Does seeding online communities with baseline information spur contributor activity and follow-on knowledge production? I shed light on this question by examining data from OpenStreetMap, a Wikipedia-style, digital map-making community that was seeded by the US Census TIGER map at its inception. I estimate the causal effects of information seeding on OpenStreetMap by leveraging a novel dataset of over 350 million contributions made by about 577,000 contributors and a natural experiment, where an oversight caused only about 60% of the counties in the US to be seeded with a complete version of the TIGER map. Rather than increasing follow-on contributions, I find that information seeding significantly lowered follow-on knowledge production and contributor activity on OpenStreetMap. Further, counties that benefited from a higher level of information seeding demonstrated lower levels of quality in the long run, despite their early advantage in this regard. I argue and find empirical support for the mechanism that the TIGER basemap crowds-out contributors’ ability to develop ownership over baseline knowledge and disincentivizes follow-on contributions, which could explain why information seeding stifles rather than spurs knowledge production in online communities.