Peter,
Sorry to miss your earlier email. It snuck into my junk folder!
I agree completely that checksums play an important role in interoperability and data integrity and are important for datasets (DOIs) that are composed of a single file. Not sure how they would be handled in datasets with multiple files unless each had a separate DOI.
In the single file case, because the checksums are unique and related only to a single instance of a particular file, I think the existing alternateIdentifier element of the schema could be a reasonable place for checksums…
Does that make sense?
Ted