The reason I don't like the safe analogy is that it is something physical, which the courts have repeatedly said is fair game. They can compel you to open a safe, they can not compel you to turn over a password. Something you have vs something you know. Which is why I also added the crypto example. The feds can not force me to decrypt my own crypto, I have 5th amendment protection against that. The 5th does not extend to something you have / are, which is why fingerprints to unlock an iphone can be compelled by the courts but they can't force you to turn over your password.
> The feds can not force me to decrypt my own crypto, I have 5th amendment protection against that.
That depends on what the gov't knows about what encrypted files you have. In the worst case, they don't need to ask you for your password, they just need you to decrypt your data.
In Re Boucher is the most famous case where the government ruled that the 5th amendment couldn't prevent the government from compelling someone to decrypt their data. On the flip side, Wikipedia has this short blurb on it's article on U.S. V Hubbell: "The Supreme Court ruled in favor of Hubbell. The Court held that the Fifth Amendment privilege against self-incrimination protects a witness from being compelled to disclose the existence of incriminating documents that the Government is unable to describe with reasonable particularity. The Court also ruled that if the witness produces such documents, pursuant to a grant of immunity, the government may not use them to prepare criminal charges against him."
Well, they can "compel" in the sense that they can demand the owner release it.
The owner can refuse and suffer the legal consequences.
In such a case, has there ever been a precedent of going to the safe manufacturer and compelling them to break into the safe? Have they ever then required all safe makers to make exploitable safes with master keys?
In the case where the safe maker has special ability to break the existing safe, there is plenty of precedent. Every application of the All Writs Act follows that analogy, including this San Bernardino case.
> They can compel you to open a safe, they can not compel you to turn over a password.
Note that they can, however, force a third party to turn over a password. The privilege against compelled testimony only applies to self-incrimination.
That's a tough analogy, because unbreakable safes aren't real and are probably impossible to make.
I prefer to look at the history of cryptography. Unbreakable cryptography isn't new. PGP is 25 years old. Classic techniques like one-time pads and book ciphers go back a very long time. Yet they never forced these things to be made insecure.
The government can listen, we can try to hide things, and they can try to un-hide them (with a warrant). Their ability to un-hide them depends on how well we do the hiding. It has always been this way, but they're not happy with it.
Unbreakable cryptography isn't real either. It's not impossible to reverse engineer a private key it's just impractical. It would require immense computing power and only yield that individual private key broken.
What the president is proposing would allow a key for law enforcement that would, presumably, work on many or all devices. Once cracked, it would give hackers access to all of those devices. This elevates risk considerably because it's no longer impractical to reverse engineer the encryption.
"If the key is truly random, is at least as long as the plaintext, is never reused in whole or in part, and is kept completely secret, then the resulting ciphertext will be impossible to decrypt or break"
That's not a great example in this case, because any useful system that incorporated OTP would have to store the really long secret keystream somewhere. That somewhere would be vulnerable to subpoena.
When that immense computing power is such that it would require turning the entire observable universe into computronium dedicated to the task, and running it for longer than the heat death of the universe, I feel comfortable calling it "unbreakable."
This debate is not about the direct ask imo, but more about the precedent that it would set and what the fbi could compel a tech company to do under the AWA. While in tech we tend to turn up our noses to the 'its always been done that way' mentality, the law takes a different approach, once you comply with the first request, it makes it harder, if not impossible to stop the subsequent requests. Which is why the fbi waited until something related to 'terrorism' to make this ask. I would be willing to be every last dollar I have in the bank that there is no useful info on that phone, and that if there were useful info, the nsa would have already had access to it.
To me, the really fun mental exercise is to identify when this would be acceptable. If there was a bomb about to go off and hypothetically the location was written on the phone, and only on the phone, would we agree that apple should break in? What if the bomb was a dirty bomb, what about a nuke? Now the slippery slope is where does that stop, a backpack bomb? Bomb making materials, texts to people about a bomb, what about a gun? What about a knife....
Honestly, it wouldn't take very much work for Apple to circumvent the lockout mechanism. It might even be possible just by commenting out a couple lines of code. The thing the FBI needs from Apple is the use of their private key.
Not with the new iOS, right? There's no Apple's private key. There's only your own private key. If you loose it - you're loosing your data and nothing can help you. I think you have a choice of backing up your keychain in iCloud. If you haven't backed it up there's nothing that Apple can do to access data - except by changing security design. If you backed it up on iCloud then I don't know - maybe the whole keychain is just protected by password, in which case they could try to brute force it or something. Not feasible if your password is too long.
Essentially there are two keys used to encrypt data on an iphone - one linked to the hardware (hardcoded and unique to each iphone, cannot be erased) and one linked to the software that's stored in NAND. After N attempts the NAND is erased and the key is lost thus all the data on the device is gone (one could try to brute force using a rainbow table style attack but at this point it's a lost cause because the amount of time it would take to decrypt the device). The NSA uses the old technique of copying this NAND to a backup and restoring it every time the device gets erased. This would give them infinite attempts at cracking the passcode with as little device meddling as possible
edit Apologies forgot to mention the passcode/fingerprint is tied into the 2 key process.
Ok, right, that makes sense. That's probably doable - making iOS being able to run auto-update when locked. Once for example. Update still needs to be signed of course.
It would probably make all crooks use long passwords for keychain, but who knows, maybe it would help in some cases.
If I'm not mistaken, the FBI wants a version of the OS that circumvents those limits and runs in RAM. Apple said it will take their team something like 2 weeks to build.
What they have to build and how long it will take them are moot points and detract from the overall argument which is can the gvt use the AWA to compel a 3rd party to create something that didn't exist before.
A box for which they can't make an individual key, only a master-key (at great cost of man-hours).