Introduction: Why We Make Bad Decisions
It’s an important truth often overlooked: the human brain is naturally bad at probabilities. Most of us default to deterministic thinking—focusing on a single outcome or “what do I do tomorrow”. But in a complex world, this approach guarantees error.
High-quality decision-making—in business, in life, and especially as we transition to a world driven by AI—requires shifting our mindset to probabilistic thinking. Instead of predicting one future, we must “build the graph of 30 years” to explore all potential outcomes.The Core Framework for High-Quality Decision Making
Richard Arnold, a long-time student of human judgment and reasoning (a field within cognitive psychology), outlines a powerful framework, inspired by physicist Subash Gupta, to move past our deterministic limitations:
1. Frame the Question Correctly
Framing a decision happens automatically, whether you think about it or not. It is critical to define the scope: Is this a big decision or a small one? Are you viewing it from your personal perspective or a global one?
2. Depositioning: Drop the Initial Idea
The first critical step is to “get off the idea that caused you to realize you have to make a decision”. If your initial thought is, “I want to go to Hawaii for vacation,” the decision should not be Should we go to Hawaii? but Where do I go for vacation?. Starting with the original thought too severely constrains the analysis.
3. Clarify Objectives and Constraints
Define the essential goals (e.g., have a great time, be warm, stop thinking about work). Arnold recommends focusing on three or four key objectives, not 15.
Next, define the constraints: budget, time limits, or travel logistics.
4. Identify Data Needs
Instead of asking “what data do I have to help me,” ask: “what data do I need to help me?”. This ensures you gather information necessary to evaluate a broad range of ideas.
5. Ideation
Generate broad, expansive ideas based on your objectives and constraints. Once the ideas are generated, you can refine the specific data needed to evaluate them (e.g., temperatures or flight prices for a new location).Case Study: From Christmas Cards to Customer Success Arnold shares a powerful example of how the framework can transform a simple decision:
When a company’s customer success team suggested sending Christmas cards, the decision was depositioned to “What are we really trying to accomplish?” The goal was to show the customer they were thinking of them.
This led to the implementation of Shama (inspired by the Jain sect’s practice). At the end of the year, the company would:
-
Identify harm caused: Review the relationship, acknowledge specific failings (e.g., slow bug fixes).
-
Apologize properly and resolve to fix: State what they are doing to fix the underlying process to prevent future harm.
This change, born from an analysis that moved beyond the “Christmas card” idea, became the way they ran customer success, resulting in a phenomenally effective process and leading to a billion-dollar global business.
The Gotchas: Human Frailties in Decision Making
Our deterministic nature leaves us vulnerable to specific cognitive biases, especially when new information is introduced:
-
The Monty Hall Problem: This classic riddle shows that after being given new information that changes the odds (Monty opening a door), people overwhelmingly stick to their original, deterministic decision even when switching would give them a two-thirds chance of winning.
-
Anchoring: Arbitrary starting points—or anchors—can “badly screw up human decision making”. In one experiment, people asked if the average July temperature in San Francisco was “more or less than 150 degrees” gave drastically higher estimates than those who were asked if it was “more or less than 60 degrees”.
We also tend to overestimate our ability to estimate. In a 10-question test where participants were asked to provide a range that had a 90% probability of containing the correct numerical answer (e.g., the length of the Nile or the unladen weight of a Boeing 747), the average person only gets four or five correct, demonstrating a profound lack of probabilistic awareness.
Conclusion: Aligning with the Future of AI
The world is moving toward Artificial Intelligence, which is inherently a probabilistic and stochastic engine. It is rapidly replacing human workers who currently make decisions within deterministic frameworks.
For instance, the sophisticated systems that determine hotel room pricing and availability globally do not know if the Pope is coming; they simply apply probabilistic mathematics to changes in demand and set prices better than a human revenue manager ever could.
To thrive in the future, we must build personal operating systems and organizational processes that acknowledge these human frailties and intentionally incorporate probabilistic frameworks and data-driven systems.
Watch the full episode here: https://dub.sh/iDaH1wg
