0 00:00:00,000 --> 00:00:30,000 Dear viewer, these subtitles were generated by a machine via the service Trint and therefore are (very) buggy. If you are capable, please help us to create good quality subtitles: https://c3subtitles.de/talk/495 Thanks! 1 00:00:08,970 --> 00:00:11,069 My proudly present, our next 2 00:00:11,070 --> 00:00:12,479 speaker. 3 00:00:12,480 --> 00:00:14,579 He's done a lot of exciting things and 4 00:00:14,580 --> 00:00:16,619 I will just give you some very brief 5 00:00:16,620 --> 00:00:17,909 examples. 6 00:00:17,910 --> 00:00:20,189 It's things like running a company 7 00:00:20,190 --> 00:00:22,319 communication company in 8 00:00:22,320 --> 00:00:24,629 Iraq and Afghanistan, 9 00:00:24,630 --> 00:00:27,119 the things about building a data haven 10 00:00:27,120 --> 00:00:29,189 on a platform in the 11 00:00:29,190 --> 00:00:31,199 Atlantic Ocean and the North Sea. 12 00:00:31,200 --> 00:00:33,479 So I'm 13 00:00:33,480 --> 00:00:34,469 very proud to have him here. 14 00:00:34,470 --> 00:00:36,659 Please give a warm welcome to Ryan 15 00:00:36,660 --> 00:00:37,660 Lekki. 16 00:00:46,500 --> 00:00:47,759 Thank you very much for the introduction 17 00:00:47,760 --> 00:00:49,919 and it's great to be here, C.C.C. 18 00:00:49,920 --> 00:00:51,629 is my favorite conference out of all the 19 00:00:51,630 --> 00:00:54,209 conferences I go to great crowd 20 00:00:54,210 --> 00:00:56,579 first day. It's already really exciting, 21 00:00:56,580 --> 00:00:58,109 so looking forward to it. 22 00:00:58,110 --> 00:00:59,639 So I'm going to talk about data havens 23 00:00:59,640 --> 00:01:01,799 from Haven Koja today. 24 00:01:01,800 --> 00:01:03,239 Just a quick overview. 25 00:01:03,240 --> 00:01:05,669 Who am I? What's a data haven? 26 00:01:05,670 --> 00:01:06,629 Where do they come from? 27 00:01:06,630 --> 00:01:08,309 Why would you why have they failed? 28 00:01:08,310 --> 00:01:09,989 And then some ways to be successful. 29 00:01:09,990 --> 00:01:12,089 I really like this idea of having a 30 00:01:12,090 --> 00:01:13,379 conference track on things that have 31 00:01:13,380 --> 00:01:14,879 failed because normally people are 32 00:01:14,880 --> 00:01:16,169 ashamed to talk about things that have 33 00:01:16,170 --> 00:01:17,999 failed or try to minimize how much things 34 00:01:18,000 --> 00:01:19,229 and things that have gone wrong. 35 00:01:19,230 --> 00:01:21,119 And it's really the only way to learn. 36 00:01:21,120 --> 00:01:22,319 The only way to learn is to make 37 00:01:22,320 --> 00:01:24,449 mistakes. And it's better for someone 38 00:01:24,450 --> 00:01:25,739 else to make those mistakes rather than, 39 00:01:25,740 --> 00:01:27,209 you know, rather than making them over 40 00:01:27,210 --> 00:01:28,210 and over again. So. 41 00:01:33,530 --> 00:01:35,239 So just as background, I've been 42 00:01:35,240 --> 00:01:37,399 interested in crypto and cash, 43 00:01:37,400 --> 00:01:39,079 specifically anonymous electronic cash 44 00:01:39,080 --> 00:01:41,539 since the early 1990s, the cypherpunks 45 00:01:41,540 --> 00:01:42,799 mailing list and a bunch of stuff like 46 00:01:42,800 --> 00:01:44,959 that started the offshore data Haven 47 00:01:44,960 --> 00:01:46,459 Haven, which we're going to talk about in 48 00:01:46,460 --> 00:01:48,619 two thousand with some friends of mine, 49 00:01:48,620 --> 00:01:50,419 then work in some war zones doing 50 00:01:50,420 --> 00:01:51,889 satellite communications, using some of 51 00:01:51,890 --> 00:01:53,299 the experience I had from the Haven 52 00:01:53,300 --> 00:01:54,469 cooperation. 53 00:01:54,470 --> 00:01:55,969 Then I needed a trusted computing startup 54 00:01:55,970 --> 00:01:57,739 working on trust and cloud computing. 55 00:01:57,740 --> 00:01:59,239 And now I work at CloudFlare, which 56 00:01:59,240 --> 00:02:00,709 bought my trust in cloud computing 57 00:02:00,710 --> 00:02:01,609 company. 58 00:02:01,610 --> 00:02:03,439 So first thing is, what does the data 59 00:02:03,440 --> 00:02:05,119 haven? A lot of people use the term. 60 00:02:05,120 --> 00:02:06,589 They use it for different things. 61 00:02:06,590 --> 00:02:08,569 We're going to talk about the Wikipedia 62 00:02:08,570 --> 00:02:10,159 definition, which is now, I would say, 63 00:02:10,160 --> 00:02:11,719 the canonical way to find the definition 64 00:02:11,720 --> 00:02:13,999 of something, a refuge for uninterrupted 65 00:02:14,000 --> 00:02:15,349 or unregulated data. 66 00:02:15,350 --> 00:02:16,879 So that's really two parts. 67 00:02:16,880 --> 00:02:18,259 And there's a lot of factors that go into 68 00:02:18,260 --> 00:02:20,809 it, uninterrupted meaning 69 00:02:20,810 --> 00:02:23,089 against natural disaster, against normal 70 00:02:23,090 --> 00:02:24,979 service interruptions, things like that, 71 00:02:24,980 --> 00:02:26,479 but also against active attacks, 72 00:02:26,480 --> 00:02:28,339 censorship, things like that. 73 00:02:28,340 --> 00:02:30,979 There's a concept of regulatory arbitrage 74 00:02:30,980 --> 00:02:32,419 where you have one set of laws in one 75 00:02:32,420 --> 00:02:33,949 place, another set of laws in another 76 00:02:33,950 --> 00:02:36,079 place, and you're trying to pick 77 00:02:36,080 --> 00:02:37,729 and pick and choose which pieces of law 78 00:02:37,730 --> 00:02:39,169 are the best for you and the best for 79 00:02:39,170 --> 00:02:40,069 your operation. 80 00:02:40,070 --> 00:02:41,779 And there's all sorts of traits that go 81 00:02:41,780 --> 00:02:43,999 into a data haven. 82 00:02:44,000 --> 00:02:45,769 There's a physical, legal, operational 83 00:02:45,770 --> 00:02:48,079 cryptographic data haven 84 00:02:48,080 --> 00:02:50,629 could be both a physical data haven 85 00:02:50,630 --> 00:02:52,939 or a software, a data haven 86 00:02:52,940 --> 00:02:54,709 sort of cryptographic approach or some 87 00:02:54,710 --> 00:02:56,269 combination or hybrid approach to those 88 00:02:56,270 --> 00:02:57,079 things. 89 00:02:57,080 --> 00:02:59,179 So the origin, as with 90 00:02:59,180 --> 00:03:00,829 a lot of things before someone but one 91 00:03:00,830 --> 00:03:02,209 people have talked about it and thought 92 00:03:02,210 --> 00:03:03,709 about what could happen. 93 00:03:03,710 --> 00:03:05,869 Fiction, science fiction is generally 94 00:03:05,870 --> 00:03:08,359 a great predictor of science fact 95 00:03:08,360 --> 00:03:10,549 and science fiction is talked about 96 00:03:10,550 --> 00:03:11,869 this for a long time. 97 00:03:11,870 --> 00:03:13,369 There's also analogies and some very 98 00:03:13,370 --> 00:03:15,439 similar fields that are outside of 99 00:03:15,440 --> 00:03:17,239 the data or data processing world that 100 00:03:17,240 --> 00:03:18,139 are very similar. 101 00:03:18,140 --> 00:03:19,429 And there are some people doing data 102 00:03:19,430 --> 00:03:21,859 havens, at least in some sense, before 103 00:03:21,860 --> 00:03:23,839 Haven. But Haven is really the first 104 00:03:23,840 --> 00:03:26,359 purpose built physical data haven 105 00:03:26,360 --> 00:03:28,579 that that I certainly know about. 106 00:03:28,580 --> 00:03:30,829 So in fiction, there's some great science 107 00:03:30,830 --> 00:03:32,899 fiction. As we move down this list, 108 00:03:32,900 --> 00:03:34,129 they get more and more precise. 109 00:03:34,130 --> 00:03:35,959 Early on, it was more about the concept 110 00:03:35,960 --> 00:03:38,419 of crypto software things. 111 00:03:38,420 --> 00:03:40,339 Then it was sort of passing references to 112 00:03:40,340 --> 00:03:42,139 it, I would say. Islands in the Net by 113 00:03:42,140 --> 00:03:44,059 Bruce Sterling, which is a great book, is 114 00:03:44,060 --> 00:03:46,309 probably the first story 115 00:03:46,310 --> 00:03:48,589 that really references data havens and 116 00:03:48,590 --> 00:03:49,939 multiple data havens and how they 117 00:03:49,940 --> 00:03:51,289 interoperate. If you haven't read these, 118 00:03:51,290 --> 00:03:52,249 all of these are great books. 119 00:03:52,250 --> 00:03:53,989 They're part of the canon of science 120 00:03:53,990 --> 00:03:55,939 fiction. So I would definitely read them. 121 00:03:55,940 --> 00:03:57,919 Cryptonomicon has a really interesting 122 00:03:57,920 --> 00:03:59,749 history with HEREINTO. 123 00:03:59,750 --> 00:04:02,029 It was started probably ninety eight 124 00:04:02,030 --> 00:04:03,559 or so who've started writing it. 125 00:04:03,560 --> 00:04:05,089 I didn't find out about it until after 126 00:04:05,090 --> 00:04:06,319 we'd already started having code, but 127 00:04:06,320 --> 00:04:07,909 they were sort of completely parallel 128 00:04:07,910 --> 00:04:10,159 evolution and huge numbers 129 00:04:10,160 --> 00:04:12,469 of weird parallels like people's names. 130 00:04:12,470 --> 00:04:14,179 A lot, a lot of factors about it were 131 00:04:14,180 --> 00:04:15,529 very, very similar. So I guess it was 132 00:04:15,530 --> 00:04:16,730 just the right place at the right time. 133 00:04:17,959 --> 00:04:19,819 And then as you move down this list, 134 00:04:19,820 --> 00:04:21,199 Dayman and Freedom are probably my two 135 00:04:21,200 --> 00:04:22,519 favorite sci fi books. 136 00:04:22,520 --> 00:04:24,619 They're the very 137 00:04:24,620 --> 00:04:25,969 persistently referenced the idea of 138 00:04:25,970 --> 00:04:27,229 software that can't be terminated. 139 00:04:27,230 --> 00:04:29,719 So it solves this uninterrupted problem. 140 00:04:29,720 --> 00:04:31,459 But I would say that now data havens are 141 00:04:31,460 --> 00:04:33,349 like a concept that, ah, that's used in 142 00:04:33,350 --> 00:04:34,669 fiction quite, quite frequently. 143 00:04:35,750 --> 00:04:37,219 And then there are some analogies that 144 00:04:37,220 --> 00:04:39,409 are outside of the sort of data 145 00:04:39,410 --> 00:04:41,389 processing world free trade zones in 146 00:04:41,390 --> 00:04:43,399 countries. I think U.A.E is probably the 147 00:04:43,400 --> 00:04:45,829 the biggest exponent of these 148 00:04:45,830 --> 00:04:47,209 are a great analogy where you have a 149 00:04:47,210 --> 00:04:48,859 country that has one set of laws but in a 150 00:04:48,860 --> 00:04:51,019 certain geographic area or a certain set 151 00:04:51,020 --> 00:04:52,159 of entities, they're allowed to have 152 00:04:52,160 --> 00:04:53,089 different laws. 153 00:04:53,090 --> 00:04:55,129 And the idea is that foreign businesses 154 00:04:55,130 --> 00:04:57,019 will locate there, take advantage of 155 00:04:57,020 --> 00:04:58,429 these favorable laws and they wouldn't 156 00:04:58,430 --> 00:04:59,559 otherwise have been in the country and it 157 00:04:59,560 --> 00:05:00,889 doesn't affect the country. 158 00:05:00,890 --> 00:05:02,089 There's also the concept of offshore 159 00:05:02,090 --> 00:05:04,189 banking, the whole theoretical Swiss 160 00:05:04,190 --> 00:05:06,439 bank account where you are a 161 00:05:06,440 --> 00:05:08,329 political leader in some foreign country 162 00:05:08,330 --> 00:05:09,739 and you invested so much of money from 163 00:05:09,740 --> 00:05:11,389 your country and sent it off to another 164 00:05:11,390 --> 00:05:13,459 country, or you live 165 00:05:13,460 --> 00:05:14,929 in a country that's horribly repressive 166 00:05:14,930 --> 00:05:16,639 and we'll take away all your assets and 167 00:05:16,640 --> 00:05:17,989 you store them in a place like this. 168 00:05:17,990 --> 00:05:19,819 It's really a value neutral kind of 169 00:05:19,820 --> 00:05:21,889 thing, the concept of tax havens. 170 00:05:21,890 --> 00:05:23,749 I mean, people talk, at least in America, 171 00:05:23,750 --> 00:05:25,609 a lot about the way Apple shelters a lot 172 00:05:25,610 --> 00:05:27,079 of their income through foreign entities, 173 00:05:27,080 --> 00:05:28,459 even though it's earned over sort of 174 00:05:28,460 --> 00:05:30,259 overseas. But these have existed for a 175 00:05:30,260 --> 00:05:32,299 long time. And then shipping, I think 176 00:05:32,300 --> 00:05:34,639 basically all ships of 177 00:05:34,640 --> 00:05:36,199 large commercial ships are registered in 178 00:05:36,200 --> 00:05:37,999 a small number of jurisdictions that are 179 00:05:38,000 --> 00:05:40,219 different from the beneficial owners of 180 00:05:40,220 --> 00:05:41,929 that thing, like Liberia. 181 00:05:41,930 --> 00:05:43,969 Countries like that have huge numbers of 182 00:05:43,970 --> 00:05:45,859 ships but don't really have businesses 183 00:05:45,860 --> 00:05:47,209 there for them. And that's just because 184 00:05:47,210 --> 00:05:48,649 they have very favorable laws. 185 00:05:48,650 --> 00:05:50,179 And then there's gambling centers like 186 00:05:50,180 --> 00:05:51,289 McCallan next to Hong Kong. 187 00:05:51,290 --> 00:05:53,419 And China is not really catering 188 00:05:53,420 --> 00:05:55,129 to the local population. 189 00:05:55,130 --> 00:05:57,229 It's people flying there to gamble. 190 00:05:57,230 --> 00:05:58,939 So there's that. 191 00:05:58,940 --> 00:06:00,709 And then there were some data havens pre 192 00:06:00,710 --> 00:06:01,549 have go. 193 00:06:01,550 --> 00:06:03,859 So like the late 80s, early 194 00:06:03,860 --> 00:06:04,910 90s, mid 90s, 195 00:06:06,740 --> 00:06:08,659 you know, community, there was the sort 196 00:06:08,660 --> 00:06:10,789 of like top state system, which 197 00:06:10,790 --> 00:06:12,259 these were servers that were relatively 198 00:06:12,260 --> 00:06:14,449 well protected by usually being run 199 00:06:14,450 --> 00:06:15,979 by an admin on the side or something like 200 00:06:15,980 --> 00:06:17,089 that. 201 00:06:17,090 --> 00:06:18,949 There was a company, Offshore Information 202 00:06:18,950 --> 00:06:21,139 Services Limited, run by Vince Kate, who 203 00:06:21,140 --> 00:06:23,479 is one of the 204 00:06:23,480 --> 00:06:25,549 top people from the cypherpunks community 205 00:06:25,550 --> 00:06:26,779 creating this kind of stuff. 206 00:06:26,780 --> 00:06:29,779 He started a business offshore before 207 00:06:29,780 --> 00:06:31,669 almost anyone from the US had considered 208 00:06:31,670 --> 00:06:33,939 it in. Field in Anguilla, 209 00:06:33,940 --> 00:06:35,019 he was actually my next door neighbor 210 00:06:35,020 --> 00:06:36,379 while I lived there. 211 00:06:36,380 --> 00:06:37,959 He said it was a relatively small 212 00:06:37,960 --> 00:06:39,249 business. He had a bunch of, like, car 213 00:06:39,250 --> 00:06:42,279 batteries as a UPS and 214 00:06:42,280 --> 00:06:44,409 10 base to connection to me and 215 00:06:44,410 --> 00:06:45,759 t one connection or one point five 216 00:06:45,760 --> 00:06:46,749 megabit connection. 217 00:06:46,750 --> 00:06:47,859 And he had a couple of really big 218 00:06:47,860 --> 00:06:49,929 clients. But this was to some extent one 219 00:06:49,930 --> 00:06:51,459 of the first data havens because it 220 00:06:51,460 --> 00:06:53,229 hosted data from the US that was, under 221 00:06:53,230 --> 00:06:55,239 US law, not allowed to be retained for a 222 00:06:55,240 --> 00:06:57,189 period of time. It basically was a 223 00:06:57,190 --> 00:06:58,809 driver's license database from, I think, 224 00:06:58,810 --> 00:07:00,849 the state of Texas, where they couldn't 225 00:07:00,850 --> 00:07:02,379 retain the documents longer than a year. 226 00:07:02,380 --> 00:07:03,459 But there's this other company that would 227 00:07:03,460 --> 00:07:04,869 retain them forever and let people search 228 00:07:04,870 --> 00:07:07,479 against them again, value neutral. 229 00:07:07,480 --> 00:07:08,979 And then there's the whole, like, dark 230 00:07:08,980 --> 00:07:10,299 period of the 90s, which I'm afraid we're 231 00:07:10,300 --> 00:07:12,519 going back to, where crypto software 232 00:07:12,520 --> 00:07:14,829 was banned from export, or at least 233 00:07:14,830 --> 00:07:16,029 you couldn't give it to a foreign 234 00:07:16,030 --> 00:07:17,709 nationals. So effectively, you had to do 235 00:07:17,710 --> 00:07:18,849 all your crypto development if you wanted 236 00:07:18,850 --> 00:07:20,619 to be open outside of the United States. 237 00:07:21,910 --> 00:07:23,829 And we're sort of going back in that 238 00:07:23,830 --> 00:07:25,990 direction with the exploits world. 239 00:07:27,040 --> 00:07:28,599 But in that case, you had to be outside 240 00:07:28,600 --> 00:07:30,459 of the US and a non US citizen to do a 241 00:07:30,460 --> 00:07:31,539 lot of the development. 242 00:07:31,540 --> 00:07:33,069 There was a system that Ross Anderson 243 00:07:33,070 --> 00:07:35,379 from Cambridge had Triniti and then 244 00:07:35,380 --> 00:07:37,659 a few people, two 245 00:07:37,660 --> 00:07:40,359 of whom are on the possibly Satoshi list, 246 00:07:40,360 --> 00:07:41,679 created versions of it. 247 00:07:41,680 --> 00:07:43,599 And I created a pretty bad version of it. 248 00:07:43,600 --> 00:07:44,499 TasRail Weber. 249 00:07:44,500 --> 00:07:45,969 These are all software systems at the 250 00:07:45,970 --> 00:07:47,259 end. And then of course, there's the 251 00:07:47,260 --> 00:07:49,029 concept of availability from a high end 252 00:07:49,030 --> 00:07:51,159 data center. You've got great data 253 00:07:51,160 --> 00:07:52,929 centers where people spent three sixty 254 00:07:52,930 --> 00:07:54,669 five men in San Francisco, two billion 255 00:07:54,670 --> 00:07:56,199 dollars building this data center. 256 00:07:56,200 --> 00:07:57,669 The idea being that it'll stay up through 257 00:07:57,670 --> 00:07:59,289 any sort of disaster. It's isolated from 258 00:07:59,290 --> 00:08:00,639 earthquakes. It's got a bunch of diesel 259 00:08:00,640 --> 00:08:01,569 fuel onsite. 260 00:08:01,570 --> 00:08:02,829 Of course, there's a software bug and all 261 00:08:02,830 --> 00:08:04,449 the UPSs. So they all crash at the same 262 00:08:04,450 --> 00:08:07,179 time. But, yeah, minor data 263 00:08:07,180 --> 00:08:09,129 and there's been uses of this censorship. 264 00:08:09,130 --> 00:08:10,449 Resistance is the first thing that most 265 00:08:10,450 --> 00:08:11,739 people think of when they think of a data 266 00:08:11,740 --> 00:08:13,089 haven. They think of data that someone's 267 00:08:13,090 --> 00:08:14,559 actively trying to shut down. 268 00:08:14,560 --> 00:08:15,999 But part of it is just uncertainty. 269 00:08:16,000 --> 00:08:18,099 Like until very recently, we're still 270 00:08:18,100 --> 00:08:19,479 ongoing. We don't really know the legal 271 00:08:19,480 --> 00:08:21,399 status of things like Bitcoin. 272 00:08:21,400 --> 00:08:23,019 So if you can pick a place that has 273 00:08:23,020 --> 00:08:24,549 decided what the legal status is, it's a 274 00:08:24,550 --> 00:08:25,689 lot less risky. 275 00:08:25,690 --> 00:08:27,189 And there's also just durability and 276 00:08:27,190 --> 00:08:28,659 reliability. If you're spending a lot of 277 00:08:28,660 --> 00:08:29,979 money to have a service that needs to be 278 00:08:29,980 --> 00:08:31,399 up all the time, you probably don't want 279 00:08:31,400 --> 00:08:32,439 to host it in your basement. You want to 280 00:08:32,440 --> 00:08:34,298 find a physically secure facility for it. 281 00:08:34,299 --> 00:08:35,589 And there's a lot of things. 282 00:08:35,590 --> 00:08:36,548 And then there's, of course, choice of 283 00:08:36,549 --> 00:08:38,859 law like gambling in the US is 284 00:08:38,860 --> 00:08:41,019 a super big problem to have any 285 00:08:41,020 --> 00:08:43,029 connection to, especially recently purely 286 00:08:43,030 --> 00:08:44,030 for 287 00:08:45,670 --> 00:08:47,679 protection of the big existing casino 288 00:08:47,680 --> 00:08:49,779 industry. There's no moral really issues 289 00:08:49,780 --> 00:08:52,359 with it or anything particularly, but 290 00:08:52,360 --> 00:08:53,529 you can't have any of that stuff touch 291 00:08:53,530 --> 00:08:54,849 the US. You have to pick a jurisdiction 292 00:08:54,850 --> 00:08:55,850 where you can. 293 00:08:56,850 --> 00:08:57,959 So then there's this place called 294 00:08:57,960 --> 00:09:00,239 Sealand, so World 295 00:09:00,240 --> 00:09:02,399 War two, anti-aircraft anti-aircraft 296 00:09:02,400 --> 00:09:04,889 fortress in the North Sea, which 297 00:09:04,890 --> 00:09:07,139 if you just take that statement, it's 298 00:09:07,140 --> 00:09:08,140 crazy. 299 00:09:08,610 --> 00:09:10,199 So during World War two, the British had 300 00:09:10,200 --> 00:09:11,819 a problem with German bombers coming in 301 00:09:11,820 --> 00:09:12,809 and bombing London. 302 00:09:12,810 --> 00:09:14,339 So they decided they wanted to intercept 303 00:09:14,340 --> 00:09:15,779 them not over London, but over the North 304 00:09:15,780 --> 00:09:18,239 Sea. They started building these 305 00:09:18,240 --> 00:09:20,339 floating platforms out in 306 00:09:20,340 --> 00:09:22,079 the North Sea. And we're not floating, 307 00:09:22,080 --> 00:09:23,039 but they're anchored platforms in the 308 00:09:23,040 --> 00:09:24,090 North Sea. And 309 00:09:25,110 --> 00:09:26,110 these 310 00:09:27,240 --> 00:09:28,679 sort of got left behind at the end of the 311 00:09:28,680 --> 00:09:30,749 war. And they were sort of legal 312 00:09:30,750 --> 00:09:32,259 curiosities. Most of them got torn down 313 00:09:32,260 --> 00:09:33,209 and one of them didn't. 314 00:09:33,210 --> 00:09:34,769 The U.K. also did not have commercial 315 00:09:34,770 --> 00:09:36,929 radio. So 316 00:09:36,930 --> 00:09:38,849 people were doing pirate pirate radio 317 00:09:38,850 --> 00:09:40,259 broadcasts from ships. 318 00:09:40,260 --> 00:09:41,789 Then they started cracking down on pirate 319 00:09:41,790 --> 00:09:43,379 radio broadcasts from ships. 320 00:09:43,380 --> 00:09:45,449 And so people started moving farther and 321 00:09:45,450 --> 00:09:47,549 farther out and they eventually started 322 00:09:47,550 --> 00:09:48,569 looking at places like this. 323 00:09:48,570 --> 00:09:49,979 As far as I know, they never actually did 324 00:09:49,980 --> 00:09:51,359 pirate radio operations from this 325 00:09:51,360 --> 00:09:52,589 particular fort. They did it from other 326 00:09:52,590 --> 00:09:54,179 locations. But it was basically this 327 00:09:54,180 --> 00:09:56,249 place that was in at the 328 00:09:56,250 --> 00:09:58,469 time, international waters occupied 329 00:09:58,470 --> 00:10:00,269 by somebody who declared it sovereign. 330 00:10:00,270 --> 00:10:01,679 And that if you look at international 331 00:10:01,680 --> 00:10:03,599 law, which is kind of crazy, is how you 332 00:10:03,600 --> 00:10:06,359 start a country. And so 333 00:10:06,360 --> 00:10:08,849 it's probably maybe sort of a country 334 00:10:08,850 --> 00:10:10,919 under the technical definition, although 335 00:10:10,920 --> 00:10:12,419 the population is very small, territory 336 00:10:12,420 --> 00:10:13,949 is very small or some other things. 337 00:10:13,950 --> 00:10:15,599 No one really cared about it because it 338 00:10:15,600 --> 00:10:17,939 was a relatively upstanding British 339 00:10:17,940 --> 00:10:19,139 family that had it. They didn't do 340 00:10:19,140 --> 00:10:20,140 anything bad with it. 341 00:10:21,090 --> 00:10:22,319 They didn't really cause too much 342 00:10:22,320 --> 00:10:23,789 trouble. The British are pretty 343 00:10:23,790 --> 00:10:25,379 accommodating of that kind of thing. 344 00:10:25,380 --> 00:10:26,969 And it was sort of interesting legal 345 00:10:26,970 --> 00:10:28,619 curiosity, but that was about it. 346 00:10:30,000 --> 00:10:32,429 And then then we came around, 347 00:10:33,600 --> 00:10:34,600 there was this 348 00:10:35,790 --> 00:10:38,219 set of people who will name 349 00:10:38,220 --> 00:10:40,379 who had done business in 350 00:10:40,380 --> 00:10:42,449 various places, Anguilla, various 351 00:10:42,450 --> 00:10:44,639 other places, and thought, oh, Sealand 352 00:10:44,640 --> 00:10:45,899 will be a great place to do it and having 353 00:10:45,900 --> 00:10:47,099 them out there. 354 00:10:47,100 --> 00:10:48,719 So that's that's our story. 355 00:10:48,720 --> 00:10:49,889 So the founders are Sean and Joe 356 00:10:49,890 --> 00:10:52,049 Hastings, who have done crypto software 357 00:10:52,050 --> 00:10:53,069 for a long time. 358 00:10:53,070 --> 00:10:55,049 And me, we'd worked on crypto software 359 00:10:55,050 --> 00:10:57,239 actually in Anguilla about a year 360 00:10:57,240 --> 00:10:59,339 apart and worked from 361 00:10:59,340 --> 00:11:00,959 there for the Ayata reasons. 362 00:11:00,960 --> 00:11:03,479 Samir Parekh, who was the first SSL 363 00:11:03,480 --> 00:11:05,519 license vendor in the US in the 90s, and 364 00:11:05,520 --> 00:11:07,409 Avi Friedman, who was an early guy 365 00:11:07,410 --> 00:11:09,359 working on Internet stuff and then at 366 00:11:09,360 --> 00:11:11,069 Akamai to much other places, were very 367 00:11:11,070 --> 00:11:13,079 helpful. Early investors and we had 368 00:11:13,080 --> 00:11:14,609 experienced first hand with Anguilla, 369 00:11:14,610 --> 00:11:15,629 this little tiny country. 370 00:11:15,630 --> 00:11:17,729 It's maybe seven thousand people in the 371 00:11:17,730 --> 00:11:19,679 Caribbean that didn't really have 372 00:11:19,680 --> 00:11:20,849 favorable laws in any way. 373 00:11:20,850 --> 00:11:22,379 It just didn't have a lot of laws. 374 00:11:22,380 --> 00:11:24,479 And we just moved there because it 375 00:11:24,480 --> 00:11:26,219 was sort of a nice place to live, sort 376 00:11:26,220 --> 00:11:27,419 of, if you like, Caribbean islands. 377 00:11:27,420 --> 00:11:29,669 But there was no Internet, shady legal 378 00:11:29,670 --> 00:11:30,839 system, all sorts of stuff like that. 379 00:11:30,840 --> 00:11:31,840 So. 380 00:11:33,220 --> 00:11:34,599 Purely by accident, basically, we were 381 00:11:34,600 --> 00:11:35,600 all there at the same time, 382 00:11:37,510 --> 00:11:39,609 so one of the things 383 00:11:39,610 --> 00:11:40,959 that happened was, one, it was really 384 00:11:40,960 --> 00:11:42,609 boring. I don't really like little 385 00:11:42,610 --> 00:11:43,929 Caribbean islands. I'd be much happier in 386 00:11:43,930 --> 00:11:45,309 a place with lots of fast Internet. 387 00:11:46,690 --> 00:11:48,669 And it's one of those tourist places 388 00:11:48,670 --> 00:11:50,139 where you're only busy and, what, one 389 00:11:50,140 --> 00:11:51,140 month out of the year, 390 00:11:52,450 --> 00:11:53,949 the law was unsettled and a lot of other 391 00:11:53,950 --> 00:11:56,199 places we couldn't really go there. 392 00:11:56,200 --> 00:11:58,329 We looked at some 393 00:11:58,330 --> 00:11:59,529 other countries we might be able to go 394 00:11:59,530 --> 00:12:01,149 to. 395 00:12:01,150 --> 00:12:03,549 So we left and we left island or we left 396 00:12:03,550 --> 00:12:04,550 Anguilla. 397 00:12:05,290 --> 00:12:07,719 We rented the house from a 398 00:12:07,720 --> 00:12:10,179 government officials brother at above 399 00:12:10,180 --> 00:12:12,039 market rate. And then the other political 400 00:12:12,040 --> 00:12:13,809 party got elected in and none of our work 401 00:12:13,810 --> 00:12:14,889 permits were valid anymore. 402 00:12:14,890 --> 00:12:16,359 So we basically got kicked off. 403 00:12:16,360 --> 00:12:18,249 It was exactly what you'd expect from a 404 00:12:18,250 --> 00:12:19,250 tiny country like that. 405 00:12:20,380 --> 00:12:22,689 So we all left and we were all in Oakland 406 00:12:22,690 --> 00:12:24,369 again. So Oakland, California. 407 00:12:24,370 --> 00:12:25,569 So we were like hanging out, trying to 408 00:12:25,570 --> 00:12:26,979 figure out what to do next and like, oh, 409 00:12:26,980 --> 00:12:28,239 we can't go to another little Caribbean 410 00:12:28,240 --> 00:12:29,529 island or something like that. 411 00:12:29,530 --> 00:12:31,089 We have to find some better solution to 412 00:12:31,090 --> 00:12:32,409 this thing because we all wanted to build 413 00:12:32,410 --> 00:12:34,599 anonymous electronic cash, which is, I 414 00:12:34,600 --> 00:12:36,069 would say, probably the most difficult 415 00:12:36,070 --> 00:12:36,999 application to build. 416 00:12:37,000 --> 00:12:38,439 It meets all the requirements later in 417 00:12:38,440 --> 00:12:39,440 the talk. 418 00:12:40,060 --> 00:12:41,589 So we were looking at either existing 419 00:12:41,590 --> 00:12:43,389 countries, we could use free trade zones, 420 00:12:43,390 --> 00:12:44,979 we could possibly negotiate with certain 421 00:12:44,980 --> 00:12:46,239 countries. 422 00:12:46,240 --> 00:12:47,889 And then this concept of microstates. 423 00:12:47,890 --> 00:12:50,019 So like places like sea land, 424 00:12:50,020 --> 00:12:51,729 there's a place in Australia, the river 425 00:12:51,730 --> 00:12:54,249 province, all sorts of legal curiosities. 426 00:12:54,250 --> 00:12:55,929 We found this book called How to Start 427 00:12:55,930 --> 00:12:57,459 Your Own Country, which was Illume 428 00:12:57,460 --> 00:12:59,529 Panic's press, which is sort 429 00:12:59,530 --> 00:13:00,909 of an alternative. 430 00:13:00,910 --> 00:13:03,129 Interesting stuff. Press and they 431 00:13:03,130 --> 00:13:04,359 publish it. They mentioned Zealand. 432 00:13:04,360 --> 00:13:05,679 We contacted them and did that. 433 00:13:05,680 --> 00:13:07,899 But one of our other rejected ideas 434 00:13:07,900 --> 00:13:10,269 was to get a bunch of ships 435 00:13:10,270 --> 00:13:12,039 except toxic waste from a bunch of 436 00:13:12,040 --> 00:13:14,169 countries, put servers on the same barge 437 00:13:14,170 --> 00:13:16,359 as the toxic waste such that they 438 00:13:16,360 --> 00:13:17,829 couldn't really do anything to the barge 439 00:13:17,830 --> 00:13:19,089 without causing like an environmental 440 00:13:19,090 --> 00:13:22,309 catastrophe. So, yes. 441 00:13:22,310 --> 00:13:23,310 Yeah. 442 00:13:25,950 --> 00:13:28,079 Yeah, I'm kind of glad 443 00:13:28,080 --> 00:13:29,080 we didn't do that, actually, 444 00:13:30,180 --> 00:13:32,399 so we found this place called Sealand and 445 00:13:32,400 --> 00:13:33,749 emailed the guy. 446 00:13:33,750 --> 00:13:35,669 It was a family that there were like two 447 00:13:35,670 --> 00:13:37,499 people living out on the place part time, 448 00:13:37,500 --> 00:13:39,719 and they ran 449 00:13:39,720 --> 00:13:42,149 a fishing business in the in the coast 450 00:13:42,150 --> 00:13:42,569 nearby. 451 00:13:42,570 --> 00:13:44,429 So it was it was basically just used 452 00:13:45,840 --> 00:13:48,089 it had a lot of stuff left over from 453 00:13:48,090 --> 00:13:49,349 World War Two, like they did build it 454 00:13:49,350 --> 00:13:50,399 right after the war. So there are no, 455 00:13:50,400 --> 00:13:52,109 like, actual useful weapons or anything 456 00:13:52,110 --> 00:13:54,479 on it. But it had like 457 00:13:54,480 --> 00:13:56,219 tools and a bunch of stuff left over from 458 00:13:56,220 --> 00:13:58,349 them. And it was about five thousand, ten 459 00:13:58,350 --> 00:14:00,659 thousand square feet falling apart. 460 00:14:00,660 --> 00:14:01,799 Lots of lots of issues. 461 00:14:01,800 --> 00:14:03,389 All the rooms look sort of like this, if 462 00:14:03,390 --> 00:14:04,469 not worse. This was actually one of the 463 00:14:04,470 --> 00:14:05,470 cleanest rooms 464 00:14:06,840 --> 00:14:08,639 and they did have guns, but they were 465 00:14:08,640 --> 00:14:10,709 these like four inch deck guns 466 00:14:10,710 --> 00:14:13,229 or something that were like rusted 467 00:14:13,230 --> 00:14:15,149 solid and ended up getting cut and thrown 468 00:14:15,150 --> 00:14:16,150 overboard. 469 00:14:17,440 --> 00:14:18,999 And this is sort of the layout of the 470 00:14:19,000 --> 00:14:21,009 thing, it was a structure to cylindrical 471 00:14:21,010 --> 00:14:22,689 towers on top of a concrete barge. 472 00:14:22,690 --> 00:14:24,309 The whole thing's made out of harassment 473 00:14:24,310 --> 00:14:26,199 and a superstructure. 474 00:14:26,200 --> 00:14:27,939 There were at one point three hundred 475 00:14:27,940 --> 00:14:29,049 people living on this thing. 476 00:14:29,050 --> 00:14:30,009 The most we ever got on. 477 00:14:30,010 --> 00:14:32,169 It was about 20, but 478 00:14:32,170 --> 00:14:33,219 usually around two. 479 00:14:34,480 --> 00:14:37,089 And it was everything 480 00:14:37,090 --> 00:14:39,219 you would fear a 50 481 00:14:39,220 --> 00:14:41,859 or 60 year old sort of abandoned 482 00:14:41,860 --> 00:14:43,089 structure would be like, 483 00:14:44,470 --> 00:14:46,479 yeah, so they assembled it in place and 484 00:14:46,480 --> 00:14:47,890 they sunk it. It was pretty crazy 485 00:14:49,090 --> 00:14:51,069 and it was decorated in high British 486 00:14:51,070 --> 00:14:52,070 fashion. 487 00:14:52,490 --> 00:14:53,559 Yeah. 488 00:14:53,560 --> 00:14:55,629 So, yeah. And it's where it is. 489 00:14:55,630 --> 00:14:57,909 It's so they changed the law on 490 00:14:57,910 --> 00:14:59,979 what, international waters where it used 491 00:14:59,980 --> 00:15:01,809 to be three nautical miles and you were 492 00:15:01,810 --> 00:15:03,519 international. They changed it with 12, 493 00:15:03,520 --> 00:15:06,009 but it was after the declaration 494 00:15:06,010 --> 00:15:07,629 of sea land being an independent country. 495 00:15:07,630 --> 00:15:10,029 So it's probably 496 00:15:10,030 --> 00:15:12,249 OK, but just adds 497 00:15:12,250 --> 00:15:13,149 lots of uncertainty. 498 00:15:13,150 --> 00:15:14,829 And if you see the town, Felixstowe, we 499 00:15:14,830 --> 00:15:17,079 had a six story building there that we 500 00:15:17,080 --> 00:15:18,669 beamed our communications using a point 501 00:15:18,670 --> 00:15:19,839 to point wi fi shot to. 502 00:15:19,840 --> 00:15:22,029 So it was a pretty close to shore, not 503 00:15:22,030 --> 00:15:23,739 that far. But the North Sea is not a fun 504 00:15:23,740 --> 00:15:24,809 place to be on a boat. 505 00:15:25,870 --> 00:15:27,789 So there's sort of stages of the island, 506 00:15:27,790 --> 00:15:29,679 Evancho adventure, all the ceiling stuff 507 00:15:29,680 --> 00:15:31,179 that happened before I got involved 508 00:15:31,180 --> 00:15:32,949 happened, but not a whole lot happened. 509 00:15:32,950 --> 00:15:34,659 There was some crazy legal things, but 510 00:15:34,660 --> 00:15:37,239 not very much physical infrastructure. 511 00:15:37,240 --> 00:15:38,559 During the starting up phase. 512 00:15:38,560 --> 00:15:40,839 We did a lot of physical stuff 513 00:15:40,840 --> 00:15:41,949 and then we set up a business structure 514 00:15:41,950 --> 00:15:44,019 in parallel. And I was like 19 years 515 00:15:44,020 --> 00:15:45,459 old at the time. This was also during the 516 00:15:45,460 --> 00:15:46,359 first dotcom boom. 517 00:15:46,360 --> 00:15:47,709 So we didn't really have a lot of precedent 518 00:15:47,710 --> 00:15:49,809 to sort of copy things 519 00:15:49,810 --> 00:15:51,909 from. It'd be really easy to do most of 520 00:15:51,910 --> 00:15:53,739 the stuff today. And technology was not 521 00:15:53,740 --> 00:15:54,819 quite where it should have been. 522 00:15:54,820 --> 00:15:55,989 But yeah, this is the structure as we 523 00:15:55,990 --> 00:15:57,009 went up to it. 524 00:15:57,010 --> 00:15:59,049 This is this is actually an older photo 525 00:15:59,050 --> 00:16:00,849 or a newer photo and cleaned up a little 526 00:16:00,850 --> 00:16:02,889 bit. But it was this basically bare 527 00:16:02,890 --> 00:16:03,890 structure. 528 00:16:07,980 --> 00:16:10,259 Getting out to it was on these little 529 00:16:10,260 --> 00:16:11,729 rigid inflatable 530 00:16:13,440 --> 00:16:15,629 rib boats, and you would lift 531 00:16:15,630 --> 00:16:17,129 the entire boat out of the water using a 532 00:16:17,130 --> 00:16:19,559 crane, which was an exciting process. 533 00:16:19,560 --> 00:16:21,659 I learned to use pelican dry 534 00:16:21,660 --> 00:16:23,099 cases because if you don't, then your 535 00:16:23,100 --> 00:16:25,409 stuff gets wet. And it's really fun 536 00:16:25,410 --> 00:16:27,089 when you have a non backed up sunny via 537 00:16:27,090 --> 00:16:29,189 laptop to have to disassemble 538 00:16:29,190 --> 00:16:31,259 it, run everything through DNA as 539 00:16:31,260 --> 00:16:32,699 ionized water or the closest thing you 540 00:16:32,700 --> 00:16:34,979 have, and then like remove 541 00:16:34,980 --> 00:16:36,269 the electronic squirt from a hard drive 542 00:16:36,270 --> 00:16:37,979 because your only copies of keys were on 543 00:16:37,980 --> 00:16:39,449 that because somebody had taken your 544 00:16:39,450 --> 00:16:40,679 laptop out of the drive back. 545 00:16:40,680 --> 00:16:41,909 But yeah. 546 00:16:41,910 --> 00:16:43,619 So things like that. 547 00:16:43,620 --> 00:16:44,519 This is what it looks like when you're 548 00:16:44,520 --> 00:16:46,769 getting carried up and it's really small 549 00:16:46,770 --> 00:16:48,839 from the air helicopter 550 00:16:48,840 --> 00:16:50,099 trips out for this thing because it's an 551 00:16:50,100 --> 00:16:51,569 officer structure where like three 552 00:16:51,570 --> 00:16:53,819 thousand pounds, so about five thousand 553 00:16:53,820 --> 00:16:55,379 dollars or so each trip because they have 554 00:16:55,380 --> 00:16:56,589 to be twin engine and everything else. 555 00:16:56,590 --> 00:16:57,959 So we would only really is a helicopter 556 00:16:57,960 --> 00:16:59,549 if the press were paying, if we were 557 00:16:59,550 --> 00:17:00,879 doing it. We use these little boats or 558 00:17:00,880 --> 00:17:02,449 used a fishing boat. 559 00:17:02,450 --> 00:17:04,019 This is sort of the process of getting 560 00:17:04,020 --> 00:17:05,639 winched up. In some cases you wear a 561 00:17:05,640 --> 00:17:07,259 harness and it would get attached to the 562 00:17:07,260 --> 00:17:08,549 back of your head and that. 563 00:17:08,550 --> 00:17:10,799 And later we added a 500 pound 564 00:17:10,800 --> 00:17:12,899 concrete ball as ballast, which made 565 00:17:12,900 --> 00:17:14,429 the whole process even more dangerous 566 00:17:14,430 --> 00:17:15,959 because if that thing hit you, also the 567 00:17:15,960 --> 00:17:17,759 sea is pitching up and down like 15 feet. 568 00:17:17,760 --> 00:17:19,828 So there's a timing 569 00:17:19,829 --> 00:17:22,709 aspect to this. It's crazy. 570 00:17:22,710 --> 00:17:23,818 Yeah. 571 00:17:23,819 --> 00:17:25,379 So the really interesting part is the 572 00:17:25,380 --> 00:17:26,818 datalink. So this is the first thing we 573 00:17:26,819 --> 00:17:28,889 set up. This was a Teche on one point two 574 00:17:28,890 --> 00:17:31,289 meter dish, part of a network 575 00:17:31,290 --> 00:17:32,549 which I later worked on that kind of 576 00:17:32,550 --> 00:17:33,570 stuff in Iraq and Afghanistan. 577 00:17:34,780 --> 00:17:36,929 You could get maybe two megs down, 578 00:17:36,930 --> 00:17:39,269 maybe a quarter mag up, 579 00:17:39,270 --> 00:17:41,219 and it was a shared network across a lot 580 00:17:41,220 --> 00:17:43,079 of networks. The interesting thing about 581 00:17:43,080 --> 00:17:44,669 this is I had never really set up a 582 00:17:44,670 --> 00:17:45,569 satellite network before. 583 00:17:45,570 --> 00:17:47,969 And there's this whole polarization 584 00:17:47,970 --> 00:17:49,829 angle thing where you have horizontal and 585 00:17:49,830 --> 00:17:51,809 vertical polarization on signals. 586 00:17:51,810 --> 00:17:53,369 It turns out that the alternate 587 00:17:53,370 --> 00:17:55,079 polarization was actually the credit card 588 00:17:55,080 --> 00:17:57,119 processor for like all the gas stations 589 00:17:57,120 --> 00:17:58,139 in Europe. 590 00:17:58,140 --> 00:17:59,549 And I didn't have the thing turn 591 00:17:59,550 --> 00:18:02,249 correctly. So I blocked out satellite 592 00:18:02,250 --> 00:18:03,539 credit card processing for like 10 593 00:18:03,540 --> 00:18:04,649 minutes when I first saw this thing up, 594 00:18:04,650 --> 00:18:06,199 which was kind of scary. 595 00:18:06,200 --> 00:18:07,379 You can do that with like one small 596 00:18:07,380 --> 00:18:09,299 satellite dish. But so that was our sort 597 00:18:09,300 --> 00:18:11,369 of a backup link because VSAT systems go 598 00:18:11,370 --> 00:18:13,589 to geostationary orbit and there's about 599 00:18:13,590 --> 00:18:15,389 six hundred milliseconds, a thousand 600 00:18:15,390 --> 00:18:17,819 milliseconds of latency added, which is 601 00:18:17,820 --> 00:18:19,349 not so great for communications. 602 00:18:20,520 --> 00:18:21,929 Then at the very top of this, sorry, I 603 00:18:21,930 --> 00:18:22,919 didn't have any better photos. 604 00:18:22,920 --> 00:18:25,139 I didn't take as many photos back then 605 00:18:25,140 --> 00:18:27,729 as I would have as I should have. 606 00:18:27,730 --> 00:18:29,369 There was a microwave link. 607 00:18:29,370 --> 00:18:30,659 It was originally a wi fi link. 608 00:18:30,660 --> 00:18:32,339 I think it was an eight or two eleven B 609 00:18:32,340 --> 00:18:34,469 link with a PCMCIA card and 610 00:18:34,470 --> 00:18:35,759 a bunch of cables and a bunch of like 611 00:18:35,760 --> 00:18:37,649 completely hacked together junk going 612 00:18:37,650 --> 00:18:39,089 from this to the building on shore. 613 00:18:39,090 --> 00:18:40,739 We later replaced it with a thirty 614 00:18:40,740 --> 00:18:42,929 thousand four by one system for no 615 00:18:42,930 --> 00:18:44,619 good reason. That didn't work as well. 616 00:18:44,620 --> 00:18:46,679 So yeah, the wi fi version works 617 00:18:46,680 --> 00:18:48,389 better, way better. Then we had FreeBSD 618 00:18:48,390 --> 00:18:50,909 boxes sitting on shore that had 619 00:18:50,910 --> 00:18:52,739 E one cards because I was really anti 620 00:18:52,740 --> 00:18:53,969 Cisco at the time because I wanted to 621 00:18:53,970 --> 00:18:55,649 open source routers. So I ran everything 622 00:18:55,650 --> 00:18:57,839 on Xebra and Free 623 00:18:57,840 --> 00:19:00,389 BSD boxes and 624 00:19:00,390 --> 00:19:01,769 the whole thing was like crazy because 625 00:19:01,770 --> 00:19:03,749 you could we didn't have enough 626 00:19:03,750 --> 00:19:04,769 redundancy and power cycle. 627 00:19:04,770 --> 00:19:06,599 And so there were times where the power 628 00:19:06,600 --> 00:19:07,829 would go out in this building or 629 00:19:07,830 --> 00:19:08,879 something would crash. And we have to 630 00:19:08,880 --> 00:19:10,889 like take a boat to go to a location a 631 00:19:10,890 --> 00:19:12,029 bunch of times where I did, like a 632 00:19:12,030 --> 00:19:14,669 McWorld upgrade without 633 00:19:14,670 --> 00:19:16,739 it working and the machine wouldn't 634 00:19:16,740 --> 00:19:18,599 come back up, stupid stuff like that that 635 00:19:18,600 --> 00:19:20,219 I would never do again. 636 00:19:20,220 --> 00:19:22,349 But you can you can 637 00:19:22,350 --> 00:19:23,699 make this work with, like, really crazy 638 00:19:23,700 --> 00:19:25,419 stuff today. You could do this with us. 639 00:19:25,420 --> 00:19:26,759 You're actually within range of a cell 640 00:19:26,760 --> 00:19:28,139 phone, I'm sure. So you can just put a 641 00:19:28,140 --> 00:19:29,849 cell phone data card in your laptop and 642 00:19:29,850 --> 00:19:31,319 do largely the same thing. 643 00:19:31,320 --> 00:19:32,609 The way we set up our network was 644 00:19:32,610 --> 00:19:34,709 actually intentionally this 645 00:19:34,710 --> 00:19:36,719 way. We had a transport session that went 646 00:19:36,720 --> 00:19:38,819 between us and London Telehouse and 647 00:19:38,820 --> 00:19:40,709 another appearing facility so that we 648 00:19:40,710 --> 00:19:42,599 publicly appeared with people in these 649 00:19:42,600 --> 00:19:44,099 high bandwidth locations. 650 00:19:44,100 --> 00:19:45,119 We did a lot of filtering and then 651 00:19:45,120 --> 00:19:47,519 brought it back to the To Haven CO 652 00:19:47,520 --> 00:19:48,989 via links that we control that we could 653 00:19:48,990 --> 00:19:50,309 obfuscate because those links were much 654 00:19:50,310 --> 00:19:51,389 harder to replace. 655 00:19:51,390 --> 00:19:52,440 Something I talked about a little bit. 656 00:19:54,120 --> 00:19:55,349 OK, so we did launch. 657 00:19:56,730 --> 00:19:58,919 There was this Wired magazine has 658 00:19:58,920 --> 00:20:00,779 like a four month lead time on 659 00:20:00,780 --> 00:20:01,709 publications. 660 00:20:01,710 --> 00:20:03,839 So we were telling them what we 661 00:20:03,840 --> 00:20:06,119 were going to do in advance of actually 662 00:20:06,120 --> 00:20:08,499 doing it because it had to be ready 663 00:20:08,500 --> 00:20:10,649 like we wanted it ready before the 664 00:20:10,650 --> 00:20:11,609 the press hit. 665 00:20:11,610 --> 00:20:14,129 So there's a lot of speculative stuff. 666 00:20:14,130 --> 00:20:15,809 Probably the dumbest quote was the 667 00:20:15,810 --> 00:20:18,089 nitrogen filled data center quote, where 668 00:20:18,090 --> 00:20:20,099 conceivably you could do that, but we had 669 00:20:20,100 --> 00:20:21,869 no money to do that kind of thing. 670 00:20:21,870 --> 00:20:23,369 So we had a secret plan to all of this, 671 00:20:23,370 --> 00:20:25,589 which was to get a lot of press and using 672 00:20:25,590 --> 00:20:27,479 all that press in the peak of the market, 673 00:20:27,480 --> 00:20:29,249 be able to negotiate with another country 674 00:20:29,250 --> 00:20:30,929 to set up the second version of this, 675 00:20:30,930 --> 00:20:32,639 because our thinking was that doing the 676 00:20:32,640 --> 00:20:34,019 first one of these would be really hard 677 00:20:34,020 --> 00:20:35,429 to have any country say this would be a 678 00:20:35,430 --> 00:20:36,479 worthwhile thing to do. 679 00:20:36,480 --> 00:20:37,949 But getting another country to say, oh, 680 00:20:37,950 --> 00:20:39,239 we'll be the backup site for that would 681 00:20:39,240 --> 00:20:41,309 be really easy. So we get like a a more 682 00:20:41,310 --> 00:20:43,079 real place like, say, Hong Kong to be our 683 00:20:43,080 --> 00:20:44,399 secondary data center. 684 00:20:44,400 --> 00:20:45,629 And then once you've got to, you can do 685 00:20:45,630 --> 00:20:47,039 like ten of them really easily. 686 00:20:47,040 --> 00:20:49,139 So that was a reasonable plan, I think. 687 00:20:49,140 --> 00:20:50,249 But we didn't get to that point. 688 00:20:50,250 --> 00:20:52,139 We got about three hundred major press 689 00:20:52,140 --> 00:20:53,609 articles. We had press flying out there 690 00:20:53,610 --> 00:20:55,409 all the time. It was a height of dotcom 691 00:20:55,410 --> 00:20:57,149 frenzy. But even at this point, it was 692 00:20:57,150 --> 00:21:00,119 team disorganization and stuff. 693 00:21:00,120 --> 00:21:02,189 We had like a sales email box that we 694 00:21:02,190 --> 00:21:03,779 didn't answer. We just let, like, 695 00:21:03,780 --> 00:21:05,669 accumulate because we were arguing over 696 00:21:05,670 --> 00:21:07,809 which tech. Getting system we should use, 697 00:21:07,810 --> 00:21:09,219 so we just didn't answer it, which was 698 00:21:09,220 --> 00:21:10,220 yeah, yeah, 699 00:21:11,950 --> 00:21:14,229 so yeah. And this is one of our watch 700 00:21:14,230 --> 00:21:16,479 photos and this is me 701 00:21:16,480 --> 00:21:17,919 and this is Michael Bates, who was at the 702 00:21:17,920 --> 00:21:20,259 time the son of the Prince of Sealants, 703 00:21:20,260 --> 00:21:21,309 who's like the second. 704 00:21:21,310 --> 00:21:23,539 I don't know what the real thing is, but. 705 00:21:23,540 --> 00:21:24,540 Yeah. 706 00:21:25,060 --> 00:21:27,339 This is yet more examples of lovely 707 00:21:27,340 --> 00:21:29,200 British decor for this place, 708 00:21:30,670 --> 00:21:32,019 press corps. 709 00:21:32,020 --> 00:21:34,629 So, as you can expect, 710 00:21:34,630 --> 00:21:36,519 given as a failed talk, there was there's 711 00:21:36,520 --> 00:21:38,139 a crash. So there was the dotcom 712 00:21:38,140 --> 00:21:39,489 collapse, which I don't know how many 713 00:21:39,490 --> 00:21:41,589 people remember the details of it, but 714 00:21:41,590 --> 00:21:43,899 it was multiple things that happened 715 00:21:43,900 --> 00:21:44,889 in a row. 716 00:21:44,890 --> 00:21:46,509 It was sort of bad in two thousand. 717 00:21:46,510 --> 00:21:48,519 And it got a lot bad in 2001 and it was 718 00:21:48,520 --> 00:21:49,629 bad for different sectors. 719 00:21:49,630 --> 00:21:51,039 There were all sorts of crazy things. 720 00:21:51,040 --> 00:21:53,079 The thing I really remember is Nortel 721 00:21:53,080 --> 00:21:54,080 stock. 722 00:21:54,700 --> 00:21:56,709 Apparently, if you had bought beer in 723 00:21:56,710 --> 00:21:58,839 Canada and save the cans for the 724 00:21:58,840 --> 00:22:00,789 collection like the deposit, you would 725 00:22:00,790 --> 00:22:02,199 have more money than if you had bought 726 00:22:02,200 --> 00:22:03,879 Nortel stock because it went to like 727 00:22:03,880 --> 00:22:04,880 almost nothing. 728 00:22:05,920 --> 00:22:07,239 So we ran out of money. We thought we 729 00:22:07,240 --> 00:22:08,319 could raise more money. We didn't have 730 00:22:08,320 --> 00:22:09,279 any more money. 731 00:22:09,280 --> 00:22:11,559 We were burning two 732 00:22:11,560 --> 00:22:13,359 big fifty five gallon drums of diesel 733 00:22:13,360 --> 00:22:15,369 fuel every day. We had a lot of staff, 734 00:22:15,370 --> 00:22:16,480 all sorts of crazy stuff like that. 735 00:22:17,560 --> 00:22:19,119 So we didn't we didn't have the ability 736 00:22:19,120 --> 00:22:20,619 to raise money from third parties. 737 00:22:20,620 --> 00:22:22,779 So we sort of refinanced. 738 00:22:22,780 --> 00:22:24,609 And we originally had a contract where we 739 00:22:24,610 --> 00:22:26,379 could buy all of Celan for about five 740 00:22:26,380 --> 00:22:28,539 million dollars in six to 12 741 00:22:28,540 --> 00:22:29,949 months after launch, thinking, oh, it'll 742 00:22:29,950 --> 00:22:32,379 be really easy to raise whatever number 743 00:22:32,380 --> 00:22:33,609 of billions of dollars. 744 00:22:33,610 --> 00:22:35,679 And today or in 1999, that would be 745 00:22:35,680 --> 00:22:37,299 true. Not true in two thousand twenty 746 00:22:37,300 --> 00:22:39,369 one. So and instead we ended up 747 00:22:39,370 --> 00:22:41,259 bringing the royal family of Sealand in 748 00:22:41,260 --> 00:22:42,489 as a partner. 749 00:22:42,490 --> 00:22:44,439 They had run this place since like 750 00:22:44,440 --> 00:22:46,269 nineteen sixty six. 751 00:22:46,270 --> 00:22:48,429 It was their main 752 00:22:48,430 --> 00:22:49,899 asset, their main pride and joy. 753 00:22:49,900 --> 00:22:50,919 They thought about a lot. 754 00:22:50,920 --> 00:22:52,749 So they thought about this entirely 755 00:22:52,750 --> 00:22:54,339 differently than we did. 756 00:22:54,340 --> 00:22:55,869 We were willing to push the limits on one 757 00:22:55,870 --> 00:22:58,059 data center in order to 758 00:22:58,060 --> 00:22:59,799 expand the model to things like that. 759 00:22:59,800 --> 00:23:02,019 They were much more conservative, 760 00:23:02,020 --> 00:23:04,089 which I can't really fault them for. 761 00:23:04,090 --> 00:23:06,009 And then we ran on a shoebox budget, my 762 00:23:06,010 --> 00:23:07,389 friend Abby, one of our investors. 763 00:23:07,390 --> 00:23:09,639 So he put in, I think, like a million 764 00:23:09,640 --> 00:23:10,719 dollars or so originally. 765 00:23:10,720 --> 00:23:12,189 And then every time I'd fly to Boston, 766 00:23:12,190 --> 00:23:13,599 he'd give me a bunch of cash, like a 767 00:23:13,600 --> 00:23:14,859 bunch of cash, like ten grand or 768 00:23:14,860 --> 00:23:16,509 something. And I would use that to pay 769 00:23:16,510 --> 00:23:18,189 bills. And we didn't really keep track of 770 00:23:18,190 --> 00:23:19,989 it. It was like the ultimate thing you 771 00:23:19,990 --> 00:23:21,399 don't do with investments of like 772 00:23:21,400 --> 00:23:22,959 throwing good money after bad. 773 00:23:22,960 --> 00:23:24,529 I just kept it running, which was awesome 774 00:23:24,530 --> 00:23:26,799 with him, but yeah, not a good financial 775 00:23:26,800 --> 00:23:27,800 decision 776 00:23:29,260 --> 00:23:29,679 then. 777 00:23:29,680 --> 00:23:32,049 So we basically run out of money. 778 00:23:32,050 --> 00:23:34,299 We're spending like 30, 50 K a week or so 779 00:23:34,300 --> 00:23:35,229 on renovations. 780 00:23:35,230 --> 00:23:36,699 And then we got down to the point where 781 00:23:36,700 --> 00:23:38,619 we had to pay for things from our server 782 00:23:38,620 --> 00:23:40,779 hosting and we had maybe 783 00:23:40,780 --> 00:23:42,999 like a grand left over every month. 784 00:23:43,000 --> 00:23:45,039 We had this food which was mostly left 785 00:23:45,040 --> 00:23:47,469 over from pre Haven, CODIS, 786 00:23:47,470 --> 00:23:48,729 and this was. 787 00:23:48,730 --> 00:23:51,009 Yeah, yeah, it was it was exciting 788 00:23:51,010 --> 00:23:52,629 of those things. There's some corned beef 789 00:23:52,630 --> 00:23:54,549 from Argentina that is actually the most 790 00:23:54,550 --> 00:23:56,379 wretched thing I have eaten. 791 00:23:56,380 --> 00:23:58,119 And we ate like most of the stuff 792 00:23:59,260 --> 00:24:01,239 in the place. 793 00:24:01,240 --> 00:24:03,579 So as far as servers go, 794 00:24:03,580 --> 00:24:05,289 these are a bunch of Celeron five thirty 795 00:24:05,290 --> 00:24:06,759 three boxes. 796 00:24:06,760 --> 00:24:07,760 We have this. 797 00:24:08,780 --> 00:24:10,669 I think fairly justifiable fear that if 798 00:24:10,670 --> 00:24:12,739 people saw how little infrastructure 799 00:24:12,740 --> 00:24:13,909 we had, nobody would actually buy 800 00:24:13,910 --> 00:24:15,289 anything. 801 00:24:15,290 --> 00:24:17,569 This was our showcase data room of what 802 00:24:17,570 --> 00:24:19,459 the rest of the rooms would look like. 803 00:24:19,460 --> 00:24:20,749 This was, in fact, the only one. 804 00:24:20,750 --> 00:24:22,909 So, yeah, 805 00:24:22,910 --> 00:24:24,049 it got a little bit more populated. 806 00:24:24,050 --> 00:24:26,119 But this is about like two thirds of the 807 00:24:26,120 --> 00:24:28,189 peak, very small number of servers, 808 00:24:28,190 --> 00:24:30,079 some UPSs. We've got a bigger GPS later, 809 00:24:30,080 --> 00:24:32,329 but all basically FreeBSD and Linux 810 00:24:32,330 --> 00:24:33,769 stuff. 811 00:24:33,770 --> 00:24:35,599 Pretty reasonable, but yeah, very, very 812 00:24:35,600 --> 00:24:36,619 small scale. 813 00:24:36,620 --> 00:24:38,299 And we never had more than eight megs of 814 00:24:38,300 --> 00:24:39,559 bandwidth going to the place. 815 00:24:39,560 --> 00:24:41,179 We had aggressive caching and then much 816 00:24:41,180 --> 00:24:42,469 more aggressive caching after I left, 817 00:24:42,470 --> 00:24:43,999 which is an interesting story. 818 00:24:44,000 --> 00:24:46,369 This is about peak capacity, 819 00:24:46,370 --> 00:24:47,370 so. 820 00:24:48,720 --> 00:24:51,149 Sitting in and the rooms are 20 821 00:24:51,150 --> 00:24:53,579 or they're like six meter cylinders, 822 00:24:53,580 --> 00:24:55,319 so you've got a circular set of desks 823 00:24:55,320 --> 00:24:56,699 around them in this room. 824 00:24:56,700 --> 00:24:58,349 And that was where I spent most of my 825 00:24:58,350 --> 00:25:00,299 time, mostly on Iasi, which is the main 826 00:25:00,300 --> 00:25:01,619 thing you do when you're on a little 827 00:25:01,620 --> 00:25:02,620 island like that. 828 00:25:04,150 --> 00:25:06,449 Yeah. So speaking of aggressive 829 00:25:06,450 --> 00:25:08,309 caching, so we have this whole model 830 00:25:08,310 --> 00:25:10,559 where we would host things where rather 831 00:25:10,560 --> 00:25:12,089 we would host things on sea and then we 832 00:25:12,090 --> 00:25:14,219 would have hearing sessions 833 00:25:14,220 --> 00:25:16,529 and transit purchased 834 00:25:16,530 --> 00:25:18,239 in places like London, Telehouse and then 835 00:25:18,240 --> 00:25:19,779 a place in New York, and then we bring 836 00:25:19,780 --> 00:25:20,789 the transport back. 837 00:25:20,790 --> 00:25:22,889 Those boxes are really tempting 838 00:25:22,890 --> 00:25:23,910 to use for caching 839 00:25:24,960 --> 00:25:26,759 after I left, which wasn't the most 840 00:25:26,760 --> 00:25:28,169 friendly of circumstances at the end of 841 00:25:28,170 --> 00:25:30,239 2002, I think they decided 842 00:25:30,240 --> 00:25:31,949 it would be a lot cheaper to just not 843 00:25:31,950 --> 00:25:33,419 have anything on the island anymore and 844 00:25:33,420 --> 00:25:34,889 put everything in those locations and not 845 00:25:34,890 --> 00:25:35,969 tell anyone. 846 00:25:35,970 --> 00:25:37,619 It was really obvious for two reasons. 847 00:25:37,620 --> 00:25:38,999 One, the pink times were zero 848 00:25:39,000 --> 00:25:41,079 milliseconds from the edge to this 849 00:25:41,080 --> 00:25:42,599 noise, about two milliseconds by speed of 850 00:25:42,600 --> 00:25:44,639 light. And then there was a huge fire 851 00:25:44,640 --> 00:25:46,379 because the original structure was all 852 00:25:46,380 --> 00:25:48,029 like Feresten and everything else. 853 00:25:48,030 --> 00:25:50,099 The new generator room, which 854 00:25:50,100 --> 00:25:52,289 they stored a bunch of oily rags in, 855 00:25:52,290 --> 00:25:53,549 was made of tarpaper. 856 00:25:53,550 --> 00:25:55,409 And the predictable thing happened after 857 00:25:55,410 --> 00:25:56,819 like five years. I don't know why it took 858 00:25:56,820 --> 00:25:58,319 five years, but it eventually completely 859 00:25:58,320 --> 00:26:00,449 caught on fire as a huge fire and nothing 860 00:26:00,450 --> 00:26:01,469 happened on the server. 861 00:26:01,470 --> 00:26:03,719 So, yeah, 862 00:26:03,720 --> 00:26:05,949 there's that. So given 863 00:26:05,950 --> 00:26:08,069 many, I would say the failure here 864 00:26:08,070 --> 00:26:09,899 was overdetermined. So we have lots of 865 00:26:09,900 --> 00:26:12,239 reasons why it failed and we need to 866 00:26:12,240 --> 00:26:14,189 piece apart why those things are some of 867 00:26:14,190 --> 00:26:16,259 them are totally idiosyncratic to a tiny 868 00:26:16,260 --> 00:26:18,509 platform. LNC those are less interesting 869 00:26:18,510 --> 00:26:19,949 unless you for some reason want to build 870 00:26:19,950 --> 00:26:21,119 something on a tiny platform. 871 00:26:21,120 --> 00:26:22,529 See, there's other things that are much 872 00:26:22,530 --> 00:26:25,169 more general as data haven issues, 873 00:26:25,170 --> 00:26:27,239 the core reasons where economics, the 874 00:26:27,240 --> 00:26:29,039 product itself wasn't all that great. 875 00:26:29,040 --> 00:26:30,569 And to be honest, we were not the 876 00:26:30,570 --> 00:26:31,769 greatest team. 877 00:26:31,770 --> 00:26:33,149 None of us had any experience running 878 00:26:33,150 --> 00:26:35,039 large businesses or even small businesses 879 00:26:35,040 --> 00:26:36,040 or really anything. 880 00:26:36,990 --> 00:26:39,029 And then we did have the market like it's 881 00:26:39,030 --> 00:26:40,559 really lame to blame the market for it, 882 00:26:40,560 --> 00:26:42,749 but I think the twenty two thousand one 883 00:26:42,750 --> 00:26:44,099 collapse in bandwidth prices was a 884 00:26:44,100 --> 00:26:45,659 fundamental driver. 885 00:26:45,660 --> 00:26:47,159 So what happened is we had an inherently 886 00:26:47,160 --> 00:26:49,169 high cost, which was totally fine in two 887 00:26:49,170 --> 00:26:50,249 thousand. 888 00:26:50,250 --> 00:26:51,479 There were two reasons for high cost. 889 00:26:51,480 --> 00:26:52,589 It was high cost because it was on this 890 00:26:52,590 --> 00:26:53,729 little tiny platform. 891 00:26:53,730 --> 00:26:55,049 It was also high cost because we didn't 892 00:26:55,050 --> 00:26:56,309 have any scale. When you're buying eight 893 00:26:56,310 --> 00:26:58,169 megabits of Internet and transporting it 894 00:26:58,170 --> 00:27:00,239 over, once you pay a pretty high cost 895 00:27:00,240 --> 00:27:02,249 per megabit, if you're buying one hundred 896 00:27:02,250 --> 00:27:03,329 and fifty five, which is what we'd 897 00:27:03,330 --> 00:27:05,519 originally ordered, the cost per megabit 898 00:27:05,520 --> 00:27:06,520 is a lot lower. 899 00:27:07,560 --> 00:27:09,209 And then we had the issue that in 900 00:27:09,210 --> 00:27:10,589 nineteen ninety nine two thousand, the 901 00:27:10,590 --> 00:27:12,720 price for I guess Akamai 902 00:27:13,770 --> 00:27:15,539 service was like two to three thousand 903 00:27:15,540 --> 00:27:16,679 dollars and megabits per second. 904 00:27:16,680 --> 00:27:18,749 Effectively transit was one to 905 00:27:18,750 --> 00:27:20,759 two K or something, a megabit. 906 00:27:20,760 --> 00:27:22,259 Then the market collapsed and people 907 00:27:22,260 --> 00:27:23,729 would start selling below their own cost 908 00:27:23,730 --> 00:27:24,929 because they had to. 909 00:27:24,930 --> 00:27:26,249 Sometimes they sell below their marginal 910 00:27:26,250 --> 00:27:27,449 cost, which is crazy, but usually 911 00:27:27,450 --> 00:27:28,529 bandwidth doesn't have a marginal cost. 912 00:27:30,150 --> 00:27:31,799 So basically, like the market price went 913 00:27:31,800 --> 00:27:33,479 to like ten dollars and our cost was five 914 00:27:33,480 --> 00:27:34,799 hundred and we built everything on like 915 00:27:34,800 --> 00:27:36,239 three thousand. So that was a serious 916 00:27:36,240 --> 00:27:38,639 problem. And then there was another more 917 00:27:38,640 --> 00:27:40,529 additionally serious problem is we are 918 00:27:40,530 --> 00:27:42,029 missing some key components to make this 919 00:27:42,030 --> 00:27:43,529 a great product. And then we didn't have 920 00:27:43,530 --> 00:27:45,659 any money to do anything differently 921 00:27:45,660 --> 00:27:47,429 because we only had only only eight megs. 922 00:27:47,430 --> 00:27:48,929 We had to really ration bandwidth. 923 00:27:48,930 --> 00:27:50,159 There's a lot of cool stuff you can do 924 00:27:50,160 --> 00:27:51,299 that you'd want to have more bandwidth 925 00:27:51,300 --> 00:27:53,639 for. This was also pre virtualization. 926 00:27:53,640 --> 00:27:56,009 So you had the physical servers and one 927 00:27:56,010 --> 00:27:57,779 server per customer was like a crazy 928 00:27:57,780 --> 00:27:58,780 thing. 929 00:27:59,130 --> 00:28:00,689 The biggest problem I would say is we had 930 00:28:00,690 --> 00:28:02,279 no way for our customers to handle 931 00:28:02,280 --> 00:28:04,469 payments. If you were a purely 932 00:28:04,470 --> 00:28:06,599 cipher space business, you still had 933 00:28:06,600 --> 00:28:08,729 to go incorporate somewhere to get a bank 934 00:28:08,730 --> 00:28:10,619 account to accept credit card processing. 935 00:28:10,620 --> 00:28:12,749 There was no Bitcoin back then, and 936 00:28:12,750 --> 00:28:14,459 I proposed the beginning of this thing 937 00:28:14,460 --> 00:28:15,659 that we fund building anonymous 938 00:28:15,660 --> 00:28:17,609 electronic cash as the first enabler for 939 00:28:17,610 --> 00:28:19,229 this thing. Because if you have anonymous 940 00:28:19,230 --> 00:28:20,939 electronic cash and a secure place to put 941 00:28:20,940 --> 00:28:22,409 your server, you don't actually need to 942 00:28:22,410 --> 00:28:23,549 incorporate anywhere. You can just have a 943 00:28:23,550 --> 00:28:25,319 key as your your thing. 944 00:28:25,320 --> 00:28:27,179 But we didn't do that. 945 00:28:27,180 --> 00:28:28,739 And then we never really found a single, 946 00:28:28,740 --> 00:28:30,419 really solid application for this thing. 947 00:28:30,420 --> 00:28:32,459 So all the things from doing start ups 948 00:28:32,460 --> 00:28:34,140 that I know you shouldn't do, we did. 949 00:28:35,160 --> 00:28:37,439 And then the team and structure issue. 950 00:28:37,440 --> 00:28:40,019 Fundamentally, the issue was that 951 00:28:40,020 --> 00:28:42,149 the ceiling people were more 952 00:28:42,150 --> 00:28:44,759 traditional, much, much more legally 953 00:28:44,760 --> 00:28:46,799 averse to risk uncertainty, things like 954 00:28:46,800 --> 00:28:48,929 that. And then there was me 955 00:28:48,930 --> 00:28:51,089 and I was very willing 956 00:28:51,090 --> 00:28:54,269 to push the limits on stuff because 957 00:28:54,270 --> 00:28:55,529 I just walk away from it if it failed. 958 00:28:55,530 --> 00:28:57,299 So like, I was trying to get to success. 959 00:28:57,300 --> 00:28:58,439 There's a whole model in like venture 960 00:28:58,440 --> 00:29:00,359 capital where it's OK to fail like one 961 00:29:00,360 --> 00:29:01,859 hundred times if you're one success is 962 00:29:01,860 --> 00:29:04,079 like a ten thousand times bigger 963 00:29:04,080 --> 00:29:05,080 return. 964 00:29:05,550 --> 00:29:07,049 So we were pushing for that kind of 965 00:29:07,050 --> 00:29:08,669 thing. And that isn't the kind of thing 966 00:29:08,670 --> 00:29:09,809 you do if it's your your house, 967 00:29:09,810 --> 00:29:10,810 basically. 968 00:29:11,660 --> 00:29:13,649 Then we had lots of internal team issues, 969 00:29:13,650 --> 00:29:14,939 politics like that. And also it's just 970 00:29:14,940 --> 00:29:16,499 boring. Like this place was like five or 971 00:29:16,500 --> 00:29:17,519 ten thousand square feet. 972 00:29:17,520 --> 00:29:19,469 It was kind of cool when I got to leave 973 00:29:19,470 --> 00:29:20,879 every week or two. But there was a period 974 00:29:20,880 --> 00:29:22,199 where I was out there for six months 975 00:29:22,200 --> 00:29:23,279 because I didn't have any money to go 976 00:29:23,280 --> 00:29:25,019 anywhere else. So I was basically living 977 00:29:25,020 --> 00:29:27,569 on a tiny little one, 978 00:29:27,570 --> 00:29:28,769 tiny little platform. 979 00:29:28,770 --> 00:29:30,869 There was one other person there who was 980 00:29:30,870 --> 00:29:32,609 like a security guard, like a six year 981 00:29:32,610 --> 00:29:33,959 old British guy, a security guard. 982 00:29:33,960 --> 00:29:36,149 And I arranged to have like an offset 983 00:29:36,150 --> 00:29:37,349 by 12 hour shift from him. 984 00:29:37,350 --> 00:29:38,819 So I would not actually see another 985 00:29:38,820 --> 00:29:40,139 person for like three months at a time 986 00:29:40,140 --> 00:29:42,359 was it was probably not the most 987 00:29:42,360 --> 00:29:44,489 psychologically awesome thing to do, 988 00:29:44,490 --> 00:29:46,289 but I had irises, so that totally made it 989 00:29:46,290 --> 00:29:47,519 better. 990 00:29:47,520 --> 00:29:49,109 Yeah. So there's things that you would 991 00:29:49,110 --> 00:29:50,819 expect from building a data haven that 992 00:29:50,820 --> 00:29:52,169 would be the reasons for failure, and 993 00:29:52,170 --> 00:29:53,219 they actually, in our case, were not 994 00:29:53,220 --> 00:29:54,749 reasons for failure. We don't know that 995 00:29:54,750 --> 00:29:56,219 these things are not reasons for failure 996 00:29:56,220 --> 00:29:57,449 in general, but they were not our reasons 997 00:29:57,450 --> 00:29:58,679 for failure. Maybe we didn't get to them. 998 00:29:58,680 --> 00:29:59,609 Maybe we were lucky. 999 00:29:59,610 --> 00:30:01,919 So legal or regulatory pressure never 1000 00:30:01,920 --> 00:30:03,070 actually was an issue at all. 1001 00:30:04,320 --> 00:30:06,689 We got some very, very 1002 00:30:06,690 --> 00:30:09,149 cursory legal threats type things, 1003 00:30:09,150 --> 00:30:11,129 but they were mostly from civil things. 1004 00:30:11,130 --> 00:30:12,309 We didn't do anything really bad. 1005 00:30:12,310 --> 00:30:14,489 We had a very correctly chosen 1006 00:30:14,490 --> 00:30:15,569 for their risk model. 1007 00:30:15,570 --> 00:30:17,189 We had acceptable use policy. 1008 00:30:17,190 --> 00:30:19,499 So no spam, no child porn, 1009 00:30:19,500 --> 00:30:21,839 no hacking other people's servers. 1010 00:30:21,840 --> 00:30:23,999 And I think 1011 00:30:24,000 --> 00:30:25,859 we added no terrorism on 9/11. 1012 00:30:25,860 --> 00:30:27,629 It wasn't really I thought before that. 1013 00:30:27,630 --> 00:30:29,099 And that was like the entirety of our 1014 00:30:29,100 --> 00:30:30,179 acceptable use policy. 1015 00:30:30,180 --> 00:30:31,439 We were sort of in the gray area about 1016 00:30:31,440 --> 00:30:33,059 copyright. We also had the benefit that 1017 00:30:33,060 --> 00:30:34,919 our cost was so high for servers that you 1018 00:30:34,920 --> 00:30:36,479 would not be able to put a file sharing 1019 00:30:36,480 --> 00:30:38,279 server on it and have it be profitable. 1020 00:30:38,280 --> 00:30:39,929 If we'd had much lower costs, that would 1021 00:30:39,930 --> 00:30:41,489 have been an issue. 1022 00:30:41,490 --> 00:30:43,019 There was no real competition for a 1023 00:30:43,020 --> 00:30:44,489 physical data haven at the time. 1024 00:30:44,490 --> 00:30:45,959 There were secure facilities, but there 1025 00:30:45,960 --> 00:30:48,419 was no data haven as such, 1026 00:30:48,420 --> 00:30:49,799 and no one had a great software 1027 00:30:49,800 --> 00:30:51,569 replacement for this kind of thing. 1028 00:30:51,570 --> 00:30:53,699 No one hacked us. As far as I know, we 1029 00:30:53,700 --> 00:30:55,529 had very little infrastructure ourselves, 1030 00:30:55,530 --> 00:30:57,599 so it's unlikely that that got hacked. 1031 00:30:57,600 --> 00:30:58,829 Some of our customers might have gotten 1032 00:30:58,830 --> 00:30:59,849 hacked, but we wouldn't know. 1033 00:30:59,850 --> 00:31:01,079 But there was no like major hacking 1034 00:31:01,080 --> 00:31:02,399 incident or anything and there were no 1035 00:31:02,400 --> 00:31:04,529 fundamental technical issues here. 1036 00:31:04,530 --> 00:31:06,029 Getting bandwidth out to this place would 1037 00:31:06,030 --> 00:31:07,889 not have been that hard with like five or 1038 00:31:07,890 --> 00:31:09,419 ten million dollars versus two million 1039 00:31:09,420 --> 00:31:11,609 dollars, lots 1040 00:31:11,610 --> 00:31:12,629 of things like that. So these weren't really 1041 00:31:12,630 --> 00:31:13,889 recent failure. But we don't know if 1042 00:31:13,890 --> 00:31:15,149 there's these are the things that I'm 1043 00:31:15,150 --> 00:31:16,709 much more concerned about are is there an 1044 00:31:16,710 --> 00:31:18,479 actual demand for data center, a data 1045 00:31:18,480 --> 00:31:19,529 haven type services? 1046 00:31:19,530 --> 00:31:20,759 I think there is, but we don't really 1047 00:31:20,760 --> 00:31:22,679 know for sure how can you make a viable 1048 00:31:22,680 --> 00:31:23,969 product that people will actually be able 1049 00:31:23,970 --> 00:31:25,439 to pay for? Like there's plenty of people 1050 00:31:25,440 --> 00:31:26,759 who would love to have a server that 1051 00:31:26,760 --> 00:31:28,169 could never be shut off. 1052 00:31:28,170 --> 00:31:29,849 But are they people who will be able to 1053 00:31:29,850 --> 00:31:31,139 pay for that service at the cost 1054 00:31:31,140 --> 00:31:33,419 required? And then the biggest 1055 00:31:33,420 --> 00:31:35,519 problem with this whole thing is it's 1056 00:31:35,520 --> 00:31:38,099 really easy to have like one or two 1057 00:31:38,100 --> 00:31:40,649 below the radar. 1058 00:31:40,650 --> 00:31:42,179 If you got some services that don't 1059 00:31:42,180 --> 00:31:43,529 attract any negative attention, you're 1060 00:31:43,530 --> 00:31:45,659 fine until you have like 1061 00:31:45,660 --> 00:31:47,489 Lavabit or something hosted on you and 1062 00:31:47,490 --> 00:31:49,169 then legal stuff happens. 1063 00:31:49,170 --> 00:31:50,579 So basically, the more successful you 1064 00:31:50,580 --> 00:31:52,829 get, the higher the odds of some horrible 1065 00:31:52,830 --> 00:31:54,479 incident happening and then having to 1066 00:31:54,480 --> 00:31:56,489 resist it, which is sort of the opposite 1067 00:31:56,490 --> 00:31:58,049 of a lot of other models where the bigger 1068 00:31:58,050 --> 00:31:59,159 you get, the easier it is to be 1069 00:31:59,160 --> 00:32:00,299 successful. So that's that's a 1070 00:32:00,300 --> 00:32:01,799 fundamental thing. And there's a question 1071 00:32:01,800 --> 00:32:03,539 of can you do this better in software if 1072 00:32:03,540 --> 00:32:04,859 you can ever do something in software 1073 00:32:04,860 --> 00:32:06,659 rather than spending money on platforms 1074 00:32:06,660 --> 00:32:07,859 and offshore data havens, it's totally 1075 00:32:07,860 --> 00:32:08,459 worth it. 1076 00:32:08,460 --> 00:32:10,409 And then the other question is, even if 1077 00:32:10,410 --> 00:32:12,449 you can do all the stuff, should we do it 1078 00:32:12,450 --> 00:32:14,249 like not everything you can do should be 1079 00:32:14,250 --> 00:32:15,659 done by people. 1080 00:32:15,660 --> 00:32:18,839 I think free speech trumps the other 1081 00:32:18,840 --> 00:32:20,129 disincentives to doing those things. 1082 00:32:20,130 --> 00:32:22,139 I think data havens should exist, but it 1083 00:32:22,140 --> 00:32:23,399 is an open question and people have 1084 00:32:23,400 --> 00:32:25,199 different opinions on that. 1085 00:32:25,200 --> 00:32:26,609 So we're not the only data here. 1086 00:32:26,610 --> 00:32:27,749 And there have been data havens since 1087 00:32:27,750 --> 00:32:29,969 then. And based 1088 00:32:29,970 --> 00:32:31,829 on the earlier definition, there's 1089 00:32:31,830 --> 00:32:32,879 different ones. So I would say that a 1090 00:32:32,880 --> 00:32:34,979 conventional great data center, 1091 00:32:34,980 --> 00:32:37,769 like a Tier one or whatever facility is 1092 00:32:37,770 --> 00:32:39,239 a data haven in terms of keeping your 1093 00:32:39,240 --> 00:32:40,259 servers available. 1094 00:32:40,260 --> 00:32:41,519 You're subject to the laws in whatever 1095 00:32:41,520 --> 00:32:42,419 country you are. 1096 00:32:42,420 --> 00:32:44,339 And the nice thing is, like the US has 1097 00:32:44,340 --> 00:32:46,209 great laws for certain things. 1098 00:32:46,210 --> 00:32:48,359 Ireland, Germany have great laws for 1099 00:32:48,360 --> 00:32:49,799 other things and you can pick and choose 1100 00:32:49,800 --> 00:32:50,969 your application. 1101 00:32:50,970 --> 00:32:52,619 There's a company called the Bunker and 1102 00:32:52,620 --> 00:32:53,819 there's a couple other things where 1103 00:32:53,820 --> 00:32:56,099 they've taken World War 1104 00:32:56,100 --> 00:32:58,259 Cold War bunkers and refitted them 1105 00:32:58,260 --> 00:32:59,459 as data centers. 1106 00:32:59,460 --> 00:33:01,259 They're in the countries that have 1107 00:33:01,260 --> 00:33:03,239 certain laws, so they'll have either UK 1108 00:33:03,240 --> 00:33:05,309 law or Swedish law 1109 00:33:05,310 --> 00:33:07,019 or Swiss law or the things like that. 1110 00:33:07,020 --> 00:33:09,239 But they're nice facilities. 1111 00:33:09,240 --> 00:33:11,819 They're generally have weaker network 1112 00:33:11,820 --> 00:33:13,979 connectivity than the center of town 1113 00:33:13,980 --> 00:33:15,479 data centers like Telehouse Wood. 1114 00:33:15,480 --> 00:33:17,369 So there's a trade off there. 1115 00:33:17,370 --> 00:33:18,809 There's a bulletproof hosting model. 1116 00:33:18,810 --> 00:33:20,219 So there's this whole world of the other. 1117 00:33:20,220 --> 00:33:21,220 All we had is no spam. 1118 00:33:22,290 --> 00:33:24,599 Speaking of bulletproof hosting, the 1119 00:33:24,600 --> 00:33:27,119 response, you URLs that people 1120 00:33:27,120 --> 00:33:29,249 go to from Spam get shut 1121 00:33:29,250 --> 00:33:30,419 down by people all the time. 1122 00:33:30,420 --> 00:33:32,099 Malware, you URLs get shut down all the 1123 00:33:32,100 --> 00:33:33,509 time. There's this concept of a 1124 00:33:33,510 --> 00:33:35,489 bulletproof host that will stay up 1125 00:33:35,490 --> 00:33:36,539 against this kind of attack. 1126 00:33:36,540 --> 00:33:38,459 Usually the term bulletproof host is used 1127 00:33:38,460 --> 00:33:39,959 specifically around this kind of like 1128 00:33:39,960 --> 00:33:42,239 nuisance type stuff versus free speech. 1129 00:33:42,240 --> 00:33:43,619 But you could use it in any way. 1130 00:33:43,620 --> 00:33:44,789 And there's a famous company in the 1131 00:33:44,790 --> 00:33:46,829 Netherlands that was rated has some crazy 1132 00:33:46,830 --> 00:33:48,149 stuff. And then there's distributed 1133 00:33:48,150 --> 00:33:49,959 software systems and there's application 1134 00:33:49,960 --> 00:33:51,299 specific systems that are designed around 1135 00:33:51,300 --> 00:33:52,589 redundancy. 1136 00:33:52,590 --> 00:33:54,659 So, yeah, this is a 1137 00:33:54,660 --> 00:33:56,969 two billion dollar data center in 1138 00:33:56,970 --> 00:33:58,919 downtown San Francisco, which is later 1139 00:33:58,920 --> 00:33:59,969 got sold for like twenty five million 1140 00:33:59,970 --> 00:34:01,169 dollars or something. There were the 1141 00:34:01,170 --> 00:34:02,999 generators going out. 1142 00:34:03,000 --> 00:34:04,659 This is an awesome facility. 1143 00:34:04,660 --> 00:34:06,749 Punin Bonlac Data Center, which I've 1144 00:34:06,750 --> 00:34:08,099 never been to, but I'd love to get a 1145 00:34:08,100 --> 00:34:09,100 server there someday. 1146 00:34:10,170 --> 00:34:12,359 Telehouse, the probably the 1147 00:34:12,360 --> 00:34:14,579 premier Internet Interconnection Center 1148 00:34:14,580 --> 00:34:16,829 and all of Europe, which is also 1149 00:34:16,830 --> 00:34:19,229 great. I used to live like five 1150 00:34:19,230 --> 00:34:20,459 blocks away from that place, which I 1151 00:34:20,460 --> 00:34:21,569 thought would be awesome. But it's actually 1152 00:34:21,570 --> 00:34:22,919 more of a pain to move a server five 1153 00:34:22,920 --> 00:34:24,359 blocks than it is to move it like across 1154 00:34:24,360 --> 00:34:25,829 town, because you feel bad about taking a 1155 00:34:25,830 --> 00:34:27,209 taxi that distance and you can't really 1156 00:34:27,210 --> 00:34:29,369 carry it. So that was not a decision. 1157 00:34:29,370 --> 00:34:31,019 And this is the bunker. The cyber bunker, 1158 00:34:31,020 --> 00:34:32,069 sorry, the cyber bunker. 1159 00:34:33,270 --> 00:34:34,289 So there have been successes. 1160 00:34:34,290 --> 00:34:36,689 So PAVLENKO not a complete 1161 00:34:36,690 --> 00:34:38,579 failure. We did accomplish the goal of 1162 00:34:38,580 --> 00:34:41,039 making things popular ised 1163 00:34:41,040 --> 00:34:42,509 and things like that. But it was not a 1164 00:34:42,510 --> 00:34:43,529 commercial success in any way. 1165 00:34:43,530 --> 00:34:44,968 I lost like a quarter of a million 1166 00:34:44,969 --> 00:34:47,099 dollars by spending a bunch of money not 1167 00:34:47,100 --> 00:34:48,229 getting reimbursed on credit card. 1168 00:34:48,230 --> 00:34:49,230 Which is not awesome, 1169 00:34:50,300 --> 00:34:51,349 but there have been people who have been 1170 00:34:51,350 --> 00:34:53,419 successful Bitcoin, as 1171 00:34:53,420 --> 00:34:55,099 far as we can tell so far, like it's not 1172 00:34:55,100 --> 00:34:56,689 the finished book, but like it's been a 1173 00:34:56,690 --> 00:34:58,459 pretty successful Pirate Bay has been 1174 00:34:58,460 --> 00:35:00,649 successful, sort of like they've moved 1175 00:35:00,650 --> 00:35:01,699 their servers around and have been 1176 00:35:01,700 --> 00:35:02,839 relatively successful. 1177 00:35:02,840 --> 00:35:04,789 BitTorrent has been incredibly successful 1178 00:35:04,790 --> 00:35:07,039 at keeping things 1179 00:35:07,040 --> 00:35:08,479 available, but has not been commercially 1180 00:35:08,480 --> 00:35:09,559 terribly successful. 1181 00:35:09,560 --> 00:35:11,569 WikiLeaks has remained online despite 1182 00:35:11,570 --> 00:35:13,429 doing stuff that the most powerful 1183 00:35:13,430 --> 00:35:15,019 governments in the world don't like. 1184 00:35:15,020 --> 00:35:16,909 And the really exciting thing they did 1185 00:35:16,910 --> 00:35:19,069 was it's insurance file concept. 1186 00:35:19,070 --> 00:35:20,689 Silk Road was I guess there's an 1187 00:35:20,690 --> 00:35:22,699 asterisk. So was basically successful 1188 00:35:22,700 --> 00:35:24,979 technically, except for some user 1189 00:35:24,980 --> 00:35:26,809 admin issues or something. 1190 00:35:26,810 --> 00:35:28,309 And Tor has been pretty successful, 1191 00:35:28,310 --> 00:35:29,449 especially hidden services, which are 1192 00:35:29,450 --> 00:35:30,450 valid in this case. 1193 00:35:31,520 --> 00:35:33,439 So really, if we so we don't want to just 1194 00:35:33,440 --> 00:35:35,329 like blindly do the same thing again, we 1195 00:35:35,330 --> 00:35:37,099 want to figure out what went wrong and 1196 00:35:37,100 --> 00:35:38,239 fix it. 1197 00:35:38,240 --> 00:35:39,289 So how not to fail. 1198 00:35:39,290 --> 00:35:40,699 You want to think about your application 1199 00:35:40,700 --> 00:35:42,199 model for the technical side of it. 1200 00:35:42,200 --> 00:35:43,489 You want to think about who your threats 1201 00:35:43,490 --> 00:35:44,490 are and 1202 00:35:45,590 --> 00:35:47,119 adversaries. You definitely need to think 1203 00:35:47,120 --> 00:35:49,039 about your business model and useful 1204 00:35:49,040 --> 00:35:51,499 technologies that can help you. 1205 00:35:51,500 --> 00:35:53,989 So as far as application, you want to do 1206 00:35:53,990 --> 00:35:56,119 as little hard stuff as possible 1207 00:35:56,120 --> 00:35:58,459 ever. You always want to do easy stuff 1208 00:35:58,460 --> 00:36:00,499 and you want to make sure it's like the I 1209 00:36:00,500 --> 00:36:01,669 guess I could use a zip quote or 1210 00:36:01,670 --> 00:36:02,689 something, but you always want to meet 1211 00:36:02,690 --> 00:36:04,609 the enemy on the ground where you 1212 00:36:04,610 --> 00:36:05,959 control, but your advantage, you don't 1213 00:36:05,960 --> 00:36:07,309 want to fight in your weak spot. 1214 00:36:07,310 --> 00:36:09,529 So do things that are easy, like 1215 00:36:09,530 --> 00:36:11,749 static data is pretty easy to keep 1216 00:36:11,750 --> 00:36:12,859 resistent from censorship. 1217 00:36:12,860 --> 00:36:14,419 You just make lots of copies and is also 1218 00:36:14,420 --> 00:36:16,969 resistant against accidental deletion, 1219 00:36:16,970 --> 00:36:18,979 whatever. So if you have a lot of copies, 1220 00:36:18,980 --> 00:36:20,389 the cool thing with the insurance file 1221 00:36:20,390 --> 00:36:22,789 is, is distributing widely and encrypted 1222 00:36:22,790 --> 00:36:24,409 file and then distributing the small key 1223 00:36:24,410 --> 00:36:26,449 later because basically they won't censor 1224 00:36:26,450 --> 00:36:27,559 it because they can't go back in time. 1225 00:36:27,560 --> 00:36:29,419 So it's kind of awesome. 1226 00:36:29,420 --> 00:36:31,039 You end up with a lot of general 1227 00:36:31,040 --> 00:36:32,359 distributed systems problems. 1228 00:36:33,470 --> 00:36:34,609 Do you need to have immediate 1229 00:36:34,610 --> 00:36:36,349 consistency, some sort of global lock? 1230 00:36:36,350 --> 00:36:37,819 Can you do eventual consistency? 1231 00:36:37,820 --> 00:36:39,259 All these things that like computer 1232 00:36:39,260 --> 00:36:40,549 scientists deal with and then Web 1233 00:36:40,550 --> 00:36:41,989 application developers and application 1234 00:36:41,990 --> 00:36:43,879 developers deal with are super relevant 1235 00:36:43,880 --> 00:36:44,880 in the data haven world. 1236 00:36:45,860 --> 00:36:47,929 And then there's the hardest 1237 00:36:47,930 --> 00:36:49,579 thing possible is to build a 1238 00:36:50,630 --> 00:36:52,429 legacy so you can't build a custom 1239 00:36:52,430 --> 00:36:53,599 client. You have to use like a regular 1240 00:36:53,600 --> 00:36:55,339 Web browser, globally synchronized 1241 00:36:55,340 --> 00:36:56,239 transaction system. 1242 00:36:56,240 --> 00:36:57,499 That is the hardest thing to build. 1243 00:36:57,500 --> 00:36:58,639 So if you can build a data haven that 1244 00:36:58,640 --> 00:37:00,709 will work with that model, you have one 1245 00:37:00,710 --> 00:37:02,839 maybe threats and 1246 00:37:02,840 --> 00:37:04,909 adversaries depending on who your 1247 00:37:04,910 --> 00:37:06,349 threats are. There's a lot of techniques 1248 00:37:06,350 --> 00:37:07,339 you can do and they're different against 1249 00:37:07,340 --> 00:37:09,799 different people. If you're worried about 1250 00:37:09,800 --> 00:37:11,539 a government in the Middle East, 1251 00:37:11,540 --> 00:37:13,039 government in Africa or something else 1252 00:37:13,040 --> 00:37:15,349 going after your community, your 1253 00:37:15,350 --> 00:37:16,699 global diaspora community for human 1254 00:37:16,700 --> 00:37:18,349 rights stuff, put your servers in a place 1255 00:37:18,350 --> 00:37:19,639 like Germany or in the United States 1256 00:37:19,640 --> 00:37:21,769 because those governments will be happy 1257 00:37:21,770 --> 00:37:23,899 to stand up to a dictator, at 1258 00:37:23,900 --> 00:37:25,759 least as far as not giving them the 1259 00:37:25,760 --> 00:37:26,760 copies of your data. 1260 00:37:27,920 --> 00:37:29,749 You can split up across jurisdictions. 1261 00:37:29,750 --> 00:37:31,879 I've seen as I did the work in Iraq and 1262 00:37:31,880 --> 00:37:33,049 Afghanistan, just doing like satellite 1263 00:37:33,050 --> 00:37:34,579 Internet for people. But I got to meet 1264 00:37:34,580 --> 00:37:35,809 some of the law enforcement people and 1265 00:37:35,810 --> 00:37:37,009 see the trouble they go through when 1266 00:37:37,010 --> 00:37:38,299 they're trying to deal with servers. 1267 00:37:38,300 --> 00:37:39,859 They're in multiple jurisdictions and a 1268 00:37:39,860 --> 00:37:41,179 lot of cases they don't even bother if 1269 00:37:41,180 --> 00:37:42,439 it's a particularly difficult 1270 00:37:42,440 --> 00:37:44,659 jurisdiction like, say, Eastern Europe. 1271 00:37:44,660 --> 00:37:46,069 But if you can make that so it's like the 1272 00:37:46,070 --> 00:37:47,389 Russian dolls problem of multiple 1273 00:37:47,390 --> 00:37:50,269 servers, they get bored pretty fast 1274 00:37:50,270 --> 00:37:52,489 and you can use disposable finance, 1275 00:37:52,490 --> 00:37:53,539 which has been sort of the WikiLeaks 1276 00:37:53,540 --> 00:37:55,729 model where the pieces that you have 1277 00:37:55,730 --> 00:37:57,229 the most of and that you can easily 1278 00:37:57,230 --> 00:37:59,629 replace, you make those 1279 00:37:59,630 --> 00:38:01,069 pop up. Those are the only things that 1280 00:38:01,070 --> 00:38:02,749 are exposed. And then you have servers 1281 00:38:02,750 --> 00:38:04,489 and back end your big data repository 1282 00:38:04,490 --> 00:38:05,989 processing. That's much harder to 1283 00:38:05,990 --> 00:38:08,149 replace. You keep shielded and then you 1284 00:38:08,150 --> 00:38:09,989 minimize the bulletproof computing base. 1285 00:38:09,990 --> 00:38:11,179 So it's sort of the inverse of a trusted 1286 00:38:11,180 --> 00:38:12,409 computing base or sort of related to 1287 00:38:12,410 --> 00:38:14,479 trusting, committed base where the parts 1288 00:38:14,480 --> 00:38:16,429 that actually need to be resilient 1289 00:38:16,430 --> 00:38:17,689 against all these attacks should be as 1290 00:38:17,690 --> 00:38:18,949 small as possible for application 1291 00:38:20,630 --> 00:38:21,589 laws and politics. 1292 00:38:21,590 --> 00:38:22,489 These always change. 1293 00:38:22,490 --> 00:38:23,929 I know much more about technology than I 1294 00:38:23,930 --> 00:38:26,059 do about laws, but the issue we 1295 00:38:26,060 --> 00:38:27,499 had in the nineties was that law wasn't 1296 00:38:27,500 --> 00:38:29,479 really settled. We had both. 1297 00:38:29,480 --> 00:38:31,669 I mean, there's a concept, at least in 1298 00:38:31,670 --> 00:38:33,139 common law, countries like Black Letter 1299 00:38:33,140 --> 00:38:35,239 Law and then court law or case 1300 00:38:35,240 --> 00:38:36,979 law. There were lots of cases where there 1301 00:38:36,980 --> 00:38:38,749 was neither black letter law nor case 1302 00:38:38,750 --> 00:38:40,459 law. Now, at least there's case law and a 1303 00:38:40,460 --> 00:38:41,779 lot of cases. There's black letter law 1304 00:38:41,780 --> 00:38:44,089 about it, but 1305 00:38:44,090 --> 00:38:45,649 there wasn't a lot of that stuff back in 1306 00:38:45,650 --> 00:38:47,839 the day then. 1307 00:38:47,840 --> 00:38:49,909 Well, Saelens was still going on during 1308 00:38:49,910 --> 00:38:52,129 terrorism. I woke up, I was like 1309 00:38:52,130 --> 00:38:53,599 asleep. I'd gotten like a weird sleep 1310 00:38:53,600 --> 00:38:55,819 schedule. So I woke up at local 1311 00:38:55,820 --> 00:38:57,469 three p.m. or so, which was like an hour 1312 00:38:57,470 --> 00:38:59,659 after the 9/11 World Trade 1313 00:38:59,660 --> 00:39:01,879 Center attacks, woke up, saw the 1314 00:39:01,880 --> 00:39:03,439 TV of, like, these things crashing in and 1315 00:39:03,440 --> 00:39:06,049 thought one that sucks and too 1316 00:39:06,050 --> 00:39:08,089 well, they're going to completely ruin 1317 00:39:08,090 --> 00:39:09,719 like any possibility of doing offshore 1318 00:39:09,720 --> 00:39:10,639 data and stuff. 1319 00:39:10,640 --> 00:39:12,409 So I set up anonymous remailer later that 1320 00:39:12,410 --> 00:39:13,879 day and a bunch of other stuff, like push 1321 00:39:13,880 --> 00:39:15,049 the limits a little bit more. 1322 00:39:15,050 --> 00:39:16,969 But it was pretty clear that terrorism 1323 00:39:16,970 --> 00:39:19,039 was going to get used to beat down 1324 00:39:19,040 --> 00:39:20,839 any form of anonymity, even for things 1325 00:39:20,840 --> 00:39:22,969 that are totally unrelated to terrorism. 1326 00:39:22,970 --> 00:39:24,679 In the nineties, we had the four horsemen 1327 00:39:24,680 --> 00:39:27,349 of child porn, money laundering, 1328 00:39:27,350 --> 00:39:29,359 terrorism and tax evasion. 1329 00:39:29,360 --> 00:39:31,459 And it's a lot easier to scare people 1330 00:39:31,460 --> 00:39:32,959 with terrorism than it is to scare people 1331 00:39:32,960 --> 00:39:34,669 with like evil tax evaders or something. 1332 00:39:34,670 --> 00:39:35,670 So 1333 00:39:36,920 --> 00:39:38,179 so we were afraid of that. 1334 00:39:38,180 --> 00:39:39,139 The other thing is you want to have a 1335 00:39:39,140 --> 00:39:40,969 preemptive, positive legal campaign if 1336 00:39:40,970 --> 00:39:42,559 you are behind the curve and the first 1337 00:39:42,560 --> 00:39:43,849 thing anyone hears about you is something 1338 00:39:43,850 --> 00:39:45,169 negative, like you're used for child 1339 00:39:45,170 --> 00:39:47,329 porn. You have lost no matter if the law 1340 00:39:47,330 --> 00:39:48,939 on your side. They'll just change the law 1341 00:39:48,940 --> 00:39:50,549 or they'll change the interpretation or 1342 00:39:50,550 --> 00:39:52,449 your funding pool or something like that, 1343 00:39:52,450 --> 00:39:53,609 you want to make sure that you've got 1344 00:39:53,610 --> 00:39:55,439 good stories out there of how you're 1345 00:39:55,440 --> 00:39:57,539 helping people 1346 00:39:57,540 --> 00:39:59,979 escape horrible situations, 1347 00:39:59,980 --> 00:40:01,529 things like that first. 1348 00:40:01,530 --> 00:40:02,879 And then you pick some cases that have 1349 00:40:02,880 --> 00:40:03,899 really good, good optics. 1350 00:40:03,900 --> 00:40:05,729 So you help open source projects. 1351 00:40:05,730 --> 00:40:07,859 You help like the overseas diaspora 1352 00:40:07,860 --> 00:40:09,089 of a country where the country is 1353 00:40:09,090 --> 00:40:10,649 monitoring those people's communications, 1354 00:40:10,650 --> 00:40:12,389 things like that. The other key thing, 1355 00:40:12,390 --> 00:40:13,979 which I think is these are these are all 1356 00:40:13,980 --> 00:40:14,999 fairly obvious. 1357 00:40:15,000 --> 00:40:16,889 That one that I thought was the most 1358 00:40:16,890 --> 00:40:19,409 useful is pick one known 1359 00:40:19,410 --> 00:40:21,479 main adversary, one threat, and make sure 1360 00:40:21,480 --> 00:40:23,609 you can defeat that. So in our case, we 1361 00:40:23,610 --> 00:40:25,169 could pick something like gambling and we 1362 00:40:25,170 --> 00:40:26,849 could be the the awesome place where you 1363 00:40:26,850 --> 00:40:28,229 can host your gambling servers. 1364 00:40:28,230 --> 00:40:29,999 No one really cared about gambling except 1365 00:40:30,000 --> 00:40:31,469 for the US government. And there were 1366 00:40:31,470 --> 00:40:32,669 five thousand miles away from us. 1367 00:40:32,670 --> 00:40:34,379 So we didn't really have to worry about 1368 00:40:34,380 --> 00:40:36,029 them very much. We would not have picked 1369 00:40:36,030 --> 00:40:38,159 something that a lot of countries hated 1370 00:40:38,160 --> 00:40:40,079 or that the British specifically hated 1371 00:40:40,080 --> 00:40:41,609 because they're much closer to us. 1372 00:40:41,610 --> 00:40:43,709 So pick one thing to do and just 1373 00:40:43,710 --> 00:40:45,899 do that thing rather than like and taking 1374 00:40:45,900 --> 00:40:47,189 on every every other country in the 1375 00:40:47,190 --> 00:40:48,190 world. 1376 00:40:48,930 --> 00:40:50,399 Then there's an open question about 1377 00:40:50,400 --> 00:40:51,689 business model. You really want to have a 1378 00:40:51,690 --> 00:40:53,369 working business model before you scale 1379 00:40:53,370 --> 00:40:55,559 up your business. So keep your costs 1380 00:40:55,560 --> 00:40:56,849 low and building something like this so 1381 00:40:56,850 --> 00:40:59,069 you can pick a model. 1382 00:40:59,070 --> 00:41:01,229 A lot of the problems are the interesting 1383 00:41:01,230 --> 00:41:03,779 customers are usually not really 1384 00:41:03,780 --> 00:41:05,189 able to pay very much. 1385 00:41:05,190 --> 00:41:06,899 The boring customers are slow to move. 1386 00:41:06,900 --> 00:41:08,519 So you can get interesting customers with 1387 00:41:08,520 --> 00:41:10,649 not a lot of money early on, but 1388 00:41:10,650 --> 00:41:12,509 you can't really serve them unless you 1389 00:41:12,510 --> 00:41:14,609 either have a lot of money or have very 1390 00:41:14,610 --> 00:41:16,829 low costs. And there's that. 1391 00:41:16,830 --> 00:41:18,749 The non-intuitive thing is to build a 1392 00:41:18,750 --> 00:41:20,729 system that works really, really, really 1393 00:41:20,730 --> 00:41:22,709 well for one specific application. 1394 00:41:22,710 --> 00:41:24,149 So you could build like anonymous 1395 00:41:24,150 --> 00:41:26,460 retailers or one of the most secure, most 1396 00:41:27,750 --> 00:41:29,729 effective against very high level threat 1397 00:41:29,730 --> 00:41:31,289 systems out there because they only dealt 1398 00:41:31,290 --> 00:41:33,269 with email, whereas TOR has a much harder 1399 00:41:33,270 --> 00:41:34,739 problem because it deals with arbitrary 1400 00:41:34,740 --> 00:41:36,809 protocols or an arbitrary date. 1401 00:41:36,810 --> 00:41:38,129 And as more things so you can build a 1402 00:41:38,130 --> 00:41:39,449 single application that you know a lot 1403 00:41:39,450 --> 00:41:41,429 about, it's pretty easy or relatively 1404 00:41:41,430 --> 00:41:43,319 it's much easier, not pretty easy, 1405 00:41:44,520 --> 00:41:45,959 and then solve something that's on the 1406 00:41:45,960 --> 00:41:47,429 efficient frontier of risk and value. 1407 00:41:47,430 --> 00:41:49,439 Don't put like hundreds of millions of 1408 00:41:49,440 --> 00:41:51,029 dollars into solving something where 1409 00:41:51,030 --> 00:41:53,189 you're only like a tiny bit better 1410 00:41:53,190 --> 00:41:54,719 than somebody else either. 1411 00:41:54,720 --> 00:41:56,129 Don't spend a lot of money to be a tiny 1412 00:41:56,130 --> 00:41:57,449 bit better or 1413 00:41:58,890 --> 00:42:01,229 spend a lot of money but be amazingly 1414 00:42:01,230 --> 00:42:03,089 awesome at it and then cross subsidize. 1415 00:42:03,090 --> 00:42:04,319 If you build a system for your own 1416 00:42:04,320 --> 00:42:06,599 application, you might make really 1417 00:42:06,600 --> 00:42:08,459 high margins on, say, offering email 1418 00:42:08,460 --> 00:42:10,649 addresses and stuff where 1419 00:42:10,650 --> 00:42:12,329 the hosting can be higher than it is in 1420 00:42:12,330 --> 00:42:13,619 other places because you have less cost 1421 00:42:13,620 --> 00:42:14,969 for other things and then hybrid 1422 00:42:14,970 --> 00:42:16,289 solutions where you have like a P2P 1423 00:42:16,290 --> 00:42:18,779 model, a software system and a data haven 1424 00:42:18,780 --> 00:42:19,799 that are working together or things like 1425 00:42:19,800 --> 00:42:21,509 that, or an open source thing, or you 1426 00:42:21,510 --> 00:42:23,369 have something like fact's where it's not 1427 00:42:23,370 --> 00:42:24,689 an objectional application. 1428 00:42:24,690 --> 00:42:25,859 You use that to demonstrate your 1429 00:42:25,860 --> 00:42:27,059 technology and then build something else 1430 00:42:27,060 --> 00:42:28,060 later. 1431 00:42:28,710 --> 00:42:30,329 So there's a lot of useful technologies 1432 00:42:30,330 --> 00:42:31,979 that didn't really exist in the late 90s 1433 00:42:31,980 --> 00:42:33,269 when we're doing the stuff that do exist 1434 00:42:33,270 --> 00:42:34,289 now. 1435 00:42:34,290 --> 00:42:36,449 Tor one of the more useful tools here. 1436 00:42:36,450 --> 00:42:37,709 We have a lot of people to talk about. 1437 00:42:37,710 --> 00:42:38,819 There is no point in me talking matau 1438 00:42:39,900 --> 00:42:40,799 heavyweight clients. 1439 00:42:40,800 --> 00:42:42,839 There was a period in the early 2000s 1440 00:42:42,840 --> 00:42:44,669 where it was basically just Web clients. 1441 00:42:44,670 --> 00:42:47,249 No one wanted to run local 1442 00:42:47,250 --> 00:42:48,689 software on the machine and it was sort 1443 00:42:48,690 --> 00:42:50,189 of a JavaScript was at the level where 1444 00:42:50,190 --> 00:42:52,139 you could run a real client there. 1445 00:42:52,140 --> 00:42:53,759 Two things have killed that model, one 1446 00:42:53,760 --> 00:42:55,469 Ajax JavaScript stuff, so you can build a 1447 00:42:55,470 --> 00:42:56,609 pretty heavyweight client in a Web 1448 00:42:56,610 --> 00:42:58,289 browser. Even that doesn't require 1449 00:42:58,290 --> 00:42:59,369 network access. 1450 00:42:59,370 --> 00:43:01,439 Even better, you have mobile phones and 1451 00:43:01,440 --> 00:43:03,389 things. We can have an actual application. 1452 00:43:03,390 --> 00:43:04,709 So we've gotten back to the point where 1453 00:43:04,710 --> 00:43:05,969 you can build an application, has local 1454 00:43:05,970 --> 00:43:07,109 state, which lets you do much more 1455 00:43:07,110 --> 00:43:08,579 interesting protocols and you can build 1456 00:43:08,580 --> 00:43:10,199 new protocols. You can build a protocol 1457 00:43:10,200 --> 00:43:11,939 that's much more resistant to censorship 1458 00:43:11,940 --> 00:43:14,819 than http to a static address. 1459 00:43:14,820 --> 00:43:16,679 And I'm a big fan of Messers based 1460 00:43:16,680 --> 00:43:17,969 systems rather than Conexion oriented 1461 00:43:17,970 --> 00:43:20,009 systems. If your fundamental task is 1462 00:43:20,010 --> 00:43:22,169 message based, you can pass these 1463 00:43:22,170 --> 00:43:23,699 messages around like in the U.S. 1464 00:43:23,700 --> 00:43:25,559 model or the anonymous remailer model 1465 00:43:25,560 --> 00:43:26,729 rather than opening a pipe. 1466 00:43:26,730 --> 00:43:28,169 It's much more anonymous, much more 1467 00:43:28,170 --> 00:43:30,389 secure, much more reliable than that, and 1468 00:43:30,390 --> 00:43:31,769 use latency in that kind of situation 1469 00:43:31,770 --> 00:43:33,839 here. Advantage, however, there's a bunch 1470 00:43:33,840 --> 00:43:35,069 of missing technology. So we still have 1471 00:43:35,070 --> 00:43:36,539 not, I would say, solve the anonymous 1472 00:43:36,540 --> 00:43:37,589 electronic cash problem. 1473 00:43:37,590 --> 00:43:39,989 And Bitcoin is 1474 00:43:39,990 --> 00:43:42,509 a decent system, but is not anonymous, 1475 00:43:42,510 --> 00:43:44,219 certainly not anonymous against 1476 00:43:44,220 --> 00:43:46,079 determined effort to find ARBITRATES 1477 00:43:46,080 --> 00:43:47,699 actions not anonymous by default. 1478 00:43:47,700 --> 00:43:49,649 So zero coin and some systems like that 1479 00:43:49,650 --> 00:43:50,579 might be sufficient. 1480 00:43:50,580 --> 00:43:52,859 I'm still a true believer in Chami 1481 00:43:52,860 --> 00:43:55,439 and electronic digital blinded cash. 1482 00:43:55,440 --> 00:43:57,539 It has a long and kind of 1483 00:43:57,540 --> 00:43:59,129 sad history, but I think someone will 1484 00:43:59,130 --> 00:44:00,029 eventually do this and I'll be 1485 00:44:00,030 --> 00:44:01,169 successful. 1486 00:44:01,170 --> 00:44:02,789 I think we also need to reboot the 1487 00:44:02,790 --> 00:44:04,049 anonymous remailer network. 1488 00:44:04,050 --> 00:44:05,639 I mean, one of the sad things was Len 1489 00:44:05,640 --> 00:44:07,859 Sesemann was the main 1490 00:44:07,860 --> 00:44:09,959 remailer guy and he's no longer with us. 1491 00:44:09,960 --> 00:44:11,789 So we don't have a remailer network that 1492 00:44:11,790 --> 00:44:14,429 is as good as it was in like twenty 1493 00:44:14,430 --> 00:44:16,499 three. So we need at least get back to 1494 00:44:16,500 --> 00:44:17,759 that level and maybe build something 1495 00:44:17,760 --> 00:44:20,279 better and then cloud computing you 1496 00:44:20,280 --> 00:44:22,049 if you're building on the stuff for Data 1497 00:44:22,050 --> 00:44:24,209 Haven and then you've got us is your back 1498 00:44:24,210 --> 00:44:26,009 end for your application, it's really 1499 00:44:26,010 --> 00:44:27,869 easy to send a subpoena to there or do 1500 00:44:27,870 --> 00:44:29,069 whatever else. 1501 00:44:29,070 --> 00:44:30,599 Having a trustworthy cloud where the 1502 00:44:30,600 --> 00:44:32,489 operator can't modify your computing 1503 00:44:32,490 --> 00:44:34,649 would be great. And it's not really cost 1504 00:44:34,650 --> 00:44:36,239 effective to have dedicated physical 1505 00:44:36,240 --> 00:44:37,799 servers for each machine, especially in 1506 00:44:37,800 --> 00:44:38,849 an offshore data haven. 1507 00:44:38,850 --> 00:44:40,619 So there's that and then secure client 1508 00:44:40,620 --> 00:44:42,779 devices. As we've seen with the Silkroad 1509 00:44:42,780 --> 00:44:44,939 example, no matter how great your 1510 00:44:44,940 --> 00:44:47,159 server security is, if your client 1511 00:44:47,160 --> 00:44:48,259 device is. 1512 00:44:48,260 --> 00:44:50,419 Captured, unencrypted, or whatever else, 1513 00:44:50,420 --> 00:44:51,420 you've got serious problems. 1514 00:44:52,520 --> 00:44:54,289 So, yeah, that's basically 1515 00:44:55,430 --> 00:44:56,749 I have a bunch of URLs, this will be up 1516 00:44:56,750 --> 00:44:58,339 on the Web somewhere. 1517 00:44:58,340 --> 00:44:59,929 A lot of stuff. There's a bunch of 1518 00:44:59,930 --> 00:45:02,359 articles, some interesting 1519 00:45:02,360 --> 00:45:04,309 legal analysis of this that has been done 1520 00:45:04,310 --> 00:45:05,569 by people and that. 1521 00:45:05,570 --> 00:45:07,609 So, yeah, data humans have existed in 1522 00:45:07,610 --> 00:45:09,469 concept and practice for. 1523 00:45:09,470 --> 00:45:11,359 I would I think there's probably examples 1524 00:45:11,360 --> 00:45:12,649 in the 50s and 60s, certainly from the 1525 00:45:12,650 --> 00:45:14,749 70s and 80s and 1526 00:45:14,750 --> 00:45:16,039 genuinely mixed results. 1527 00:45:16,040 --> 00:45:17,299 So there's lots of work to do in the 1528 00:45:17,300 --> 00:45:19,069 future. I'd be very interested in any 1529 00:45:19,070 --> 00:45:21,079 questions or comments or anyone has. 1530 00:45:21,080 --> 00:45:22,420 We have about 15 minutes left. 1531 00:45:23,760 --> 00:45:24,760 Please use. 1532 00:45:34,680 --> 00:45:36,420 So any questions, anyone? 1533 00:45:41,730 --> 00:45:43,909 A question from the Internet, in his 1534 00:45:43,910 --> 00:45:45,979 opinion, a little out of place does 1535 00:45:45,980 --> 00:45:48,199 the unwanted attention, drugs, politics 1536 00:45:48,200 --> 00:45:50,419 and poorly supervised business and 1537 00:45:50,420 --> 00:45:52,579 data redundancy models like Siyabonga 1538 00:45:52,580 --> 00:45:54,219 outweigh the benefits of a data haven? 1539 00:45:56,020 --> 00:45:57,069 Please repeat 1540 00:45:59,290 --> 00:46:01,059 your opinion, does the unwanted 1541 00:46:01,060 --> 00:46:03,129 attention, drugs and politics 1542 00:46:03,130 --> 00:46:05,319 and poorly supervised business 1543 00:46:05,320 --> 00:46:07,599 and data redundancy models like Cyber 1544 00:46:07,600 --> 00:46:09,669 Bunker outweigh the benefits of a 1545 00:46:09,670 --> 00:46:10,419 data haven? 1546 00:46:10,420 --> 00:46:12,639 Yeah, I think 1547 00:46:12,640 --> 00:46:14,739 the the need for a data haven 1548 00:46:14,740 --> 00:46:16,899 in nineteen ninety eight was 1549 00:46:16,900 --> 00:46:18,699 very, very clear because the laws would 1550 00:46:18,700 --> 00:46:20,439 not allow lots of very legitimate 1551 00:46:20,440 --> 00:46:22,390 applications in 1552 00:46:23,800 --> 00:46:24,879 and September 10th. 1553 00:46:24,880 --> 00:46:26,259 Two thousand. 1554 00:46:26,260 --> 00:46:28,419 I would probably maybe have answered 1555 00:46:28,420 --> 00:46:30,489 the other way that laws in the 1556 00:46:30,490 --> 00:46:32,259 US and in Europe were pretty good at the 1557 00:46:32,260 --> 00:46:33,260 time. 1558 00:46:33,790 --> 00:46:36,309 However, Patriot Act 1559 00:46:36,310 --> 00:46:38,229 or IP in the UK, lots of things have been 1560 00:46:38,230 --> 00:46:39,579 pushing in the other direction. 1561 00:46:39,580 --> 00:46:41,649 So while there are severe negatives 1562 00:46:41,650 --> 00:46:43,779 to spam, abusive 1563 00:46:43,780 --> 00:46:45,999 use, things like that, I think 1564 00:46:46,000 --> 00:46:48,309 actual legitimate free speech use is a 1565 00:46:48,310 --> 00:46:50,079 sufficiently at risk that the value of 1566 00:46:50,080 --> 00:46:52,419 data havens is, if not 1567 00:46:52,420 --> 00:46:53,589 absolute, today. 1568 00:46:53,590 --> 00:46:55,659 Yes, there's a very easy 1569 00:46:55,660 --> 00:46:56,649 projection where it is. 1570 00:46:56,650 --> 00:46:58,629 So I think we need the technology, even 1571 00:46:58,630 --> 00:47:00,879 if we use it for the equivalent of 1572 00:47:00,880 --> 00:47:03,129 better latency reduction, having 1573 00:47:03,130 --> 00:47:05,379 servers close to people, we should build 1574 00:47:05,380 --> 00:47:06,510 that and then. Yeah. 1575 00:47:09,330 --> 00:47:11,549 Please, if you leave the room now, please 1576 00:47:11,550 --> 00:47:13,079 leave silent. 1577 00:47:13,080 --> 00:47:15,269 Thank you very much and take your garbage 1578 00:47:15,270 --> 00:47:16,270 with you. 1579 00:47:16,760 --> 00:47:18,169 Thanks, please. 1580 00:47:18,170 --> 00:47:20,539 On a scale from one to 10, 1581 00:47:20,540 --> 00:47:23,509 how much bullshit would you say is in the 1582 00:47:23,510 --> 00:47:25,699 ecosystem around the 1583 00:47:25,700 --> 00:47:27,889 Bitcoin and coin block 1584 00:47:27,890 --> 00:47:30,769 and block chain technology 1585 00:47:30,770 --> 00:47:32,899 right now and also in the startup world 1586 00:47:32,900 --> 00:47:34,399 in general? 1587 00:47:34,400 --> 00:47:36,349 Yes, that is a good question. 1588 00:47:37,870 --> 00:47:39,389 Ten being the most. 1589 00:47:39,390 --> 00:47:41,689 OK, so I think there 1590 00:47:41,690 --> 00:47:43,789 is clearly value in both a lot of 1591 00:47:43,790 --> 00:47:45,709 the startups and in Bitcoin. 1592 00:47:45,710 --> 00:47:48,529 Bitcoin is an awesome solution to 1593 00:47:48,530 --> 00:47:50,629 distributed systems problem that 1594 00:47:50,630 --> 00:47:52,249 has been open for a long time. 1595 00:47:52,250 --> 00:47:55,159 Bitcoin itself as a currency 1596 00:47:55,160 --> 00:47:57,079 does not really personally excite me. 1597 00:47:57,080 --> 00:47:59,269 I own two point three Bitcoin after 1598 00:47:59,270 --> 00:48:01,019 winning a bet about the North Korea 1599 00:48:01,020 --> 00:48:02,839 hackers and Sony thing. 1600 00:48:02,840 --> 00:48:05,089 So I'm not 1601 00:48:05,090 --> 00:48:07,069 I mean, I think it is not the final 1602 00:48:07,070 --> 00:48:08,539 system. I don't think Bitcoin as it is 1603 00:48:08,540 --> 00:48:09,919 today is going to be the system that does 1604 00:48:09,920 --> 00:48:11,059 everything we wanted to do. 1605 00:48:11,060 --> 00:48:13,519 But I think some anonymous 1606 00:48:13,520 --> 00:48:15,409 electronic payment system or some form of 1607 00:48:15,410 --> 00:48:17,659 value will and that might be block chain. 1608 00:48:17,660 --> 00:48:19,099 It might be Bitcoin over time. 1609 00:48:19,100 --> 00:48:20,629 It might change. 1610 00:48:20,630 --> 00:48:22,009 And yeah, there's a lot of hype with 1611 00:48:22,010 --> 00:48:23,719 startups, but especially in Silicon 1612 00:48:23,720 --> 00:48:26,059 Valley, where it seems like everyone 1613 00:48:26,060 --> 00:48:27,919 does it by default. But there's also a 1614 00:48:27,920 --> 00:48:29,599 lot of value. 1615 00:48:29,600 --> 00:48:31,459 The contrary to that is look look at the 1616 00:48:31,460 --> 00:48:32,779 big companies and how much innovation 1617 00:48:32,780 --> 00:48:34,219 they have. It seems like they have 1618 00:48:34,220 --> 00:48:36,049 outsourced all of the innovation to 1619 00:48:36,050 --> 00:48:37,309 startups other than by. 1620 00:48:37,310 --> 00:48:39,169 So maybe it's a fundamental shift in how 1621 00:48:39,170 --> 00:48:41,029 business works. But, yeah, there's a lot 1622 00:48:41,030 --> 00:48:42,199 there's there's a lot of good and a lot 1623 00:48:42,200 --> 00:48:43,250 of bad. And we don't really know. 1624 00:48:46,510 --> 00:48:48,669 I think maybe maybe there's another 1625 00:48:48,670 --> 00:48:50,739 axis like an axis or something here, 1626 00:48:50,740 --> 00:48:51,749 but yeah, yeah. 1627 00:48:55,390 --> 00:48:57,519 Have you considered the benefits of 1628 00:48:57,520 --> 00:48:59,799 distributing a decentralized 1629 00:48:59,800 --> 00:49:01,989 conflict set up just like the Pirate Bay 1630 00:49:01,990 --> 00:49:04,089 did while the servers were 1631 00:49:04,090 --> 00:49:05,979 taken down from everywhere they popped 1632 00:49:05,980 --> 00:49:08,139 up? Because I 1633 00:49:08,140 --> 00:49:10,209 think that superior 1634 00:49:10,210 --> 00:49:11,859 states like the United States, they would 1635 00:49:11,860 --> 00:49:13,989 just like Zealand if it was 1636 00:49:13,990 --> 00:49:16,149 just a great threat to the geopolitical 1637 00:49:16,150 --> 00:49:18,519 agenda, if it didn't have public support 1638 00:49:18,520 --> 00:49:20,529 and military to back it up. 1639 00:49:20,530 --> 00:49:21,609 Yeah, absolutely. 1640 00:49:21,610 --> 00:49:23,859 The level of protection you can get 1641 00:49:23,860 --> 00:49:26,079 from a physical location is 1642 00:49:26,080 --> 00:49:28,179 up to how angry you make people 1643 00:49:28,180 --> 00:49:30,399 that have the ability to bomb you 1644 00:49:30,400 --> 00:49:31,929 hosting gambling servers. 1645 00:49:31,930 --> 00:49:33,549 We never would have gotten to that point. 1646 00:49:33,550 --> 00:49:35,079 Our most objective, the weirdest thing 1647 00:49:35,080 --> 00:49:36,549 that I learned was that there's a lot of 1648 00:49:36,550 --> 00:49:38,059 stuff that's legal and like set of 1649 00:49:38,060 --> 00:49:39,399 country and then a lot of stuff. 1650 00:49:39,400 --> 00:49:41,079 It's legal instead of country B, but then 1651 00:49:41,080 --> 00:49:42,639 there's like the combination of that. 1652 00:49:42,640 --> 00:49:43,779 I guess I can talk about that now. 1653 00:49:43,780 --> 00:49:45,219 But like we had a customer that did this 1654 00:49:45,220 --> 00:49:47,529 weird bidding or betting 1655 00:49:47,530 --> 00:49:48,429 on porn images. 1656 00:49:48,430 --> 00:49:50,649 So porn in countries that are 1657 00:49:50,650 --> 00:49:52,359 OK with gambling is usually bad. 1658 00:49:52,360 --> 00:49:53,589 So there were very few countries that had 1659 00:49:53,590 --> 00:49:55,329 both porn and gambling being OK. 1660 00:49:55,330 --> 00:49:57,789 You'd bet on which of six softcore images 1661 00:49:57,790 --> 00:49:59,409 would be popular, the most popular among 1662 00:49:59,410 --> 00:50:00,369 the people that week. 1663 00:50:00,370 --> 00:50:02,529 And you would then 1664 00:50:02,530 --> 00:50:03,789 win if it was the most popular thing, 1665 00:50:03,790 --> 00:50:04,959 which is actually pretty awesome, I think 1666 00:50:04,960 --> 00:50:06,459 would be a fun thing to recreate. 1667 00:50:06,460 --> 00:50:08,559 But they couldn't find an 1668 00:50:08,560 --> 00:50:10,659 acceptable gaming jurisdiction and 1669 00:50:10,660 --> 00:50:11,869 an acceptable porn jurisdiction. 1670 00:50:11,870 --> 00:50:13,089 That was all in one. But no one's going 1671 00:50:13,090 --> 00:50:15,639 to be famous for that. So, yeah. 1672 00:50:15,640 --> 00:50:17,529 Yeah, I agree that the people like 1673 00:50:17,530 --> 00:50:19,689 WikiLeaks, I am certain that if the 1674 00:50:19,690 --> 00:50:22,269 US government could quietly kill 1675 00:50:22,270 --> 00:50:24,429 the people involved in WikiLeaks and not 1676 00:50:24,430 --> 00:50:26,289 get caught for it or not be attributable, 1677 00:50:26,290 --> 00:50:27,969 it would have happened. So the only thing 1678 00:50:27,970 --> 00:50:30,399 that kept people alive and successful was 1679 00:50:30,400 --> 00:50:31,749 being distributed. And that sort of 1680 00:50:31,750 --> 00:50:34,119 system that is the most resistant system 1681 00:50:34,120 --> 00:50:36,279 against a large that like that or finding 1682 00:50:36,280 --> 00:50:38,409 a counterweight nation state or something 1683 00:50:38,410 --> 00:50:40,329 like going to Russia is the solution. 1684 00:50:40,330 --> 00:50:41,349 They're not going to go to war with 1685 00:50:41,350 --> 00:50:43,119 Russia over Snowden, but they would have 1686 00:50:43,120 --> 00:50:44,919 potentially done more pressure against my 1687 00:50:44,920 --> 00:50:47,259 country. So there's a crazy 1688 00:50:47,260 --> 00:50:49,239 geopolitical thing involved. 1689 00:50:49,240 --> 00:50:51,329 But yeah, software crypto that night. 1690 00:50:51,330 --> 00:50:52,629 I think the thing that got me interested 1691 00:50:52,630 --> 00:50:54,939 in crypto when I was like 11 or 12 1692 00:50:54,940 --> 00:50:57,279 years old was knowing that I 1693 00:50:57,280 --> 00:50:59,529 was like in a house in a 1694 00:50:59,530 --> 00:51:00,939 suburban house. And I had like very 1695 00:51:00,940 --> 00:51:02,979 little resources. I could do something 1696 00:51:02,980 --> 00:51:04,089 that no one could undo. 1697 00:51:04,090 --> 00:51:06,159 Like cryptographically the computing 1698 00:51:06,160 --> 00:51:08,229 power on my machine was enough that with 1699 00:51:08,230 --> 00:51:09,789 the right algorithm, you couldn't decrypt 1700 00:51:09,790 --> 00:51:11,649 it, even if you had all the resources of 1701 00:51:11,650 --> 00:51:13,329 everyone that would ever exist in any 1702 00:51:13,330 --> 00:51:15,609 part of the universe. So that is a really 1703 00:51:15,610 --> 00:51:16,569 awesome concept. 1704 00:51:16,570 --> 00:51:17,619 And if you can use that to your 1705 00:51:17,620 --> 00:51:18,609 advantage, you go for it. 1706 00:51:18,610 --> 00:51:19,749 But there's a lot of things you can't do 1707 00:51:19,750 --> 00:51:21,249 that transaction systems make it much 1708 00:51:21,250 --> 00:51:22,250 harder to. 1709 00:51:24,270 --> 00:51:25,799 It seems like in a way to avoid and 1710 00:51:25,800 --> 00:51:28,139 services has achieved some of what 1711 00:51:28,140 --> 00:51:30,469 you intended to achieve with a haven. 1712 00:51:30,470 --> 00:51:32,789 So do you see any other 1713 00:51:32,790 --> 00:51:34,349 practical alternatives to Tauhid and 1714 00:51:34,350 --> 00:51:35,999 services right now or things coming up 1715 00:51:36,000 --> 00:51:37,439 that might be better than to wait and 1716 00:51:37,440 --> 00:51:38,579 services yet? 1717 00:51:38,580 --> 00:51:40,679 Tor Hidden Services is 1718 00:51:40,680 --> 00:51:43,079 a great system within 1719 00:51:43,080 --> 00:51:45,329 the security parameters of the Tor 1720 00:51:45,330 --> 00:51:46,679 network. I think if you had a 1721 00:51:46,680 --> 00:51:49,409 sufficiently dangerous application 1722 00:51:49,410 --> 00:51:51,209 posted on a service or if you made a 1723 00:51:51,210 --> 00:51:53,219 mistake, you could compromise the entire 1724 00:51:53,220 --> 00:51:56,099 system. TOR is not designed to resist a 1725 00:51:56,100 --> 00:51:58,229 really determined, active global 1726 00:51:58,230 --> 00:52:00,419 adversary and that is the 1727 00:52:00,420 --> 00:52:02,669 adversary that we face. 1728 00:52:02,670 --> 00:52:04,079 They're willing to modify packets, 1729 00:52:04,080 --> 00:52:05,669 they're willing to do whatever. So I 1730 00:52:05,670 --> 00:52:07,589 don't think you could run a long running 1731 00:52:07,590 --> 00:52:09,569 tor hidden service with a Tor network as 1732 00:52:09,570 --> 00:52:11,969 it is today, with 1733 00:52:11,970 --> 00:52:14,159 something that the US government really, 1734 00:52:14,160 --> 00:52:16,109 really, really cared about defeating to 1735 00:52:16,110 --> 00:52:17,249 the point where they would break arbitrary 1736 00:52:17,250 --> 00:52:19,529 laws for like if the location 1737 00:52:19,530 --> 00:52:21,419 of Osama bin Laden had been discoverable 1738 00:52:21,420 --> 00:52:23,669 by defeating all of Tor 1739 00:52:23,670 --> 00:52:25,739 in 2005 or so, 1740 00:52:25,740 --> 00:52:27,479 it would have been defeated. 1741 00:52:27,480 --> 00:52:29,729 So that's a partially 1742 00:52:29,730 --> 00:52:30,929 a problem with the system as part of the 1743 00:52:30,930 --> 00:52:33,089 problem of scale. But I think there 1744 00:52:33,090 --> 00:52:35,369 are systems that you could build with 1745 00:52:35,370 --> 00:52:37,409 feasible resources that would resist that 1746 00:52:37,410 --> 00:52:39,329 threat. I think the systems that would be 1747 00:52:39,330 --> 00:52:41,189 the easiest to build the most principle 1748 00:52:41,190 --> 00:52:43,229 are message based systems like anonymous 1749 00:52:43,230 --> 00:52:44,279 remailer. 1750 00:52:44,280 --> 00:52:46,409 Tim may have this awesome thing, 1751 00:52:46,410 --> 00:52:48,599 black net back in like the mid 1752 00:52:48,600 --> 00:52:50,909 90s where you would send 1753 00:52:50,910 --> 00:52:52,859 anonymous mail to anonymous remailer. 1754 00:52:52,860 --> 00:52:54,539 It would then wait like a week or so and 1755 00:52:54,540 --> 00:52:56,189 do some operation on an email, you back a 1756 00:52:56,190 --> 00:52:57,929 response or send it to a Usenet posting 1757 00:52:57,930 --> 00:52:59,699 group that's much more secure against a 1758 00:52:59,700 --> 00:53:01,919 global passive or global global active 1759 00:53:01,920 --> 00:53:03,279 adversary, then a connection orated 1760 00:53:03,280 --> 00:53:05,099 system with connection systems. 1761 00:53:05,100 --> 00:53:06,899 You can just do crazy stuff like if you 1762 00:53:06,900 --> 00:53:09,269 think someone is the the list, 1763 00:53:09,270 --> 00:53:10,379 like ten thousand people that are 1764 00:53:10,380 --> 00:53:11,849 possible candidates, you look at their 1765 00:53:11,850 --> 00:53:13,619 travel patterns, you arrest like ten of 1766 00:53:13,620 --> 00:53:15,149 them. If you thought of their high probability, 1767 00:53:15,150 --> 00:53:16,769 you see if a service goes offline, you go 1768 00:53:16,770 --> 00:53:17,939 after servers individually. 1769 00:53:17,940 --> 00:53:18,989 There's lots of stuff you can do if 1770 00:53:18,990 --> 00:53:20,279 you're willing to be a bad person to 1771 00:53:20,280 --> 00:53:21,719 uncover important services. 1772 00:53:21,720 --> 00:53:22,889 But it's an awesome system and it's like 1773 00:53:22,890 --> 00:53:25,019 the it's the best practical 1774 00:53:25,020 --> 00:53:26,039 thing we have today. 1775 00:53:29,190 --> 00:53:31,289 It's pleas from the Internet. 1776 00:53:31,290 --> 00:53:33,389 Yes, there's a question, uh, could an 1777 00:53:33,390 --> 00:53:35,129 established multinational company 1778 00:53:35,130 --> 00:53:37,019 theoretically build a data haven within 1779 00:53:37,020 --> 00:53:38,020 itself currently? 1780 00:53:39,560 --> 00:53:41,959 Yes, so and multinational 1781 00:53:41,960 --> 00:53:44,269 organizations often do have things 1782 00:53:44,270 --> 00:53:45,619 that are very close to date havens, like 1783 00:53:45,620 --> 00:53:47,689 I think I've done a lot of computer 1784 00:53:47,690 --> 00:53:49,549 security stuff over the past two decades. 1785 00:53:49,550 --> 00:53:51,469 There's a crazy fact of like if you're a 1786 00:53:51,470 --> 00:53:52,999 regulated industry that has to meet a 1787 00:53:53,000 --> 00:53:54,679 certain security objective, like a 1788 00:53:54,680 --> 00:53:56,419 government contractor, a government 1789 00:53:56,420 --> 00:53:58,489 entity, somebody who has 1790 00:53:58,490 --> 00:54:00,259 some external regulation, you will meet 1791 00:54:00,260 --> 00:54:01,939 that regulation, but you won't go any 1792 00:54:01,940 --> 00:54:03,349 beyond that regulation. 1793 00:54:03,350 --> 00:54:04,909 The people who have like an open ended 1794 00:54:04,910 --> 00:54:06,079 liability if their things are 1795 00:54:06,080 --> 00:54:08,179 compromised, do actually in some cases an 1796 00:54:08,180 --> 00:54:09,679 exceptionally good job of security. 1797 00:54:09,680 --> 00:54:11,359 Pharma companies doing drug discovery 1798 00:54:11,360 --> 00:54:13,279 work where the molecule is like the most 1799 00:54:13,280 --> 00:54:14,629 secret thing in the world for them. 1800 00:54:14,630 --> 00:54:17,029 They have actually good security 1801 00:54:17,030 --> 00:54:18,169 proprietary trading firms. 1802 00:54:18,170 --> 00:54:19,339 They keep their algorithms relatively 1803 00:54:19,340 --> 00:54:21,649 secure against like compared 1804 00:54:21,650 --> 00:54:22,969 to what I would say any government agency 1805 00:54:22,970 --> 00:54:25,099 has done. So there 1806 00:54:25,100 --> 00:54:27,169 are organizations that do 1807 00:54:27,170 --> 00:54:28,339 a pretty good job of this. 1808 00:54:28,340 --> 00:54:29,899 Usually those are those are usually 1809 00:54:29,900 --> 00:54:31,819 static data hosting systems, not 1810 00:54:31,820 --> 00:54:33,139 transaction processing systems. 1811 00:54:33,140 --> 00:54:34,549 And they're for internal use. 1812 00:54:34,550 --> 00:54:35,989 It's a much easier model because you 1813 00:54:35,990 --> 00:54:36,889 don't really have to worry about 1814 00:54:36,890 --> 00:54:39,169 censorship or denial of service, 1815 00:54:39,170 --> 00:54:40,909 attack or other action. 1816 00:54:40,910 --> 00:54:42,709 You just need to keep integrity and 1817 00:54:42,710 --> 00:54:44,719 operation. But yeah, corporations 1818 00:54:44,720 --> 00:54:46,309 internally can build very secure systems. 1819 00:54:46,310 --> 00:54:47,959 However, most corporations, as everyone 1820 00:54:47,960 --> 00:54:49,609 is aware, have pretty horrible internal 1821 00:54:49,610 --> 00:54:50,610 systems. 1822 00:54:52,490 --> 00:54:53,490 OK. 1823 00:54:54,530 --> 00:54:55,459 One question. 1824 00:54:55,460 --> 00:54:57,199 Is there any system you would recommend 1825 00:54:57,200 --> 00:54:58,760 to watch out for in the future, 1826 00:55:00,590 --> 00:55:02,929 Zucco is working on zero point 1827 00:55:02,930 --> 00:55:05,299 zero cash. Well, I think that's public. 1828 00:55:05,300 --> 00:55:07,279 I hope that's the thing. 1829 00:55:07,280 --> 00:55:08,690 And I'm personally the most excited about 1830 00:55:10,040 --> 00:55:11,869 there's. 1831 00:55:11,870 --> 00:55:13,789 Yeah, I think that's the most exciting 1832 00:55:13,790 --> 00:55:15,889 system tour, HP will 1833 00:55:15,890 --> 00:55:16,969 continue development. 1834 00:55:16,970 --> 00:55:19,159 I think they have to they're on a curve 1835 00:55:19,160 --> 00:55:20,869 that is not going to get to where you 1836 00:55:20,870 --> 00:55:22,669 really need to be for data haven unless 1837 00:55:22,670 --> 00:55:25,069 they dramatically 1838 00:55:25,070 --> 00:55:26,239 change or improve. 1839 00:55:26,240 --> 00:55:28,129 I think there needs to be some 1840 00:55:28,130 --> 00:55:29,419 improvement and change. 1841 00:55:29,420 --> 00:55:31,159 Part of that is coming up with legitimate 1842 00:55:31,160 --> 00:55:34,009 applications. So if if 1843 00:55:34,010 --> 00:55:36,049 horrible governments get elected, US, 1844 00:55:36,050 --> 00:55:38,149 Europe, everywhere else, such that the 1845 00:55:38,150 --> 00:55:40,309 need for data havens is increased 1846 00:55:40,310 --> 00:55:42,139 and more people see it as a mainstream 1847 00:55:42,140 --> 00:55:43,669 thing that we have to have, we'll get 1848 00:55:43,670 --> 00:55:44,569 great data. 1849 00:55:44,570 --> 00:55:46,009 However, we'll live in a world that has 1850 00:55:46,010 --> 00:55:47,749 really shitty US and European governments 1851 00:55:47,750 --> 00:55:49,129 where people get like abducted and killed 1852 00:55:49,130 --> 00:55:50,209 in all sorts of stuff like that. 1853 00:55:50,210 --> 00:55:51,529 It's not really a trade off at once. 1854 00:55:51,530 --> 00:55:53,149 So what we really want is a data haven 1855 00:55:53,150 --> 00:55:55,099 that's really secure in an environment 1856 00:55:55,100 --> 00:55:57,349 where you don't really need them. 1857 00:55:57,350 --> 00:55:59,149 And that's sort of a fundamental quandary 1858 00:55:59,150 --> 00:56:01,249 of if you don't need it, what no one's 1859 00:56:01,250 --> 00:56:02,279 going to spend the money doing it. 1860 00:56:02,280 --> 00:56:03,259 You're the best people in the world 1861 00:56:03,260 --> 00:56:04,189 aren't going to work on it. They're going to 1862 00:56:04,190 --> 00:56:06,379 work on something like delivering 1863 00:56:06,380 --> 00:56:07,909 faster traffic rather than more secure 1864 00:56:07,910 --> 00:56:09,079 traffic. So, yeah. 1865 00:56:10,400 --> 00:56:12,350 OK, one last question from the Internet. 1866 00:56:14,030 --> 00:56:15,979 Is there a data haven, do it yourself, 1867 00:56:15,980 --> 00:56:17,749 how to online? If not, would you like to 1868 00:56:17,750 --> 00:56:20,329 put online? And is there a data haven 1869 00:56:20,330 --> 00:56:22,259 software as a service? 1870 00:56:22,260 --> 00:56:24,599 Yes, there 1871 00:56:24,600 --> 00:56:26,759 is not a good document for this I'm 1872 00:56:26,760 --> 00:56:28,979 working on, I bought a couple of cabinets 1873 00:56:28,980 --> 00:56:30,869 of space and I'm building out what I 1874 00:56:30,870 --> 00:56:33,419 think the best way to build hosting 1875 00:56:33,420 --> 00:56:35,759 and just sort of like best practice 1876 00:56:35,760 --> 00:56:37,379 for that will be I'll document all that 1877 00:56:37,380 --> 00:56:38,039 and put it up. 1878 00:56:38,040 --> 00:56:40,829 But until we get a 1879 00:56:40,830 --> 00:56:43,079 virtualization platform that can do 1880 00:56:43,080 --> 00:56:44,129 remote attestations. 1881 00:56:44,130 --> 00:56:46,379 So I have this company before 1882 00:56:46,380 --> 00:56:48,629 I joined CloudFlare Crypto SEAL 1883 00:56:48,630 --> 00:56:50,789 that was doing cloud computing 1884 00:56:50,790 --> 00:56:52,379 where you could remotely attest to the 1885 00:56:52,380 --> 00:56:54,389 integrity of the container and the VM and 1886 00:56:54,390 --> 00:56:56,189 everything else. There was no market for 1887 00:56:56,190 --> 00:56:57,779 it. It was very hard to build. 1888 00:56:57,780 --> 00:56:59,339 Private Core, which got the Facebook, was 1889 00:56:59,340 --> 00:57:00,360 working on very similar stuff 1890 00:57:01,920 --> 00:57:04,169 until we have better hardware 1891 00:57:04,170 --> 00:57:06,239 platforms, until SGX is a start, but 1892 00:57:06,240 --> 00:57:08,189 a lot of other stuff, we aren't gonna be 1893 00:57:08,190 --> 00:57:09,659 able to build a virtual virtualize 1894 00:57:09,660 --> 00:57:10,919 platform that is secure enough for this 1895 00:57:10,920 --> 00:57:12,839 kind of stuff. And without that, it's 1896 00:57:12,840 --> 00:57:13,919 going to be very hard to make this 1897 00:57:13,920 --> 00:57:15,779 commercially viable if you have to use 1898 00:57:15,780 --> 00:57:17,009 dedicated hardware for every single 1899 00:57:17,010 --> 00:57:18,659 customer. It'll be pretty challenging. 1900 00:57:18,660 --> 00:57:20,790 So, yeah, maybe. 1901 00:57:23,770 --> 00:57:25,149 OK, thank you, Ron. 1902 00:57:36,660 --> 00:57:39,149 OK, for all the others to please, please 1903 00:57:39,150 --> 00:57:41,249 take your garbage with you and 1904 00:57:41,250 --> 00:57:43,199 we will continue in 15 minutes, memory 1905 00:57:43,200 --> 00:57:45,089 corruption, why can't we have nice 1906 00:57:45,090 --> 00:57:46,090 things?