[andyferris]

Not at all.

GraphQL has the API servers (database clients) marshalling the data into heirarchies (nested data) and managing consistent results (e.g. streaming updates).

The article is arguing that the database should be more capable of understanding queries that request nested data (and separately Jamie has spoken in favour of databases that encourage end clients getting up-to-date/internally-consistent results).

My two cents is that it's a crutch having SQL rely on JSON instead of it's own first-class nested data (which would be any given value (field, aka cell) can itself be a relation - at least postgres has array values but they are challenging to index/use keys programmatically).