agree!

> Exactly. I rather trust well proven math more than people or infrastructure. One famous example nowadays is Bitcoin ... nobody was able to break the fundamental math behind it.

Well, there was the integer overflow bug years ago where someone could essentially create money out of thin air. But that's the only one I know of and it's a pretty amazing security track record for such a high-profile and lucrative target.

That said, this is just me being pedantic, I agree I'd much rather trust solid crypto than a promise from a person somewhere, even if that promise is in writing.

Depends on how long you want your data to be private, though. There's no guarantee that the encryption won't be broken in a decade or three. And, even if it's not mathematically broken, increased computing power (quantum?) could make brute-forcing fairly trivial.

True, Tarsnap is pretty on-point there, but it's also not cheap. $0.25/GB is much more than S3 ($0.023) or B2 ($0.005) - the tarsnap dev says it's because he does blocking and that makes it so much more valuable. But and there are other tools that can do encrypted backups with blocking like Duplicati and can be used with cheaper services. With this considered, Tarsnap is 50x the price of B2 - and that's without counting bandwidth.

Or if you're a cheap fuck like me, you want even lower and you go to OVH Hubic which is $50 for 10 TB for a year, with no additional bandwidth cost.

hm. Security policies is something you can break with one way or another. You cannot break mathematics that way.