Abundance Hurts Innovation When Leaders Cannot Decide What Not to Do

David Epstein’s central claim is that constraints can make people and organizations more creative, not less. The important qualifier is that the constraint still has to leave room for agency and surprise. A limit that forces a team to clarify priorities or search in unfamiliar places can be useful. A limit that makes success impossible has simply gone too far.

Epstein illustrates the distinction with NASA’s LCROSS mission, a story he says came from Ed Hoffman, NASA’s former Chief Knowledge Officer. The engineers, Epstein says, found themselves with “half the time and half the money that they expected.” Their first response was ordinary: they complained. Then they reframed the problem: “If we were going to get this done anyway, how would we do it?”

That question changed the solution space. They could not build everything from scratch, so they borrowed. According to Epstein, the team used imaging equipment from Army tanks and engine temperature sensors from NASCAR to build a probe that confirmed water on the moon. The source showed NASA-attributed LCROSS simulation imagery during this example: a 3D rendering of the satellite over Earth and, later, the spacecraft shown in NASA’s “Eyes on the Solar System” interface alongside other missions.

The point was not that every budget cut is good. Epstein explicitly stops short of that. If the team had had a quarter of the time and money rather than half, he says, he does not know whether the mission would have been possible. The useful line, in his formulation, is whether people can still plausibly surprise themselves.

Jeff Berman presses Epstein on whether there is a difference between constraints imposed from outside and constraints leaders create for their teams. Epstein’s answer is partly yes and partly no. Functionally, both kinds of constraints can do two valuable things: force a clarification of priorities and launch people into “productive exploration.” Emotionally, however, externally imposed constraints can leave people feeling that something is being done to them. Epstein hopes the argument in Inside the Box can help reframe some of those situations as what psychologists call a “desirable difficulty”: something hard that may still bring out better performance.

David Epstein’s interest in constraints began partly as a self-diagnosis. After Range, he was independent as a writer and decided he would not write another book unless he found the perfect topic. Because his interests were wide, he kept exploring possibilities without choosing one. A quote from Mihaly Csikszentmihalyi, the researcher who coined the term “flow,” helped him see the problem. Csikszentmihalyi was talking about marriage, but Epstein applied it to work: when you commit by choice, you can stop wondering how to live and start spending your energy living.

Epstein realized he was doing the opposite with book topics. He was fascinated by constraints, but he kept asking what else might be around the corner. He decided to write a proposal on constraints the next day. Two weeks later, he says, he was “10 times as interested in it.” His problem was not a lack of ideas. It was deciding what to execute.

That personal problem led into a business example Epstein treats as emblematic: General Magic. He first encountered the company through a documentary and was struck by footage of an organization that had extraordinary talent, money, and ambition. As Epstein describes it, General Magic was “basically making the iPhone, you know, 20 years too early.” Goldman Sachs took it public as what he calls the first “concept IPO” in Silicon Valley history: an offering based on an idea rather than a finished product.

Resources poured in. Talent poured in. The company could do anything, and, in Epstein’s telling, that became the problem. “Every good idea they had someone built it,” he says. The product became incoherent, and users did not know what to do with it. When he interviewed former employees, the refrain was that they could not figure out “what not to do.”

That phrase becomes one of the most important managerial claims in the discussion. In abstract models of rational decision-making, Epstein says, more choice, more opportunity, more resources, and more talent always look better. But research into how people actually decide, focus, and find satisfaction points in the other direction. Too much can leave people unsatisfied and adrift. It can make organizations unable to concentrate their effort.

Epstein’s broader warning is that it has “never been easier to do too much” than it is now. That applies to companies, but also to individuals. General Magic is not presented as a failure of ambition. It is presented as a failure to constrain ambition into a usable form.

David Epstein contrasts General Magic with Pixar because both were pursuing audacious goals at roughly the same time. Pixar was trying to make the world’s first fully computer-animated feature film. It is often imagined as a place of unfettered imagination, but Epstein argues that its creative output depended on rules, guardrails, and visible constraints.

One example is Pixar’s “three pitches rule.” Epstein says Pixar observed that directors pitching stories tended to attach too quickly to their first idea, even though it was usually not their best. He connects this to the “creative cliff illusion”: the belief that the best idea comes first or not at all. In practice, he says, the first idea is often merely the convenient one.

The rule forced people to bring three ideas. That made it harder to over-identify with the first and gave the organization a structure for discovering better options. Epstein’s interpretation is that Pixar had unusually good insight into human psychology and designed rules to “save people from themselves.”

The same logic appears in what Pixar co-founder Ed Catmull called the “beautifully shaded penny problem.” A director might fixate on a tiny background detail, such as the shading on a penny, that the audience might never notice. Animators would continue working on it while attention and labor were diverted from more consequential parts of the film.

Pixar’s solution, as Epstein describes it, was physical and simple: popsicle sticks Velcroed to a board. Each stick represented the amount of work one animator could do in one week. If a director wanted more work on the penny, sticks had to be removed from other characters or priorities. The issue was not whether the penny could be improved. It was what else would receive less attention if the penny consumed more work. The managerial lesson Epstein draws is direct: visible capacity constraints convert taste debates into resource-allocation decisions.

The current AI moment might seem to weaken the old startup warning that “more startups die of indigestion than starvation,” a quote Jeff Berman attributes to Bill Gurley. If AI agents can do much of the work, Berman asks, should leaders rethink the tradeoff between taking on too much and doing too little?

David Epstein’s answer is not anti-AI. He says AI has cut some tasks in his own work from ten hours to one. He expects smaller organizations to do things that once would have required much larger teams. But he also sees a new version of the same old problem: an “infinite ability now to start a million things that we’re not going to finish.”

Over the prior year, Epstein says, he spent time with an AI company that helps other organizations implement AI. His observation from that experience was that many companies were implementing AI in a sprawling way. They felt they had to move quickly because everyone else was moving quickly, but the work did not always connect clearly to strategy. He says researchers are beginning to call this “work slop”: a high volume of output whose relationship to actual priorities is unclear.

The organizations that seemed more successful, in Epstein’s view, began by defining the problem well. Only then did they ask what tool matched the problem. Some mapped the “jobs to be done”: what work actually needed doing, which parts a tool could perform, and which humans might be freed to think more strategically.

For Epstein, AI strengthens rather than weakens the need for constraints. If tools make it easier to generate work, leaders need stronger ways to decide what work matters. Problem definition becomes a guardrail against motion that looks like progress.

Epstein also points to organizational guardrails around AI use. Employees will use these tools whether they are officially allowed to or not, he says, and using public models can create risks for organizations. His suggested posture is not blanket prohibition. Leaders should create the playing field, set guardrails, and then invite experimentation: “Please experiment and let the flowers bloom,” but within a defined space rather than a disjointed set of individual practices.

Asked what business leaders can learn from scientists, David Epstein focuses less on technical expertise than on how scientists should, at their best, handle assumptions. He describes research from Inside the Box in which founders received different kinds of training in market research and value proposition evaluation. Some, in Epstein’s account, were trained in a version of the scientific method: state a hypothesis about the value they thought they provided, commit in advance to a way of testing it, and define a decision rule for what would count as being right or wrong.

Those founders, Epstein says, often discovered that an assumption about their value proposition was wrong. They had misread the market or their customers. Because the prediction and test were specified in advance, they were more likely to pivot and more likely to succeed.

By contrast, Epstein says founders who received more standard training were likelier to retrofit whatever they heard back into the story they already believed. They pivoted less often. The difference was not that one group heard feedback and the other did not. It was that one group had created a constraint on interpretation before the feedback arrived.

Epstein is careful not to idealize scientists. He says scientists themselves need to do better at prospective prediction, and he links some failures of replication to cases where researchers have not made predictions in advance and stuck to them. But the business application he draws is direct: before trying something with a team, predict what you think will happen. Write it down. When the result comes in, the record makes it harder to say, after the fact, “I basically got that right.”

That habit also connects back to Range, where Epstein wrote about people who improve at forecasting. One of their most important habits, he says, is recording predictions. Without a record, people unconsciously revise their stories. With a record, they can update their model of the world, the market, or the team.

Jeff Berman frames Range as a data-backed case for diversifying how people spend their time, especially in a culture that pushes children and workers toward early skill-building and specialization. David Epstein extends that argument with what he describes as a paper in Science aggregating roughly 30,000 careers across musicians, scientists, and athletes.

According to Epstein, the paper supported the thesis of Range: indicators of elite youth performance were “basically negatively associated” with indicators of elite adult performance. You can optimize for the short term and produce the “best kid,” he says, but that often undermines long-term development. The claim applies, in his account, to individual skills, career selection, and other domains, and he argues it becomes even more important when the world changes quickly and people cannot assume next year will look like last year.

The point is not that expertise is unimportant. It is that the path to high adult performance may include a period of sampling, general skill development, and delayed specialization. Epstein later returns to this through his debate with Malcolm Gladwell over the so-called 10,000 hours rule.

In his first book, Epstein criticized the research underlying that rule and Gladwell’s popularization of it. The question was not whether elite performers practice a great deal. Epstein says elite athletes do spend more time in deliberate practice than lower-level athletes. But studies tracking athletes across development show, in his telling, that early on, future elites spent less time in deliberate practice in their eventual sport than peers who plateaued at lower levels. They had what scientists call a sampling period: a broad variety of activities that developed general skills, helped them learn their interests and abilities, and delayed specialization.

Epstein and Gladwell came to call this the “Roger versus Tiger problem”: Tiger Woods as the early specialist, Roger Federer as the early diversifier. Epstein says that framing grew out of their debates and became the opening of Range.

The discussion of Gladwell becomes a broader argument about disagreement. David Epstein says Gladwell could have “crushed” him in their initial debate. Epstein was not especially well known, while Gladwell was one of the most famous nonfiction writers in the world. Instead, after the debate, Gladwell acknowledged that Epstein’s evidence did not fit what he had thought and suggested they run together in New York, since both had been national-level middle-distance runners.

They began running together on weekends. Epstein says Gladwell later told him, “I have the luxury of learning from my critics.” That line stayed with Epstein. It became a model for how to treat an earnest critic: not as someone to defeat, but as someone who may have found something real.

Jeff Berman connects that to a practice in his own marriage and work life: when someone offers constructive criticism, the first response is, “Thank you for telling me.” He says the phrase helps suppress defensiveness and creates a more normal environment for thoughtful criticism.

Epstein links that practice to Wernher von Braun’s leadership of the Saturn V rocket program. According to Epstein, von Braun’s engineering leads circulated their questions each week on one sheet of paper. Von Braun would write notes in the margins and recirculate them. Sometimes he would write “congrats” when someone found a major anomaly, because he wanted people to see him thanking those who surfaced problems.

Epstein also applies the lesson to himself. Adam Grant once told him he was a “defensive pessimist”: someone who gets excited about an idea, becomes convinced it will fail, and then works hard to prevent failure. Epstein says the description stung because he thinks of himself as optimistic about the world and human capacity. But he concluded Grant was right about his own projects, and the criticism became useful.

The condition, for Epstein, is that criticism has to come from an earnest critic and that people need enough relationship to respect one another as human beings. “If you’re only social media posting at each other,” he says, “that’s not realistic.”

Asked what he would tell political leaders in a divided country, David Epstein starts skeptically: they would have to want to fix it. He says the political landscape has in many ways become an extension of influencer culture, where incentives often reward making people angry or offering shortcut solutions that are not true.

His more constructive answer is that leaders need more real contact across divides. Epstein says people he knows who used to be in Congress have told him that members once spent more time in Washington, D.C. Over time, political incentives pushed them to spend more time in their districts. That may sound closer to constituents, but Epstein argues it also means members’ children are not in the same schools, they are not on the same baseball teams, and politicians do not form relationships across the aisle. They do not work together because they do not know one another.

Jeff Berman reinforces the point with his own Capitol Hill experience. A little over 20 years earlier, he says, he worked for a member of the Senate Democratic leadership while his counterpart on a subcommittee was Jeff Sessions’s chief counsel, Ed Haden. They had lunch every month in the Senate cafeteria and were criticized by people on both sides for doing it. Berman says he came to know Haden as bright, patriotic, and committed to the country, even though they saw the world very differently.

Epstein broadens the point beyond politics. One chapter of Inside the Box, he says, discusses social norms as constraints on human behavior that make strangers more predictable to one another. In Epstein’s telling, economic history shows those norms helped people trust strangers enough to collaborate beyond kinship networks. That trust preceded much of the technological innovation that led to shared prosperity.

When norms erode, Epstein says, people trust strangers less. He says the proportion of people who report trusting strangers is tightly related to per capita GDP at the national level. He also cites a Pew survey that, in his account, showed America as the only country where a small majority of adults said other people have bad morals. To him, that is a bad sign.

The political lesson, as Epstein frames it, is not only about civility as etiquette. Public decorum is a constraint that protects the conditions for cooperation. Political leaders are role models whether they want to or not. If trust between strangers degrades, Epstein says, shared prosperity is at risk regardless of the policy agenda.