I had an interesting conversation this week with the folks behind https://biofactoid.org, a service to allow researchers to openly make available the semantics behind pathways they study in their paper. Their service is working quite well at the moment because they hide all of the complexity from researchers. They told me that even using terminology like "database" or "identifier" reduces the likelihood of researchers completing the task. I say this because I think the idea that researchers will use a self generate public key won't work at the the scale of our scholarly ecosystem (roughly 5M papers peer reviewed papers published annually).