As things stand, users have to query both the Errata service and the ESGF API to get information about which data has been retracted, as its not clear that all publishers are filing Errata in a timely fashion or at all. (1) might be worth investigating how many retracted datasets don't have an errata. (2) Could an automated process create an errata for these, with perhaps an errata generated for the retracted dataset with no corresponding errata - could be "Unspecified issue" ?