“When I am in the car listening to the radio, I often check other stations to see if something better is playing, even if I am relatively satisfied with what I’m listening to.”
In November 2001, 548 graduating students at 11 American colleges and universities rated that sentence from 1 to 9. They had just started looking for jobs. The researchers, Sheena Iyengar and Rachael Wells of Columbia Business School and Barry Schwartz of Swarthmore College, used it and ten other items to sort them. High scorers were “maximizers,” who look for the best. Low scorers were “satisficers,” who look for good enough.
The maximizers searched harder. They expected to send out more applications and leaned more on rankings, career offices, family and comparisons with peers. In February, they agreed more strongly that they often fantasized about jobs quite different from the ones they were chasing.
By May, the two groups had different pay. Split at the median, the maximizers’ accepted offers averaged $44,515 a year. The satisficers’ averaged $37,085, a gap that held after controlling for university rank, grades and major.
Then the researchers asked both groups how they felt about the jobs they had taken.
Better offers, lower ratings
The maximizers felt worse. Asked how satisfied they were with the offer and how confident they were in their choice, they averaged 7.02 on a 9-point scale. Satisficers averaged 7.58. At every survey, the maximizers also reported feeling more pessimistic, stressed, anxious and overwhelmed. They wished more strongly that they had pursued still more options.
The authors’ explanation was the options themselves. Hunting for the best kept more of them in view, raised expectations, and left more jobs behind to wonder about. Statistically, the satisfaction gap was accounted for by two things: dwelling on the jobs not taken, and wishing they had looked further. Their paper’s own picture was a channel-surfer facing 400 cable channels, who spends so long deciding that little time is left to watch. The satisficer surfs until the first acceptable show, puts down the remote, and watches it.
Decide what good enough means, take the first option that clears it, and stop looking.
The word came from Herbert Simon, who won the 1978 Nobel Prize in economics. In two papers in 1955 and 1956, he described a chooser who forms an aspiration for how good an option should be. “As soon as he discovered an alternative for choice meeting his level of aspiration, he would terminate the search and choose that alternative,” Simon recalled in his Nobel lecture. “I called this mode of selection satisficing.” Simon offered it as an account of how choice can actually be made “with reasonable amounts of calculation, and using very incomplete information,” when nobody hands you the options in advance. Schwartz’s scale later turned it into a difference between people. In Simon’s account the aspiration moves, too. “In a benign environment that provides many good alternatives, aspirations rise; in a harsher environment, they fall.”
What the study can’t tell you
The job study found associations among the students who kept answering, not causes. Its authors noted that salary was only one measure of a good outcome. And the scale itself has been questioned. In 2008, researchers at Bowling Green State University and Denison Consulting argued that Schwartz’s Maximization Scale also picks up indecision and emotional instability. On a new scale they built and tested with 191 people, they found that maximizers, aside from being more prone to regret, “are just as happy as satisficers.”
In this study, the higher pay was a real difference too: about $7,430 a year. That doesn’t mean the harder search itself caused the gap. Some choices deserve a long search, especially the ones that are hard to undo, the one-way doors.
The software trial that never ends
Imagine you run a twelve-person design studio and need new bookkeeping software. Before you look, you write the bar on one line: it must handle two currencies, export in your accountant’s format, and cost under a set monthly price. The third product you trial meets all of it. You switch, and invoices go out on time that month.
Now imagine you skip the line. You trial nine products over six weeks, and each one is better at something. While you compare, your invoices go out late. You pick the one with the best reports, and spend the next quarter wondering about the one with the better app.
Picking the best of a thousand
People satisfice because they can’t check everything. An AI system can generate candidate answers far faster than anyone could read them, so the old reason to stop looking seems to vanish. A 2022 experiment by three OpenAI researchers, Leo Gao, John Schulman and Jacob Hilton, suggests a different reason.
They had a model write many answers and used a scoring model to pick the highest-scoring one, a method called best-of-n. The scorer was trained to imitate a larger “gold” scoring model, which stood in for human judgment in this synthetic setup. As n grew, the scorer’s grade for its pick kept rising. The gold score rose at first, then fell. In the authors’ words, because the scorer “is an imperfect proxy, optimizing its value too much can hinder ground truth performance.”
That was one lab’s setup, with one model standing in for people. You can run a small version on your own work. Before you ask a chatbot for twenty taglines, write down what a good one must do. Then read them in order and stop at the first that passes. If you also ask the AI to pick its favorite, check its winner against your same line.
Sometimes the winner is the one you would have stopped at. Sometimes it’s the twentieth.