4 ms·
Hold on - what has any of this got to do with openness? How is that in any way part of the argument? Also, I believe you just called Tim Cook a flat out liar,
by zepto 11y ago
Hold on - what has any of this got to do with openness? How is that in any way part of the argument?
Also, I believe you just called Tim Cook a flat out liar, since he has repeatedly, publicly talked about his and the companies beliefs about crypto.
- mindslight 11y agoAn open design for a trusted computing platform would not contain an arbitrary fixed signing key. It would instead allow modifying the trust root subject to some time delay or proof of work. It would be thoroughly documented, giving the manufacturer no special insight into the use or working that any other developer did not have access to. It would thus not be specifically vulnerable to attacks by Apple, and by extension, USG. Apple believes that proper crypto can be used to provide the best security for their users, and Tim Cook may personally believe that crypto provides the best rights to individuals. But corporations operate in terms of what is legal. If proper crypto were made illegal, that just shifts the playing field for everyone. Apple would still exist and try to provide the best security they legally could - they don't have a mortal stake in this fight.
- zepto 11y agoSo basically it has nothing to do with what we're talking about. You just think a completely hypothetical system which has never existed, let alone been subjected to any kind of attack, would be a better alternative, and so you think that what Apple is doing is in error.
- mindslight 11y agoIt has everything to do with what we're talking about. We can analyze the security properties of systems in advance. It's called engineering. This attack isn't some discovered flaw, but an explicit property of their system design. There are widely-used systems that hold up to the attack in question. For example, Linux's LUKS/dmcrypt does not have this vulnerability. Apple wishes to use lower entropy passcodes, necessitating the use of trusted hardware. Few have gone down this road, so I've detailed properties of a hypothetical design that could be free of manufacturer backdoors. Apple needs to either stop marketing their devices as secure against nation state attackers that can compel Apple, or actually build trusted hardware that's secure from manufacturer as I'm describing. Cryptography is not a "best effort" endeavor.
- zepto 11y agoAccepting your first paragraph as true, even though it is debatable - nobody has even done that with the kind of system you are proposing. Until they do, you are arguing favor of an imaginary system over a real one. It should be obvious that this is not an approach to security.
- mindslight 11y agoYou had asked what the current situation has to do with open/closed, and I detailed an open alternative. You're now arguing against my explanation as if it affects the main point. I didn't need to describe a better way in order to analyze the failings of Apple's system. If you aren't interested in an example of how open trusted hardware could function, then just ignore it and judge the system on its own merit. The fundamental flaw remains that Apple utilizes closed-design trusted hardware based on a manufacturer backdoor. This type of system is insecure against the manufacturer (and by extension USG), and marketing it as secure against such is willfully negligent.
- zepto 11y agoAs I said they have never marketed anything in the way that you claim. If you want to persist in that assertion, you should easily be able to point to the marketing you are refering to.
- zepto 11y agoJust to be clear: 1. Apple does not market their devices as secure against nation state attackers. 2. You have not detailed any alternative that is.
- mindslight 11y ago> Apple does not market their devices as secure against nation state attackers. Then what exactly is this court case for? > You have not detailed any alternative that is As I had just said, Linux's LUKS/dmcrypt. The authors simply do not have the power to unlock other peoples' volumes. Some other properties are traded off, but since you are unwilling to discuss system models ("completely hypothetical"), then I don't see the point of detailing the landscape. A system can have a vulnerability without needing a nearly-identical system without the vulnerability for comparison.