>I always said that in-device encryption that you cannot verify is useless and considered those SATA drives with built-in encryption to be useless. Famous HN luminaries disagreed with me.

I mean, yeah? Throwing around words like "useless" without context is almost universally unserious and irritating in a security conversation since security is about economics and threat scenarios. I'm aware of the 2019 paper "Self-Encrypting Deception: Weaknesses in the Encryption of Solid State Drives" [0], but not a lot since then on the state of Opal and any other SED mechanisms and real world effects. Even for the drives discussed there, it's not clear it actually affects what much of the population cared about in terms of FDE, which would more be along the lines of "it'd be hard for a random casual person to pop the drive into an enclosure and just mount it and copy everything off, I don't have to use a super long password all the time, and it doesn't cost noticeable speed".

As another practical issue, encryption is not something most people can verify anyway whether they have the code or not. The pure cold data at rest symmetric use-case means that a lot of hot attacks and side channel stuff isn't so important, even so I sure wouldn't trust myself to either write it or verify it. I mean, I guess I could spot some truly dumb-tier backdoor hard coded keys but not anything subtle in terms of weakness, and even then not every single software update. For most people/organizations they're all black boxes based on trust in the manufacturer/developer(s).

I mean, I don't disagree with you that personally I go with software, less from an encryption pov (though it's not nothing) and more from a recovery-if-things-go-fubar pov. I'd agree the incentives are better, and that while many-eyes is not a panacea lower barriers for security researchers to poke around is still better on balance. Perhaps the LLM-era further pushes it that way, and really will make it more practical for more people to check over software solutions.

But still seems like kind of a stretch for you to assert that Apple's (or many others) in-device encryption is "useless"? We've seen some pretty powerful organizations worldwide really want to get at protected data and then fail even when stakes were high.

----

0: https://ieeexplore.ieee.org/document/8835339

Lenovo's 'encrypted' fingerprint USB is just an unencrypted SD card in a case