Pick out anything that sits at an absolute extreme—the best month, the worst quarter, the team on top of the table, the happiest week of your life—and the next time you measure it, it will almost certainly be closer to ordinary. Not because anything went wrong, but simply because extremes do not hold still.
Statisticians call this regression to the mean. The reason it tricks us so consistently is that when something moves back toward normal, human beings automatically look for what pushed it. We want a narrative. And there is usually a convincing candidate standing right there, ready to take the credit or the blame.
Just ask the flight instructors of the 1960s Israeli Air Force.
When a young psychologist named Daniel Kahneman stood in front of them and explained that research proved praise worked better than punishment, the instructors pushed back. They had years of hard evidence to the contrary. Praise a cadet for a beautiful landing, they argued, and his next one is worse. Scream at a cadet for a terrible landing, and his next one is better. Every instructor in the room had watched it happen.
They were absolutely right about the pattern. The landings really did move that way. What they had wrong was the cause. A beautiful landing is a good pilot plus a good day, and the good day does not come back for the next one. Those landings would have done exactly the same thing if the instructors had stood on the tarmac and said nothing at all.
Skill plus luck
Every performance is exactly two things added together: how good you are, and how the day went.
How good you are is stable. It shows up every time you do. How the day went is volatile—it is the sudden gust of wind, the lost hour of sleep, the specific question that happened to be on the exam, or the shot that hits the post instead of going in.
When you isolate the absolute extremes—the flawless landing, the disastrous quarter, the team at the very top of the league—you are never just looking at the most skilled people. You are looking at people whose skill and luck temporarily pointed in the exact same direction.
The skill stays. The luck evaporates. So when the next measurement inevitably drifts back toward the middle, it looks for all the world like something intervened.
But nothing happened to them. You just made a judgment on a number that had luck built into it, and then watched the luck run out.
Try the slider. When none of the score is luck, the top group stays exactly where it was — there is nothing to fall back from. Turn luck up to half and the group gives back about half the distance to the middle, every time, with no coaching, no criticism, no change of any kind. Then switch to the bottom 10% and watch them improve by the same amount. That is the flight instructors’ entire body of evidence, generated by a random number generator.
Where this fools people for real
Medicine. People go to the doctor when their symptoms are at their absolute worst. That is almost the definition of deciding to go. So whatever they do next—a real drug, a herbal tea, a healing crystal—is almost guaranteed to be followed by improvement. This is not a small psychological effect. It is the entire reason controlled trials exist. You need a group of people who were equally miserable and got nothing, or you cannot tell your new treatment from a normal Tuesday.
Road safety. A busy junction has a terrible year for accidents, so the city installs a traffic camera. The next year has fewer accidents, and the camera gets all the credit. But the junction was chosen precisely because it had an unusually bad year, and unusually bad years are followed by ordinary ones all on their own. Some cameras actually work. You just cannot learn which ones from that comparison.
Money. The mutual fund at the top of the leaderboard this year is there partly because of skill, but mostly because its wildest bets happened to land perfectly. Next year, it is ordinary. The leaderboard is not lying to you, and neither is the manager. You are simply reading a list that was sorted by skill-plus-luck, and you are mistakenly expecting only the skill part to persist.
The peak experience. Think of the absolute happiest week of your life. The project launches perfectly, the relationship is electric, the world feels completely conquered. The human ecstasy of that moment is real, and it tricks you into thinking you have unlocked a permanent new baseline. But a peak emotional high is exactly like a flawless airplane landing: it is your baseline effort plus a massive, temporary spike of luck. It requires every single volatile variable in your life to line up perfectly at the exact same time. Your baseline stays, but the luck evaporates. When the following weeks inevitably feel a little flatter, it isn’t because you did something wrong or lost your edge. The universe just rarely rolls sixes twice in a row.
Everything with a “curse.” This is the magazine cover jinx, the sophomore album slump, and the second-season collapse. You get put on the cover of a magazine precisely because you just hit that peak emotional and professional high. By the time the issue actually hits the newsstands, the luck has already evaporated. The inevitable comedown isn’t a curse; it is just statistical gravity.
Hiring and school. The most charming interviewer of the day, the student who topped the entrance exam, the worst-performing employee you threw onto a mandatory training program. All of them will drift back toward the middle on their next evaluation. And every time, someone in management will confidently take the credit or the blame.
What to do with it
One question does most of the work: was this picked for being extreme?
If it was — the top of the leaderboard, the worst-performing team, the patient at their sickest, the junction with the terrible year — then expect it to drift toward the middle on its own, and be suspicious of whatever happens to be standing nearby taking the credit. The only way to know whether the drug, the camera or the training program actually did anything is to compare it against a group that was equally extreme and got nothing at all. That is the entire purpose of a control group.
The same goes for people. Judge them on more than one reading. A single measurement is skill plus luck; several measurements are mostly skill.
And nothing here is being corrected or repaid. A coin that has come up tails five times is not due for heads, and a bad quarter does not earn you a good one. You simply went looking in an unusual spot, and the next look was always going to be calmer.