🧠 Model Thinker Dashboard

20 Mental Models from Scott Page's "The Model Thinker"

"All models are wrong, but ensembles of wrong models can be RIGHT."
— Condorcet Jury Theorem

📊 Overview

🔍 All Models

💡 Your Use Cases

🎯 How to Use

⚡ Justin's Quick Reference: DonorBureau & TLH Applications

Your Question	Model to Apply	What to Look For
Why do some creatives 10x?	Power Law	Top 10% driving 60%+ = home run game
How do winning hooks spread?	Diffusion	Pattern reuse, R₀ > 1 = viral spread
Optimal A/B test strategy?	Multi-Armed Bandit	Explore/exploit ratio, UCB allocation
Donor lifecycle states?	Markov Chain	Transition probabilities between states
Best ask string?	Signaling	What amounts communicate to donors
Why do donors lapse?	Threshold/Tipping	Fatigue accumulation until they tip
TLH geographic patterns?	Spatial Models	County clustering, underserved areas
Competitor strategy?	Game Theory	Nash equilibrium, best response
Are we stuck on local peak?	Rugged Landscape	Incremental vs radical innovation
Resource allocation?	Colonel Blotto	Which battles to concede vs win

📊 Models by Category

📈 Distribution Models — How outcomes are distributed across a population

Power Law Distribution

Few winners dominate; long tail of small performers. 80/20 rule on steroids.

💡 "If top X% drives Y% of results, you're in a 'home run' game, not an 'average improvement' game."

Normal Distribution

Bell curve — most items cluster around average, few extremes.

💡 "If normally distributed, improving the average matters more than chasing outliers."

🕸️ Network Models — How connections affect dynamics

Diffusion & Contagion

How information, behaviors, or patterns spread through connected networks.

💡 "If R > 1, the pattern will spread exponentially until saturation."

Network Centrality

Identifying key nodes — who/what is most connected or influential.

💡 "Node X has disproportionate influence due to position Y."

🔄 Dynamic Models — How systems change over time

Markov Chains

State transitions where future depends only on current state.

💡 "Given current state X, there's P% probability of moving to state Y."

Tipping Points

Small changes accumulate until a threshold triggers dramatic change.

💡 "System is stable until threshold T is crossed, then rapid phase change."

Path Dependence

History matters; early choices constrain future options (lock-in).

💡 "Current state exists because of historical choice, not because it's optimal."

🎮 Strategic Models — How interactions between agents play out

Game Theory

Strategic interaction where outcomes depend on others' choices.

💡 "Given opponent strategy X, our best response is Y."

Signaling

Actions communicate information; costly signals are credible.

💡 "A $X ask signals donor capacity/commitment level Y."

Colonel Blotto

Resource allocation across multiple battlefields; can't win everywhere.

💡 "Optimal strategy concentrates resources on X fronts, concedes Y."

🧠 Learning Models — How agents adapt and improve

Multi-Armed Bandit

Balancing exploration (trying new things) vs exploitation (using what works).

💡 "Optimal strategy balances exploring unknowns vs exploiting known winners."

Rugged Landscapes

Multiple local peaks; incremental optimization may miss global optimum.

💡 "We're on peak X but peak Y may be higher — need big jumps to find it."

💡 Your Real-World Use Cases

📱

DonorBureau Creative Analysis

You have 500 RNC creatives. Top 10% drive 60% of revenue. How do you optimize?

Apply These Models:

Power Law Multi-Armed Bandit Rugged Landscape Diffusion

Ensemble Insight: You're in a "home run" game, not an "optimization" game. Increase creative volume (more lottery tickets), study outliers intensely, and deliberately propagate winning hooks to new creative. Don't over-optimize — hunt for new peaks.

🏠

TLH Orange County Expansion

SD is working (25% win rate). OC is struggling (8%). Where do you focus?

Apply These Models:

Spatial/Geographic Colonel Blotto Game Theory Path Dependence

Ensemble Insight: South OC (San Clemente, Dana Point) is underserved — concentrate there. Don't spread thin across all OC. Kill PMAX in OC (competitors dominating), pivot to Search. Your DR 29 is a moat — exploit it with SEO content.

📧

Donor Retention Analysis

Why do donors lapse after 3-4 asks? How do you prevent churn?

Apply These Models:

Markov Chain Threshold/Tipping Signaling Learning

Ensemble Insight: Model donor states (Active → Cooling → Lapsed → Recovered). Identify the "fatigue threshold" — how many asks before tipping? Vary ask amounts (signaling) and message types to reset the fatigue counter. Build in "rest periods" to avoid triggering the threshold.

⚔️

Competitive Fundraising Strategy

TMA Direct is crushing you at NRSC (37% vs your 8%). What's your move?

Apply These Models:

Game Theory Colonel Blotto Network Centrality

Ensemble Insight: Don't fight head-on where they dominate. Find the "structural hole" — are there underserved committee types? Concentrate resources on 2-3 clients where you can dominate rather than spreading across all. What's your asymmetric advantage (e.g., MMS when they're SMS-focused)?

🤖

AI Consulting Pricing

How do you price "The AI Operator" tiers? $1,500 / $3,000 / $5,000?

Apply These Models:

Signaling Game Theory Nonlinear (S-curve)

Ensemble Insight: Price is a signal — too low = "not real AI" / too high = "enterprise only". $3K/mo is the Goldilocks zone for mid-market. Structure tiers as separating equilibrium: Starter for "AI curious", Operator for "I want results", War Room for "be my fractional CMO".

📝

Content Strategy (Substack)

22K subs, down from $70K to $21K ARR. How do you reignite growth?

Apply These Models:

Diffusion Tipping Point Path Dependence Random Walk

Ensemble Insight: You're path-dependent on COVID content — audience may not follow to AI. Launch "Hello AI" as fresh diffusion seed. Need a new R₀ > 1 moment (viral piece). Don't assume RG decline is random — analyze if there's a tipping point you crossed (e.g., topic fatigue).

🎯 How to Use Model Thinker

Classify Your Problem

Is this about Prediction, Explanation, Design, Decision, or Strategy?

Pick 3-5 Relevant Models

Ask: "What model might apply here?" Use the Quick Reference table for common scenarios. More models = more perspectives.

Apply Each Model's Key Question

Each model has a "key question" it helps answer. Run your data through that lens. What do you see?

Synthesize the Ensemble

Where do models agree? = High confidence insight
Where do they disagree? = Interesting tension or tradeoff to explore
What's the composite action? = Your recommendation

The REDCAPE Framework

Every model can serve 7 purposes:

Reason

Deduce logical implications

Explain

Why did this happen?

Design

How to build it?

Communicate

Convey understanding

Act

Guide decisions

Predict

Forecast outcomes

Explore

Investigate hypotheticals

🗣️ Trigger Phrases for Railstote

Say any of these to invoke many-model analysis:

"Apply model thinker lens to this"
"What models could explain this?"
"Give me different perspectives"
"What am I missing in this analysis?"
"Run this through multiple models"

🔍 All 20 Models

Click any model for full details.