Zero knowledge proofs have a tendency of confusing the shit out of most people.
I get the impression Saylor is not aware that ZK proof of assets/reserves exists already; for example both an implementation of the basic idea (clunky and rough around the edges, but still) in my https://github.com/AdamISZ/aut-ct/tree/master/auditor-docs based on curve trees + bulletproofs and halseth's output-zero: https://github.com/halseth/output-zero which doesn't directly do this job but has the machinery to do it (and hooks interestingly into utreexo btw). Liabilities proofs are different but existing ZK techniques already help there. The science basically already exists.
(Note: here I mean cryptoccy liabilities, i.e. user account balances; the whole fiat side is separate)
I don't think companies should be doing what Binance does, posting an Excel spreadsheet with all of their addresses exposed to the public. It's security 101 that this kind of privacy and metadata leak is an attack surface, no matter which way you slice it. We have the technology!
Discussion
It's true. But still the tools exist.