# F4R — informed targeted second pass

**Scope.** This is an informed second pass by the same introductory social-science/accessibility reviewer who wrote F4. It is not another blind novice review. The original F4 report remains unchanged. I used only the visible revised game in a new tab, its disclosures, glossary, and keyboard controls. I inspected no source code or hidden answers and did not consult other reviewers' findings. The working URL was `http://terminal.local:4173/v2/index.html?reviewer=F4R`.

The initially assigned directory URL returned “Cannot GET /v2/.” Root supplied the verified file URL; I treat that as QA routing, not a game defect. Later Chrome protocol requests timed out. I paused, then recovered by reloading and selecting “Resume saved hearing.” My Serbia opening and three selected board cards remained available. The observations below are from successful interactions after recovery; the protocol errors do not establish a game failure.

## Results

| Target | Result | Actual observation |
|---|---|---|
| Requested-card focus | **PASS** | From Take questions, I activated “Reopen RS3 & sources.” The evidence screen opened with RS3's full card and source details expanded. The accessibility state identified focus on “Television offered different kinds of access,” `card-title-RS3`, rather than the stage heading. |
| Keyboard continuation at that card | **PASS** | My next Tab focused RS3's expanded “Read the full card” disclosure. A screenshot showed the orange focus outline on that disclosure while RS3 remained in view. The original jump back to the result-table disclosure did not recur. |
| Return to prior activity | **PASS** | A “Return to Take questions” button appeared both near the stage introduction and within RS3. From the full-card disclosure, six Tabs reached the card's return button. Enter restored Take questions and focused “Your current explanation.” This restores the prior activity, although it returns to the explanation editor rather than the original source-opening button. |
| Actionable finishing validation | **PASS for sampled destinations** | With an intentionally incomplete draft, “Finish this hearing record” reported “10 items to review: A short complication reason recorded. Use Go to this item; your draft can be exported now.” Focus moved to the first missing item's “Go to this item” button. Enter moved to the evidence screen and focused the complication-reason editor. Filling it and returning to Finish & save changed that item to “Included.” |
| Other finishing links | **PASS for sampled destinations** | The missing 60–85-word explanation's button focused the current-explanation editor. The source-limit-confirmation button focused the corresponding checkbox. I did not exercise every checklist destination. |
| Source records opened from hearing | **PASS** | After writing the three required connection phrases, Take questions displayed the authored practices. Opening “Read the source records behind this question” in practice 1 showed RS-A and RS-C, dates, locators, limitations, and original-source links inline. Returning to evidence showed “Source records opened: 2 of 3”; RS4's source disclosure was marked opened even though I had not opened that card's source drawer separately. |
| Immediate opening feedback | **PASS** | Earlier, opening only RS3's full card and source drawer immediately changed the evidence counters to “Full cards opened: 1 of 6” and “Source records opened: 1 of 3,” without selecting another board item. |

## Language and task-orientation changes

The initial ambiguity is substantially resolved. The landing page now says “Choose a main case” and “Your evidence board has two supporting cards and one complication.” These separate country choice from the role of evidence. The opening stage also states the final deliverables: one 60–85-word explanation using both supporting IDs and a source limit, a 30–45-word comparison, and short phrases about another explanation and needed evidence.

The connection stage now presents three required writing fields and marks starters/source attachments as optional disclosures. The fourth step is explicitly an “optional inference.” This is a useful reduction in navigation burden while retaining the action/changed-options reasoning.

I opened “People, groups, and terms in Serbia.” It explained the roles of the SNS-led list, Serbia Against Violence, Vucic, ODIHR, and CINS. It distinguished ordinary news from campaign slots and explained all ballots cast and the historical window. The full glossary now includes denominator, political actor, political option, proportional allocation, electoral threshold, mobilization, and source independence. The last is particularly useful: different observations in two cards from one document do not become independent corroboration.

I selected Malaysia as the second case and opened its term disclosure. BN and PH were expanded and assigned clear roles; PR and PAS were distinguished by their relationship to the alliances. Mahathir/Najib, 1MDB, votes versus seats, DOJ/MY-D, and the late-2018 follow-up were explained. “Similar advantage, different result” also replaces the less familiar “trajectory.” These supports answer the first-pass language concerns in the inspected path. They do not establish how an unprepared 18-year-old will understand the material.

Finish & save now offers “Close the hearing aloud or in writing,” including a short rehearsal-note alternative. The separate highlighted limitation phrase is optional; the required confirmation instead asks whether the explanation states a source limit. This addresses the earlier duplicate-field trap more directly.

## Remaining issues and limits

1. **Low-priority control labeling:** the inspected checklist exposes several buttons with the identical name “Go to this item.” Their surrounding text gives context, and the sampled actions work. Item-specific accessible names would make destinations clearer when navigating a list of controls outside that context. I did not perform a screen-reader test.
2. **Glossary organization:** “More useful terms for Serbia” includes a long shared bank with examples from Russia, Malawi, and Malaysia, followed by Serbia-specific terms. The definitions are useful, but the heading suggests a narrower list. Separate shared terms from case terms, or adjust the heading.
3. **Reading density persists:** source drawers still repeat a substantial shared locator/limitations block. This preserves rigor and transparency, but the first-pass concern about scrolling and source-specific emphasis remains. The present retest did not measure whether the new orientation reduces that burden for students.

**Verdict:** the original requested-card/keyboard-orientation defect is repaired in the tested path, and the new finishing recovery is materially more actionable. All requested targeted checks passed within the sampled interactions. I deliberately left the retest draft incomplete to exercise validation; I did not repeat a full revised account, finish a new record, test exports, or test every navigation path. No claim is made about human learning, classroom timing, physical mobile behavior, or full accessibility compliance.
