I prefer the term "schema-on-read" to "schema-free" or "schema-less". One needs to correctly understand the implications of the fact that all data has SOME schema and that if it changes you need to be able to read the old and new.
I agree. My employer Treasure Data (http://www.treasuredata.com/product) has been offering exactly this (SQL on schema-on-read data) for the last three years as a cloud service. But in reality, people "get" schema-less much better than "schema-on-read" thanks to all the marketing that MongoDB folks have done for "schemaless".
Heard a brilliant quote on this topic at a meetup last year: "If you think your data's schemaless, what that really means is that you don't know what the schema is, and then you really have problems".