[00:01.860 --> 00:03.400]  Please remain calm.
[00:07.840 --> 00:10.740]  Our host tonight is Bob Stratton.
[00:14.480 --> 00:16.560]  Hi everybody, I'm Stratton.
[00:16.880 --> 00:24.680]  I have to make the obligatory blurb that I work for a big internet provider and what I say here is me and not them.
[00:26.180 --> 00:28.760]  Just because they make me say that whenever I talk at these things.
[00:28.760 --> 00:42.020]  That having been said, a lot of the things I'll say I think are shared by a significant portion of the networking community and people that do run big networks and things like that and people that have common sense.
[00:45.100 --> 00:46.320]  Let me get my brain here.
[00:48.640 --> 00:52.340]  Basically, it's kind of a tough call figuring out just what ground to cover.
[00:52.340 --> 01:01.760]  So, I may ask you guys a couple of things because I'm trying to get a feel for what the knowledge base is and what people need to know about and what people want to know about.
[01:05.240 --> 01:06.180]  Let's see.
[01:08.260 --> 01:09.700]  First off...
[01:09.700 --> 01:10.500]  Introduction.
[01:10.900 --> 01:11.580]  Yes.
[01:12.160 --> 01:19.580]  To my right is Mr. Eric Hughes, who is a cypherpunk and a mathematician and what else are you?
[01:19.580 --> 01:20.220]  A lot.
[01:25.100 --> 01:26.380]  An anarchist.
[01:27.020 --> 01:27.360]  That's right.
[01:28.500 --> 01:33.440]  And to his right is Mr. Matt Blaze, renowned cryptographer.
[01:38.780 --> 01:42.180]  Publishing papers, I think, qualifies you as being a renowned something or other.
[01:44.780 --> 01:51.360]  I work in the network operations center of UUNet Technologies or Alternet, whatever our name is this week.
[01:52.680 --> 02:04.980]  But when I'm not doing that, I speak about privacy and security issues to people like state general assemblies and city councils and things like that.
[02:05.820 --> 02:10.080]  Frequently with Dave Banasar, who is much better about this policy stuff than I am.
[02:11.560 --> 02:15.000]  And sometimes these guys actually want to know technical things.
[02:15.040 --> 02:20.740]  And it does my heart good every time they do because sometimes they don't realize that they're about to do really evil things.
[02:21.300 --> 02:27.840]  And explaining to them technically why it's bad actually occasionally helps them make different decisions.
[02:30.820 --> 02:57.040]  Cryptography or cryptology, and there's a distinction there, cryptography being the science of obscuring your writing, and cryptology being the study of codes and ciphers, is an amazing branch of mathematics and other technologies because you get quite a few different things depending on what you're looking for.
[02:57.040 --> 03:10.520]  You can go looking for a tool to preserve your privacy and in the process have, as a pleasant side effect, a tool to help you authenticate things and know where things came from.
[03:10.680 --> 03:12.360]  By the same token, the converse is true.
[03:12.580 --> 03:26.840]  You could be looking for a thing that helps you know that the person you're talking to is really who you think they are and as a pleasant side effect discover that in fact you have concealed your traffic from prying eyes, be it voice or data or whatever.
[03:27.760 --> 03:41.760]  As a result, applying the technology can be really interesting and controlling the technology can be even more interesting because you frequently wind up with unintended major wins to a project.
[03:43.300 --> 03:50.940]  So basically, the two major areas where I think people think about using this stuff are privacy and authentication.
[03:51.240 --> 03:56.760]  And they're very different things and very different groups have reasons to use them.
[03:56.900 --> 04:04.240]  For example, there are a lot of people that may be interested in privacy but really, really, really don't want authentication for a particular message.
[04:04.400 --> 04:06.780]  Anonymous remailers are kind of an example of that.
[04:07.320 --> 04:13.900]  You may want to send a thing and you may want to know that somebody can't read it but you may also have a very good reason to conceal where it came from.
[04:16.600 --> 04:34.340]  By the same token, there are times when all you really want to know is who is talking to you or who did a thing or who has a thing and you're really not interested in concealing the contents of or concealing that information and in fact, it may actually work against your purpose.
[04:35.440 --> 04:38.960]  Examples like that are signing an electronic mail message.
[04:39.980 --> 04:50.440]  When the day comes that we're doing business regularly through the internet and buying and selling Zamfir or the pan flute cassettes from Time Life Records or whatever over the internet...
[04:50.440 --> 04:52.120]  Except you'll be able to get it from Finland, too.
[04:53.480 --> 04:55.860]  Well, more like mainland China, but I digress.
[04:56.040 --> 04:56.420]  Exactly.
[04:56.800 --> 05:01.900]  You may want to sign a thing, you may want authentication, but you may not want privacy.
[05:02.460 --> 05:06.820]  That may be a bad example, but you may not want to conceal the information in there.
[05:08.940 --> 05:22.240]  Another concept that you'll occasionally hear bandied about is non-repudiation, which is kind of a silly-sounding phrase that means I'm using a technology that makes it impossible for me to say I didn't say that.
[05:23.760 --> 05:33.220]  And there are various tools or applications where you need that, especially, for example, if you're ever going to be doing business because you sign a contract.
[05:33.220 --> 05:39.400]  It would be really nice to know that the person who signed that contract can't just automatically hit a key and undo that.
[05:40.380 --> 05:46.900]  So you need a way to say, yes, in fact, I did this thing, and anyone who ever checks in the future can discover that I did this thing.
[05:48.740 --> 05:55.520]  Other pleasant side effects are things like, is it the New York Times that the timestamping service is published in?
[05:55.520 --> 05:55.760]  Yeah.
[05:55.880 --> 06:01.640]  In the Sunday New York Times and the classifieds, there are some really weird-looking little messages.
[06:02.520 --> 06:16.500]  And those are actually from a service that allows you to publish a statement mathematically that says, I wrote this document at a particular date and time.
[06:16.740 --> 06:17.620]  On or before.
[06:17.780 --> 06:18.580]  On or before.
[06:18.760 --> 06:19.520]  A particular date and time.
[06:19.800 --> 06:23.420]  And in fact, this information is registered such that you can verify this.
[06:23.740 --> 06:37.580]  Say you have just written the killer application, and you really, really, really want to be sure that somebody can't swipe it from you later on, so you want to basically copyright it or patent it or do any of those things to protect your intellectual property rights.
[06:37.580 --> 06:55.120]  You can actually send information to a service which will hash it down into a small set of numerical information and apply a function, a cryptographic function, that can always be backed out and verified against a copy of that information later on.
[06:55.880 --> 07:02.960]  And people have a record that you published this at this particular time, and it's very difficult for someone to say otherwise.
[07:03.960 --> 07:07.960]  So that's actually a regular business that is being run.
[07:08.320 --> 07:12.560]  And in fact, in any Sunday New York Times in the classified section, you can find that stuff.
[07:13.320 --> 07:28.640]  So these are all little side effects out of cryptology that convince us as we look at this that we will always keep finding neat new tools.
[07:29.660 --> 07:39.760]  And the more we keep looking at this and the more the general public starts using this, which is my personal gung-ho motive, that we'll find even more neat things.
[07:40.720 --> 07:43.820]  A couple of other places where you'll see it.
[07:44.280 --> 08:01.260]  The general magic, Telescript, well I guess it's both a product and a service, that will have these agents running around for you doing things like checking your stock quotes and dropping you mail when your favorite stock goes above a certain price.
[08:01.260 --> 08:09.200]  Or that will roam out looking for information and finding it for you and returning it and sending it to your pocket widget thing automatically.
[08:09.740 --> 08:15.980]  Those are basically enabled with some authentication technology so that you can figure out who that belongs to.
[08:16.160 --> 08:21.940]  So that they can say, I am Joe's agent looking for this information.
[08:21.940 --> 08:28.640]  So that's another application where you'll probably see authentication technology happening an awful lot.
[08:30.780 --> 08:36.020]  How many people here understand the distinction between public and private keys?
[08:38.020 --> 08:38.740]  Good.
[08:38.940 --> 08:39.160]  Time?
[08:39.680 --> 08:39.960]  Hmm?
[08:40.840 --> 08:41.240]  Say...
[08:41.940 --> 08:43.140]  I'm just...
[08:44.820 --> 08:49.120]  The general conceptual, I'm just trying to get a feel for what we know about here.
[08:50.160 --> 08:51.760]  Which means, let's not review it.
[08:52.260 --> 08:52.420]  Right.
[08:53.260 --> 08:55.460]  I won't go over it if people don't want me to.
[08:55.660 --> 08:55.940]  That's why.
[08:57.120 --> 08:57.680]  Please do.
[08:58.020 --> 08:59.240]  Do you want to give a tutorial on this?
[08:59.600 --> 09:00.600]  Do you want to write now?
[09:01.240 --> 09:01.900]  No, no, no, no.
[09:02.420 --> 09:03.640]  Do you want to do a separate tutorial?
[09:05.120 --> 09:05.720]  Yeah, sure.
[09:05.900 --> 09:06.100]  All right.
[09:06.300 --> 09:06.940]  Go for it.
[09:07.200 --> 09:07.760]  This was easy.
[09:07.920 --> 09:12.800]  Matt has a half hour presentation called Everything You Need to Know About Crypto in 30 Minutes or Less.
[09:12.980 --> 09:16.580]  And we're going to do it not here, but in the next room, kind of after the programs are over.
[09:16.580 --> 09:24.560]  Okay, so if you want a real crypto tutorial, which is more technical than we're going to get into now, find Matt when he's over at the next room holding forth.
[09:25.080 --> 09:25.160]  Okay.
[09:25.920 --> 09:27.300]  That was easy now, wasn't it?
[09:29.260 --> 09:43.480]  So, a couple of things that I'm going to talk about are what's out there right now, what you need to worry about, what the tools are to help you protect yourselves, and who the people are that have an influence over your ability to use those tools.
[09:44.920 --> 09:54.260]  With respect to voice information, a lot of us have used cordless phones for a long time and know that anybody can listen to those because they're on the radio.
[09:54.260 --> 10:00.800]  A lot of people use cellular phones and know that an old black and white television set on Channel 77 will sometimes get you that stuff.
[10:01.720 --> 10:09.200]  And that there are an awful lot of people who have decided that, in fact, legislative measures are all that's necessary to protect your privacy.
[10:11.940 --> 10:13.680]  That's some of the reaction I was hoping for.
[10:15.500 --> 10:30.760]  Those people are acting in the influence of some very large interests, including some people who understand that to include parts in every single thing that they shipped at the time would maybe cost 50 cents a unit or something.
[10:30.920 --> 10:31.380]  Oh, my God.
[10:32.960 --> 10:43.280]  And that there is an administrative overhead and that they didn't want to provide those services and that they, in fact, in some cases claimed quite blatantly that these services were private.
[10:43.280 --> 10:51.540]  But I actually have some old cellular telephony advertising where the advertisers said, this is private.
[10:52.420 --> 10:52.880]  So...
[10:55.020 --> 10:55.720]  It's the White House.
[10:58.600 --> 11:09.620]  Now, a lot of people here, I think, understand that legislative protections will probably stop some casual people and maybe even a 13-year-old or two who are kind of scared from listening to your stuff.
[11:09.620 --> 11:14.740]  But those of us who actually want to take some responsibility for ourselves understand that that's not appropriate.
[11:16.360 --> 11:27.520]  For a long time, the most readily available tools for obscuring voice information was a thing called the frequency inversion scrambler.
[11:28.080 --> 11:36.000]  And the radio heads will smile and nod because this thing basically took your highs and made them low and your lows and made them high in your speech.
[11:36.300 --> 11:39.360]  And anybody who actually practiced for a while could listen to this stuff.
[11:39.560 --> 11:41.160]  It sounded kind of like Donald Duck.
[11:41.740 --> 11:50.200]  And if you trained yourself, there are people I know who, with a little practice, could actually decipher what was being said without even filtering it or doing anything like that.
[11:50.200 --> 11:51.560]  There is a full break anyway.
[11:51.720 --> 11:51.960]  Yeah.
[11:52.400 --> 11:55.220]  Oh, it's real simple even to build the circuit to do that.
[11:55.320 --> 12:02.280]  But for a really long time, and I might even argue to this day, that was the most readily available stuff for people.
[12:04.620 --> 12:06.660]  And there are a couple of reasons for that.
[12:07.680 --> 12:24.380]  Part of that is that the American government, for many reasons, has decided that cryptology and the way the national security directors are written, they actually cover cryptology, which means the entire science and study of this stuff.
[12:24.660 --> 12:28.220]  Not just cryptography, which is actually using it to write things.
[12:28.920 --> 12:45.600]  Cryptology was actually such a sensitive technology that, and this is again my estimation, that the national security interest outweighed any other personal privacy interest in terms of spreading the technology.
[12:45.600 --> 12:52.100]  And it took the banks screaming bloody murder, actually, to get a national standard.
[12:53.140 --> 13:04.820]  The financial institution has sort of been the leader in this, because it didn't take a whole hell of a lot of common sense to say, wow, you know, maybe I don't really want my wire transfers going around in the clear where some guy with a butt set and a modem can,
[13:05.000 --> 13:07.440]  or a data scope can watch this stuff.
[13:08.420 --> 13:12.820]  But you need to think about the magnitude of that, because these are the banks.
[13:13.700 --> 13:19.140]  Now, Joe Blow, with very good business or personal reasons, was kind of out of luck.
[13:20.520 --> 13:24.480]  Since then, the genies come out of the bottle, and a lot of things have happened.
[13:24.980 --> 13:43.100]  For example, I discovered that Bell Atlantic had an encrypted cellular service that they would sell to you, because before the Secure Telephone Unit 3 came out, which was a standard for telephones for the government, basically, that allowed you, with a key that you plugged into the side,
[13:43.360 --> 13:48.220]  to encrypt your phone calls at various levels of security.
[13:49.340 --> 13:56.460]  Before that was available, there were certain things within the government, but this private sector usually didn't have them.
[13:56.460 --> 13:59.160]  So if the government wanted to talk to the private sector, it was a real pain.
[13:59.940 --> 14:04.580]  And even within the government, there were several different tools and several different ways of using them.
[14:05.760 --> 14:19.100]  One of the things, because this was so difficult, apparently the White House decided that they wanted a cellular system that would allow for protection of confidential information or classified information.
[14:19.740 --> 14:21.380]  And Bell Atlantic built one.
[14:22.160 --> 14:24.660]  And not many people know about it, because they don't really advertise it.
[14:25.300 --> 14:37.840]  It has since been supplanted by the Secure Telephone Unit 3, and it turns out that they will, they claim, some people claim, they'll sell you service on this as long as you buy the phone, because they have all these phones that the government doesn't use anymore,
[14:38.320 --> 14:43.180]  that cost quite a bit to build at one point, but they don't know what to do with them.
[14:44.020 --> 14:46.560]  So, I called them up, trying to buy the service.
[14:47.640 --> 14:51.680]  And at the time, I was a government contractor, although I certainly didn't work for the government.
[14:51.880 --> 14:53.660]  And I got this amazing run around.
[14:53.900 --> 14:56.400]  It was really interesting, because they didn't deny that they had it.
[14:56.460 --> 15:08.040]  Well, the first person denied that they had it, and I just sort of blew them off and went on and started wending my way through the Bell Atlantic Mobile sales force, and customer support, and whoever else I could get to.
[15:09.280 --> 15:11.520]  And I discovered that, in fact, they did have this.
[15:11.640 --> 15:12.540]  They admitted to having this.
[15:12.640 --> 15:17.860]  They even admitted that the service rates were the same as for the regular cellular service, but that you had to buy the phone.
[15:18.000 --> 15:21.840]  And when I got to the part about buying the phone, it kind of just vanished.
[15:22.340 --> 15:33.140]  And it got to the point where I chewed out the general manager of the entire Bell Atlantic Mobile Service Department, or customer service department, and was promised that people would get back.
[15:33.140 --> 15:36.160]  And to this day, I have not been able to get anybody to sell me one of these phones.
[15:36.160 --> 15:42.560]  So this was last summer, and I have periodically pinged them ever since.
[15:42.820 --> 15:53.100]  So there are actually cellular companies who do understand that they have to protect their customers' privacy, but apparently they don't care about us, the public, enough to actually sell it to us.
[15:53.840 --> 15:55.020]  There are exceptions.
[15:56.000 --> 16:03.340]  I would probably say what you were about to say, but say one here has a cellular cell service right now.
[16:03.500 --> 16:08.480]  Now, whether digital necessarily needs it's encrypted, I don't know, but it's still more secure now.
[16:08.620 --> 16:09.660]  Digital does not...
[16:09.660 --> 16:18.360]  We lost the battle for secure digital telephony about two years ago in committee when the NSA put pressure on a bunch of the service providers and equipment manufacturers to not put encryption in.
[16:21.240 --> 16:36.500]  Now, that having been said, GTE, believe it or not, the great telephone experiment in part of California, actually started selling units made by a company called Transcript, which is a very sharp bunch of guys.
[16:36.500 --> 16:40.500]  And you could actually buy their widget, plug it into your phone.
[16:40.580 --> 16:41.920]  It worked with Oki's pretty well.
[16:42.420 --> 16:44.860]  And I know there are a lot of people here with Oki's.
[16:46.720 --> 16:50.980]  And they actually decided that their customers mattered enough to do this.
[16:51.040 --> 16:55.160]  So there is the possibility that if we scream loud enough, we can actually get something.
[16:56.180 --> 16:56.520]  But...
[16:56.520 --> 16:57.420]  Go ahead.
[16:57.540 --> 17:05.700]  It was customer phone to the mobile telephone switching office.
[17:05.700 --> 17:07.320]  So it was not just the cell site.
[17:07.440 --> 17:12.680]  It was actually all the way back to the mid-cell, which is where all the cells for that area plug into.
[17:15.000 --> 17:15.640]  Oh, yeah.
[17:15.880 --> 17:16.140]  Yeah.
[17:16.620 --> 17:19.040]  People with TV sets wouldn't be able to hear your stuff.
[17:19.220 --> 17:21.300]  And this was on a regular analog cellular system.
[17:21.480 --> 17:28.020]  The reason I bring that up is that if we convince people that there is a market force for this stuff, they'll use it.
[17:28.240 --> 17:34.580]  Now, that having been said, the voice stuff is fairly straightforward.
[17:36.440 --> 17:42.000]  But the data stuff is sort of where I think a lot of us are getting concerned and getting more concerned every day.
[17:42.100 --> 17:45.160]  And, of course, I work for an internet provider and I think about this stuff all the time.
[17:46.260 --> 17:55.440]  How many of you send email to people, your loved ones or business associates, that you would rather not have people read?
[17:58.160 --> 18:00.000]  How many of you didn't raise your hands?
[18:01.280 --> 18:01.840]  All right.
[18:01.980 --> 18:03.460]  I was just finding the lazy people.
[18:03.600 --> 18:03.940]  Okay.
[18:05.980 --> 18:14.220]  And here's the... How many of you believe... How many of you actually have confidence that that's reasonably private?
[18:15.740 --> 18:16.180]  Really?
[18:16.440 --> 18:17.360]  I use PGP.
[18:19.240 --> 18:19.920]  That's a start.
[18:20.160 --> 18:22.620]  No, it's really interesting because it... that's fine.
[18:23.180 --> 18:30.980]  The important... one of the important things to think about when you're using this stuff, and Robert Steele touched on it this morning, and it's a really, really important concept.
[18:30.980 --> 18:37.480]  And Winn Schwartau in his book, Information Warfare, which is very good for scaring yourself, talks about this too.
[18:38.600 --> 18:41.860]  You need to think about what your information is worth.
[18:42.260 --> 18:46.960]  Because only if you think about what your information is worth can you decide how to adequately protect it.
[18:47.460 --> 18:53.840]  There are tools that are very, very effective that are a very big pain to use.
[18:54.000 --> 18:56.240]  And in some cases are perfectly justified.
[18:56.240 --> 19:03.360]  And the risk in having your stuff violated is far less than the discomfort that you endure.
[19:03.600 --> 19:04.980]  However, it's a continuum.
[19:05.300 --> 19:13.780]  Some things are really, really easy to use, and some things are not necessarily as robust against the determined attacker.
[19:14.020 --> 19:17.920]  But for casual privacy, might be perfectly reasonable.
[19:18.320 --> 19:27.620]  You know, I'd really like to think that when I send a casual note to the people that I know on the net, then, you know, I don't have to think about somebody reading it.
[19:27.660 --> 19:30.680]  Now, that having been said, there are a couple of things, ways that we can do this.
[19:31.720 --> 19:39.620]  The way that most people probably use now, or think about using now, is PGP, which is a great package.
[19:39.840 --> 19:44.900]  And now, apparently, it's like even easier to get and even easier to give to other people.
[19:44.900 --> 19:52.640]  And has an awful lot of really good people trying to make it into something that everyone can get to easily.
[19:55.020 --> 20:04.140]  My contention is that with respect to the internet, that is just a tiny little piece of the overall equation.
[20:05.860 --> 20:13.560]  If I'm worried about people reading my mail, why shouldn't I be worried about people watching my terminal sessions to my favorite public access UNIX?
[20:13.800 --> 20:17.900]  Why shouldn't I be worried about people watching the stuff that I FTP around?
[20:18.200 --> 20:23.620]  Why shouldn't I be worried about people looking at which dirty GIFs that I'm downloading over Mosaic?
[20:23.960 --> 20:26.200]  I mean, these things are no different.
[20:26.440 --> 20:31.320]  So, if you care enough to protect your electronic mail, you should care enough to protect everything else.
[20:31.320 --> 20:36.880]  Now, for the average person, there's not much they can necessarily do to do that.
[20:37.220 --> 20:47.860]  But the important thing to know is that the new specifications for TCP IP, specifically for IP in particular, mandate the security tools.
[20:48.840 --> 20:54.560]  And as they've currently been spec'd out, and there are some things that were deliberately left...
[20:54.560 --> 20:55.280]  Hanging.
[20:55.480 --> 20:57.080]  Yeah, hanging is a good word.
[20:57.080 --> 20:59.060]  And there are some very good reasons for it.
[20:59.100 --> 21:00.420]  And we can get into that if you want.
[21:01.780 --> 21:03.920]  The encryption is at a packet level.
[21:04.560 --> 21:10.640]  Which means your Telnet sessions, your FTP sessions, your mail, all of that stuff is going over...
[21:10.980 --> 21:12.860]  Well, anything that's TCP based, at least.
[21:13.200 --> 21:17.760]  All of that stuff is going over a pipe that nobody, hopefully, will be able to read.
[21:18.820 --> 21:24.040]  And that is a fundamental, quantifiable difference from the internet that we have enjoyed today.
[21:24.220 --> 21:28.300]  And it also explains why certain people in the government are so scared about some of this technology.
[21:28.300 --> 21:35.260]  Because all of a sudden now, you not only have people who can make, horror of horrors, private phone calls to each other, but you could have people who could actually have private...
[21:36.180 --> 21:36.980]  IRC sessions.
[21:37.460 --> 21:38.520]  If you really wanted to.
[21:40.560 --> 21:46.300]  So, if you... if you run a public access site, you really, really need to think about this stuff.
[21:46.300 --> 21:51.020]  If you use a public access site, you really, really need to talk to your provider about this stuff.
[21:52.320 --> 21:57.560]  Anyone who sends a plain text password across the internet should consider it basically blown.
[21:59.640 --> 22:01.400]  I mean, it's a simple statement.
[22:01.720 --> 22:03.220]  It sounds kind of hysterical.
[22:03.220 --> 22:08.060]  But the fact is, the state of the world right now, if you use a password that's the same...
[22:08.060 --> 22:11.860]  the next time you type it is the first time you type it, you've just lost the battle.
[22:13.260 --> 22:17.300]  So, if you have a place that you use, that uses those, you need to have them fix that.
[22:17.380 --> 22:18.120]  Let's take a poll.
[22:18.440 --> 22:20.400]  How many of you have used /dev/nit?
[22:23.740 --> 22:29.200]  Okay, for those of you who don't know what /dev/nit is, /dev/nit puts the Ethernet adapter of a Sun into promiscuous mode...
[22:29.200 --> 22:31.000]  and turns it into a password sniffer.
[22:32.300 --> 22:34.720]  Yeah, on any Ethernet it's attached to.
[22:34.980 --> 22:37.500]  And there are many of them running.
[22:37.720 --> 22:43.460]  In fact, the Department of Defense has recently released some information indicating that as many...
[22:43.460 --> 22:45.380]  Actually, there have been two releases.
[22:45.660 --> 22:52.840]  The most recent I saw was the Department of Energy estimating that 11,000 passwords got compromised in a couple of months.
[22:54.020 --> 22:55.920]  11,000 passwords.
[22:56.120 --> 23:02.640]  That's 11,000 people that basically had potentially all of their information given to the world.
[23:18.540 --> 23:27.540]  So, I mean, to some degree, you know, I mean, while I fully realized that, you know, that security is totally...
[23:27.540 --> 23:37.000]  I mean, anybody can sniff anything and it's totally vulnerable, you have to realize that the chances of you individually being sniffed are fairly low.
[23:37.000 --> 23:46.500]  When you think about it, if you said there were 20 million, or if you said there were 11,000 passwords, there were 20 million users on the internet...
[23:46.500 --> 23:52.080]  That's a good question, and let us answer that.
[23:52.460 --> 23:53.500]  I have a one-word answer.
[23:53.940 --> 23:54.420]  Filtering.
[23:55.040 --> 23:56.560]  No, here's the question.
[23:56.660 --> 24:03.000]  The question was, there's so much data going over the internet, what are the chances of somebody actually finding my stuff?
[24:03.000 --> 24:07.620]  Well, you know, the neat thing about the internet is, there is one-stop shopping.
[24:08.400 --> 24:10.460]  If I happen to want...
[24:10.460 --> 24:11.980]  Okay, look, think about it this way.
[24:12.400 --> 24:20.320]  How many of you have used a public access service that has dial-ups all over the country?
[24:22.720 --> 24:24.060]  Okay, quite a few of those.
[24:24.680 --> 24:32.580]  Well, I don't have to get on the ANS, the NSF net backbone at T3 speed and look for this stuff.
[24:32.700 --> 24:37.280]  I can go to your friendly neighborhood provider and, oh, it's like scooping eggs out of a nest.
[24:38.080 --> 24:42.120]  Because it's all right there on his local Ethernet, and I get all of his users.
[24:42.420 --> 24:45.900]  And if I'm lucky, I might even get people signing up with their credit card information.
[24:47.080 --> 24:51.760]  So it's not a question of, how likely am I to be hit necessarily?
[24:51.760 --> 24:54.160]  It's a question of, where do I travel?
[24:54.500 --> 24:57.760]  It's a question of, how much do I trust the people I do business with?
[24:58.000 --> 25:03.360]  Yeah, I mean, one of the things you've got to remember is that you can easily cut down the amount of data you're looking at.
[25:03.800 --> 25:06.240]  I mean, it's not a random search process.
[25:06.320 --> 25:08.000]  You can go out and you can filter on stuff.
[25:08.000 --> 25:14.840]  In fact, one of the sophisticated things I've seen is a thing that monitors ongoing TCP connections.
[25:15.160 --> 25:23.380]  And what they do is they merely log off the first kilobyte of information on a TCP connection going both ways, and then they cut it off for later analysis.
[25:23.660 --> 25:27.100]  That'll catch every R login session and every Telnet session in existence.
[25:27.440 --> 25:35.740]  By the way, there are sniffers running on the internet right now that look for mail sessions and suck the mail out and forward it to a third address.
[25:36.080 --> 25:37.160]  Just thought I'd mention that.
[25:37.160 --> 25:44.540]  Yeah, I mean, also, you know, the way you analyze a threat like this is figure out the risks and figure out what you're protecting.
[25:44.960 --> 25:49.940]  You know, it's certainly true that if you send out a password, it's not a guarantee that someone has looked at it.
[25:50.020 --> 25:56.600]  But on the same token, saying, well, I saw somebody smoke a cigarette last week and he doesn't have cancer.
[25:56.780 --> 26:01.300]  So you can't necessarily conclude from that that there is no causal link.
[26:04.800 --> 26:16.460]  As the network gets used for more and more serious things, the number of seriously bad people trying to do seriously bad things with the information is going to go up and up and up.
[26:17.180 --> 26:26.920]  And, you know, the people in this room are in a very good position to make this network safe for real serious use, which it just isn't right now.
[26:29.440 --> 26:30.340]  Do you want the easel?
[26:36.090 --> 26:36.590]  Are you?
[26:36.890 --> 26:37.610]  Should I cue you?
[26:38.010 --> 26:39.130]  Tell me what you're going to do.
[26:39.610 --> 26:40.390]  I'm going to draw pictures.
[26:40.890 --> 26:41.450]  This is great.
[26:42.530 --> 26:43.150]  Are we on?
[26:43.350 --> 26:43.970]  Wait, it's not on.
[26:44.210 --> 26:45.070]  You were just on.
[26:50.360 --> 26:51.340]  Any other questions?
[26:51.700 --> 26:52.240]  Real quick?
[26:52.240 --> 26:53.020]  Off.
[26:53.920 --> 26:54.740]  Off is on.
[26:54.880 --> 26:55.140]  Off is on.
[26:55.140 --> 26:55.620]  Can you talk more?
[26:55.620 --> 26:56.380]  Oh, that makes sense.
[26:56.580 --> 26:57.100]  Off is on.
[26:57.700 --> 26:58.600]  That is good.
[26:59.740 --> 27:00.860]  Plain text is ciphertext.
[27:03.200 --> 27:04.040]  You ready, Eric?
[27:04.380 --> 27:04.500]  Yeah.
[27:04.660 --> 27:05.000]  Go ahead.
[27:05.280 --> 27:05.560]  Okay.
[27:06.600 --> 27:08.980]  I'm going to talk about something different than what Bob's talking about.
[27:09.740 --> 27:13.580]  There's two basic classes of things you want to keep secret.
[27:13.800 --> 27:16.380]  There's content secrecy and transaction secrecy.
[27:16.820 --> 27:16.900]  Okay.
[27:17.100 --> 27:21.020]  Content secrecy is what everyone understands what regular cryptography to be.
[27:21.020 --> 27:25.180]  That is, what you say doesn't get out to the rest of the world.
[27:25.420 --> 27:29.400]  Now, what does get out to the rest of the world is the fact that I was talking to you.
[27:29.560 --> 27:29.960]  All right?
[27:30.120 --> 27:32.280]  Which is a first-class fact all by itself.
[27:32.560 --> 27:35.220]  Now, this is of serious cause for concern.
[27:35.320 --> 27:36.900]  And I have lots of scare stories.
[27:36.980 --> 27:37.780]  I'll only give you one.
[27:37.960 --> 27:42.460]  The Nazis in France rounded up Jews by looking up at telephone call records.
[27:42.460 --> 27:47.300]  Which is why the European telephone systems use tick billing, which has no state.
[27:47.820 --> 27:48.080]  Okay?
[27:48.080 --> 27:49.420]  They took clerks.
[27:49.540 --> 27:51.000]  They took long logs of records.
[27:51.160 --> 27:52.980]  They took known enemies of the state.
[27:53.200 --> 27:53.960]  Jews in this case.
[27:54.280 --> 27:56.280]  And they found out who they were calling most often.
[27:56.620 --> 27:57.780]  They took those people.
[27:57.920 --> 27:59.160]  They found out who they were calling more.
[27:59.360 --> 28:04.840]  And starting from existing people, they could create a connection graph of enemies of the state that they wish to take down.
[28:06.320 --> 28:08.380]  Transaction data is enormously powerful.
[28:08.380 --> 28:12.380]  One of the interesting things about transaction data is that it comes pre-compressed.
[28:12.860 --> 28:17.240]  You don't have to do voice recognition on transaction data because it's already encoded.
[28:17.440 --> 28:23.960]  If you have a username or a user ID or a machine name or an IP number, this is already done for machine recognition.
[28:23.960 --> 28:27.640]  So, in fact, transactional data is easier to use than content data.
[28:28.760 --> 28:29.160]  Okay?
[28:29.240 --> 28:30.420]  That's extremely significant.
[28:30.860 --> 28:33.960]  Another thing about... there's transactions in terms of communications.
[28:34.220 --> 28:41.280]  There's also transactions in terms of monetary transactions, which is not a big deal on the internet now, but will be.
[28:41.500 --> 28:45.320]  Especially after I describe some of the things we're going to... I want to build in the next few years.
[28:46.120 --> 28:48.080]  For various reasons, they're not working now.
[28:49.420 --> 28:53.140]  I will give you another... one example about financial privacy.
[28:53.520 --> 29:06.980]  Three years ago, the South Korean government ordered Visa, the credit card company, to turn over tapes of every transaction in which one end occurred in South Korea or be forbidden for doing business in that country forever.
[29:07.720 --> 29:10.700]  They turned tapes in to the South Korean government.
[29:11.140 --> 29:14.720]  Now, what have you bought on a credit card that you didn't wish to go?
[29:14.720 --> 29:21.220]  The FBI is known to buy transaction data for people who buy anything from mail-order, grow-light places.
[29:23.100 --> 29:23.700]  Okay?
[29:24.260 --> 29:34.040]  That's... I will... my favorite scare story that I've used, which really gets people's attention, is the handmaiden's tale scenario where the Christian right takes over the government.
[29:34.280 --> 29:38.080]  They subpoena Visa for all records of condom purchase.
[29:41.560 --> 29:45.320]  Those are clearly enemies of the state because they're not breeding enough.
[29:46.160 --> 29:47.440]  We don't want this kind of thing.
[29:47.740 --> 29:58.160]  So I'm going to talk about particular kinds of systems that we are designing and trying to figure out how to implement right and trying to figure out an economic structure in order to implement them.
[29:58.320 --> 30:00.000]  The first one are called remailers.
[30:00.160 --> 30:03.820]  How many of you have used any of the cypherpunk remailers?
[30:04.720 --> 30:05.880]  Okay, quite a few.
[30:06.400 --> 30:09.040]  The cypherpunk... there are two forms of remailers here.
[30:09.260 --> 30:12.320]  There are stateless remailers and remailers with state.
[30:12.660 --> 30:15.400]  The Pennet server in Finland is a remailer with state.
[30:15.580 --> 30:22.620]  It has stored information that contains a mapping between a nominal address and something else.
[30:23.140 --> 30:28.220]  Now, a remailer with state is vulnerable to compromise because you have valuable data.
[30:29.060 --> 30:38.400]  The data is valuable per se because it gives you a mapping between a nominal address, who you know, a pseudonym, and an actual address, who you don't.
[30:38.560 --> 30:44.060]  So there is economic reason for... economic and political reason for these things to be at risk.
[30:44.320 --> 30:50.560]  So one of the things you want to do as a general design principle is to remove state from the systems which provide you communication.
[30:51.320 --> 30:55.680]  Okay, so what the remailer... the original cypherpunks remailer is a stateless remailer.
[30:55.860 --> 31:01.000]  Okay, which means it's not good for replies, except in certain kinds of ways, which we'll get to.
[31:01.300 --> 31:03.100]  The basic principle is the following.
[31:03.280 --> 31:13.720]  We assume a model in which you have machines, which I'll represent with boxes, which have links to each other, and you have little messages flowing across them.
[31:14.020 --> 31:16.400]  Okay, and you can see that the messages are flowing across.
[31:16.400 --> 31:21.740]  Now on the internet, these would be SMTP connections, and you can see the TCP connection startup.
[31:21.940 --> 31:23.260]  I'm going to talk in internet terminology.
[31:23.260 --> 31:27.880]  If you don't follow it, just the geometric picture will give you most of the information.
[31:28.060 --> 31:32.260]  I'm giving you specifics for the people who know, and I'm assuming that that's about half the people in the room.
[31:32.400 --> 31:38.460]  So if you can't follow me, come up and ask me later, take notes, and we'll do questions at the mic.
[31:39.340 --> 31:42.820]  So you have messages flowing along, and you can see the boundaries of the messages.
[31:42.960 --> 31:48.580]  Now that's actually quite important, because if you can't see the boundaries of the message, you may not know how much traffic is going across here.
[31:48.940 --> 31:50.260]  There are devices called link encryptors.
[31:51.000 --> 31:57.800]  And what a link encryptor is, is a device which takes a connection like this, and sends full-time random-looking noise across the channel.
[31:58.140 --> 32:00.460]  You can't determine that there are boundaries in that message.
[32:00.860 --> 32:10.100]  A link encryptor is a much more secure thing, but it will be extremely expensive to build link encryptors on the internet, because we will have per-packet billing on the backbone within five years.
[32:10.320 --> 32:13.400]  Right now you could get away with simply simulating a link encryptor.
[32:13.600 --> 32:14.980]  We may not necessarily agree on that.
[32:14.980 --> 32:17.520]  Well, it is certainly a concern.
[32:18.740 --> 32:19.180]  Now...
[32:19.180 --> 32:20.460]  And we also sell...
[32:20.800 --> 32:23.680]  If you want to buy link encryptors at T1 speed, you can talk to me afterwards.
[32:24.120 --> 32:24.300]  Yes.
[32:25.160 --> 32:25.920]  These are real...
[32:25.920 --> 32:29.760]  He's selling real link encryptors, not the simulated kind that I'm referring to.
[32:29.860 --> 32:34.020]  Anyway, we will assume a model in which the opponent...
[32:34.020 --> 32:41.820]  Now, I have a rant here, because a lot of cryptographers have this professional disease of paranoia, and I think it makes them stupid.
[32:43.740 --> 32:46.320]  Because paranoia makes you think wrong things.
[32:46.500 --> 32:58.220]  If you think the NSA is going to spy on you, and you take actions designed to stop the NSA from spying on you, and in fact the NSA thinks you're a worthless piece of shit, then you've done things which are a waste of your time.
[32:59.080 --> 32:59.800]  Very literally.
[32:59.800 --> 33:06.380]  If you can't make an accurate analysis of who your opponents are, you will not be able to think well about your problem.
[33:07.040 --> 33:08.120]  Is this what this book is about?
[33:08.180 --> 33:08.300]  Yeah.
[33:08.460 --> 33:09.880]  Wynn's book talks about this stuff.
[33:10.080 --> 33:11.580]  Again, this is a great book.
[33:12.840 --> 33:13.280]  Okay.
[33:13.440 --> 33:18.480]  So, it's always important to remember who you're trying to defend against.
[33:18.820 --> 33:19.060]  All right?
[33:19.200 --> 33:20.500]  This is vitally important.
[33:20.660 --> 33:31.160]  We've had the Cypherpunks list for two and a half years now, and only recently are we beginning to get people who, as a matter of course, talk about who the opponent is, and still half the people don't do it.
[33:31.320 --> 33:39.580]  We have one guy from England right now who's talking about this remailer subject, and he can't get straight in his mind why he's doing the things he's doing.
[33:39.680 --> 33:41.780]  He says, well, if he did that, then this would happen.
[33:42.200 --> 33:44.380]  I says, what do you mean this would happen?
[33:44.480 --> 33:46.880]  Who are you talking about that would actually do this?
[33:47.220 --> 33:52.940]  Because what will happen is some things are easier to defend against than others, and some things you probably can't defend against reasonably.
[33:53.520 --> 33:53.640]  All right?
[33:53.880 --> 33:55.460]  So, there are a few classes of opponents.
[33:55.460 --> 33:57.540]  One is the person you're talking to.
[33:58.780 --> 33:59.220]  Okay?
[33:59.700 --> 34:09.040]  Now, this is not a silly thing, because if you simply send someone regular mail from an account which has your real name on it, you have not protected your privacy against the person you've been talking to.
[34:09.220 --> 34:16.800]  So, a simple remailer of any sort is worth 80% of your security, since most of the time you simply want the person that you're talking to not to know who you are.
[34:17.100 --> 34:17.440]  Okay?
[34:17.600 --> 34:19.720]  The second worst people are sysops.
[34:20.400 --> 34:20.840]  Okay?
[34:21.880 --> 34:23.120]  Or people with root.
[34:23.380 --> 34:24.560]  Let's generalize that.
[34:33.880 --> 34:38.300]  And these people have access to the RAM of the processes that are running your machine.
[34:38.420 --> 34:46.560]  And they can grab clear text out of the communication, unless you're running non-existent end-to-end communications across your modem line.
[34:46.560 --> 34:49.760]  The next people you have to worry about are the net providers.
[34:50.720 --> 34:52.560]  Now, Bob is a nice guy.
[34:54.060 --> 34:55.680]  Bob works for a network provider.
[34:57.140 --> 34:59.020]  Bob is not a universal example.
[35:01.400 --> 35:04.000]  We will assume that there are bad people in network providers.
[35:05.060 --> 35:07.600]  The fourth person you have to worry about is the government.
[35:08.240 --> 35:08.620]  Okay?
[35:08.880 --> 35:10.120]  And there are two forms of government.
[35:10.240 --> 35:10.840]  I'm going to distinguish them.
[35:10.980 --> 35:12.400]  We have government which is law enforcement.
[35:12.940 --> 35:13.300]  Okay?
[35:13.480 --> 35:15.420]  And then you have the government which is national security.
[35:17.460 --> 35:17.840]  Okay?
[35:18.020 --> 35:20.580]  And these people really do have different capabilities.
[35:21.160 --> 35:30.260]  There's this wonderful phrase, which if you ever want to really snow someone, if they ask you a question that you don't want to answer, it says, I'm sorry, that would be revealing national technical means.
[35:31.940 --> 35:41.320]  And national technical means are things like spy satellites and listening devices and things which are the mechanisms by which listening takes place.
[35:41.700 --> 35:51.120]  Now, these things are considered so secure that they are not used for law enforcement because to use them for law enforcement would be to give away how the information got there, which you'd have to present in a court of law.
[35:51.360 --> 35:51.720]  Okay?
[35:51.900 --> 35:54.960]  So it really is important to distinguish between the two kinds of government.
[35:55.260 --> 35:57.480]  Government is not a monolithic thing which is your enemy.
[35:58.980 --> 36:00.900]  It's a multitudinous thing which is your enemy.
[36:08.830 --> 36:15.550]  Each of these opponents has different requirements in order to provide secure communication against them.
[36:16.190 --> 36:16.250]  Okay?
[36:17.950 --> 36:22.910]  The systems we're describing here are at least good enough for the first four of these.
[36:23.230 --> 36:23.330]  Okay?
[36:23.630 --> 36:27.230]  People think that they're good enough against the NSA and I'm not yet convinced of that.
[36:27.230 --> 36:30.530]  There's some really subtle mathematical things at work here which I don't want to go into.
[36:30.950 --> 36:32.510]  But let's talk about what remailers do.
[36:32.690 --> 36:34.230]  Now that we've got a threat model.
[36:34.410 --> 36:38.110]  This is the kind of crypto academic usage for this.
[36:38.110 --> 36:39.130]  This is a threat model.
[36:39.370 --> 36:42.170]  So we have a remailer which looks like this.
[36:42.310 --> 36:43.070]  And I'm over here.
[36:43.170 --> 36:46.630]  I'm at Machine A and I want to send a message to this remailer which we'll call X.
[36:46.770 --> 36:48.390]  We'll call this one Y and we'll call this one Z.
[36:51.470 --> 36:53.290]  And I'm sending a message to B.
[36:53.430 --> 36:55.210]  You're also standing in front of the mic.
[36:55.230 --> 36:55.630]  Thank you.
[36:56.030 --> 36:58.190]  Well, I'm also having to hold the mic which is getting...
[36:59.190 --> 36:59.590]  Okay.
[37:00.010 --> 37:04.030]  So suppose you use one remailer and I send it to X and X sends it directly to B.
[37:04.250 --> 37:04.590]  Okay?
[37:04.830 --> 37:07.770]  Here we have a threat model which is we have the correspondent.
[37:07.770 --> 37:18.570]  Well, the correspondent doesn't see who we are because the remailer has stripped off my identifying information and put a substitute which says that the anonymous remailer at X is sending it which contains no useful information.
[37:20.210 --> 37:23.310]  It's not good against the sysop of X.
[37:23.350 --> 37:25.910]  Although it is good against the sysops of A and B.
[37:26.610 --> 37:26.670]  Okay?
[37:27.010 --> 37:27.750]  A and B don't...
[37:27.750 --> 37:30.710]  A sees that this is going out encrypted to be...
[37:30.710 --> 37:32.050]  By the way, I should assume that...
[37:32.050 --> 37:34.350]  I'm assuming in all of this and this is not...
[37:34.350 --> 37:35.710]  I didn't make this obvious which I really should.
[37:35.710 --> 37:40.150]  That content secrecy is taken care of in all of these communications.
[37:40.850 --> 37:41.150]  Okay?
[37:41.410 --> 37:43.390]  That's a necessary prerequisite for all of this.
[37:43.490 --> 37:48.070]  So in other words, this message from A to B is encrypted so that people can't read it in transit.
[37:48.270 --> 37:52.190]  So what we're entirely concerned with is protecting the transactional data involved here.
[37:52.450 --> 37:54.090]  A sees a message go out to X.
[37:54.230 --> 37:58.630]  The message between A and X is encrypted with X's public key so that only X can read it.
[37:58.810 --> 38:02.290]  X opens it up and sees that it says, Please remail this to B.
[38:02.490 --> 38:03.870]  X sends the message to B.
[38:03.870 --> 38:08.550]  Now this is secure against the root operators of both A and B.
[38:08.910 --> 38:09.350]  Okay?
[38:09.570 --> 38:12.290]  Because A sees that it's only going to X and not to B.
[38:12.490 --> 38:15.070]  B sees that it's only coming from X and not from A.
[38:15.270 --> 38:17.890]  So the sysops of A and B have protection against that.
[38:17.970 --> 38:20.070]  What it doesn't give you protection against is X.
[38:20.370 --> 38:31.310]  If X is run by the NSA, which is a perfectly reasonable thing for an intelligence agency to do to gather data, is to operate the service in which they want to get the data for, you don't have protection against X.
[38:31.550 --> 38:31.910]  Okay?
[38:32.310 --> 38:33.250]  Now that's a real threat.
[38:33.370 --> 38:35.030]  This is the threat of node compromise.
[38:35.350 --> 38:37.010]  There are other ways of compromising the things.
[38:37.230 --> 38:38.690]  The NSA doesn't have to run this thing.
[38:38.770 --> 38:44.070]  They merely have to threaten them to turn over logging information of the thing or face results.
[38:46.530 --> 38:50.290]  We don't know what these are, but lots of people are open to pressure.
[38:50.450 --> 38:53.830]  So we should assume that no particular node is secure.
[38:54.010 --> 38:57.930]  And this is a threat model of the sysop root operators of the remailers themselves.
[38:58.710 --> 38:59.070]  Okay?
[38:59.290 --> 39:02.370]  So the solution to this is to use multiple remailers.
[39:02.530 --> 39:09.890]  And in this case, what you're doing is you're asserting you can send it from A to X to Y to Z to B.
[39:10.450 --> 39:10.770]  Okay?
[39:10.770 --> 39:13.310]  Going through three hops, as we call it.
[39:14.150 --> 39:20.190]  And by going through three hops, any of the remailers which provide you security will provide you...
[39:20.190 --> 39:24.350]  Any of these which actually do not log information and have what I call trust in silence.
[39:24.690 --> 39:29.630]  The silence that these people will forget the data and destroy it when it's no longer used internally.
[39:29.890 --> 39:34.610]  If you get trust in silence in any of these three, then you have a secure message.
[39:34.790 --> 39:38.570]  Because the chain is as strong as the strongest link in this case.
[39:38.570 --> 39:43.130]  Because only one of these has to actually work in order for your message to be blinded.
[39:43.490 --> 39:45.150]  This is a really important concept.
[39:45.470 --> 39:46.610]  It's not as weak as the weakest link.
[39:49.250 --> 39:50.610]  It's as strong as the strongest.
[39:51.050 --> 39:56.550]  It's an abnormal kind of situation because you're trying to forget information instead of trying to gain information.
[39:56.550 --> 40:01.550]  And since you're trying to forget information, you have the kind of con versus what you usually have in failure modes.
[40:02.550 --> 40:03.010]  Okay?
[40:04.710 --> 40:08.570]  That same security of multiple hops is also good against law enforcement.
[40:08.970 --> 40:09.430]  Okay?
[40:09.650 --> 40:11.870]  Simply because they don't have access unless they have...
[40:11.870 --> 40:18.790]  One of the things about stateless remailers is if you're served a subpoena, the operator of XA is served a subpoena, there's no data there to subpoena.
[40:19.250 --> 40:19.650]  Okay?
[40:19.830 --> 40:25.250]  You can comply completely with the letter of the subpoena and give them all your records.
[40:25.490 --> 40:29.870]  You can give them a blank sheet of paper and a user's manual for how the software works.
[40:31.230 --> 40:36.450]  And what you have given them is something which complies with what they asked for but isn't useful.
[40:37.970 --> 40:38.370]  Okay?
[40:38.410 --> 40:39.450]  This is a wonderful technique.
[40:39.610 --> 40:41.850]  It'll come up over and over again in some of these other things we'll talk about.
[40:42.870 --> 40:45.370]  This system is probably not good against the NSA.
[40:45.570 --> 40:47.670]  In fact, it's certainly not good for several reasons.
[40:47.870 --> 40:50.550]  One, the message from A to B is a variable length.
[40:50.830 --> 40:52.530]  And so you can correlate on length.
[40:53.350 --> 40:59.390]  The second thing is that if the content doesn't change between A to X, X to B, then you can correlate on content.
[40:59.510 --> 41:07.490]  In other words, if the message is forward without doing an encryption or decryption operation, then you can just see that the messages look the same and you'll assert that they're the same message.
[41:07.770 --> 41:09.610]  The third one is the subtlest of them all.
[41:09.610 --> 41:12.990]  If A comes in... I have to draw some other diagrams now.
[41:13.130 --> 41:23.090]  If you have a message one coming in here and message two coming in here, you have a message one prime coming out and a message two prime coming out.
[41:23.370 --> 41:31.090]  Now, if the machine immediately processes the message and then sends it out immediately, you get time correlation effects.
[41:31.110 --> 41:36.430]  Where you can see that this message came in and then 12 milliseconds later, this message went out.
[41:36.430 --> 41:49.110]  And we can conclude by looking at the characteristic times for messages to pass through the system and things like under load and various factors like that, exactly how long the messages take to pass through and which messages correlate to which other one.
[41:49.310 --> 41:51.810]  So what you want to do is mix.
[41:53.010 --> 41:56.150]  And mixing is... or reorder them.
[41:56.970 --> 42:03.590]  And this is like the thing that on the Cypherpunks mailing list which people just are not quite fully aware of yet.
[42:03.590 --> 42:07.790]  Is that you need to make sure that the messages come in in a different order than they go out.
[42:08.170 --> 42:10.230]  So the easiest way to do that is the following.
[42:10.430 --> 42:12.890]  Suppose we have a reordering factor of two.
[42:13.150 --> 42:16.330]  So that every message that goes out could be one of two other messages.
[42:16.650 --> 42:24.510]  The easiest algorithm for this is you wait until you get two messages and then you pick... you flip a coin and you send out message one if it's heads and message two if it's tails first.
[42:24.810 --> 42:25.990]  And then you send out the other message.
[42:26.410 --> 42:29.510]  Okay, this gives you a privacy diffusion of factor of two.
[42:29.510 --> 42:32.030]  Now, those algorithms can be extended.
[42:32.190 --> 42:33.250]  There's some more sophisticated ones.
[42:33.370 --> 42:40.810]  But it's not... it would not be uncommon in a well-used network to get diffusion factors of a hundred with fairly little latency and enough state.
[42:41.310 --> 42:41.530]  Okay.
[42:42.150 --> 42:47.910]  Now, I want to point out now that I've been talking about email this whole time because email is the easiest way to implement this.
[42:48.950 --> 42:55.110]  You'll now see that all these are boxes and they could just as well do packet routing this way as well as email.
[42:56.030 --> 42:58.210]  Now, I want that to sink in for just a second.
[42:58.370 --> 43:00.590]  Because what you're doing is you're sending datagrams.
[43:00.790 --> 43:05.630]  And a datagram can either be fully formatted as an email message or it can be simply an IP packet.
[43:06.030 --> 43:06.370]  Okay.
[43:06.610 --> 43:09.330]  And so we have what... Matt Blaze uses this wonderful phrase.
[43:09.550 --> 43:12.910]  And I'm going to... we can make what he calls packet laundries.
[43:17.690 --> 43:18.250]  Okay.
[43:18.610 --> 43:24.010]  And a packet laundry is simply something that takes the data in and puts it out with different IP addresses.
[43:24.310 --> 43:28.930]  And then... and keeps some... keeps some state while a connection is up and does the reverse mapping.
[43:29.550 --> 43:31.650]  Now, let's talk about internationalization.
[43:33.170 --> 43:33.730]  Okay.
[43:34.310 --> 43:41.510]  There's a wonderful phrase I got from reading a financial book which is called regulatory arbitrage.
[43:45.510 --> 43:48.850]  Now, regulatory arbitrage is also called forum shopping.
[43:49.230 --> 43:49.570]  Okay.
[43:49.790 --> 43:51.410]  And what this is is the fact...
[43:51.410 --> 43:51.530]  Okay.
[43:51.690 --> 43:51.950]  You...
[43:51.950 --> 43:53.870]  I'll... I'll give you an example that you all know of.
[43:54.510 --> 43:54.910]  Um...
[43:54.910 --> 43:58.670]  Chrysler wants to locate a car manufacturing plant in some state.
[43:58.910 --> 44:01.390]  So they go over to the governor of that state and say, Hey, look.
[44:01.550 --> 44:03.350]  We want to put lots of jobs in your state.
[44:03.510 --> 44:06.250]  But, you know, we're really big and we can go anywhere we want.
[44:06.410 --> 44:09.730]  So we want you to give us lots of money so that we'll give jobs in your state.
[44:09.730 --> 44:11.910]  And the state will go, Well, okay.
[44:12.330 --> 44:12.510]  Maybe.
[44:12.770 --> 44:13.610]  How much money are you giving in?
[44:13.710 --> 44:17.110]  And what the states end up doing is bidding on who gets Chrysler?
[44:17.610 --> 44:18.590]  Who gets the plant?
[44:18.870 --> 44:23.230]  This is... this is forum shopping going from state to state, county to county, city to city.
[44:23.530 --> 44:27.570]  It works just as well in international territories as well as everything else.
[44:27.790 --> 44:35.110]  So one of the things which is going to make actual practical security in the future for privacy is the internationalization of communications.
[44:35.670 --> 44:37.930]  I really... I don't think I can stress that enough.
[44:38.190 --> 44:44.370]  The fact is, as soon as you move outside US jurisdiction, it costs tons more to litigate a case.
[44:44.670 --> 44:48.070]  Now, I have a project that I don't know yet how to pay for.
[44:48.290 --> 44:50.810]  Because this would have to be a serious project.
[44:50.810 --> 44:53.270]  It would need serious funding in order to make it truly secure.
[44:53.410 --> 44:54.750]  But I want to do the following.
[44:55.670 --> 44:56.110]  Okay.
[44:56.930 --> 44:59.110]  This is... this is a bad map of California.
[45:00.550 --> 45:02.130]  And here's where I live, up here.
[45:02.710 --> 45:03.730]  This is San Diego.
[45:04.510 --> 45:07.450]  And right across the border from San Diego is Tijuana.
[45:07.910 --> 45:08.270]  Yay!
[45:11.350 --> 45:15.010]  So, we can... we can make a network topology which looks like following.
[45:23.810 --> 45:24.830]  Do you all understand that?
[45:24.870 --> 45:26.390]  This is the internet at large.
[45:26.610 --> 45:27.350]  This is San Diego.
[45:27.550 --> 45:28.270]  And this is Tijuana.
[45:28.970 --> 45:32.790]  Now, I can make a packet laundry of the Mexican packet laundry, of course.
[45:36.730 --> 45:39.450]  which comes in the internet, comes from the internet here.
[45:39.650 --> 45:41.090]  It gets routed through San Diego.
[45:41.090 --> 45:42.310]  It goes through Tijuana.
[45:42.590 --> 45:45.670]  It gets laundered in Tijuana and goes back out.
[45:47.370 --> 45:51.170]  Now, I can rent, say, machine room space in Tijuana.
[45:51.350 --> 45:53.870]  Or I can rent web space in Tijuana.
[45:54.050 --> 45:56.490]  Or FTP or FSP space in Tijuana.
[45:56.630 --> 45:58.510]  You can rent anything you want in Tijuana.
[46:07.180 --> 46:10.720]  But let me talk about some of the legality here.
[46:11.080 --> 46:18.400]  I had a friend in San Francisco whose BBS was seized in a civil action by Sega.
[46:18.680 --> 46:20.040]  How many of you have heard of this?
[46:20.040 --> 46:27.300]  It did not get a lot of press, unfortunately, because, well, there were some dirty lawyers involved in it.
[46:27.380 --> 46:28.680]  It's really quite a nasty story.
[46:28.940 --> 46:29.560]  But here...
[46:29.560 --> 46:30.560]  By Sega.
[46:30.820 --> 46:32.660]  This was not Steve Jackson.
[46:32.880 --> 46:34.340]  This was not a criminal prosecution.
[46:34.340 --> 46:35.820]  This was a civil case.
[46:36.200 --> 46:40.520]  Sega was asserting copyright infringement and software piracy, basically.
[46:40.520 --> 46:41.740]  And they had the U.S.
[46:41.860 --> 46:43.160]  Marshal seize his BBS.
[46:43.380 --> 46:44.760]  This happened in December of 93.
[46:45.800 --> 46:46.000]  Hmm?
[46:46.700 --> 46:47.100]  Blame.
[46:47.440 --> 46:47.840]  Blame.
[46:49.180 --> 46:51.420]  Well, let me explain exactly what happened.
[46:51.480 --> 46:53.020]  Because I read all the court papers for this.
[46:54.400 --> 46:55.400]  All they had...
[46:55.400 --> 47:02.160]  Basically, it's the standard bad hacker thing that you should all read Bruce Sterling's Hacker Crackdown and see why it's stupid to rat on each other.
[47:02.300 --> 47:03.640]  But some guy ratted on him.
[47:03.760 --> 47:06.020]  And the only thing the guy had was his phone number.
[47:06.020 --> 47:09.680]  So, the informant gave him the phone number.
[47:10.040 --> 47:11.080]  Representatives from Sega.
[47:11.320 --> 47:12.380]  Someone in some law office.
[47:12.680 --> 47:13.760]  Logged onto these boards.
[47:14.360 --> 47:15.120]  Downloaded stuff.
[47:15.600 --> 47:18.080]  And then drew up this huge voluminous allegations of...
[47:18.660 --> 47:21.720]  Which included screen dumps and purchases made from other companies.
[47:21.880 --> 47:23.060]  And assertions of conspiracy.
[47:23.360 --> 47:24.040]  And all this stuff.
[47:24.280 --> 47:26.760]  And they asked the court for, like, a bunch of things.
[47:26.860 --> 47:29.100]  They asked for emergency powers of search and seizure.
[47:29.320 --> 47:32.680]  Which means they didn't have to notify the guy that they were about to steal his computer.
[47:33.900 --> 47:34.320]  Okay?
[47:34.700 --> 47:40.100]  And the argument they used was in analogy with a Gucci counterfeiting case of all things.
[47:40.240 --> 47:47.000]  Where they felt that if they notified the counterfeiter in advance, that they'd simply move the stock out before they had a chance to seize.
[47:47.720 --> 47:52.280]  Probably a reasonable thing to ask for from their point of view.
[47:52.800 --> 47:56.020]  Now, what they did though, is that all they had was a phone number.
[47:56.020 --> 47:59.400]  So the court granted them all this power to do this.
[47:59.500 --> 48:02.080]  And they served a subpoena on Pacific Bell.
[48:04.960 --> 48:09.800]  They served a subpoena on Pacific Bell asking them for the physical termination for this phone number.
[48:11.580 --> 48:12.020]  Alright?
[48:12.200 --> 48:14.400]  Because they didn't have an address to give to the U.S.
[48:14.520 --> 48:16.740]  Marshal Service from where to conduct the seizure.
[48:16.740 --> 48:18.400]  So this is lesson one.
[48:21.920 --> 48:23.060]  Well, no, no.
[48:23.260 --> 48:27.980]  An actual real practical lesson from this is rent the phone at your neighbor's address.
[48:29.480 --> 48:29.920]  Okay?
[48:30.140 --> 48:33.020]  Now this, in fact, is a perfectly good jurisdiction hopping.
[48:33.360 --> 48:33.580]  Alright?
[48:33.740 --> 48:35.440]  Because what you've done is the term...
[48:35.440 --> 48:36.920]  I'm dead serious about this.
[48:37.000 --> 48:40.540]  The terms of the subpoena said, we are allowed to search at this address.
[48:41.100 --> 48:41.540]  Alright?
[48:41.780 --> 48:44.000]  Because this is the address where the phone line terminates.
[48:44.000 --> 48:45.960]  Even if it's next door, the U.S.
[48:46.100 --> 48:50.380]  Marshal, with a writ that says 665, can't go into the number that says 666.
[48:51.320 --> 48:51.680]  Alright?
[48:51.800 --> 48:55.120]  It just doesn't work that way because they are not authorized by law to do so.
[48:55.280 --> 48:57.840]  We still have a reasonably good government about that.
[48:57.960 --> 49:00.960]  And we don't have random midnight knocks and disappearances.
[49:01.160 --> 49:03.000]  So play it up while it lasts.
[49:06.380 --> 49:08.940]  So that's the first lesson is to rent the things in your neighbor's room.
[49:09.020 --> 49:11.020]  Actually, you want to rent on the same side of the street.
[49:11.020 --> 49:15.140]  Because you can run wires on private property all you want as long as you don't cross public lines.
[49:15.600 --> 49:16.020]  So if you...
[49:16.020 --> 49:22.580]  And in fact, if you don't know your neighbor, you really don't have to tell your neighbor if you're in the right space.
[49:22.840 --> 49:25.900]  Because, at least in California, they run these network interface units.
[49:25.900 --> 49:29.740]  These little green boxes with the 530-second center-drilled Allen wrenches you have to use to get in the back of them.
[49:30.540 --> 49:30.940]  Anyway...
[49:30.940 --> 49:32.960]  Watch the movie pump up the volume.
[49:33.060 --> 49:36.600]  There's actually a cute hack along the same lines actually depicted really well.
[49:36.600 --> 49:39.500]  Yeah, you can actually just say that you want it installed...
[49:39.500 --> 49:46.480]  Say you're living at 666 and the neighbor of the beast, 668, is your neighbor.
[49:46.560 --> 49:50.320]  And what you do is you just have a phone installed at 666-668-A.
[49:51.460 --> 49:53.780]  And you say, well, we'll take care of the inside wiring.
[49:53.920 --> 49:55.200]  Just put it on the outside of the building.
[49:55.580 --> 49:55.880]  Okay?
[49:56.100 --> 49:58.360]  Now, if your neighbor is clueless, which, you know...
[49:59.280 --> 50:01.800]  You can simply run wires underground over to your house.
[50:02.040 --> 50:02.160]  All right?
[50:02.320 --> 50:04.460]  Now, this is a perfectly good form of jurisdiction hopping.
[50:05.040 --> 50:07.180]  How many of you want to live next door to Eric?
[50:10.760 --> 50:13.860]  I actually have it even better, Matt, because I live in a duplex house.
[50:14.020 --> 50:16.160]  And I could install a line in my neighbor downstairs.
[50:17.080 --> 50:19.220]  And the phone would terminate in the basement.
[50:21.140 --> 50:22.020]  I digress.
[50:24.220 --> 50:25.680]  I haven't actually done that.
[50:26.320 --> 50:28.480]  But the second thing to realize is that...
[50:29.860 --> 50:33.260]  Assuming that Pac Bell does have accurate information about the mapping...
[50:33.260 --> 50:35.700]  About the mapping between address and phone number.
[50:35.880 --> 50:37.360]  That you want to make sure that that...
[50:39.620 --> 50:41.440]  That's state information which is valuable.
[50:41.920 --> 50:44.220]  So one of the things you want to do is to make that information valueless.
[50:44.340 --> 50:46.380]  Well, the subpoena...
[50:46.380 --> 50:48.140]  One is to use a stateless mechanism.
[50:48.500 --> 50:48.600]  All right?
[50:48.740 --> 50:52.620]  I'm going to talk about a way of doing synchronization for packet remailing.
[50:52.760 --> 50:54.140]  Which gets around some of the state problems.
[50:54.320 --> 50:55.080]  But in just a second.
[50:55.080 --> 50:56.720]  But the other thing to realize is that...
[50:56.720 --> 50:58.720]  If they had to go to a Mexican court...
[50:58.720 --> 51:01.060]  And get a court order...
[51:01.060 --> 51:05.160]  To search and seize a particular location...
[51:05.160 --> 51:05.500]  And...
[51:05.500 --> 51:06.300]  It would be...
[51:06.300 --> 51:07.240]  More difficult.
[51:09.640 --> 51:12.540]  If this were, say, in the Netherlands...
[51:12.540 --> 51:13.900]  It would be even more difficult.
[51:14.240 --> 51:16.800]  And if it were in Singapore or Hong Kong...
[51:16.800 --> 51:17.560]  It would be...
[51:17.560 --> 51:18.580]  Really difficult.
[51:19.960 --> 51:20.880]  As soon as...
[51:20.880 --> 51:21.880]  It would be a Chinese laundry.
[51:22.040 --> 51:22.720]  Yes, it would...
[51:22.720 --> 51:23.960]  It would be a Chinese laundry.
[51:26.340 --> 51:26.780]  I...
[51:27.480 --> 51:30.060]  I swear I have not thought of that before.
[51:32.280 --> 51:34.860]  I must have a constitution against bad puns.
[51:35.740 --> 51:36.180]  Okay.
[51:37.500 --> 51:37.940]  So...
[51:37.940 --> 51:38.740]  This is...
[51:38.740 --> 51:40.700]  This is a very basic form of jurisdiction hopping.
[51:40.920 --> 51:41.440]  If you want...
[51:41.440 --> 51:43.780]  If you have things that you really want to hide...
[51:43.780 --> 51:44.880]  Just use another layer.
[51:46.780 --> 51:47.220]  Because...
[51:47.220 --> 51:47.560]  They can...
[51:47.560 --> 51:49.180]  They can go to the machine in Tijuana...
[51:49.180 --> 51:50.180]  And they'll merely find out...
[51:50.180 --> 51:51.500]  That it's forwarding packets...
[51:51.500 --> 51:52.500]  To Guatemala.
[51:56.040 --> 51:56.620]  Okay.
[51:56.960 --> 51:57.220]  And...
[51:57.220 --> 51:58.600]  And this works really well...
[51:58.600 --> 51:58.780]  In...
[51:58.780 --> 51:59.840]  In places...
[51:59.840 --> 52:00.520]  Like...
[52:00.520 --> 52:01.480]  Central America...
[52:01.480 --> 52:02.340]  North and South America...
[52:02.340 --> 52:04.020]  Which have lots of small countries.
[52:05.960 --> 52:06.540]  Now...
[52:06.540 --> 52:06.800]  I...
[52:06.800 --> 52:07.140]  I...
[52:08.020 --> 52:09.920]  If you spend time in libraries like I do...
[52:09.920 --> 52:10.620]  You'll know that...
[52:10.620 --> 52:13.000]  The white CIA world fact books...
[52:13.000 --> 52:14.920]  Contain loads of really great data...
[52:14.920 --> 52:15.560]  On tele...
[52:16.130 --> 52:18.020]  Telecom connectivity of the third world.
[52:18.800 --> 52:20.280]  They're on CD-ROM now too.
[52:21.740 --> 52:22.320]  It's very...
[52:22.320 --> 52:23.140]  It's very good information.
[52:23.300 --> 52:24.600]  And that's where I found out that...
[52:24.600 --> 52:26.460]  In the Caribbean...
[52:28.860 --> 52:29.260]  Cuba.
[52:29.920 --> 52:31.040]  Well there's Cuba over here.
[52:31.200 --> 52:32.240]  But there's Jamaica...
[52:32.240 --> 52:32.880]  Sitting right about here.
[52:32.980 --> 52:34.280]  And Jamaica has connections...
[52:34.280 --> 52:35.240]  Both to Florida...
[52:35.240 --> 52:36.180]  And to Panama.
[52:37.820 --> 52:38.620]  Underground lines.
[52:38.900 --> 52:39.040]  Okay.
[52:39.300 --> 52:40.440]  Which means that...
[52:40.440 --> 52:41.780]  Jamaica is a really nice place...
[52:41.780 --> 52:42.680]  To put a service like this.
[52:42.800 --> 52:43.200]  Because...
[52:43.200 --> 52:45.180]  If they try to cut the wire at one end...
[52:45.180 --> 52:46.380]  You've got two point failure.
[52:50.120 --> 52:50.600]  That's...
[52:50.600 --> 52:50.940]  That's...
[52:50.940 --> 52:52.140]  That's way cool.
[52:54.640 --> 52:55.120]  Now...
[52:55.120 --> 52:55.840]  As you can see...
[52:55.840 --> 52:57.060]  As soon as you start buying...
[52:57.060 --> 52:58.140]  T1 connectivity...
[52:58.140 --> 52:58.760]  Transatlantic...
[52:58.760 --> 53:00.360]  You need serious money to do this.
[53:00.580 --> 53:02.200]  And so I'm trying to figure out...
[53:02.200 --> 53:03.880]  Where the serious money lies.
[53:05.040 --> 53:05.840]  Well there's...
[53:05.840 --> 53:06.800]  There's grants.
[53:07.920 --> 53:08.720]  There is...
[53:08.720 --> 53:08.920]  Actually...
[53:09.440 --> 53:10.500]  I have an idea.
[53:11.800 --> 53:12.640]  No, no, listen...
[53:12.640 --> 53:13.500]  Wait, wait, wait...
[53:13.500 --> 53:14.080]  Wait, wait, wait...
[53:14.080 --> 53:14.660]  This is...
[53:14.660 --> 53:15.420]  This is a...
[53:15.420 --> 53:15.960]  This is another idea.
[53:16.080 --> 53:16.420]  And it's...
[53:16.420 --> 53:17.220]  It's premature...
[53:17.220 --> 53:18.440]  But I'm going to talk about it anyway.
[53:18.660 --> 53:20.120]  One of the things we also work with is...
[53:20.120 --> 53:21.880]  We've talked about transaction locking...
[53:21.880 --> 53:22.680]  For financial data.
[53:23.780 --> 53:25.240]  And one of the things we do with that...
[53:25.240 --> 53:27.340]  Is try to make what are called...
[53:27.340 --> 53:28.200]  Digital cash protocols.
[53:28.520 --> 53:30.240]  Which are electronic equivalent of...
[53:30.240 --> 53:31.900]  Dollar bill transfer across the network.
[53:31.900 --> 53:34.220]  Where there's no record possible...
[53:34.220 --> 53:35.740]  To be made of who talked to whom.
[53:36.340 --> 53:37.120]  I'm also working...
[53:37.120 --> 53:37.240]  That...
[53:37.240 --> 53:38.600]  That's the money in motion protocol.
[53:38.780 --> 53:39.720]  I've developed a thing called...
[53:39.720 --> 53:40.520]  Encrypted open books.
[53:40.720 --> 53:42.280]  Which allows you to...
[53:42.280 --> 53:43.600]  Operate a bank where the bank doesn't know...
[53:43.600 --> 53:44.500]  What your bank balance is.
[53:44.600 --> 53:45.560]  But they know the sum total...
[53:45.560 --> 53:47.040]  Of all the bank balances that they have.
[53:47.820 --> 53:49.040]  Yes, it is possible.
[53:50.140 --> 53:50.440]  Okay.
[53:50.800 --> 53:51.160]  Yeah.
[53:51.300 --> 53:53.180]  Matt has...
[53:53.180 --> 53:55.380]  About five minutes to do so.
[53:55.600 --> 53:55.860]  Oh.
[53:56.660 --> 53:57.100]  I'm...
[53:57.100 --> 53:58.060]  I don't know how much time we have.
[53:58.480 --> 53:59.060]  I don't either.
[53:59.140 --> 54:00.520]  What's the session after this one?
[54:00.520 --> 54:02.740]  We'll just go until we're told to shut up.
[54:04.220 --> 54:05.380]  And then I'll let Matt talk.
[54:09.180 --> 54:11.340]  So here's a market which uses some of these things.
[54:11.480 --> 54:12.240]  I think I'll...
[54:12.240 --> 54:13.820]  I've talked enough about this kind of stuff.
[54:13.860 --> 54:14.520]  You get the idea.
[54:14.780 --> 54:16.560]  I have lots more stuff about this.
[54:16.880 --> 54:18.840]  Not all of you are going to be completely interested.
[54:18.840 --> 54:21.280]  I can do tutorials later tonight or something.
[54:21.600 --> 54:27.380]  Or you can buy these two books which will tell you an awful lot of what you need to know about any of this stuff.
[54:27.380 --> 54:27.760]  Yeah.
[54:28.040 --> 54:28.440]  And you...
[54:28.440 --> 54:31.000]  This was the definitive text for many, many years.
[54:31.220 --> 54:32.200]  And this is sort of...
[54:32.200 --> 54:32.340]  Yeah.
[54:32.420 --> 54:34.540]  It's David Kahn's The Code Breakers, which is the old book.
[54:34.640 --> 54:41.640]  And Bruce Schneier Applied Cryptography, which will teach you all the basic crypto that you need to actually understand how this stuff works and think about how to design it yourself.
[54:41.640 --> 54:51.180]  Or Dave Banasar is putting together a cryptography and privacy source book, which I believe is for sale as we speak or will be very, very shortly.
[54:51.180 --> 54:53.520]  So you should also buttonhole him and ask him about it.
[54:54.060 --> 54:54.340]  Okay.
[54:54.600 --> 54:56.440]  Last thing and then I'll let Matt talk.
[54:56.540 --> 54:57.700]  I'm sorry to take up Matt's time.
[54:58.600 --> 55:01.220]  So we need serious money to develop a network like this.
[55:01.220 --> 55:09.340]  And it would be nice to have a legal market where there's nothing governmentally bad going on.
[55:10.100 --> 55:13.000]  It would be nice to have a market which required anonymity...
[55:13.000 --> 55:22.480]  It would be nice to have a system that required stateless things because someone is going to sue you asking for your records.
[55:22.480 --> 55:27.200]  Okay, I think I found the perfect market for this, which is Intel microprocessor allocations.
[55:28.300 --> 55:39.560]  Now, those of you who know how Intel works, they sell their microprocessors on allocation, which means that a particular company has...they can buy...they're allowed to buy, you know, 5,000 CPUs per year.
[55:39.740 --> 55:43.000]  If they don't buy all 5,000, they can't buy 5,000 next year.
[55:43.340 --> 55:48.340]  In other words, there's like a one-way ratchet down in your allocations, which only goes up very slowly.
[55:48.340 --> 56:00.400]  And so, there are three companies, AMD, Cyrix, and Nugen, who have Intel-compatible microprocessors who are having a hard time breaking into the market because of these sales practices of Intel.
[56:00.620 --> 56:05.300]  It would be really nice if a company could sell off its excess Intel processors.
[56:05.840 --> 56:08.740]  Now, you can't really...how do you go about doing this?
[56:08.900 --> 56:10.240]  Because Intel doesn't approve of this.
[56:10.320 --> 56:12.900]  Intel will just simply cut your allocation if they find out you're doing this.
[56:13.000 --> 56:15.100]  So you want anonymity, okay?
[56:15.220 --> 56:21.420]  You need to know that you actually have microprocessors to sell, so you need a vetting agent to determine that, so you need pseudonymity.
[56:21.700 --> 56:24.980]  You need to make transactions in the future, okay?
[56:25.120 --> 56:26.400]  So you need anonymous markets.
[56:26.600 --> 56:34.920]  You need to be able to pay for the thing, so you need at least some form of doing clearing in the money, which works with the pseudonymity, okay?
[56:35.520 --> 56:43.160]  How many of you know about this Vietnamese gang about a year ago who made a heist of a truck full of Intel 46 processors?
[56:44.020 --> 56:46.020]  Do you remember how much money was in that truck?
[56:48.400 --> 56:50.100]  Several tens of millions of dollars.
[56:50.980 --> 56:54.740]  And the thing weighed only a few hundred pounds, okay?
[56:55.100 --> 56:57.040]  I mean, there's a lot of money in CPUs.
[56:57.140 --> 57:03.900]  And, in fact, I'm going to investigate when I get home in September the possibility of actually using systems like this.
[57:03.900 --> 57:10.220]  And it will probably require going out of the country in some fashion in order to avoid certain litigious entities.
[57:10.820 --> 57:11.800]  Oh, I feel like Intel.
[57:13.180 --> 57:13.860]  To do this.
[57:14.060 --> 57:15.500]  So this is what I'm thinking about.
[57:15.580 --> 57:16.560]  This is what Cypherpunks is about.
[57:16.680 --> 57:22.920]  If you want to subscribe to our mailing list and talk about this stuff, send a blank email message to majordomo at toad.com.
[57:23.240 --> 57:24.160]  T-O-A-D.
[57:24.520 --> 57:25.420]  Can you write that down?
[57:25.420 --> 57:26.540]  Oh, I can write that down.
[57:37.240 --> 57:39.140]  And just say subscribe Cypherpunks.
[57:39.260 --> 57:42.620]  Or send a blank message and you can use the Majordomo server and ask what lists are there.
[57:42.820 --> 57:45.600]  And there's a bunch of ancillaries, but Cypherpunks is the main one.
[57:45.900 --> 57:47.900]  And you can talk about all this.
[57:48.020 --> 57:49.280]  My email address...
[57:49.920 --> 57:51.840]  You can find out my email address, I'm sure.
[57:51.920 --> 57:52.680]  I won't insult you.
[57:53.860 --> 57:55.080]  D-O-M-O.
[57:55.180 --> 57:55.920]  D-O-M-O.
[57:56.100 --> 57:56.460]  Majordomo.
[57:56.580 --> 57:58.360]  It's the major list server thing.
[57:59.300 --> 58:00.560]  Matt Blaze is going to talk to you.
[58:00.660 --> 58:01.500]  Matt is a researcher.
[58:02.320 --> 58:02.760]  No, no, no.
[58:03.200 --> 58:05.600]  I don't work for anybody, at least not tonight.
[58:06.900 --> 58:07.780]  Oh, I'm...
[58:08.720 --> 58:10.640]  Matt works for Crypto.com.
[58:13.340 --> 58:15.680]  Yeah, so just a quick question.
[58:16.300 --> 58:28.340]  How many of you use cryptography or consider yourself cryptography hackers or use PGP or play with this stuff in any way?
[58:28.340 --> 58:29.160]  shape or form?
[58:30.140 --> 58:30.760]  Okay.
[58:30.860 --> 58:31.680]  How many of you don't?
[58:32.900 --> 58:33.520]  Okay.
[58:34.040 --> 58:35.380]  Sorry, I have a...
[58:35.380 --> 58:45.600]  Just a sort of quick sales pitch, which is why, you know, people who are hackers really should be interested in cryptography.
[58:48.120 --> 58:51.080]  Cryptography is an enormous amount of fun.
[58:51.560 --> 58:55.920]  And it allows you to do some sort of seemingly impossible things.
[58:55.960 --> 59:08.320]  And that's the sort of hackish essence of where a lot of capability lies, particularly the kind of cryptography that's emerged over the last 20 years in the civilian world.
[59:08.320 --> 59:08.520]  Right?
[59:08.600 --> 59:15.380]  There are things that are possible to do or that we know how to do today that simply didn't exist and weren't possible.
[59:15.640 --> 59:15.700]  Right?
[59:16.180 --> 59:17.540]  Just as with computers.
[59:17.700 --> 59:35.520]  Well, the same thing with this branch of what is essentially mathematics and its application, which 20 years ago involved things that only very, very specialized people in very, very specialized places would have even heard about, is now something that's available,
[59:35.960 --> 59:39.220]  at least in theory, to everyone.
[59:40.080 --> 59:41.840]  But the practice is...
[59:41.840 --> 59:42.320]  Where's the crime?
[59:42.780 --> 59:43.180]  Yeah.
[59:43.420 --> 59:43.600]  Yeah.
[59:43.720 --> 59:44.600]  The practice...
[59:44.600 --> 59:45.620]  I'm tired of this nerd shit.
[59:45.780 --> 59:46.740]  Where's the crime?
[59:47.120 --> 59:49.200]  We came here by crime.
[59:52.420 --> 59:53.180]  Yeah.
[59:53.640 --> 59:56.380]  There's plenty of opportunity for crime.
[59:58.360 --> 01:00:09.780]  The point is, in the last 20 years, the basic techniques to do all sorts of impossible-seeming things have suddenly emerged.
[01:00:10.240 --> 01:00:10.300]  Right?
[01:00:10.620 --> 01:00:11.180]  If you...
[01:00:11.180 --> 01:00:11.580]  Sorry!
[01:00:12.180 --> 01:00:12.300]  Yeah.
[01:00:12.460 --> 01:00:12.720]  All right.
[01:00:12.780 --> 01:00:13.240]  Never mind.
[01:00:14.560 --> 01:00:15.220]  All right.
[01:00:15.300 --> 01:00:16.540]  The point is...
[01:00:16.540 --> 01:00:17.420]  The vet is...
[01:00:17.420 --> 01:00:18.120]  Yeah.
[01:00:18.820 --> 01:00:19.100]  Yeah.
[01:00:19.100 --> 01:00:19.160]  Right.
[01:00:19.320 --> 01:00:19.500]  Yeah.
[01:00:20.500 --> 01:00:21.080]  All right.
[01:00:21.980 --> 01:00:25.360]  So it's possible to do things that don't seem possible.
[01:00:26.480 --> 01:00:26.600]  All right.
[01:00:27.420 --> 01:00:28.100]  You can...
[01:00:30.440 --> 01:00:31.320]  Never mind.
[01:00:32.160 --> 01:00:33.560]  Use a random number.
[01:00:33.700 --> 01:00:33.840]  Yeah.
[01:00:33.920 --> 01:00:34.620]  Go to jail.
[01:00:34.620 --> 01:00:34.980]  Yeah.
[01:00:35.820 --> 01:00:36.020]  Yeah.
[01:00:36.760 --> 01:00:37.160]  They...
[01:00:37.160 --> 01:00:37.660]  Yeah.
[01:00:38.120 --> 01:00:38.400]  All right.
[01:00:38.880 --> 01:00:41.400]  So, the first thing is that this is a lot of fun.
[01:00:41.660 --> 01:00:43.580]  It's a lot of fun for a lot of reasons.
[01:00:43.580 --> 01:00:44.840]  It's orange!
[01:00:45.280 --> 01:00:45.580]  All right.
[01:00:47.140 --> 01:00:47.600]  All right.
[01:00:48.420 --> 01:00:49.320]  All right.
[01:00:49.700 --> 01:00:50.140]  All right.
[01:00:50.680 --> 01:00:51.140]  All right.
[01:00:51.160 --> 01:00:52.080]  All right.
[01:00:53.700 --> 01:00:53.800]  All right.
[01:00:53.800 --> 01:00:55.080]  Well, there's no point in going on.
[01:00:58.500 --> 01:00:59.660]  Take care of him.
[01:01:01.840 --> 01:01:03.080]  Take it into your own hands.
[01:01:03.280 --> 01:01:03.500]  All right.
[01:01:03.840 --> 01:01:05.540]  I can't speak over all this noise.
[01:01:05.800 --> 01:01:07.100]  It's just too much work.
[01:01:07.940 --> 01:01:08.880]  Go for it.
[01:01:09.120 --> 01:01:09.420]  Go for it.
[01:01:09.800 --> 01:01:10.240]  All right.
[01:01:10.320 --> 01:01:10.580]  The...
[01:01:10.580 --> 01:01:10.660]  The...
[01:01:10.660 --> 01:01:11.080]  The...
[01:01:13.040 --> 01:01:14.600]  Oh, here's the crime.
[01:01:14.780 --> 01:01:15.520]  That's a...
[01:01:15.520 --> 01:01:15.800]  Yeah.
[01:01:16.380 --> 01:01:17.060]  All right.
[01:01:17.640 --> 01:01:21.120]  So the basic techniques to do things that just don't seem possible.
[01:01:21.340 --> 01:01:34.720]  And essentially, it's hackers who have the... who have the ability and the inclination to do things that aren't available to other... to people who haven't had the inclination to play with them.
[01:01:34.720 --> 01:01:42.140]  In other words, this is a brand new field with the opportunity to do things by playing that other people can't do.
[01:01:42.300 --> 01:01:45.420]  And in fact, lots of other people don't really want you to do.
[01:01:45.560 --> 01:01:52.660]  It's an opportunity to protect yourself in very personal ways without having to trust and rely on other people.
[01:01:53.220 --> 01:01:58.360]  Now, there are really three ways in which you can play with this.
[01:01:58.360 --> 01:02:06.660]  And one is at the very basic low level of trying to figure out mathematical techniques for doing these impossible things.
[01:02:06.840 --> 01:02:13.520]  The problem with playing with cryptography at the really low level is that it's amazingly hard.
[01:02:13.700 --> 01:02:14.260]  Right?
[01:02:14.260 --> 01:02:20.240]  In fact, almost everybody who tries to invent new ciphers gets it wrong.
[01:02:20.820 --> 01:02:29.360]  Even people with enormous amounts of specialized training and lots of other incredibly smart people to help them and look at it still get it wrong.
[01:02:29.520 --> 01:02:30.120]  Right?
[01:02:30.300 --> 01:02:33.200]  This is incredibly dangerous to play with.
[01:02:33.200 --> 01:02:51.620]  And unlike other kinds of dangerous things you do, typically when you try to invent your own cipher, you end up exposing your own information to somebody who's just a little bit smarter than you or somebody who has the access to information that you just didn't happen to know.
[01:02:51.620 --> 01:03:01.120]  But fortunately, good techniques for doing this sort of stuff have come out in the last 20 years that seem to be pretty good.
[01:03:01.120 --> 01:03:06.300]  So it's not necessary to understand how to invent new ciphers in order to play with this.
[01:03:06.300 --> 01:03:30.100]  So there's sort of a next level of thing that hasn't been out there and hasn't expanded out into the world of communications and computing, which is understanding the basic techniques for using these building blocks, these mathematical techniques, to accomplish secure things.
[01:03:30.220 --> 01:03:34.360]  That's inventing the protocols for using cryptography in good ways.
[01:03:34.360 --> 01:03:40.560]  Well, that's easier, but it turns out that's also really hard and you have to be really careful.
[01:03:41.220 --> 01:03:46.220]  There are ways to do subtle things that cause stuff to break all of a sudden.
[01:03:46.660 --> 01:03:47.140]  Right?
[01:03:47.700 --> 01:03:59.620]  So it's possible to be very subtle and be a little bit more clever than the person who designed the way to use encryption or signature techniques.
[01:03:59.620 --> 01:04:06.240]  And it turns out that there's something very simple you can do that lets you subvert the original purpose.
[01:04:06.500 --> 01:04:14.320]  So it's dangerous, but not quite as dangerous as inventing new ciphers, as inventing new ways of using cryptographic techniques.
[01:04:14.820 --> 01:04:19.860]  But it turns out that there are well-established techniques for doing lots of things like that.
[01:04:19.860 --> 01:04:32.860]  The part where almost nothing has been done, and we pretty much know how to do it, but it still isn't being done, is actually using cryptography to protect things.
[01:04:33.040 --> 01:04:41.820]  So we have these basic techniques for doing seemingly magical things like sending a secret message to somebody you haven't spoken to before.
[01:04:42.460 --> 01:04:44.320]  And we know how to do that.
[01:04:44.480 --> 01:04:48.440]  But it turns out that in real life, it's just not used.
[01:04:49.400 --> 01:04:49.480]  Right?
[01:04:51.580 --> 01:05:02.480]  There's a wonderful opportunity to get a grasp on what these basic techniques are and put them into practice where they just don't exist.
[01:05:02.480 --> 01:05:11.740]  And you can put this into practice where it just doesn't exist to protect your own communication and your own information.
[01:05:11.840 --> 01:05:27.280]  And to secure things like messages that you want to establish came from you actually having come from you in secure ways, preventing forgery and so on.
[01:05:27.440 --> 01:05:31.480]  Just by applying some of these techniques and putting them into real systems, right?
[01:05:31.480 --> 01:05:34.620]  That's what everybody in this room is good at, right?
[01:05:34.820 --> 01:05:37.960]  We're precisely the people who are good at doing this.
[01:05:38.120 --> 01:05:45.620]  And essentially, we've been handed on a silver platter these basic tools to do it, and it's just a matter of putting them into real systems.
[01:05:45.860 --> 01:05:49.800]  So my sort of advertisement is go get excited about these techniques.
[01:05:50.100 --> 01:05:57.860]  There's a little bit to learn in order to use them safely, but there are good books on this and good information out there.
[01:05:57.860 --> 01:06:00.920]  Applied cryptography by Bruce Schneier is a wonderful book.
[01:06:02.260 --> 01:06:07.140]  It's available in Barnes and Noble that has all the building blocks.
[01:06:07.400 --> 01:06:16.040]  I'll be doing a tutorial on this tomorrow for about half an hour on some of the very basics of the technique for those of you who aren't familiar with them.
[01:06:16.040 --> 01:06:21.520]  So the opportunities are there, and this is an awful lot of fun to play with.
[01:06:22.140 --> 01:06:22.640]  So, anyway.
[01:06:23.300 --> 01:06:23.520]  Thanks.
[01:06:23.680 --> 01:06:27.680]  And for those of you who want to see a crime, Bernie S.
[01:06:28.000 --> 01:06:37.120]  has in his possession in front of us two honest-to-god AT&T 3600 devices with your favorite and my chip inside.
[01:06:37.120 --> 01:06:38.040]  So...
[01:06:38.040 --> 01:06:38.340]  Yeah!
[01:06:39.140 --> 01:06:40.640]  I think that's a crime.
[01:06:40.800 --> 01:06:41.580]  What do you guys think?
[01:06:48.040 --> 01:06:49.760]  And an actual clipper chip.
[01:06:52.640 --> 01:06:53.800]  That's an actor flipper joke.
[01:06:54.000 --> 01:06:55.140]  Why does this say name in the theory?
[01:06:55.380 --> 01:06:55.740]  I don't know.
[01:06:57.800 --> 01:06:58.500]  There it is.
[01:06:58.800 --> 01:06:59.080]  You don't.
[01:07:00.720 --> 01:07:01.140]  I don't know.
[01:07:01.840 --> 01:07:03.660]  I don't know.
[01:07:03.660 --> 01:07:04.280]  I don't know.
[01:07:05.400 --> 01:07:06.440]  I don't know.
[01:07:09.800 --> 01:07:11.380]  Don't let the static charge.
[01:07:12.020 --> 01:07:12.600]  Yeah, really.
[01:07:15.980 --> 01:07:17.480]  You might want to use this one.
[01:07:17.480 --> 01:07:19.140]  That one's good.
[01:07:19.260 --> 01:07:19.700]  That one's good.
[01:07:21.420 --> 01:07:23.020]  What is the social engineering?
[01:07:24.220 --> 01:07:26.900]  I had to do a lot to get these clipper chips, believe me.
[01:07:31.080 --> 01:07:34.300]  We're all aware how much controversy there's been about these things.
[01:07:36.760 --> 01:07:46.140]  And it occurred to me when Emmanuel and I were planning this conference that this was going to be a topic of conversation.
[01:07:46.140 --> 01:08:02.060]  And no one I knew or even heard of had ever seen a clipper chip, used a clipper chip, held a clipper chip, anything, had really any contact with a clipper chip.
[01:08:02.280 --> 01:08:08.220]  And it seemed like this ambiguous thing that none of us really had a grasp on.
[01:08:08.220 --> 01:08:11.240]  I mean, we read a lot about it, but it wasn't a concrete thing.
[01:08:11.360 --> 01:08:13.080]  It was just sort of like this ghost out there.
[01:08:13.080 --> 01:08:24.680]  So I decided to find out for myself what this thing is, get my hands on it, and run it through its paces, and find out for myself what this is about.
[01:08:24.800 --> 01:08:31.440]  Try to make it seem more like a concrete thing than just this specter that seems to be floating around us.
[01:08:31.440 --> 01:08:38.680]  So I did some research and I found out that AT&T manufactures a device.
[01:08:38.980 --> 01:08:49.500]  In fact, to my knowledge, and maybe I'll be corrected on this, AT&T is still the only commercial company using the clipper chip in a commercial product.
[01:08:49.520 --> 01:08:50.140]  Is that right?
[01:08:50.780 --> 01:08:53.200]  I think they're the only one that's announced one.
[01:08:53.200 --> 01:08:53.680]  Yeah.
[01:08:54.120 --> 01:08:55.100]  No, they're not, actually.
[01:08:55.260 --> 01:08:55.520]  Oh, really?
[01:08:55.680 --> 01:08:56.040]  Who else?
[01:08:56.200 --> 01:08:58.160]  There's some really low-end stuff.
[01:08:58.300 --> 01:08:58.760]  Don't worry about it.
[01:08:58.820 --> 01:08:59.960]  They're the only major manufacturers.
[01:09:00.020 --> 01:09:00.260]  Okay.
[01:09:02.300 --> 01:09:10.780]  My research turned out the only company I could find out who was making anything that used a clipper chip was AT&T, a friend of mine.
[01:09:13.960 --> 01:09:27.200]  And I think I remember reading something to the effect that there was some sort of controversy about AT&T signing the EFF.
[01:09:28.660 --> 01:09:44.900]  There are people at AT&T on the Digital Privacy and Security Working Group, which is a group of people who ostensibly are telecommunications providers, equipment providers, service providers, who are interested in fighting things like...
[01:09:44.900 --> 01:09:45.840]  Clipper chip.
[01:09:45.900 --> 01:09:46.380]  Clipper chip.
[01:09:48.000 --> 01:09:52.520]  And I found that to be quite hypocritical.
[01:09:52.760 --> 01:09:54.360]  But anyway, I wanted to get to the bottom of it.
[01:09:54.640 --> 01:09:59.480]  So I contacted AT&T and found out what their product line consisted of.
[01:09:59.480 --> 01:10:02.660]  And they actually make several clipper chip-based products.
[01:10:03.720 --> 01:10:25.500]  According to Dennis Troilo, the head of the AT&T Secure Communications Products Division, their most popular product is the AT&T Surety, that's S-U-R-I-T-Y Surety Telephone Device 3600 for the protection of sensitive information.
[01:10:25.500 --> 01:10:26.740]  That's the name of the product.
[01:10:29.680 --> 01:10:35.220]  Its engineering name is just the 3600 C, C meaning clipper.
[01:10:35.980 --> 01:10:42.600]  They make other encryption products that don't have the C after them, which don't use a clipper chip.
[01:10:42.840 --> 01:10:44.920]  So that's the most popular product.
[01:10:45.080 --> 01:10:55.720]  They make another product that's designed specifically for cellular communications security, which we all know is a real issue.
[01:10:56.340 --> 01:11:05.240]  This consists of an actual cellular telephone, which is manufactured by AT&T, which has been modified to contain a clipper chip.
[01:11:05.260 --> 01:11:08.960]  And it actually encrypts the audio inside the phone itself.
[01:11:08.980 --> 01:11:11.680]  It's not an add-on device like the 3600 C is.
[01:11:11.680 --> 01:11:22.440]  They also manufacture two other devices, the 3700 and 3710, which are designed specifically for facsimile encryption and decryption.
[01:11:23.480 --> 01:11:39.660]  When I asked Dennis Troilo about this, why there were all these different products, why couldn't they make one that would both encrypt facsimile and voice communications, since they're both analog signals, he said that their engineers just tried and tried,
[01:11:39.680 --> 01:11:40.860]  and they just couldn't do it.
[01:11:41.520 --> 01:11:48.520]  So anyway, I guess they couldn't afford the best engineers at AT&T.
[01:11:49.800 --> 01:12:04.980]  But anyway, so I inquired about what it would actually cost me to obtain a pair of these 3600 C clipper chip devices, which are intended for the, for, let's see what, I think it, this little guide here, it says, who the intended audience for this is.
[01:12:05.840 --> 01:12:07.040]  What did it say here?
[01:12:07.460 --> 01:12:10.080]  It was in one of their little advertising blurbs.
[01:12:14.460 --> 01:12:25.640]  Easy to use, as portable as a calculator, simple to install, compatible design and superior performance, security you can count on, comprehensive service and support, and et cetera.
[01:12:32.640 --> 01:12:47.800]  They described the types of people who would be planning to use this, but, I can't find it right now, but basically it was like a business people, corporate executives who did not want their confidential business information overheard, et cetera.
[01:12:49.380 --> 01:13:05.280]  One thing I found ironic is that while they say it's easy to use, simple to install, as portable as a calculator, security you can count on, there was this interesting little disclaimer that I saw in the back of the owner's manual in really fine print,
[01:13:05.280 --> 01:13:22.800]  which reads as the following, quote, AT&T makes no warranty that the 3600 will prevent cryptanalytic attack on any encrypted transmission by any government agency, its agents, or any third parties.
[01:13:23.440 --> 01:13:25.460]  That pretty much includes everybody, I think.
[01:13:28.740 --> 01:13:31.160]  Like only that five-year-olds can't decode it, I guess.
[01:13:31.300 --> 01:13:39.880]  Furthermore, AT&T makes no warranty that the 3600 will prevent any attack on any communication by methods which bypass encryption.
[01:13:40.900 --> 01:13:51.340]  Notwithstanding any other contractual provisions, the above rights and remedies are exclusive, and all implied warranties of merchantability and fitness for a particular purpose are excluded from any obligation contained in this warranty.
[01:13:51.580 --> 01:13:55.640]  In addition, AT&T should not be liable for any inconsequential incidental damages.
[01:13:55.660 --> 01:14:00.120]  In other words, they're not responsible in any way, shape, or form if this doesn't work out for you.
[01:14:00.640 --> 01:14:03.120]  It could be an empty box with two phone jacks on it, it wouldn't it?
[01:14:03.220 --> 01:14:03.620]  Exactly.
[01:14:04.120 --> 01:14:12.020]  Paradine actually got in trouble for that at one point when they were selling boxes to the government that had blinky lights and someone noticed that the data coming out was the same as the data going out.
[01:14:16.300 --> 01:14:23.420]  So, I decided to take the plunge and find out what I'd have to lay out to obtain one of these things.
[01:14:23.580 --> 01:14:28.780]  Obviously, one of them wasn't going to do many good to test because I would have to get two of them, catch 22.
[01:14:29.100 --> 01:14:37.320]  So, it turns out the suggested list price and the only... that is the price because you cannot buy it from anyone other than AT&T directly.
[01:14:37.320 --> 01:14:41.320]  They do not sell through dealers, this particular product anyway.
[01:14:42.280 --> 01:14:44.100]  They're $1,300 each.
[01:14:45.820 --> 01:14:47.600]  And that's this little thing here.
[01:14:47.940 --> 01:14:50.000]  And it adds up to $2,600.
[01:14:58.080 --> 01:15:00.120]  Some kind of symbiosis going on there.
[01:15:00.680 --> 01:15:01.220]  Anyway.
[01:15:03.180 --> 01:15:07.840]  So, after much faxing back and forth...
[01:15:07.840 --> 01:15:10.360]  Actually, it took me a while to...
[01:15:10.360 --> 01:15:11.540]  Oh, never mind.
[01:15:11.720 --> 01:15:18.620]  I did a bunch of faxing back and forth with this head of the product sales for this particular product at AT&T.
[01:15:18.880 --> 01:15:22.920]  And convinced him that I was a contractor.
[01:15:22.920 --> 01:15:24.720]  I am a consultant, a computer consultant, freelance.
[01:15:25.280 --> 01:15:29.740]  I convinced him that I had certain clients who did work for the government, which is true.
[01:15:30.500 --> 01:15:35.120]  And they were concerned about the security of their confidential information, which is also true.
[01:15:35.260 --> 01:15:37.980]  It really had nothing to do with what I did for them, but I didn't lie to them.
[01:15:39.360 --> 01:15:44.600]  Bottom line is I said that I really needed these to communicate with this client, which wasn't necessarily true.
[01:15:44.600 --> 01:15:47.180]  And we needed to evaluate them.
[01:15:48.860 --> 01:15:57.080]  So, he said that they could send them to me on a 30-day net basis, provided I supply them with appropriate credit references, which I was able to do, fortunately.
[01:15:57.700 --> 01:15:59.580]  And about a week later, these two...
[01:15:59.580 --> 01:16:06.380]  This large box from AT&T arrived in the mail, which was about three by three feet cube.
[01:16:06.520 --> 01:16:07.240]  A three-foot cube.
[01:16:07.780 --> 01:16:09.880]  And in it were two of these.
[01:16:10.480 --> 01:16:13.480]  And a whole lot of styrofoam peanuts.
[01:16:15.580 --> 01:16:23.820]  And actually, just to give you an idea of what the commercial packaging of this product looks like, this is the box that it comes in.
[01:16:25.560 --> 01:16:29.180]  The Surety telephone device, intended for the consumer market.
[01:16:29.300 --> 01:16:30.940]  Looks like my SOC box.
[01:16:31.060 --> 01:16:35.280]  Has a lot of really interesting codes on the side.
[01:16:35.380 --> 01:16:39.780]  It says comcode, which next to the serial number, this is like barcode.
[01:16:40.140 --> 01:16:42.940]  Evidently, maybe you can make something out of that.
[01:16:42.940 --> 01:16:44.160]  I actually have done work...
[01:16:44.160 --> 01:16:44.900]  Tell me what that is.
[01:16:45.020 --> 01:16:50.140]  The comcode is a standard product designation for almost anything that exists within AT&T.
[01:16:50.380 --> 01:16:50.680]  Oh, okay.
[01:16:50.920 --> 01:16:51.900]  Standard ID code.
[01:16:52.200 --> 01:16:52.540]  Gotcha.
[01:16:52.840 --> 01:16:53.740]  Yeah, here you go.
[01:16:54.060 --> 01:16:56.200]  They have a lot of stuff, and they need to keep track of it.
[01:16:58.660 --> 01:16:59.700]  It's not a scary one.
[01:16:59.760 --> 01:17:01.760]  Available in AT&T white or black.
[01:17:01.960 --> 01:17:02.200]  Yes.
[01:17:02.980 --> 01:17:05.220]  Now, if I'd known it came in black, I would have gotten it.
[01:17:05.280 --> 01:17:06.180]  But they sent me the white ones.
[01:17:06.380 --> 01:17:07.220]  Why is it in white and black?
[01:17:07.620 --> 01:17:07.780]  Yes.
[01:17:08.260 --> 01:17:09.640]  No, it's AT&T white.
[01:17:10.160 --> 01:17:10.480]  Yeah.
[01:17:10.700 --> 01:17:11.440]  AT&T white.
[01:17:11.600 --> 01:17:11.760]  Not...
[01:17:11.760 --> 01:17:11.840]  It's white.
[01:17:12.180 --> 01:17:12.820]  It's a special...
[01:17:13.200 --> 01:17:15.360]  They have copyrighted that color, by the way.
[01:17:15.560 --> 01:17:15.760]  Yeah.
[01:17:15.940 --> 01:17:17.860]  They got a patent on it, I think, just recently.
[01:17:18.340 --> 01:17:18.800]  It's in white.
[01:17:19.220 --> 01:17:19.900]  It's in white.
[01:17:20.060 --> 01:17:20.220]  Yeah.
[01:17:21.440 --> 01:17:22.880]  So these things came in the mail.
[01:17:23.020 --> 01:17:24.140]  And oh boy, was I excited.
[01:17:26.540 --> 01:17:27.780]  While I was waiting...
[01:17:27.780 --> 01:17:30.780]  Actually, it took about two weeks for them to send them to me, because they were...
[01:17:31.980 --> 01:17:33.160]  They did some...
[01:17:33.160 --> 01:17:34.120]  No, they weren't backordered.
[01:17:34.220 --> 01:17:39.020]  They evidently double-check everybody they send these things to.
[01:17:40.620 --> 01:17:44.340]  And I don't know what affiliations they came up with in regards to myself.
[01:17:45.580 --> 01:17:50.700]  But I was able to allay this gentleman's fears through a lot of social engineering.
[01:17:50.880 --> 01:17:53.280]  But anyway, they came about two weeks later.
[01:17:53.280 --> 01:17:55.140]  Meanwhile, while I was waiting for things to...
[01:17:55.580 --> 01:18:02.360]  Hoping for these things to come, I thought, wouldn't it be great to actually get one of these clipper chips?
[01:18:02.740 --> 01:18:03.680]  The chip itself.
[01:18:03.900 --> 01:18:06.300]  Not a box that happened to have one soldered inside it.
[01:18:06.400 --> 01:18:09.700]  But just the chip to hold it in my hand and just gawk at it.
[01:18:10.080 --> 01:18:14.080]  So I found out, you know, where these things come from.
[01:18:14.340 --> 01:18:16.280]  I found out where I thought they came from.
[01:18:17.920 --> 01:18:21.000]  And it sort of went around the rosy on that thing.
[01:18:21.000 --> 01:18:26.800]  It turns out the NSA contracted with Mycotronics to develop the actual silicon.
[01:18:27.140 --> 01:18:27.560]  Is that correct?
[01:18:27.860 --> 01:18:28.100]  Yeah.
[01:18:29.080 --> 01:18:39.680]  If you go to the machine, ftp.csua.berkeley.edu, there's a directory on the clipper chip of a bunch of the original press releases and some of the press coverage.
[01:18:39.800 --> 01:18:43.400]  And some dumpster dive information that a guy got out of Mycotronics.
[01:18:43.400 --> 01:18:43.720]  Cool.
[01:18:44.520 --> 01:18:51.860]  By the way, if you want to check their dumpsters, their dumpsters are located at 357 Van Nez Way, Suite 200 in Torrance, California.
[01:18:54.240 --> 01:18:57.700]  It's ftp.csua.berkeley.edu.
[01:18:59.020 --> 01:19:05.380]  While I'm talking, the event here is, the Society for Electronic Access is having me speak on Tuesday night.
[01:19:05.380 --> 01:19:07.120]  So that's what the event is.
[01:19:07.220 --> 01:19:08.040]  I just wanted to give the plug.
[01:19:08.160 --> 01:19:09.540]  And I'll give you the ftp address in the bottom of there.
[01:19:10.000 --> 01:19:10.180]  Sorry.
[01:19:10.360 --> 01:19:10.680]  That's all right.
[01:19:15.850 --> 01:19:16.570]  Where is that pen?
[01:19:16.730 --> 01:19:17.210]  There you go.
[01:19:17.330 --> 01:19:17.910]  You got it?
[01:19:18.930 --> 01:19:19.250]  No.
[01:19:20.070 --> 01:19:21.610]  So while you're doing that...
[01:19:22.210 --> 01:19:30.050]  So I thought, wouldn't it be great to actually have a clipper chip and be able to show that to people here at the HOPE conference and do with it as we pleased?
[01:19:32.190 --> 01:19:39.850]  So I decided that was going to be my goal, to actually get a raw clipper chip that wasn't built into some device and I wouldn't be able to see it.
[01:19:41.330 --> 01:19:50.250]  So I contacted Mycotronics first, who evidently was contracted by the National Security Agency, NSA, to develop this product.
[01:19:50.250 --> 01:20:01.410]  And after talking with some of the people at Mycotronics, I was told that the chip is actually manufactured by VLSI Technology out in San Jose, California.
[01:20:02.230 --> 01:20:07.270]  This is something I had not read anywhere else until I got this information from the folks inside of Mycotronics.
[01:20:07.430 --> 01:20:20.950]  So I contacted the folks at VLSI Technology and went through like three levels of management before I determined that they just didn't have the authorization to give me a clipper chip as a sample or to look at or whatever.
[01:20:22.030 --> 01:20:23.730]  And I got kicked back to Mycotronics.
[01:20:23.890 --> 01:20:27.850]  And finally, I reached the right person at Mycotronics, who I convinced that...
[01:20:27.850 --> 01:20:28.490]  John Droege.
[01:20:30.430 --> 01:20:31.550]  Exactly, yes, John Droege.
[01:20:31.630 --> 01:20:33.030]  John Droege, product manager.
[01:20:33.050 --> 01:20:38.790]  John Droege is the vice president of program development at Mycotronics, and this particular program being the clipper chip.
[01:20:39.250 --> 01:20:41.310]  How do you know this guy, John Droege?
[01:20:41.410 --> 01:20:41.690]  I'm curious.
[01:20:42.790 --> 01:20:49.610]  We had long... one of my business partners had long conversations with people John Droege about clipper technical specs when it first came out.
[01:20:51.150 --> 01:21:03.210]  So I actually was able to at least make John feel that I was a trusted friend and talked to him for about a week or so back and forth, and we've shared some common interests and so forth.
[01:21:05.210 --> 01:21:06.090]  Social engineering.
[01:21:06.510 --> 01:21:11.410]  And bottom line is I said, you know, there's so much negative publicity about this thing floating around, John.
[01:21:11.410 --> 01:21:13.810]  It would be great if I could just show people...
[01:21:13.810 --> 01:21:16.710]  This little chip is not such a sinister thing after all.
[01:21:25.750 --> 01:21:27.390]  And he couldn't agree with me more.
[01:21:34.770 --> 01:21:45.030]  And I said, wouldn't it be great if we could actually pass this thing around and show probably 1,000 specialists in the computer field...
[01:21:48.270 --> 01:21:50.090]  ...MIS directors and such...
[01:21:50.090 --> 01:21:54.310]  ...that this was something that could really benefit all of them.
[01:21:54.950 --> 01:21:56.730]  And enhance your market share also.
[01:21:58.350 --> 01:22:00.330]  So that... he took the bait.
[01:22:00.950 --> 01:22:04.970]  And FedEx arrived the next day with two clipper chips.
[01:22:05.430 --> 01:22:06.490]  How big was a box?
[01:22:06.970 --> 01:22:08.410]  It was just a FedEx envelope.
[01:22:13.740 --> 01:22:16.560]  Evidently not as much in these things as I had expected.
[01:22:16.660 --> 01:22:17.460]  I expected some...
[01:22:17.460 --> 01:22:23.460]  You know, if they're manufactured by VLSI out in San Jose, I expected this very large-scale integrated, you know, like this...
[01:22:25.100 --> 01:22:31.880]  The VLSI technology they're using is not actually for the large-scale integration but for the anti-tamper properties in the chip.
[01:22:32.140 --> 01:22:33.240]  There's some...
[01:22:33.240 --> 01:22:40.740]  As far as we can tell, and these are classified details so far, and we haven't gotten enough clipper chips in large quantities to start ashing them.
[01:22:41.880 --> 01:22:44.980]  We have friends at FABs.
[01:22:46.280 --> 01:22:47.320]  We'll say no more now.
[01:22:47.520 --> 01:22:54.980]  Anyway, the part of the cryptography algorithm is kept in special memory cells which can't be scanned in from outside.
[01:22:55.120 --> 01:22:58.360]  And VLSI has a lot of very specific silicon expertise in doing this.
[01:22:58.360 --> 01:23:01.220]  This is why they're using VTI rather than another FAB.
[01:23:04.140 --> 01:23:05.460]  So I was glad they were plastic.
[01:23:05.600 --> 01:23:12.360]  I'd hoped to use some sort of solvent to dissolve the plastic, which Simuramic wouldn't do.
[01:23:12.460 --> 01:23:15.660]  But I thought it might make a nice tie tack or piece of jewelry.
[01:23:15.660 --> 01:23:18.160]  So anyway, I got these two clipper chips.
[01:23:18.680 --> 01:23:24.500]  And then about a week after the chips arrived, the AT&T Surety device...
[01:23:24.500 --> 01:23:29.760]  The AT&T Surety telephone device 3600 for the protection of sensitive information arrived in the mail in that big box.
[01:23:30.100 --> 01:23:32.540]  So I went to my friend John's house.
[01:23:33.380 --> 01:23:34.920]  We wanted to test this thing out.
[01:23:35.000 --> 01:23:48.040]  And since I can't talk to myself and talk and hear myself at the same time very easily, I need another party to evaluate this since it's got to be a two-way communication between two people to really thoroughly evaluate it.
[01:23:49.020 --> 01:23:51.120]  And so I went up to his house rather than...
[01:23:51.120 --> 01:23:53.520]  I would have had to deliver, you know, mail it to him.
[01:23:53.580 --> 01:24:00.360]  So I just went up to his house and he and I both own a device called a Phantom central office simulator.
[01:24:00.360 --> 01:24:02.980]  We use it to test modems and things on the road.
[01:24:03.080 --> 01:24:06.900]  And what it basically does is it generates a line voltage.
[01:24:07.300 --> 01:24:15.300]  And when you pick one phone, if you connect two telephones to it and take one off the hook, then it sends a ringing voltage to the other phone.
[01:24:15.620 --> 01:24:18.720]  And you can say, hi there.
[01:24:19.540 --> 01:24:20.880]  And it provides talk voltage.
[01:24:21.140 --> 01:24:27.740]  So I thought that would be an ideal way for us to test this in a controlled environment and see how well it worked.
[01:24:28.380 --> 01:24:29.600]  So we took them out of the box.
[01:24:29.620 --> 01:24:32.500]  We read the instructions and plugged them in.
[01:24:33.080 --> 01:24:36.720]  And the way they actually are configured is this small box.
[01:24:37.220 --> 01:24:39.540]  It's kind of interesting the way they have this thing designed.
[01:24:39.720 --> 01:24:41.440]  I don't think it makes a lot of sense, frankly, but...
[01:24:41.440 --> 01:24:42.280]  I can explain that.
[01:24:42.420 --> 01:24:42.720]  Go ahead.
[01:24:42.880 --> 01:24:43.140]  Okay.
[01:24:44.620 --> 01:24:48.320]  The device does not actually connect between the telephone line and the telephone itself.
[01:24:48.320 --> 01:24:52.280]  It connects between the telephone handset and the telephone.
[01:24:52.280 --> 01:25:04.560]  And so that way, without getting into too many technical details, it just focuses on the audio signals without having to deal with the two to four wire line conversion in a phone and all that stuff.
[01:25:05.460 --> 01:25:11.720]  So, first of all, you must use this only with telephones that have modular handset jacks.
[01:25:11.720 --> 01:25:19.900]  A lot of telephones, which may well be modular, have handset jacks, which have handsets which are permanently attached or maybe one piece phones.
[01:25:20.040 --> 01:25:21.520]  You cannot use this with any of those phones.
[01:25:21.600 --> 01:25:27.500]  It has to be a phone that has the handsets that are detachable like this.
[01:25:31.300 --> 01:25:32.900]  I scared myself there.
[01:25:36.200 --> 01:25:37.820]  I went through...
[01:25:38.480 --> 01:25:44.640]  My friend and I, John, we went through three different types of telephones to no avail, trying to get these damn things to work.
[01:25:44.920 --> 01:25:46.920]  We couldn't get squat through them.
[01:25:47.440 --> 01:25:49.560]  No audio would pass through them at all.
[01:25:49.920 --> 01:25:51.180]  So, I thought...
[01:25:51.180 --> 01:26:02.180]  Well, and the other complexity to the design of this thing is evidently the telephone industry did not arrive at any kind of standards in regards to telephone handset voltages.
[01:26:02.460 --> 01:26:03.460]  Let me shut this thing off.
[01:26:04.060 --> 01:26:17.720]  So, you could have, you know, a hundred different telephone manufacturers and a hundred different protocols or standards in regards to how much voltage and current goes through these and impedance on these handset jacks.
[01:26:17.860 --> 01:26:30.540]  So, what AT&T decided to do to make this thing as they thought as compatible as possible with different phones in the market is to come up with these different modules.
[01:26:30.920 --> 01:26:40.820]  And these modules, the 3600C comes with five such modules, which basically are a cartridge about the size of a video game cartridge that contain a ROM.
[01:26:41.220 --> 01:26:51.380]  And this ROM or PROM contains the information that tells the 3600C what personality or what type of telephone it's being used with.
[01:26:51.380 --> 01:26:52.180]  Did you open them up?
[01:26:52.920 --> 01:26:53.680]  No, but...
[01:26:53.680 --> 01:26:55.340]  Oh, but they are.
[01:26:55.480 --> 01:26:57.040]  This is the point you were making before.
[01:26:58.940 --> 01:27:00.660]  This will not stay whole for long.
[01:27:02.660 --> 01:27:08.240]  The reason they have these compatibility modules, as they call them, is that they wanted to go after the corporate networks here.
[01:27:08.300 --> 01:27:11.180]  And a lot of corporations are running digital PBXs.
[01:27:11.320 --> 01:27:16.280]  And they didn't want to have to demodulate all the different varieties of digital data going over the two wires.
[01:27:16.280 --> 01:27:20.380]  So instead, what they did is they did... these are mostly passive devices, I understand.
[01:27:20.500 --> 01:27:22.560]  Maybe they have op amps for voltage following in them.
[01:27:22.640 --> 01:27:26.420]  But it just does line equalization and voltage stuff.
[01:27:27.360 --> 01:27:33.300]  So these things, both these devices came with three, or six, five handset modules.
[01:27:33.720 --> 01:27:38.060]  And after trying different telephones, first we had a mixture of telephones.
[01:27:38.060 --> 01:27:40.140]  And we tried every combination of phones we could think of.
[01:27:40.240 --> 01:27:51.180]  And with five different handset modules on two different types of phones, you end up with five times, four times, three times, two combinations, times three different kinds of phones.
[01:27:51.980 --> 01:27:58.060]  Add that up, and you realize it took us a good part of an evening to try to get these things to work without any success.
[01:27:58.480 --> 01:28:04.620]  So I thought, well, we can limit the combinations to this puzzle by getting two identical phones.
[01:28:04.960 --> 01:28:07.980]  So I went out to a local hardware store, and I picked up two...
[01:28:09.300 --> 01:28:14.200]  Unfortunately, Bell AT&T phones were not in stock at this Rickles hardware store I went to.
[01:28:14.300 --> 01:28:17.340]  But they had Northwestern Bell phones.
[01:28:17.460 --> 01:28:21.200]  It's like, you know, cheap $25 phones.
[01:28:21.420 --> 01:28:24.220]  I bought a set of them, and I figured, well, they're going to have to use...
[01:28:24.220 --> 01:28:28.040]  If they're going to work with any of these handset modules, they're going to at least use the same one, because it's two of the same phone.
[01:28:28.180 --> 01:28:31.480]  So that limits it to five combinations instead of a couple of hundred.
[01:28:32.340 --> 01:28:34.380]  So came back with those, we plugged them in.
[01:28:34.560 --> 01:28:38.500]  We went through all five modules, and of course, it was the last one that we finally got to work.
[01:28:40.500 --> 01:28:44.620]  And we were not too impressed with the performance of these, and I'll begin to explain why.
[01:28:45.520 --> 01:29:02.240]  Coincidentally, as John and I were testing these devices on the pool table of his rec room, the local telephone company, and I live in Philadelphia, Bell Atlantic, was there fixing a crosstalk problem in John's house.
[01:29:02.400 --> 01:29:05.960]  So there was a telephone lineman there from Bella, Pennsylvania.
[01:29:07.160 --> 01:29:11.500]  And he was looking at all our stuff down there in the basement, and it was like...
[01:29:11.500 --> 01:29:14.920]  He just had this strange look on his face, like, what are these guys doing?
[01:29:15.720 --> 01:29:26.840]  Because we had multiple computers down there, cellular telephones that we were cracking, all kinds of stuff we were working on, and these strange looking boxes from AT&T.
[01:29:28.220 --> 01:29:29.740]  And I said, what are you guys doing?
[01:29:29.880 --> 01:29:37.520]  And we told him, he said, we're testing and evaluating this new product from AT&T, which costs $1,300 a pop, and you have to have at least two of them to make them work.
[01:29:39.660 --> 01:29:41.160]  And he said, well, what do they do?
[01:29:41.260 --> 01:29:45.180]  And I said, well, they're supposed to make your conversation secure so they can't be overheard.
[01:29:45.520 --> 01:29:48.000]  And he said, well, I can overhear them with my butt set.
[01:29:48.120 --> 01:29:52.820]  And he showed me his little hand, you know, the orange big telephone.
[01:29:53.060 --> 01:30:00.420]  And I said, no, what this thing actually does is it converts your voice to data, much like a telephone modem, which he understood what it was.
[01:30:01.720 --> 01:30:03.880]  And all you're going to hear is, like, noise.
[01:30:04.100 --> 01:30:05.300]  And he said, oh, okay.
[01:30:05.500 --> 01:30:07.160]  So he said, well, show me how they work.
[01:30:07.600 --> 01:30:13.540]  So we plugged in handset module number five, which was finally the one we finally got it to work with.
[01:30:13.960 --> 01:30:19.980]  And the best we could get through these things was a very low level of audio.
[01:30:20.660 --> 01:30:23.340]  We could hear each other, but you pretty much had to shout.
[01:30:23.560 --> 01:30:34.460]  And the conclusion that Bella, Pennsylvania lineman came to was, even if these things were secure, all the FBI would have to do is stand outside your house and listen to your yelling through this thing.
[01:30:35.760 --> 01:30:37.420]  And they could still find out what you were saying.
[01:30:39.840 --> 01:30:40.780]  So what's in there?
[01:30:41.780 --> 01:30:44.580]  Oh, it's mostly passive components and what we think is an op amp.
[01:30:44.840 --> 01:30:45.340]  Which would, yeah.
[01:30:49.120 --> 01:30:50.280]  It's a Motorola chip.
[01:30:51.060 --> 01:30:52.460]  Read out the numbers, see if anybody knows it.
[01:30:52.460 --> 01:30:53.300]  You wrote them down?
[01:30:53.900 --> 01:30:55.500]  And two chips in there.
[01:30:55.760 --> 01:30:55.940]  Anyway.
[01:31:01.840 --> 01:31:04.680]  Well, MC34074D and then the Motorola logo.
[01:31:04.840 --> 01:31:07.780]  Next to the Motorola logo is XAF305.
[01:31:08.180 --> 01:31:08.880]  No, it can't be.
[01:31:08.980 --> 01:31:13.300]  It's, it's like a, it's a, it's a 16 pin dip.
[01:31:13.420 --> 01:31:14.840]  So it's not a, I doubt it's a regulator.
[01:31:15.600 --> 01:31:17.340]  And the regulator would be on the device anyway.
[01:31:17.500 --> 01:31:19.020]  It wouldn't be in the handset module, I wouldn't think.
[01:31:19.060 --> 01:31:22.080]  That sounds like one of their funky regulators, but it might not be.
[01:31:22.100 --> 01:31:22.420]  Who knows?
[01:31:22.540 --> 01:31:24.160]  There's an awful lot of capacitors on here.
[01:31:24.220 --> 01:31:24.980]  I don't know why that would be.
[01:31:25.120 --> 01:31:26.020]  But anyway.
[01:31:27.820 --> 01:31:28.180]  Could be.
[01:31:28.280 --> 01:31:28.920]  Yeah, that must be it.
[01:31:30.160 --> 01:31:31.080]  I don't know if that makes sense.
[01:31:32.260 --> 01:31:32.540]  Anyway.
[01:31:32.800 --> 01:31:38.680]  So the best we could get out of these things was a, a very weak, um, transfer of audio from one to the other.
[01:31:38.960 --> 01:31:40.300]  Pretty much had to yell through them.
[01:31:40.820 --> 01:31:43.220]  And one really annoying aspect of it.
[01:31:43.360 --> 01:31:53.100]  Well, there was one distracting aspect to them, which was, uh, since we could hear each other through the air, since we're only like 10 feet away from me, 10, 15 feet away from the other on the, in his basement.
[01:31:53.500 --> 01:32:01.200]  Um, there was an annoying echo because, uh, what, the way these things basically work is they convert your, your speech into a digital data stream.
[01:32:01.520 --> 01:32:04.140]  Um, and there's some encryption, there's encryption involved after that.
[01:32:04.140 --> 01:32:10.740]  But the processing, the, the delay in processing this information amounts to, we estimated about a quarter of a second.
[01:32:11.300 --> 01:32:20.980]  And so you had this really annoying delay, um, an echo effect between, um, what I would hear him saying through the air and what would arrive a quarter second later.
[01:32:21.120 --> 01:32:23.360]  It doesn't sound like a lot, but a quarter second, but...
[01:32:23.360 --> 01:32:23.920]  Do you know how these work?
[01:32:24.360 --> 01:32:24.900]  I'm sorry?
[01:32:24.980 --> 01:32:26.820]  Did they give you the architecture of how these are constructed?
[01:32:27.860 --> 01:32:28.300]  Who?
[01:32:28.700 --> 01:32:29.100]  Oh, anybody.
[01:32:30.340 --> 01:32:36.060]  Um, no, they didn't, they were very, uh, um, hesitant, to give me much in the way of technical information.
[01:32:36.220 --> 01:32:43.820]  The reason for, the reason for the delay is, is that it, it digitizes your voice, compresses it, and then sends it out over a 9600 baud modem.
[01:32:44.120 --> 01:32:44.600]  Is it 48?
[01:32:44.600 --> 01:32:45.580]  Actually, it's 48 hundred.
[01:32:45.680 --> 01:32:46.520]  It's 48 in this device.
[01:32:46.820 --> 01:32:51.280]  It's, it's, it's V.32 compliant, according to the specifications in the manual.
[01:32:51.420 --> 01:32:55.080]  So one, so the quarter second delay is all the digital signal processing due to the voice compression.
[01:32:55.200 --> 01:32:55.600]  Right.
[01:32:55.700 --> 01:32:58.600]  The question is, you said it was low, but how was the audio quality?
[01:32:59.260 --> 01:32:59.740]  So...
[01:32:59.740 --> 01:33:00.720]  That was even worse.
[01:33:01.000 --> 01:33:01.480]  Okay.
[01:33:01.760 --> 01:33:02.080]  Thank you.
[01:33:02.320 --> 01:33:03.360]  To answer your question.
[01:33:03.660 --> 01:33:24.420]  Uh, in regards to audio quality, um, because of, uh, all the audio compression and, uh, subsequent expansion that gets involved, they had to cut some corners to, to cut the actual volume of information down to make it manageable from the standpoint of transmitting all this information over the phone line at 48 hundred baud in full duplex,
[01:33:24.580 --> 01:33:25.380]  in full duplex mode.
[01:33:25.380 --> 01:33:25.940]  I think, I think...
[01:33:25.940 --> 01:33:26.840]  In full duplex mode.
[01:33:27.280 --> 01:33:33.740]  So, the corners they cut, they, the, according on, they did go into some, some technical details, which I was surprised.
[01:33:34.300 --> 01:33:35.520]  Um, let's see.
[01:33:35.580 --> 01:33:47.220]  It does say that it uses, uh, uh, the voice mode, in secure voice mode is, uh, 48 hundred baud, full duplex, relaxed code, excited linear prediction, or R-C-E-L-P.
[01:33:47.420 --> 01:33:47.780]  R-C-E-L-P.
[01:33:47.780 --> 01:33:47.960]  R-C-E-L-P.
[01:33:48.160 --> 01:33:48.420]  Yeah.
[01:33:49.080 --> 01:33:55.820]  I, I'll, I'll offline, I'll tell you all, anything you want to know about, uh, voice compression, which is probably not all your questions, but everything I know.
[01:33:56.780 --> 01:33:58.380]  You want to talk about that every minute?
[01:33:58.380 --> 01:33:58.440]  Oh, what?
[01:33:58.960 --> 01:33:59.320]  No.
[01:33:59.620 --> 01:33:59.800]  Okay.
[01:34:00.180 --> 01:34:01.840]  It might just bore everybody, because it could...
[01:34:01.840 --> 01:34:02.520]  Bottom line.
[01:34:02.580 --> 01:34:06.120]  It's really, it's very clever, and it's really complex, and it's not really relevant right now.
[01:34:06.460 --> 01:34:06.800]  Okay.
[01:34:07.000 --> 01:34:07.260]  Um...
[01:34:07.260 --> 01:34:08.400]  It costs a lot in CPU.
[01:34:08.540 --> 01:34:09.920]  That's the really only...
[01:34:09.920 --> 01:34:19.340]  And it also, uh, the modem, according to this thing, features, uh, near-slash-far echo cancellation, and frequency offset compensation.
[01:34:19.340 --> 01:34:22.280]  If they didn't have the echo consolation, it would just be unbearable, I would imagine.
[01:34:22.800 --> 01:34:27.540]  Um, and the frequent, frequency offset compensation, um, maybe you could explain what that would be.
[01:34:27.820 --> 01:34:29.560]  Oh, that, that's modem engineering.
[01:34:29.840 --> 01:34:32.720]  They're not doing it, that's not on the audio channel, that's on the modem channel.
[01:34:32.820 --> 01:34:34.680]  It just means that they have a fancy modem.
[01:34:36.160 --> 01:34:36.780]  Good enough.
[01:34:37.280 --> 01:34:46.940]  So what's interesting is when, uh, when we initiated a, uh, communication between these two things, we could actually hear a bit of crosstalk in the handset of the phones of the data being sent.
[01:34:47.500 --> 01:34:52.520]  And so you could hear what sounded like normal, uh, modem communications being established.
[01:34:52.640 --> 01:34:53.420]  You could hear the carrier.
[01:34:54.020 --> 01:34:58.840]  Um, actually what it does first is it sends, uh, it sends four touch tones.
[01:34:59.220 --> 01:35:13.340]  When, when you take the phone off, the first phone off the hook, and, well actually when, when you dial another phone, and, uh, you wish to establish secure communications with that person at the other end, you tell the other person, uh, we need to establish secure communications.
[01:35:13.900 --> 01:35:19.220]  Um, either you or I need to push the red button on the device, which puts you into secure mode.
[01:35:19.640 --> 01:35:23.240]  When one or the other party pushes that, both parties don't need to push it, only one.
[01:35:23.620 --> 01:35:26.120]  In fact, only one party can push that button or it won't work.
[01:35:28.120 --> 01:35:36.400]  The unit whose button was pushed sends a string of DTMF digits, or touch tones, to the receiving unit.
[01:35:36.580 --> 01:35:47.100]  And that struck me as kind of curious, so I got out my handy touch tone decoding device and decoded those tones, and there were 2587.
[01:35:47.580 --> 01:35:53.880]  And it didn't seem to matter which handset module you used or how many times you used it, it was always the same digit string.
[01:35:53.880 --> 01:35:56.160]  I don't know why that was, but go figure.
[01:35:56.720 --> 01:35:58.840]  I tried to do a little bit of hacking as much as I could.
[01:36:00.540 --> 01:36:10.100]  So after this four digit string is sent to the other unit, that evidently is sort of a wake-up signal to the receiving modem that it's supposed to start to...
[01:36:10.100 --> 01:36:10.680]  Yeah?
[01:36:14.380 --> 01:36:24.260]  I tried that, and it did indeed go into secure mode, but it wouldn't work because the sending unit had not itself gone into secure mode.
[01:36:24.260 --> 01:36:31.860]  So it did initialize the remote modem, and of course it didn't hear a handshake carrier, so no communication was established.
[01:36:32.160 --> 01:36:34.220]  Sounds like a new ToneLoc module to me.
[01:36:34.620 --> 01:36:35.140]  Sorry?
[01:36:35.340 --> 01:36:37.740]  Sounds like a new addition that ToneLoc needs.
[01:36:37.940 --> 01:36:38.280]  Oh, yeah.
[01:36:40.980 --> 01:36:41.500]  Yeah?
[01:36:44.000 --> 01:36:52.820]  Can I try maybe doing that with your own modem to one of the secure phone Hoosie-Watsi interfaces?
[01:36:53.160 --> 01:36:54.700]  Yes, the Hoosie-Watsi interface.
[01:36:54.960 --> 01:36:58.320]  We did check that out, and it just was garbage.
[01:36:58.680 --> 01:37:00.040]  Wait, that's the best we could get.
[01:37:00.160 --> 01:37:02.320]  It seemed pretty random garbage.
[01:37:02.320 --> 01:37:03.920]  2800 is the obvious question.
[01:37:04.020 --> 01:37:12.540]  Yeah, we were able to get our Hays modem to recognize that the V.32, that it was nothing that we could recognize as intelligible.
[01:37:12.640 --> 01:37:19.740]  Not that even if it were clear that we would know what normal voice waveforms look like in digital form by looking at them, but anyway.
[01:37:20.320 --> 01:37:20.800]  Yes?
[01:37:21.480 --> 01:37:26.580]  Yeah, you mentioned that it's using some sort of handshake to keep two modes communicating with each other.
[01:37:27.080 --> 01:37:36.260]  Did you try to experiment, like, introducing, like, a piece of blind noise into it to see what happens if the data communications are screwed up and it's become lost?
[01:37:36.340 --> 01:37:36.980]  Yes, we did.
[01:37:37.060 --> 01:37:43.900]  As a matter of fact, we tried blowing into the microphones, and that, it handled that pretty well.
[01:37:44.080 --> 01:37:49.600]  But what surprised me was, I thought, well, how would it handle a pure signal?
[01:37:49.600 --> 01:37:55.040]  Um, like a sine wave or a combination of sine waves like a touch tone.
[01:37:55.420 --> 01:38:02.820]  And it turns out any combination of sine waves or single pure waveforms makes this thing crash completely.
[01:38:03.100 --> 01:38:07.980]  It just, it just hangs up at the other end and, and, and it goes out of secure mode.
[01:38:08.160 --> 01:38:14.500]  And, and what's really odd is when it, when it goes out of secure mode, when you push a touch tone button on, on one or the other phone, it doesn't matter which.
[01:38:14.500 --> 01:38:28.620]  When the other unit goes out of secure mode, you hear in the earpiece a truly demonic sounding noise, which the way my friend John put it sounded like something straight out of the Exorcist.
[01:38:28.680 --> 01:38:30.020]  It was really evil.
[01:38:30.200 --> 01:38:32.140]  So we decided to call this noise the Clipper Demon.
[01:38:32.920 --> 01:38:34.700]  It was truly evil sounding.
[01:38:34.700 --> 01:38:50.280]  It was what happens when, I guess, secure voice goes back into normal voice, but it was about a half to three-quarters of a second of garbly voice, and it sounded really evil.
[01:38:50.860 --> 01:38:53.520]  That's to make sure you know it went unsecure.
[01:38:53.700 --> 01:38:54.120]  I guess.
[01:38:56.260 --> 01:38:58.200]  How are the phones doing key distributions?
[01:38:58.680 --> 01:38:59.740]  Who's the Oscar agent?
[01:39:00.020 --> 01:39:00.520]  Sort of.
[01:39:00.640 --> 01:39:01.340]  A little worse, though.
[01:39:03.220 --> 01:39:04.320]  I'm sorry, I didn't hear you?
[01:39:12.860 --> 01:39:18.060]  It's my understanding that Treasury and the National Institute of Standards and Technology are the key escrow.
[01:39:19.040 --> 01:39:19.900]  No, not at all.
[01:39:19.960 --> 01:39:20.700]  They don't talk about that.
[01:39:20.820 --> 01:39:25.520]  In fact, they don't even mention anywhere in the literature that this uses anything even called a Clipper chip.
[01:39:26.040 --> 01:39:27.800]  I mean, there's a C after the...
[01:39:27.800 --> 01:39:28.640]  You know, they even...
[01:39:28.640 --> 01:39:29.340]  It's interesting.
[01:39:29.460 --> 01:39:31.840]  In the manual, they even omit the C.
[01:39:32.040 --> 01:39:33.160]  It must be classified.
[01:39:34.440 --> 01:39:37.060]  They can't tell anybody that there's actual Clipper chip in this thing.
[01:39:37.200 --> 01:39:43.920]  But the marketing folks refer to it as 3600C, the C suffix, meaning it's Clipper chip.
[01:39:44.020 --> 01:39:46.560]  They evidently have another model, 3600, which doesn't have the Clipper.
[01:39:46.600 --> 01:39:46.960]  Is that right?
[01:39:47.400 --> 01:39:48.280]  Is there some other...
[01:39:50.060 --> 01:39:52.460]  There are actually... I think there are actually two other models.
[01:39:52.600 --> 01:39:54.700]  One is...
[01:39:55.560 --> 01:39:58.980]  There's a model that... the original model used DES as the cipher.
[01:39:59.080 --> 01:40:02.500]  And there's another model for export, which uses really bad cryptography.
[01:40:03.900 --> 01:40:07.100]  And there's another one, evidently, which is a acoustic coupler version.
[01:40:07.300 --> 01:40:07.600]  Yeah.
[01:40:07.920 --> 01:40:09.260]  One I forgot to mention.
[01:40:09.460 --> 01:40:10.380]  There's a question over on the wall here, which is...
[01:40:10.380 --> 01:40:10.840]  Oh, I'm sorry.
[01:40:11.300 --> 01:40:11.620]  Yeah.
[01:40:12.140 --> 01:40:14.120]  This is just a theory I want to advance to you.
[01:40:14.240 --> 01:40:14.400]  Mm-hmm.
[01:40:14.420 --> 01:40:15.200]  Five, eight, seven.
[01:40:15.580 --> 01:40:19.980]  If you check, two, the five, and the eight are all directly underneath each other.
[01:40:20.080 --> 01:40:20.120]  Mm-hmm.
[01:40:20.120 --> 01:40:21.060]  And the seven's off to the left.
[01:40:21.880 --> 01:40:22.980]  And seven's off to the left.
[01:40:23.000 --> 01:40:23.160]  Mm-hmm.
[01:40:23.160 --> 01:40:38.020]  By sending those three tones, you establish a phase tone, and you pretty much build at least four rows of your matrix, of your frequency matrix, to check clock cycles, to try and get them... to try and get the clock cycles to sync up.
[01:40:38.020 --> 01:40:38.360]  Training?
[01:40:38.740 --> 01:40:39.140]  Training.
[01:40:39.400 --> 01:40:43.240]  You mean, like, similar to fax training for equalization of the line?
[01:40:43.360 --> 01:40:43.540]  Yeah.
[01:40:44.040 --> 01:40:44.640]  That's a good idea.
[01:40:44.720 --> 01:40:45.420]  That's an interesting idea.
[01:40:45.560 --> 01:40:46.580]  You're taking, like...
[01:40:46.580 --> 01:40:47.120]  That's very good.
[01:40:48.300 --> 01:40:49.020]  Very sharp.
[01:40:49.200 --> 01:40:49.720]  Yeah, that could be...
[01:40:49.720 --> 01:40:51.680]  Yeah, but it didn't even occur to me, but that makes sense.
[01:40:51.980 --> 01:40:52.740]  That makes sense.
[01:40:54.740 --> 01:41:02.080]  The theory is that the two, five, eight, and seven are all on the right rows of the keypad matrix, so that you get a...
[01:41:02.080 --> 01:41:02.980]  Left one.
[01:41:04.020 --> 01:41:05.020]  Whatever, it doesn't matter.
[01:41:05.180 --> 01:41:05.640]  The correct...
[01:41:05.640 --> 01:41:06.120]  Same side.
[01:41:06.980 --> 01:41:13.860]  In other words, that you can train the... you can equalize the line to figure out what the frequency response of the actual phone connection is like.
[01:41:14.220 --> 01:41:15.540]  It's similar to what faxes do.
[01:41:16.420 --> 01:41:29.780]  The original... those high tones in the fax thing are actually... they're doing analog analysis of that signal to determine where the holes in the... in the frequency response of the line you're using on, and they'll pick some response characteristics in the fax to make it sure it works when it actually connects.
[01:41:31.860 --> 01:41:33.820]  So we weren't too pleased with these devices.
[01:41:34.000 --> 01:41:35.900]  We even let the Bell telephone lineman try it.
[01:41:35.940 --> 01:41:37.700]  And he said, these things suck.
[01:41:40.540 --> 01:41:44.220]  So we thought, well, maybe it had something to do with our central office simulator.
[01:41:44.220 --> 01:41:50.460]  Maybe it was not providing proper line voltages, although we had no problem ever testing telephones or modems with it before.
[01:41:51.200 --> 01:41:57.380]  And we had two different models of the central office simulator, and both of them gave the same lousy results.
[01:41:58.520 --> 01:42:00.940]  So I decided I would take the things home.
[01:42:00.940 --> 01:42:02.960]  I have three phone lines at my home.
[01:42:03.620 --> 01:42:06.820]  One fax line, one business line, and one personal line.
[01:42:06.960 --> 01:42:12.480]  So I figured, well, I can call myself and try to do it through looping through my central office.
[01:42:12.900 --> 01:42:26.360]  So I did that the next day, and it took me close to four hours to get it to work at all, just between the two phones that were the same phone, just looping through my CO.
[01:42:26.540 --> 01:42:39.060]  So I called the number that's in the manual, and it says, if you have any trouble, with your device, let me find that little cheat sheet that came with it, a little...
[01:42:44.320 --> 01:42:45.100]  Here it is.
[01:42:45.300 --> 01:42:55.660]  It says, for assistance, call the AT&T Secure Communications Products Customer Service Center at, you might want to write this down, 1-800-952-4082.
[01:42:56.480 --> 01:43:01.060]  This office is evidently located in a very rural backwards area of North Carolina.
[01:43:01.060 --> 01:43:06.900]  And I don't know why they do this down there, but maybe they pay the people less there or something.
[01:43:07.600 --> 01:43:09.860]  Oh, I should mention one thing, since you're mentioning North Carolina.
[01:43:10.000 --> 01:43:22.900]  The guy who is the head cryptographer of this division is a guy named Dave Mayer, who's one of the five people on the Clipper Review panel that Dorothy Denning chaired last year, just for your conspiracy theory mind to churn on.
[01:43:22.900 --> 01:43:24.640]  It all fits together.
[01:43:25.960 --> 01:43:31.160]  So I called there, and I went through, first, sort of a screening person.
[01:43:31.320 --> 01:43:32.140]  Oh, question over there.
[01:43:35.360 --> 01:43:37.320]  Is he a teacher?
[01:43:37.640 --> 01:43:38.480]  Dave Mayer?
[01:43:39.320 --> 01:43:40.420]  I don't think he...
[01:43:40.420 --> 01:43:43.840]  I think he's just an AT&T employee.
[01:43:44.820 --> 01:43:45.380]  Greensboro?
[01:43:45.840 --> 01:43:46.160]  Yeah.
[01:43:46.540 --> 01:43:47.220]  He's a mathematician.
[01:43:47.580 --> 01:43:50.520]  Yeah, I mean, he's trained as a mathematician, and he's very, very sharp.
[01:43:52.000 --> 01:43:54.880]  He does, however, you know, build a Clipper phone.
[01:43:56.760 --> 01:43:57.080]  Hmm?
[01:43:57.280 --> 01:43:58.260]  Do you know what he looks like?
[01:43:59.040 --> 01:43:59.360]  Um...
[01:43:59.360 --> 01:44:00.160]  I don't know if we do.
[01:44:00.840 --> 01:44:01.480]  I've met him.
[01:44:01.540 --> 01:44:02.620]  I've met him once, yeah.
[01:44:03.020 --> 01:44:04.040]  I know what he looks like.
[01:44:04.060 --> 01:44:04.760]  Matt knows what he looks like.
[01:44:05.020 --> 01:44:05.760]  What's the question?
[01:44:05.920 --> 01:44:06.980]  Do you have, like, short hair?
[01:44:09.280 --> 01:44:09.600]  No.
[01:44:09.600 --> 01:44:11.740]  We'll get a mug shot in 2600.
[01:44:11.840 --> 01:44:14.620]  We'll have a police artist interview you after the session.
[01:44:19.380 --> 01:44:33.480]  So, anyway, I tried to get these things working in my own home, and called the AT&T Secure Communications Products Customer Service Center, and went through a woman named Teresa, who was absolutely no help at all.
[01:44:33.680 --> 01:44:45.440]  She tried, on three separate occasions, to establish a secure link between herself and myself, and we went through all five handset modules, and nothing seemed to work at all.
[01:44:45.720 --> 01:44:47.580]  It would not go into secure mode.
[01:44:47.720 --> 01:44:54.920]  What it seems to do is it tries, wants to go into secure mode for about 10 seconds, which is normally how long it's supposed to take, according to the manual.
[01:44:55.480 --> 01:45:03.160]  And then it would say, try a second time, and take another 10 seconds, and then it would just finally say, push clear, meaning you have to reset it, because it just didn't take.
[01:45:04.160 --> 01:45:05.560]  She was downright puzzled.
[01:45:05.640 --> 01:45:12.100]  She said, I'm going to have to talk with an engineer, and I said, well, that's good, because I could probably, hopefully speak his language, because I have a degree in engineering.
[01:45:12.320 --> 01:45:22.160]  So, I left a voicemail for this one engineer, and it took him a couple of days to get back to me.
[01:45:22.480 --> 01:45:40.300]  Finally he did, and he was friendly, and rather helpful, but he came to the conclusion that his assumption was, there was too much line generated, too much line noise generated between my, my drop lines in my house and the central office.
[01:45:40.520 --> 01:45:49.460]  Now, I've never had any problem with noise and anything else, including V.fast modems, which are four times faster than these things, and never had a problem.
[01:45:50.000 --> 01:45:51.980]  So, you seem to have the same problems, so...
[01:45:51.980 --> 01:45:52.720]  I'm sorry?
[01:45:52.820 --> 01:45:54.500]  So, you're just as much sure what the life simulator...
[01:45:54.500 --> 01:45:58.140]  Yeah, and then we had the same repeatable problems with a central office simulator.
[01:45:58.300 --> 01:46:00.080]  So, you didn't have them send you two more?
[01:46:00.740 --> 01:46:05.960]  I thought about that, but it was, the time of the HOPE conference was drawing near, and there wasn't time for them to get them to me.
[01:46:06.120 --> 01:46:08.880]  So, I thought about actually doing that, but there wasn't time.
[01:46:09.060 --> 01:46:09.880]  There's still time.
[01:46:10.080 --> 01:46:10.540]  Ah, yes.
[01:46:12.020 --> 01:46:13.180]  A couple questions up here?
[01:46:14.100 --> 01:46:15.840]  I have two questions, really.
[01:46:16.060 --> 01:46:19.680]  First of all, are the encoding boxes in matched pairs?
[01:46:20.000 --> 01:46:20.100]  Or do
[01:46:26.360 --> 01:46:28.000]  you want three people to lose?
[01:46:28.140 --> 01:46:29.400]  You haven't had six long pairs?
[01:46:29.940 --> 01:46:32.840]  The answer to the second question is no, not to my knowledge.
[01:46:32.840 --> 01:46:34.700]  Is there any method of providing a...
[01:46:34.700 --> 01:46:41.220]  In fact, I know almost for a fact that there's no way you could do that, because it uses something called a spoofing code, and I'll go into that in a minute.
[01:46:41.440 --> 01:46:44.780]  Yeah, these units don't have particular names.
[01:46:44.780 --> 01:46:50.280]  They use this algorithm to generate an encryption key on the fly at the beginning of the phone call.
[01:46:51.140 --> 01:46:51.820]  So they...
[01:46:51.820 --> 01:46:53.680]  So there's no keying material in the box.
[01:46:53.800 --> 01:46:56.840]  It's not like a Stu-3 that has little ignition keys on the side of it.
[01:46:57.180 --> 01:46:59.280]  Someone's spoofing from the very beginning...
[01:46:59.280 --> 01:47:01.720]  Well, they're not in matched pairs.
[01:47:02.040 --> 01:47:03.740]  I'll explain to you the algorithm later.
[01:47:03.740 --> 01:47:04.800]  It's quite clever.
[01:47:05.520 --> 01:47:06.700]  It was patented.
[01:47:07.520 --> 01:47:08.540]  Another question over here?
[01:47:08.920 --> 01:47:09.220]  Yeah.
[01:47:09.500 --> 01:47:10.480]  As far as
[01:47:17.800 --> 01:47:18.400]  YouTube...
[01:47:18.400 --> 01:47:18.760]  Right.
[01:47:19.620 --> 01:47:20.040]  Mm-hmm.
[01:47:26.850 --> 01:47:33.530]  Well, his concern was it was affecting the error rate, I guess, and it was just not enough for it to establish a link.
[01:47:33.670 --> 01:47:35.190]  I'm sorry?
[01:47:35.890 --> 01:47:37.330]  Wouldn't it have to be about rates?
[01:47:37.610 --> 01:47:38.010]  Yes.
[01:47:38.330 --> 01:47:38.730]  It would.
[01:47:39.150 --> 01:47:40.730]  And that was my response.
[01:47:48.070 --> 01:47:49.190]  I'm sorry?
[01:47:54.650 --> 01:47:55.210]  Right.
[01:47:55.650 --> 01:48:00.250]  There is distortion in the signal or something accurate, but...
[01:48:03.470 --> 01:48:04.590]  The receiving...
[01:48:04.590 --> 01:48:09.950]  It's a good theory, but the receiving unit did indeed receive those tones and go into secure mode or go into...
[01:48:10.350 --> 01:48:11.850]  the attempt to go into secure mode.
[01:48:11.950 --> 01:48:14.430]  So it did receive the command at the remote end to do that.
[01:48:14.790 --> 01:48:16.050]  And we've got a question up here now.
[01:48:16.210 --> 01:48:18.890]  Did you check to see what happens when you get a call waiting signal?
[01:48:19.510 --> 01:48:19.870]  No.
[01:48:19.990 --> 01:48:22.190]  I refused to pay extra to have my calls interrupted.
[01:48:23.230 --> 01:48:24.150]  Here, here, here.
[01:48:25.690 --> 01:48:26.050]  Let's...
[01:48:26.050 --> 01:48:30.870]  I would suggest that we stop trying to diagnose the specific problem because it sounds like they have a bad unit.
[01:48:31.410 --> 01:48:38.270]  No, actually, the guy finally admitted to me that they were having a lot of problems with these things.
[01:48:38.670 --> 01:48:42.910]  And that my problems were not unusual.
[01:48:43.270 --> 01:48:44.850]  I did open them up, as a matter of fact.
[01:48:45.010 --> 01:48:45.450]  I wanted to...
[01:48:45.450 --> 01:48:46.650]  In fact, I'll open one up.
[01:48:46.770 --> 01:48:48.690]  It's kind of you can people come up and see what one looks like inside.
[01:48:49.750 --> 01:48:50.110]  But...
[01:48:50.110 --> 01:48:52.170]  We should take general questions for a bit.
[01:48:52.410 --> 01:48:52.710]  Yeah.
[01:48:53.450 --> 01:48:56.650]  The bottom line is I determined that...
[01:48:56.650 --> 01:48:58.430]  Well, I finally did get a secure link established.
[01:48:58.430 --> 01:49:03.630]  It turns out there was an undocumented feature in the manual, which I discovered completely by accident.
[01:49:03.830 --> 01:49:18.850]  And that is if you hit the clear button, just prior to pushing the secure button, it puts it into what's called plus mode, which enhances or boosts the audio on one end and one end only.
[01:49:19.990 --> 01:49:26.090]  And that provided enough of a boost, evidently, to get a legitimate data...
[01:49:26.090 --> 01:49:28.670]  enough of a data stream together to establish handshaking.
[01:49:28.890 --> 01:49:29.770]  And that worked.
[01:49:29.850 --> 01:49:31.150]  But the audio was still pretty weak.
[01:49:31.330 --> 01:49:34.050]  And we just tested it here, and it's still pretty darn weak.
[01:49:34.250 --> 01:49:40.230]  So, I feel pretty confident in saying that after running a bunch of tests on these things, they're not that great.
[01:49:40.370 --> 01:49:45.410]  And they're certainly not worth $1,300 a piece, especially since I don't even trust the security aspect of them.
[01:49:46.310 --> 01:49:47.110]  Do you have a question over here?
[01:49:47.110 --> 01:49:47.370]  Question over here?
[01:49:47.530 --> 01:49:48.690]  Yeah, but what if...
[01:49:53.220 --> 01:49:54.780]  Would that...
[01:49:57.960 --> 01:49:58.320]  If...
[01:49:58.320 --> 01:50:00.020]  It would work if...
[01:50:02.990 --> 01:50:05.230]  No, I tried that, and it didn't work at all.
[01:50:05.390 --> 01:50:13.570]  In fact, they told me that the AT&T Secure Communications Products Customer Service Center, that that definitely could not work.
[01:50:13.670 --> 01:50:14.090]  It just...
[01:50:14.090 --> 01:50:16.310]  It would just cause some sort of conflict.
[01:50:16.530 --> 01:50:18.430]  But I tested it anyway, just to be sure.
[01:50:19.610 --> 01:50:22.050]  So, I did everything I could think of to get these things to work.
[01:50:22.050 --> 01:50:23.010]  And they just...
[01:50:23.010 --> 01:50:25.070]  It's piss poor performance, frankly.
[01:50:25.590 --> 01:50:27.990]  Not to mention all the other concerns we have about them.
[01:50:28.470 --> 01:50:37.450]  So, I'm really glad that they sent these things net 30 because I have no intention of paying $2,600 or even $26 for these two things.
[01:50:37.570 --> 01:50:38.770]  But they were fun to play with.
[01:50:38.930 --> 01:50:39.550]  I have a...
[01:50:39.550 --> 01:50:48.550]  Okay, as a bit of news, I have a friend in California who I knew from Cypherpunks who quit his job at HP two months ago in order to create a company he calls ComSec Partners.
[01:50:48.550 --> 01:50:50.530]  They're ComSec.com, if you want to finger them.
[01:50:50.690 --> 01:50:54.390]  And they're coming out with a non-Clipper version of this thing.
[01:50:54.530 --> 01:50:56.110]  And they should be out by the end of the year.
[01:50:56.190 --> 01:50:57.770]  And they're going to cost less than $1,000.
[01:50:58.050 --> 01:51:01.010]  And I'm sure we can get Bernie a pair of them without having to social engineer.
[01:51:01.310 --> 01:51:01.650]  Yay!
[01:51:01.930 --> 01:51:02.990]  It's not easy for encryption.
[01:51:03.370 --> 01:51:03.690]  Yeah, right.
[01:51:04.750 --> 01:51:06.290]  They're using good cryptography.
[01:51:06.350 --> 01:51:08.470]  I don't want to go into the details because they're quite numerous.
[01:51:08.470 --> 01:51:10.610]  But I've looked at the stuff and they're...
[01:51:10.610 --> 01:51:12.330]  Something at least as good as PGP Voice?
[01:51:13.470 --> 01:51:13.890]  Yes.
[01:51:14.410 --> 01:51:15.070]  Sounds good.
[01:51:16.070 --> 01:51:20.710]  In fact, the algorithm of this unit will probably be the standard upon which PGP Voice is based.
[01:51:20.870 --> 01:51:24.050]  PGP Voice is kind of this project that's getting lots of talk but no code.
[01:51:26.810 --> 01:51:35.010]  Before I wrap this up, I want to actually make a telephone call between these two things and allow you to hear the data stream and hear our reactions to the audio.
[01:51:35.150 --> 01:51:39.510]  Unfortunately, I don't have any way of coupling the handset audio to your PA system here.
[01:51:41.130 --> 01:51:41.890]  It's too weak.
[01:51:42.010 --> 01:51:43.390]  I guarantee it's going to be too weak.
[01:51:43.490 --> 01:51:47.670]  I'll try it but it's barely audible so it's that bad.
[01:51:47.830 --> 01:51:52.590]  But you'll at least be able to hear us talking and hear the encryption through the loudspeaker here.
[01:51:52.690 --> 01:51:53.750]  I have it hooked up through an amplifier.
[01:51:54.850 --> 01:51:56.570]  So I'm going to pick up this phone.
[01:51:56.950 --> 01:51:58.490]  I've got to turn on the central office simulator.
[01:52:00.330 --> 01:52:02.790]  And I've got to put this in...
[01:52:03.570 --> 01:52:04.490]  You want this one too?
[01:52:05.770 --> 01:52:08.230]  Well, I can just hold the handset up to this mic here.
[01:52:08.290 --> 01:52:09.350]  I was thinking about...
[01:52:09.350 --> 01:52:09.930]  Oh, yeah.
[01:52:10.470 --> 01:52:11.610]  It's going to be pretty loud actually.
[01:52:11.730 --> 01:52:12.890]  But yeah, put it up there.
[01:52:13.090 --> 01:52:13.530]  It can't hurt.
[01:52:17.190 --> 01:52:18.590]  Oh, yeah, you'll definitely hear that.
[01:52:27.160 --> 01:52:28.240]  Okay, here it goes.
[01:52:32.820 --> 01:52:33.820]  This is clear audio.
[01:52:35.360 --> 01:52:35.880]  Hello.
[01:52:36.380 --> 01:52:36.840]  Hello.
[01:52:37.140 --> 01:52:38.180]  This is clear audio.
[01:52:38.960 --> 01:52:45.740]  Now, I have a Radio Shack recording adapter connected between these two lines.
[01:52:45.860 --> 01:52:47.660]  And that's the audio I'm feeding into the amplifier.
[01:52:47.660 --> 01:52:49.460]  So you're hearing clear audio now.
[01:52:49.540 --> 01:52:54.960]  I'm going to put this thing into first plus mode to boost the audio before it goes into encrypted mode.
[01:52:54.960 --> 01:52:57.580]  And I'm going to put it in encrypted mode now.
[01:53:08.880 --> 01:53:10.180]  Now, that's the handshaking.
[01:53:18.580 --> 01:53:20.100]  Oh, it's trying a second time.
[01:53:31.180 --> 01:53:33.720]  The manual says it's supposed to take about 10 seconds.
[01:53:33.920 --> 01:53:35.740]  But it's already on the second trial.
[01:53:44.520 --> 01:53:45.540]  Oh, it's trying to turn it down.
[01:53:45.540 --> 01:53:45.580]  Oh, it's trying to turn it down.
[01:53:45.960 --> 01:53:47.300]  Okay, now it says it gave up.
[01:53:47.400 --> 01:53:48.100]  It says push clear.
[01:53:48.280 --> 01:53:54.180]  So we're going to try this again from your end to my end.
[01:53:54.180 --> 01:53:55.100]  Yeah.
[01:53:57.300 --> 01:53:58.300]  Hot product.
[01:54:04.460 --> 01:54:05.140]  Hello.
[01:54:08.280 --> 01:54:08.960]  Ready?
[01:54:09.300 --> 01:54:10.820]  Oh, we had to...
[01:54:10.820 --> 01:54:11.780]  He didn't clear the unit.
[01:54:12.180 --> 01:54:12.860]  You ready?
[01:54:13.200 --> 01:54:13.500]  Yeah.
[01:54:14.540 --> 01:54:14.940]  Okay.
[01:54:15.040 --> 01:54:15.760]  Now we have clear audio.
[01:54:16.040 --> 01:54:17.300]  And push...
[01:54:17.300 --> 01:54:19.080]  Just push the clear button.
[01:54:19.160 --> 01:54:19.700]  Not the clear button.
[01:54:19.780 --> 01:54:20.980]  Just the red button on yours.
[01:54:21.720 --> 01:54:22.120]  Okay.
[01:54:22.200 --> 01:54:23.440]  That's the...
[01:54:23.440 --> 01:54:25.420]  It may also be to punch through PBX.
[01:54:34.910 --> 01:54:35.310]  Ah.
[01:54:35.470 --> 01:54:36.090]  That's really interesting.
[01:54:36.210 --> 01:54:37.490]  We have a secure link established.
[01:54:37.810 --> 01:54:38.570]  Now, what happens...
[01:54:38.570 --> 01:54:39.590]  What happens when this...
[01:54:39.590 --> 01:54:41.850]  We have a secure link established now.
[01:54:42.010 --> 01:54:42.870]  What happens is...
[01:54:43.870 --> 01:54:44.010]  Yeah.
[01:54:46.050 --> 01:54:47.490]  You hear someone else talking through it?
[01:54:47.490 --> 01:54:49.390]  Does it sound like a demonic voice?
[01:54:50.230 --> 01:54:51.530]  Who does it sound like?
[01:54:51.990 --> 01:54:52.850]  Someone kind of...
[01:54:52.850 --> 01:54:53.450]  Kevin Mitnick?
[01:54:57.390 --> 01:54:59.550]  It sounds kind of like every operator you've ever heard.
[01:54:59.930 --> 01:55:01.410]  Like every operator you've ever heard.
[01:55:01.590 --> 01:55:04.010]  Oh, you mean it just sounds like a disembodied voice?
[01:55:04.250 --> 01:55:04.690]  Like...
[01:55:04.690 --> 01:55:06.050]  Kind of like me, but it's not really me?
[01:55:06.290 --> 01:55:06.450]  Yeah.
[01:55:06.530 --> 01:55:06.730]  Yeah.
[01:55:07.090 --> 01:55:07.710]  Well, this...
[01:55:07.710 --> 01:55:08.970]  The prediction...
[01:55:09.490 --> 01:55:11.650]  The method of compression this thing uses...
[01:55:11.650 --> 01:55:13.430]  I think is a CVSD.
[01:55:13.670 --> 01:55:15.830]  Continuously variable slope detection.
[01:55:15.830 --> 01:55:20.190]  And what that basically does is it cuts corners in...
[01:55:20.810 --> 01:55:23.830]  Literally cuts corners with the waveforms of your voice.
[01:55:24.150 --> 01:55:28.390]  And ends up causing you to lose all the inflection in your voice.
[01:55:28.490 --> 01:55:34.910]  Now, if you heard me talking in just monotone, it wouldn't have the same effect on...
[01:55:36.010 --> 01:55:37.990]  In the content of my speech.
[01:55:37.990 --> 01:55:46.930]  And CVSD has this really weird effect in making it sound almost not human.
[01:55:48.190 --> 01:55:50.510]  It just takes all the inflection out of your voice.
[01:55:50.770 --> 01:55:51.390]  And I don't...
[01:55:51.390 --> 01:55:52.990]  You folks are probably...
[01:55:54.210 --> 01:55:54.430]  Yeah.
[01:55:54.710 --> 01:55:55.570]  It's just...
[01:55:55.570 --> 01:55:57.370]  It makes you feel uncomfortable.
[01:55:57.470 --> 01:55:58.690]  And I'm gonna hold the microphone up.
[01:55:58.690 --> 01:55:59.770]  I'm gonna turn the...
[01:56:01.010 --> 01:56:02.010]  This amp down.
[01:56:02.390 --> 01:56:06.290]  And we're gonna have you shout into the microphone.
[01:56:06.530 --> 01:56:08.390]  And we're gonna see if this picks up this end.
[01:56:08.530 --> 01:56:10.850]  See if you can turn the gain up as much as possible on this mic.
[01:56:11.010 --> 01:56:12.230]  And turn his all the way off.
[01:56:12.610 --> 01:56:13.910]  And I'll hold this up here.
[01:56:14.470 --> 01:56:16.010]  And shout off like in that direction.
[01:56:19.150 --> 01:56:21.010]  That's one, two, three...
[01:56:21.010 --> 01:56:22.070]  See if you can whistle into it.
[01:56:25.970 --> 01:56:27.250]  Now I can hear it in my ear.
[01:56:27.410 --> 01:56:28.410]  But it's just not...
[01:56:28.950 --> 01:56:30.630]  It's not enough for the mic to pick up even.
[01:56:32.390 --> 01:56:32.790]  Now...
[01:56:33.670 --> 01:56:34.510]  Could you hear something?
[01:56:34.890 --> 01:56:36.810]  Did you hear the delay effect?
[01:56:37.930 --> 01:56:39.010]  If you heard anything...
[01:56:40.170 --> 01:56:43.190]  Well, you would have noticed the delay effect if you were picking it up.
[01:56:43.190 --> 01:56:44.290]  Do some words.
[01:56:44.490 --> 01:56:44.810]  Do some words.
[01:56:45.470 --> 01:56:45.770]  Do some words.
[01:56:46.070 --> 01:56:46.910]  Hand me the hand.
[01:56:47.050 --> 01:56:47.730]  Yeah, that's better.
[01:56:49.390 --> 01:56:49.830]  Hello.
[01:56:50.290 --> 01:56:50.730]  Hello?
[01:56:51.010 --> 01:56:51.850]  Hello, NSA?
[01:56:52.310 --> 01:56:52.990]  Yes, yes.
[01:56:53.130 --> 01:56:53.510]  This is Eric.
[01:56:55.230 --> 01:56:55.670]  No.
[01:56:55.950 --> 01:56:56.290]  No.
[01:56:56.470 --> 01:56:58.730]  I don't have any clipper chips except for right now.
[01:56:59.390 --> 01:56:59.830]  No.
[01:57:00.930 --> 01:57:01.370]  No.
[01:57:03.250 --> 01:57:04.010]  I'm sorry.
[01:57:05.490 --> 01:57:06.270]  Can you hear...
[01:57:06.730 --> 01:57:07.970]  Is this coming up in the audience?
[01:57:08.230 --> 01:57:09.310]  Yeah, I can hear you.
[01:57:09.470 --> 01:57:09.910]  There's like a...
[01:57:09.910 --> 01:57:11.730]  There's a weird quarter second delay.
[01:57:13.310 --> 01:57:16.150]  Let's get an independent judge from the audience here.
[01:57:16.250 --> 01:57:17.570]  Here, you just got a...
[01:57:18.630 --> 01:57:19.610]  Here, give us a listen.
[01:57:21.730 --> 01:57:23.530]  Tell me how this sounds to you.
[01:57:24.030 --> 01:57:24.830]  We're on the moon.
[01:57:25.530 --> 01:57:26.690]  Can you hear that we're on the moon?
[01:57:26.890 --> 01:57:28.070]  It's a satellite delay.
[01:57:31.090 --> 01:57:31.950]  It's very weak.
[01:57:32.550 --> 01:57:33.350]  It's really bad.
[01:57:35.350 --> 01:57:35.770]  Here.
[01:57:37.250 --> 01:57:38.230]  Talk to us.
[01:57:38.230 --> 01:57:38.970]  Hey.
[01:57:39.570 --> 01:57:40.230]  Can you hear it?
[01:57:40.830 --> 01:57:41.130]  Yeah.
[01:57:41.290 --> 01:57:41.490]  Yeah.
[01:57:42.890 --> 01:57:43.310]  Yeah.
[01:57:44.170 --> 01:57:45.350]  You can barely hear it.
[01:57:45.490 --> 01:57:46.370]  It really is weak.
[01:57:46.530 --> 01:57:48.050]  You can make it out, but it's just really weak audio.
[01:57:49.610 --> 01:57:56.530]  And not only is it weak, but it's got that really annoying disembodied aspect to it.
[01:57:56.530 --> 01:57:58.450]  Does the demonic voice get any louder?
[01:57:58.630 --> 01:57:58.870]  No, no.
[01:57:58.870 --> 01:57:59.430]  Check this out.
[01:57:59.510 --> 01:58:00.350]  Now I'm going to press...
[01:58:00.350 --> 01:58:00.730]  Here.
[01:58:00.870 --> 01:58:11.950]  You might be able to hear the demonic voice thing if you put the earpiece of that handset on that mic and have the gain all the way up on that microphone and turn mine all the way down.
[01:58:11.950 --> 01:58:17.450]  I'm going to push a touch tone, which will probably blast you out of your seats, but right after that, you're probably going to hear some really weird sounds.
[01:58:18.390 --> 01:58:18.710]  Ready?
[01:58:18.870 --> 01:58:19.350]  Have you heard this noise?
[01:58:19.890 --> 01:58:20.110]  Yes.
[01:58:26.230 --> 01:58:26.550]  Ooh.
[01:58:28.330 --> 01:58:29.450]  Man, I heard it at my end.
[01:58:29.550 --> 01:58:31.030]  It was really evil, but...
[01:58:31.030 --> 01:58:32.010]  Let's try that again.
[01:58:39.430 --> 01:58:40.950]  I guess I'll have to do it on my end here.
[01:58:41.650 --> 01:58:41.910]  All right.
[01:58:42.050 --> 01:58:43.130]  Turn his gain all the way down.
[01:58:46.070 --> 01:58:47.030]  And mine all the way up.
[01:58:50.470 --> 01:58:51.550]  It's just not...
[01:58:57.290 --> 01:58:57.890]  Say something.
[01:58:58.150 --> 01:58:59.930]  Put something in the buffer if there's a buffer in this thing.
[01:59:00.110 --> 01:59:00.730]  Is there a buffer?
[01:59:02.170 --> 01:59:03.370]  Can you call my mom and call her?
[01:59:03.370 --> 01:59:04.010]  Is there a buffer?
[01:59:07.390 --> 01:59:08.550]  Unfortunately, I didn't have to.
[01:59:08.830 --> 01:59:09.150]  No.
[01:59:09.290 --> 01:59:10.210]  I went out of secure mode.
[01:59:10.390 --> 01:59:13.430]  I don't think you would have picked it up, but basically, I'll try to imitate it.
[01:59:13.510 --> 01:59:14.050]  It went...
[01:59:14.770 --> 01:59:16.390]  But it was far worse.
[01:59:16.510 --> 01:59:18.110]  It was Linda Blair and the Exorcist.
[01:59:18.190 --> 01:59:18.850]  That's all I can say.
[01:59:19.030 --> 01:59:19.930]  It was that bad.
[01:59:21.230 --> 01:59:23.070]  I think we should let the next set...
[01:59:23.070 --> 01:59:24.090]  Is there a next panel?
[01:59:24.330 --> 01:59:24.490]  Yeah.
[01:59:24.970 --> 01:59:25.170]  There is.
[01:59:25.270 --> 01:59:25.550]  Oh, well.
[01:59:26.130 --> 01:59:26.690]  Is that one?
[01:59:27.510 --> 01:59:28.030]  I think so.
[01:59:28.130 --> 01:59:29.330]  So, that's the consensus.
[01:59:29.350 --> 01:59:30.090]  It sucks.
[01:59:31.110 --> 01:59:31.590]  Lauren, have you?
[01:59:31.790 --> 01:59:32.370]  Lauren, have you?
[01:59:33.510 --> 01:59:34.010]  Thank you.
[01:59:36.750 --> 01:59:43.890]  I do want to say that tomorrow we'll be giving away, as a door prize, one of the two clipper chips that I was able to social engineer out of Mycopatronics.
[01:59:43.890 --> 01:59:52.990]  And we'll be giving that lucky person the opportunity to detonate that particular chip with a small explosive device that I brought with me.
[01:59:53.970 --> 01:59:55.610]  Can we give it to our friend in Russia?
[01:59:55.830 --> 01:59:56.490]  That's a good question.
[01:59:56.750 --> 01:59:57.070]  I don't know.
[01:59:57.430 --> 02:00:01.010]  On the chip in the box, if we open the box here...
[02:00:01.010 --> 02:00:02.330]  Is it the same?
[02:00:02.330 --> 02:00:03.050]  Yeah.
[02:00:03.310 --> 02:00:04.910]  On a label that's paced over it.
[02:00:04.930 --> 02:00:05.810]  So maybe they outsourced it.
[02:00:06.010 --> 02:00:06.830]  Which, that's a fine...