The Confederacy of Heterogeneous Software Organizations and Heterogeneous Developers: Field Experimental Evidence on Sorting and Worker Effort

Abstract

Software development occurs in a patchwork or “confederacy” of different types of institutions (universities, small start-ups, multinational enterprises, government agencies, etc.) utilizing varied work approaches. Here we speculate on one possible explanation for this organizational heterogeneity: it may reflect inherent heterogeneity of the software workforce, in terms of which kinds of organizations individual workers prefer to work within (“institutional preference”). We take very preliminary steps towards investigating this possibility by devising a novel 10-day field experiment to estimate the differences in behavior that are created by sorting workers into their preferred institutional regimes versus having them unsorted. The experiment involved assigning 1040 elite software developers to either a competitive or a cooperative work regime to create software for NASA’s Space Life Sciences Directorate. Half of the subjects - the “sorted” group - were assigned according to their institutional preferences; the other half - the “unsorted” group - were assigned without regard to their preferences. Assignment was done in a manner that sorted and unsorted groups had identical distributions of raw problem-solving ability. We find a remarkably large effect of institutional preference-based sorting on the effort exerted. Sorting on institutional preferences roughly doubled effort within the competitive regime and increased effort by roughly half in the cooperative regime, while accounting for incentives. Our experimental approach and results indicate the importance of accounting for worker preferences in creative activities that drive the rate and direction of inventive activity in the economy.