5 ms·

I think what you would do is make it computationally infeasible for them to create a separate network. The network exists independent of your individual secret

I think what you would do is make it computationally infeasible for them to create a separate network. The network exists independent of your individual secret and may be working towards the release of many soon to be dead men's secrets in a peer to peer fashion. Sure they could avoid the dead man injecting something, but then they would get hit with the full workload. If you coupled this with an economic incentive like bitcoin mining you could get the miners to allow you to drastically increase the amount of work required and make it keep up with the state of the art in technology. I couldn't say whether you can string enough primitives together to build something like this but it would be really cool if you could. Maybe I should go file a patent or something :-P

betterunix 13y ago
"I think what you would do is make it computationally infeasible for them to create a separate network." If the goal of the network is to release my secret when I fail to participate then there must be a way for the network to operate without my participation. What stops the parties from ignoring the messages I send, thus recovering my secret by simply pretending I died?
- arielweisberg 13y ago
  I think that is one of the novel pieces of functionality that needs to be created. The soon to be dead man looks like any other miner to the network and submits what appears to be valid work, but he is secretly poisoning the work going to release his secret, and of course he has to be the only one capable of doing that. Maybe later on the network realizes the work it was doing was poisoned and rolls back the poison change and resumes processing. Every time a poison pill is introduced the network must do some amount of work to determine that that the pill must be discarded and that amount is tunable by the soon to be dead man. "What stops the parties from ignoring the messages I send, thus recovering my secret by simply pretending I died?" They would have to control enough of the network and know who you are and how you are interfacing with the network to stop you. If the network has an incentive like bitcoin mining this could be infeasible for many adversaries. Tampering is a problem with real dead man's switches as well such as a script that you have to ping or an associate. I think big computational networks with incentives unlock some interesting doors. If you can assume that the computational majority of the network is playing ball you can ask it to do some interesting things if you have the right private key.
  - simbolit 13y 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
    - arielweisberg 13y 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." 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.

betterunix 13y ago

"I think what you would do is make it computationally infeasible for them to create a separate network." If the goal of the network is to release my secret when I fail to participate then there must be a way for the network to operate without my participation. What stops the parties from ignoring the messages I send, thus recovering my secret by simply pretending I died?

arielweisberg 13y ago
I think that is one of the novel pieces of functionality that needs to be created. The soon to be dead man looks like any other miner to the network and submits what appears to be valid work, but he is secretly poisoning the work going to release his secret, and of course he has to be the only one capable of doing that. Maybe later on the network realizes the work it was doing was poisoned and rolls back the poison change and resumes processing. Every time a poison pill is introduced the network must do some amount of work to determine that that the pill must be discarded and that amount is tunable by the soon to be dead man. "What stops the parties from ignoring the messages I send, thus recovering my secret by simply pretending I died?" They would have to control enough of the network and know who you are and how you are interfacing with the network to stop you. If the network has an incentive like bitcoin mining this could be infeasible for many adversaries. Tampering is a problem with real dead man's switches as well such as a script that you have to ping or an associate. I think big computational networks with incentives unlock some interesting doors. If you can assume that the computational majority of the network is playing ball you can ask it to do some interesting things if you have the right private key.
- simbolit 13y 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
  - arielweisberg 13y 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." 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.