"and switching to plain text"
Plain text is good for humans only. Not for computers. Plain text is exactly hard for computers to read - and an UTF document will never be in "plain text" - and even ASCII depends on what a computer thinks about whatever is beyond 127 (and it's utterly incapable of representing anything beyond English properly).
Performing math on plain text data is utterly inefficient. Navigation, search, etc. too.
If specialist had to re-create documents from plain text data only, they would spend a lot of time to understand which formatting they should apply.
IT has already became far less efficient since web developers, unable to understand proper data structures, used only inefficient text representation of everything. Let's not go down that hill any more....