The CF Conventions define metadata that promote sharing of climate and forecasting data and facilitate automated processing by computers. Since work on CF began in 1999, the development, maintenance, and evolution of the conventions have mainly been provided by voluntary community contributors. Because of growing engagement and use of the conventions, an organizational framework was established in 2006, which relies on established rules and encourages public input and discussion (currently conducted on GitHub), providing for smooth evolution of the standard to cover a wider range of data. Decisions are made by consensus, which takes time but achieves a better result. The CF standard has been essential to the success of high-profile internationally-coordinated modeling activities (e.g, the Coupled Model Intercomparison Project, which hosts more than 30 million files and more than 15 petabytes of data, all compliant with CF). CF has grown beyond its original purpose of describing gridded data from climate and forecast models, having been adopted widely in the geosciences for both observations and model output. CF's self-explanatory metadata thus facilitates transparent data-sharing and enables interdisciplinary collaborative science.