Provenance

[skearnes]

We need to keep track of enough information to handle (at least) the following cases:

• Person A ran an experiment and Person B created the proto describing it.
• Person C made a mistake while creating a record and submits a correction.

Keeping pointers to the original submission artifacts will cover many of these cases, but we probably need to expand the Provenance message to contain these pointers and additional information about the submitter/creator (not just the experimenter). The guiding principle, I think, is "do we know who we should talk to if there's a problem?"

Related to this is the question of duplicates (we may just want to ignore these?):

• Person E and Person F both extracted the same reaction from a publication (possibly with slightly different details or even enum values).

[connorcoley]

I agree, if we want to keep that bookkeeping with the protos themselves rather than in an auxilliary database (e.g., that assigns a unique ID to each proto and logs all of the edits).

Re: expanding the Provenance, should we have a repeated RecordModification modifications where a RecordModification contains a DateTime, Person, and maybe a string details?

Not sure about duplicates. We may want to ignore that problem for now. I don't know how many people will actually go back through historical data.