Smith had been paying a piece rate—a rate per kilogram of fruit picked—to his fruit pickers, but he had come to the conclusion that his pay plan wasn't working the way he wanted . . . or, perhaps, the way the workers themselves might have preferred (???).
Farmer Smith tried to adjust the piece rate each day so that it was always adequate but never generous: The more the work force picked, the lower the piece rate. But his workers were outwitting him by keeping an eye on each other, making sure nobody picked too quickly, and thus collectively slowing down and cranking up the piece rate.
In the first year, the economists' proposed pay scheme "increased productivity (kilograms of fruit per worker per hour) by about 50 percent."
The next summer, the researchers turned their attention to incentives for low-level managers, who would also be temporary immigrant workers but who would be responsible for on-the-spot decisions such as which workers were assigned to which row. The researchers found that managers tended to do their friends favors by assigning them the easiest rows. This made life comfortable for insiders but was unproductive since the most efficient assignment for fruit picking is for the best workers to get the best rows. The researchers responded by linking managers' pay to the daily harvest. The result was that managers started favoring the best workers rather than their own friends, and productivity rose by another 20 percent.
Small wonder that the economists were invited back for another summer. They proposed a "tournament" scheme in which workers were allowed to sort themselves into teams. Initially, friends tended to group themselves together, but as the economists began to publish league tables and then hand out prizes to the most productive teams, that changed. Again, workers prioritized money over social ties, abandoning groups of friends to ally themselves with the most productive co-workers who would accept them. In practice, that meant that the fastest workers clustered together, and again, productivity soared—by yet another 20 percent.
How far can this kind of improved efficiency go? We don't know. After three seasons, the economists figured they had run out all the improvements they were going to get. Farmer Smith, on the other hand, has now hired a consultant to try to continue to improve his workers' efficiency.
If this research is of interest to you, I would recommend you not only read the original article, but the more scholarly University College London papers referred to within it as well: The Evolution of Cooperative Norms: Evidence from a Natural Field Experiment and Incentives for Managers and Inequality Among Workers.