> "I don't think the work can ever speak for itself; it is simply too hard to verify the data." It wouldn't work in every case, however it is possible to have

> "I don't think the work can ever speak for itself; it is simply too hard to verify the data." It wouldn't work in every case, however it is possible to have a reproducible experiment as part of a scientific paper. If the experiment can be reproduced (by publicly known scientists), then the identity of the author is not that important. The only time this wouldn't work is when experiments were too costly and/or time-consuming to reproduce. > "Finally, anonymity precludes collaboration, the very basis of science. If I don't know who did something I can't call them up to extend their work, get their raw data, get help optimizing their method, etc." Could go down the pseudo-anonymous route instead (similar to sites like HN and GitHub). That way the contact information could be specific to a single project, or could span multiple projects.