T-Rec 4.4 Mapping repositories should allow accessibility to mapping metadata even when the mapping is no longer available

[linikujp]

Mappings and crosswalks often become deprecated over time due to the cost-intensive maintenance they require. To avoid the inconvenience of searching for mappings and crosswalks that no longer exist, it is advisable to keep the associated metadata accessible. Storing metadata is relatively cost-effective, and it enables searchers to explore additional information or resources described in the metadata related to the sought-after mappings or crosswalks.

[graybeal]

I realize there is a FAIR principle that uses these words, but I think in the case of mappings it should be particularly unconcerning. Mappings are unlike 'research data' in several relevant ways: mappings themselves are pretty simple and self-explanatory, especially SSSOM-compliant mappings; mappings take relatively little space even if millions of them are being autogenerated; they might be updated over time but they don't have to be, in which case, they require no maintenance and incur no growth or other costs; and there is enough obvious value in deprecated mappings that I'm having trouble seeing repositories desperate to get rid of their deprecated mapping files.

If someone has decided the "sought-after mappings or crosswalks" should be deprecated, why would they care about the associated metadata? I mean, the metadata is no more cost-effective than storing the actual data—there can be way more metadata about a mapping instance than the actual data in the instance—and as noted mappings and crosswalks require no maintenance if they are not being updated. They just sit there.

So this ticket seems to assume there is a well-maintained metadata storage system that understands mappings and crosswalks, but no corresponding data storage system for the original data. It also assumes the two must be kept separately, which itself encourages the loss of half the data. (See #66 etc.)

I could agree with this ticket if its focus is shifted to preserving access to the metadata of deprecated mappings and the recovery of the actual mappings. The use case I see is that the author deprecates a mapping set or mapping, so they aren't routinely visible, but later scientists want to see or use or analyze the mappings anyway. The only way they will know the mapping(set) exists is if they can search for and discover the deprecated items, and if they find one they want to access, to get access to it. In other words, repositories should always offer the feature to 'search previous and deprecated content'. (Of course, it's non-trivial in most cases so they don't, but we can point out it's only FAIR, pun intended.)