Hézuò

Cooperation is the collective practice of contributing knowledge or effort toward a shared result, and the book treats it as a condition that must be organized rather than assumed. Its strongest examples show different routes to the same outcome: researchers share evidence without a commanding authority, decision-makers are encouraged to rely on independent information, and participants sustain common resources when they expect others to do their part. Cooperation falters when people can profit by withholding effort or manipulating the rules.

Cooperation and Collective Knowledge

The global investigation of SARS shows how cooperation can turn dispersed expertise into a result no participant could establish alone. After reports of the disease emerged, the World Health Organization contacted eleven laboratories across countries and invited them to find and analyze its cause. The laboratories began working together on March 17, sharing electron-microscope photographs, analyses, and test results on a WHO website, exchanging virus samples, and discussing their work in daily teleconferences. Because they could work simultaneously on the same samples, they tested possibilities in parallel rather than waiting for one laboratory to finish before another began.

The sequence of findings depended on this shared effort. Scientists at Hong Kong University and at the United States Centers for Disease Control separately isolated promising viruses on March 21. Laboratories in Germany, the Netherlands, and Hong Kong then sequenced the coronavirus, and experiments in a Netherlands laboratory found that infected monkeys developed SARS. By April 16, the network was confident that the coronavirus caused the disease. Cynthia Goldsmith first spotted the coronavirus, but identifying a candidate was not the same as proving that it caused SARS: weeks of shared work, including the elimination of other possible causes, established the finding. The discovery belonged to the collaborating laboratories collectively, and took weeks rather than the months or years a single laboratory might have needed.

The arrangement was coordinated but not commanded. The World Health Organization created the network, yet did not dictate which viruses or samples each laboratory should examine or how the work should be divided. The laboratories agreed to share relevant data and speak each morning, then worked out the collaboration among themselves. The example makes cooperation more than a synonym for centralized control: autonomy and information exchange could operate together, with the absence of top-down direction allowing the laboratories to find an efficient division of labor.

This scientific collaboration rests on a broader practice of making knowledge available to others. Researchers generally publish discoveries and provide data for inspection so that colleagues can reassess conclusions, develop new hypotheses, and design experiments. A failed hypothesis can also help by showing others where not to look. Knowledge therefore depends not only on accumulated discoveries but on contemporaries’ ability to use one another’s work; each scientist relies on a shared pool that others help replenish.

Trust, Evidence, and Scientific Judgment

Open exchange does not remove the need for trust. Although experiments should in principle be replicable, most scientists will not repeat most other researchers’ experiments. They rely on colleagues’ reports and data, and an accepted hypothesis is generally one that the scientific community finds credible, not one that every scientist has personally tested. When a researcher fails to reproduce an established experiment, the first suspicion may properly fall on the researcher’s own technique rather than on the result.

That dependence is practical, not a claim that truth is merely a matter of opinion. The coronavirus caused SARS before the scientific community reached agreement about it; broad scrutiny and acceptance made the evidence usable as scientific knowledge. Once that consensus formed, laboratories around the world could pursue tests and possible vaccines on the basis that SARS was caused by a coronavirus. Scientific cooperation thus joins fair dealing with data to collective scrutiny: researchers trust one another enough to build on reported work, while the community’s acceptance determines which claims enter its common fund of knowledge.

Independent Judgment and Group Decisions

Cooperation can also require people to resist copying one another. In an experiment devised by Angela Hung and Charles Plott, participants drew colored marbles in sequence from one of two urns. Urn A contained twice as many light marbles as dark ones, while Urn B contained twice as many dark marbles as light ones. Each participant privately saw a marble’s color and publicly guessed which urn was being used. The private draw and the guesses already made by others could point in different directions, creating a choice between personal evidence and the apparent group consensus.

When students were rewarded for identifying the correct urn individually, information cascades began in 78 percent of trials: participants often followed earlier guesses even when their own marble offered contrary evidence. Hung and Plott changed the incentive so that students were paid according to whether the group’s majority answer was correct. The change made each participant’s independent information more valuable to the group. Although relying on an imperfect private signal could make an individual guess wrong, the group became significantly more accurate than groups caught in cascades. In this case, helping the group meant paying less attention to what others had already said.

The experiment distinguishes cooperation from uniformity. A group’s judgment improves when it incorporates many different signals, not when every participant repeats the same visible opinion. Organizations can therefore support cooperation by having people offer judgments simultaneously or nearly so, reducing the influence of choices made earlier in sequence. The experiment also reveals the difficulty of autonomy: members must be able to attend to their own information even while working toward a shared decision.

Incentives and the Fragility of Cooperation

The public-goods experiments of Ernst Fehr and Simon Gächter make the conflict between individual advantage and shared benefit concrete. Four players each begin with twenty tokens and play four rounds. A player can contribute tokens to a public pot or keep them. Each contributed token costs its contributor one token but earns 0.4 tokens for each of the four group members, a total group return of 1.6. If everyone keeps their tokens, each ends with twenty; if everyone contributes everything, each ends with thirty-two. Yet the ordinary temptation is to keep one’s own tokens and benefit from other people’s contributions.

Participants typically begin by contributing about half their tokens, but contributions decline as players observe others free riding. By the end of the experiments, 70 to 80 percent of players are free riding, leaving the group poorer than it would have been with sustained contributions. The deterioration is not simply a failure to recognize the shared payoff: the rules make noncontribution individually attractive, especially when others continue to contribute. Cooperation depends on expectations about other participants as well as on the design of the activity.

The book connects this conditional willingness to cooperate with tax compliance. People are more willing to pay when they believe others are paying and that evaders have a reasonable chance of being caught and punished. If they think rule breakers are getting away with it, they may feel that compliance makes them suckers. Across both taxation and the public-goods game, shared participation is vulnerable to perceived unfairness: cooperation is easier to sustain when people can expect reciprocity and accountability.

Organizational Incentives and Their Costs

Cooperation also requires information to move honestly through organizations. A hierarchy may give decision-makers authority, but it can discourage subordinates from offering unwelcome information: bosses may resist opposition, while employees already have reason to avoid conflict. The result can be an appearance of agreement that prevents an organization from learning what is actually happening.

Donald Roy’s early-1950s account of machine-shop work shows how incentives can directly reward strategic behavior rather than shared productivity. As a lathe operator, Roy was paid under a piece-rate system in which the rate rose after workers reached specified targets and increased again at a second hurdle before being capped. Workers knew that producing too quickly could lead the company to raise the hurdles. They therefore restricted output and concentrated on managing the payment system rather than working as productively as possible; Roy called the practice “Goldbricking in a Machine Shop.”

The same logic can shape executive targets. When bonuses depend on surpassing expectations, managers may set targets low by understating prospects, then use accounting maneuvers to meet them at the expense of future performance. The system can reward the appearance of success instead of accurate reporting and durable results. These cases sharpen the book’s central distinction: people’s capacity to cooperate is real, but it is not protected by good intentions alone. Institutions must make honest contributions, independent judgment, and reciprocal effort more worthwhile than concealment, imitation, or free riding.