Hacker News Reader

Comment by canoebuilder | Hacker News Reader

>I wish it were possible to make a service that lets you upload encrypted notes that will be made public after your death. But those two requirements conflict.

That's a very interesting idea. What do you mean by the "requirements conflict"?

Couldn't it be implemented with a dead man's switch where one would check in every year or so. Of course there would need to be some verification methods baked in to thwart pranksters.

View on HN

> What do you mean by the "requirements conflict"?

Possibly that such a dead man's switch would require the keys to be outside of your control to some extent - after all, if you die the day after resetting the switch, any infrastructure owned, maintained, or paid for by you might not exist for the switch to fire. If you give the keys to a third-party, you have to trust they aren't reading the contents and that their security measures are sufficient.

It could be done on an ad-hoc basis if you have a trusted third-party (close friend, for instance), but it's difficult to see how it could be scaled into a service. Perhaps it could be split into two parties - you upload the encrypted notes to the service, but they don't get the keys. You pass the keys to a trusted third-party (i.e. the close friend), who can make the notes public by giving the keys to the notes holder, who (ideally) would verify the death via public records. This avoids giving the secrets directly to the friend (in case of disagreements etc.), and you can reverse the decision whilst you are still alive (by deleting the notes).

Also, depending on the nature of the secrets, it does provide an incentive for death.

You could make a time triggered death man switch that needs to be reset every year. Else your secret is decrypted

But if the system can decrypt itself in your absence it must have access to the decryption keys. So you can't guarantee that the subsequent maintainer of the system won't release the keys after your death. I suppose you could hide the system somewhere (anonymous VPS) but then the encryption isn't actually buying you anything over what the anonymity is getting you.

You could use public key system to your advantage. Devise a cryptosystem that takes x time to crack, write a program to crack it, let it roll. If you're alive at the time of vulnerability being found, you can adjust or change the key.

Likewise it would be a good case for a proof of work system with specialized single machine input. You have to prove x amount of work every so often to keep the data encrypted.

There are no keys to launch infrastructure in Russian military. The few things amounting to that are the key codes stored in sealed containers that open code locks on control panels with physical launch buttons. The steel plate that cover the launch panel at Topol TEL can be easily cut open with a power tool.

Topol missiles have literally only 5 control inputs available to the launch crew:

1. raise the launch tower, 2. lower the launch tower, 3. arm the missile, 4. disarm the missile, 5. launch the missile

All missiles are programmed at the factory. Precise start coordinates are held by a separate navigation receiver at the missile TEL, and if the receiver fails, there are multiple redundant fallbacks.

The only way a crew can mess up the launch is to override the start coordinates with manual input.

https://www.deadmansswitch.net/

If the notes are encrypted, you hold the keys. If the service held the keys, no one would use it. But that means the notes can't be decrypted upon your death, since the keys would need to be uploaded at that point.

Simple solution: subcutaneous key implant. Someone has to physically cut you open to decrypt your notes.

Hopefully your attackers are not too determined... but then again, any system where your death results in decryption has that particular flaw. A bigger problem is if you die in such a way that your body can't be recovered.

I don't know if that's what he meant by "conflict", but it does provide an incentive for one's death

Reply on news.ycombinator.com