At 8:47 AM on a Tuesday morning, Dr. Sarah Chen stands in the emergency room of Massachusetts General Hospital, facing three simultaneous decisions. In Trauma Bay 1, a motorcycle accident victim needs immediate surgery—delay could mean death, but the wrong surgical approach could also kill him. In Room 3, a child presents with unusual symptoms that could be anything from a rare genetic disorder to common flu. At the nurses' station, her phone buzzes with a text from her daughter's school: "Emma forgot her science project. Can you bring it?"
Each decision demands attention. Each has consequences. But treating them the same way would be disastrous. The trauma requires instant action based on pattern recognition. The diagnostic puzzle needs methodical analysis and possibly multiple consultations. The forgotten homework needs a quick cost-benefit calculation—is the disruption to her shift worth preventing her daughter's disappointment?
Dr. Chen navigates these decisions effortlessly, unconsciously matching each to an appropriate decision strategy. She doesn't analyze the trauma case or act impulsively on the diagnosis. She's developed what decision scientists call "meta-decision capability"—the ability to decide how to decide.
Most of us lack this capability. We apply the same decision-making approach regardless of context, like using a hammer for every household repair. We agonize over restaurant choices while making career moves on impulse. We research phone purchases exhaustively but choose retirement investments randomly. We treat reversible decisions like death sentences and irreversible ones like casual experiments.
This chapter develops a taxonomy of decisions—a classification system that matches decision types to optimal strategies. By understanding the fundamental dimensions along which decisions vary, we can choose appropriate tools for each situation. The goal isn't to make every decision perfectly but to stop using the wrong approach for the wrong situation.
Jeff Bezos popularized the most practical distinction in decision-making: Type 1 versus Type 2 decisions. Type 1 decisions are one-way doors—once you walk through, you can't come back. Type 2 decisions are two-way doors—if you don't like what you find, you can return¹.
This distinction seems obvious, yet we constantly miscategorize decisions. Marriage feels like a Type 1 decision, but divorce exists. Accepting a job feels irreversible, but people change jobs every 4.2 years on average². Starting a business feels like a one-way door, but most entrepreneurs start multiple ventures. Conversely, decisions that seem reversible often aren't. Posted social media content lives forever in screenshots. First impressions, once made, rarely fully reset. Time spent can never be recovered.
The reversibility dimension has profound implications for decision strategy:
Type 1 Decisions merit extensive analysis, multiple perspectives, and careful deliberation. These are the decisions where gathering more information, seeking expert input, and considering multiple scenarios pays off. The cost of error is high and correction is impossible or expensive. Examples include:
- Major surgical procedures
- Large irreversible investments
- Public statements on controversial topics
- Burning bridges with people or organizations
- Regulatory compliance decisions
Type 2 Decisions should be made quickly with minimal analysis. Since you can correct mistakes, the cost of delay usually exceeds the cost of error. Speed and experimentation beat precision. Examples include:
- Product feature priorities
- Marketing messages
- Hiring for probationary periods
- Restaurant choices
- Most daily operational decisions
The tragedy in organizations is that most decisions are Type 2, but we treat them like Type 1. Amazon combats this by pushing Type 2 decisions down to the lowest possible level and making them quickly. As Bezos writes, "Type 2 decisions can and should be made quickly by high judgment individuals or small groups. As organizations get larger, there seems to be a tendency to use the heavyweight Type 1 decision-making process on most decisions... The end result of this is slowness, unthoughtful risk aversion, failure to experiment sufficiently, and consequently diminished invention."³
But reversibility isn't binary—it's a spectrum. Consider these gradations:
Easily Reversible: Changes that can be undone with minimal cost or effort. Software features can be toggled off. Prices can be adjusted. Meetings can be rescheduled.
Reversible with Friction: Changes that can be undone but require effort, cost, or social capital. Returning a major purchase. Backing out of a verbal commitment. Changing strategic direction.
Partially Reversible: Decisions where some but not all consequences can be undone. You can sell a house but not recover transaction costs. You can leave a job but not erase the resume gap. You can apologize but not unsay words.
Practically Irreversible: Technically reversible but practically impossible. Rebuilding burned relationships. Recovering from reputational damage. Reversing major surgery.
Truly Irreversible: No possibility of undoing. Having children. Amputations. Criminal acts. Death.
Understanding where a decision falls on this spectrum determines how much time, effort, and resources to invest in making it.
In 1972, Stanford psychologist Walter Mischel offered four-year-old children a choice: one marshmallow now or two marshmallows in fifteen minutes. The "Marshmallow Test" became famous when follow-up studies found that children who waited scored higher on SATs, earned more money, and had lower rates of addiction decades later⁴.
But the experiment reveals something deeper about decision-making: the fundamental tension between immediate and delayed consequences. Every decision involves a time horizon—the period over which costs and benefits play out. This temporal dimension dramatically affects optimal strategy.
Immediate Decisions have consequences that manifest within minutes to hours:
- Emergency medical interventions
- Trading opportunities in volatile markets
- Customer service responses
- Crisis management
- Real-time negotiations
These decisions reward speed, intuition, and pattern recognition. Analysis is a luxury you can't afford. The best decision-makers in immediate contexts develop what Gary Klein calls "recognition-primed decision making"—the ability to quickly match situations to patterns and execute practiced responses⁵.
Short-term Decisions play out over days to weeks:
- Project task prioritization
- Inventory management
- Content publication schedules
- Travel planning
- Minor purchases
These decisions benefit from basic analysis but don't merit extensive research. Simple frameworks like pro/con lists, basic financial models, or quick experiments usually suffice.
Medium-term Decisions unfold over months to a few years:
- Job changes
- Major purchases (cars, appliances)
- Annual strategic planning
- Educational courses
- Relationship commitments
These decisions justify significant information gathering and analysis. The consequences are substantial but not life-defining. This is where most decision-making frameworks apply well.
Long-term Decisions have impacts spanning many years or decades:
- Career path choices
- Having children
- Retirement planning
- Major investments
- Educational degrees
These decisions involve fundamental uncertainty because the future context in which consequences play out is unknowable. Scenario planning, optionality preservation, and robust strategies become crucial.
Intergenerational Decisions affect not just you but future generations:
- Environmental policies
- Genetic modifications
- Institutional design
- Cultural preservation
- Dynastic wealth management
These decisions require considering values, ethics, and uncertainties that transcend individual experience. They demand humility about our ability to predict long-term consequences.
The time horizon affects not just how we should decide but what we should optimize for:
- Immediate decisions optimize for survival and loss prevention
- Short-term decisions optimize for efficiency and productivity
- Medium-term decisions optimize for goal achievement and growth
- Long-term decisions optimize for resilience and adaptability
- Intergenerational decisions optimize for sustainability and values preservation
A venture capitalist once told me, "I'd rather make a hundred small bets than one big bet, even if the expected value is the same. With a hundred bets, the law of large numbers is my friend. With one bet, I'm at the mercy of chance."
This captures a crucial but often overlooked dimension: decision frequency. How often you make similar decisions fundamentally changes the optimal approach.
One-shot Decisions occur once or rarely in a lifetime:
- Choosing a life partner
- Major surgery decisions
- Selling a company you founded
- Whistleblowing
- End-of-life choices
With one-shot decisions, you can't learn from experience or rely on averages. Each decision is unique, making statistical thinking less useful. These decisions reward extensive deliberation, scenario planning, and seeking wisdom from others who've faced similar choices.
Infrequent Decisions occur a handful of times:
- Buying houses
- Having children
- Major career pivots
- Divorce
- Starting businesses
With limited repetitions, you can learn between decisions but can't develop deep expertise. These decisions benefit from structured approaches, external advice, and careful documentation of lessons learned.
Regular Decisions occur monthly to yearly:
- Performance reviews
- Budget allocations
- Vacation planning
- Insurance renewals
- Investment rebalancing
Regular decisions allow for systematic improvement. You can develop templates, refine processes, and track outcomes. This is where decision journals and formal frameworks pay the highest dividends.
Frequent Decisions occur weekly to daily:
- Meeting scheduling
- Email responses
- Task prioritization
- Content curation
- Routine purchases
Frequent decisions benefit from automation, delegation, and simple heuristics. The cost of optimizing each decision exceeds the benefit. Better to develop good-enough rules and focus energy elsewhere.
Continuous Decisions are made constantly:
- Attention allocation
- Energy management
- Communication style
- Emotional regulation
- Habit execution
Continuous decisions shape life through accumulation. They're too numerous to make consciously, so the goal is programming good defaults through environment design, habit formation, and systematic practice.
The frequency dimension intersects with learning in powerful ways. High-frequency decisions in stable environments allow rapid skill development—what researchers call "kind learning environments." Low-frequency decisions in changing contexts prevent effective learning—"wicked learning environments"⁶.
This explains many expertise paradoxes. Firefighters develop remarkable intuition because they make similar decisions repeatedly with clear feedback. Psychiatrists predicting patient violence show no improvement over decades because they make such predictions rarely with delayed, ambiguous feedback⁷.
In 1962, during the Cuban Missile Crisis, President Kennedy faced decisions where mistakes meant nuclear war. Every option risked millions of lives. In the same year, he also decided what to serve at state dinners and which paintings to hang in the Oval Office. The stakes could not have been more different.
Stakes—what you stand to gain or lose—should determine how much resources to invest in a decision. Yet we routinely violate this principle, agonizing over trivial choices while rushing through consequential ones.
Micro-stakes Decisions have minimal impact:
- Coffee order choices
- Outfit selection (usually)
- TV show selection
- Restaurant appetizers
- Social media posts (usually)
These decisions deserve minimal time and energy. Satisficing—choosing the first acceptable option—is optimal. The cost of optimization exceeds any possible benefit.
Low-stakes Decisions have noticeable but minor impact:
- Small purchases (<1% of income)
- Routine work assignments
- Casual social commitments
- Entertainment choices
- Daily routes and routines
Low-stakes decisions benefit from simple heuristics and quick choices. Develop rules of thumb and stick to them.
Medium-stakes Decisions have significant but not life-altering impact:
- Major purchases (1-10% of annual income)
- Job project choices
- Important relationships
- Health interventions
- Skill investments
Medium-stakes decisions merit structured thinking and moderate research. This is where most decision frameworks apply well.
High-stakes Decisions can substantially alter life trajectory:
- Career defining choices
- Major financial commitments (>10% of net worth)
- Life partner selection
- Major medical decisions
- Legal strategies
High-stakes decisions justify extensive analysis, multiple opinions, and careful deliberation. The cost of error is substantial.
Existential-stakes Decisions affect survival or fundamental identity:
- Life-threatening medical choices
- Decisions affecting children's welfare
- Choices risking financial ruin
- Moral stands with severe consequences
- Decisions affecting many lives
Existential-stakes decisions demand maximum care, extensive consultation, and often, spiritual or philosophical reflection. These are the decisions where no amount of preparation feels like enough.
But stakes aren't always obvious. A casual comment can destroy a relationship. A small investment can compound into fortune. A minor habit can accumulate into life-defining pathology. Part of decision-making wisdom is recognizing when seemingly small decisions have large hidden stakes.
Some decisions suffer from too little information, others from too much. Understanding where a decision falls on the information spectrum determines whether to gather more data or act on what you have.
Information-Scarce Decisions lack crucial data:
- Unprecedented situations
- Emerging markets or technologies
- Decisions about other people's private thoughts
- Long-term predictions
- Creative or innovative choices
When information is truly scarce, gathering more has high value. But beware false scarcity—often we think information doesn't exist when we just haven't looked.
Information-Adequate Decisions have sufficient data for good choices:
- Routine operational decisions
- Well-understood domains
- Decisions with clear precedents
- Choices between comparable options
- Situations with reliable feedback
When information is adequate, the focus shifts from gathering to processing. Good frameworks and clear thinking matter more than additional data.
Information-Rich Decisions have abundant relevant data:
- Data-driven optimization problems
- Decisions with extensive historical precedent
- Choices in transparent markets
- Situations with measurable outcomes
- Domains with established science
Information-rich decisions reward analytical approaches, statistical thinking, and systematic methods. The challenge is synthesis, not collection.
Information-Overwhelming Decisions drown in data:
- Big data problems
- Decisions with infinite variables
- Choices in noisy environments
- Situations with conflicting expertise
- Domains with information warfare
When information overwhelms, the solution is filtering, not gathering. Focus on signal extraction, complexity reduction, and simple robust strategies.
No decision exists in a social vacuum. Even seemingly personal choices affect and are affected by others. The social dimension of a decision determines whose input to seek and how to manage social dynamics.
Individual Decisions primarily affect only you:
- Personal health choices (usually)
- Private financial decisions
- Individual skill development
- Personal belief formation
- Solo travel choices
Individual decisions benefit from self-knowledge and personal values clarification. Outside input is optional.
Dyadic Decisions involve one other person:
- Relationship negotiations
- One-on-one conflicts
- Bilateral agreements
- Mentoring relationships
- Partnership formations
Dyadic decisions require understanding the other party's perspective, finding mutual benefit, and managing interpersonal dynamics.
Small Group Decisions involve 3-8 people:
- Family choices
- Small team strategies
- Friend group activities
- Board decisions
- Founding teams
Small groups enable rich discussion but risk groupthink. Structured processes like devil's advocacy and rotating leadership improve outcomes.
Large Group Decisions involve dozens to hundreds:
- Organizational strategies
- Community initiatives
- Shareholder votes
- Union negotiations
- Political campaigns
Large groups require formal processes, clear communication, and often, representative democracy rather than direct participation.
Societal Decisions affect entire populations:
- Public policy
- Environmental regulations
- Public health measures
- Educational systems
- Cultural norms
Societal decisions involve irreducible value conflicts, massive uncertainty, and complex feedback loops. They require humility, extensive consultation, and robust processes for incorporating diverse views.
As we discussed in Chapter 1, different types of uncertainty require different approaches. The certainty dimension captures how much we know about probabilities and outcomes.
Deterministic Decisions have known outcomes:
- Mathematical calculations
- Physical laws
- Logical deductions
- Rule-based systems
- Algorithmic processes
When outcomes are deterministic, the decision reduces to computation. The challenge is calculation, not judgment.
Risk Decisions have known probabilities:
- Casino games
- Insurance pricing
- Statistical sampling
- Weather-dependent choices
- Actuarial calculations
Risk decisions reward statistical thinking, expected value calculations, and portfolio approaches.
Uncertainty Decisions have unknown probabilities:
- New market entries
- Innovation investments
- Relationship formations
- Career changes
- Medical treatments for rare conditions
Uncertainty decisions benefit from scenario planning, robust strategies, and preservation of optionality.
Ambiguity Decisions have unknown outcomes and probabilities:
- Paradigm shifts
- Black swan events
- Revolutionary technologies
- Unprecedented situations
- Creative endeavors
Ambiguity decisions require imagination, experimentation, and comfort with not knowing.
Ignorance Decisions involve unknown unknowns:
- Decisions affecting complex systems
- Long-term consequences
- Unintended effects
- Emergence and evolution
- Consciousness and meaning
Ignorance decisions demand humility, reversibility when possible, and respect for the limits of knowledge.
Real decisions don't fall cleanly along one dimension—they have profiles across all dimensions. Consider these examples:
Emergency Surgery: Irreversible, immediate timeline, one-shot, high-stakes, information-scarce, dyadic (patient-doctor), uncertainty. Optimal approach: Rapid pattern recognition by experienced practitioner with patient input when possible.
Startup Pivot: Partially reversible, medium-term, infrequent, high-stakes, information-adequate, small group, ambiguity. Optimal approach: Structured experimentation with clear hypotheses and kill criteria.
Retirement Investment: Irreversible contributions, long-term, regular, high-stakes, information-rich, individual, risk. Optimal approach: Statistical optimization with periodic rebalancing based on life changes.
Daily Commute: Reversible, immediate, continuous, micro-stakes, information-rich, individual, risk. Optimal approach: Automated habit with occasional optimization.
Marriage Proposal: Socially irreversible, long-term, one-shot, high-stakes, information-adequate, dyadic, uncertainty. Optimal approach: Deep reflection on values and compatibility with trusted advisor input.
By profiling decisions across dimensions, we can match them to appropriate strategies rather than applying one-size-fits-all approaches.
The dimensions don't operate independently—they interact in important ways:
Reversibility × Stakes: High-stakes reversible decisions are rare but valuable. They allow bold experimentation with limited downside—the essence of optionality.
Frequency × Time Horizon: High-frequency, long-term decisions (like daily exercise) have massive cumulative impact. Small improvements compound dramatically.
Information × Certainty: Information-rich uncertainty is dangerous—lots of data creates false confidence about fundamentally unknowable outcomes.
Social × Stakes: High-stakes decisions affecting many people require extensive process safeguards to prevent both groupthink and analysis paralysis.
Time × Reversibility: Immediate irreversible decisions are the most dangerous—no time to think and no ability to correct. These require pre-commitment and standard operating procedures.
Just as financial advisors recommend portfolio diversification, we should diversify our decision approaches. Most people have a default decision style they apply everywhere:
- Analysts overthink everything, from career moves to lunch orders
- Impulsives act quickly on everything, from marriages to investments
- Avoiders delay everything, from difficult conversations to opportunities
- Consensus-seekers crowdsource everything, from personal values to creative choices
- Maximizers optimize everything, from email responses to life partners
The goal isn't to abandon your natural style but to develop range. Like a golfer who needs different clubs for different shots, you need different decision approaches for different situations.
This requires:
- Recognition: Quickly categorizing decisions along key dimensions
- Repertoire: Having multiple decision approaches available
- Matching: Selecting appropriate approaches for each situation
- Execution: Implementing the chosen approach effectively
- Learning: Improving both recognition and execution over time
Based on our taxonomy, here are specific frameworks matched to decision profiles:
How will I feel about this in 10 minutes, 10 months, 10 years? If the answer is "I won't care," decide quickly and move on⁸.
Assume your preferred option is unavailable. What would you do? This often reveals fixation on one path and surfaces creative alternatives⁹.
Create explicit rules based on past outcomes. Example: "Always accept meetings with people I haven't talked to in >6 months, decline recurring meetings without clear agendas."
Make small reversible moves to gather information, then adjust based on response. Think dating before marriage, internships before careers¹⁰.
Gather independent opinions before group discussion to avoid anchoring and conformity. Aggregate, share, discuss, then repeat¹¹.
Jeff Bezos used this to decide to start Amazon: "When I'm 80, will I regret not trying this?" Minimizing future regret often clarifies present choices¹².
Imagine the decision led to failure. Work backward to identify what went wrong. This surfaces risks that optimism bias usually hides¹³.
Before making any significant decision, ask:
- What type of decision is this? (Profile it across dimensions)
- What approach fits this type? (Match to appropriate strategy)
- What resources should I invest? (Time, money, attention, social capital)
- Who should be involved? (Individual, advisors, group)
- When must I decide? (Real vs. artificial deadlines)
- How will I know if I was right? (Feedback mechanisms and learning)
This meta-decision—deciding how to decide—is often more important than the decision itself. A good process with mediocre judgment beats brilliant judgment with a bad process.
Understanding decision types helps avoid common errors:
Treating Reversible as Irreversible: Creates paralysis and missed opportunities. Most decisions are more reversible than they appear.
Treating Irreversible as Reversible: "We'll figure it out later" works for software features, not for reputation or trust.
Optimizing Low-Stakes Decisions: Spending hours researching the best $20 purchase while making five-figure investments casually.
Rushing High-Stakes Decisions: Artificial urgency causes real mistakes. Few decisions are as urgent as they seem.
Solo Deciding Social Decisions: Making unilateral choices that affect others breeds resentment and misses valuable input.
Crowdsourcing Individual Decisions: Seeking consensus on personal values or creative vision leads to mediocrity.
Analyzing Intuitive Decisions: Some domains reward gut feelings. Over-analysis can worsen outcomes.
Intuiting Analytical Decisions: Other domains punish intuition. Feelings don't predict statistical outcomes.
Like any skill, recognizing decision types improves with practice. Try these exercises:
List 10 recent decisions. Profile each across all dimensions. Notice patterns in your decision types and approaches.
Identify three decisions you regret. Did you mismatch the decision type and approach? What approach would have fit better?
Before making a decision, predict:
- Which dimensions matter most
- What approach you'll use
- How long it will take
- How you'll feel afterward
Check your predictions afterward. Calibrate over time.
Identify your default decision style. For one week, deliberately use different approaches:
- If you're analytical, make three quick intuitive decisions
- If you're impulsive, analyze three decisions thoroughly
- If you're a solo decider, seek input on three decisions
- If you're consensus-driven, make three decisions alone
Keep a decision journal with:
- Decision description
- Type profile (dimensions)
- Approach used
- Time invested
- Outcome
- Lessons learned
Review monthly to identify patterns and improve type-matching.
Every decision is unique, but decision types repeat. By recognizing patterns across dimensions—reversibility, time horizon, frequency, stakes, information, social context, and certainty—we can match approaches to situations.
The emergency room doctor who opened this chapter succeeds not because she makes perfect decisions but because she makes appropriate decisions. She doesn't analyze the trauma or rush the diagnosis. She matches her approach to the decision type.
This taxonomic thinking extends beyond individual decisions to life strategy. Some people optimize for reversibility, keeping options open at all costs. Others optimize for stakes, focusing only on big decisions. Still others optimize for frequency, building systems for repeated choices. Understanding your current optimization and consciously choosing your focus is itself a meta-decision.
In the next chapter, we'll dive deep into one of the most powerful tools for navigating uncertainty: probabilistic thinking. While we can't eliminate uncertainty, we can quantify it, reason about it, and make peace with it. The shift from binary to probabilistic thinking—from "will happen/won't happen" to probability distributions—is perhaps the single most important upgrade to human decision-making.
The taxonomy tells us what type of decision we face. Probabilistic thinking tells us how to reason about uncertainty within that type. Together, they form the foundation of decision-making under uncertainty.
¹ Bezos, J. (2016). Letter to Amazon Shareholders. Amazon.com Annual Report.
² Bureau of Labor Statistics. (2020). "Employee Tenure in 2020." U.S. Department of Labor.
³ Bezos, J. (2016). Letter to Amazon Shareholders. Amazon.com Annual Report.
⁴ Mischel, W., Ebbesen, E. B., & Raskoff Zeiss, A. (1972). "Cognitive and attentional mechanisms in delay of gratification." Journal of Personality and Social Psychology, 21(2), 204-218.
⁵ Klein, G. (1998). Sources of Power: How People Make Decisions. Cambridge, MA: MIT Press.
⁶ Hogarth, R. M. (2001). Educating Intuition. Chicago: University of Chicago Press.
⁷ Dawes, R. M., Faust, D., & Meehl, P. E. (1989). "Clinical versus actuarial judgment." Science, 243(4899), 1668-1674.
⁸ Welch, S. (2009). 10-10-10: A Life-Transforming Idea. New York: Scribner.
⁹ Heath, C., & Heath, D. (2013). Decisive: How to Make Better Choices in Life and Work. New York: Crown Business.
¹⁰ Snowden, D. J., & Boone, M. E. (2007). "A leader's framework for decision making." Harvard Business Review, 85(11), 68-76.
¹¹ Dalkey, N., & Helmer, O. (1963). "An experimental application of the Delphi method to the use of experts." Management Science, 9(3), 458-467.
¹² Bezos, J. (2008). Interview with Academy of Achievement. Washington, D.C.
¹³ Klein, G. (2007). "Performing a project premortem." Harvard Business Review, 85(9), 18-19.