Hacker News Reader

Comment by btilly | Hacker News Reader

The automated version results in strong incentives for enemies of the USA to kill Snowden.

Therefore the version that he did is safer for him.

I don't understand why those same enemies are now disincentivized from killing him.

Perhaps the assumption is that the people to whom Snowden distributed the information wouldn't release the information (at least immediately) if Snowden were killed by an enemy of the USA. Of course, that's assuming a lot, including the fact that it would even be clear who killed him and why.

And how would these people be able to ascertain that it was US's enemies and not a black flag operation?

That's what I meant by my last sentence.

As someone commented on Schneier's site, US would have an incentive to keep Snowden protected because if any enemy of US would kill Snowden, then the information would become public, and I guess US wouldn't want that. Personally, I haven't invested a lot of thought into this, just wanted to point out an interesting angle. :)

Except that Snowden is not in the States, which presumably makes it harder to protect him on every step.

They are not disincentivized from killing him. They merely are not being incentivized by a guarantee that the USA will have lots of sensitive documents released.

It should also be noted that an active dead man's switch opens the possibility that surveillance of Snowden will establish the mechanism of the switch, allowing the CIA to take over the switch and then do away with Snowden with impunity. By contrast the passive distributed key storage he likely has provides no active traffic that current surveillance can trace.

He could also distribute encrypted versions to a bunch of confederates, and have the dead man switch send them the encryption key (but not release it to the public). That way he both has to be dead and the confederates have to be satisfied that the documents ought to be released before they are published.

Just curious here -- how would a dead man's switch be created?

I'm thinking something like a python script which is scanning for particular words and phrases on google news, like "Snowden killed", or "Snowden captured"? That seems like something I could build easily -- would a kill switch indeed be something as simple as that?

You could have a server running a timer that needs to be reset periodically and can only be reset by the person with the password.

How about a script that watches for such phrases (perhaps using Google Alerts), which sets off a timer for the release of the key. Then Snowden could disable it if there's a false positive.

The only problem is, that complicates the system with a critical point of failure.

My understanding is that's exactly what he's done. He's already distributed the encrypted documents, the kill switch will distribute the key.

Let's say he has N friends who have agreed to help. He doesn't want to allow any single individual to have the power to act alone and release information. And, he doesn't want to require all N of them to act together. What if one dies, is arrested, etc.? Could he encrypt the documents with, say, all combinations of three friends' own public keys such that a quorum of three friends would have to cooperate in the release? He would end up publishing N choose 3 bundles.

He doesn't have to make separate encrypted copies - all he needs to do is encrypt it with one master key, and split the master key between his confederates with a secret sharing scheme[0].

[0] https://en.wikipedia.org/wiki/Shamir%27s_Secret_Sharing

With the automated version if they kill him documents leak. How is that an incentive to kill him?

It's an incentive for enemies of the USA.

Aha. Yeah, didn't think of that. Yikes.

Ie: if you WANT the documents leaked, why NOT kill him (if you are not barred from doing so because of ethical or pratical considerations)

Perhaps he established both types.

Reply on news.ycombinator.com