SwayReports

Sway Thread Showcase: Pre-Release Review Report

Reviewed: 2026-02-19 Scope: threads.html landing page + all 6 linked student thread pages Purpose: Determine readiness for public release to showcase Sway to prospective college instructors Prior report status: Critical PII issues from prior audit (real name “leyu” in trans_athletes, surname “Wesley Kibet” in ethical_egoism, fabricated Guide quote, stray backtick, wrong-student quiz question) have all been fixed.

EXECUTIVE SUMMARY

The thread showcase is close to release-ready but has several issues that should be addressed first. De-identification is now clean across all 6 threads – no real student names were detected. The threads collectively demonstrate Sway’s core value proposition well: respectful cross-partisan dialogue, effective AI facilitation, and measurable opinion shifts. However, there are factual errors on the landing page, a broken mobile link, and a few transcript items that merit attention.

Verdict: Fix the MUST-FIX items below, then release.

MUST-FIX ISSUES

1. Broken Mobile Link (CRITICAL)

File: threads.html, line 607

The mobile overlay links to student_threads.html?standalone=true. This file does not exist anywhere in the repository. Mobile users who tap “Open in new tab” will get a 404.

Fix: Change to threads.html?standalone=true.

2. Trans Athletes Card Description is Factually Wrong (HIGH)

File: threads.html, line 650

The card says:

“Emery’s position shifts notably from strong disagreement to moderate”

Emery’s actual shift was Moderately agree (6) -> Somewhat disagree (3), a -3 shift toward disagreement. The description has the starting position, direction, and endpoint all wrong.

Fix: Replace with something like: “Emery’s position shifts notably from moderate agreement to somewhat disagree, demonstrating how structured disagreement can move deeply held views.”

3. Gender Inequality Blockquote is Truncated (MEDIUM)

File: threads.html, lines 676-678

Landing page blockquote:

“Women were subordinated to men in the past, but they have full legal, social, and economic equality in the contemporary United States.”

Actual statement in the thread:

“Women were subordinated to men in the United States in the past, and they are subordinated in other countries today, but they have full legal, social, and economic equality in the contemporary United States.”

The missing middle clause changes the framing significantly.

Fix: Match the full statement from the thread.

4. Professor’s Book Reference – De-anonymization Risk (MEDIUM)

File: student_threads/gender_inequality.html, line 2576

Rowan says: “Which is why I really enjoy reading the professor’s book, especially the chapter where they talk about gender neutral restrooms.”

This references a specific professor’s textbook with a specific chapter topic. Someone familiar with courses using Sway could identify the professor and institution, potentially narrowing down the students.

Fix: Genericize to something like “reading the course material” or “the textbook.”

DE-IDENTIFICATION ASSESSMENT

Overall: PASS – No real names detected

Previously-reported name leaks (“leyu” and “Wesley Kibet”) have been successfully fixed. All 6 threads now use only plausible pseudonyms with no emails, university names, instructor names, or other PII.

Thread	Pseudonyms	PII Status	Notes
Trans Athletes	Emery, Wesley	Clean	Guide uses she/her for Emery without explicit gender disclosure; minor risk
Immigration	Samantha, Ethan	Clean	Cited organizations (American Immigration Council) are public sources
Gender Inequality	Rowan, Zoe	See Must-Fix #4	Professor’s book reference is indirect de-anonymization vector
Gun Control	Harper, Lillian	Clean	“Have a great thanksgiving break!” is temporal context but very low risk
Ethical Egoism	Wesley, Ava	Clean	Prior surname leak fixed
AI & Autonomy	Jordan, Zoe	Clean

Pseudonym Reuse Across Threads

“Wesley” appears in both trans_athletes and ethical_egoism. “Zoe” appears in both gender_inequality and ai_autonomy. These are different students with the same pseudonym. A reader might mistakenly think the same student participated in two threads. Consider whether this is confusing enough to warrant changing one instance of each.

THREAD-BY-THREAD QUALITY ASSESSMENT

Ranking (Best to Weakest for Showcase)

Rank	Thread	Tier	Opinion Shift	Key Strength	Key Weakness
1	Trans Athletes	STRONG	Emery: -3, Wesley: -1	Most dramatic opinion change; richest evidence-based dialogue; explicit “my view has changed”	Guide’s confident “debunked” claim needs verification; quiz misattributes a claim
2	Immigration	STRONG	Samantha: -2, Ethan: +2	Textbook symmetrical depolarization; Guide’s constitutional challenge is excellent	Ends abruptly without Guide closing; Ethan uses “illegals”
3	Gun Control	STRONG	Harper: -4, Lillian: -1	Harper’s -4 is the largest single shift in the dataset; warm closure	Arguments cycle repetitively; Guide gendered as “he” in quiz
4	Gender Inequality	GOOD	Rowan: -2, Zoe: +1	Genuine convergence to same position; Guide catches real contradictions	Low narrative tension; professor’s book reference
5	Ethical Egoism	MODERATE	Wesley: N/A, Ava: +2	Guide’s philosophical challenges are sharp; high satisfaction ratings	Wesley’s arguments repetitive, writing quality uneven; no conversation closure at all
6	AI & Autonomy	MODERATE	No post-chat data	Single most quotable Guide moment across all threads (“choice architecture”)	Neither student completed post-chat opinion; extensive typos in Zoe’s messages

Thread 1: Transgender Athletes in Sports (FEATURED)

Tier: STRONG – Rightfully featured

Strengths:

Both students genuinely shift positions. Emery moves from “hormone therapy is sufficient” to acknowledging sport-specific nuances. Wesley concedes some sports (gymnastics, shooting, equestrian) are fine for trans women.
Emery explicitly states “My view has changed by this chat” – the clearest evidence of transformation in any thread.
Seven Guide interventions, all well-timed. Standout moments: clarifying testosterone vs. puberty-effects; challenging a dubious source; the Socratic “Should we exclude tall women from basketball?”
Opinion shift: Emery -3 (dramatic, crosses the midpoint).

Issues to consider:

Guide states the “900 medals” claim “has been debunked” with high confidence. If not formally debunked by a recognized fact-checker, consider softening to “has been widely questioned.”
A quiz question misattributes a claim about lung capacity/grip strength to Guide when it was Emery who raised it.
Wesley implies people transition “for the sole purpose” of sports participation – a harmful stereotype the Guide doesn’t address.
Guide uses she/her for Emery without Emery explicitly stating gender, slightly weakening de-identification.

Thread 2: Immigration & Labor Policy

Tier: STRONG – Best depolarization data

Strengths:

Both students move exactly 2 points toward the center: Samantha from “Strongly agree” to “Somewhat agree,” Ethan from “Strongly disagree” to “Somewhat disagree.” This symmetrical convergence is textbook depolarization.
Guide’s constitutional challenge (“Ethan, you keep asserting this is ‘unconstitutional’ – but can you explain which provision this would violate?”) is excellent Socratic questioning.
Samantha provides specific statistics from named sources. Ethan eventually concedes Congress’s legal authority.

Issues to consider:

Ethan uses “illegals” as a noun (line 2331). Guide doesn’t flag it. Defensible – the Guide kept focus on the substantive debate – but an instructor might notice.
Ethan’s constitutional argument is legally incorrect (cites Equal Employment Opportunity Act, which doesn’t apply to immigration admissions). Samantha correctly rebuts this.
No Guide closing message. The conversation just stops after Ethan’s final concession.

Thread 3: Gun Laws & Mass Shootings

Tier: STRONG – Largest single opinion shift

Strengths:

Harper shifts 4 points (the largest in the dataset), from “Strongly agree” (gun control won’t help) to “Somewhat disagree.” A full reversal across the midpoint.
Warm, genuine closure. Harper: “I do agree with you about how we should use both to help stop the mass shooter!”
Guide alternates challenges evenly between both students. Both rate Guide as “Strongly Agree” on improving the discussion.
Lillian’s open feedback: “It was very effective.”

Issues to consider:

Guide is gendered as “he” in a quiz question (“what key distinction did he highlight?”). Should use “the Guide.”
Word count metadata (Guide: “10 messages”) doesn’t match visible transcript (6 messages). Could confuse readers.
Arguments cycle somewhat repetitively around the same core points.

Thread 4: Gender Inequality in America

Tier: GOOD – Solid but less compelling

Strengths:

Genuine convergence: Rowan moves -2, Zoe moves +1, both landing on “Somewhat disagree.”
Guide excels at catching contradictions. Best moment: “Rowan, you just said women and men are equal ‘in those aspects’ but then immediately said ‘socially, there does seem to be a bit of tension.’ Social equality was one of the three areas listed.”
Both students express appreciation for the experience in their final messages.

Issues to consider:

Professor’s book reference (see Must-Fix #4).
Rowan has a confusing typo: “most of the time they will earn more money than men” (meant women earn less). Authentic but reads awkwardly.
Students are very agreeable throughout, creating low narrative tension. The Guide has to work hard to generate productive friction.

Thread 5: Ethical Egoism vs. Competing Moral Theories

Tier: MODERATE – Guide is strong but conversation is uneven

Strengths:

Guide’s philosophical challenges are the thread’s highlight. Best: “Wesley, if relationships and cooperation are part of self-interest, doesn’t that suggest self-interest itself isn’t actually the fundamental moral principle?”
Ava moved +2 toward agreement with ethical egoism despite appearing to win the argument – showing genuine intellectual influence.
High satisfaction ratings where completed (mostly “Strongly Agree”).

Issues to consider:

Wesley’s writing quality is noticeably below Ava’s (frequent grammar errors: “ethical egoism make that,” “blacklash,” “if care bind even when harms you”). Creates an uneven reading experience.
Wesley repeats the same “long-term self-interest promotes cooperation” argument roughly five times without deepening it.
No closing whatsoever. The thread simply stops after Ava’s care ethics defense. No Guide summary, no mutual acknowledgment. Weakest ending of any thread.
Wesley did not complete the post-chat opinion survey, so his shift is unknown.

Thread 6: AI & Autonomy

Tier: MODERATE – Outstanding Guide moment, but missing opinion shifts

Strengths:

Contains the single best Guide quote across all threads: “If an AI summarizes a 50-page report, filters your news feed, or suggests which job candidates to interview, how autonomous is your ‘final call’ when the AI has already shaped what you see and how you see it?”
Both students engage with genuine intellectual respect and make sophisticated arguments about choice architecture and corporate concentration.
Highly relevant topic for the instructor audience.
Solid survey data: both students rate Guide as “Agree” on improving the discussion, both “Agree” on Guide supporting both sides equally, Jordan rates partner as “Strongly Agree” on respectfulness.

Issues to consider:

No post-chat opinion data. Both students have pre-chat positions displayed (Zoe: Moderately disagree; Jordan: Somewhat agree) but neither completed the post-chat opinion survey, so the shift column shows dashes. The opinion shift visualization is present but shows no movement data.
Zoe’s messages contain many distracting typos: “cross chechking,” “the coundaries,” “fetecting,” “autonomy doesn;r evolove,” “the stadards.” Volume is notably higher than other threads.
Conversation ends without resolution or Guide closing.

LANDING PAGE ISSUES (COMPLETE LIST)

#	Severity	Issue	Location
1	CRITICAL	Mobile overlay links to non-existent `student_threads.html`	Line 607
2	HIGH	Trans Athletes card: Emery’s shift description is wrong	Line 650
3	MEDIUM	Gender Inequality blockquote omits significant clauses	Lines 676-678
4	LOW	Trans Athletes blockquote is a paraphrase, not exact quote (despite quotation marks)	Lines 647-648
5	LOW	AI Autonomy blockquote: “chatbots” vs “Chatbots” capitalization mismatch	Line 722
6	INFO	Pseudonyms “Wesley” and “Zoe” each appear in two thread cards	Could confuse readers
7	INFO	`ai_job_loss.html` and `climate_change.html` exist but are unlinked	May be intentional
8	INFO	noindex/noarchive/nofollow meta tags block all search engines	Presumably intentional

CROSS-CUTTING ISSUES ACROSS ALL THREADS

Cloudflare Script Artifacts

All 6 thread HTML files contain a Cloudflare challenge-platform script at the bottom of </body>. Dead code from page export. Not harmful but unnecessary. Recommend removing from all files.

Guide Gendered as “He”

In at least the gun_control quiz, Guide is referred to as “he.” Guide is an AI. Recommend using “the Guide” in all quiz questions.

Missing Guide Closing Messages

Several threads (immigration, ethical_egoism, ai_autonomy) end without a Guide wrap-up. Threads with closings (trans_athletes, gun_control) are noticeably stronger. If Guide closing messages can be added, it would improve the showcase.

Incomplete Survey Responses

Every thread has at least some unanswered survey questions. Most complete: gun_control and immigration.

Message Count Discrepancies

4 of 6 threads show message counts that don’t match the visible transcript (counts include filtered/system messages). Consider aligning displayed counts with visible messages.

WHAT THE SHOWCASE DOES WELL

Topic diversity is excellent. Policy (immigration, gun control), gender/identity (trans athletes, gender inequality), philosophy (ethical egoism), technology (AI autonomy). Demonstrates Sway’s flexibility.
Opinion shift data is compelling. Four of six threads show meaningful shifts. Harper’s -4 and the immigration thread’s symmetrical +2/-2 depolarization are powerful data points.
Guide moderation is consistently strong. Across all threads, Guide maintains balance, asks incisive Socratic questions, and catches contradictions. Survey data confirms students rate Guide positively.
Students model respectful disagreement. Zero instances of flagged unconstructive messaging across all threads. Students are polite, engage with each other’s arguments, and express appreciation.
The three-section structure (Survey + Transcript + Quiz) works well. Multiple lenses for understanding each conversation.

BEST GUIDE QUOTES FOR MARKETING

AI & Autonomy – “Choice Architecture”:

“You say autonomy is preserved when ‘we set the rules and make the final call.’ But consider: if an AI summarizes a 50-page report, filters your news feed, or suggests which job candidates to interview, how autonomous is your ‘final call’ when the AI has already shaped what you see and how you see it?”
Gender Inequality – Catching a contradiction:

“Rowan, you just said women and men are equal ‘in those aspects’ but then immediately said ‘socially, there does seem to be a bit of tension.’ Social equality was one of the three areas listed. So are women socially equal to men or not?”
Ethical Egoism – Philosophical precision:

“Wesley, if relationships and cooperation are part of self-interest, doesn’t that suggest self-interest itself isn’t actually the fundamental moral principle? If ethical egoism has to keep appealing to cooperation, fairness, and relationships to avoid problematic conclusions, maybe those things – not self-interest – are doing the real moral work.”
Immigration – Constitutional challenge:

“Ethan, you keep asserting that prioritizing immigrants for specific job categories is ‘unconstitutional’ – but can you explain which constitutional provision this would violate?”
Gun Control – Synthesis question:

“Does Harper’s argument require that we choose only behavioral intervention, or could stricter gun laws serve as a complementary regulatory baseline even if intervention is the more effective prevention tool?”
Trans Athletes – Source challenge:

“Wesley, that source is highly questionable. ‘Nfox14’ isn’t a credible outlet…”

RECOMMENDED ACTION ITEMS

Must-Fix Before Release

Fix broken mobile link: student_threads.html -> threads.html (line 607)
Correct trans_athletes card description of Emery’s opinion shift (line 650)
Fix truncated gender_inequality blockquote (lines 676-678)
Genericize professor’s book reference in gender_inequality transcript (line 2576)

Strongly Recommended

Note: AI & Autonomy thread has pre-chat opinion data but no post-chat data (dashes for shifts). Consider whether this matters enough to swap with ai_job_loss.html or climate_change.html
Fix Guide gendered as “he” in quiz questions (gun_control and ethical_egoism)
Remove Cloudflare script artifacts from all 6 thread files
Consider renaming duplicate pseudonyms (Wesley x2, Zoe x2)

Nice to Have

Match trans_athletes blockquote to exact statement wording
Align “chatbots”/”Chatbots” capitalization
Verify Guide’s “debunked” claim about 900-medals article in trans_athletes; consider softening
Fix quiz misattribution in trans_athletes (attributes lung capacity/grip strength claim to Guide; was Emery)
Align message count stats with visible transcripts across all threads