Decision making

Important book: Choose Wisely: Rationality, Ethics, and the Art of Decision‑Making by Barry Schwartz and Richard Scheffler.

1. The Specification Problem

The first mistake is what the authors call the specification problem. The decision‑making framework they critique—rational choice theory (RCT)—is the most widely used model in economics, political science, and many other fields. Its influence has crept into your life whether you realize it or not.

RCT says that when you face a decision, you list your options, assign numerical values to the outcomes, estimate the probability of each outcome, multiply the two together, and pick the option with the highest expected utility.

But the authors argue—and I’ve argued elsewhere in my own work on value theory—that this framework is the wrong tool for almost every real decision you will ever face.

To illustrate this, imagine two scenarios. First, you wake up with a completely free day. What should you do? Second, imagine a high‑school student deciding which college to attend.

In both cases, RCT tells you how to choose between options, but it says nothing about how to generate those options. It assumes you already have a well‑defined preference set. But where does that set come from?

The student doesn’t wake up with a list of ten colleges and clear preferences. He has to decide what even counts as a relevant option—what kind of life he wants, what he values in an education, and who he wants to become. RCT gives him none of that.

And for your free day, the problem is even more obvious. The list of possible activities is infinite. The only thing that narrows it down is your own judgment—something RCT neither provides nor acknowledges as necessary. Before the calculation even begins, the most important decision has already been made without help from the framework.

2. The Quantification Problem

Suppose you push past the specification problem and list your options. RCT then tells you to assign numerical values to your preferences. But how?

Do you prefer the beach to the museum? Fine. But do you prefer it 40% more? Not 39%, not 41%—exactly 40%. That’s what the equation requires. Any number you choose feels arbitrary.

Consider the college student again. He rates the academic program at College A as an 8 out of 10 and the social life as an 8 out of 10. But are those two eights measuring the same thing? Almost certainly not. They may be incommensurable values—values that share no common metric. Adding them up is not rational; it’s mathematical theater that hides the complexity of the decision.

3. The Interaction Problem

Even if you could assign accurate numbers, you still can’t simply add them up because utilities interact.

A family deciding between the beach and the museum illustrates this. The father prefers the museum. So the utility of going to the beach isn’t just the child’s joy—it’s the child’s joy minus the father’s reluctance, minus the tension that reluctance creates between the parents, which then affects the child’s enjoyment.

For the college student, the utility of attending College A depends on whether her best friend gets in, whether the distance affects her relationship, whether her parents approve, and so on. These factors interact in ways that can’t be pre‑specified. RCT has no tools for handling this.

4. Radical Uncertainty

The fourth mistake is radical uncertainty. Standard decision theory assumes that even if you don’t know what will happen, you can assign probabilities to outcomes. But many real‑world decisions don’t allow this.

Economist Frank Knight distinguished between probabilistic risk and radical uncertainty—situations where you cannot meaningfully assign probabilities. Daniel Ellsberg illustrated this with his famous urn experiment: people prefer known probabilities over unknown ones, even when the actual odds may be identical.

Now ask yourself: what is the probability that attending College A leads to a fulfilling life? Not a successful career—a fulfilling life. There is no honest answer. You’re not drawing from a well‑specified container. You’re making a choice whose consequences ripple across decades and across a self whose preferences will change in unpredictable ways. RCT simply does not apply.

5. The Index Problem

The fifth mistake is the index problem, which often has the darkest consequences. When you try to measure something important, you often end up optimizing for the measure rather than the thing itself.

The U.S. military in Vietnam used body counts as an index of success. Body counts went up. The U.S. still lost the war.

College rankings use graduation rates, faculty publications, and endowment size as proxies for educational quality. Schools optimize for those proxies. Rankings go up. Education quality does not necessarily improve.

So when a student consults rankings, he’s not seeing where he’ll thrive. He’s seeing how well institutions have learned to game the indices.

6. The Framing Leakage Problem

The sixth mistake is the framing leakage problem. RCT assumes your preferences exist independently of how a decision is framed. But framing doesn’t just affect the decision—it affects the experience of the outcome.

In one study, beef labeled “75% lean” was rated as tasting better than beef labeled “25% fat,” even though it was identical. The label changed the experience.

For the college student, if she frames her choice as “the school I settled for,” she will experience it as settling. The story you tell about your decision becomes part of the decision itself.

Alternative

After dismantling rational choice theory, Schwartz and Scheffler propose an alternative rooted in Aristotle’s concept of the golden mean. The goal is not to maximize expected utility but to exercise judgment guided by your values to find what is good enough across the dimensions that matter.

They offer five guiding questions:

1. What kind of person am I trying to become? This shifts the focus from rules to virtues and reframes the entire decision.
2. How does this choice fit into the story of my life? People measure their lives narratively, not quantitatively.
3. How am I framing this decision, and does that frame reflect what I value? Framing is a tool, not a bias.
4. What is genuinely good enough here, rather than objectively best? Maximization often makes people worse off.
5. Am I letting calculation substitute for judgment? Overreliance on spreadsheets and rankings atrophies your capacity for genuine judgment.