Cross Document Reasoning

AI Research Test/
  Christianity and Western Morality/
    Holland - Dominion notes.md
    Siedentop - Inventing the Individual notes.md
    Freeman - Closing of the Western Mind notes.md
    Nietzsche - slave morality notes.md
    Human rights without God.md
    Secular moral realism.md
    Enlightenment morality.md

AI Model Analysis/
  Assumptions/
    AI token demand assumptions.md
    TSMC capacity assumptions.md
    Nvidia revenue assumptions.md
    Open-source model commoditization.md
    Frontier lab moat thesis.md
  Spreadsheets/
    AI infrastructure model.xlsx
    TSMC capacity model.xlsx
    GPU demand model.xlsx

I want you to act as a research analyst.
 
Analyze all the attached notes/files as a single body of thought. Do not merely summarize each file separately.
 
Build a master view of my beliefs and assumptions.
 
Return:
 
1. My central thesis
2. The recurring claims I keep making
3. The assumptions that seem to drive my thinking
4. The authors/sources doing the most work in my reasoning
5. The strongest evidence I rely on
6. The weakest or least-supported claims
7. Internal contradictions or tensions
8. Questions I have not resolved yet
9. Missing authors, books, or perspectives I should add
10. A revised, more rigorous version of my thesis
 
Where possible, cite the file or note that supports each point.

Analyze these notes for contradictions and unresolved tensions.
 
Focus especially on:
 
- places where I seem to accept two claims that may conflict
- places where one author undermines another author
- places where I use a concept inconsistently
- places where I need to distinguish historical influence from philosophical justification
- places where I conflate genealogy, causation, and moral validity
 
For each tension, give:
 
1. The conflicting claims
2. Where each claim appears
3. Why they conflict
4. Whether the conflict is real or only apparent
5. How I could revise the argument to remove the problem

Review these notes and identify which authors or sources appear to be highest-signal for my project.
 
By "highest-signal," I mean sources that:
 
- explain many other ideas
- recur across different notes
- clarify the central debate
- help predict where the argument goes
- force me to revise or sharpen my thesis
- are cited by multiple other authors
- expose weaknesses in competing explanations
 
Rank the top 10 sources.
 
For each source, explain:
 
1. Why it is high-signal
2. What part of my thesis it supports
3. What part of my thesis it challenges
4. Whether I should read it deeply, skim it, or use it only as reference

Analyze these notes chronologically.
 
I want to understand how my thinking changed over time.
 
Return:
 
1. My earliest position
2. My current position
3. Major turning points
4. Authors or arguments that shifted my view
5. Claims I repeated consistently
6. Claims I abandoned or weakened
7. New distinctions I learned to make
8. A timeline of my intellectual development on this topic
9. The next 5 questions I should investigate

Turn the analysis into an Obsidian-ready note.
 
Format it with:
 
- YAML frontmatter
- clear headings
- internal links using [[double brackets]]
- tags
- key claims
- counterclaims
- source list
- unresolved questions
- related notes I should create
- a final section called "Permanent Notes to Create"
 
Do not flatten the argument. Preserve tensions and disagreements.

Create an Obsidian MOC for this topic with links to the major notes I should create.

Act as an investment research analyst.
 
I am giving you several models, notes, and assumptions about AI infrastructure, token demand, frontier models, open-source commoditization, TSMC capacity, Nvidia/AMD chips, and hyperscaler capex.
 
Create a master view of the thesis implied by these files.
 
Return:
 
1. The core investment thesis
2. The key assumptions
3. The most important variables
4. The causal chain
5. The biggest upside drivers
6. The biggest downside risks
7. Contradictions between assumptions
8. What would have to be true for the thesis to work
9. What would falsify the thesis
10. A bear case, base case, and bull case
11. The 5 assumptions I should track over time

Now stress-test the thesis.
 
Assume the following possibilities:
 
1. Open-source models close the gap faster than expected.
2. Inference costs fall faster than token demand rises.
3. TSMC capacity becomes the bottleneck.
4. Frontier models continue to extend their lead.
5. Agentic workflows increase token consumption by 10x.
 
For each scenario, explain what changes in the thesis.

Stage 1: Choose one focused research domain.
Example: Christianity and Western Morality.
 
Stage 2: Gather 10–20 notes/PDFs.
Use Holland, Siedentop, Freeman, Nietzsche, Enlightenment morality, secular human rights, natural law.
 
Stage 3: Ask for a master view.
Do not ask for summaries.
 
Stage 4: Ask for contradictions.
Force the model to critique your implied thesis.
 
Stage 5: Ask for missing sources.
Use it to improve your reading list.
 
Stage 6: Ask for Obsidian output.
Turn the results into MOCs, permanent notes, and argument maps.

Do not summarize file-by-file.
 
Reason across the entire set.
 
Find the hidden structure, assumptions, contradictions, strongest evidence, weakest claims, and missing perspectives.
 
Then produce a revised version of the thesis that is more precise than the one implied by the notes.

Now critique your own analysis. Where might you have overfit the pattern, misread my notes, or imposed coherence that is not actually there?