# F5 — independent first-use faculty review

I approached this as a skeptical political science instructor concerned with assessment and game-based learning. I had not seen the game, its development history, or other reviews. I used only the visible game at `http://terminal.local:4173/?reviewer=F5`, including help, glossary and source records. After completing play, I read the permitted facilitator guide. I did not inspect application code, hidden answers or other reviews. This is an expert walkthrough, not evidence from human students or a measured learning/time study.

Before opening help, I recorded this understanding: I was preparing a fictional public hearing about how an advantage changed another actor’s options; I would choose one historical anchor, read freely, foreground two anchors and a complication, and later test an explanation in another country. Outcomes were given, and decisions affected analysis rather than history. That understanding came directly from the landing screen. Its central question, “How did an advantage change what another actor could do?”, was intelligible, although “advantage” initially suggested incumbent resources more strongly than institutional constraints on incumbents. Malawi subsequently made the broader meaning clear.

## What I actually did

I chose Malawi, “For class” and Solo. I retained the provisional opening “A new contest was enforced.” Help clarified the task particularly well: “The browser checks whether items are included; a person must judge the interpretation.” I opened all six full Malawi cards and all six card source records, as well as table limitations. I selected MW3, the court order surviving appeal, as Anchor 1; MW5, implementation by other institutions, as Anchor 2; and MW6, a higher incumbent vote share nevertheless ending in defeat, as the complication. I wrote reasons for all three selections.

I authored an actor/action/changed-option chain connecting courts and implementing institutions to voters having another contest. I wrote a bounded explanation separating the creation of an opportunity from the eventual winner. At the first hearing question, I deliberately tested the overstatement “Once judges issued an order, no other institution’s action mattered to whether a new vote occurred.” The feedback identified MW5’s Parliament, commission and security roles and said, “Implementation belongs in the chain.” I then selected the warranted sequence and received an explanation distinguishing opportunity, implementation and outcome. Both attempts remained in the final preview. This was a deliberate feedback probe, not a naturally observed student error.

I applied a defense with an evidence-linked reason. On the second question I chose the qualified comparison of official figures, then used “Narrow a step” to revise the possible-consequence field. The interface confirmed “Board step changed” and displayed the revised wording in the persistent exhibit. I separately revised the evolving prose, which carried forward to filing. One automated input became disconnected after a decision button rerendered the controls; a fresh state check and a semantic field selection recovered it. I did not observe lost work or a substantive browser blocker.

For the second case I chose Serbia and supplied my own question: “Does the ability to challenge an election itself create an implemented fresh contest?” I read RS2 and RS5 with their source records, selected RS5, and wrote a 40-word comparison distinguishing ongoing challenges in Serbia’s stated window from Malawi’s implemented reopening. I requested later court/implementation records to distinguish unfinished proceedings from durable constraints. I also answered the optional fictional Vale source test.

My first filing attempt left the separate limitation field blank, despite limitations appearing in my prose. Filing responded, “1 inclusion items still need review. Your draft can be exported now.” I supplied the phrase and filed successfully. The screen confirmed “Hearing filed · interpretation awaits discussion.” I opened the final preview and verified the explanation, comparison, chain, decisions, attempt history and source limitations. There was no final blocker. I did not download or test import/export round trips.

The game counted my filed Malawi explanation as 69 words:

> MW3 shows courts setting aside the declared 2019 result, preventing Mutharika from relying on that declaration alone. MW5 describes institutions implementing a fresh election, giving voters another contest. This explains an opening, not Chakwera’s victory: MW6 combines alliances, turnout and majority rules. Retrospective administrator reports cannot isolate each cause. The annulled 2019 declaration cannot reconstruct a fair count, and differing 2020 totals remain unresolved despite agreement at one decimal.

## Strengths and limits of the learning task

The strongest design choice is making a limited public exhibit distinct from access to information. “The limit is what you foreground. Every source remains open” avoids teaching that additional evidence is a purchased resource or that research ends when slots fill. A complication has a real argumentative job. Defending a bounded account is explicitly legitimate; students are not rewarded merely for changing their minds. My defense and narrowing produced visible records, and narrowing actually altered a chain field. These are consequential choices within an analytical exercise, although they do not constitute a political simulation with strategic opponents or changing historical outcomes.

Malawi’s file contains unusually responsible cautions. It distinguishes declared results from a reconstructed fair count, administrator retrospection from independent observation, and implementation from the explanation of victory. Source records disclose that original judgments were not retrieved and identify reliance on scholarly analysis and MEC accounts. Two inconsistent official 2020 totals are acknowledged rather than silently harmonized. Dates, locators and source roles make claims answerable to documents. I assessed these visible disclosures; I did not independently audit each linked PDF against the paraphrase.

The corrective feedback is explanatory, specific and recoverable. My wrong answer did not merely turn red: it named the omitted institutions and their relevance. The repeated statement that this is source-scope feedback rather than a grade is valuable. However, the multiple-choice alternatives were very uneven. “The judges directly appointed Chakwera president, making voting unnecessary” is easy to reject without close reading. The warranted option is conspicuously nuanced while distractors use absolute claims. An average eighteen-year-old may learn the test’s verbal pattern faster than the distinction. My immediate correction demonstrates usable feedback, not independent understanding or durable learning.

Completion is handled more cautiously than many educational games. The preview says, “Included does not mean correct.” The filing page explicitly disclaims judgment of “quality, truth, originality, or logic.” Those are substantive safeguards, not incidental footnotes. Nevertheless, the inclusion list requires only one anchor-file source record opened, and opening is not reading. Supplied opening accounts and causal starters can produce a fluent structure with limited independent synthesis. I did not submit nonsense or demonstrate a fabricated-prose exploit; the visible structural checks themselves justify the inference that an instructor must inspect meaning. A filed record is evidence of submitted work, not mastery.

The late-read facilitator guide reinforces that caution. It calls the timetable a design estimate, distinguishes AI walkthroughs from student trials, recommends first-attempt inspection and a later fresh case, and warns against inferring effects from improved scaffolded responses. Its research discussion labels design inferences and acknowledges differences between cited interventions and this lesson, including limited source access for some studies. That is appropriately restrained. The proposed participation scale still needs instructor examples of what counts as a “substantive attempt”; numbers such as 100/70/0 can otherwise turn a nuanced rubric into completion credit.

## Clarity, presentation and practical fixes

The calm cream, dark-green and serif presentation is readable and credible. Numbered stages and the persistent exhibit maintain orientation. The causal panel’s vertical steps communicate sequence. These visuals organize work more than they illustrate political processes. Long expanded source records repeat substantial caveats, and single-line exhibit notes visibly truncate my sentences. A novice can lose the argumentative thread while scrolling through provenance, especially when also managing a short word limit.

The glossary gives useful plain definitions of observation, inference, complication and source limitation. One concrete problem is “Seat and majority”: it defines a majority only as more than half of legislative seats, whereas Malawi repeatedly discusses a majority of valid presidential votes. Add the general meaning and both contexts. “Denominator” is repeatedly demanded but not directly defined; “retrospective,” “annulment,” “petition” and “implementation” also warrant brief contextual explanations. The opening’s definitions of incumbent and opposition are well placed.

My priorities are:

1. **Strengthen assessment without adding an essay.** Require a short explanation before revealing the source-test options, or ask students to identify the exact observation supporting their selection. Add plausible, partially correct rivals rather than absolute caricatures. Preserve attempts and the separation between machine checks and human judgment.
2. **Make the human rubric usable.** Supply two concise annotated accounts: one with accurate facts but an unsupported causal jump, one with a warranted narrow inference. Include an instructor calibration example for participation credit and a genuinely new delayed task. Do not advertise validated gains before classroom evidence exists.
3. **Reduce source friction while retaining rigor.** Keep a brief claim-specific limitation beside each card, with fuller provenance available on expansion. Show that two cards citing the same administrator are not automatically independent corroboration. Use multiline exhibit notes and a missing-item link that takes the learner directly to the blank filing field.
4. **Clarify mechanism and scope.** Add a compact Malawi chronology separating petition, ruling, appeal, implementation and vote. Explain majority/denominator contextually. At comparison, make the historical-window mismatch an explicit question, since my Serbia contrast cannot establish what happened after 6 February.

A 50–60-minute class is plausible with a facilitator, tightly bounded reading and protected comparison/debrief time. Reading every expanded record, independently checking originals and conducting substantial peer challenges would strain that estimate. My automated walkthrough supplies no human timing estimate. I would pilot this as a serious, attractive evidence workshop with game constraints, and use its final records for discussion and formative assessment. I would not yet adopt it as a standalone measure of causal reasoning or evidence of learning gains.
