[Cryptography] Sadly predictable: Terrorism used as excuse to attack encryption

[Benjamin Kreuter]

On Tue, 2015-11-17 at 18:41 +0100, Miroslav Kratochvil wrote:
> To explain: Common people with reasonable operating systems/browsers
> are now using bulk encryption on every single HTTP request they make,
> on every single disk block they have, making SPF handshake with each
> person they IM, etc.. Observe that only a really tiny amount of the
> data is actually confidental

How do we even make that determination?  People do not think about
operational security when they use their computers and software is not
even remotely intelligent enough to automate the process (and think of
all the electricity such an AI would require).

>  business data,

Case in point:  how does my laptop know whether or not I am dealing with
business data?  That is why we use FDE -- even if most files are not
confidential, there is no way people are going to remember to encrypt
truly sensitive files, so we just encrypt it all to leave no room for
error.

> Think about what bulk encryption means for the consumption of computing
> power (RSA ain't free, I'd actually expect more than gigawatts).

I would expect it is nearly negligible compared to what is being spent
on other, far more questionable things.

> Think about what it means for law-enforcement agencies -- they can't even
> simply prove that given single user is _not_ a suspect

In my country, the police have to prove that someone *is* a suspect
before they are allowed to conduct surveillance (or at least that is the
theory; in practice there are tons of exceptions and violations, *which
is why we need more encryption*).

> I'd prefer something less drastic before the ban comes, like forcing
> the user/software selectively choose (by some smart API or a correctly
> designed UI) what to encrypt, leaving the rest (most) of data
> "ecologic" and "law-enforcement friendly".

Forcing users to choose what should be encrypted is the definition of a
bad UI.  Users are not going to think about what needs to be encrypted,
they are just going to use their computer to accomplish their tasks with
the minimal effort necessary.  If you force the choice (think Windows
Vista), you'll wind up training users to click "no" (after all, most of
the time, there is no reason to encrypt), and then when it matters
you'll see tons of mistakes.

> PS. In no way I suggest simply "turning SSL off", but there could be a
> way that just authenticates the data without doing encryption.

Who is going to decide what requires authentication and what requires
encryption?  Getting programmers to use TLS *in the first place* is like
pulling teeth.  You want to force programmers to decide whether the data
needs authentication or encryption?

> Method for easily marking the "secret bits" of the stream would be cool as
> well.

It is easy to predict the result:  most developers will mark nothing as
secret because they are too busy trying to get their software to work
correctly.  Again, just getting people to use TLS is an uphill battle;
adding more complexity is completely counterproductive.

-- Ben
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part
URL: <http://www.metzdowd.com/pipermail/cryptography/attachments/20151117/738486b0/attachment.sig>