Can Jupyter stop using JSON and look for a new structure (not only about the file format but also the data fields representation)? The current format is unorder

Can Jupyter stop using JSON and look for a new structure (not only about the file format but also the data fields representation)? The current format is unordered and contains huge binary blobs which makes it very inefficient and version control is simply a pain. I believe a new design worth sacrificing backward compability.