What you call question-begging is really just an assumption I'm making - that git's feature set isn't available (or practically feasible) for any binary format.

What you call question-begging is really just an assumption I'm making - that git's feature set isn't available (or practically feasible) for any binary format. You might disagree with that assumption, but if so I'd challenge you to show me (or build) a piece of software that (a) uses a binary storage file format and (b) has a feature set comparable to that of git for files in that format. I know an argument from lack of existence doesn't prove anything. It certainly is within the universe of the possible to implement git's feature set in MS Word. But I highly doubt MS will ever do that, as that wouldn't make sense from a business standpoint. A great general version control system is most likely to be implemented on formats that are the most interoperable. Nothing is more interoperable than plaintext. Perhaps if someone designs a structured general-use binary format that becomes widely adopted, then version control will be implemented for it. But I bet if someone does try to build such a binary format, someone else will follow it up by building a similar or better plaintext format. Plaintext is more attractive for non-proprietary software, because you get a lot for free. In fact, I believe the only long-term business benefit of a binary format is vendor lock-in. You might say developers are spoiled -- we enjoy power and freedom in our workflows that others don't even know is possible. I for one prefer the power and freedom that plaintext formats (both data and code) afford. Although I do sometimes use MS Word, I almost never commit a non-plaintext file to a git repo when there's a plaintext-format alternative.