# R7 — Independent legal and comparative-democratization content review

This is an AI role simulation of a political scientist specializing in law, courts and comparative democratization, not participation by an actual professional reviewer. I read the player text and How to Play first, then the Source Register and Facilitator Guide. I did not inspect source code, previous reviews or interactive behavior. References below are to the frozen `Player_Screens.txt`; repeated second-case versions should receive equivalent changes.

My initial understanding was that the task concerns a bounded mechanism: an actor changes another actor's options. It does not require explaining every vote or identifying a single cause of turnover. That framing works particularly well for Malawi, where judicial reopening, implementation and electoral victory are different outcomes. The supplied materials establish no clear numerical contradiction or demonstrated false legal proposition that warrants a blocking finding. The concerns below distinguish local presentation problems from interpretive risks; an inference that the packet explicitly rejects is not evidence that the packet makes that error.

The strongest features deserve preservation. Malaysia's 2013 and 2018 alliances are distinguished, PAS is not folded into universal opposition unity, and PH's 113 seats explicitly exclude Warisan. The DOJ action is consistently described as civil allegations rather than a contemporary conviction of Najib. MY6 is repeatedly identified as a post-election epilogue. Malawi separates declared 2019 figures from an authoritative reconstruction of a fair count, discloses the two official 2020 series, and does not explain their discrepancy away. Most importantly, the Malawi hearing rejects the proposition that holding a fresh election explains its winner, or that an administrator's retrospective report establishes why officials complied. These are substantive protections, not cosmetic disclaimers.

1. **Moderate — Put the judicial-source limitation on the court card itself.** Location: MW3, lines 1947–1964, compared with MW2's MW-D record at 1939–1942 and Source Register p. 3. MW3 carries the annulment, constitutional interpretation and appellate outcome, but its attached records are only MEC reports. The legal scholarship and explicit statement that original judgments were unavailable appear elsewhere. A student following the instruction to open the records for the card used can therefore miss the most relevant legal-source limitation. MEC is also identified as an appellant, so this is not equivalent to reading an independent judicial account. **Minimal fix:** attach MW-D to MW3 and repeat its existing direct-judgment disclosure. This is a provenance improvement, not a finding that MEC's chronology is false.

2. **Moderate — Separate the remedy, reported irregularities and majority interpretation.** Location: MW3's full paragraph, line 1951, and connection starter “Ordered a fresh election and required a majority,” line 2086. These are distinct legal propositions compressed into one sequence. “The commission's reports describe altered and incorrect result forms” identifies what the administrator reports, but does not specify which propositions the court found, what legally justified annulment, or whether personal culpability was adjudicated. The text does not actually accuse Mutharika of directing fraud; that should not be alleged as an existing error. **Minimal fix:** divide the paragraph into the reported court order, the separately attributed irregularity account, and the constitutional interpretation. State that this summary does not establish personal responsibility for wrongdoing; verify any stronger statement about judicial findings against the cited legal analysis before adding it.

3. **Moderate — Distinguish a threshold's strategic effects from its mechanical application.** Location: Malawi opening option at line 1805, MW4 at 1971, and MW6 at 2006. The packet correctly says several things changed at once. Nevertheless, listing the majority rule alongside turnout and the joint ticket as possible explanations of the different winner can encourage a simple rule-change explanation. The displayed 59.3% for Chakwera is both an absolute majority and the largest share. On those reported figures, plurality versus majority alone does not select a different 2020 winner. **Minimal fix:** add, “Chakwera's reported 2020 share satisfied either threshold; any effect of the majority requirement through alliance strategy needs separate evidence.” Preserve the Guide p. 3 distinction between constitutional interpretation and a newly amended constitutional text.

4. **Low to moderate — Move the voter-transfer warning into student evidence.** Location: opening “A joint opposition ticket,” line 1804; MW4, line 1971; MW0, lines 1859–1875. A joint ticket is an observed change in the choices offered. It is not an observation that everyone who previously supported Chilima transferred to Chakwera, or that adding 2019 candidate shares measures the ticket's contribution. The Guide p. 3 already states precisely this limit, and MW6 acknowledges simultaneous changes. Thus the concern is unequal placement of an existing caution, not an unsupported alliance claim in the text. **Minimal fix:** put “The ticket does not establish how individual 2019 voters voted in 2020” on MW4 or its source limit. A useful evidence request would concern individual transitions or localized turnout and support, while acknowledging the annulled baseline.

5. **Low — Make the implementation record's limit match its claim.** Location: MW5, lines 1987–1997; MW-C's attached limitation; hearing alternatives around lines 2184–2188. MW5 appropriately calls the procedures the administrator's retrospective descriptions. The hearing also explicitly rejects an inference from implementation to officials' motives. However, the MW-C limitation principally discusses exact vote totals, although this card is about implementation. Multiple MEC cards do not independently corroborate universal delivery of result copies, posting or security practices. **Minimal fix:** append a claim-specific sentence: “This is the administrator's account of implementation; it does not independently verify every instance or establish why institutions complied.” Keep this a modest improvement: the current packet already recognizes source independence in the glossary and does not claim to prove motive.

6. **Moderate — Explain what the vote-to-seat rule does and does not establish about advantage.** Location: MY1, lines 1041–1058, especially “BN therefore kept the parliamentary majority” at 1045; Malaysia's guiding question at 933. The district rule explains how a national popular-vote majority can coexist with a parliamentary defeat. It does not, from aggregate totals alone, identify the cause of the spatial distribution or demonstrate that district boundaries were designed to benefit BN. The anti-fraud warning is good, but fraud is not the only overinterpretation available. **Minimal fix:** add, “These figures show the conversion of votes into seats; they do not by themselves identify why that distribution favored BN or establish biased districting.” No additional allegation of gerrymandering is needed. This keeps a rule of allocation from becoming an unsupported explanation of the source of incumbent advantage.

7. **Low — Align MY4's headline and source label with its careful legal text.** Location: MY4 at lines 1091–1110, repeated at 2417–2434. “Parliamentary record” and “The campaign involved costs and corruption” are less discriminating than the full card, which combines an IPU election narrative with a U.S. executive enforcement announcement alleging misappropriation. The full text, source record and glossary adequately distinguish allegations from conviction; there is no basis for saying the game treats the complaint as a judicial finding. **Minimal fix:** use “Election account and legal allegations” as the label and “Campaign concerns included costs and the 1MDB controversy” as the heading. This makes the skimmable version preserve the status distinctions already present below it.

8. **Low to moderate — Explicitly name MY6's alternative outcome when used as support.** Location: MY6, lines 1126–1142; Malaysia guiding question at 933; connection inference “Election victory left some later institutional constraints,” line 1226. Chronology is already clear: August passage and September blockage cannot cause May victory. The remaining issue is analytical target. MY6 can directly support a claim about post-turnover legislative opportunities and constraints, while the displayed country question concerns defeating the incumbent. **Minimal fix:** add, “If you use this as supporting evidence, explain later reform capacity rather than the cause of May's victory.” It remains a legitimate complication for an electoral explanation. Do not remove a useful epilogue merely because it answers a different, explicitly named question.

9. **Moderate — Specify the comparison's election and outcome before declaring similarity or difference.** Location: Malawi → Malaysia comparison, lines 2308–2318 and 2340–2343; reverse comparison around 1411–1420. The sample asks “Similar advantage, different result?” while both displayed final endpoints are opposition turnover. Malawi's main question also concerns reopening an election, whereas Malaysia's concerns winning office. This can be a defensible comparison of the 2019 declaration with Malaysia 2013, or of different routes to eventual turnover, but those are different comparisons. **Minimal fix:** require a short named target, such as “another vote became possible” or “control changed after the election,” and name the contest/year. For these two final endpoints, offer “Similar result, different routes?” as the starting wording. The existing warning against comparing percentages is necessary but does not resolve a mismatched outcome.

I would prioritize findings 1, 2 and 9 because they improve the reader's ability to identify the legal claim, the evidence carrying it and the outcome being compared. Findings 3 and 4 are compact protections against common causal shortcuts. The remaining edits should stay short; the exercise already carries substantial source guidance within a tightly bounded writing task.

This review evaluates supplied content, not whether every underlying judgment, broadcaster statistic or administrative procedure has been independently verified. No external fact-check or direct judgment inspection was undertaken, and the Source Register's candid retrieval limitation remains material. Nor does a two-case comparison establish a general causal effect of courts, electoral rules or opposition cooperation on democratization. The Guide correctly describes these selected cases as a primer rather than a controlled test, and distinguishes AI review from a human classroom trial. The evidence exercise is strongest when its final claim remains correspondingly bounded: a defensible account of one action, one changed opportunity and the limits of the observations used to support it.
