Another (transactional) lawyer here. While I generally agree with this, there are still circumstances in which something like a VC system would be useful. It's not uncommon for documents to go through a dozen or more (sometimes significantly more) iterations before being finalized. It would sometimes be useful to know in which version a particular fragment of language was inserted. But this can be worked out "manually", and indeed I think the real issue is that the value add of a VC system is probably small enough that it's just not worth any significant additional complexity. Having said that, I've long considered doing a lunchtime brownbag session on git, just to blow people's minds about what is possible.

I do disagree about "Track Changes" though. That quickly becomes an incomprehensible mess. Better to avoid Track Changes (except in very small/simple agreements where you don't expect more than one or two turns, in which case it can be useful) and send a "clean" and redline version (based on the immediately preceding version) on each turn.