In 1963, a physicist and historian of science named Derek J. de Solla Price published a set of lectures delivered at Brookhaven National Laboratory under the title Little Science, Big Science. He had been studying the distribution of scientific output — who published, how much, and who got cited — and he found something that violated his intuitions about how intellectual communities work.
The distribution was not a bell curve. It was not even close. A small fraction of researchers were responsible for approximately half of all published work in any given field. Price formalised this as a hypothesis: the square root of the number of authors in a domain produces roughly 50 percent of its output.
The observation became known as Price’s Law. It has since been applied well beyond scientific publishing — to corporate productivity, creative output, economic contribution, and the dynamics of any complex system in which skill, consistency, and accumulated advantage interact over time.
It has also been substantially challenged. Subsequent empirical studies across multiple scientific disciplines found that actual publication distributions do not reliably follow the square root pattern Price described. The precise mathematical formulation, it turns out, is less robust than its initial elegance suggested.
What survived the empirical critique, however, is the underlying structural observation: output in complex human systems tends to concentrate. Not randomly, and not because systems are rigged — though they sometimes are — but because of the way compounding works when multiple contributing factors interact over time.
The simple version of the equation looks like this: output is not a function of effort alone. It is a function of effort multiplied by compounding factors — feedback quality, domain fit, starting conditions, time — and that multiplication means small early differences produce large late differences. Most people are applying effort to a linear model. The actual model is multiplicative. That mismatch is the problem this article is about.
This synthesis — taking the distributional finding from scientometrics, combining it with what the behavioural science literature actually says about skill acquisition, and drawing out the practical implication — does not appear often in the places people encounter advice about effort and persistence. The individual components are known in their respective fields. The combined picture is not what most people are working from.
Why output concentrates regardless of fairness
The behavioural science literature on skill acquisition offers a cleaner account of why output concentrates than Price’s Law alone provides.
The foundational research by K. Anders Ericsson, published in the Psychological Review in 1993, proposed that expert performance is primarily a function of deliberate practice — structured, effortful, feedback-rich engagement with a domain over time. The implication was broadly democratising: given sufficient practice of the right kind, exceptional performance was within reach of most people.
Later research complicated this significantly. A series of meta-analyses by Macnamara, synthesising data across domains including music, chess, sports, and academic performance, found that deliberate practice accounted for a highly variable and often modest portion of the variance in expert performance — substantial in some domains like chess and music, and strikingly small in others like professional work and education, where it explained as little as 1 to 4 percent of the variance. Subsequent research has identified starting age, working memory capacity, and genetic predispositions as contributing independently to performance outcomes.
What this produces, in any domain where multiple contributing factors each compound over time, is a distribution that concentrates at the top regardless of whether the system is designed to be fair. Equal starting resources, equal training, equal time — and outcomes still stratify. Each factor multiplies the others. Small initial differences in any one of them accumulate into large differences in outcome as time passes.
For those whose combination of factors does not reach the compounding threshold — who are applying effort without the feedback structures, domain fit, or accumulated base that allows it to multiply — output does not climb at the rate the effort deserves. It accumulates more slowly, cycling without the multiplication. This is not failure of will. It is what the structure produces when the multiplicative conditions are not in place.
This is not a moral claim. It is a structural one. The distribution is a product of how compounding arithmetic works, not of intention or design.
The equation most people are actually solving
Most people operate on an implicit model in which effort maps linearly to outcome. Work twice as hard, get approximately twice the result. Improve by ten percent each year, be ten percent better in a year. This model is intuitive, it is what most educational and institutional systems implicitly promise, and it is broadly incorrect for complex skill domains.
The actual relationship between consistent effort and outcome in most demanding domains is nonlinear. In the early stages of skill development, output is flat relative to effort — the curve climbs slowly, foundational competencies are being built, and results are not yet visible. This is the phase at which most people stop. The effort feels disproportionate to the return. The feedback loop is weak. The rational calculation, applied to a linear model, says to redirect effort elsewhere.
What the compounding model shows is that the return on early investment only becomes legible after a threshold that looks, from inside the flat section of the curve, like it may never arrive. The people whose output eventually concentrates at the top of any distribution are often not those who found the work easier in the early stages. They are those who continued to apply consistent effort through the phase where the linear model would have told them to stop.
The role of domain selection
The evidence on deliberate practice also points toward something Price’s Law does not address directly: not all effort compounds equally across all domains for all people.
Ericsson’s original framework emphasised practice structure over innate capacity. The Macnamara meta-analyses found that the predictive power of deliberate practice varied significantly by domain — it explained more variance in performance in some fields than others, and less in domains where early starting age and working memory capacity played larger independent roles.
This creates a more nuanced prescription than either the “work harder” or “talent is destiny” readings typically suggest. The question is not simply whether someone is working hard enough or whether they have the right genetic endowment. It is whether the compounding of their specific combination of factors — starting point, working memory, domain responsiveness to practice, available feedback structures, time — is more favourable in the domain they are currently working in than in alternatives they have not considered.
Consider two people with equivalent hours invested in skill-intensive work. One is in a domain with fast, accurate feedback and strong responsiveness to deliberate practice — their effort is multiplying. The other is in a domain where feedback is delayed, ambiguous, and where working memory plays a larger independent role they happen not to have — their effort is accumulating linearly at best. Same input. Structurally different output trajectories. The difference is not motivation.
What this means for how effort is allocated
The practical implication for anyone working in a domain that rewards accumulated expertise — which includes most knowledge work, research, creative production, and skill-intensive professional fields — follows from the structure of the compounding curve rather than from Price’s Law specifically.
Volume matters in the early stages not because most output will be excellent, but because the pattern that eventually compounds cannot be identified without sufficient exposure to the distribution of one’s own work. As writer Kaguura Gichuru documented using his own Substack data, approximately 17 pieces out of 730 total outputs drove the majority of his subscriber growth — roughly 2.3 percent. But those 17 pieces could not have been identified in advance. They became identifiable only retrospectively, after the volume had been produced and the pattern had emerged.
This is a form of empirical self-knowledge that the linear effort model forecloses. If each piece of output is treated as an independent event to be maximised, the distributional pattern never becomes visible. If early volume is treated as the sampling process by which the productive subset is eventually identified, the math changes.
The second implication is about domain specificity. The research on deliberate practice suggests that effort invested in activities that provide immediate, accurate feedback and push just beyond current capability compounds more efficiently than equivalent time in activities that do not. Most people, most of the time, do not have access to either the feedback quality or the difficulty calibration that deliberate practice requires. The effort is being expended. The structure that allows it to compound is not in place.
The distributional reality and what it leaves open
Price’s original hypothesis has not held up cleanly to empirical scrutiny. The precise square root relationship between contributors and output does not generalise reliably across domains. What generalises is the shape of the observation: output in complex human systems concentrates, and it concentrates because multiple compounding factors interact in ways that amplify small early differences over long timescales.
This is uncomfortable in the same way that honest accounts of inequality tend to be uncomfortable. It does not resolve into a simple prescription. It does not promise that consistent effort will produce exceptional outcomes for everyone who applies it. It does not vindicate either the pure talent view or the pure practice view.
What it does is describe the actual structure of the problem more clearly than the linear model does. Most people are solving an equation that assumes output scales proportionally with input. The behavioural science literature, and the distributional evidence from any complex human system, suggests the equation is different. The relationship between early effort and eventual output is mediated by compounding, by domain fit, by feedback quality, and by the length of time over which consistent application is maintained.
Understanding that equation does not guarantee a different outcome. It does change which variables are worth examining first.
