Price's law
Price law · square root law
Derek de Solla Price's claim that half the output of a group comes from the square root of the number of contributors — so of a hundred people, ten produce half the work. The pattern is real; the square-root form is much weaker evidentially than it is usually presented as being.
In practice
Contribution in most creative and technical work is not distributed evenly and is not close to even. Planning as though a team of forty is forty equal units produces a plan that does not survive anyone leaving.
The common mistake
Quoting it as an established law. Price observed the pattern in scientific publication counts in the 1960s; the precise square-root relationship does not hold generally, and the number of papers someone publishes is a poor stand-in for value.
Derek de Solla Price was a historian of science who counted things, and one of the things he counted was how many papers each scientist in a field published. The distribution was extremely uneven: a small number of people accounted for an enormous share of the output. His formulation was that half the output comes from the square root of the number of contributors — of a hundred publishing scientists, about ten produce half the papers.
The observation of unevenness is solid and widely reproduced. The square root is not, and the distinction is worth insisting on.
What is actually established
That output in creative, technical and scientific work follows a heavily skewed distribution rather than a normal one. This shows up almost everywhere it is measured: papers, patents, software commits, book sales, hit records, startup returns.
That is the same family of observation as the Pareto principle, and it is what a steeply declining contribution curve looks like when you add it up — see diminishing marginal utility for the shape.
What is not
The square root specifically. It is one curve among several that fit skewed data, and it is not clear it fits better than the alternatives in most domains. Price fitted it to publication counts in mid-century science; treating it as a general law of human organizations is an extrapolation nobody has earned.
That counting outputs measures contribution. Papers published is a proxy, and a bad one — it rewards volume, salami-slicing and authorship conventions, and it is exactly the kind of indicator that deforms under pressure. See Campbell's law. The same objection applies to commits, tickets closed and lines of code.
That the top contributors are a fixed set. The skew can be produced by a stable elite, and it can also be produced by cumulative advantage, where early success attracts the resources that produce later success. Those look identical in a snapshot and have completely different implications for what to do.
Why it is worth knowing anyway
Because the practical consequences follow from the unevenness and not from the exponent.
Planning by headcount is wrong. A team of forty is not forty units of anything, and a plan built on the average is a plan that fails when a specific few people are unavailable. This is owner dependency with a distribution behind it.
Adding people has sharply diminishing returns. If contribution is heavily skewed, the marginal hire is far below the average of the existing team, which is one of the mechanisms behind Brooks's law.
Averages describe nobody. Reporting the mean of a skewed distribution produces a number no individual is near, and decisions made on it are made about a person who does not exist.
The use that is not honest
It gets quoted to justify concentrating rewards, on the reasoning that a few people produce most of the value and should capture most of the return. That may be right and it does not follow from the law, for two reasons.
The measured output is a proxy chosen because it was countable, and the correlation between it and value is assumed rather than shown. And cumulative advantage means today's distribution is partly a record of who got resources earlier, so using it to allocate resources now is a feedback loop that produces the distribution it claims to be responding to.
The observation is a reason to plan for unevenness. It is not, on its own, a finding about desert.
Concept web
Open the full webQuestions
What is Price's law?
Derek de Solla Price's claim that half a group's output comes from the square root of the number of contributors — so of a hundred people, about ten produce half the work.
Is Price's law accurate?
The unevenness is well established; the square-root form is not. Price fitted it to publication counts in mid-century science, and treating it as a general law of organizations is an extrapolation the evidence does not support.
How is Price's law different from the Pareto principle?
They describe the same family of skewed distribution. Pareto is the general observation; Price's law makes a specific and much stronger claim about the exact relationship, which is where the evidence thins out.