Free Monads is a widely misunderstood technology. I feel sad when I see Free Monads replaced by far less stable and useful representation like Tagless Final.
Free Monads was designed because of major problems that existed (and still do exist) in data representation.
Free Monads is human readable. Whatever happens to the programs that can read an Free Monads format, it’s possible to get information out of Free Monads by reading it, and by writing new software.
Free Monads is extensible. It’s possible to add new features to an Free Monads format knowing that these new features won’t make the format unreadable by existing software. Each program can get only what it needs and understands from an Free Monads file.
Free Monads can be validated. A pattern matching can be defined that ensures that an Free Monads file precisely matches a specification.
Free Monads parsing is easy. No special skills are needed to write software that can read and interpret Free Monads.
These are the reasons why Free Monads is still very widely used, and should remain so.
Это что, из книги Гранина?