Limits to the wisdom of the crowd in idea selection

Paper highlights

Large crowds are useful for generating diverse ideas, but selecting the best idea is a different problem. If a firm can identify a few unusually accurate judges, adding many weaker judges can make a majority vote worse. The wisdom of the crowd therefore depends on who enters the crowd and how the decision is aggregated, not on size alone.

The model’s practical result is that near-best performance usually requires far fewer people than exact optimization does. Very large crowds are necessary only when everyone available has low individual accuracy.

How the model works

The formal model draws voters from a population whose members differ in their probability of choosing correctly. It varies the population size, the distribution of accuracy, the number recruited, and the firm’s ability to identify the most accurate people. The crowd selects between alternatives by majority vote.

Main results

• Increasing crowd size can reduce decision accuracy when additional members are worse judges than those already selected.
• A small group can achieve almost all of the available performance gain, even when a larger group is technically optimal.
• Better recruiting ability lowers the best crowd size because the firm can concentrate its strongest judges in a small group.
• When no superior judges exist and all individuals are only slightly better than chance, a large crowd becomes useful for canceling errors.

Why it matters

• Firms recruit from a finite pool whose members differ in accuracy; the order in which judges enter the crowd therefore matters.
• The exact optimum may require many voters for tiny incremental gains, while a much smaller group often captures nearly all available performance.
• Firms should first identify judges with relevant predictive skill and only then decide how many judgments to aggregate.

How to use this paper

Cite this for

• A formal model of crowd size in idea selection rather than idea generation.
• Conditions under which adding more voters can reduce majority-vote accuracy.
• The finding that near-best performance often requires a much smaller crowd than exact optimization.

Useful for teaching

• Why the wisdom of crowds depends on recruitment and aggregation, not size alone.
• The difference between generating many ideas and selecting the best one.
• How the marginal value of another judge depends on that judge’s accuracy relative to the selected group.

Careful claim

Large crowds help idea selection mainly when added members provide useful independent judgments or cancel errors; when stronger judges can be identified, adding many weaker voters can lower accuracy.

Abstract

An emerging management trend is to use the “wisdom of the crowd” to make decisions traditionally made by the top management alone. Research on this phenomenon has focused mainly on the capacity of crowds to generate ideas, but much less is known about a crowd’s capacity to select ideas.

To study crowd-based idea selection in firms, this chapter develops a mathematical model of a crowd that makes decisions by majority voting. The model takes into account contingencies that are of particular importance to firms, namely: the size of the population from which the crowd is drawn, the distribution of accuracy among members of the population, and the firm’s ability to recruit the population’s most accurate individuals.

The results show that: (1) under relatively common conditions, increasing the size of the crowd may actually reduce performance; (2) near-optimal performance can usually be achieved by a much smaller crowd than the one required to achieve optimal performance; (3) determining the best crowd size depends critically on the firm’s ability to recruit “accurate” individuals; and (4) good performance does not require large crowds unless all population members exhibit low levels of accuracy.

Other links

• SSRN preprint

Last updated 2026-06-21