|
|
|
|
|
|
by simbolit
4635 days ago
|
|
|
Your proposal is no better than the 3 letter server farm working alone. If the poisoner is not inside the wealthy attacker, then the attacker knowing this scheme is better off not cooperating with anyone, including you. Even if that were not the case, your scheme relies on your identity being a secret, which essentially is "security through obscurity". We have to assume that a dedicated attacker knows no greater incentive than breaking the encryption. Anything else is foolish, IMHO. Edit: clarification |
|
|
Having the attacker be better off alone is the point.
First of all your assuming that the 3 letter server farm is your adversary. This approach has utility even if some adversaries (which don't happen to be yours) might have enough compute power. The approach has utility if it can ratchet up the difficulty enough that the adversary you care about can't trigger early release. It also means you don't have to maintain any infrastructure that is traceable to you. If an attacker can find and disable the switch before it triggers what good does it do you?
Another advantage of this approach is that you can ratchet up the difficulty to the point that the 3 letter server farm can't complete the task in a reasonable amount of time. I suppose there is hard upper bound in the sense that the network is only going to be a few orders of magnitude more powerful than the 3 letter server farm.
Another assumption is that an adversary finding out the secret and releasing it is the worst case. It really depends which adversary right? One thing you are trying to influence is public distribution, and the people with the incentive to apply the compute power may not be the people with an incentive to act on the information by distributing it publicly. The secret is just going to be an encryption key for decrypting the payload you have put out through other means. It's not black and white whether the utility is zero.
In summation, it's a dead man's switch, and the important part is that it fires if you don't do your thing. There are a lot of other desirable attributes, but real world dead man's switches share many of the same flaws.
"Even if that were not the case, your scheme relies on your identity being a secret, which essentially is "security through obscurity"
I don't actually follow why anonymity is necessary. Dead man's switches are usually because you fear you are not or may not remain anonymous. Anonymity is of course a desirable attribute for a lot of reasons, but if it is blown this approach would still work. Only the person possessing the secret key can generate the poison pills that delay the release of the secret.
Tor is an example of ways to attempt anonymity while participating in a network. I view that as a separate problem space.
I think the real competitor would be a switch that actively monitors a piece of state that has plausible deniability, but that runs afoul of the whole running infrastructure traceable to you thing.