I’d say that it’s not necessarily plain text files, but some text based formats, preferably open. Easy to parse, read and improve.
I follow something similar here, but with a difference that I don’t do anything in closed or binary formats, except for images and such binary only things, that I use only open formats. It’s being working fine - I also have everything in a source control (moved to git, currently).
As someone suggested here, there’s no better way to see changes over time when you need to see a previous version.