I dont think so. It will be hybrid. Because for some application we need more structured and predefined model. now a days mostly the data in net are model-less / structured-on-the -go approach.
Not unless data is reduced to it's simplest representation, being that of 'binary data objects', wherein no complexity has been added to constrain the object as 'text' or 'integer' or 'varchar'. I'm working on it :)
The idea of separating the schema from data storage was proposed in an article in ACM Transactions on Database Systems in 2000 (Parsons and Wand 2000). In this instance-based data model, the schema can guide either the population of the database or as a mechanism to view (query) the data, or neither. However, instances can be accessed independent of any schema/classification. The model is compatible with column-based and graph databases.
Parsons, J. and Y. Wand (2000). Emancipating Instances from the Tyranny of Classes in Information Modeling. ACM Transactions on Database Systems, 25(2), 228-268.