# R8 — ordinary first-use student review

**Reviewer role:** A simulated ordinary 18-year-old community-college student: willing, impatient, uneven reader, no prior knowledge of these cases, no exceptional research skills. This is one constructed first-use perspective, not evidence about all Gen Z students or a measured student learning outcome.

**Scope:** Only the visible game at `http://terminal.local:4173/?reviewer=R8`. Serbia main case; Russia comparison. For class, Solo, no name. I did not read code, design notes, previous reviews, other agents' work, original external documents, or teacher materials. I did not edit the app. I used the visible people/terms help and glossary. I finished the record and closed my QA tab.

## My understanding from the first page, before entering

“I pick a country, choose evidence that explains why the group in power had an advantage, deal with one thing that makes my explanation less certain, then compare another country. I am supposed to explain something, not change who won.”

The two-supporting-cards/one-complication rule was understandable. I did not yet know what an actor, electoral list, or parliamentary majority meant, or whether a hearing would involve speaking to someone, clicking through a story, or writing an essay. “Decide what your evidence can carry” sounded serious but did not tell me an action to take. “The outcomes are known” and “your choices change the analysis, not history” usefully limited what I expected the game to do.

The initial screenshot showed a large title/image section and the playing-mode form. The country choices and Enter button were below the initial viewport; the start of the country section was just at the bottom. Those controls existed in the accessibility view and were actionable. I did not establish a number of physical scrolls needed. There was no observed broken entry button. My first substantive action was selecting Serbia and entering.

## What I actually did and observed

### 1. Opening view

- Read the Serbia outcome: governing list 129 of 250 seats, opposition alliance 65; Vucic campaigned but was not a parliamentary candidate.
- Opened **People, groups, and terms in Serbia**. It explained SNS-led list, Serbia Against Violence, Vucic, ODIHR, CINS, ordinary news versus campaign slots, all ballots cast, and historical window.
- Selected **Unequal access and pressure**, left confidence at **A first idea · I expect to test it**, and clicked **Record my first explanation**.
- The game preserved the supplied first idea and explicitly said I would write my current paragraph myself. That distinction worked.
- The opening already disclosed the final outputs: 60–85-word explanation with supporting IDs/source limit, 30–45-word comparison, and short rival/evidence phrases. Helpful, but a lot to remember while learning the case.

### 2. Choose evidence

The page put a result table, table notes, reminders about percentages, a reading counter, six cards, and my board together. I skimmed the six headings and previews. I fully opened only **RS3, RS4, RS2**, and their relevant source details.

- **RS3**, television: read that positive presidential/governing-party coverage dominated ordinary news; separate campaign slots and critical outlets also existed; the monitoring measured content, not audience belief. Opened the RS-A source record, including page locations and limits. Chose RS3 as support 1.
- **RS4**, call center: read that recruiters linked paid election-day work to voting for the governing party and wanted a ballot photograph. Read the cash-payment detail and the warning that one investigation was not a national vote count or court conviction. Opened the CINS source record. Chose RS4 as support 2.
- **RS2**, opposition organization: read that genuine alternatives and generally available expression/assembly existed, opposition parties organized an alliance, and Vucic campaigned for the governing list. Opened its source details. Chose RS2 as complication.

The call-center evidence was the most engaging because it described a person being asked to do something concrete. “Unequal access” became easier to picture there than in the opening options.

I wrote these board notes:

> RS3: People saw more good news about the government.
>
> RS4: Workers were asked to prove their vote.
>
> RS2: The opposition was still allowed to organize, so they were not completely shut out.

I left both support dropdowns at their default **An action**. I did not use evidence attachments. The supporting reasons were labelled optional; the complication asked for a short phrase.

The reading counter reached **Full cards opened: 3 of 6; Source records opened: 2 of 3**. Opening RS-A on RS3 caused that shared source to display “opened” on other cards, including ones whose full card I had not read. That is defensible for a shared document, but as a student I could mistake the badge for having done all of that card's source work. The page correctly warned that opening a record did not prove reading. I did not open any original-source link, so I cannot judge whether those originals are accessible or usable.

### 3. Make a connection

I filled all three required fields and the optional consequence:

> Who acted? The call center recruiters
>
> Action: They said paid election work was linked to voting for the governing party and wanted a photo.
>
> Changed option: Workers could vote for someone else, but then might miss out on the paid job.
>
> Possible effect: It could get the governing party more votes.

The library-hours example helped show what “changed option” meant. The “before” part of the required prompt was harder: the card did not give me a clean before/after account of an actual worker's choices, so I wrote a plausible conditional claim.

I wrote a first current explanation. The app counted it as **66 words**. It cited RS3 and RS4, connected paid work to pressure on workers, and said one investigation could not establish how many national votes changed.

**Early board-change notice — distinct from the later actual revision:** On first reaching **Make a connection**, above the paragraph editor, I saw:

> Your board changed. Review your explanation below; we have not rewritten it.
>
> Check nearby steps, source attachments, and board notes as well. You can defend your explanation, make it more precise, or change it.

There was a **Review my explanation** button. At that moment I had only initially populated the board; the current explanation was empty and counted **0 words**. The message therefore asked me to review writing I had not written. It was not evidence that any writing had been overwritten. It was an inaccurate-feeling prompt at that point in the workflow.

### 4. Take questions

The page explicitly said the proposed claims were supplied practice, not claims detected in my paragraph. That prevented a potentially unfair feeling of “I never said that.” However, the repeated explanations of this distinction added a lot of text before and between questions.

**Question 1:** I first chose the statement saying monitoring described typical voters' exposure. Clicked **Consider this reading**. Feedback said it went beyond the sources: selected-outlet monitoring records broadcast content, not typical exposure or persuasion. This was a believable mistake for this role, and the correction was understandable.

I reconsidered and chose the broadcast-content/one-operation answer. The feedback then said that reading stayed within the sources. Both attempts were retained.

I chose **Make a step more precise**. This revealed a note, a selector containing my actual current causal wording, and a new-wording field. I changed the optional consequence to:

> This could pressure some workers, but RS4 does not count how many votes changed.

My reason was:

> RS4 is one call center, not all voters.

I clicked **Apply to my board**.

**Actual revision handoff:** The decision now displayed **Applied: made a step more precise**, plus:

> Before: It could get the governing party more votes.
>
> Now: This could pressure some workers, but RS4 does not count how many votes changed.

It also said to check whether a source should be attached. The lower revision notice repeated **“Your board changed. Review your explanation below; we have not rewritten it.”**, now with this real before/now change. My 66-word paragraph remained exactly as I had written it. Clicking **Review my explanation** focused that paragraph field; the observed accessibility tree did not change. The notice did not visibly become a completed review acknowledgment merely from that click.

This later handoff made sense. “Board” still covered several things—card slots, causal fields, and hearing decisions—so the button name alone did not tell me what it would change. The actual-current-wording selector and before/now text were the clearest parts.

**Question 2:** I selected the answer that said the 129 seats followed 46.75% of **valid votes**. Feedback caught the denominator error: it was all ballots cast, including invalid ballots. I did not retry this second question. I kept my explanation with the reason:

> RS3 and RS4 are about campaign advantages. I am not trying to explain the seat maths.

Applied that decision. This is a completed practice attempt, not evidence that I mastered the vote/seat issue.

For the required human-reviewed responses I wrote:

> Another explanation: People might just prefer the governing party and vote for it anyway.
>
> Needed evidence: A survey asking why people voted that way and if they were offered work.

I did not open **Discuss or reflect** or the source-record drawers inside the questions. I relied on the records I had read in evidence selection.

### 5. Test another case

Selected Russia and changed the comparison purpose from its default to **Similar result, different ways to act**. A new case summary, survey table, six cards, source-reading counter, and comparison editor appeared. The scale of that page felt like being asked to start the research task again for a very short paragraph.

I read **RU1** and its source details. It covered refusal to register Nadezhdin, survey recognition/support, and the limit that a hypothetical preference could not tell us support under unrestricted entry/campaigning. The **Survey** label initially surprised me on a card headed “One possible challenger could not enter”; the full card explained the connection.

I did not read the five other Russia full cards or originals. I wrote this comparison, counted as **36 words**:

> In Russia, the commission stopped Nadezhdin from being on the ballot (RU1). In Serbia workers could still vote for opposition but risked paid work. This shows an advantage can restrict the choices before voters even decide.

I used **Glossary**, saw a searchable list of **35 terms**, and filtered **denominator** to one definition. That definition was clear. The initial list also contained terms from other cases, such as civil forfeiture, which were irrelevant to my immediate confusion. I did not use outside help.

### 6. Finish & save

The inclusion checklist initially identified the two unchecked self-confirmations. It clearly said the software checked items, lengths, IDs, and my confirmations, not meaning or quality.

At this point the self-check made me notice that my paragraph should actually mention the complication, so I manually added that the opposition could still organize and was not totally blocked. The final paragraph was counted as **76 words**:

> The government had an advantage because it got more positive TV news (RS3), while the call center connected paid work to voting for the governing party (RS4). Workers could feel they had to vote that way to get the job, so the opposition had a harder time winning support. The opposition could still organize (RS2), so it wasn't totally blocked. But one investigation does not show that most voters were pressured or how many votes changed.

Checked both self-confirmations and clicked **Finish this hearing record**. All inclusion items displayed **Included**, and I verified:

> Hearing finished · explanation ready for discussion.
>
> You can reopen and revise. Changes return it to draft.

The top status said **Saved on this device**. I did not test HTML download, JSON export/import, preview, print, or copying. I did not do the optional fictional Vale source test or the closing spoken/written rehearsal. A finished local record is the outcome I verified; I did not submit anything to a teacher. The header's “Next useful move” still instructed me to review missing items after the finished state, which weakened the sense of being done.

## Language that slowed or confused this student

These are the terms or phrases encountered on this path that needed translation or made the intended action unclear. This is not a claim that every student would find them difficult.

| Wording | My first-use difficulty / observed help |
|---|---|
| “what your evidence can carry” | Metaphor before a concrete task; I wanted “what your sources let you claim.” |
| “political actor,” “another actor,” “political alternatives” | Actor initially sounded like a performer; alternatives could mean parties, candidates, or answer options. Glossary later clarified actor/option. |
| “incumbent-backed electoral list,” “parliamentary majority,” “seat” | Three unfamiliar concepts in the Serbia entry option. Incumbent was defined after entering; the full glossary explained list and seat later. |
| “SNS-led,” “ODIHR,” “OSCE,” “CINS,” “REC” | Case help and source details helped with several names; the number of abbreviations remained a burden. REC in the source details was not clear to me from the help I opened. |
| “organization and recorded backing,” “mobilized sufficient recorded electoral backing” | Formal way of saying organizing and counted votes; hard to turn into a first idea. |
| “ordinary news,” “campaign slots” | Needed the case help to understand they were different kinds of TV access. |
| “democratic rationale,” “dissolve parliament” | Seen on RS1's preview; I could not confidently paraphrase either without help. I did not open RS1. |
| “proportionally,” “nationwide district,” “3% threshold,” “national-minority lists” | RS6's preview introduced several rules at once. I did not fully investigate them. |
| “all ballots cast,” “valid votes,” “denominator,” “allocation” | I made the valid/all error in practice. Feedback and later denominator search helped. I would still need a concrete vote-to-seat demonstration. |
| “complication,” “source limit” | Initially sounded like a source proving me wrong. The explicit explanation that it could expose a missing step or limit was useful. |
| “action” versus “advantage” versus “limit” | The board's default categorization did not feel self-explanatory; I left the defaults. |
| “observation,” “inference,” “exposure,” “persuasion,” “scale” | Distinct research concepts compressed into question options. The first feedback made content/exposure clearer than the options alone. |
| “changed option,” “before ... after” | The library example helped. My historical source did not show a clean observed before/after worker story, so I had to infer it. |
| “hypothetical preferences,” “unrestricted entry and campaigning” | In RU1's source limit, abstract phrasing for what voters might choose if the challenger could actually run. |
| Card IDs RS3/RS4 versus source IDs RS-A/RS-C | Both are citations, but the writing requirement wanted card IDs. I had to keep two naming systems straight. |
| “Apply to my board” | Could mean swap a card, change a causal step, edit my paragraph, or just save a decision. The revealed controls and before/now text clarified the actual step edit. |
| “Review my explanation” | Moved focus; did not visibly say whether review was completed. Particularly odd above an empty first paragraph. |
| “filing length,” “record,” “Finish & save” | I understood the word count but not a classroom hand-in destination. Finished locally was clear only after the success panel. |
| HTML and JSON | The page explained readable report versus editable save, which helped; the format names are still an extra choice when I just want to hand in my work. |

## Five useful fixes

| Priority / severity | Concrete fix | Why it matters on the observed path |
|---|---|---|
| **1 — High** | Make the first action and case meaning concrete: a short Serbia hook (“Workers were asked for a photo of their vote—what could that change?”), a plain-language result (“the ruling group won 129 of 250 places in parliament”), and at-point definitions for list, seat, actor, and incumbent. Keep the outcome and uncertainty honest. | Entry's action was below a large hero, and the country option itself required election vocabulary. The concrete worker story was what engaged me later. |
| **2 — High** | Present the hearing as one challenge at a time with a compact question, source observation, answer, and explanation. Collapse repeated directions and the full board by default. In comparison, let me choose a relevant second-case lead before presenting every table/card. Keep every source reachable. | Evidence, hearing, comparison, and final pages are long reading stacks. I lost the feeling of a live hearing. A brief challenger turn or paired defense would make the reasoning feel consequential while retaining source evaluation, a rival, causal explanation, and comparison. |
| **3 — Medium** | Distinguish initial board setup from revision. Before writing, say “Your board is ready—draft your explanation.” After an actual change, name the changed field, show before/after, state that the paragraph is unchanged, and offer “Edit paragraph” or “Keep paragraph—explain why.” | The early notice asked me to review 0 words. The later actual before/now handoff was useful and should be retained. A focus-only review button leaves the status ambiguous. |
| **4 — Medium** | Make source/card relationships visible next to the evidence: “RS3 is a paraphrase from source RS-A; this source also appears on RS2/RS5/RS6.” Use “Source record previously opened” for shared badges. Put the observation and its most relevant limit side by side; show a short contextual source excerpt where permitted, with the original link still available. | Source evaluation is central, but repeated full source records and two ID systems add work. My RS3 board note still wrongly said “People saw,” even after I corrected the practice answer; a teacher needs to be able to spot that gap. |
| **5 — High** | Give Finish one clear completed state and classroom handoff: “Record complete for human review; save this report / submit it using your teacher's instructions.” Update the next-move text after success, put optional transfer/rehearsal work after the primary finish/save area, and label exported report versus resume file in plain language. Preserve the warning that inclusion is not correctness. | I could finish despite an unreconsidered wrong second practice answer, appropriately for practice, but the page still mentioned missing items and offered more tasks. I knew the record was finished, not whether class work had been handed in. |

## What worked, what I liked, and what I did not

**Worked / liked:** Explicit finite board; all evidence remains readable; no timer; a preserved first view; a manageable final writing length; concrete call-center evidence; a clear source-limit distinction; meaningful error feedback; permission to reconsider; actual-current-wording controls; no automatic rewrite of my paragraph; search in the glossary; explicit human review; save recovery that kept my board.

**Did not like:** Too much serious explanatory text before and around each action; several almost-identical-sounding multiple-choice answers; a second country page that felt like a second assignment; formal phrasing that made me work out what the question meant before working on history; repeated notices telling me what software had not done; a final page that offered another test and rehearsal before I felt released. It felt like a careful research worksheet with a hearing theme more than a game. I do not need points or faster clicking; I need the evidence dispute to feel like an exchange with a clear next action.

## Limits, unfinished tasks, and what the teacher needs next

No human learning/time claim is supported by this simulation. I did not measure completion time, read all available cards, verify original sources, or test export/import. No unfinishable product task was observed on my chosen path. My second practice answer remained incorrect; the game still correctly classified the record by inclusion rather than claiming its reasoning was correct.

A teacher should inspect my actual work, especially the leap from one call center to opposition difficulty, the difference between possible pressure and observed changed votes, the surviving “People saw” board note, and the thin way my added complication qualifies the main explanation. The generic voter survey I requested also needs discussion about whether it would actually distinguish pressure from preference. My comparison identifies different kinds of restriction but does not establish their relative effects.

The teacher needs a short visible rubric, one ordinary adequate example, a plain introduction to seats versus ballots, a clear hand-in route, and an explanation of whether the closing briefing is expected or optional. They also need to read the reasoning rather than treat a green completion list or practice selection as a score. A next pilot should include real students with varied reading confidence and ask them to explain what they think “finished” means; this review alone cannot establish student engagement, accessibility, or learning.

## Browser limits and recovery

- Initial browser connection returned **“Attempted to exit cancel scope in a different task than it was entered in.”** No game tab existed yet. One connection retry succeeded.
- After adding RS3, a batch attempting the next evidence interaction returned **“Protocol error (DOM.scrollIntoViewIfNeeded): Node is detached from document.”** I made one reload recovery, then clicked **Resume saved hearing**. The resumed game retained Serbia, the recorded opening, RS3 in support 1, and the opened source/card state. I then continued with freshly observed controls.
- These are browser-automation protocol limits, not established game defects. They did not prevent completion after recovery. The QA tab was closed after verifying the finished state. I did not retest a changed candidate or access prior feedback.
