The aipuria Times

Propaganda Wars is a browser-based, single-player media-literacy simulation. The student steps into the role of the incoming editor of a fictional newspaper — The Jaipuria Times — that is being out-published by a rival broadsheet, The Daily Contrarian. Over three editions (rounds), the rival releases an article laced with well-known propaganda techniques. The student must:

1. Detect which propaganda techniques the rival used.
2. Write a counter-editorial that rebuts the manipulation and restores public trust.
3. Watch the sentiment meter — the game's proxy for "did the public believe you or the rival?".

One full session lasts 15–25 minutes. Students may play unlimited sessions, but only the last 3 sessions are considered for the grade — the highest score among those three is what the faculty sees. Extra plays beyond three push older attempts out of the counted window.

• Media literacy is a survival skill. Students consume slanted content daily on WhatsApp, Instagram and YouTube. Passive lectures on "bias" don't stick — active recognition does.
• Writing is the assessment. The counter-editorial forces the student to argue, not just recognise. Argumentation reveals whether comprehension is real.
• AI-assisted, not AI-graded blindly. The AI (Gemini) drafts the rival's article and scores the student's writing against a public rubric, but every score is reviewable by faculty and every editorial is stored verbatim.
• Repetition + variation. Career, SDG, industries and ideology are picked by the student each session, so the rival's articles vary — students can't memorise their way to marks.

These are the eight classical techniques the rival paper draws from. Students must spot which are present in each article.

1. Landing — masthead, brief, "Start a new edition" button.
2. Career — the student picks a reporting lens (Investigative, Business, Tech, Sports, Political, Culture). Shapes tone.
3. SDG Focus — one UN Sustainable Development Goal. Shapes the topic space.
4. Industries — up to three industries the story touches (e.g. Energy, Health, EdTech). Shapes specifics.
5. Ideology — the editorial stance the student's paper writes from. Shapes the counter-editorial's angle.
6. Round 1, 2, 3 — three sequential rounds, each with three phases:
   • Analysis — the rival's article is displayed. Student ticks the techniques they detect. False positives hurt; misses hurt.
   • Writing — student writes a counter-editorial (≈ 40+ meaningful words). Paste is disabled to enforce original thought.
   • Stop Press — sentiment meter animates based on detection accuracy + editorial quality; a short editor's note explains what moved.
7. Reveal — final rubric breakdown, round-by-round scores, editor's qualitative notes, AI-authorship verdict.

Each round produces four sub-signals that combine into the round's score, then all three rounds combine into the final rubric.

The rubric is holistic — it grades the whole session, not each round separately.

Aggregation: the dashboard ranks students by the highest rubric_total among their LAST 3 sessions. If a student plays more than three, older attempts drop out of the counted window (they remain visible when drilling into the student but do not affect the headline grade).

Written for readers who have never seen the codebase — the key ideas only.

• One website, three roles. The same URL serves students, faculty (mentors) and admins (Editorial Director, Consulting Editor, Editors-in-Chief). What you see depends on your role.
• Sign-up needs an access code + a Jaipuria email. Roll numbers and emails are unique — no duplicates possible. Email confirmation is mandatory before playing.
• Every session is logged. Career, SDG, industries, ideology, all three editorials, all detections, the AI's rubric, the AI-authorship verdict, and the qualitative editor note are stored on the server as a row in the game_runs table.
• Row-level security ensures a student only sees their own history, a mentor only sees their own mentees, and only Editors-in-Chief and above see across faculties or campuses.
• Grades are re-computed on the fly. The dashboard doesn't store a "final grade" — each time you load it, the code picks the highest rubric_total among the student's 3 most recent sessions. Older attempts stay in the database for audit but do not affect the headline score.

Three layers of intelligence sit behind the scoring — treat them as indicative, not infallible:

1. Rival article generator. Google Gemini writes each round's rival article live, seeded by the student's chosen career + SDG + industries + ideology. That's why no two sessions read the same.
2. Rubric grader. Gemini re-reads all three editorials against the 10-mark rubric and returns the two component scores plus a qualitative editor's note. A gibberish floor (heuristic word-count + character-entropy check) hard-caps nonsense to 0 before the AI ever sees it.
3. AI-authorship check. Gemini classifies each editorial as Human, Mixed signals, or Likely AI-assisted, with a confidence % and a one-line reason per round. Shown to faculty as a coloured badge (green / amber / red) in the Evaluations tab and inside the session report.

Anti-cheat surface (deterrent, not fortress):

• Paste, copy, cut and right-click are disabled on the two writing inputs — reflection and counter-editorial.
• Drag-and-drop into those inputs is blocked.
• The AI-authorship check gives faculty a second opinion on suspiciously fluent prose.

1. Get your access code. Ask any admin (Group Editorial Director or Consulting Editor — Tech) for the current teacher access code.
2. Sign up as faculty. Open the site → Sign in / Sign up → Faculty tab → enter your Jaipuria email, name, and the teacher code. Confirm the email link.
3. Get the student access code. Same admin will share the student access code — you'll read it out in class or drop it on the LMS.
4. 15 minutes before class: open the site once yourself and click through one demo round so you know the flow.
5. In class (10-min setup):
   • Show the Student Playbook URL on the projector.
   • Walk through the eight techniques on this page for two minutes.
   • Give the student access code verbally.
   • Let them sign up. Answer "did the confirmation email arrive?" questions live.
6. Homework: ask for at least 3 sessions per student. Their grade is the best of their last 3 attempts, so extra plays never hurt — they simply replace the weakest of the counted three.
7. Grading day: open your dashboard → Evaluations tab. Every mentee shows up with their best rubric, qualitative summary, and AI-check verdict. Click a row to drill into any single session.

• Evaluations — your primary view. One row per mentee: name, roll no, sessions played, best rubric /10, qualitative feedback, AI-check verdict. Click Open sessions to see every session that student played.
• Top Performers — same dataset, ranked by best rubric.
• Faculty-wise Overview — best-of aggregates grouped by mentor (admins see all faculty).
• Campus Summary — network-wide rollups by campus. Visible only to Group Editorial Director, Editors-in-Chief and Consulting Editor — Tech.
• Student dossier (click a student row) — every session as a card: date, career, SDG, industries, per-round scores. Click a card to open the full session report with the actual counter-editorials, marking sheet, editor notes and AI-authorship section.

1. Frame the brief (2 min): "You are the editor of The Jaipuria Times. A rival paper is publishing slanted articles. Your job: spot the trick and rebut it."
2. Show the 8 propaganda techniques (3 min): walk through Bandwagon, Fear Appeal, Glittering Generality, Name-Calling, Plain Folks, Testimonial, Card Stacking, Transfer with one-line examples from Section 3.
3. Demonstrate one round live (5 min): log in with a demo account, pick a Career → SDG → Industries → Ideology, complete Analysis → Writing → Stop Press.
4. Hand out access (2 min): share the student access code and the sign-up URL. Rules: Jaipuria email only, roll number JN/JL/JJ/JI…, confirm the email before playing.
5. Set the rules (2 min): at least 3 sessions, no gibberish (auto-floored to zero), grade is the best of their last 3 attempts (older plays drop out of scoring).
6. Open the floor (1 min): let them sign up while you stay on screen to help with confirmation-email issues.

• One identity per student. Roll number + email are unique; duplicates are rejected at signup.
• Email confirmation is mandatory. Unconfirmed accounts cannot play.
• AI judge is consistent but not infallible. Use the qualitative editor notes as the source of truth, not the raw number, when in doubt.
• Best of the last 3, not average. Only each student's 3 most recent sessions count — the top one among those wins. Encourage replays: a stronger fresh attempt bumps out the weakest of the counted three.
• Plagiarism shows. Off-topic, copy-pasted, or gibberish editorials get capped low automatically.
• Access codes are role-scoped: STUDENT_ACCESS_CODE shared with the class, TEACHER_ACCESS_CODE with mentors, ADMIN_ACCESS_CODE with the Editorial Director / Consulting Editor. Ask Shiva or Gourab for the current values.

The generator and the AI judge both operate under a hard content policy. Rival articles will never reference party politics, religion, communal/caste identity, or live public controversies (Kashmir, Palestine/Israel, Ukraine, sectarian violence, sexuality/gender-identity debates), and real politicians / religious institutions are never named. If a generated article ever drifts, the platform auto-scrubs it and serves a neutral choice-aware fallback.

The same policy is applied to student submissions. Rubric scores are unchanged (we grade the essay-planning craft, not the topic), but any editorial that veers into party politics, religion, or communal territory receives a ⚠ Policy note in the feedback. Faculty see the same note attached to the student's row in the dashboard. Use it to redirect students back to their SDG + industry brief.

• Off-topic-but-neutral prose is capped low by the topical-relevance rule (already in scoring).
• Policy-flagged prose is additionally annotated so you can spot systemic issues in a cohort at a glance.
• No student is penalised for the flag itself — it is a coaching signal, not a demerit.