4 ms·

To prevent backdoor abuse, you could have a shared key controlled by various parties. A scheme such as Shamir's Secret Sharing - "An algorithm in cryptography

by w8rbt 9y ago

To prevent backdoor abuse, you could have a shared key controlled by various parties. A scheme such as Shamir's Secret Sharing - "An algorithm in cryptography created by Adi Shamir. It is a form of secret sharing, where a secret is divided into parts, giving each participant its own unique part, where some of the parts or all of them are needed in order to reconstruct the secret." Source - https://en.wikipedia.org/wiki/Shamir's_Secret_Sharing https://en.wikipedia.org/wiki/Shamir's_Secret_Sharing

gbrown 9y ago
Mathematically, you're correct. Unfortunately, real world systems have bugs, and even multiple parties can be compromised. Back doors are a bad idea.
- kefka 9y ago
  But that's explicitly not a backdoor. That's straight forward, "If permission is granted from 2 opposing parties, crypt can be reversed". And that would have to be added to the crypt on cyphertext creation. And the Bitcoin protocol has a similar thing, with escrow key permissions. Again, intended behavior, not some "super sekret backdoor".
  - AnimalMuppet 9y ago
    If I'm the encrypter, and I don't want the two parties to read it, that's a backdoor.
    - kefka 9y ago
      That's being rather obtuse. I can easily parse that the contents are encrypted to X keys. I'm thinking of something obvious like GPG here. I would accept the idea of a backdoor IFF the program encrypting hid the fact that it also encrypted to an escrow unawares to you, along with keeping that a secret.
      - AnimalMuppet 9y ago
        No, I'm not being obtuse. I understand what you're saying; I believe that I understand why you think it's reasonable; and I totally reject the claim that it is not a backdoor. I don't care how many keys there are, and I don't care what the underlying technology is. If a group of people, no matter how many (other than the sender and receiver) can choose, by a designed feature of the encryption, to decrypt the communication, that's a backdoor, by definition. And I assert that, no matter how carefully designed, a backdoor is always a bad idea. Having two parties that have to agree is a mitigating feature. It makes it less bad. It doesn't make it good, though.
crystalmeph 9y ago
I've thought of systems like this as well. To be clear, I do not want James Comey, Donald Trump, Joe Biden, or any "heroes" or "villains" of our power structures to be given any access to my encrypted data, whether they call it "lawful" or not. That said, I don't think the "all backdoor implementations will inevitably have bugs" argument holds water in all cases, so this battleground will move to a philosophical one based on reputation and the ideals we want to uphold. This means we can't just keep shouting, "you're too stupid to make a sufficiently secure system" at the government over and over again. So what would an effective and "secure" backdoor system look like? Say the manufacturer creates a device with a protection similar to Apple's Secure Enclave, where there is a key i burned into the system at manufacture, which is not accessible in any way after it has been programmed in. A second, independent, random key j is created at the same time, and the value i' = "i xor j" can be retrieved from the device with physical access only, e.g. by programming i' into a separate memory section inside the IC itself that is not accessible on the data bus while the system is running normally, it is only powered on by the hardware if one of the microprocessor's pins is jumpered at system boot. Now, to retrieve the value i, you need both i' and j, and you can only get i' with hardware access. So even if the database containing all the j-values ever created gets compromised, nobody can break into your phone without physical access. Now, of course, if the list of j-values got published, that would be a huge embarrassment for the US, but it would not mean that everybody's bank account and private emails would be accessible to every script kiddie overnight. This is not a 100% risk-free solution, but it is exactly the kind of so-called "balanced" approach that the government is going to try and sell us, because it does actually ensure that the gub'mint can't break into your encrypted data, at least without physical access. So the argument to actually use against Jim Comey is simply to point out that rest of the free world won't require this of their manufacturers, so nobody outside the US will ever buy an Apple product again since they don't feel any US manufacturer can be trusted, even if the system is technically secure. Also, any terrorist can still create an unbreakable crypto system by running custom software on a $20 Raspberry Pi, and good luck getting ISIS to cooperate with a lawful warrant for assistance in decrypting their operative's device.

by w8rbt 9y ago

gbrown 9y ago

Mathematically, you're correct. Unfortunately, real world systems have bugs, and even multiple parties can be compromised. Back doors are a bad idea.

kefka 9y ago
But that's explicitly not a backdoor. That's straight forward, "If permission is granted from 2 opposing parties, crypt can be reversed". And that would have to be added to the crypt on cyphertext creation. And the Bitcoin protocol has a similar thing, with escrow key permissions. Again, intended behavior, not some "super sekret backdoor".
- AnimalMuppet 9y ago
  If I'm the encrypter, and I don't want the two parties to read it, that's a backdoor.
  - kefka 9y ago
    That's being rather obtuse. I can easily parse that the contents are encrypted to X keys. I'm thinking of something obvious like GPG here. I would accept the idea of a backdoor IFF the program encrypting hid the fact that it also encrypted to an escrow unawares to you, along with keeping that a secret.
    - AnimalMuppet 9y ago
      No, I'm not being obtuse. I understand what you're saying; I believe that I understand why you think it's reasonable; and I totally reject the claim that it is not a backdoor. I don't care how many keys there are, and I don't care what the underlying technology is. If a group of people, no matter how many (other than the sender and receiver) can choose, by a designed feature of the encryption, to decrypt the communication, that's a backdoor, by definition. And I assert that, no matter how carefully designed, a backdoor is always a bad idea. Having two parties that have to agree is a mitigating feature. It makes it less bad. It doesn't make it good, though.

crystalmeph 9y ago

I've thought of systems like this as well. To be clear, I do not want James Comey, Donald Trump, Joe Biden, or any "heroes" or "villains" of our power structures to be given any access to my encrypted data, whether they call it "lawful" or not. That said, I don't think the "all backdoor implementations will inevitably have bugs" argument holds water in all cases, so this battleground will move to a philosophical one based on reputation and the ideals we want to uphold. This means we can't just keep shouting, "you're too stupid to make a sufficiently secure system" at the government over and over again. So what would an effective and "secure" backdoor system look like? Say the manufacturer creates a device with a protection similar to Apple's Secure Enclave, where there is a key i burned into the system at manufacture, which is not accessible in any way after it has been programmed in. A second, independent, random key j is created at the same time, and the value i' = "i xor j" can be retrieved from the device with physical access only, e.g. by programming i' into a separate memory section inside the IC itself that is not accessible on the data bus while the system is running normally, it is only powered on by the hardware if one of the microprocessor's pins is jumpered at system boot. Now, to retrieve the value i, you need both i' and j, and you can only get i' with hardware access. So even if the database containing all the j-values ever created gets compromised, nobody can break into your phone without physical access. Now, of course, if the list of j-values got published, that would be a huge embarrassment for the US, but it would not mean that everybody's bank account and private emails would be accessible to every script kiddie overnight. This is not a 100% risk-free solution, but it is exactly the kind of so-called "balanced" approach that the government is going to try and sell us, because it does actually ensure that the gub'mint can't break into your encrypted data, at least without physical access. So the argument to actually use against Jim Comey is simply to point out that rest of the free world won't require this of their manufacturers, so nobody outside the US will ever buy an Apple product again since they don't feel any US manufacturer can be trusted, even if the system is technically secure. Also, any terrorist can still create an unbreakable crypto system by running custom software on a $20 Raspberry Pi, and good luck getting ISIS to cooperate with a lawful warrant for assistance in decrypting their operative's device.