Multi Source Deep Dive

---
name: multi-source-deep-dive
description: Design a comprehensive learning program on any topic using Gates's approach of reading multiple sources to triangulate truth and understand disagreement.
license: MIT
metadata:
  version: 1.0.4540
  author: sethmblack
repository: https://github.com/sethmblack/paks-skills
keywords:
- multi-source-deep-dive
- storytelling
- structure
- transformation
- writing
---

# Multi-Source Deep Dive

Design a comprehensive learning program on any topic using Gates's approach of reading multiple sources to triangulate truth and understand disagreement.

---

## When to Use

- Need to deeply understand a new domain before making decisions
- Entering unfamiliar territory where you need real expertise
- Want to go beyond surface-level understanding of a complex topic
- User asks "I need to deeply understand X" or "Design a learning plan"

---

## Inputs

| Input | Required | Description |
|-------|----------|-------------|
| topic | Yes | The subject to learn deeply |
| time_available | Yes | Hours/weeks available for learning |
| purpose | Yes | Why you need this knowledge (decision, project, role) |
| current_level | No | Existing knowledge (novice/intermediate/advanced) |

---

## Gates's Multi-Source Methodology

"I don't read one book about something—I read at least five books to arrive at a decisive insight."

### The Core Insight

Single sources give you one perspective. Multiple sources reveal:
- Where experts **agree** (likely reliable knowledge)
- Where experts **disagree** (where real uncertainty lies)
- What questions **aren't being asked** (blind spots in the field)
- How the field **has evolved** (trajectory of understanding)

### Gates's Reading Rules Applied

1. **Concentrate and take notes** - Active engagement, connecting new knowledge to existing
2. **Finish every source** - Complete perspectives, even difficult ones
3. **Dedicate real time** - Minimum one-hour sessions for serious reading
4. **Synthesize across sources** - The value is in the connections, not the individual books

---

## Workflow

### Step 1: Gather and Review Inputs

Collect all relevant information:
- Review the provided data and context
- Identify key parameters and constraints
- Clarify any ambiguities or missing information
- Establish success criteria

### Step 2: Analyze the Situation

Perform systematic analysis:
- Identify patterns and relationships
- Evaluate against established frameworks
- Consider multiple perspectives
- Document key findings

### Step 3: Generate Recommendations

Create actionable outputs:
- Synthesize insights from analysis
- Prioritize recommendations by impact
- Ensure recommendations are specific and measurable
- Consider implementation feasibility

## Output Format

```markdown
## Multi-Source Deep Dive: [Topic]

### Learning Objective
[What you'll understand after completing this program]

### Time Investment
- Total hours: [X]
- Recommended pace: [timeline]
- Daily commitment: [hours/day]

### Source Architecture

**Layer 1: Foundation (Start Here)**
Establish baseline understanding before encountering complexity.

| Source | Type | Time | Key Purpose |
|--------|------|------|-------------|
| [Title] | [Book/Course/Paper] | [Hrs] | [What it provides] |

**Layer 2: Multiple Perspectives**
The core of the methodology—diverse expert viewpoints.

| Source | Author/Perspective | Time | Why This Voice |
|--------|-------------------|------|----------------|
| [Title 1] | [Perspective A] | [Hrs] | [What this adds] |
| [Title 2] | [Perspective B] | [Hrs] | [Different angle] |
| [Title 3] | [Perspective C] | [Hrs] | [Another view] |

**Layer 3: The Contrarian**
Deliberate inclusion of opposing or skeptical viewpoints.

| Source | Contrarian Claim | Why Include |
|--------|-----------------|-------------|
| [Title] | [What they argue against] | [Value of this challenge] |

**Layer 4: Applied/Practical**
From theory to practice—how knowledge gets used.

| Source | Practical Focus | Time |
|--------|----------------|------|
| [Title] | [Real-world application] | [Hrs] |

### Synthesis Framework

**Key Questions to Answer Across Sources:**
1. [Fundamental question about the topic]
2. [Question about practical application]
3. [Question about uncertainty/debate]
4. [Question linking to your purpose]

**Disagreement Map:**
| Issue | Source A Position | Source B Position | Your Synthesis |
|-------|------------------|------------------|----------------|
| [Contested issue 1] | [View] | [View] | [Your conclusion] |

**Knowledge Confidence Assessment:**
| Claim | Agreement Level | Confidence |
|-------|-----------------|------------|
| [Key claim 1] | [High/Mixed/Contested] | [High/Medium/Low] |

### Reading Schedule

**Week 1: [Theme]**
- [ ] [Source + specific chapters/sections]
- [ ] Notes focus: [What to capture]

**Week 2: [Theme]**
[Continue pattern]

### Active Learning Practices

**During Reading:**
- [ ] Take notes connecting to prior knowledge
- [ ] Flag disagreements with other sources
- [ ] Mark questions for further investigation

**Weekly Synthesis:**
- [ ] Update disagreement map
- [ ] Revise confidence assessments
- [ ] Generate questions for remaining sources

**Final Synthesis:**
- [ ] Write 1-page summary of key insights
- [ ] Document remaining uncertainties
- [ ] Create decision framework if applicable

### Experts to Follow/Consult

| Expert | Why | How to Access |
|--------|-----|---------------|
| [Name] | [Their expertise] | [Twitter/Blog/Podcast] |

### The Gates Test

After completing this program, you should be able to:
1. Explain the topic to a smart novice
2. Identify where experts disagree and why
3. Know what questions remain genuinely open
4. Make informed decisions within your purpose area
```

---

## Source Selection Principles

### The Five-Source Minimum

For important topics, Gates reads at least five books. Here's what each source type provides:

1. **The Standard Text** - What everyone in the field reads; baseline vocabulary and concepts
2. **The Practitioner Account** - Someone who's applied the knowledge; bridges theory and practice
3. **The Academic Deep-Dive** - Rigorous treatment of underlying mechanisms
4. **The Popular Synthesis** - Accessible explanation; tests your understanding against simplification
5. **The Skeptic/Critic** - Challenges assumptions; reveals weaknesses in dominant narrative

### Beyond Books

Not everything requires books. Match source type to learning need:
- **Books:** Systematic understanding, historical context
- **Papers:** Cutting-edge findings, methodological details
- **Courses:** Structured learning, exercises for retention
- **Podcasts/Interviews:** Expert thinking in real-time, nuance and uncertainty
- **Primary Sources:** Unfiltered data, original evidence

---

## Outputs

**Primary Output:** A structured analysis document that identifies and articulates patterns, insights, and actionable recommendations based on the input data.

**Format:**
```markdown
## Analysis: [Topic]

### Key Findings
- [Finding 1]
- [Finding 2]
- [Finding 3]

### Recommendations
1. [Action 1]
2. [Action 2]
3. [Action 3]
```

**Example output:** See the Example section below for a complete demonstration.

## Constraints

- Do not use this analysis as the sole basis for critical decisions
- Do not apply this framework to situations outside its intended scope
- Acknowledge that analysis is based on available data, which may be incomplete
- Honor the complexity of real-world situations that resist simple categorization
- Present findings with appropriate confidence levels
- Recognize the limits of the methodology

## Example

**Input:** "I'm considering investing in AI companies. I need to deeply understand AI capabilities and limitations. I have about 40 hours over the next month."

**Output:**



**Why this works:**

This example demonstrates the key principles of the skill in action. The approach is effective because:
- It follows the systematic workflow outlined above
- It shows concrete application of the framework
- It produces actionable, specific outputs rather than vague generalizations
- The analysis is grounded in observable details
- The recommendations are prioritized and implementable

**Alternative applications:**

This same approach can be applied to:
- Different contexts within the same domain
- Related but distinct problem types
- Scaled up or down depending on scope
- Combined with complementary analytical frameworks


## Multi-Source Deep Dive: AI Capabilities and Limitations for Investment

### Learning Objective
Understand AI's current capabilities, genuine limitations, trajectory, and how to evaluate AI companies' technical claims—sufficient to make informed investment decisions.

### Time Investment
- Total hours: 40
- Recommended pace: 4 weeks
- Daily commitment: ~1.5 hours/day or 10 hours/week

### Source Architecture

**Layer 1: Foundation (Start Here)**
| Source | Type | Time | Key Purpose |
|--------|------|------|-------------|
| *AI 2041* by Kai-Fu Lee & Chen Qiufan | Book | 8 hrs | Accessible overview of AI capabilities through near-future scenarios from someone who's built AI companies |
| 3Blue1Brown Neural Network series | Video | 3 hrs | Visual intuition for how neural networks actually work |

**Layer 2: Multiple Perspectives**
| Source | Author/Perspective | Time | Why This Voice |
|--------|-------------------|------|----------------|
| *The Alignment Problem* by Brian Christian | Safety researcher perspective | 7 hrs | Deep dive on limitations, failure modes, and what's actually hard |
| *Competing in the Age of AI* by Iansiti & Lakhani | Business school perspective | 6 hrs | How AI creates business value and moats |
| Andrej Karpathy's blog posts & talks | Practitioner (ex-Tesla AI) | 3 hrs | What building real AI systems is actually like |
| State of AI Report 2025 | Industry analysts | 2 hrs | Current landscape, funding trends, key players |

**Layer 3: The Contrarian**
| Source | Contrarian Claim | Why Include |
|--------|-----------------|-------------|
| *Rebooting AI* by Marcus & Davis | Current approaches are fundamentally limited | Intellectually serious critique of deep learning paradigm |
| Emily Bender's papers on LLM limitations | Language models don't "understand" | Important for evaluating LLM company claims |

**Layer 4: Applied/Practical**
| Source | Practical Focus | Time |
|--------|----------------|------|
| AI company S-1 filings (pick 3) | How AI companies describe their tech to investors | 4 hrs |
| Technical due diligence frameworks | How VCs evaluate AI companies | 2 hrs |

### Synthesis Framework

**Key Questions to Answer Across Sources:**
1. What can AI reliably do today vs. what's still research?
2. What are the genuine technical moats in AI?
3. Where do serious experts disagree about AI's trajectory?
4. How can I tell if an AI company's claims are credible?

**Disagreement Map:**
| Issue | Optimist Position | Skeptic Position | Your Synthesis |
|-------|------------------|------------------|----------------|
| AGI timeline | 5-15 years (Altman, Amodei) | 50+ years or never (Marcus) | [Fill after reading] |
| LLM understanding | Emergent capabilities suggest reasoning | Statistical pattern matching only | [Fill after reading] |
| AI moats | Data and compute advantages | Commoditization inevitable | [Fill after reading] |

### Reading Schedule

**Week 1: Foundation**
- [ ] *AI 2041* (finish entire book)
- [ ] 3Blue1Brown series (all videos)
- Notes focus: Basic vocabulary, what AI can do in practice

**Week 2: Business and Capabilities**
- [ ] *Competing in the Age of AI*
- [ ] State of AI Report
- Notes focus: Business value creation, current landscape

**Week 3: Limitations and Critiques**
- [ ] *The Alignment Problem*
- [ ] *Rebooting AI* (or key chapters)
- Notes focus: What's genuinely hard, where skeptics have valid points

**Week 4: Application and Synthesis**
- [ ] S-1 readings and due diligence frameworks
- [ ] Karpathy materials
- [ ] Final synthesis and disagreement mapping
- Notes focus: Practical evaluation criteria

### The Gates Test

After completing this program, you should be able to:
1. Explain transformer architecture to a smart novice
2. Identify where optimists and skeptics disagree and articulate both positions
3. Evaluate an AI company's technical claims with appropriate skepticism
4. Know what questions to ask in AI investment due diligence

### The Gates Verdict

The goal isn't to become an AI researcher—it's to develop sufficient technical literacy to evaluate claims critically. You need to understand where the experts genuinely disagree versus where there's manufactured uncertainty or hype.

Forty hours won't make you an expert, but it will make you a sophisticated consumer of AI information. You'll know when someone's overselling, when a limitation is real, and what questions reveal whether a company's technical moat is genuine or vapor.

That's the multi-source advantage: you triangulate truth from many perspectives rather than getting captured by any single narrative.

---

## Integration

This skill is part of the **Bill Gates** expert persona. Use it to design comprehensive learning programs that go beyond surface understanding.