JSON is a more popular data type and is easily serialized. I find that we are constantly adding schema to support taking all of the attributes to describe whatever data we are integrating onto appropriate objects, whether they are custom objects or existing objects, like users.
In some objects in the IDM world, we are simply using JSON text strings to handle these more complex data types without exploding schema for every possible permutation of attributes. The issue we run into with dumping data into JSON text attributes is that we are unable to index them for search, without a ton of overhead, so that we can search for specific key:value pairs. It would be interesting if we had a JSON attribute type that we could set upon schema creation.
Currently, we are splitting this data out into separate schema attributes for each key:value pair, so that we can assign all of the values that way, then index the appropriate ones. If we could just dump the blob into a JSON element, then query and index specific key:values, it would make modelling data in the directory significantly more flexible.
Back in the days there was an effort to standardize similar features for XML: https://tools.ietf.org/html/draft-legg-xed-matching-03
Maybe one could combine these ideas with current JSON data type support like in PostgreSQL: https://www.postgresql.org/docs/12/datatype-json.html