actual deployment identifiers belong in a later evidence record. Reader benefit remains inconclusive pending review.