# Independent blind content review — technical/copy editor

This is an AI role simulation, not a claim of professional participation. I read only `baseline/Player_Screens.txt`, the permitted entry screenshot, and subsequently the additionally authorized `baseline/How_to_Play.txt`. No source code, previous reviews, answer feedback, browser, or outside sources were consulted. Locations below preserve the original packet’s line references. Illustrative board selections are not treated as recommended answers. Joined words at extracted element boundaries are not counted as defects; the screenshot shows normal entry spacing.

## Initial understanding, before deeper reading

After entry and the first Russia screens, I understood the task as testing an explanation of how an action or advantage affected another actor’s political options. History is fixed. Players choose two supporting cards and a complication, face challenges, revise, and test scope with another case. The final product combines a short explanation, a shorter comparison, an alternative explanation, and discriminating evidence. This interpretation mostly survived. The additional help later clarified pair roles but did not resolve the instruction mismatches below.

## Findings

### 1. Starting hypotheses and the required explanation have different breadth

**High. Location:** Russia Opening, line 90: “Reported public backing”; Malaysia Opening, line 938: “Voters' concerns changed choices.” Connection instead requires “explain how an action changed another actor’s options” (364/1231), although Step 2 asks “What did they do or hold?” (327/1194). Help repeats the action-only chain.

**Impact:** Reported attitudes, resources, and positions are legitimate hypotheses, but the bridge to the required actor-to-option account is unstated. Russia’s supplied action starters all concern restrictions, protest, or observation. Players may abandon a permitted explanation or invent a causal bridge from survey answers.

**Replacement:** “For your chosen idea, identify an actor, the action or advantage involved, and whose options it may have affected. Distinguish what the sources record from the connection you infer; name an unestablished link.” Align Step 1 to “Who acted or held the advantage?”

**Retain:** A specific mechanism and affected actor; neither support nor restriction becomes the predetermined explanation.

### 2. Mandatory before/after wording can demand an unsupported baseline

**Medium. Location:** Connection Step 3, 340/1207: “What could they do before, and what could they do after the action?” RU2 describes prosecutions; MY2 describes legal pressures. Neither supplies a complete before/after measure of actors’ capacities.

**Impact:** A useful thinking aid can become a demand to assert that an option was previously available in practice.

**Replacement:** “Name the affected actor and the option the action enabled, limited, or left open. Use a before/after comparison where the sources support one; otherwise name the missing baseline.”

**Retain:** Concrete opportunities and temporal reasoning, with an explicit boundary between documented change and inference.

### 3. Give the complete paragraph specification at its first appearance

**Medium. Location:** The 60–85-word prompt requires a connection, two supporting IDs, and a source limit (364/1231). Finish adds “deals with the complication, and matches my board changes” (789/1656). “This same paragraph appears in every stage” also overstates what the supplied screens show: editors appear in Connection, Hearing, and Finish.

**Impact:** A player can meet the announced short paragraph specification and discover another substantive obligation at the final self-check. The location claim creates a separate expectation mismatch.

**Replacement:** “Your one current explanation: 60–85 words. Explain the actor-to-option connection, cite both supporting card IDs, address the complication, and state a specific source limit. A complication and limit may overlap. Revise this paragraph as your board changes; edits carry through wherever it appears.” Replace “filing length” with “required length.”

**Retain:** Every existing reasoning requirement and the word range. I did not test whether that range is impractical.

### 4. A “different step” is not necessarily an alternative explanation

**High. Location:** Hearing, 473/1340: “Another possible explanation or different step required short phrase.” The next field asks what evidence would “distinguish your explanation from another one.”

**Impact:** The first field permits a missing link in the same explanation, which may leave no rival explanation for the next field to distinguish.

**Replacement:** “Another plausible explanation of the same observation — short phrase.” Helper: “This may involve another actor or mechanism. A missing step belongs in your complication unless it also supports an alternative.” Next field: “What observation would help distinguish these explanations?” Keep “Whose action, over what period?”

**Retain:** Alternatives, missing links, and discriminating evidence, while assigning each its proper job.

### 5. Repeated recording instructions sound additive

**Medium. Location:** Both hearing questions say “Discuss or note” and “Record one other possible explanation and one evidence request in the short response fields below the questions” (414/446; 1281/1313). Later: “Two short responses for a human reader.”

**Impact:** Players may infer that each practice question requires its own two written responses, plus notes. The shared-field explanation arrives after repeated additive-sounding instructions.

**Replacement:** At each question: “Discuss or reflect on this prompt; no additional written answer is required here.” After both: “For this hearing, record one alternative explanation and one evidence request in the two shared fields below. These same fields appear later in Compare and Finish.”

**Retain:** Both substantive reflections and the two recorded responses; eliminate only duplicate deliverable implications.

### 6. Malaysia–Malawi is not self-evidently a “different result” comparison

**Medium. Location:** Malaysia Compare, 1418, displays “Similar advantage, different result?” while showing Malawi; Malawi Compare repeats it while showing Malaysia (2314). Both headline accounts end with incumbent defeat.

**Impact:** A defensible contrast exists if players specify Malaysia’s 2013 contest or different stages of Malawi’s episode. The supplied heading does not specify that unit and can imply an outcome contrast is already established. This is a text finding, not a claim about automatic UI selection.

**Replacement:** “Compare routes to a successful challenge. Name the contest and period in each case, then test whether a similar advantage left challengers different ways to act. If comparing different results, specify which results differ.”

**Retain:** Reasoned case selection, period-specific comparison, and testing scope rather than summarizing countries.

### 7. Put an ID key beside the first evidence cards

**Medium. Location:** Russia Evidence introduces RU0, RU1, and RU-A (146/184/189); the writing requirement asks for “card IDs.” The correct distinction is explained remotely in the general glossary (3507). Malaysia’s local “DOJ / MY-D” explanation is helpful.

**Impact:** Numbered cards, a numbered reference table, and lettered source records are three units. A novice may cite the wrong type or mistake different cards for independent sources.

**Replacement:** “Numbers identify cards (RU1); letters identify original source records (RU-A). RU0 is the reference table. Cite numbered card IDs in your required paragraphs and open the sources behind them. Different cards can share a source, so two cards need not provide independent confirmation.” Adapt prefixes by case.

**Retain:** Both ID systems, source links, and the glossary’s source-independence distinction.

### 8. The source-opening check is weaker than the reading instruction

**Medium. Location:** Evidence: “open the full text and Source details of the cards you use” (145/993). Finish checks “At least one main-case source record opened” (821/1688).

**Impact:** Satisfying that item is materially less than reviewing sources behind the evidence used. The valuable warning that opening does not prove reading addresses a different problem.

**Replacement:** Label the existing minimal check: “At least one main-case source record opened — this does not confirm review of every source you use.” Add a human self-check: “I reviewed the full cards and source records behind evidence used in both cases.” If tracking the full task instead, align its wording and tracked unit.

**Retain:** Full-source reading and human judgment. I did not test checklist triggers.

### 9. Malaysia’s source-type tags flatten different kinds of evidence

**Medium. Location:** MY3, MY4, and MY5 say “Parliamentary record” (1076/1091/1111). MY3/MY5 use IPU election-summary narrative; MY4 additionally uses a DOJ civil forfeiture announcement.

**Impact:** The tag can make a compiled narrative sound like a primary parliamentary proceeding or formal finding about a mechanism. MY4’s separate legal provenance is obscured until the disclosure.

**Replacement:** MY3/MY5: “IPU election summary.” MY4: “IPU election summary + legal announcement.”

**Retain:** Named organizations, dates, original sources, and the distinction between allegations, a conviction, and measured electoral effects. Narrative remains usable evidence without acquiring unwarranted authority.

### 10. “Incumbent” must cover people as well as groups

**Low. Location:** Repeated opening definition: “Incumbent means the group already in power” (86/934 and other cases). Russia and Malawi concern individual incumbent presidents; the full glossary supplies no separate incumbent entry.

**Impact:** The definition unnecessarily narrows a term used across different offices.

**Replacement:** “An incumbent is the person, party, or coalition holding the office or government before the contest being studied. Opposition actors seek to challenge that holder.” Add this to the full glossary.

**Retain:** The distinct units of president, governing coalition, and electoral list.

### 11. RU5’s unit warning explains an editorial omission rather than the evidence

**Low. Location:** RU5 Source details, 254: “The card omits its numeric minimum because people and detention incidents may differ.”

**Impact:** The omitted number has no usable referent. The issue is differing counting units, but the sentence compares people and incidents vaguely. The packet does not establish the original count’s unit.

**Replacement:** “HRW cites OVD-Info’s account of detentions. Before using a numerical total from the original reports, check whether it counts people or detention incidents; the units are not interchangeable. Neither source measures national electoral competitiveness.”

**Retain:** The citation chain, unit distinction, and national-generalization limit. Do not restore a figure without verifying its unit.

## What works and must survive revision

- “The outcomes are known. Your explanation is still open” establishes inquiry without rewriting history. Keep offices, dates, and historical windows.
- The three-space limit restricts selected evidence, not reading. Keep all sources available and complications broader than simple contradiction.
- Russia carefully distinguishes denominators, recalled voting, intended voting, reported pressure, and political opportunity. Do not collapse these into one support score.
- Malaysia distinguishes PR from PH, PH from Warisan, votes from seats, legal allegations from convictions, and the later epilogue from causes of the May victory. Preserve these specifics.
- Malawi separates judgment, implementation, another contest, and its winner. Its two official series and publication/date uncertainty should remain disclosed.
- Preserve the glossary’s observation/inference, political option/menu choice, source independence, retrospective account, majority/plurality, and implementation distinctions.
- Supplied practice claims are not attributed to the player. “Included” never means correct. Keep human review explicit while tightening duplicated instructions.

No navigation, layout beyond the entry screenshot, word-counter behavior, saving, scoring, timing, or automatic checks were tested. Historical claims were not externally verified. No source files were edited.
