A Broken Arm Forced a Perfect French Score: Constraints Are the Engine of Creativity
After David Epstein broke his arm he was forced to use image-based memory instead of note-taking, and his French scores went to perfect. A two-day London Underground strike led many people to better commutes. Constraints aren't obstacles — they're the switch that forces you to experiment.
The video won't play here. Listen to the audio instead:
The argument · tap a timestamp to hear it
A broken arm forced a memory method into existence
In eighth grade, playing American football, David Epstein threw too hard and his arm snapped on the follow-through; his right arm was bound from shoulder to hand against his torso. His French class had a weekly dictation test: listen to a native speaker's recording and fill in the blanks in a text, and he was mediocre. With his right hand bound, he couldn't write fast enough with his left, so he had to hold the flying words in his head first, by linking each word to something sports-related — he didn't know at the time that this was called a mnemonic device. As a result he could remember every word, then slowly fill them in with his left hand, and he started getting perfect scores. The key mechanism: his most convenient path was blocked, forcing him to find an alternative, and the alternative was better.
— David EpsteinA two-day subway strike improved commutes for years
During a London Underground strike, people were forced to find different routes to work. The strike lasted only 48 hours, but a substantial share of people discovered a better commute than the one they had used for weeks, months or even years, and kept using it afterward. Epstein recalls the study saying this collectively saved about 1,500 hours of commuting per day. His inference: we normally experiment too little; useful constraints drive people to run experiments they otherwise wouldn't, and the payoff from those experiments can far exceed the inconvenience.
— David EpsteinGeneral Magic died of too many resources
General Magic was a late-80s, early-90s startup with nearly unlimited talent, money and resources, and it judged the direction of technology for the coming decades correctly — on paper, they invented the iPhone. But Epstein says more startups die of indigestion than of starvation: because they could do anything, they did everything, and projects kept ballooning out of control. They never defined clearly who the customer was (the so-called Joe Sixpack — nobody knew who that was), and in the end they were building for each other rather than solving a specific problem. Internally, a low-level service engineer built a prototype of eBay and handed it to the company, and it was rejected as "too small, it'll never amount to anything." Epstein argues they should have started from smaller goals and smaller milestones.
— David EpsteinA world without constraints is a fiction
Tim Harford raises a disagreement: General Magic wasn't without constraints, it's just that all its constraints were long-term, so they could pretend the constraints didn't exist until the money and time actually ran out. Epstein partly agrees, but corrects that they didn't actually lack money — they had more than they could spend — what they lacked was time. He cites Oliver Burkeman's view: believing that with the right productivity tools you can get everything done may just be a refusal to face your own mortality, a refusal to prioritize ruthlessly. About two-thirds of the former General Magic employees he interviewed said they simply couldn't figure out what not to do. Epstein treats this as a parable for the AI era: when the cost of starting something approaches zero, it erases the hard work of deciding what not to do, which may be one reason adoption has risen unprecedentedly while productivity hasn't risen correspondingly.
— David EpsteinThe Hemingway principle against urgency
Epstein calls his own writing method the Hemingway principle: Hemingway stopped mid-sentence, so that the next day he knew which important place to return to. The last thing Epstein does at the end of each workday is designate which important place to start from the next day. He says this guards against the two main ways a workday goes off the rails: first, falling into an algorithmic feed or inbox first thing in the morning; second, the so-called mere urgency effect — people prioritize what feels urgent (like answering email) over what's truly important, even when the important thing objectively deserves more time.
— David EpsteinPre-registration makes miracle drugs disappear
Before 2000, psychology was full of counterintuitive findings: thinking of Florida makes you walk more slowly, power poses raise testosterone, hurricanes with female names are more dangerous. Medicine was the same — NIH-funded cardiovascular drug trials almost all found positive results; after 2000, new trials were almost all negative. The turning point was the NIH forcing researchers to constrain their behavior: write down your prediction in advance and stick to it. Epstein explains the mechanism: researchers would originally make a prediction, and when it didn't hold, they'd dig through a large dataset for a correlation, and once they found one, retroactively invent a prediction, so that at publication it looked like they'd been looking for this from the start. Digging through data amounts to running nearly infinite tests, and each test raises the probability of a false positive, so you're almost guaranteed to find a false-positive association.
— David EpsteinEverything in the fridge both causes and prevents cancer
Epstein's favorite "dark humor paper" aggregated all nutrition research and found that almost every food you can think of has been shown in different studies to both cause and prevent cancer, with only bacon as the exception — it was only found to cause cancer. He also tells the case of nutrition star researcher Brian Woznick: he did the famous bottomless soup bowl experiment, where people ate far more when an unseen tube kept refilling their soup. Later he wrote a blog post describing his research method: when a prediction didn't hold, he'd hand the data to a graduate student and say "find me something positive," and she never failed. Epstein's analogy: it's like firing randomly at a wall, then finding where the bullet holes cluster and drawing a bullseye and a ten-ring there, and passersby marvel at the marksmanship. Most of Woznick's life's work was eventually retracted, and his research had been cited in a Nobel laureate's book.
— David EpsteinEntrepreneurs who write predictions fail more but succeed more
Epstein extends pre-registration to personal life: when making a behavior change (say, wanting to exercise more), first write down what you plan to try and predict what it will bring, then do it, then check whether the prediction held — this educates your intuition. He cites a randomized controlled study: entrepreneurs randomly assigned to a "scientific method training" group or an ordinary market research group; those trained falsified their own prior beliefs more often, discovering that the value they thought they were adding to the market didn't hold, and as a result they pivoted more often and succeeded more often.
— David EpsteinDr. Seuss was forced into rhythm by 50 words
Green Eggs and Ham came out of Dr. Seuss (Theodore Geisel) betting he could write a book using only 50 words. The constraint made it impossible for him to experiment with vocabulary, so he turned to rhythm. The earlier origin was that someone asked him to write a children's book using about 200 words from a children's vocabulary list; he looked through it and found very few adjectives, complained to his wife that "this is like making strudel without strudel," then threw up his hands and said he'd use the first two rhyming words on the list — cat and hat. This forced out his signature jaunty rhythm, because the conventional storytelling path was blocked. Cognitive scientist Daniel Willingham puts it this way: you might think the brain is for thinking, but actually it's for preventing you from thinking as much as possible, because thinking burns energy. Without constraints, we take the path of least resistance.
— David EpsteinTime pressure can destroy or make you
Epstein admits it's hard to judge what a good constraint is. Take time pressure alone: research shows it can both destroy and enhance creative problem-solving, depending on how people use it — if used to multitask and try to do everything at once, it destroys; if used to single-task and focus on the most important part of the task, it enhances. He also offers a rough criterion: if, facing a constraint, you can still ask "can I still surprise myself," it still has value; if you can't, it's definitely a bad constraint. Invention experiments confirm this too: give 100 parts to use freely and creativity is low; give 20 parts and ask for furniture and creativity rises; but require that it must be a chair and creativity falls again — because you've specified both what to make and how to make it.
— David EpsteinHalf a budget built a lunar impactor
Former NASA chief knowledge officer Ed Hoffman told Epstein about the LCROSS mission: the engineers were given only half the budget and time expected. They complained first, then a leader asked "if we had to make it work no matter what, how would we do it," which led them to borrow technology — taking imaging equipment straight off Army tanks and engine temperature sensors off NASCAR race cars — and they ultimately built a probe that confirmed there is water on the Moon. Epstein says that without the halved time and budget, they would never have thought that way. But he also draws a boundary: if only a quarter of the time and budget were left, he doesn't think they could have done it. So the key is "titration" — finding the enabling constraint that drives productive experimentation, rather than one that makes the task impossible.
— David EpsteinInfinite choice makes people more bored
Harford mentions the conclusion from his own research on online dating: seemingly infinite partner choice isn't beneficial; people constantly second-guess themselves — "is this the right one, I can always go back and find someone else" — and end up lonelier. Epstein says this is an analogy for everything today: the brain is to some degree a comparison engine, and now you can compare every aspect of your life against the entire internet, which is a recipe for dissatisfaction. He cites a Haruki Murakami short story: a man and a woman discover they are a 100% perfect match for each other, but doubting "can happiness that comes this easily really be okay," they agree to part and to marry if they meet again by fate — and they never meet again. He also mentions an experiment: people were randomly assigned to choose from 20 videos, or to watch only 1 designated one; those who could watch only 1 were less bored — the mere thought that "you might be doing something else better" spoils the present experience.
— David EpsteinIn their own words · checked verbatim
I threw a ball as hard as I possibly could. And on the follow through of the throw, my arm snapped.
David Epstein5:28
only once it was so desperate that I couldn't do it at all was I forced to look for another way. So it was very much necessity being the mother of invention.
David Epstein6:30
more startups die of indigestion than starvation.
David Epstein10:38
I just couldn't figure out what not to do.
David Epstein13:42
they had to make a plan ahead of time that they had to stick to
David Epstein18:44
You may think that your brain is made for thinking, but it's actually made for preventing you from having to think whenever possible because thinking is energetically costly.
David Epstein30:53
if you have a constraint and to say, can I still surprise myself? Then there's still value there. If not, then it's definitely bad.
David Epstein35:04
Figures
| Commuting time collectively saved during the London Underground strike | about 1,500 hours/day | 8:35 |
| Length of the London Underground strike | 48 hours | 8:35 |
| Vocabulary size used to write Green Eggs and Ham | 50 words | 29:53 |
| Size of the vocabulary list for Dr. Seuss's early children's books | about 200 words | 29:53 |
| Groups in the invention experiment's constraint conditions | 100 parts free choice / 20 parts to make furniture / 20 parts to make a chair | 34:00 |
| Time and budget given to the LCROSS mission | half of what was expected | 36:05 |
| Group that was less bored in the video choice experiment | people who could watch only 1 video were less bored than those who could choose from 20 | 37:06 |
Glossary
- mere urgency effect
- People prioritize what feels urgent over what is objectively more important.
- pre-registration
- Writing down your hypothesis before a study begins and sticking to it, preventing after-the-fact invented predictions.
- HARKing
- Hypothesizing After the Results are Known — inventing a hypothesis after seeing the results.
- replication crisis
- The phenomenon in which a large number of famous research findings cannot be reproduced.
- Hemingway principle
- Stopping mid-sentence to leave a clear starting point for the next day.
How to listen
Founders making product trade-offs, growth operators running experiments, and anyone interested in research integrity and decision-making methods.
The opening about the 10,000-hour rule and the personal feud with Malcolm Gladwell can be skipped.