Research of natural environments often requires extensive field studies beyond the capacity of a single individual. In CONE Welder, we explore harnessing the power of human computation to aid ongoing research on avian activity and possible links to climate changes. Our approach utilizes a singly deployed pan-tilt-zoom web camera to gather observational data. The camera supports multiple operators at any given time. Our frame selection technique weights requests by a dissatisfaction quotient, which quantifies how well a request has been fulfilled since submission. We calculate fulfillment by considering video-specific expectations, including latency, coverage time, and coverage area. Each participant can take snapshots of the live video feed, generating a photographic history of the site's activity. An accompanying game extracts information from these photographs, creating a self-sustaining ecosystem for players and researchers. Players identify animals in a photograph by defining zones, or rectangular selections. Each zone is an opportunity to earn points by correctly classifying the enclosed species. When a consensus classification has been reached, the game rewards the original classifiers. Our scoring model extends beyond classifications, awarding behaviors and activities based on how closely they match our researchers' desires. We deliver all interaction through a unified Flash-based client, which can run on many common computing platforms. This client, in conjunction with our video relay server, implements an enhanced M-JPEG decoder that efficiently extracts frames. Deployed in Spring 2008, the project has already produced a vast quantity of data during its short lifetime. Over 200 participants have taken nearly 2,400 photographs and classified 28 species within them. Participants have provided good temporal coverage of the site, documenting animal activity over all hours. Consensus classifications have identified some species of particular interest to our researchers. This paper describes the key concepts, interface design, and implementation details of CONE Welder. We conclude with early results from our launch and a discussion of future directions.