Explore potential for EID "Presubmission"

[acoffman]

Some journals require variant submission to an open database as a precondition for publication. CIViC would like to be able to fulfill that role. Currently however, every EID must be linked to an already existing, peer-reviewed publication and we want to retain that level of provenance.

In the interest of satisfying both goals we have discussed the idea of an unlisted EID status that lacks a formal source and will not appear in the UI or data exports, but will allow authors to submit an EID pending acceptance of a publication, mint a CIViC ID and generate a link to show journal editors. (@ahwagner compared it to an unlisted YouTube video). Upon publication, authors can return to CIViC, add the formal source ID, and the EID will move into the normal 'submitted' state and become visible on the site and ready for normal civic review.

This would allow authors to show journals they're in compliance with the submission requirements and allow CIViC to fill the role of a prepublication deposit for journals while retaining our core value that all evidence that users encounter on the site link back to a peer reviewed publication.

There are some open questions about implementation that I have (and I'm sure plenty more):

• Do we require a preprint, or since its unlisted, is no source at all fine to start?

• If an editor has access to a preprint or other copy of the paper, can they "pre-approve" the EID so that it goes from "provisional' to "accepted" rather than "submitted" upon addition of the source ID.

• Is there any potential for integrating with the journals themselves to make the process smoother? Or will a user just create an EID on our site and then paste the URL into their submission somewhere?

• Right now EIDs are monotonically increasing integers that directly reflect our primary keys. While this is handy, it does have some undesirable properties for the "unlisted" use case as someone naively iterating through IDs could discover (on purpose or inadvertently) unlisted EIDs quite easily. Often UUIDs are used for this reason but those are not user friendly to say out loud, in text, or as a curie. There are lots of alternate ID schemes (nanoid is one such example) that strike a better balance between simplicity and "non-discoverability." Perhaps we should explore that idea? (Obviously existing EID assignments would remain unchanged)