The Geometry of Luck
In 2006, three researches at Columbia ran one of the cleanest experiments anyone has ever run on success. They put fourteen thousand participants in front of a list of forty-eight songs by unknown bands. Every visitor could listen, rate from one star to five, then download what they liked.
Half them saw only the names of the songs and the bands. The other half saw, beside each song, a small counter showing how many people before them had downloaded it. That was the entire experiment. One number, on or off.
The researchers Salganik, Dodds and Watts (2006) did one more clever thing. They split the second half into eight separate "worlds," each starting from zero, each sealed off from the others and then let the worlds run.
What came back has been reshaping how we should think about success. Inequality went up sharply when people could see what others were doing. Hits became bigger hits. Flops became bigger flops. The more striking finding was that the worlds disagreed with each other. The same song could top the chart in one world and sink to the bottom in another. Quality only set a soft envelope as the very best songs almost never bombed and the very worst almost never won. But across the whole middle range, anything was possible. Early downloads pull in more downloads and leaders feed off being the leaders. It was all about which song happened to get clicked on first.
This is on of the cleanest empirical demonstration on how outcomes in any market are shaped by social influence and therefore partly path-dependent. The successful song is successful because it caught the early tailwind.
What this seems to mean
The Music Lab is a study about songs, but it could also have been your career, your business, your reputation, your life. Run the world again from the same starting point and a different version of you ends up somewhere else entirely, doing roughly the same things, getting roughly the opposite result.
That thought is hard to sit with. It threatens a story most of us are quietly telling about our own lives, in which the good outcomes were earned and the bad ones were instructive. If the Music Lab is right, much of what we credit to ourselves was the early click that happened to land, and much of what we blame ourselves for was the click that happened to miss.
The temptation, once that lands, is to draw a larger conclusion than the data supports. If the same effort produces wildly different outcomes depending on the branch, then effort is theatre. We are performing agency inside a process that has already decided. Some version of this reaction is honest. Most people who read the Music Lab carefully have a moment of it.
The problem is that the conclusion does not follow. The study shows that effort does not guarantee. It does not show that effort does not matter. Those are very different claims, and the slide from one to the other is the most common philosophical mistake you can make about luck.
A guarantee is a one-to-one map from input to output. The Music Lab destroys that map. Nothing in the study, though, says that the distribution of possible outcomes is the same regardless of what you do. The very best songs almost never bombed. The very worst almost never won. Quality set a soft envelope. Inside the envelope, anything was possible, but the envelope itself was real, and which envelope you were in was decided by the work.
That is the move the rest of this essay turns on. Effort does not pick the outcome. Effort picks the distribution. Within the distribution, luck does the rest.
The vertigo
Once that thought lands, it can be hard to recover from. There are two ways the vertigo usually expresses itself.
The nihilistic one says that if outcomes are largely path-dependent, there's no point. The same effort produces wildly different results depending on which random branch you happen to be running in.
The fatalistic one says that if you happened to start in a low-luck branch, you're not just behind. You're inside a process that systematically widens the gap. The losers in world were merely later to the party.
Both reactions follow from the same misreading. They take the empirical finding that outcomes are not deterministic and turn it into a metaphysical claim that effort is therefore impotent. The leap is unjustified. The Music Lab does not show that effort doesn't matter. It shows that effort doesn't guarantee. Those are very different claims. Confusing them is the most common philosophical mistake you can make about luck.
A different frame
Everyone knows Victor E. Frankl, one of his closes pupils is far less known. Alfried Längle worked with him for many years and allthoug with some differences carried forroward many of this ideas. He once in a conversation introduced me to his definition of "success", which he also explains bliliantly in his Book Sinvoll leben (Längle, 2007).
A flower blooming in the forest is a good outcome but not a success. A flower blooming in your garden because you tended it is. The word is reserved for outcomes where human effort is a real causal contributor. That already excludes the two extremes. Things that would have happened anyway are not successes. Things that had nothing to do with you are not successes. Both fall outside the category.
What's left is a middle band. On one side of that band is what Längle calls Bemühen, deliberate effort, including skill, practice, attention and persistence. On the other side is Glück, the favourable conjunction of circumstance that "falls to" the effort from elsewhere. Every real success is a product of both. The German word Glück, he notes in passing, originally meant to close the gap. Effort reaches up to a certain height. What closes the rest of the distance is what we call luck.
Two consequences. Since success requires both halves, neither can carry the full weight of explanation. Crediting yourself entirely for a success is as wrong as blaming yourself entirely for a failure. The depressive who reads every setback as personal incompetence and the narcissist who reads every win as personal genius are making the same error in opposite directions.
And since the halves multiply rather than add, more effort raises the distribution of possible outcomes without ever guaranteeing any single one of them. We can prepare the ground. We can't make the seed sprout.
The deeper move comes when Längle redefines what we are actually after.
Drawing on Frankl, he argues that what looks like a desire for success is almost always a desire for fulfilment. The two are not the same. Success is the achievement of a chosen goal. Fulfilment is the experience of having committed yourself to something genuinely worth doing. Success requires the favourable cooperation of the world. Fulfilment doesn't. A premiere that flops doesn't retroactively make the work meaningless if the work was worth doing.
What you can actually do
There is still a practical question. Given that luck is real and partly outside our reach, are there ways to raise the probability of favourable luck events?
The answer is yes. Luck has a surface area. The larger the surface, the more opportunities pass through it.
Sam Matla (2024), building on Jason Roberts (2010), frames it as a multiplication. Luck is the product of three things. Surface area, the size of the target you present to the world. Position, the part of the world you are in. And staying power, the duration over which the previous two are sustained. Each factor has zero as a possible value. Any zero collapses the product. A large surface in a poor location is wasted. A perfect location with no staying power produces nothing.
Surface area is built primarily through what Roberts called doing-and-telling. Doing alone is not enough. The world cannot find work it does not know exists. Telling alone is not enough. There is nothing for the world to find. The product of the two scales the probability that the right person sees the right thing at the right time.
The mechanism behind this is the same one that drove the Music Lab cascades. Networks reward existing visibility with more visibility. The implication is that volume matters in a way that intuition resists. Skill compounds with reps. Signal emerges from variance. The single best video, post or product typically outperforms the next-best by an order of magnitude. Strategy cannot identify in advance which one it will be. Volume is how that lottery is run.
Position is built through environment and agency. The high-agency person, in George Mack's phrasing, treats the story given to them about what is possible as a story rather than a fact. They court rejection. They ask for unreasonable things. They follow up. They fly to meet someone in person when the meeting matters. They assume what looks fixed is in fact learnable. None of these moves looks dramatic in isolation. Aggregated over years, they shift the social and geographic region in which luck can find them.
Staying power is the simplest of the three to articulate and the hardest to live. Most people abandon a project at article ten, video twenty or year three, just before the regime that would have rewarded their persistence kicks in. Endurance over long horizons is the price of admission for any non-linear payoff. Curiosity is what makes the endurance bearable. Energy is what makes the curiosity sustainable. The three feed each other. None substitutes for the others.
What the luck is worth
Jim Collins ran a different kind of experiment. He and Morten Hansen (Collins & Hansen, 2011) studied pairs of public companies in volatile industries. One member of each pair was an outlier success. The other was a similar peer that failed to thrive. The hypothesis was that the outliers had been luckier. The data did not support it. When they coded discrete luck events across each pair's history, both members had received roughly comparable numbers of good and bad ones.
What separated the winners was something Collins called Return on Luck. The outliers got a higher return on every event of luck they encountered, both by extracting more upside from the good ones and by absorbing more of the impact of the bad ones.
Bill Gates and the IBM contract is the classic case. Several other people in 1980 had access to the same configuration of circumstances. Few had Gates's combination of preparation, paranoia and execution speed. The luck was widely distributed. The capacity to convert it was not.
If outcomes are partly random, the relevant skill is not the elimination of randomness, which is impossible. It is the conversion of randomness into outcome. Discipline, financial conservatism, what Collins calls productive paranoia, the readiness to act decisively at the moments that matter.
Two implications follow. The value of preparation is measured at the points of inflection rather than continuously. Most preparation looks wasted most of the time. It earns its keep in the rare moments when an opportunity arrives that the unprepared cannot capitalise on and the underprepared cannot survive. And surviving a bad luck event that destroys your peers is itself a high return on negative luck. Antifragility, in Taleb's vocabulary (2012), is the same idea seen from a different angle.
How to decide
The Music Lab tells us the world is path-dependent. Maybe you already noticed it, path-dependency is also a key feature of non-ergodicity - one of the three lenses my essay continue to circle The same starting point does not produce the same destination. Längle tells us the right orientation for a life lived inside such a world is values rather than outcomes, since outcomes are partly outside our reach but values are not. Surface area tells us we can shift the probability distribution of outcomes by widening our exposure to favourable ones. Return on Luck tells us that the same widening is wasted without the discipline to convert what arrives.
Three principles for making decisions inside this geometry follow. They bind together rather than stand alone.
The decision-maker who measures themselves on outcomes in a path-dependent world will be either crushed by negative variance they did not deserve or inflated by positive variance they did not earn. The decision-maker who measures themselves on the quality of their inputs, which are within reach, has an evaluable signal that is independent of whether any single bet pays off. This is the operational meaning of Längle's redefinition of fulfilment. It is also what good poker players do. They keep making the right bets and trust the long run. Bad players make the right bet, lose the hand and conclude the bet was wrong. Act in ways that increase the probability that good luck finds you, while staying honest about the fact that you cannot make it find you.
Längle's line captures it. Ich kann nur das mir Mögliche tun, dies aber ganz. I can only do what is possible for me. And that, completely. The rest is the suspense of being alive.
References
Collins, J., & Hansen, M. T. (2011). Great by choice: Uncertainty, chaos and luck. Why some thrive despite them all. HarperBusiness.
Längle, A. (2007). Sinnvoll leben. Residenz Verlag.
Matla, S. (2024). How to increase luck surface area [Video]. YouTube.
Roberts, J. (2010). How to increase your luck surface area. codusoperandi.com.
Salganik, M. J., Dodds, P. S., & Watts, D. J. (2006). Experimental study of inequality and unpredictability in an artificial cultural market. Science, 311(5762), 854–856.
Taleb, N. N. (2012). Antifragile: Things that gain from disorder. Random House.