Search Mailing List Archives


Limit search to: Subject & Body Subject Author
Sort by: Reverse Sort
Limit to: All This Week Last Week This Month Last Month
Select Date Range     through    

[liberationtech] CryptoParty Handbook

Jacob Appelbaum jacob at appelbaum.net
Tue Oct 9 05:46:43 PDT 2012


Bernard Tyers - ei8fdb:
> 
> On 8 Oct 2012, at 23:46, Jacob Appelbaum wrote:
> 
>> Asher Wolf:
>>> The argument everyone is politely avoiding - while pondering the 
>>> numerous ways CryptoParty will expose already compromised
>>> individuals - is whether the masses SHOULD use crypto.
>>> 
> 
>> I'm not ignoring it and most of the world has been using crypto
>> for online stuff since SSLv2 was released to the world.
> 
>>> Rain-check: it's happening - or at least, the users are are
>>> trying - regardless of whether they're are doing it right, or
>>> regardless of whether more experienced ppl are willing to offer
>>> their advice or not, and completely separate to the
>>> philosophical, technical and security related-discussions that
>>> are currently swirling.
>>> 
>>> Basically: hello crypto, the users are here.
>>> 
> 
>> I'm sorry to say it but a lot of the users have been here for a
>> while - most people that use crypto just don't know they're doing
>> it. Ironically, if users don't get good advice, they'll just be in
>> the same spot - thinking they're safe when they're not - all over
>> again!
> 
> Yes, the users have been here for a long time. They are hanging
> around in the corner trying to talk to the guy who just seems like
> such a dick, but in fact is just unable to talk to people who don't
> speak his lingo, and he is not interested in talking to them.

I'm not into your analogy but I understand it.

> 
> Ultimately some people will want to talk to him, and other won't, but
> at least get over the awkward introductions.

Are these the users using HTTPS? :-)

> 
> 
> 
>>>> From experience, most of the non-tech ppl who attended
>>>> Melbourne's
>>> Cryptoparty had previously attempted to install various tools by 
>>> themselves and had either (a) failed (b) installed them
>>> incorrectly (c) couldn't figure out how to configure them (d)
>>> given up 'til now.
>>> 
>>> CryptoParty is essentially the user saying: We are working
>>> together to trying to figure out how to do it better. We need
>>> these tools.
>>> 
>>> Whatever the best-practice model actually is, it'll be
>>> crowdsourced, because people are unwilling to wait for easy
>>> 'crypto manna from heaven', offered up on a plate.
>>> 
>>> And frankly, the users have much to learn from the crypto experts
>>> and it'd be a damn shame if knowledgeable people refused to teach
>>> or share their expertise because ppl are "doing it wrong."
> 
>> I think that the real changes belong in the platforms - anything
>> that requires configuration is probably already doomed to fail and
>> screw a user. That's generally the approach that I've seen work -
>> everything that requires 0) user education and 1) realistic honesty
>> about threats or risks results in 2) denial or mistakes or a
>> bork'ed tool shooting the user in the foot.
> 
> While you've probably got a really good reason for saying 0) and 1)
> above, I'd like to hear them, just because I used to think similar
> things as 0) and 1).
> 

A few mail clients would ask you for your username/password - the user
would type them in and then the mail client would be setup.

Sadly - those mail clients didn't use crypto and from the very first
step - they were owned.

I advocate that no user should ever need that education (hit cancel,
setup the account by hand manually, etc).

Realistic honesty means that people who fall back to saying "well, I'm
not worried anyway" or "I have nothing to hide" need our support. These
kinds of justifications are part of a coping mechanism that kicks in as
a result of user education, pride or often, shame.

> The more I have been reading about [NORMAL USERS] the more I think we
> (the people who know about the complicated shit) are not going to
> save them all. It may sound disingenuous to say but, there is no
> point trying. Humans are humans. Getting people to think themselves
> about these topics it a lot more successful than giving them tools
> and hoping they use them.
> 

This is why a platform fix will help every single user who decides to
use it. When people use Tails - I think they're way better off over say,
an old Windows system with Tor Browser for anonymous browsing.

> I'm surprised to hear you say everything that requires user education
> fails. I (think) I know your reasoning.
> 

The key is that a user may learn to click cancel in the above example
but I don't think they *should* be required to do so to be safe.

> 
>>> 
>>> We've known we've been doing it wrong for a long time now and
>>> going back to Facebook to organise is no longer an option.
>>> 
>>> The creation of CryptoParty was a spontaneous, viral storm. It
>>> was NOT a concerted, centrally-organised campaign, with funding
>>> or even a best-practice model. My hope is that experts contribute
>>> to eventually provide a best-practice model, and that users give
>>> the necessary feedback allowing for tweaks in tools and creation
>>> of more accessible crypto.
>>> 
> 
>> Since clearly a few loud people were bent out of shape by my
>> comments - they have no idea that I encouraged you or tried to help
>> out; so let me set the record straight: go you!
> 
>> I think it is *great* to make the book and I think it is great to
>> do it with a set of unifying principles - it will help to ensure
>> that good stuff gets into the book and crappy stuff stays out of
>> the book or is so noted as crappy or contentious. I think that
>> means that peer review is essential before rushing to publish.
> 
> (This may sound insulting, but its not meant to.)
> 
> Absolutely 100% completely agree. It should have been peer-reviewded
> from the start.
> 
> It was (IMO) wrong, and slightly misguided to write a handbook which
> is ultimately about a complex and technical set of topics, without
> having some expert advice or input. That is not to insult the work
> the people did. I just think it was executed incorrectly.
> 
> To continue my previous self-bashing thread, technical people need to
> be more easier to deal with. We need to be more intelligent in how we
> explain what are complicated subjects.
> 
> Saying "data confidentiality, authentication, threat-model,
> non-repudiation," to most people means squat. As someone explained,
> "every time you walk across a road, you make a
> threat-model/assessment" makes more sense.
> 

I think the term "safety plan" or "evacuation plan" or "legal strategy"
are often used in different communities. The first is very often used in
the domestic violence world - they don't talk about threat models at
all, not with that name anyway.

> We also need to understand that most people don't really care about
> the super complicated stuff thats involved in crypto/security.
> 

That is why a platform (or an application) is a nice place to fix things
- users learn "TextSecure is safe for x uses" and "WhatsApp is not safe
ever" without learning about ECDSA. :) It is of course good to write a
guide that tries to explain the details - like a manual of sorts. But if
it doesn't work by default and the manual goes unread, users are screwed.

> On the other hand, to bash the people who are not crypto/security
> experts, they have to understand that the subjects are involved are
> complicated, not easily reduced. They also need to understand the
> implications of giving wrong/bad information.
> 
> They also need to know when something the technical people say is not
> just hot-air, or BS.
> 

That's an epistemological point, I think. It is lost of many people and
it is a core aspect of Real Literacy - believing a thing on faith is
dangerous but not knowing that an assumption of faith is involved is deadly.

> 
> One giant PDF for this cryptoparty handbook was the wrong approach.
> It should have been Git or a closed wiki. It should have been broken
> down into chapters and had people who know about the topics working
> on it too. Sure it would have taken more time, but it would have
> saved all this thrashing about.
> 
> Git is not something most people would want to use (or indeed heard
> of) but for an editing process its a good option. There should have
> been a change log all the while.
> 
> When its finished, or ready for purpose, then publish it in
> PDF/mobi/whatever.
> 
> 
> Learning how crypto / security works is not rocket science. Anyone
> can learn the approaches used, the right or wrong ways to do
> something. You may not become a leading developer of it but thats ok.
> Not everyone will.
> 
> People can learn how to install PGP, Enigmail, the Tor browser
> bundle, Truecrypt, but if they aren't educated on how best to use
> them (to think for themselves) then the cryptoparty is wasting their
> time.
> 
> I was present (and helped out a bit ) at the Tor talk the London
> crytoparty and it was wonderful to see discussions with the
> "non-technical" (I hate saying that but I don't know what else to
> say) people who attended to Tor talks. They were thinking about
> things. They just needed some help.

Indeed - most of what they want is intention based - imagine the
surprise of people when they find out a phone can be tapped at all. It
is a largely inconceivable problem - even Justices of the US Courts are
clueless and they grant wiretapping permission!

> 
> It was clear to see there were people present who had thought about
> these tools. Their issue was not "Should I should use this?", it was
> "How the fuck do I know when/where to use it?"
> 

I think the answer is - always and by default - if it was in the
platform by default, we'd see apps on other platforms for interoperability.

> 
>> I really encourage you to put in a few chapters about the
>> following:
> 
>> social and technical compartmentalization targeted exploitation
>> realities (from Core Impact to Metasploit) threat modeling 
>> intention/goal based risk analysis physical security risks 
>> practical information on real surveillance/censorship systems 
>> getting involved going from a user (to a translator or...) to a
>> developer outlining the currently missing tools that we need to
>> build
> 
> 
> The list of topics you added above, to me, are topics that show
> cryptography and security is not just about whizz-bang technical
> stuff - they are about looking at a subject critically, with rational
> thought and making decisions based on logic rather than trust or
> "it'll be ok.

Indeed - the name of the things doesn't matter either - what matters is
the content. I'd suggest talking about the many names present in our
world for such concepts.

> 
> Once you understand the threats, then you yourself can come to a
> decision as to what tool is right for the job.
> 

Part of the debate is that some people don't get that warrant-less
wiretapping, a total lack of mobile privacy and insecure software are a
threat to society itself. We need a network wide impact and sometimes
that requires changes in the platform where we don't blame the user
(ever) for not understanding the threats. The threats are artificial and
constructed by the (telecoms and the state) people producing "solutions"
to problems.

All the best,
Jacob



More information about the liberationtech mailing list