Course 10 Implementation Spec: The World Is Being Rebuilt

1. Title and source files used

Course title: Course 10: "The World Is Being Rebuilt" Owned deliverable: 10-world-is-being-rebuilt/10_world_is_being_rebuilt_course_content.md

Source files used

• 10-world-is-being-rebuilt/curriculum.md
• 10-world-is-being-rebuilt/website-prompt.md

Purpose of this document This is the implementation-ready course-content specification for Course 10. It translates the curriculum into an operational teaching plan, assessment system, artifact map, and AI-assisted grading framework that a facilitator, program operator, or content-production team can execute without inventing core instructional details.

2. Design decisions at the top

Core design choices

1. This course is a capstone seminar, not a survey course.

The curriculum positions Course 10 as the culmination of the Project Agni arc. Content therefore prioritizes synthesis, thesis formation, authorship, and responsibility over broad coverage.

1. The primary student outcome is a defensible personal building thesis.

Students should leave with a coherent worldview, not just exposure to ideas. Every session therefore moves toward a final manifesto and oral defense.

1. The course runs on intellectual rigor plus personal stakes.

The curriculum explicitly rejects passive seminar participation and "safe answers." Activities must force specificity: what students believe, what they are building, what tradeoffs they accept, and what evidence could prove them wrong.

1. The four-lens structure is used as a framing engine, not a content silo.

Complexity science, Buddhist economics, indigenous intelligence traditions, and the Qeng Ho cosmology are treated as analytical lenses students repeatedly apply to case studies and their own work.

1. Speaking and writing are equally assessed.

The final experience is not just a written capstone. Students must also present, withstand questioning, and question peers well. This mirrors real-world builder environments where clarity under pressure matters.

1. LLMs support evaluation, but not as sole judges.

Because this course includes philosophy, identity, and ambition, evaluation must preserve nuance. The AI framework below is designed for structured first-pass assessment, evidence extraction, rubric scoring, and feedback drafting, with human override on edge cases.

1. The learning environment must feel serious, spacious, and high-trust.

The website prompt describes a "library crossed with a manifesto." In delivery terms, that means fewer shallow activities, more long-form reading, deliberate silence, and disciplined discussion.

Instructional principles

• Favor depth over coverage.
• Push students from abstraction into concrete claims.
• Reward intellectual humility when it sharpens conviction rather than replacing it.
• Treat blindness, tradeoffs, and second-order effects as part of excellence, not optional add-ons.
• Maintain a standard that students can explain what they are betting against.

3. Delivery model assumptions

Default format

• Length: 4 weeks
• Sessions: 8
• Session length: 90 minutes each
• Total contact time: 12 hours
• Recommended cohort size: 8-16 students
• Delivery mode: In-person preferred; synchronous online acceptable if cameras are on and breakout rooms are reliable
• Prerequisite assumption: Students have completed at least 2 prior Project Agni courses

Compression option

The curriculum allows a 2-week intensive version. This spec assumes the 4-week default. For the intensive version:

• Keep all sessions.
• Reduce between-session reading volume by pre-distributing excerpts.
• Merge office hours into Session 6 and add asynchronous annotation requirements.
• Preserve the final manifesto and presentation standards.

Classroom and tooling assumptions

• Whiteboard or shared digital board
• Projector or screen
• Shared document workspace for submissions
• LMS or folder structure that can collect:
• reading notes
• framework drafts
• manifesto drafts
• final manifesto
• presentation outline
• peer questions
• Optional LLM stack for assessment:
• one model pass for evidence extraction and rubric pre-scoring
• one model pass for feedback drafting
• human review before release on high-stakes grades

Facilitation assumptions

• Facilitator is comfortable with Socratic questioning
• Facilitator has read the assigned materials in advance
• Facilitator can redirect vague moralizing into concrete claims
• Facilitator is prepared to manage emotionally charged identity, purpose, and ambition discussions without collapsing rigor

Student workload assumptions

• Reading: 40-70 pages or equivalent excerpts per week
• Writing: 3,500-5,000 words total across framework and manifesto work
• Discussion prep: Students come with marked passages and at least 2 written questions per reading block
• Presentation prep: 2-3 hours outside class

4. Detailed course content broken down by module, session, lesson, activity, timing, facilitator moves, and student outputs

Module 1: Where Are We? The Civilizational Frame

Module outcomes

By the end of Module 1, students can:

• explain the course thesis that the current era is a rebuilding moment
• distinguish between change and first-principles rebuilding
• compare the four intellectual lenses at a useful level of precision
• begin articulating which lens most attracts and distorts their own thinking

Session 1: The Rebuilding

Session objective Establish the course thesis, build urgency without hype, and move students from "AI is changing everything" clichés toward a civilizational frame.

Pre-work

• No major reading required before Session 1
• Students complete a 10-minute written intake:
• What is being rebuilt right now?
• Who seems to be shaping that rebuilding?
• What role, if any, do you want in it?

Lesson 1.1: Opening provocation

• Timing: 0-15 min
• Activity: Silent write, then pair-share
• Prompt: "What is the difference between living through change and living through rebuilding?"
• Facilitator moves:
• Start in silence; do not lecture immediately
• Press for examples of foundations, not trends
• Capture student language on the board under two columns: surface changes and foundational rewrites
• Student outputs:
• 1 paragraph silent write
• 1 verbally shared example of a foundational rewrite

Lesson 1.2: Mini-lecture on rebuilding moments

• Timing: 15-35 min
• Activity: Historical framing lecture with guided note-taking
• Content anchors:
• Agricultural Revolution
• Industrial Revolution
• Digital Revolution
• AI revolution as "digitization of reasoning itself"
• Facilitator moves:
• Use the exact curricular question: what foundations are being rebuilt, and who are the builders?
• Avoid triumphalist framing; include instability and dislocation
• Pause every 5-7 minutes for students to identify what changed in coordination, labor, value, and identity
• Student outputs:
• Rebuilding timeline notes
• Completed four-column chart:
• foundation
• what changed
• who gained power
• who lost power

Lesson 1.3: Builder mapping exercise

• Timing: 35-55 min
• Activity: Small-group mapping
• Task: Groups choose one domain:
• energy
• education
• governance
• biology
• media
• labor

Then answer:

• what is being rebuilt?
• who are current builders?
• what assumptions are being discarded?
• Facilitator moves:
• Push groups away from naming only celebrities or founders
• Ask where incentives, protocols, and infrastructure matter more than charismatic individuals
• Ask what "first principles" means in the chosen domain
• Student outputs:
• 1 poster or shared board map per group
• 2-minute group report-out

Lesson 1.4: Course arc and reading launch

• Timing: 55-75 min
• Activity: Whole-class debrief and reading briefing
• Facilitator moves:
• Introduce the idea that four traditions illuminate different dimensions of building
• Tell students that the course ends with a manifesto and public defense
• Explain that reading quality will directly affect speaking quality
• Student outputs:
• Annotated reading plan
• Personal commitment statement: "One question I want this course to sharpen"

Lesson 1.5: Exit ticket

• Timing: 75-90 min
• Activity: Written exit and share
• Prompt: "If the world is being rebuilt, what kind of builder do you fear becoming?"
• Facilitator moves:
• Normalize that ambition can produce blindness
• Save strong anonymous lines for Session 4
• Student outputs:
• Exit ticket submission

Between-session assignment

• Read:
• Vernor Vinge, "The Coming Technological Singularity"
• Peter Diamandis, Abundance introduction and Chapter 1
• E.F. Schumacher, "Buddhist Economics"
• Prepare:
• 3 annotated passages
• 2 discussion questions
• 1 claim each author makes that the student thinks is wrong, incomplete, or dangerous

Session 2: The Traditions - Four Lenses

Session objective Give students a working comparative frame for the four lenses and establish that every lens reveals some truths while hiding others.

Pre-work check

• Short reading quiz or annotation check at entry

Lesson 2.1: Reading accountability

• Timing: 0-15 min
• Activity: Low-stakes written quiz
• Sample prompts:
• What does Schumacher critique in conventional economics?
• What is the core acceleration claim in Vinge?
• What assumption about abundance does Diamandis rely on?
• Facilitator moves:
• Score for completion and fidelity, not polish
• Use answers to spot shallow reading
• Student outputs:
• Quiz response

Lesson 2.2: Four-lens guided tour

• Timing: 15-40 min
• Activity: Structured lecture-discussion
• Lens sequence:
• Complexity science and emergence
• Buddhist economics
• Indigenous intelligence traditions
• Qeng Ho cosmology / legacy time horizon
• Facilitator moves:
• For each lens, define:
• core claim
• unit of analysis
• moral center
• likely blind spot
• Keep the Qeng Ho discussion grounded in time horizon and civilizational persistence rather than fandom
• Student outputs:
• Completed comparison matrix with five fields per lens:
• what it optimizes for
• what it notices
• what it ignores
• what kind of builder it produces
• what kind of failure it risks

Lesson 2.3: Lens stations

• Timing: 40-65 min
• Activity: Rotating discussion stations
• Task: At each station students answer:
• What becomes visible through this lens?
• What becomes morally urgent?
• What dangerous simplification might followers of this lens make?
• Facilitator moves:
• Require evidence from readings or lecture notes
• Stop students from reducing indigenous knowledge to vague "harmony with nature"
• Stop students from reducing complexity to "everything is unpredictable"
• Student outputs:
• Station notes
• 1 strongest insight and 1 blind spot per lens

Lesson 2.4: Personal attraction and blindness

• Timing: 65-82 min
• Activity: Individual reflection and triads
• Prompt: "Which lens most attracts you, and what does that attraction make you likely to miss?"
• Facilitator moves:
• Ask students to name a concrete missed stakeholder, timescale, or tradeoff
• Reward specificity over moral performance
• Student outputs:
• Reflection memo, 250-300 words

Lesson 2.5: Bridge to case studies

• Timing: 82-90 min
• Activity: Instructor close
• Facilitator moves:
• Preview that Session 3 tests the lenses against real builders
• Assign case-study preparation
• Student outputs:
• Selected case-study preference ranking

Between-session assignment

• Read or watch assigned material on:
• XPRIZE
• Yamanaka factors
• Ethereum governance
• Yawanawa / ancestral intelligence
• Write a 1-page case brief:
• what was built or reconfigured?
• what problem definition drove it?
• which lens best interprets it?
• what blindness accompanied it?

Module 2: The Builders

Module outcomes

By the end of Module 2, students can:

• analyze builders and systems builders using the four lenses
• identify blind spots built into ambitious projects
• apply the conscientious builder framework to cases and to themselves

Session 3: Case Studies in Civilizational Building

Session objective Help students see that "building" includes incentives, protocols, discovery frameworks, and cultural memory, not just products or companies.

Lesson 3.1: Case brief lightning round

• Timing: 0-20 min
• Activity: 90-second student briefings
• Facilitator moves:
• Enforce time
• Ask each speaker to name the system-level shift, not just the biography
• Student outputs:
• Oral case brief
• Uploaded written case brief

Lesson 3.2: Deep dive A and B

• Timing: 20-40 min
• Activity: Whole-class discussion on XPRIZE and Yamanaka factors
• Facilitator moves:
• Ask: did this builder create an object, a protocol, an incentive, or a new possibility space?
• Push on "rewrite vs improve"
• Ask what second-order effects would count as success or failure
• Student outputs:
• Comparative notes on incentive redesign vs scientific first-principles rewrite

Lesson 3.3: Deep dive C and D

• Timing: 40-60 min
• Activity: Whole-class discussion on Ethereum and Yawanawa
• Facilitator moves:
• Keep governance and legitimacy central in Ethereum discussion
• Keep transmission, reciprocity, and long-horizon stewardship central in Yawanawa discussion
• Require students to avoid romanticization and cynicism alike
• Student outputs:
• Comparative notes on non-state governance and ancestral intelligence

Lesson 3.4: Builder typology workshop

• Timing: 60-78 min
• Activity: Small-group synthesis
• Task: Build a typology of builders:
• system designers
• protocol designers
• discovery unlockers
• stewardship builders
• narrative shapers
• Facilitator moves:
• Ask what each builder type is accountable for
• Ask which type students instinctively respect most and why
• Student outputs:
• Group typology chart
• Ranked list of builder types by civilizational leverage and risk

Lesson 3.5: Written synthesis

• Timing: 78-90 min
• Activity: Individual writing
• Prompt: "Which case most resembles the kind of building you want to do, and what warning should you take from it?"
• Facilitator moves:
• Collect this as a seed for Session 5 framework work
• Student outputs:
• 400-word synthesis memo

Between-session assignment

• Read facilitator-provided notes on:
• builder hubris
• perverse incentives
• reversibility and feedback loops
• Write a short analysis:
• Describe one important builder failure or blindness in a domain you care about.

Session 4: The Counterargument - Builders and Their Blindnesses

Session objective Teach students that serious ambition requires serious self-critique, and give them a usable framework for responsible building.

Lesson 4.1: Anonymous fear lines

• Timing: 0-10 min
• Activity: Facilitator reads selected anonymous exit lines from Session 1
• Facilitator moves:
• Use these to re-open the moral stakes
• Invite students to notice recurring fears: domination, vanity, abstraction, detachment, unintended harm
• Student outputs:
• Quick write: "Which fear feels most real to you now?"

Lesson 4.2: Builder blindness seminar

• Timing: 10-35 min
• Activity: Discussion using failure cases
• Facilitator moves:
• Ask: what did conviction make possible here?
• Ask: what did conviction make invisible?
• Require students to name the mechanism of blindness, not just say "they were overconfident"
• Student outputs:
• Blindness mechanism notes:
• incentive blindness
• stakeholder blindness
• scale blindness
• time-horizon blindness
• measurement blindness

Lesson 4.3: Conscientious Builder Framework

• Timing: 35-55 min
• Activity: Direct instruction plus examples
• Framework components:
• build with feedback mechanisms
• build for reversibility
• build with affected people, not just for them
• build at the right scale
• Facilitator moves:
• Give one concrete good and bad example for each component
• Clarify that "right scale" can mean smaller
• Student outputs:
• Conscientious builder checklist

Lesson 4.4: Stress test lab

• Timing: 55-78 min
• Activity: Students apply the framework to their own emerging area of interest
• Prompt set:
• What harm signal would tell you your project is going wrong?
• What part of your project should remain reversible?
• Who needs to be inside the design loop early?
• What part of your ambition is oversized?
• Facilitator moves:
• Interrupt generic answers
• Ask students to identify one operational safeguard
• Student outputs:
• 1-page project stress test worksheet

Lesson 4.5: Transition to framework document

• Timing: 78-90 min
• Activity: Assignment briefing
• Facilitator moves:
• Present the framework template from the curriculum
• Explain that Version 1 must be sharp enough to revise, not polished enough to hide uncertainty
• Student outputs:
• Outline for framework document

Between-session assignment

• Draft My Building Framework v1
• Length: 800-1,200 words
• Must include:
• where we are
• what being a builder means to me
• the work
• why it matters beyond me
• my lens/lenses
• my blindness
• revisit date

Module 3: Your Position in the Rebuilding

Module outcomes

By the end of Module 3, students can:

• state a clear personal thesis about this historical moment
• name a project, problem space, or search direction they are committing to
• explain why their work matters beyond personal success
• revise their thesis in response to critique without diluting it into vagueness

Session 5: The Personal Framework Workshop

Session objective Convert abstract engagement into a first coherent self-positioning document.

Lesson 5.1: Framework gallery walk

• Timing: 0-18 min
• Activity: Students post printed excerpts or shared digital cards from their framework
• Facilitator moves:
• Ask peers to leave comments under three headings:
• sharpest claim
• missing lens
• biggest risk
• Student outputs:
• Commented framework excerpt

Lesson 5.2: Mini-lesson on thesis quality

• Timing: 18-30 min
• Activity: Facilitator models weak vs strong thesis statements
• Facilitator moves:
• Contrast:
• weak: "AI will change everything, so I want to help people"
• strong: "If reasoning becomes cheap, trust and coordination become scarcer; I want to build tools that help small communities govern shared decisions without defaulting to institutional bloat"
• Emphasize falsifiability, specificity, and beyond-self logic
• Student outputs:
• Revised thesis sentence

Lesson 5.3: Paired framework review

• Timing: 30-55 min
• Activity: Structured pair exchange
• Review protocol:
• identify strongest paragraph
• identify missing lens
• ask one Socratic question beginning with "What would you say if you're wrong about..."
• identify one phrase that sounds impressive but means little
• Facilitator moves:
• Circulate and sharpen reviewer comments
• Ban empty praise
• Student outputs:
• Peer review sheet
• Received annotated framework

Lesson 5.4: Revision sprint

• Timing: 55-78 min
• Activity: Silent revision
• Facilitator moves:
• Hold 3-minute desk-side conferences
• Ask each student: "What are you betting against?"
• Student outputs:
• Framework v1.5

Lesson 5.5: Public claim circle

• Timing: 78-90 min
• Activity: One-sentence oral shares
• Prompt: "The future I am trying to help build is..."
• Facilitator moves:
• Keep tempo high
• Note which students are still hiding behind abstraction
• Student outputs:
• Oral claim statement

Between-session assignment

• Expand framework into capstone manifesto draft
• Target length by Session 6: 1,800-2,500 words
• Include:
• thesis
• supporting arguments
• personal narrative component
• forward-looking commitment
• articulation of acceptable downside if wrong

Session 6: The Writing Workshop

Session objective Support students in turning a framework into a rigorous, personal, long-form manifesto.

Lesson 6.1: Manifesto architecture briefing

• Timing: 0-15 min
• Activity: Direct instruction
• Recommended manifesto structure:
• opening claim about the historical moment
• analysis of forces reshaping the world
• personal position and why it is earned
• what the student is building or seeking
• beyond-self significance
• blindnesses and constraints
• concrete commitment
• Facilitator moves:
• Clarify that personal does not mean confessional
• Clarify that rigorous does not mean academic jargon
• Student outputs:
• Manifesto outline check

Lesson 6.2: Writing block 1

• Timing: 15-40 min
• Activity: Silent drafting
• Facilitator moves:
• Conference with 4-5 students
• For each conference ask:
• what is your thesis in one sentence?
• what evidence or reasoning holds it up?
• what would challenge it?
• Student outputs:
• Draft pages

Lesson 6.3: Midpoint critique clusters

• Timing: 40-58 min
• Activity: Groups of three read thesis and one key section aloud
• Feedback protocol:
• I believe you when...
• I stop believing you when...
• I need one more concrete example of...
• Facilitator moves:
• Encourage students to cut grandiosity unsupported by reasoning
• Student outputs:
• Critique notes

Lesson 6.4: Writing block 2

• Timing: 58-83 min
• Activity: Revision and expansion
• Facilitator moves:
• Conference with remaining students
• Direct stuck students to clarify either:
• what problem they think matters
• who they think benefits
• what tradeoff they accept
• Student outputs:
• Draft v2 progress

Lesson 6.5: Presentation prep launch

• Timing: 83-90 min
• Activity: Presentation briefing
• Facilitator moves:
• Explain that students will read a selected excerpt, not summarize broadly
• Explain peer questioning expectations
• Student outputs:
• Selected excerpt plan

Between-session assignment

• Submit full manifesto draft, target 3,000-5,000 words
• Submit presentation excerpt, 800-1,200 words
• Submit 3 questions the student hopes peers will ask

Module 4: The Presentations

Module outcomes

By the end of Module 4, students can:

• defend their ideas in public
• ask rigorous questions of peers
• articulate the costs, beneficiaries, and falsifiers of their own thesis
• leave with a durable statement of purpose and a revisit plan

Session 7: Capstone Presentations, Part 1

Session objective Run formal oral defenses for half the cohort while maintaining seriousness and generosity.

Lesson 7.1: Norm reset

• Timing: 0-8 min
• Activity: Facilitator frames expectations
• Facilitator moves:
• State that questioning is a form of respect
• Remind peers that weak questions stay at the level of vibe; strong questions identify assumptions
• Student outputs:
• Question tracker sheet

Lesson 7.2: Presentations 1-5

• Timing: 8-88 min
• Format per student: 12 minutes total
• 7 minutes reading
• 5 minutes questioning
• Required question bank:
• What are you betting against?
• What would it take to prove you wrong?
• Who is harmed by your thesis being right?
• What does the world look like if you succeed?
• What happens if you are wrong but act anyway?
• Facilitator moves:
• Ensure every presenter receives at least one falsification question
• Step in if peer questions are vague, flattering, or evasive
• Take notes for final assessment
• Student outputs:
• Oral presentation
• Live defense responses
• Peer question submissions

Lesson 7.3: Debrief

• Timing: 88-90 min
• Activity: Exit reflection
• Prompt: "What question today sharpened your own thinking most?"
• Student outputs:
• Reflection slip

Session 8: Capstone Presentations, Part 2 plus Closing

Session objective Complete oral defenses, consolidate learning, and create a closing commitment with appropriate gravity.

Lesson 8.1: Presentations 6-10

• Timing: 0-80 min
• Activity: Same format as Session 7
• Facilitator moves:
• Keep time tightly
• Ask follow-up questions when a student answers elegantly but evasively
• Student outputs:
• Oral presentation and defense

Lesson 8.2: Final reading and closing synthesis

• Timing: 80-86 min
• Activity: Facilitator reads the closing passage referenced by the curriculum
• Facilitator moves:
• Keep this brief and earned
• Connect the reading to asking the right question before the moment passes
• Student outputs:
• Listening notes optional

Lesson 8.3: Commitment circle

• Timing: 86-90 min
• Activity: Students complete the sentence:
• "I am building [X] because [Y] - and I accept that [Z]."
• Facilitator moves:
• Require concreteness in X, Y, and Z
• End without undercutting the moment
• Student outputs:
• Final oral commitment line
• Calendarized revisit date

5. Assignments and artifacts

Required assignments

1. Reading annotations and discussion prep

• Frequency: weekly
• Artifact: marked passages plus 2 discussion questions
• Purpose: ensure students can ground claims in text

1. Case brief

• Due: before Session 3
• Length: 1 page
• Purpose: practice lens-based analysis on real builders

1. Builder failure analysis

• Due: before Session 4
• Length: 500-700 words
• Purpose: identify mechanisms of blindness

1. My Building Framework v1

• Due: before Session 5
• Length: 800-1,200 words
• Purpose: state a first coherent personal position

1. Capstone manifesto

• Due: before Session 7 or 8 presentation slot
• Length: 3,000-5,000 words
• Purpose: articulate a civilizational thesis, building thesis, and responsible commitment

1. Presentation excerpt

• Due: with manifesto
• Length: 800-1,200 words
• Purpose: provide a read-aloud section optimized for oral defense

1. Peer questioning log

• Due: after final presentation session
• Length: 1 page or structured form
• Purpose: capture the quality of questions each student asked

1. Framework revision note

• Due: 48 hours after presentation
• Length: 400-600 words
• Purpose: record how critique changed or failed to change the student's thinking

Core artifacts produced by the course

• Intake reflection
• Rebuilding timeline chart
• Four-lens comparison matrix
• Case brief
• Builder typology chart
• Stress test worksheet
• Building Framework v1 and revised version
• Final manifesto
• Presentation excerpt
• Peer questioning log
• Final commitment statement
• Revisit date entry

Artifact quality expectations

• Every major artifact should contain at least one explicit claim, one identified tradeoff, and one form of evidence or reasoning.
• The final manifesto must connect personal ambition to a beyond-self consequence.
• Students must name at least one plausible blindness in their own framework.

6. AI/LLM grading and assessment framework

Role of AI in this course

LLMs may be used for:

• rubric-aligned first-pass scoring
• evidence extraction
• consistency checking across submissions
• feedback drafting
• identifying likely vagueness, unsupported claims, or missing rubric elements

LLMs should not be used as the only authority for:

• final grade determination on borderline submissions
• judgments about sincerity, morality, or personal worth
• disciplinary action for plagiarism without additional review

Assessment model

Recommended workflow

1. Human or system ingests assignment text.
2. LLM extracts evidence snippets aligned to rubric rows.
3. LLM proposes row-level scores and confidence.
4. LLM flags ambiguity, missing evidence, or possible integrity concerns.
5. Human reviews:

• all low-confidence cases
• all failing grades
• all top-decile grades
• all integrity flags

1. LLM drafts student-facing feedback only after row-level evidence is established.

Grade weights

Aligned to curriculum, with operational detail:

• Personal Framework Document: 15%
• Capstone Manifesto: 40%
• Presentation: 25%
• Peer Questioning: 10%
• Framework Revisions: 10%

LLM evaluation heuristics by assignment

A. Reading annotations and discussion prep

Score for:

• evidence of actual reading
• ability to identify central claims
• quality of disagreement or tension noticed

Heuristics:

• strong if the student cites or paraphrases accurately and poses non-obvious questions
• weak if questions are generic, purely emotional, or answerable without reading

B. Case brief

Score for:

• accurate case description
• correct use of at least one lens
• identification of a meaningful blindness

Heuristics:

• strong if the student distinguishes product, protocol, incentive, or system redesign
• weak if the student only summarizes biography or treats the case as inspirational content

C. Framework document

Score for:

• clarity of historical thesis
• specificity of builder identity or search direction
• beyond-self logic
• self-identified blindness

Heuristics:

• strong if each section of the template contains concrete claims
• weak if the student says they want to "help people" or "change the world" without mechanism

D. Capstone manifesto

Score for:

• thesis sharpness
• argumentative coherence
• integration of course lenses or cases
• seriousness about tradeoffs and falsifiability
• quality of personal commitment

Heuristics:

• strong if the manifesto could be summarized as a crisp claim about this moment and a justified choice of work
• average if it shows conviction but lacks evidence, structure, or concrete downstream implications
• weak if it is mostly inspirational rhetoric, summary of readings, or autobiographical drift without thesis

E. Presentation

Score for:

• excerpt quality
• oral clarity
• ability to answer questions directly
• ability to articulate what is being bet against

Heuristics:

• strong if the student answers challenges without collapsing into defensiveness or vagueness
• weak if the student repeats lines from the paper without engaging the actual question

F. Peer questioning

Score for:

• relevance
• sharpness
• assumption-testing quality
• contribution to peer thinking

Heuristics:

• strong questions surface falsifiers, harmed parties, hidden assumptions, or missing stakeholders
• weak questions are compliments framed as questions or requests for generic inspiration

G. Revision note

Score for:

• evidence the student processed critique
• ability to distinguish critique worth incorporating from critique worth rejecting

Heuristics:

• strong if the student updates claims precisely or defends a non-change with clear reasoning
• weak if the student says they "took all feedback into account" without identifying any real shift

LLM confidence and escalation rules

The model must attach confidence labels:

• high: direct evidence supports the score
• medium: evidence exists but interpretation could vary
• low: score depends on nuance, intent, or unclear writing

Mandatory human review when:

• confidence is low on any rubric row
• overall score is below passing
• integrity concerns are flagged
• the submission contains vulnerable personal disclosures that may distort grading
• the model detects possible strong originality but unusual form

7. Rubrics, scoring criteria, and evaluator prompt guidance

Master scoring scale

• 4 - Excellent: precise, rigorous, insightful, and well-supported
• 3 - Proficient: clear and competent with some depth, but limited sharpness or nuance
• 2 - Developing: partially meets expectations; too vague, uneven, or under-supported
• 1 - Insufficient: misses core requirements or shows little evidence of course learning

Rubric: Personal Framework Document

Rubric: Capstone Manifesto

Rubric: Presentation and defense

Rubric: Peer questioning

Scoring conversion

Recommended row average conversion:

• 3.70-4.00 = A
• 3.30-3.69 = A-
• 3.00-3.29 = B+
• 2.70-2.99 = B
• 2.30-2.69 = B-
• 2.00-2.29 = C
• 1.50-1.99 = D
• below 1.50 = F

Evaluator prompt guidance

Use evaluator prompts that require evidence-first grading. The model should quote or paraphrase short submission snippets before assigning scores.

Recommended system prompt for rubric scoring

You are evaluating a student submission for Project Agni Course 10: "The World Is Being Rebuilt."
Grade only against the supplied rubric. Do not reward length, confidence, or style unless the rubric calls for it.
First extract evidence for each rubric criterion. Then assign a 1-4 score for each criterion.
If evidence is missing, say so explicitly. Do not infer mastery from tone.
Flag low confidence if the writing is ambiguous, unusually experimental, or emotionally vulnerable in ways that make grading uncertain.
Return structured JSON only.
```

**Recommended user prompt template**

Assignment type: [framework / manifesto / presentation reflection / peer questioning] Rubric: [paste rubric rows] Student submission: [paste text]

Return JSON with:

• criterion_scores: array of {criterion, score_1_to_4, evidence, confidence}
• overall_summary: short paragraph
• strengths: array
• gaps: array
• followup_questions_for_human_reviewer: array
• integrity_flags: array

### Additional evaluator guardrails

- Do not punish a student for taking an unconventional position if it is reasoned.
- Do not reward moral language that substitutes for argument.
- Do not assume optimism is stronger than pessimism, or vice versa.
- Distinguish uncertainty from incoherence.
- Distinguish intensity from precision.

## 8. Feedback strategy: what strong/average/weak responses look like and how an LLM should respond

### Feedback philosophy

Feedback should make students sharper, not merely encouraged. The correct tone is serious, specific, and developmental. It should name what is working, where the argument weakens, and what next revision would most improve the work.

### Strong response profile

Strong responses usually:
- make a clear and arguable claim about the historical moment
- connect that claim to a specific kind of building
- explain who benefits and what persists beyond the student
- identify a real blindness or downside
- respond to course material through synthesis rather than summary

**How the LLM should respond to strong work**
- confirm the precise strength, not generic praise
- identify one pressure test that would make the work even stronger
- avoid overcorrecting or inventing weaknesses for balance

**Example feedback pattern**
- "Your strongest move is the claim that cheap reasoning increases the value of trustworthy coordination. That gives your manifesto a real center."
- "The next useful pressure test is to specify who loses if your thesis succeeds and how you would recognize that early."

### Average response profile

Average responses usually:
- show genuine engagement
- contain a thesis, but too broad or partially supported
- gesture at beyond-self significance without mechanism
- mention blindness without operationalizing it
- lean on inspiring language when reasoning gets thin

**How the LLM should respond to average work**
- identify the strongest available core claim
- point to the exact place where abstraction takes over
- recommend one high-leverage revision move

**Example feedback pattern**
- "You have the beginning of a thesis in the sentence about rebuilding trust infrastructure, but much of the essay stays at the level of aspiration."
- "Revise by choosing one concrete domain and showing how your lens changes what should be built there."

### Weak response profile

Weak responses usually:
- summarize readings without a position
- rely on slogans like "help people" or "change the world"
- make no clear claim about what the student is building or seeking
- avoid tradeoffs, harmed parties, or falsifiers
- confuse emotional intensity with argument

**How the LLM should respond to weak work**
- avoid vague encouragement
- name the missing requirements directly
- offer 2-3 concrete next steps, not a flood of advice

**Example feedback pattern**
- "This draft does not yet establish a clear thesis about the moment we are in or the work you intend to do."
- "Before revising further, define one problem you believe matters, one builder responsibility you accept, and one reason the work matters beyond your own success."

### Response style rules for the LLM

- Start with the most important judgment, not a compliment sandwich.
- Use textual evidence from the student submission.
- Do not psychoanalyze the student.
- Keep feedback actionable and tied to the rubric.
- End with one revision priority, not a long to-do list.

### Suggested feedback output format

```text
Overall judgment:
[2-4 sentences naming the current level of the work]

What is working:
- [specific strength]
- [specific strength]

What is not yet working:
- [specific gap]
- [specific gap]

Highest-leverage revision:
- [single most important next move]

LLM feedback heuristics by quality band

• If strong: challenge the edges of the thesis.
• If average: sharpen the center of the thesis.
• If weak: establish the existence of a thesis.

Red-flag cases requiring human feedback instead of automated release

• Student disclosure of severe distress, trauma, or crisis
• Possible plagiarism or ghostwriting
• Highly unconventional format where rubric fit is unclear
• Work that is politically or morally charged in ways likely to trigger model bias

Implementation notes for operators

• Keep all major prompts and rubrics stable across sections to improve LLM consistency.
• Require students to submit text in machine-readable format if AI scoring is used.
• Store row-level rubric evidence with the final grade for auditability.
• Sample-grade 15-20% of submissions by hand to monitor drift if AI is used at scale.

Definition of success for the course

This course is successful if, by the end, students can answer all five questions with unusual specificity:

• Where are we in history?
• What is being rebuilt?
• What does being a builder require?
• What am I building or seeking?
• Why does it matter beyond me?

The course is unusually successful if students can answer a sixth:

• What am I most likely to miss, and how will I know?