[throwhauser]

If you're being attacked with a wrench, wouldn't you want to avoid deniable encryption?

If there's no way to 100% establish that all the money has been extracted, an attacker might keep going indefinitely to see if there's more.

[littlestymaar]

In fact, since there's no way you can prove that you haven't used deniable encryption, you'll probably be in a really bad place anyway.

That creates an interesting game theory situation though, where nobody has any incentive to disclose anything, since it wouldn't change the outcome anyway, which ends up negating the whole point of torture: the victim needs to believe that the tormentor will stop if they disclose the truth.

(Unfortunately, the real world isn't a game theory problem…).

[dane-pgp]

This is the game theory that the Rubberhose file system (co-invented by one Julian Assange) is based on.[0] It's a pity the blog post didn't link to that article, and only linked to the one about rubber-hose cryptanalysis, since this prior art does seem to overlap significantly with the scheme that the post is proposing, as does the Owner-Free Filesystem[1].

Anyway, you're right that the real world isn't a game theory problem, but I do think that if someone is faced with being tortured for information, they should at least attempt to ask the torturer "How do I know that you will stop when I give you the information?". Or, perhaps less incriminatingly, "I don't have that information, and it doesn't matter because you'll keep torturing me regardless".

You may not be able to convince the torturer to give up on the torture (much less convince them to let you go free), but you might at least be able to convince yourself that there is no point talking or trying to come up with a lie. Having said that, it's also instructive to look at the example of Marcus McDilda who was tortured by the Japanese for information about atomic bombs, about which he knew nothing.[2] His lies may have saved not just his own life, but millions more.

[0] https://en.wikipedia.org/wiki/Rubberhose_%28file_system%29

[1] https://en.wikipedia.org/wiki/OFFSystem

[2] https://en.wikipedia.org/wiki/Marcus_McDilda

[syrrim]

You can prove you haven't used deniable encryption. Encrypted data looks random. As long as your disks and files don't contain unexplainable random sections, there must be no encrypted data there. Steganography might have been used to hide the encrypted data in otherwise meaningful data, but that is a separate concern.