I'm going to preface this with the statement that I know nothing of the technical implementation of Libra. I'd be grateful if someone who does know would comment to see how it compares to what I'm about to say.
I was perplexed as anyone about the use of blockchain for Libra, but seeing this thread I can imagine a scenario where it makes a ton of sense. If they are simply using a markov chain to create a ledger that makes public all transactions in a provable way, then this actually seems like a good idea to me. If you give FB $10 and expect to get X Libra back, you need your bank's help to prove that you gave FB $10, but you don't need anything to prove that FB did or did not give you X Libra (assuming you agree to identify yourself as the recipient of the transaction).
In fact, I don't think we actually need proof that FB is receiving funds for its Libra, because FB has tonnes of cash (almost literally). You can think of the receipt of Libra as a bond rather than as cash. You can then trade that bond for things or other cash. Because of the aforementioned tonnes of cash, this bond quite rightly should be highly rated -- in fact, I suspect it would be rated higher than many government bonds.
This may worry governments. I'm not sure -- incredibly liquid bonds would essentially be cash and could undermine weak currencies. You can imagine that if you are in trouble with hyper inflation, anybody who can will grab Libra because it will be easy to get and very, very liquid. This could sink your currency pretty quickly.
On the plus side (and I can't believe I'm actually saying this), this is practically exactly what I wanted when BTC was first released. Of course it is crap that it will be run by a single entity... but if it's liquid enough and if they don't start charging ridiculous fees it would be awesome as a payment method for online purchases.