Copy of Thinking Frame – Good Explanations

Popper & Deutsch – Main Ideas

  • Purpose: Stop wasting time on untestable, shallow, or fad-driven ideas.
  • Popper’s Filter: Only test hypotheses that can be disproven. If you can’t imagine what would show you it’s wrong, it’s not worth testing.
  • Deutsch’s “Good Explanation” Test: Good explanations are hard to vary — they don’t rely on ad‑hoc assumptions, and they work across contexts. An ad hoc assumption is something you tack onto an idea, theory, or explanation only to protect it from being proven wrong — rather than because the new assumption is independently well‑supported or necessary. 
  • Practical Cue: Before any experiment, ask: Is it falsifiable? Is it a good explanation? If not, improve it before starting.

Popperian Scenarios

For each of these statements, devise a:

  1. Hypothesis
  2. Test plan
  3. Falsifier
  4. Attitude

For example:

“We’ve heard TikTok ads will dramatically increase sales for businesses like ours.”

  • Hypothesis: “Running TikTok ads will generate at least 20 additional qualified leads in 14 days at a cost per lead under $40.”
  • Test Plan: “Run $500 of ads, track clicks, leads, and cost per lead.”
  • Falsifier: “If leads are <20 or CPL >$40, we stop and reallocate budget.”
  • Attitude: “If it fails, we’ve learned where not to spend — valuable in itself.”
  1. “Customers will love it if we add a vegan menu option.”
  1. Hypothesis
  2. Test plan
  3. Falsifier
  4. Attitude

2. “My team can’t be trusted to make important decisions without me.”

  1. Hypothesis
  2. Test plan
  3. Falsifier
  4. Attitude

3. “Shorter team meetings will improve productivity.”

  1. Hypothesis
  2. Test plan
  3. Falsifier
  4. Attitude

4. “Offering remote work will destroy team culture.”

  1. Hypothesis
  2. Test plan
  3. Falsifier
  4. Attitude

The Conjectures and Refutations Mindset

The “conjectures and refutations” mindset, from Karl Popper’s philosophy of science, is both a way of thinking and a discipline for how knowledge grows. It’s as relevant to leadership and strategy as it is to physics.

The Core Idea

  • Conjecture = a bold, creative guess about how something works or what might solve a problem.

  • Refutation = the process of actively trying to prove that guess wrong through critical testing.

Popper’s point:

All our theories, strategies, and explanations are provisional. They can never be proven true once and for all — only survive repeated, serious attempts to show they’re false.

 

The Mindset in Practice

  1. Start Boldly

    • Don’t wait for perfect certainty before acting.

    • Make your best, most imaginative, well‑reasoned guess — your conjecture.

  2. Test Relentlessly

    • Design tests that could genuinely disprove your idea if it’s wrong.

    • Seek out criticism, anomalies, and contrary evidence.

  3. Value Disproof as Progress

    • If your conjecture fails, you’ve learned something real — you’re closer to the truth.

    • If it survives, it’s not proven, but it’s earned more trust until a better idea comes along.

  4. Iterate Without Ego

    • Replace “protect my idea” with “improve our understanding.”

    • Each refutation is a stepping stone to a stronger conjecture.

Why It’s Powerful for Strategy & Leadership

  • Avoids stagnation — you’re always looking for a better explanation or approach.

  • Prevents groupthink — you invite challenge instead of defending dogma.

  • Builds resilience — you expect some ideas to fail, so failure becomes fuel, not a crisis.

  • Sharpens clarity — bad ideas are eliminated faster, leaving more energy for the good ones.

Structured Testing in Strategy Work

These embed the mindset into actual leadership and planning.

  • Hypothesis‑Driven Meeting

    • Start every strategy session with: “Our current best guess is…”

    • Agree on what evidence would disprove it.

    • Purpose: Forces clarity and sets up the refutation step before groupthink sets in.

  • Small‑Bet Experiments

    • Design a reversible, low‑cost test of a strategic idea.

    • Pre‑define kill criteria and success metrics.

    • Purpose: Makes refutation safe and normal — and speeds up learning.

  • Devil’s Advocate Rotation

    • Assign one person per meeting to challenge every proposal with “What would make this fail?”

    • Rotate the role so everyone practises critical challenge.

    • Purpose: Normalises constructive dissent.

 

Real‑World Application with Feedback Loops

These take the mindset into live leadership situations.

  • Conjecture Log

    • Keep a running list of strategic bets, the reasoning behind them, and the tests you’re running.

    • Review monthly: Which survived? Which were refuted? What did we learn?

    • Purpose: Creates a visible track record of learning, not just outcomes.

  • Stakeholder Refutation Panel

    • Present your strategic conjecture to a small group of trusted peers, advisors, or even customers.

    • Ask them to only focus on what would make it fail.

    • Purpose: Surfaces blind spots early and strengthens the idea before scaling.

  • Scenario Stress‑Test

    • Take your current strategy and run it through 3–4 extreme but plausible scenarios.

    • Look for where it breaks — then adapt.

    • Purpose: Builds resilience into the plan and keeps thinking adaptive.

Ask: What did I learn about my idea? What did I learn about my thinking? 

Bootcamp

Get some more practice with conjectures and refutations thinking with a 4-week bootcamp. Download the guide.


DOWNLOAD THE GUIDE

Deutsch on Explanations

Good Explanations — Hard to Vary, Deep, and Widely Applicable

A good explanation doesn’t just fit the facts — it couldn’t easily be changed without breaking its ability to explain, and it often accounts for much more than the original problem.

Examples:

  • Seasons explained by Earth’s axial tilt

    • Why good: The tilt of Earth’s axis relative to its orbit explains not only the cycle of seasons but also why they are opposite in the two hemispheres, why day length changes, and why equatorial regions have minimal seasonal variation. Change the tilt in the model, and the predictions change in specific, testable ways.

  • Germ theory of disease

    • Why good: Explains a vast range of illnesses, predicts patterns of contagion, and leads to effective interventions (sterilisation, antibiotics, vaccines). You can’t swap “microorganisms” for “evil spirits” without losing predictive and practical power.

  • Plate tectonics

    • Why good: Accounts for earthquakes, volcanic activity, mountain formation, and the fit of continental coastlines. Altering the mechanism (e.g., replacing mantle convection with “continents drift because they feel like it”) destroys its explanatory reach.

Bad Explanations — Easy to Vary, Narrow, or Ad Hoc

A bad explanation can be tweaked endlessly without changing its fit to the facts — often because it’s vague, circular, or invokes untestable causes.

Examples:

  • Seasons explained by “the gods did it”

    • Why bad: You could change the gods, their motives, or the story details infinitely and still “explain” the seasons. It predicts nothing new and can’t be tested.

  • Illness caused by “imbalanced humours”

    • Why bad: The theory can be adjusted to fit any symptom after the fact (“too much black bile”), but it doesn’t reliably predict who will get sick or how to prevent it.

  • Earthquakes happen “when the Earth is angry”

    • Why bad: The cause is metaphorical, not mechanistic. You can vary the “anger” story endlessly without affecting the supposed explanation, and it offers no testable predictions.

 

Litmus Test: If you can change the details of the explanation without changing its ability to “fit” the facts, it’s bad. If changing the details breaks the explanation, and it also accounts for more than the original problem, it’s good.

Importance of Explanations in Leadership

In leadership, being able to tell a good explanation from a bad one isn’t just an intellectual exercise — it’s a core skill that shapes the quality of every decision, strategy, and conversation you lead.

It’s the Foundation of Sound Decisions

  • Good explanations are hard to vary and grounded in reality — they make clear, testable predictions.

  • Bad explanations are vague, ad hoc, or endlessly adjustable — they can “explain” anything after the fact but predict nothing useful.

  • Leaders who can distinguish between the two avoid building strategies on shaky foundations. They make calls that stand up under pressure instead of collapsing when conditions change.

It Speeds Up Learning and Adaptation

  • In fast‑moving environments, you don’t have time to cling to weak ideas.

  • Spotting a bad explanation early lets you refute and replace it before it drains resources.

  • This is exactly what your Adaptive Intelligence Cycle™ is designed to do — filter for quality thinking before you invest in action.

It Builds Credibility and Trust

  • Teams and stakeholders notice when your reasoning is clear, consistent, and evidence‑based.

  • When you reject flimsy explanations — even if they’re convenient — you signal integrity and intellectual honesty.

  • That trust compounds over time, making it easier to rally people behind bold moves.

It Protects Against Strategic Drift

  • Bad explanations often creep in as story‑based justifications (“Our customers are different”, “The market will bounce back because it always does”).

  • Left unchecked, they lead to reactive, short‑term thinking.

  • Good explanations anchor strategy in mechanisms that actually match how the world works.

It Creates a Culture of Rigour

  • When leaders model the habit of testing and improving explanations, teams learn to challenge ideas without ego.

  • This fosters psychological safety for dissent and discipline in decision‑making — a rare and powerful combination.

Leaders who can separate good from bad explanations make better bets, adapt faster, and inspire more trust. Those who can’t risk wasting time, money, and credibility on ideas that sound right but can’t survive reality.

Activities

Awareness & Deconstruction

Goal: Learn to spot the difference between good and bad explanations.

  • Good vs Bad Explanation Swap

    • Take a real‑world phenomenon (e.g., “Why did sales drop last quarter?”).

    • Write two explanations: one deliberately bad (vague, easy to vary) and one good (mechanistic, hard to vary).

    • Discuss why one survives Deutsch’s test and the other doesn’t.

  • Explanation Autopsy

    • Bring a past failed project or decision.

    • Dissect the original reasoning: Which parts were assumptions? Which were testable? Which were ad hoc?

Building the Skill

Goal: Practise generating strong, testable explanations.

  • Mechanism Mapping

    • For any claim, draw a simple cause‑and‑effect diagram showing how it works.

    • If you can swap out parts without breaking the logic, it’s too easy to vary — refine until changes break the explanation.

  • Prediction Challenge

    • State your explanation, then list 2–3 new predictions it makes beyond the original problem.

    • Check if those predictions are testable.

  • Constraint Builder

    • Take a draft explanation and add constraints that make it harder to vary without breaking.

    • Example: “Customer churn is due to poor service” → specify which service failures, how they cause churn, and why alternatives wouldn’t.

Live Application

Goal: Embed the habit in real strategic work.

  • Explanation First Meetings

    • Before discussing solutions, require each proposal to start with:

      1. The phenomenon/problem

      2. The proposed explanation

      3. Why it’s hard to vary

  • Refutation Round

    • Present your explanation to the team and have them try to break it.

    • If it survives, it’s stronger; if not, you’ve learned where to improve.

  • Cross‑Domain Transfer

    • Take an explanation from one domain (e.g., product adoption) and see if its mechanism applies in another (e.g., employee engagement).

Sprint

Here is a 3-week challenge for you to complete.


DOWNLOAD SPRINT

Other Helpful Ideas from Deutsch

The Principle of Optimism

Problems are inevitable, but problems are soluble.

  • Leadership relevance: Frames challenges as solvable puzzles, not fixed constraints.

  • Practical use: Recast “we can’t” conversations into “what knowledge would make this possible?”

  • Impact: Builds a culture that treats obstacles as invitations to innovate, not reasons to retreat.

Fallibilism

All knowledge is provisional — we can always be wrong.

  • Leadership relevance: Keeps ego in check and decision‑making flexible.

  • Practical use: Model “strong opinions, loosely held” — act decisively, but update quickly when better explanations emerge.

  • Impact: Reduces sunk‑cost bias and speeds adaptation.

The Reach of Explanations

  • Idea: The best explanations have reach — they apply across many contexts, not just the one they were invented for.

  • Leadership relevance: Encourages leaders to favour strategies and principles that scale and transfer, rather than one‑off fixes.

  • Practical use: In strategy reviews, ask: Where else could this explanation or approach work?

  • Impact: Builds organisational “optionality” — the ability to pivot without starting from scratch.

Open Societies & Criticism

  • Idea: Progress thrives where ideas can be freely criticised and improved.

  • Leadership relevance: Psychological safety isn’t just “being nice” — it’s creating an environment where criticism is welcomed because it strengthens ideas.

  • Practical use: Formalise “critique rounds” in strategy sessions before committing resources.

  • Impact: Prevents groupthink and surfaces better options faster.

The Beginning of Infinity Mindset

  • Idea: There’s no inherent limit to what we can understand or achieve, given the right knowledge.

  • Leadership relevance: Counters fatalism and “we’ve hit the ceiling” thinking.

  • Practical use: When facing a “hard limit,” ask: Is this forbidden by the laws of nature, or just by our current knowledge?

  • Impact: Expands ambition and keeps innovation pipelines alive.

Problems as the Raw Material of Progress

  • Idea: Every solution creates new problems — and that’s a feature, not a bug.

  • Leadership relevance: Normalises the idea that “we fixed it” is never the end; it’s the start of the next improvement cycle.

  • Practical use: In debriefs, always ask: What new problems has this solution created?

  • Impact: Keeps the Adaptive Intelligence Cycle™ turning without waiting for a crisis.

Universal Explanations

  • Idea: Some explanations are so deep they can, in principle, explain anything in their domain (e.g., evolution by natural selection, quantum theory).

  • Leadership relevance: Seek strategic principles that are universal within your industry or market mechanics.

  • Practical use: Identify 2–3 “laws” of your business environment and test all strategies against them.

  • Impact: Reduces wasted effort on ideas that violate fundamental constraints.

Download the Card

This card will help you keep these ideas in mind.


DOWNLOAD THE CARD