All right, well, I think we'll hear another somewhat technical talk about the Open True Crypt or rathe The Open Crypto Audit Project.  You guys have already seen if you use true crypt at all that there has been a little bit of a change with that project and now it's something I think we're all interested in, whether you are using it to do holistic, protect your own data or if you do things like me and do forensics and need to transfer data and need a secure, trusted way and easy way to get muggles to understand how to use crypto. Crypto usually served that purpose for me so this is a pretty important project and well we only have half of the original slated speaking team. 
I think Mr. White will do a great job for us, so let's give him a big party track welcome!
(Applause)
	>> Go Hello, DEF CON.  I want to send regrets for Matt, we thought up until about a week ago that he was going to be able to come. Unfortunately he wasn't able so it's me.  Hopefully we'll have a good lively discussion.  I have quite a bit of material and also really great swag for questions at the end.  The idea is I want to tell a little bit about the history of what we've done, how we got here and kind of the community aspect.  There is a technical element here.  I'm a security engineer, not cryptographer so we'll do a reasonably deep dive into some of the technical material but I'll be the first to raise my hand if I'm cheated by my weight class here so here is what I'd like to do.  Sort of go over some basics for security engineering.  This may be old hat for people in the room but sometimes I think people go through this cycle in their career where their foundation for ideas and sort of guiding ways to think about things that are kind of memorized but not really applied, takes a while before we start to see the elegance in I guess simple ideas. 
I want to talk a little bit about revelations, specifically around crypto and how that impacts and then obviously what people are here is for true crypt story because you can't make this stuff up and then a little bit about our project and people that are behind that.  
	Then as we go.  So just couple things about me and Matt.  I have been around the scene for a while but I haven't done tons of presenting.  My first DEF CON was about ten years ago.  I have a weird background, I just confuse recruiters in HR departments and so a lot of training in signal analysis, lot of work in brain and biomedical imaging and sort of distributed systems.  I spent a fair amount of time working on cardiac safety, ECG analysis and in that world, people think a lot about secure systems, right?  When you are evaluating medication to see if there is tiny microdifferences in heart wave rhythms, the kind of principles one applies to software assurance is a really different class.  Have done a fair about of software assurance. More recently I've been working on public cloud security so location systems for Amazon and Google Compute. 
Most recently with the OCAP project, we're working with the Linux Foundation core infrastructure initiative.  We'll talk about that and then on the private side I have some disclosures in terms of NGOs, so doctors without borders, couple of other familiar ones and I'm doing some work with them.  
	So, yes, I like to work on strange signal problems like ECG and direct safety.  Little bit about Matt.  Most people know him here by reputation but for the record his background, research interests, not his Dachshunds but he has pictures of his Dachshunds.  He does not have pictures on the website.  
	These are my two hounds, I was telling Jack Daniel I think I was channeling him.  Just to give you a little perspective on getting here, I think that yeah, so Wednesday drove across country in the wagon with two dogs.
	>> Would have!
	(Laughter)
	>> Landed, unpacked, kissed everybody good-bye and hopped on the plane.  So it's been a good journey.  This is the spirit of what I want to frame things.  I think we have a lot to learn from lots of different levels of technical background and this quote from Jack I thought was really apropos.  
	So if you have seen these I think they are called 10 rules of security engineering.  Scott Culp has a great quote and I want to emphasize this a little bit, too, because we can get so caught up I think sometimes in mechanics of these projects that we have to keep things in perspective.  I added a slightly minor addition to that which is "Even with this crypto, we have got to keep an eye out."
These are classics I think people generally know these but people that may not be familiar with security engineering.  
	So this particular one is more I guess for an Archeological statement, I think there's no one in this room that is all aware of these but if this deck turns up sometimes with the future, we'll have more context of what we're doing.  But crucially around the mass surveillance.  What drew a lot of attention in the academic crypto world were the revelations around bull run and intentionally subverting the processes and committees around our critical security systems.  EFF has a fantastic timeline so those are the entire spectrum.  
	So besides the sort of sociological and political pieces, I think a lot of people kind of woke up when this came out and said we're actually going to be this entire body in existing of cryptographic work. This was late fall of last year. 
Presence of committees implied that you know these were serious issues and then eventually with the deterministic, random regenerator, one of the official standards was just pulled so which brings us to TrueCrypt, so I think people generally know but it's a piece of software for file volume and full disc or whole disc encryption.  Sorry.  It's been downloaded about 30 million times, been around for about ten years.  In the open world, there is -- it's literally listed as Fedora and, Debian, umm, license archives  as forbidden items, you have to go out of your way to include it and we'll talk about that in a second.  
	But it's you know it's a really popular tool.  It's used by quite a lot of people.  Not just human rights workers and activists but corporations, too.  This is a snapshot from this morning, so if you want to get bulk data in and out of Amazon Web Services this is what they officially support.  So this stuff is out there.  The problem is looking at trusted systems and sort of the history of things we realize late last summer, maybe this wasn't news to some people but for many it was, the TrueCrypt network may have been audited, you see a lot of things with tech press, statements about maybe this is about compliant or maybe someone adopted it for a PCI thing or whatever, but if you cut through all that, it turns out at least through last fall there had never been a form of cryptanalysis on TrueCrypt, so people are relying on it in some cases it was basically activists putting ... there were huge stakes riding on this. 
	There were issues around the different versions of whether we got with the binary or the source, volume set of differences in the different platforms, originally by the way Windows project and it was eventually imported to Mac and Linux.  
	Bruce and Colleen a few years ago did a formal cryptanalysis on the deniability aspects so there's an idea one can create a hidden volume and then within that or adjacent to that you could also have a volume that is known so you give up your keys or password and the idea is that you can still have secure information and plausible deniability.  It's not a great  umm … the implementation was not that strong.  
	Around the time it was kicked off, just a few weeks after the project began, Xavier did an amazing job through his blog and look at the process in some cases seven and 7 or 8-year-old RSA headers and fairly obscured dependencies without which you just cannot compile this code on Windows.  The last license reviews from 2008 that attorneys from Red Hat and fedora and OpenSource groups… would have said language in this license is so unconventional we can't have any -- we can't make any assurances about litigation and just said it's not a free license.
	So there are some really, really strong elements of this.  This is from what's it called (inaudible) Jeremy's and (inaudible) company that does these monster cracking systems.  This is 8 Radeon 290X's, so the numbers on the right are hashes per second so straight, you know, UNIX, descrypt, 952 million per second, TrueCrypt with this particular chain together, 17,000.  So if the as word is strong, at least by cursory measures, you know, that's pretty interesting that the contenders is OSX, by the way, that's as I understand it actually adjusted on a permissioning basis or calibrated to your CPU.  So it's at least from people who have looked at this initially there's not… there doesn't seem to be obvious bypasses for this in terms of raw brute force.  
	One other thing that is interesting about this is the development team. If you just go on Wikipedia there's a ton of information out there, right.  So what we know is some of the original developers were interviewed, these things are on multiple gethub repositories.  There are trademark filings, and in the UK, France, China, Czech Republic, U.S., and it was set up to be an entity about a year and a half, maybe a year after the project kicked off.  It was a piece of software called encryption for the masses and you could do a whole talk on that.  But only to say that there was some controversy around how clean the code was and why it was included and there was very early on some sort of fairly serious threats of litigation against the TrueCrypt developers.  I think that in combination with what was very common at the time which was sharing and download sites would sort of highlight that this is TrueCrypt or whatever the package was, and then maybe there's bundles or other things. 
In a way, there is almost like a brand dilution.  Like people weren't really clear what the official sources of the code.  To be fair, for the first year and a half, it had sort of a you know had sort of a quirky source.  This was created, a non-profit organization, so therefore there are public IRS filings, there is the original announcements that are still archived.  Here is the thing a lot of this information is sort of hiding in plain site. So some people on this first day or second day when it was announced are still quite active on Usernet with the same e-mail addresses.  We did a little digging and there are some published academic papers with colleagues and there is actually acknowledgment and a few dissertations and thesis that say no, thank you so and so for working with me on this and on TrueCrypt.  And so the information is out there.  
	But some things were not at least I won't be quite as comfortable in sharing, why is that?  Well, remember this, right, everybody get a good laugh and a few days later this story sort of died down, the official mailing list or form or whatever, Dorian Nakamoto, people forget it was a weird story and not the best journalism but a lot of people forget this part.  Within hours of that story breaking, Dorian's house was surrounded in this and I think crucially this:  He had some serious health problems as a result of potentially serious financial problems, his life was turned upside down.  All that is just a way of saying there's a huge amount of information already out there, endless discussions and forums and things where as I said, these public records can be obtained.  I don't think it benefits anyone to try to a docsync I want to leave the last 20 minutes or so for questions so we can talk more about that if you like. 
	Back to the code.  So this was I think attributed to (inaudible) …shallow, well, in the crypto world, one thing that I have come to appreciate is it's a very, very small community.  Security engineering is fairly small anyway.  But cryptographic engineering is a very different beast.  Cryptography engineering, it's a very very special narrow area.  
	So it's not the case that with just lots of people looking, bugs will be found but some of the subtle, you know, cypher chain and implementations for random generators, you know they will be missed.  I mean, there is value in the open community and there is value in all that.  But it's not the same as really really you know highly skilled, highly trained crypto engineer.  I love this picture.  This is from 1920, Sam Reshevsky, he beat 14 chess masters at the same time.  So, I mean, the point of this is there are some skill sets which are just orders of magnitude different than others.  
	So we talked and we said, you know, this is important software.  People seem to be using it.  We have people on the project using it.  We really didn't know a think about it.  
	So let's do it.  
	So there was a conversation on Twitter with Matt and I, I sort of announced it, Matt threw some money.  We were contacted by a group called FundFill and I didn't realize at the time but they were just starting out, startup company.  And then anyway this went out so Matt noodled on it a while.  Couple weeks go by and he was like we need to talk about this.  There's something here.  So we blogged about it and it hit the front page of Why Comminator and then the Internet showed up, umm yeah.  So we were assuming the conversations would be around the code and analysis but what I didn't fully appreciate and what I now understand is the enormous amount of work is sort of the community piece of that.  We advertised rather, we sort of promoted the idea that you know this, here are basic goals, basic cryptanalysis, some bug bounties would be a good idea. 
The dream would be if we could raise enough money to have one of the few security engineering firms in the world who not only have the skills but whose reputation is with people we trust and are likely to say, yes, this is credible.  So there was no shortage of people wanting to help, no shortage of firms stepping up saying we have a free static analyzer or whatever.  
	There is a very small number of professional firms that do this sort of work.  
	Then set up a thing on indigogo and he says well maybe we can offer people T-shirts if they get to a certain level or DVDs and sneakers and things and, wow!  So, yeah, in the first, arbitrary number of days for a campaign.  We set it for 60.  This is day one.  Like in the first four hours, okay.  We really struck a chord here.  When we set a bitcoin you have to understand this was nation mechanism, sorry… this was the history through the year.  Right.  So coming out here.  So this is April and the project started around here.  And we could talk about the speculation and the market price of Bitcoin, but when people are sending literally $5 donation and $1 donations and you know really sweet elderly ladies from Yorkshire are asking for an address to send a check if they are not really online, you take a serious responsibility. 
So in that context, we are taking Bitcoin and this starts happening.  You know, we're not like some investment house, so people there's -- I don't know if it makes sense to hang on to this.  For all we know this was the monster crash that happened and there were some crashes but anyway the point of all that is to say we took it very seriously, um, the people had trusted us with you know their support.  
	Anyway now we are still in like a week into this, right, eventually within a few days our technical guys, the economist and nature pick up on this, and then one morning there's 30 grand in the Paypal account, what!  How much?  So, yeah, we hadn't really thought about it seriously.  At the point we were like, this isn't just a couple thousand dollars sitting here and there for a bug bounty this is serious, we created a non-profit corporation and one of the ideas was that it doesn't make sense to spend a significant chunk of money on taxes because it's not what people intended the funds to go for it.  We have a mission statement, a charter and by-laws and so forth, and all the standard things that you have, that's on the website.  Nice logo.  And one of the most important things we did was Matt and I said, you know, we have some interest and expertise in area X but Y,Z you know, A, B, C, D, E, F, we were out of our element. 
Ah, yes.  Okay.
(Laughter)
(Applause)
	>> Yeah, yeah, yeah.  Yeah.
	>> Very good with it.  
(Laughter)
	>> I guess we have a pause.  All right.  
	>> First-time speaker at DEF CON!  
	>> Woo-hoo!
	(Laughter)
(Applause)
	>> Oh yeah!  I like it.  
	>> Congratulations.Good Luck at getting back to your talk.  
	>> Where were we?  Dammit.  Thank you, sir.  I don't know what give short shrift to the incredible people that stepped up and offered to help in ways small and large.  I really want people to go to the Open Crypto Audit website the people page.  This is an extraordinary group of people who stepped up and offered to be technical advisors and part of what that means is we knew we'd be compiling a report.  We knew we'd be dealing with a lot of different material but we also knew that would span basic cryptography, security engineering, protocol engineering, legal reviews.  So I don't know if people know but, I mean, Jim Denaro is one of the top sort of IP and sort of security-oriented attorneys and he's also a CISSP and a practitioner which is really amazing.  Thomas Ptacek, if you don't know them, definitely check that out.  It just came out a couple days ago at BlackHat they are going through level eight. 
It's an incredible program. Runa Sandvik, formerly with Tor, Nate, I mean, it's just an incredible line up.  Moxie even said, you know, I'll take a look and see what you got and, you know, give you some thoughts and ideas as we get further along.  Jo is here from his group.  
	It's an extraordinary group and if you don't know Trevor Perrin, he's one of the brightest in the world in cryptographic engineering and crypto protocols.  So anyway, just incredibly honored to be working with people of this caliber.  
	Because we have a non-profit so we need officers and Matt and I stepped up and then we put a call to Marcia Hoffman, she is still special counsel for EFF and she's been extraordinary.  
	Then we had our first board meeting.  This is how we roll.  Take that, Man!  So, you know, lots of connections.  There is sort of the problem with the history of the project and trying to make sense of things and obviously lots and lots of questions began.  
	So we did as I said we reached out to a small number in the organization that can do this.  We had tremendous response.  If you don't know iSech, find the… you know, find someone who you really, really, really respect in the crypto world, and ask them about iSech. They've been extraordinary partners.  
	Technology fun stuff happens, we have an arrangement with iSech, we've done some, I mean we have existing contract mechanism, we're actually willing to kick in a match grant so you can double the amount of engineering hours on this and that will be our contribution.  I don't know if I actually put this in but we also did -- fairly in the audit we got a $10,000 donation from Austria.  And some other large donations, I'll get to that in a second.  
	So fast-forward, I don't know if people have seen this but it's online.  We spent many, many weeks working with iSech on standups, going through each of the different sort of pieces of the, their part of the project.  It's important to say, too, what we talked with them about and there was the group and there was pretty broad consensus with the group, there's a long history of you know security failures that have nothing to do with crypto but they are in crypto products.  So, you know, this was all sort of post Heartbleed and around this time, too, preHeartBleed but that tradition is very well known and so while crypto is hard, all the other things on sort of the foundation of this stack are also very hard and also easy to get wrong.  
	So this is what we wanted them to focus on.  
	So, you know, memory management, construction of the  different classes, data structures, these are not random number generators, not you know multiple cypher block chains and that different suite.  But more sort of the foundation elements.  
	So couple of their engineers were augmented by peer-reviewed by senior principles and looked at this and they basically came up with, well the way the volume header is derived is maybe not so strong.  But again, remember the issue a lot earlier with the number of hashes per second.  These things are all relative, right?  
	So there was some discussion about information being paged out.  So the idea is that let's say you have true crypto and you have a volume open and keys in memory or passwords in memory.  Say you hit a malicious website that is doing XML entity, explosion like a -- quadratic header where you declare a single entity in XML block and then the way that is invoked, you can basically exhaust memory, you can max out CPU.  The idea was that particularly on Windows systems it's easy to contrive very simple proofs of concept where memory is just getting paged like mad.  
	So when that is happening you could get raw key material being saved to disc, to non-encrypted disc.  The decompression system not unlike libLZ, which is very prominent int he Linux world and, I mean, we have systems all over that depend on that.  
	The particular one you used was sort of deprecated but again this has been around for ten years.  Some of the core elements were built six or seven years ago.  There were several places where they are using mem set to clear data.  So the idea is I have a password or key, do manipulation on it and say mem set and wipe that out.  I'll show you in a minute why that's a problem.  
	But the thing is none of those actually address the true crypt security model.  Because one of the reasons I spent time on the fundamentals earlier is let's say that scenario plays out and you have true crypto and the volume up and you have true material in memory.  You hit a malicious website.  Well, the fact the vector is very specialized in paging memory to somehow get those keys to disc which can be five or eight hours later, take a step back.  You have hit a malicious site that is, you know, running an exploit on your machine.  It is just not a TrueCrypt model.  It's not part of the security guarantee or the security promise.  
	What does that mean?  In aggregate, none of those vulnerabilities discovered are, I was gonna say part of the security model.  If you had physical access to the machine, to the amounted volume, like the machine was in sleep mode, they already contain exploits for that, like $200 but that's not just TrueCrypt.  That's bit locker, that's others.  Deluxe family of Linux.  All major disc and all this encryption implementations have similar attack vectors.  
	So if we're not really weak in the things that are actually you know in the security promises, again, the key derivation, we could probably on the password drive key you know increase the iteration count, you already saw in the group force, it's going to be decent.  It can be improved but there's no show-stoppers so this is around the time, you know we have sent them manuscript for view, team looked at it, I signed off for the peer review and then the great line by John Lennon, life is what happens when you're busy making other plans. TrueCrypt.org gets dark, my Twitter starts lighting up, what's wrong with the website.  Did you guys -- did you do this?  Are you kidding me?  You don't know how much pain we went through because you never know what, what people are gonna have us spend time on.  We went through a fair amount of pain to harden our site. 
It's basic like HTML and like not some ancient WordPress or something.  
	I'm looking like what are you talking about?  Our site is fine.  Not your site.  True crypt organization.  What?  Yeah, at first it didn't even show this.  The morning or afternoon this happened, I don't have a screenshot of it basically we had an error message source code saying this may be a malicious Web page and I guess they have like automatic intrusion software that says, if your volume is more than like this, and then one day it shoots towards magnitude and it's a header redirect, looks like a crawl, pop-up messages, this was the message you see on source.  
	You go to true crypt.org and get redirected to sourceforge where the project was originally launched but years ago, it sort of was not abandoned but sort of there was a -- there were basically directions, you know, the official website truecrypt.org.  But look at the language here.  Not secure, may contain unfit security issues. What? What are you talking about!  So the I got my phone and Matt's and, you know, got the whole board was just everybody is getting pinged on this, like what's the deal?  
	The short answer is we don't know.  We had an official line of communication, official e-mail contact but the mail exchange went offline, too.  Well, if your official line of communication has gone silent and you have some people's personal e-mail addresses, okay, we can try that, too.  Some other people were involved that just when the story broke we're curious too, they started contacting people and there were some other lines of communication but it wasn't really clear because some of the conversations we've had with people who are listed in the public record were like, well, the key to, well our developer keys have been destroyed.  We're not even sure if we have access to the site anymore, meaning TrueCryptOrg, how do you not know if you have access to the site?  What does that even mean?  
	And so there is all this speculation.  So we sort of took a step back and we said, okay, whatever this sort of sociological equivalent is, maybe it's middle-aged guys that just said, meh, I don't need the headache, I don't need the, you know, one of the messages was XP has been stopped; and that's sort of what it has been built around.  But there's so much, so many strange inconsistencies on here, there were just so many question marks.  But you know, the thing is we have a project.  We have, you know, we have stated goals.  We have taken support from people around the world.  So we talked about it for a while and we said what, you know, doesn't matter.  I don't care if someone flipped out or someone rage quit or an agency got to somebody, you know, I don't -- if it's an internal battle.  All these things we are sort of speculating and many of which could have been possilities, we're in the middle of an audit so we'll continue on. 
We have the findings from the iSech team, we'll start phase 2, the formal crypt analysis so one of the other things that is a real drag about this is, you know, there were lots and lots of documentation and examples and sources and materials on the website.  All that got wiped.  I mean, there are mirrors that people have set up.  One of the first questions was, well, what the hell code do you use?  You can't use the stuff on sourceforge because it was labeled 7.2.  That's read-only.  It only has popups.  If you install 7.2 TrueCrypt it has popups and you basically don't use this, it may have issues.  
	Then literally they ripped out the, you know, the right in the update code.  No!  You know, whatever issues existed the day before, and whatever security guarantees could be made the day before in the announcement.  They are still there from a pure software perspective.  So we had a couple delays and I was hoping to give more update on phase 2.  But basically Thomas and Nate are kind of mentoring or organizing phase 2 but that will be full crypt analysis that we'll be looking at the cyberchains, excuse me, at the random number generator instructions and the protocol instructions and so forth.  
	We have talked about several post-audit scenarios but what we can't do is we can't support the development of a project now that's a moving target.  The people who contributed to the project said, we want a verdict on true crypt 7.1, this is used by several million people. So we're going to give you a verdict on true crypt 7.1.  We can't tell you it's absolutely secure, rock solid or not.  But what we can say is we looked at these 10 things and didn't find any issues.  And as much as that is possible, we'll do, we can do.  
	Obviously lots of questions.  But we have had some conversations with the I think originally they were calling themselves the next crypt organization in Europe but they have sort of announced they are holding off until we finish.  You can't audit an ongoing live project which is just, you know, ripping out core classes and things.  
	So I want to spend just a minute or two on some secure coding pieces.  And I want to explain a little bit more on the Isech findings and then we can open it up for questions.  
	This is a great quote!  I think it's Maciej, founder of Pinboard, I'll let you read it but "There's no difference from the attacker's point of view, between gross and tidy errors. Both of them are equally exploitable… This lesson is very hard to internalize. In the real world, if you build a bookshelf and forget to tighten one of the screws all the way it does not burn down your house." I love that.  So one of the things people thought about is there are all kinds of tools, tons of tools.  Static analyzers, dynamic analyzers, intelligent static analyzers , and yes, some of those help.  Some are crucially needed.  The world is not for lack of auditing tools.  
	I highly recommend people take a look at the examples.  But prepare yourself before you do.  I will also go through this quickly.  This is a quote by the former director insurance director.  Basically saying "Source code is interesting and sometimes helpful but computers don't work on source code, they won't.
(Laughter)
 They work on machine code.  So real quick, consider hypothetical.  Does that work or not?  Well, depends.  Depends on the compiler, depends on the flags, your primer, tour optimizations.  I'm going to show you several real quick from the 64 group.  Visual studio, I know, they have to deal with Windows, here are Windows, example one, example 2 using that security secure set.  Mim set just gets optimized out so this is a simple example pop-up of a dialogue box but what if it are the earlier example, password, key, mem sech clear?  Meh, this is where the function would have started.  This is where the code returns.  You didn't do anything with the memory you set so therefore we won't bother to do the inline compile and actually do anything for you.  
	Secure zero memory does.  This is not a slam on GCC, it's not a slam on the studio or any particular one but these are complex tools and have lots of non obvious side effects and even if they have been around for, say, four or five years, your OS may not have anything posted, a current version.  And learning from other people.  From the Tor project, I see six.    Five, yeah, I see five key files.  That's a problem.  Yeah.  NSS, that's a problem.  Then you get to stuff like this.  Basically the issue there?  
(Pause)
All right.  So just a couple thoughts on trust.  This is true.  It's really depressing.  This is a -- it's not Unicode, I forget but there's a coding standard.  This is accepted as legit from every major mobile system, every major desktop browser.  Do you have any idea who that is?  I don't.  More on trust.  Linksys, net gear home router, see that? Pcap utility built in. Thats diagnostics. It randomly sends other frames back to the mother ship just for diagnostics.  What?  As recently as February, if you were running Red Hat, sent OS, Amazon Linux, Oracle Linux, you were actually running Hubble-open ssl because there are upstream requirements with the shift of actual controls.  
	Anyway, HeartBleed, so this is a board, cipher, which was happy to give out the goods.  So takeaways.  Again, most of the catastrophic failures, were secure coding errors, they were not crypto errors, we have a long tradition of that.  Tools report but they are not enough.  And subject matter expertise is crucial.  Also in terms of unpaid volunteers, hey, volunteers are also crucial.  They are not enough.  What we really need and we have been working on, we're in talks with the Linux Foundation, core infrastructure initiative as well as some other organizations. I'm really developing a working model for public code review.Because this is important stuff and particularly in the crypto world but security engineering broadly.  
	Because even with a great chain, um...
(Pause)
Right.  Oh, just gonna tell you, we have a few extras.  Probably not your threat model.  We can talk about that if you want.  
(Pause)
Then we can add in, so reports released, we are working on the reports.  In phase 2, so we can open it up.  Five minutes?  Five minutes for questions.  I have for a really good question, I have original vintage movie posters of war games, I have sneakers DVDs and T-shirts.  So questions, please.  
	>> So you mentioned that there is all these static analyses and demand analysis tools and they're great but they're not enough and you also mentioned that a lot of the errors are not crypto errors, they are coding errors, ignoring crypto errors for the moment, what do we need to bridge the gap between what tools can find now and just general coding secure coding errors to the point where your average coder, because there's a lot of people in the world who don't spend time learning how to code securely, so what tools are we missing to make your average coder able to run a checker and have some reasonable assurance they haven't made stupid errors?  
	>> I think one of those crucial things are project initiatives like Tessa challenge or crypto palace.  I can't say enough about the service to the community that Thomas and the group is doing.  We could talk more about it afterwards but I think is really, it's a training and visibility thing and awareness and a lot of it is things that are not obvious.  But I think there are some tools, I think awareness and training are crucial but we also need to work on a better model in general because I think the model is broken.  
	When Cisco takes a week to literally just count they had 60 plus devices post HeartBleed just to enumerate how many other systems were affected -- (Inaudible) -- AG, then the -- I mean, yeah, we have got to be rethinking the whole way we approach systems.  
	Come on up.
	>> What sort of steps have been taken to address either incoming legal or governmental requests that information not be disseminated toward or to the public or what are your plans?  Is there a fallback plan for when these challenges do occur?  
	>> You mean on the project itself?  
	>> Right.
	>> Obligations or security letters?  
	>> Uh-huh.
	>> Seen the national security letter which was already online.  We have Signed statements from the Isech team and other people on this project so we have not been served or noticed by any government agency.  I suppose if we were, we have some of the best constitutional attorneys in the world working with us.  I don't know if that's a great plan but it's our plan.  
	Come on up.
	>> Do you guys have any plans to after this is all done continue on with other open source projects and audit them, because I think that's a problem.  There are so many open source projects and if we all just go on faith that they're good --
	>> No, no, I appreciate that.  Thanks.  I didn't even mention I'm sorry we're auditing open SLL, the Open Crypto Audit Project is doing it. It's one of the most ambitious security projects in history. We're doing it.  That's great.
(Applause)
And I think by the way there are probably 12 other like runners-up that are really involved in that position, too.  I think we'll do a couple more and I'll be in the chill cafe and we can talk as long as you guys want.
	>> Real quick.  How do you get involved with this open crypto audit project?  
	>> For phase 2 it will be a mix of sort of guided sort of exercises but it's going to have much more of the community kind of full participation.  We needed the sort of phase 1 to go and it did.  The second one will not be near as much professional services although it will have some of the top cryptographers in the world working and coordinating but basically reaching out and contacting.  Matt and I are online in Twitter and crypto audit.org.  We are talking to people.  Be patient, we get flooded sometimes with requests so you know you won't hurt our feelings if you ping us again in a week or two and say what can I do to help?  This is by no means Matt and I.  There are many many people that made it possible so far.  Last one?  
	>> First of all, just want to say thanks to you guys for doing this.  Two quick questions.  One how can the community support you?  Two, can you offer any comment on the possibility of TrueCrypt support for GPT and UEFI?  
	>> You know, I think that's maybe a Matt question on the last one.  
	The first one again, sorry.
	>> How can the community support you guys while you work on it?  
	>> Get involved and stay aware of things like the core infrastructure initiative.  This is looking at every core element of the stack.  From you know the most basic, all the way up to staff.  There will be big announcements coming.  We're getting approached by sizable organizations and how they can help.  For phase 2 it will be a global you know, bug hunt, by the way we still have like $30,000 so there will be way to pay outs for the next couple months.  Ideally we get more but that's where we are so far. I think we'll stop there.  I'll take questions out and then so all that swag.  Thank you.
(Applause)