Re: Prosecutor cannot compel disclosure of encryption keys?

David Wagner wrote:
Pubkeybreaker wrote:
Anyone smart enough/knowledgeable enough to claim "just random data"
isn't going to be found out by "forensic investigators".

First of all, very few people are going to go to the bother of doing
everything that is necessary to foil a forensic investigator.

Second, your claim is by no means obvious. There are a hundred and one
ways to get tripped up in a lie; the forensic investigators only have
to find one, whereas you have to anticipate them all.

Here's another example: Suppose the investigator says "Oh, ok, you've
been using this in experiments. By the way, tell me how you use that
random data? What's the name of the program that reads in that data?
When did you use it in your experiments? Be precise." Suppose you tell
the name of your program, but later the investigator looks more deeply
at the atime on that program's executable file and discovers that is
has never been used, or its usage time doesn't match your statements,
or that there is no record in the audit logs (or the registry, or the
temporary files, or you name it) of you ever using that program at the
times you mentioned (even though every other program you used can be
found in the audit log).

Now I know what you're going to say. You're going to retort with some
method you've devised that will prevent the investigator from using the
atime, or whatever, to catch you. But that misses the point. There are
many tricks an investigator might use to try to trap you into telling a
lie that can be disproven. I have no doubts that for any trick I tell
you about, you will be able to come up with a countermeasure. But will
you be able to come up with a way that protects you from all possible
traps, without knowing in advance what tricks the investigators may use?
Especially given that you are not an expert in forensics, and it only
takes one trick to catch you? Are you absolutely certain that you know
Windows well enough to know every way that Windows might leave some trace
of what you've done in a place that a forensic investigator can find it?
And even f you do come up with comprehensive countermeasures, will most
folks have the discipline to follow them every single time?

Also, your emphasis on "proof" is misguided. The legal system doesn't
work by mathematical proof, or by proof of any kind. The credibility
of witnesses is assessed in a highly subjective way. And people are
surprisingly good at assessing the credibility of witnesses. It's like
the mother who asks her kid "Did you take a cookie from the cookie jar?",
the kid says "You can't prove a thing!", and the mother says "Don't lie
to me. You're grounded!" Proof is besides the point.

I think you are overestimating the effectiveness of the "gee, officer,
those are just random numbers" ploy. The judge is a person who is used
to dealing with liars and cheats. When an investigator is able to supply
circumstantial evidence that suggests you may be trying to pull a fast one
on her, she is is not going to be amused -- and I wouldn't rely too much
on "but you have no proof!" to keep you out of jail. People often have
a way of knowing when they're being mocked, even if they can't prove it.
Maybe you'll get away with it -- but maybe you won't. I wouldn't count
on it.

I think brilliant mathematicians and brilliant cryptographers are
particularly vulnerable to fooling themselves in this way. They're used
to knowing they can come up with a cipher that they're confident no
one can crack; but that's not the same thing as telling a lie like this
that no one can uncover. And they're used to thinking of themselves as
smarter than else in the world (which, when it comes to their mathematic
skills, is almost certainly accurate -- but which is often not accurate
when one looks outside their speciality), so there's almost a smirk of
"You can't catch me!". Whatever you think of law enforcement, it is a
mistake to gamble your freedom on their incompetence and ignorance.

I would also add that even if you do manage to cover up all traces of previous crypto use, then you have the problem of repeating the effort every time you use encryption - very tough indeed.

Relevant Pages