Preface:  As a longstanding policy, whenever I buy a new hard disk or 
decommission an old one, I immediately `dd` it from start to end with a 
pseudorandom byte stream.  The result is indistinguishable from my disk 
encryption setup, which leaves no apparent on-disk headers.  I don’t do 
this for “plausibility” reasons, but rather, 0. to assure that 
immediately upon use, any sectors written with disk encryption cannot be 
distinguished from unwritten sectors, and 1. to make things overall more 
fun for potential cryptanalysts.  I do realize the small problem that I 
can’t affirmatively prove any particular disk in my possession to *not* 
contain decryptable data; and many of them don’t!

(I think that next, I may start writing my disks with headers for LUKS, 
which I do not use...)

Whereupon, I challenge plausible deniability designers to `dd` a 6TB 
disk with pseudorandom bytes, then try walking it across the U.S. border 
until it gets searched.  What could possibly go wrong?  Should you be 
ordered to decrypt it, the disk *could* be *plausibly* filled with 
pseudorandom bytes; and you would not be committing the crime of lying 
to an officer, when you truly state that in fact, it *is* filled with 
pseudorandom bytes.

Please, I want to see this “plausible deniability” theory in action.  
You owe it to your users to test the theory empirically, in 
circumstances in which users have here reported applying it.

Now, in reply:

On 2018-01-13 at 02:11:08 +0000, Damian Williamson 
<willtech@live.com.au> wrote:
>The same problems exist for users of whole disk encrypted operating 
>systems. Once the device (or, the initial password authentication) is 
>found, the adversary knows that there is something to see.

Or PGP.  Or in a broader sense, Tor.  Or in the physical world, a 
high-security safe bolted to your floor.  Security systems attract 
attention.  Smart people develop appropriate threat models, keep their 
security systems confidential where it is practical to do so (don’t brag 
about your high-security safe), and work to increase the popularity of 
network security systems (PGP, HTTPS, Tor...) to reduce how much they 
stand out.

In the context of this discussion, it does help that Bitcoin is becoming 
popular.  It would help much more if Trezors and similar devices were as 
commonplace as iGadgets.  But when considering the potential threats to 
any specific individual, the only “plausibility” shield is to not seem 
like someone who is likely to have *much*.  Of course, this is not a 
problem specific to Bitcoin.  Depending on the threat, the same danger 
applies to owning a substantial amount of gold, cash, or even money in a 
bank.

>The objective of plausible deniability is to present some acceptable 
>(plausible) alternative while keeping the actual hidden (denied).
>
>If the adversary does not believe you, you do indeed risk everything.

And therein lies the trick.  Unsophisticated adversaries such as common 
criminals may be fooled, or may not care if they can quickly grab 
*something* of value and run away.  But if your threat model may 
potentially include any adversaries possessed of both brains and 
patience, “plausible deniability” solves nothing.  Such an adversary 
will not likely be satisfied with the standard of “plausibility”.  More 
likely, the prevailing standard will be:  “I wasn’t born yesterday, and 
I *know* that you are hiding something.”

>[snip extended prior quotations]

-- 
nullius@nym.zone | PGP ECC: 0xC2E91CD74A4C57A105F6C21B5A00591B2F307E0C
Bitcoin: bc1qcash96s5jqppzsp8hy8swkggf7f6agex98an7h | (Segwit nested:
3NULL3ZCUXr7RDLxXeLPDMZDZYxuaYkCnG)  (PGP RSA: 0x36EBB4AB699A10EE)
“‘If you’re not doing anything wrong, you have nothing to hide.’
No!  Because I do nothing wrong, I have nothing to show.” — nullius