Externalities in Knowledge Production: Evidence from a Randomized Field Experiment

TL;DR

This paper asks whether seeding knowledge on a platform like Wikipedia generates externalities in future knowledge production. It uses a randomized field experiment in 2014 that added about 2,000 characters and a photo to pages about mid-sized Spanish cities across several language editions, and tracks edits through 2018. The results show negligible long-run growth in content, with only short-run increases in editing activity and new editors that largely target the seeded content itself. The findings imply that information seeding and incentives for contributions have limited impact on long-term knowledge accumulation, guiding cost-benefit decisions for platforms that rely on user-generated content.

Abstract

Are there positive or negative externalities in knowledge production? Do current contributions to knowledge production increase or decrease the future growth of knowledge? We use a randomized field experiment, which added relevant content to some pages in Wikipedia while leaving similar pages unchanged. We find that the addition of content has a negligible impact on the subsequent long-run growth of content. Our results have implications for information seeding and incentivizing contributions, implying that additional content does not generate sizable externalities by inspiring nor discouraging future contributions.

Figures (7)

• Figure 1: Average page length in the treatment and control groups
• Figure 2: Average page length in the treatment and control groups, by language
• Figure 3: The impact of treatment on page length
• Figure 4: The impact of treatment on the monthly average number of users and edits
• Figure A.1: Robustness: Page length