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/587 Thanks! 1 00:00:10,620 --> 00:00:12,749 So welcome again to everyone 2 00:00:12,750 --> 00:00:15,299 in the audience, welcome also to everyone 3 00:00:15,300 --> 00:00:17,639 watching via livestream from around 4 00:00:17,640 --> 00:00:19,019 the world. 5 00:00:19,020 --> 00:00:21,239 We have the pleasure to welcome Christoph 6 00:00:21,240 --> 00:00:23,489 Engelmann. Now, Kristof 7 00:00:23,490 --> 00:00:26,609 is an author, a teacher and a researcher. 8 00:00:26,610 --> 00:00:28,769 He researches at the Institute 9 00:00:28,770 --> 00:00:31,139 for Advanced Studies on Media 10 00:00:31,140 --> 00:00:33,719 Culture of Computer Simulation 11 00:00:33,720 --> 00:00:36,449 that is at the Low Fana University 12 00:00:36,450 --> 00:00:38,249 in Luna Park. 13 00:00:38,250 --> 00:00:40,649 In his talk today, he will focus 14 00:00:40,650 --> 00:00:42,809 on the enabling dimensions 15 00:00:42,810 --> 00:00:43,769 of drone warfare. 16 00:00:43,770 --> 00:00:46,289 That's a little spin by especially 17 00:00:46,290 --> 00:00:48,629 looking at data gathering and 18 00:00:48,630 --> 00:00:50,630 the mapping of social graphs. 19 00:00:51,810 --> 00:00:53,969 We will have a chance for Q&A 20 00:00:53,970 --> 00:00:55,319 in the end. 21 00:00:55,320 --> 00:00:57,329 So please take notes. 22 00:00:57,330 --> 00:00:59,219 If you have questions in the end, you can 23 00:00:59,220 --> 00:01:00,629 ask them. 24 00:01:00,630 --> 00:01:02,549 We have plenty of time. 25 00:01:02,550 --> 00:01:04,859 But for now, with further ado, please 26 00:01:04,860 --> 00:01:06,989 help me welcome Christoph Engelmann 27 00:01:06,990 --> 00:01:09,119 to his talk on Graff's drones and 28 00:01:09,120 --> 00:01:11,279 phones, the role of social graphs 29 00:01:11,280 --> 00:01:13,409 for drones in the war on terror and 30 00:01:13,410 --> 00:01:14,410 stuff like. 31 00:01:21,740 --> 00:01:22,740 Thank you. 32 00:01:23,750 --> 00:01:25,640 It's exciting to be back. 33 00:01:26,780 --> 00:01:28,909 It's my third talk here, and as most of 34 00:01:28,910 --> 00:01:31,669 you know, I've previously 35 00:01:31,670 --> 00:01:34,459 talked about history 36 00:01:34,460 --> 00:01:36,709 and present day developments of 37 00:01:36,710 --> 00:01:39,289 identity, media, authentic authentication 38 00:01:39,290 --> 00:01:40,699 technology. 39 00:01:40,700 --> 00:01:42,829 Um, my last talk I gave 40 00:01:42,830 --> 00:01:45,679 in a pretty hungover state, which 41 00:01:45,680 --> 00:01:47,150 you will see if you watch the video. 42 00:01:48,380 --> 00:01:49,939 I slept better this night and had some 43 00:01:49,940 --> 00:01:51,529 coffee needmore. 44 00:01:51,530 --> 00:01:53,900 And you know, what I'm going to do today 45 00:01:55,640 --> 00:01:57,799 is pretty much ask a very 46 00:01:57,800 --> 00:01:58,949 similar question. 47 00:01:58,950 --> 00:02:01,009 Then the questions I have asked before 48 00:02:01,010 --> 00:02:03,139 in this context here is how do 49 00:02:03,140 --> 00:02:04,309 you identify people? 50 00:02:04,310 --> 00:02:06,439 But here in the context of drone warfare, 51 00:02:06,440 --> 00:02:09,228 because whenever to talk 52 00:02:09,229 --> 00:02:11,629 about the debate came to drones and 53 00:02:11,630 --> 00:02:13,909 surveillance in the past decade, 54 00:02:13,910 --> 00:02:16,339 I always wondered how do they actually 55 00:02:16,340 --> 00:02:18,529 identify the individuals they 56 00:02:18,530 --> 00:02:19,849 see in the video feeds? 57 00:02:21,180 --> 00:02:23,429 Because that's an important part of of 58 00:02:23,430 --> 00:02:25,499 the whole war on terror, discriminate 59 00:02:25,500 --> 00:02:27,809 individuals and decide 60 00:02:27,810 --> 00:02:30,210 which ones you want to kill or capture 61 00:02:31,230 --> 00:02:32,189 and which ones not. 62 00:02:32,190 --> 00:02:34,499 And many may argue that drones 63 00:02:34,500 --> 00:02:36,749 kill quite indiscriminately. 64 00:02:36,750 --> 00:02:38,909 And that is definitely true. 65 00:02:38,910 --> 00:02:40,919 But at the same time, one can see, 66 00:02:40,920 --> 00:02:43,409 especially since the publication 67 00:02:43,410 --> 00:02:45,029 of the drone papers by The Intercept in 68 00:02:45,030 --> 00:02:48,359 October, there is a fairly elaborate 69 00:02:48,360 --> 00:02:50,279 process, a bureaucratic and 70 00:02:50,280 --> 00:02:51,750 administrative process 71 00:02:53,040 --> 00:02:54,989 being put in place, at least in the 72 00:02:54,990 --> 00:02:56,069 American context, 73 00:02:58,800 --> 00:03:01,799 a process that that, um, 74 00:03:01,800 --> 00:03:04,559 is meant to positively identify 75 00:03:04,560 --> 00:03:06,689 the individuals you're tracking and 76 00:03:06,690 --> 00:03:07,860 eventually taking out 77 00:03:09,060 --> 00:03:11,099 many, many people involved, many, many 78 00:03:11,100 --> 00:03:13,380 hours of of surveillance. 79 00:03:14,630 --> 00:03:16,080 There's this idea of 80 00:03:17,160 --> 00:03:19,259 an unblinking eye of 81 00:03:19,260 --> 00:03:21,389 looking at people for weeks and 82 00:03:21,390 --> 00:03:23,609 weeks and weeks and start 83 00:03:23,610 --> 00:03:25,859 the whole observation cycle anew 84 00:03:25,860 --> 00:03:28,169 once the eye blinks, once you lose 85 00:03:28,170 --> 00:03:29,210 visual tracking. 86 00:03:30,270 --> 00:03:32,429 But of course, this visual 87 00:03:32,430 --> 00:03:35,879 tracking somehow is suspicious. 88 00:03:35,880 --> 00:03:38,249 First of all, the quality might not be 89 00:03:38,250 --> 00:03:40,529 too good. Secondly, 90 00:03:40,530 --> 00:03:42,600 I couldn't fathom that they use biometric 91 00:03:44,160 --> 00:03:45,959 technology in the context of this video 92 00:03:45,960 --> 00:03:48,149 feeds because that 93 00:03:48,150 --> 00:03:49,419 would be too tall of an order. 94 00:03:49,420 --> 00:03:51,959 So they must have been something else. 95 00:03:51,960 --> 00:03:54,539 And that is what I'm trying. 96 00:03:54,540 --> 00:03:57,209 Part of this puzzle I think I 97 00:03:57,210 --> 00:03:59,549 can offer some answers to 98 00:03:59,550 --> 00:04:00,869 that might be more that we don't know 99 00:04:00,870 --> 00:04:03,179 about as far as all of this research. 100 00:04:03,180 --> 00:04:05,189 It's difficult to do because there's 101 00:04:05,190 --> 00:04:07,289 little publicly available. 102 00:04:07,290 --> 00:04:08,609 You have to puzzle together from the 103 00:04:08,610 --> 00:04:11,249 Snowden files, from from other leaks 104 00:04:11,250 --> 00:04:13,379 and also from doctrinal papers or 105 00:04:15,090 --> 00:04:17,289 like PMA and PhD works that 106 00:04:17,290 --> 00:04:19,109 people in the military do, which is an 107 00:04:19,110 --> 00:04:20,489 interesting resource. If you start to 108 00:04:20,490 --> 00:04:22,649 look at those, um, you will find 109 00:04:22,650 --> 00:04:25,589 tons and tons of interesting papers 110 00:04:25,590 --> 00:04:28,169 tackling exactly these problems. 111 00:04:28,170 --> 00:04:30,569 So one little clue came in 2008 112 00:04:30,570 --> 00:04:32,729 or 2009 and this Wired article 113 00:04:32,730 --> 00:04:34,799 about tagging technology that's used in 114 00:04:34,800 --> 00:04:35,939 the war on terror. 115 00:04:35,940 --> 00:04:38,879 Basically, it's kind of RFID 116 00:04:38,880 --> 00:04:41,399 or, um, 117 00:04:42,600 --> 00:04:45,239 added tags or technologies. 118 00:04:45,240 --> 00:04:47,369 Does that work in some way in the 119 00:04:47,370 --> 00:04:49,619 optical spectrum by 120 00:04:49,620 --> 00:04:51,599 tagging people with paint or any kind of 121 00:04:51,600 --> 00:04:53,849 other medium that 122 00:04:53,850 --> 00:04:55,949 then will show up in certain 123 00:04:55,950 --> 00:04:58,089 spectrum? Oh, the CAT scan. 124 00:04:59,430 --> 00:05:00,430 That's a pity. 125 00:05:01,740 --> 00:05:03,240 But that, again, 126 00:05:04,500 --> 00:05:06,449 seems to be fairly difficult to do 127 00:05:06,450 --> 00:05:07,649 because you need access to the 128 00:05:07,650 --> 00:05:09,899 individuals or, you know, things 129 00:05:09,900 --> 00:05:11,999 they own, equipment they have in order 130 00:05:12,000 --> 00:05:13,019 to do that. 131 00:05:13,020 --> 00:05:14,909 And also the question is how discriminate 132 00:05:14,910 --> 00:05:16,829 can can this get? 133 00:05:16,830 --> 00:05:18,929 So, um, there must 134 00:05:18,930 --> 00:05:21,389 be, as I said, something else. 135 00:05:21,390 --> 00:05:23,969 And, uh, in order to further my 136 00:05:23,970 --> 00:05:27,149 argument that I want to make today, 137 00:05:27,150 --> 00:05:29,369 which to outline it once 138 00:05:29,370 --> 00:05:31,499 more, focus less on the drones and on the 139 00:05:31,500 --> 00:05:33,599 infrastructure, I have to take you on a 140 00:05:33,600 --> 00:05:35,879 little detour through history, more 141 00:05:35,880 --> 00:05:39,119 precisely the history of graph theory 142 00:05:39,120 --> 00:05:40,769 and those of you are trained computer 143 00:05:40,770 --> 00:05:43,019 science or maybe some sociologist, 144 00:05:43,020 --> 00:05:44,969 probably no way more than I do about 145 00:05:44,970 --> 00:05:46,709 graph theory. And I'm happy to learn from 146 00:05:46,710 --> 00:05:48,779 you in the Q&A or afterwards. 147 00:05:48,780 --> 00:05:50,550 The foundations of graph theory 148 00:05:51,600 --> 00:05:53,930 is Prager, and this is the River Prager. 149 00:05:55,740 --> 00:05:57,299 We are late in the eighteenth century by 150 00:05:57,300 --> 00:05:59,939 the Swiss mathematician Leonid Oilor, 151 00:05:59,940 --> 00:06:02,429 and he was employed at the Academy 152 00:06:02,430 --> 00:06:05,459 of Sciences in St. Petersburg, 153 00:06:06,480 --> 00:06:09,449 which had opposed the so-called 154 00:06:09,450 --> 00:06:11,699 Koenigsberg. A little problème. 155 00:06:11,700 --> 00:06:14,249 Um, and 156 00:06:14,250 --> 00:06:16,319 the question was, can one cross the seven 157 00:06:16,320 --> 00:06:18,719 bridges of Koenigsberg crossing the 158 00:06:18,720 --> 00:06:20,879 river Plager here and never 159 00:06:20,880 --> 00:06:24,269 cross the same bridge twice? 160 00:06:24,270 --> 00:06:26,369 And Euler's solution was basically to 161 00:06:26,370 --> 00:06:28,799 abstract this geographical ensemble 162 00:06:28,800 --> 00:06:30,809 into a set of points and the connections 163 00:06:30,810 --> 00:06:33,059 which can be depicted like this 164 00:06:33,060 --> 00:06:35,189 or depicted like that. 165 00:06:35,190 --> 00:06:37,499 And both 166 00:06:37,500 --> 00:06:39,869 graphs, these are graphs, 167 00:06:39,870 --> 00:06:41,489 isomorphic, they are the same. 168 00:06:41,490 --> 00:06:43,049 Even though they look different, they 169 00:06:43,050 --> 00:06:45,239 show the same relationship, the same 170 00:06:45,240 --> 00:06:46,240 topology. 171 00:06:51,560 --> 00:06:53,899 OK, and what we are seeing is four points 172 00:06:53,900 --> 00:06:56,209 called vertices or not 173 00:06:56,210 --> 00:06:57,919 in graph parlance, as well as seven 174 00:06:57,920 --> 00:07:00,439 connections which represent the bridges 175 00:07:00,440 --> 00:07:01,440 in this problem, 176 00:07:02,540 --> 00:07:03,589 which are called edges. 177 00:07:03,590 --> 00:07:05,599 So the knots of artists are the 178 00:07:05,600 --> 00:07:06,600 landmasses. 179 00:07:07,560 --> 00:07:10,019 And the edges 180 00:07:10,020 --> 00:07:12,089 are the bridges and 181 00:07:12,090 --> 00:07:13,799 oilor could show that all four landmines 182 00:07:13,800 --> 00:07:15,809 and Koenigsberg would have a have an 183 00:07:15,810 --> 00:07:17,849 uneven number of bridges and that in such 184 00:07:17,850 --> 00:07:19,979 a set up, it is impossible to find 185 00:07:19,980 --> 00:07:22,979 a way that crosses the bridge only once. 186 00:07:22,980 --> 00:07:25,079 And Gravas typically notate it like 187 00:07:25,080 --> 00:07:27,239 this, a graph is averted and an 188 00:07:27,240 --> 00:07:28,169 edge. 189 00:07:28,170 --> 00:07:30,479 And as I 190 00:07:30,480 --> 00:07:32,579 as I said, um, the 191 00:07:32,580 --> 00:07:34,809 number of edges between two, what is. 192 00:07:36,280 --> 00:07:37,719 I usually called weights and the more 193 00:07:37,720 --> 00:07:40,269 edges you have, the more weights you have 194 00:07:40,270 --> 00:07:42,249 in a given graph and despite this 195 00:07:42,250 --> 00:07:44,289 graphical representation I just showed 196 00:07:44,290 --> 00:07:45,290 here. 197 00:07:46,900 --> 00:07:49,119 It is important to remember that this 198 00:07:49,120 --> 00:07:50,799 is less a visual tool. 199 00:07:50,800 --> 00:07:52,899 It is also a visual tool, but less 200 00:07:52,900 --> 00:07:53,859 so. 201 00:07:53,860 --> 00:07:55,989 But for foremost, 202 00:07:55,990 --> 00:07:57,459 a mathematical description 203 00:07:58,600 --> 00:08:00,789 where you can do calculations 204 00:08:00,790 --> 00:08:03,249 on and graphs have found 205 00:08:03,250 --> 00:08:05,979 applications in a wide range of fields 206 00:08:05,980 --> 00:08:08,349 from chemistry, whether relations between 207 00:08:08,350 --> 00:08:11,589 atoms and molecules can be represented 208 00:08:11,590 --> 00:08:13,179 as such in linguistics or in 209 00:08:13,180 --> 00:08:14,889 neuroscience, where the connections 210 00:08:14,890 --> 00:08:16,959 between neurons are described 211 00:08:16,960 --> 00:08:18,009 via graphs. 212 00:08:18,010 --> 00:08:19,839 And some of you might have seen this 213 00:08:19,840 --> 00:08:22,389 presentation is actually by the NSA. 214 00:08:22,390 --> 00:08:24,789 Um, they have 215 00:08:24,790 --> 00:08:26,679 two slide sets publicly available. 216 00:08:26,680 --> 00:08:29,469 They have a nice little journal 217 00:08:29,470 --> 00:08:31,749 for public consumption on their website. 218 00:08:31,750 --> 00:08:34,989 And in the 2014, 219 00:08:34,990 --> 00:08:37,269 um, volume on Big Data, 220 00:08:37,270 --> 00:08:39,189 they basically explain their interest in 221 00:08:39,190 --> 00:08:41,619 graphs, um, 222 00:08:41,620 --> 00:08:44,019 and this this latest thing from there. 223 00:08:44,020 --> 00:08:46,119 So this is what 224 00:08:46,120 --> 00:08:47,469 in neuroscience parlance, you would call 225 00:08:47,470 --> 00:08:49,569 a connectome. It's the individual 226 00:08:49,570 --> 00:08:51,639 graph of the connections between 227 00:08:51,640 --> 00:08:54,309 the synapses of a given brain, 228 00:08:54,310 --> 00:08:56,469 which is a pretty big data 229 00:08:56,470 --> 00:08:58,539 problem, as you can see. 230 00:08:58,540 --> 00:08:59,540 Um. 231 00:09:02,550 --> 00:09:04,619 And let's not least, of course, you have 232 00:09:04,620 --> 00:09:06,329 graphs and computer science where they 233 00:09:06,330 --> 00:09:08,039 provide one of the most important data 234 00:09:08,040 --> 00:09:10,109 structures, useful file since the Linux 235 00:09:10,110 --> 00:09:11,579 file system, which is a tree, which is a 236 00:09:11,580 --> 00:09:13,889 special form of graphs for dependancy 237 00:09:13,890 --> 00:09:15,329 matters, for control, for 238 00:09:15,330 --> 00:09:17,429 representations, compilers and so forth. 239 00:09:18,450 --> 00:09:20,519 But there's 240 00:09:20,520 --> 00:09:22,139 another important branch of science with 241 00:09:22,140 --> 00:09:24,299 graph theory has made an impact in which, 242 00:09:24,300 --> 00:09:26,129 moreover, is important to the war on 243 00:09:26,130 --> 00:09:27,329 terror discussion. 244 00:09:27,330 --> 00:09:29,699 And the argument I want to make here, and 245 00:09:29,700 --> 00:09:30,929 this is sociology. 246 00:09:30,930 --> 00:09:33,069 And let me just give you 247 00:09:33,070 --> 00:09:35,249 a very short insight about the 248 00:09:35,250 --> 00:09:37,439 weird way this debate 249 00:09:37,440 --> 00:09:40,289 has taken in sociology. 250 00:09:40,290 --> 00:09:41,290 Um. 251 00:09:42,640 --> 00:09:44,709 So they didn't enter 252 00:09:44,710 --> 00:09:46,989 sociology, straightforward 253 00:09:46,990 --> 00:09:48,669 as usual, with new paradigms. 254 00:09:48,670 --> 00:09:50,709 They came basically from the sidelines in 255 00:09:50,710 --> 00:09:53,649 this case in the guise of Jacob Moreno, 256 00:09:53,650 --> 00:09:55,899 a flamboyant figure most 257 00:09:55,900 --> 00:09:57,429 widely known for the innovation of 258 00:09:57,430 --> 00:09:58,430 psychodrama, 259 00:09:59,680 --> 00:10:01,839 which is a sort of improvised 260 00:10:01,840 --> 00:10:03,439 theater. We had a workshop on that or a 261 00:10:03,440 --> 00:10:05,559 talk on that at a conference 262 00:10:05,560 --> 00:10:07,929 here and 263 00:10:07,930 --> 00:10:10,659 who also was a social reformer 264 00:10:10,660 --> 00:10:13,089 engaged with convicts, with orphans, 265 00:10:13,090 --> 00:10:16,059 single mothers and so forth in the 1930s 266 00:10:16,060 --> 00:10:18,159 and 40s of the last century, 267 00:10:18,160 --> 00:10:20,289 and alongside using psychodrama 268 00:10:20,290 --> 00:10:22,959 to educate and empower his clients. 269 00:10:22,960 --> 00:10:25,389 Moreno also developed a technique which 270 00:10:25,390 --> 00:10:28,509 which he called psychological Geographe 271 00:10:28,510 --> 00:10:30,789 or social Mitry. 272 00:10:30,790 --> 00:10:32,919 And basically Moreno depicted 273 00:10:32,920 --> 00:10:34,959 individuals as points interrelations as 274 00:10:34,960 --> 00:10:35,919 airlines. 275 00:10:35,920 --> 00:10:38,079 And here you can see one 276 00:10:38,080 --> 00:10:41,139 of his psychological geography's 277 00:10:41,140 --> 00:10:43,749 from his book, Um, Who shall Survive 278 00:10:43,750 --> 00:10:45,819 in Britain in nineteen thirty 279 00:10:45,820 --> 00:10:48,009 four? And that's a school class 280 00:10:48,010 --> 00:10:50,169 in their first year, the boys on 281 00:10:50,170 --> 00:10:51,939 the left and the girls on the right. 282 00:10:51,940 --> 00:10:54,189 And as you can see, there is quite 283 00:10:54,190 --> 00:10:56,259 a lot of connections going back and forth 284 00:10:56,260 --> 00:10:57,940 between the individuals 285 00:10:58,990 --> 00:11:01,359 in this class and between the genders. 286 00:11:01,360 --> 00:11:03,699 And basically Moreno 287 00:11:03,700 --> 00:11:06,519 mapped out these connections 288 00:11:06,520 --> 00:11:09,249 using, um, 289 00:11:09,250 --> 00:11:11,139 you know, questioning the people who you 290 00:11:11,140 --> 00:11:13,149 do you talk to and how often how often do 291 00:11:13,150 --> 00:11:15,669 you spend time with such and such 292 00:11:15,670 --> 00:11:16,699 and so forth. 293 00:11:16,700 --> 00:11:18,939 And after two years, the same 294 00:11:18,940 --> 00:11:21,399 psychological biography looks like this. 295 00:11:21,400 --> 00:11:23,469 Boys and girls are basically 296 00:11:23,470 --> 00:11:26,679 strictly separated, except for two 297 00:11:26,680 --> 00:11:28,509 individuals which basically cross the 298 00:11:28,510 --> 00:11:29,510 gender line here. 299 00:11:32,170 --> 00:11:34,089 And this already pretty much looks like a 300 00:11:34,090 --> 00:11:36,249 graph by 301 00:11:36,250 --> 00:11:38,469 oilor, but it needed a couple 302 00:11:38,470 --> 00:11:40,629 of trained mathematicians to make that 303 00:11:40,630 --> 00:11:41,799 connection. 304 00:11:41,800 --> 00:11:43,809 And cutting a long story short, if you 305 00:11:43,810 --> 00:11:46,269 want to read up on this, there's a nice 306 00:11:46,270 --> 00:11:48,579 book available for free online by Linton 307 00:11:48,580 --> 00:11:50,859 Friedman on social network theory, 308 00:11:50,860 --> 00:11:53,079 where you can find basically the history 309 00:11:53,080 --> 00:11:54,080 of these developments. 310 00:11:56,530 --> 00:11:58,839 So this mathematician was accessible. 311 00:11:58,840 --> 00:12:01,319 Villars, who worked together 312 00:12:01,320 --> 00:12:03,250 with, quote, Levine and other famous 313 00:12:05,200 --> 00:12:07,419 psychologists of the first half of the 314 00:12:07,420 --> 00:12:08,420 20th century 315 00:12:09,820 --> 00:12:12,189 at the MIT in the Research Center 316 00:12:12,190 --> 00:12:14,379 for Group Dynamics, and together 317 00:12:14,380 --> 00:12:16,749 with two PhD students, 318 00:12:16,750 --> 00:12:18,999 Richard Lewis and Albert Perry, 319 00:12:19,000 --> 00:12:20,799 they provided the first formal 320 00:12:22,060 --> 00:12:23,710 definition or proof 321 00:12:25,400 --> 00:12:27,549 of such, what we would call 322 00:12:27,550 --> 00:12:29,769 social graphs and 323 00:12:29,770 --> 00:12:32,349 more of a, um, invented 324 00:12:32,350 --> 00:12:34,449 more or less a class of measures which we 325 00:12:34,450 --> 00:12:37,659 now call centrality measures 326 00:12:37,660 --> 00:12:41,019 where you can mathematically, 327 00:12:41,020 --> 00:12:43,179 um, measure, 328 00:12:43,180 --> 00:12:45,519 uh, the degree of connectedness and 329 00:12:45,520 --> 00:12:47,889 the importance of a given 330 00:12:47,890 --> 00:12:49,089 word is in a graph. 331 00:12:50,300 --> 00:12:52,729 So the context they did this research 332 00:12:52,730 --> 00:12:54,379 in was basically 333 00:12:55,940 --> 00:12:58,069 companies, um, 334 00:12:58,070 --> 00:13:00,469 which had an interest in knowing 335 00:13:00,470 --> 00:13:02,119 the difference between the formal and 336 00:13:02,120 --> 00:13:04,279 informal hierarchies within 337 00:13:04,280 --> 00:13:06,109 their institutions. 338 00:13:06,110 --> 00:13:08,029 So this is at the heart of Fordism, of 339 00:13:08,030 --> 00:13:10,459 Taylor Ristic, um, uh, 340 00:13:10,460 --> 00:13:12,499 work organization methods where you have 341 00:13:12,500 --> 00:13:14,239 very, very strict hierarchies, the boss 342 00:13:14,240 --> 00:13:16,519 on the top and the worker on the lowest 343 00:13:16,520 --> 00:13:18,739 level and many middle managers in 344 00:13:18,740 --> 00:13:21,139 there. And what the research showed is 345 00:13:21,140 --> 00:13:24,049 that despite his formal organization, 346 00:13:24,050 --> 00:13:26,119 you have very important informal 347 00:13:26,120 --> 00:13:28,489 networks and early 348 00:13:28,490 --> 00:13:30,679 social network research was 349 00:13:30,680 --> 00:13:32,989 concerned, um, with 350 00:13:32,990 --> 00:13:34,459 with the relation between those two 351 00:13:34,460 --> 00:13:36,319 networks and with finding out who is 352 00:13:36,320 --> 00:13:38,809 actually central in the information flow, 353 00:13:38,810 --> 00:13:40,769 uh, in these organizations. 354 00:13:40,770 --> 00:13:42,829 Another, um, 355 00:13:42,830 --> 00:13:44,269 important field of research in this 356 00:13:44,270 --> 00:13:47,179 context was, uh, pharmaceutical 357 00:13:47,180 --> 00:13:49,339 companies who wanted to know 358 00:13:49,340 --> 00:13:51,489 how to push, um, 359 00:13:51,490 --> 00:13:52,490 um. 360 00:13:54,150 --> 00:13:55,150 Uh, 361 00:13:56,230 --> 00:13:58,759 the medications into the market and 362 00:13:58,760 --> 00:14:00,229 for this, they needed to know which 363 00:14:00,230 --> 00:14:02,629 doctors influence opinion, 364 00:14:02,630 --> 00:14:04,969 so they mapped out basically 365 00:14:04,970 --> 00:14:07,009 the informal networks between doctors in 366 00:14:07,010 --> 00:14:09,079 order to find out which doctors 367 00:14:09,080 --> 00:14:11,239 do you need to give basically or do you 368 00:14:11,240 --> 00:14:13,639 need to target for your advertisement? 369 00:14:13,640 --> 00:14:15,589 This is in the 50s and 60s. 370 00:14:15,590 --> 00:14:17,749 And this is the context where central 371 00:14:17,750 --> 00:14:18,799 measures first develop. 372 00:14:18,800 --> 00:14:21,109 They then move over into problematic 373 00:14:21,110 --> 00:14:23,509 research, basically finding 374 00:14:23,510 --> 00:14:25,639 out which academic side to. 375 00:14:27,130 --> 00:14:29,049 Which is another example for more or less 376 00:14:29,050 --> 00:14:30,360 informal networks, 377 00:14:32,290 --> 00:14:33,290 so. 378 00:14:34,940 --> 00:14:36,579 Let me skip a little bit here, 379 00:14:38,360 --> 00:14:40,369 OK, so far for the historical outline, 380 00:14:41,580 --> 00:14:43,639 um, which served to highlight the 381 00:14:43,640 --> 00:14:45,889 context of the emergence of sensuality 382 00:14:45,890 --> 00:14:48,559 measures from organizational psychology 383 00:14:48,560 --> 00:14:49,560 and sociology. 384 00:14:51,080 --> 00:14:52,849 It is the study of oral and written 385 00:14:52,850 --> 00:14:54,919 communication that is important, 386 00:14:54,920 --> 00:14:56,959 uh, in groups where these centrality 387 00:14:56,960 --> 00:14:59,209 measures emerged in Western 388 00:14:59,210 --> 00:15:00,859 morality, became the index for the 389 00:15:00,860 --> 00:15:02,749 stability or instability of an 390 00:15:02,750 --> 00:15:05,119 organization and the tools of means 391 00:15:05,120 --> 00:15:07,819 or what we today call social networks. 392 00:15:07,820 --> 00:15:10,369 Um, so many mathematical 393 00:15:10,370 --> 00:15:13,459 ideas on concepts emerged, 394 00:15:13,460 --> 00:15:16,039 um, in these discourses 395 00:15:16,040 --> 00:15:18,229 and thus way before computers came 396 00:15:18,230 --> 00:15:19,230 into the pictures. 397 00:15:20,150 --> 00:15:22,730 So jumping forward to today or basically. 398 00:15:23,750 --> 00:15:24,949 Let's let's take that back. 399 00:15:24,950 --> 00:15:26,869 Jumping forward to 2000, so 15 years 400 00:15:26,870 --> 00:15:29,059 back, we have seen 401 00:15:29,060 --> 00:15:32,329 the rise of Grauwe industrial complex. 402 00:15:32,330 --> 00:15:34,459 First of all, of course, Google proof 403 00:15:34,460 --> 00:15:35,869 that graphs and graphs and morality 404 00:15:35,870 --> 00:15:38,509 measures and page rank is basically 405 00:15:38,510 --> 00:15:40,639 a graph strategy measure that came 406 00:15:40,640 --> 00:15:43,639 out of leel cuts centrally to measure 407 00:15:43,640 --> 00:15:45,709 idea in which he invented in the 408 00:15:45,710 --> 00:15:47,779 early 50s, also in the context I just 409 00:15:47,780 --> 00:15:49,219 described. 410 00:15:49,220 --> 00:15:51,439 So Google 411 00:15:51,440 --> 00:15:52,440 proved. 412 00:15:54,020 --> 00:15:55,020 That 413 00:15:56,450 --> 00:15:57,979 that these measures are instrumental in 414 00:15:57,980 --> 00:15:59,809 managing the seemingly unmanageable 415 00:15:59,810 --> 00:16:02,029 dynamics of the Internet 416 00:16:02,030 --> 00:16:04,519 and showed that you 417 00:16:04,520 --> 00:16:06,829 could graph out billions 418 00:16:06,830 --> 00:16:08,949 of Web websites and advertisers as 419 00:16:08,950 --> 00:16:11,089 back to back and to back links as 420 00:16:11,090 --> 00:16:13,879 edges, and 421 00:16:13,880 --> 00:16:16,069 that you can provide a tool for navigate 422 00:16:16,070 --> 00:16:18,079 navigation in this context. 423 00:16:18,080 --> 00:16:20,149 And since 2005, we have 424 00:16:20,150 --> 00:16:22,219 seen a number of new actors that 425 00:16:22,220 --> 00:16:24,829 emerged, um, 426 00:16:24,830 --> 00:16:26,389 which managed not only to capture the 427 00:16:26,390 --> 00:16:28,579 relations of websites, but of individuals 428 00:16:28,580 --> 00:16:29,839 into actions and graphs. 429 00:16:31,080 --> 00:16:32,599 Facebook, LinkedIn, Friendster, 430 00:16:32,600 --> 00:16:34,369 Instagram, Tumblr and so forth. 431 00:16:34,370 --> 00:16:36,439 All rely on graphing out their respective 432 00:16:36,440 --> 00:16:37,819 populations. 433 00:16:37,820 --> 00:16:40,069 And in 2011 12, 434 00:16:40,070 --> 00:16:42,289 the consultancy Gartner offered 435 00:16:42,290 --> 00:16:44,389 the following take 436 00:16:44,390 --> 00:16:46,009 on this development. 437 00:16:46,010 --> 00:16:48,199 Um, they basically say 438 00:16:48,200 --> 00:16:50,089 there are five essential strategic graphs 439 00:16:50,090 --> 00:16:51,169 in today's economy. 440 00:16:52,280 --> 00:16:54,619 This is the paper which is paid 441 00:16:54,620 --> 00:16:56,539 content. If somebody can provide me with 442 00:16:56,540 --> 00:16:58,369 the full paper, I should be thankful I 443 00:16:58,370 --> 00:17:00,859 only have part of it. 444 00:17:00,860 --> 00:17:02,959 And so this is 445 00:17:02,960 --> 00:17:05,779 this is from from this report. 446 00:17:05,780 --> 00:17:07,170 So the five, uh, 447 00:17:08,329 --> 00:17:09,919 strategic graphs, the social graph, 448 00:17:09,920 --> 00:17:11,568 Intergraph consumption graph, entrance 449 00:17:11,569 --> 00:17:13,368 graph, a mobile graph. 450 00:17:13,369 --> 00:17:14,989 Um, search. 451 00:17:17,150 --> 00:17:19,219 Would, you know, is probably 452 00:17:19,220 --> 00:17:21,318 intent or interest, that's a little bit 453 00:17:21,319 --> 00:17:23,149 unclear, right, but that's basically what 454 00:17:23,150 --> 00:17:24,150 Google owns. 455 00:17:25,609 --> 00:17:27,919 The consumption 456 00:17:27,920 --> 00:17:29,689 graph, more or less, is owned currently 457 00:17:29,690 --> 00:17:30,989 by Amazon. 458 00:17:30,990 --> 00:17:31,990 Um, 459 00:17:33,320 --> 00:17:35,599 the mobile graph is split across 460 00:17:35,600 --> 00:17:36,800 a couple of telcos. 461 00:17:38,100 --> 00:17:40,279 And the social graph, of course, is 462 00:17:40,280 --> 00:17:42,679 more or less owned by Facebook 463 00:17:42,680 --> 00:17:44,899 and Twitter, at least in the context of 464 00:17:44,900 --> 00:17:47,419 of Western industrialized 465 00:17:47,420 --> 00:17:48,420 nations. 466 00:17:49,400 --> 00:17:51,469 So now, you know, most of you will 467 00:17:51,470 --> 00:17:53,239 probably made the connection. 468 00:17:53,240 --> 00:17:55,279 Is it just the economy that understood 469 00:17:55,280 --> 00:17:57,529 the important or strategic relevance 470 00:17:57,530 --> 00:17:58,579 of graphs? 471 00:17:58,580 --> 00:17:59,580 And certainly not. 472 00:18:00,670 --> 00:18:03,219 So after the Snowden revelations, 473 00:18:03,220 --> 00:18:05,289 the German Bundestag sort 474 00:18:05,290 --> 00:18:06,519 of parliament implemented an 475 00:18:06,520 --> 00:18:08,649 investigative committee 476 00:18:08,650 --> 00:18:11,439 to study the the extent of NSA 477 00:18:11,440 --> 00:18:13,060 and five E surveillance, 478 00:18:15,520 --> 00:18:17,529 especially in Germany. 479 00:18:17,530 --> 00:18:18,999 And part of the proceedings was the 480 00:18:19,000 --> 00:18:21,519 invitation of the NSA whistleblower 481 00:18:21,520 --> 00:18:23,739 William Binney, who also has 482 00:18:23,740 --> 00:18:26,019 been here at the Chaos 483 00:18:26,020 --> 00:18:27,249 Communication Congress, I think, two 484 00:18:27,250 --> 00:18:29,379 years ago, and asked 485 00:18:29,380 --> 00:18:30,969 by the German members of parliament what 486 00:18:30,970 --> 00:18:32,829 they actually did at the NSA. 487 00:18:32,830 --> 00:18:35,169 Vinny answered the following. 488 00:18:35,170 --> 00:18:37,359 Um, we built a relationship 489 00:18:37,360 --> 00:18:39,039 in what we call a graph, a social network 490 00:18:39,040 --> 00:18:41,199 of the world. So 491 00:18:41,200 --> 00:18:43,389 the transcripts of of of 492 00:18:43,390 --> 00:18:45,339 the sitting of the committee are 493 00:18:45,340 --> 00:18:47,139 available via WikiLeaks. 494 00:18:47,140 --> 00:18:49,299 It's 187 pages 495 00:18:49,300 --> 00:18:51,219 paper. It's really interesting to read 496 00:18:51,220 --> 00:18:54,819 because Binney basically explains 497 00:18:54,820 --> 00:18:56,589 back and forth how they graphed out the 498 00:18:56,590 --> 00:18:59,019 world and how important that was. 499 00:18:59,020 --> 00:19:01,329 Um, he mentions 500 00:19:01,330 --> 00:19:03,459 graphs and social graphs, I think 15 501 00:19:03,460 --> 00:19:04,599 times or 16 times. 502 00:19:04,600 --> 00:19:06,879 He never once gets asked actually by 503 00:19:06,880 --> 00:19:08,829 any member of the committee what it 504 00:19:08,830 --> 00:19:09,819 actually is. 505 00:19:09,820 --> 00:19:11,859 They either know and have a quite good 506 00:19:11,860 --> 00:19:13,959 understanding of this concept or 507 00:19:13,960 --> 00:19:15,579 they just ignored what they interested 508 00:19:15,580 --> 00:19:17,559 in. If you read this papers. 509 00:19:17,560 --> 00:19:20,809 Uh, it's interesting to see. 510 00:19:20,810 --> 00:19:21,810 Hmm. 511 00:19:22,480 --> 00:19:24,129 That's a lot of questions about career 512 00:19:24,130 --> 00:19:26,199 pathways within the NSA on the one 513 00:19:26,200 --> 00:19:28,509 hand. So how do you become somebody, 514 00:19:28,510 --> 00:19:30,249 you know, like him, a technical director 515 00:19:30,250 --> 00:19:33,459 at the NSA, how much money he makes? 516 00:19:33,460 --> 00:19:35,589 Um, he says 517 00:19:35,590 --> 00:19:37,659 20 percent less than a U.S. 518 00:19:37,660 --> 00:19:39,159 senator. That's the highest pay grade. 519 00:19:39,160 --> 00:19:41,979 And NSA, um, 520 00:19:41,980 --> 00:19:43,329 and then, of course, they're very 521 00:19:43,330 --> 00:19:44,920 interested in how far 522 00:19:46,180 --> 00:19:48,399 Germany was targeted in this context. 523 00:19:48,400 --> 00:19:50,559 And the other interesting tidbit is 524 00:19:50,560 --> 00:19:52,929 that the NSA, 525 00:19:52,930 --> 00:19:55,299 at least he says, uh, gave 526 00:19:55,300 --> 00:19:57,489 the, uh, Woodenness and the German 527 00:19:57,490 --> 00:19:59,709 equivalent, the source code to this, 528 00:19:59,710 --> 00:20:01,839 um, project, uh, in 529 00:20:01,840 --> 00:20:04,359 the early 2000s. 530 00:20:04,360 --> 00:20:06,849 So Binnie's 531 00:20:06,850 --> 00:20:09,159 project was called Thin Thread, 532 00:20:09,160 --> 00:20:11,349 which basically we 533 00:20:11,350 --> 00:20:13,089 don't know exactly what it is, but 534 00:20:13,090 --> 00:20:15,309 basically seems to be a huge 535 00:20:15,310 --> 00:20:16,310 graph database. 536 00:20:17,360 --> 00:20:18,360 And 537 00:20:19,970 --> 00:20:22,249 the difference to what came later 538 00:20:22,250 --> 00:20:24,949 and why Behney became an NSA 539 00:20:24,950 --> 00:20:25,950 whistleblower 540 00:20:28,310 --> 00:20:30,319 was that it basically did two things. 541 00:20:30,320 --> 00:20:32,149 First of all, it wasn't a full take 542 00:20:32,150 --> 00:20:34,249 approach like the NSA and five eyes 543 00:20:34,250 --> 00:20:36,499 do now. So they didn't take 544 00:20:36,500 --> 00:20:37,729 all of the Internet traffic. 545 00:20:37,730 --> 00:20:39,889 And secondly, it would encrypt 546 00:20:39,890 --> 00:20:41,869 the data of individuals with U.S. 547 00:20:41,870 --> 00:20:42,859 citizenship. 548 00:20:42,860 --> 00:20:43,939 Right. 549 00:20:43,940 --> 00:20:46,189 Um, if 550 00:20:46,190 --> 00:20:48,079 one is to believe Drake and Binney, 551 00:20:48,080 --> 00:20:49,899 Thomas Drake, the other, um, 552 00:20:49,900 --> 00:20:52,669 whistleblower, um, 553 00:20:52,670 --> 00:20:55,099 the successor to Synthroid 554 00:20:55,100 --> 00:20:57,439 was a project called Trailblazer, 555 00:20:57,440 --> 00:20:58,969 which was implemented by private 556 00:20:58,970 --> 00:21:01,909 contractors and basically did away 557 00:21:01,910 --> 00:21:04,009 with these two restrictions that they 558 00:21:04,010 --> 00:21:05,509 had since RedHat. 559 00:21:05,510 --> 00:21:07,639 So that's that's their basic problem, 560 00:21:07,640 --> 00:21:08,640 right? I mean. 561 00:21:10,650 --> 00:21:11,650 It's 562 00:21:12,870 --> 00:21:14,010 it's astonishing 563 00:21:15,540 --> 00:21:17,639 how strongly they oppose 564 00:21:17,640 --> 00:21:20,099 basically Trailblazer, whereas Swinford 565 00:21:20,100 --> 00:21:22,199 already, if you think about it, 566 00:21:22,200 --> 00:21:24,119 was basically graphing all the world and 567 00:21:24,120 --> 00:21:25,179 that's what he says. 568 00:21:25,180 --> 00:21:27,329 And so there's 569 00:21:27,330 --> 00:21:29,609 another interesting element in here, 570 00:21:29,610 --> 00:21:32,069 and this is the timing of this. 571 00:21:32,070 --> 00:21:33,599 Behney mentions in the NSA 572 00:21:35,010 --> 00:21:37,169 report that they started in the early 573 00:21:37,170 --> 00:21:39,239 90s with this project and had 574 00:21:39,240 --> 00:21:41,249 a working prototype in the second half of 575 00:21:41,250 --> 00:21:43,469 the 90s to anyone 576 00:21:43,470 --> 00:21:44,700 familiar with the 577 00:21:45,780 --> 00:21:47,759 infrastructural necessities of graph 578 00:21:47,760 --> 00:21:49,769 processing and to developments in the 579 00:21:49,770 --> 00:21:51,659 commercial field in the past decade. 580 00:21:51,660 --> 00:21:53,999 So what you know, these companies 581 00:21:54,000 --> 00:21:55,409 I just showed from the Gartner report 582 00:21:55,410 --> 00:21:57,509 basically did so key value 583 00:21:57,510 --> 00:21:59,219 stores, mass produce and other means of 584 00:21:59,220 --> 00:22:00,269 distributed computing. 585 00:22:01,320 --> 00:22:02,849 You will agree that it's quite 586 00:22:02,850 --> 00:22:04,589 astonishing that they already, in the 587 00:22:04,590 --> 00:22:06,689 second half of the 90s were basically 588 00:22:06,690 --> 00:22:07,709 able to do this. 589 00:22:11,750 --> 00:22:14,619 So the important part here is 590 00:22:14,620 --> 00:22:16,399 that it's not only businesses, but also 591 00:22:16,400 --> 00:22:18,379 governments that have understood the 592 00:22:18,380 --> 00:22:20,599 value of graft and acquired the means of 593 00:22:20,600 --> 00:22:22,609 generating and exploiting them. 594 00:22:22,610 --> 00:22:24,199 And I believe what has happened in the 595 00:22:24,200 --> 00:22:26,389 past 15 years, basically 596 00:22:26,390 --> 00:22:28,609 between 2000 and today is 597 00:22:28,610 --> 00:22:30,919 an ongoing race to graft the world. 598 00:22:30,920 --> 00:22:33,139 And not only a industrial complex, 599 00:22:33,140 --> 00:22:35,299 but a military academic industrial 600 00:22:35,300 --> 00:22:37,009 complex has basically emerged. 601 00:22:38,240 --> 00:22:39,889 On the commercial side, we have what 602 00:22:39,890 --> 00:22:41,839 Bruce Sterling called suspects I just 603 00:22:41,840 --> 00:22:43,939 talked about on Facebook, Google, Amazon 604 00:22:43,940 --> 00:22:45,709 and so forth. 605 00:22:45,710 --> 00:22:47,749 And if you look beyond basically the 606 00:22:47,750 --> 00:22:48,750 Western world, 607 00:22:50,510 --> 00:22:52,699 you will see that China has own 608 00:22:52,700 --> 00:22:55,279 Tencent, Renren, Baijiu, Sina Weibo 609 00:22:55,280 --> 00:22:56,419 and the Russians have V.K. 610 00:22:56,420 --> 00:22:58,549 Contactor, A Momir as 611 00:22:58,550 --> 00:22:59,569 Facebook equivalents. 612 00:22:59,570 --> 00:23:02,029 And basically these are all in 613 00:23:02,030 --> 00:23:03,859 all invitations for you to become 614 00:23:03,860 --> 00:23:04,939 graphed. Right. 615 00:23:04,940 --> 00:23:06,979 Veeco contact you get MP three for free 616 00:23:06,980 --> 00:23:09,079 and they're happy to have you as a node 617 00:23:09,080 --> 00:23:10,080 in their system. 618 00:23:11,660 --> 00:23:13,459 On the government side, we have the NSA 619 00:23:13,460 --> 00:23:15,529 and five eyes which are in the 620 00:23:15,530 --> 00:23:16,939 business of graphing the world. 621 00:23:16,940 --> 00:23:19,279 And the papers from the Snowden archives, 622 00:23:19,280 --> 00:23:21,559 if you look at them so far, 623 00:23:21,560 --> 00:23:23,899 only provide indirect 624 00:23:23,900 --> 00:23:26,029 material because they 625 00:23:26,030 --> 00:23:28,969 concentrate more on the means of 626 00:23:28,970 --> 00:23:30,499 access and collection of data. 627 00:23:30,500 --> 00:23:32,569 So PRISM Buruma, Fairview 628 00:23:32,570 --> 00:23:34,729 and all these acronyms basically 629 00:23:34,730 --> 00:23:36,859 describe how to get the data. 630 00:23:36,860 --> 00:23:39,489 And we know relatively little, 631 00:23:39,490 --> 00:23:41,989 at least on the basis of those papers 632 00:23:41,990 --> 00:23:45,079 on the aggregation and analytical tools. 633 00:23:45,080 --> 00:23:47,539 William Binney, to come back to him 634 00:23:47,540 --> 00:23:49,969 last year in 2014, 635 00:23:49,970 --> 00:23:52,249 explained the threat and the Snowden 636 00:23:52,250 --> 00:23:54,319 files to select Orlean's at a dinner in 637 00:23:54,320 --> 00:23:55,999 Washington and pointed out that Stello 638 00:23:56,000 --> 00:23:58,369 went mainly in Marena, are the graphing 639 00:23:58,370 --> 00:24:00,769 tools used for discovery 640 00:24:00,770 --> 00:24:03,409 and development of targets by the NSA. 641 00:24:03,410 --> 00:24:05,479 And this again 642 00:24:05,480 --> 00:24:07,459 slides which you will find on the net. 643 00:24:07,460 --> 00:24:09,469 It's really interesting to read through 644 00:24:09,470 --> 00:24:11,779 because it's basically a Q and A where 645 00:24:11,780 --> 00:24:13,999 Binney explains how this 646 00:24:14,000 --> 00:24:15,379 is supposed to work. 647 00:24:15,380 --> 00:24:17,479 Um, we don't really know if 648 00:24:17,480 --> 00:24:18,680 he actually tells the truth, but 649 00:24:19,970 --> 00:24:21,769 what he says is basically you have you 650 00:24:21,770 --> 00:24:23,959 graph as 651 00:24:23,960 --> 00:24:25,399 much as you can and then you use 652 00:24:25,400 --> 00:24:28,279 centrality measures in order to 653 00:24:28,280 --> 00:24:30,499 create what he calls sort of suspect 654 00:24:30,500 --> 00:24:32,599 is the, um, 655 00:24:32,600 --> 00:24:35,209 in the green 656 00:24:35,210 --> 00:24:36,829 bubble here. 657 00:24:36,830 --> 00:24:39,079 And these individuals are the ones which 658 00:24:39,080 --> 00:24:41,269 you follow up further on and profile out. 659 00:24:42,410 --> 00:24:44,629 So but what do you do once you have 660 00:24:44,630 --> 00:24:47,779 something like these sort of suspects, 661 00:24:47,780 --> 00:24:49,909 a graph of a terror network, if 662 00:24:49,910 --> 00:24:51,559 you want to call it that? 663 00:24:51,560 --> 00:24:53,809 And this is a question of much debate in 664 00:24:53,810 --> 00:24:56,509 both doctorial doctrinal literature 665 00:24:56,510 --> 00:24:58,669 as well in the academic research 666 00:24:58,670 --> 00:25:00,289 I mentioned in the beginning. 667 00:25:00,290 --> 00:25:02,420 So the military academic research 668 00:25:03,740 --> 00:25:05,509 around so-called dark networks. 669 00:25:05,510 --> 00:25:07,359 So this would be a dark network in the in 670 00:25:07,360 --> 00:25:09,919 the bubble here, the bad guys 671 00:25:09,920 --> 00:25:12,199 and a striking example of the thinking. 672 00:25:12,200 --> 00:25:14,299 So there's there's tons of papers I just 673 00:25:14,300 --> 00:25:17,059 pulled out to to show you a little bit 674 00:25:17,060 --> 00:25:18,529 how they conceptualize that 675 00:25:19,840 --> 00:25:21,409 striking example is the U.S. 676 00:25:21,410 --> 00:25:23,959 Counterinsurgency Field Manual published 677 00:25:23,960 --> 00:25:25,949 in 2006 to much public fanfare. 678 00:25:25,950 --> 00:25:27,649 I don't know how many Americans are here. 679 00:25:27,650 --> 00:25:29,389 Maybe you remember, because even The New 680 00:25:29,390 --> 00:25:32,059 York Times reported about this. 681 00:25:32,060 --> 00:25:33,889 The U.S. military I mean, this is in the 682 00:25:33,890 --> 00:25:36,949 depth of of Iraq and Afghanistan wars. 683 00:25:36,950 --> 00:25:38,989 They had no counter insurgency, 684 00:25:38,990 --> 00:25:40,759 insurgency doctrine. 685 00:25:40,760 --> 00:25:42,409 The last doctrine was was written in the 686 00:25:42,410 --> 00:25:44,839 70s in the context of the Vietnam War. 687 00:25:44,840 --> 00:25:46,939 And then after, you know, the 688 00:25:46,940 --> 00:25:49,129 way that war went, nobody wanted to do 689 00:25:49,130 --> 00:25:50,239 Quoin, as they call it, 690 00:25:50,240 --> 00:25:51,749 counterinsurgency. 691 00:25:51,750 --> 00:25:54,079 Um, but 692 00:25:54,080 --> 00:25:56,389 three or four years into 693 00:25:56,390 --> 00:25:58,069 these wars, they figured out we probably 694 00:25:58,070 --> 00:25:58,969 need a doctrine. 695 00:25:58,970 --> 00:26:01,369 So they had 696 00:26:01,370 --> 00:26:04,249 this guy, right. One, David Petraeus, 697 00:26:04,250 --> 00:26:06,740 who later became the CIA director until 698 00:26:07,970 --> 00:26:09,349 he was basically removed from that 699 00:26:09,350 --> 00:26:11,389 position in 2012. 700 00:26:11,390 --> 00:26:12,769 This is another interesting book which 701 00:26:12,770 --> 00:26:14,899 you really should read, even 702 00:26:14,900 --> 00:26:16,009 though it's an allergy. 703 00:26:16,010 --> 00:26:18,259 And, you know, Fred Kaplan is 704 00:26:19,310 --> 00:26:21,439 not a good author, but it's if only 705 00:26:21,440 --> 00:26:23,449 if only. Well, he you know, he's one of 706 00:26:23,450 --> 00:26:24,950 those military 707 00:26:26,090 --> 00:26:27,289 experts. 708 00:26:27,290 --> 00:26:29,989 But if only half of the stories 709 00:26:29,990 --> 00:26:32,179 about the Iraq war, he 710 00:26:32,180 --> 00:26:34,129 tells, are true, it's pretty amazing what 711 00:26:34,130 --> 00:26:36,649 happened in the political 712 00:26:36,650 --> 00:26:38,059 system in the U.S. at that time. 713 00:26:38,060 --> 00:26:39,949 And he also, Petraeus writes, does this 714 00:26:39,950 --> 00:26:42,079 counterinsurgency manual and of 715 00:26:42,080 --> 00:26:43,729 course, there's an appendix of sort on 716 00:26:43,730 --> 00:26:45,619 social network analysis and exploitation 717 00:26:45,620 --> 00:26:47,959 in there, which is really, really short. 718 00:26:47,960 --> 00:26:49,339 Just a couple of pictures. 719 00:26:49,340 --> 00:26:50,660 And here's an example 720 00:26:51,680 --> 00:26:53,359 of how they envision this. 721 00:26:57,030 --> 00:26:59,279 So basically, what you see 722 00:26:59,280 --> 00:27:01,499 is a graph of some kind 723 00:27:01,500 --> 00:27:03,809 of insurgent terrorists 724 00:27:03,810 --> 00:27:05,939 or whatever you call it, subgroup, 725 00:27:05,940 --> 00:27:08,579 which they map out who 726 00:27:08,580 --> 00:27:11,219 and now the important doctrinal, um, 727 00:27:11,220 --> 00:27:14,059 uh, notion is shaping. 728 00:27:14,060 --> 00:27:16,049 Right. So what you want to do in a 729 00:27:16,050 --> 00:27:17,429 counter-insurgency or war on terror 730 00:27:17,430 --> 00:27:19,649 context is shaping the graph of of 731 00:27:19,650 --> 00:27:20,699 the enemy. 732 00:27:20,700 --> 00:27:22,889 So you have 733 00:27:22,890 --> 00:27:25,139 the network on the left side 734 00:27:25,140 --> 00:27:26,839 and you implement some measure like that 735 00:27:26,840 --> 00:27:27,989 in the surge. 736 00:27:27,990 --> 00:27:30,059 Then you grab out a couple of people to 737 00:27:30,060 --> 00:27:32,339 start a food distribution and during 738 00:27:32,340 --> 00:27:34,739 this time basically map out 739 00:27:34,740 --> 00:27:37,139 the communications, 740 00:27:37,140 --> 00:27:40,019 graph out the insurgents. 741 00:27:40,020 --> 00:27:42,419 And the idea is to shape 742 00:27:42,420 --> 00:27:44,879 the graph in a way that it reveals 743 00:27:44,880 --> 00:27:46,799 information about who is important in 744 00:27:46,800 --> 00:27:47,729 this context. 745 00:27:47,730 --> 00:27:50,159 If you can use, of course, the centrality 746 00:27:50,160 --> 00:27:51,160 measures I 747 00:27:52,290 --> 00:27:53,669 mentioned in the beginning. 748 00:27:53,670 --> 00:27:55,559 So, again, just the shaping and there's 749 00:27:55,560 --> 00:27:57,659 much debate about how to do 750 00:27:57,660 --> 00:27:58,889 this, actually. 751 00:27:58,890 --> 00:28:01,049 And another interesting 752 00:28:01,050 --> 00:28:03,569 example for what I just described 753 00:28:03,570 --> 00:28:05,609 is this paper that came out of West 754 00:28:05,610 --> 00:28:07,829 Point, the Network Science Center 755 00:28:07,830 --> 00:28:10,109 there are shaping operations 756 00:28:10,110 --> 00:28:12,479 to attack robust terror networks. 757 00:28:13,640 --> 00:28:15,739 And the use of publicly available 758 00:28:15,740 --> 00:28:18,199 data set the so-called Tanzania 759 00:28:18,200 --> 00:28:20,779 data set, which is a graph of al-Qaida 760 00:28:20,780 --> 00:28:22,999 1998 in the context of the 761 00:28:23,000 --> 00:28:24,019 bombing of the U.S. 762 00:28:24,020 --> 00:28:25,369 embassy. 763 00:28:25,370 --> 00:28:27,619 And what they offer 764 00:28:27,620 --> 00:28:29,899 is basically an algorithm that 765 00:28:29,900 --> 00:28:32,029 promises to automate the 766 00:28:32,030 --> 00:28:34,429 decisions which nodes 767 00:28:34,430 --> 00:28:36,559 to attack in order to shape your 768 00:28:36,560 --> 00:28:37,519 graph. 769 00:28:37,520 --> 00:28:39,739 And the idea is to render 770 00:28:39,740 --> 00:28:41,269 the graph more 771 00:28:42,590 --> 00:28:43,590 fragile 772 00:28:45,170 --> 00:28:47,449 by generating more sensuality, 773 00:28:47,450 --> 00:28:49,759 by making it less distributed, more 774 00:28:49,760 --> 00:28:50,569 central. 775 00:28:50,570 --> 00:28:52,789 So basically, the algorithm 776 00:28:52,790 --> 00:28:54,349 tells you which nodes do you need to 777 00:28:54,350 --> 00:28:56,419 attack in order to have a star shaped 778 00:28:56,420 --> 00:28:58,729 network in the end and then know 779 00:28:58,730 --> 00:29:01,069 who is actually the most important person 780 00:29:01,070 --> 00:29:02,479 in here? 781 00:29:02,480 --> 00:29:04,189 Because, I mean, the quality is bad here 782 00:29:04,190 --> 00:29:06,049 and it's really tiny in the paper. 783 00:29:06,050 --> 00:29:08,479 But basically in a they don't know who 784 00:29:08,480 --> 00:29:10,609 who bin Laden is and they want 785 00:29:10,610 --> 00:29:12,049 to know who that is. 786 00:29:12,050 --> 00:29:14,429 And this algorithm basically offers 787 00:29:14,430 --> 00:29:16,369 attack. This notice is not this node. 788 00:29:16,370 --> 00:29:18,859 And eventually the communications 789 00:29:18,860 --> 00:29:20,419 will lead to bin Laden, which is actually 790 00:29:20,420 --> 00:29:22,729 this, uh, individuals down 791 00:29:22,730 --> 00:29:25,819 here. So they call this fragility, 792 00:29:25,820 --> 00:29:28,159 which is basically the inverse, you know, 793 00:29:28,160 --> 00:29:30,529 network the inversion, but tied 794 00:29:30,530 --> 00:29:32,509 to sensuality and 795 00:29:33,680 --> 00:29:35,389 which seeks to find a set of nodes whose 796 00:29:35,390 --> 00:29:37,399 removal maximize the networks wide 797 00:29:37,400 --> 00:29:38,629 sensuality. 798 00:29:38,630 --> 00:29:40,189 And we also include the problem of no 799 00:29:40,190 --> 00:29:41,239 strike list. 800 00:29:41,240 --> 00:29:43,039 So basically, the organization will take 801 00:29:43,040 --> 00:29:45,259 those people out, which you which are not 802 00:29:45,260 --> 00:29:46,260 supposed to attack. 803 00:29:48,020 --> 00:29:50,149 And this is because we were talking 804 00:29:50,150 --> 00:29:52,849 of terror and social networks of, uh, 805 00:29:52,850 --> 00:29:54,499 often includes restrictions against 806 00:29:54,500 --> 00:29:55,789 certain individuals and so forth. 807 00:29:55,790 --> 00:29:57,709 They also prove even prove that this is 808 00:29:57,710 --> 00:29:59,929 an entry NP complete 809 00:29:59,930 --> 00:30:00,930 problem. 810 00:30:01,920 --> 00:30:03,979 So read this paper 811 00:30:03,980 --> 00:30:05,179 is really interesting and there's many, 812 00:30:05,180 --> 00:30:06,769 many more. I think this is one of the 813 00:30:06,770 --> 00:30:07,770 more important ones. 814 00:30:09,350 --> 00:30:10,999 So this is basically an algorithm that 815 00:30:11,000 --> 00:30:13,279 helps decision making in in shaping 816 00:30:13,280 --> 00:30:14,699 grafs. Right. 817 00:30:14,700 --> 00:30:16,759 So but how do you actually 818 00:30:16,760 --> 00:30:18,769 get those graphs when people are not 819 00:30:18,770 --> 00:30:20,839 online all the time like we are and have 820 00:30:20,840 --> 00:30:22,669 those little cell phones in them and are 821 00:30:22,670 --> 00:30:24,259 on Facebook and so forth? 822 00:30:24,260 --> 00:30:26,479 And that was the problem basically that 823 00:30:26,480 --> 00:30:28,609 you had in a country like Afghanistan 824 00:30:28,610 --> 00:30:30,769 and what the US military did, what 825 00:30:30,770 --> 00:30:32,659 basically was basically to hire 826 00:30:32,660 --> 00:30:34,969 anthropologists, uh, 827 00:30:34,970 --> 00:30:36,950 they sent anthropologists in field and 828 00:30:38,330 --> 00:30:40,219 asked people those questions. 829 00:30:40,220 --> 00:30:42,589 You, uh, what are so 830 00:30:42,590 --> 00:30:44,719 the, you know, people in the villages and 831 00:30:44,720 --> 00:30:46,939 so forth, um, 832 00:30:46,940 --> 00:30:48,769 what five people here have, you know, the 833 00:30:48,770 --> 00:30:51,319 longest, etc. 834 00:30:51,320 --> 00:30:53,569 and you enter that in a tool like 835 00:30:53,570 --> 00:30:55,729 this where 836 00:30:55,730 --> 00:30:58,189 you basically generate a graph 837 00:30:58,190 --> 00:31:00,289 of the given population 838 00:31:00,290 --> 00:31:02,359 in your area and hence can 839 00:31:02,360 --> 00:31:03,259 shape it. 840 00:31:03,260 --> 00:31:05,599 This is the map map human 841 00:31:05,600 --> 00:31:07,699 terrain. They called Courtice 842 00:31:07,700 --> 00:31:09,649 the human terrain system, which was 843 00:31:09,650 --> 00:31:12,089 closed down a year ago. 844 00:31:12,090 --> 00:31:14,389 It ran from 2006 to 2014, 845 00:31:14,390 --> 00:31:15,829 very controversial. 846 00:31:15,830 --> 00:31:17,809 The American Anthropological Association 847 00:31:17,810 --> 00:31:21,259 actually came out to oppose 848 00:31:21,260 --> 00:31:23,119 the system and 849 00:31:25,160 --> 00:31:27,049 protested strongly against it. 850 00:31:27,050 --> 00:31:28,849 There's an interesting movie about it 851 00:31:28,850 --> 00:31:30,859 which I recommend to watch. 852 00:31:30,860 --> 00:31:33,019 It's actually not a um. 853 00:31:33,020 --> 00:31:34,729 I'm not sure if they actually use the 854 00:31:34,730 --> 00:31:35,899 software. 855 00:31:35,900 --> 00:31:36,900 This is just taking. 856 00:31:38,320 --> 00:31:40,419 From a handbook of the map 857 00:31:40,420 --> 00:31:43,329 human terrain system, 858 00:31:43,330 --> 00:31:45,069 but what they definitely used and what 859 00:31:45,070 --> 00:31:47,349 you will find in many, many papers is 860 00:31:47,350 --> 00:31:49,179 IBM's analyst notebook, which is 861 00:31:49,180 --> 00:31:51,369 basically a software 862 00:31:51,370 --> 00:31:53,409 tool which offers you to map all social 863 00:31:53,410 --> 00:31:55,479 graphs and you get the centrality 864 00:31:55,480 --> 00:31:57,609 measures and can use 865 00:31:57,610 --> 00:31:59,769 that in order to track 866 00:31:59,770 --> 00:32:01,149 and profile individuals. 867 00:32:01,150 --> 00:32:03,399 And of course, what they use is 868 00:32:03,400 --> 00:32:05,679 here, which I think owns the market 869 00:32:05,680 --> 00:32:07,809 for for graphing technologies in 870 00:32:07,810 --> 00:32:09,129 this context. 871 00:32:09,130 --> 00:32:10,130 Um, 872 00:32:11,410 --> 00:32:13,449 anybody who actually has worked on a 873 00:32:13,450 --> 00:32:14,529 plant workstation. 874 00:32:16,130 --> 00:32:17,690 I'm still waiting to find somebody. 875 00:32:19,370 --> 00:32:21,589 I strongly encourage you to watch 876 00:32:21,590 --> 00:32:23,899 the YouTube, the YouTube video 877 00:32:23,900 --> 00:32:26,419 on counter-terrorism that volunteers 878 00:32:26,420 --> 00:32:28,189 up, it's really interesting. 879 00:32:28,190 --> 00:32:30,229 So Pawlenty is basically appetitive, 880 00:32:30,230 --> 00:32:33,139 founded START, a startup 881 00:32:33,140 --> 00:32:35,239 found in 2006 882 00:32:35,240 --> 00:32:38,089 which has seen a meteoric rise and 883 00:32:38,090 --> 00:32:40,399 which started out as 884 00:32:40,400 --> 00:32:42,529 with a software product allowing graphic 885 00:32:42,530 --> 00:32:45,109 graphing for government entities 886 00:32:45,110 --> 00:32:46,699 and now has branched out in all kinds of 887 00:32:46,700 --> 00:32:49,219 fields, mostly fraud detection 888 00:32:49,220 --> 00:32:51,409 in banks. But they also do philanthropic 889 00:32:51,410 --> 00:32:53,689 engineering. So if you want to map out 890 00:32:53,690 --> 00:32:56,059 your clients 891 00:32:56,060 --> 00:32:57,619 as a philanthropist, philanthropist, 892 00:32:57,620 --> 00:32:59,040 Pawlenty has a solution for you. 893 00:33:00,170 --> 00:33:02,449 OK, so this 894 00:33:02,450 --> 00:33:04,339 is the analytical level. 895 00:33:04,340 --> 00:33:06,409 But of course, you know, what about 896 00:33:06,410 --> 00:33:07,910 the drones and 897 00:33:09,080 --> 00:33:11,939 what I think we have to think about, 898 00:33:11,940 --> 00:33:14,239 uh, drones as is 899 00:33:14,240 --> 00:33:16,789 less tools of visual collection, 900 00:33:16,790 --> 00:33:18,469 but more tools of collection of 901 00:33:18,470 --> 00:33:20,839 communication, metadata, basically 902 00:33:20,840 --> 00:33:21,919 cell phone data. 903 00:33:21,920 --> 00:33:24,019 And if you look in all those 904 00:33:24,020 --> 00:33:25,849 presentations which float around on the 905 00:33:25,850 --> 00:33:28,189 Web, you will stumble upon 906 00:33:28,190 --> 00:33:30,499 all those plots that the drones carry. 907 00:33:30,500 --> 00:33:32,329 Of course, they always have a camera and 908 00:33:32,330 --> 00:33:34,699 they have a visual tracking tool 909 00:33:34,700 --> 00:33:36,409 on board, but they also have the 910 00:33:36,410 --> 00:33:37,999 communications and networking. 911 00:33:38,000 --> 00:33:39,739 This is the Argos pod, which is a wide 912 00:33:39,740 --> 00:33:40,740 area. 913 00:33:42,010 --> 00:33:44,109 Surveillance system, they 914 00:33:44,110 --> 00:33:45,589 have the communications and networking 915 00:33:45,590 --> 00:33:47,799 spot, which basically surveillance in 916 00:33:47,800 --> 00:33:50,049 all the communication data in a given 917 00:33:50,050 --> 00:33:53,289 area, and that data 918 00:33:53,290 --> 00:33:54,290 gets aggregated 919 00:33:55,360 --> 00:33:56,979 and graphed out. 920 00:33:56,980 --> 00:33:59,259 So I think we have to think about drones 921 00:33:59,260 --> 00:34:01,209 as crawlers, just as the Googlebot, as a 922 00:34:01,210 --> 00:34:03,489 crawler that indexes Web pages. 923 00:34:03,490 --> 00:34:05,679 And Dilling's what drones 924 00:34:05,680 --> 00:34:07,839 basically do is they crawl an area 925 00:34:07,840 --> 00:34:09,549 in index all the communication that is 926 00:34:09,550 --> 00:34:11,829 going on, allowing you to build 927 00:34:11,830 --> 00:34:13,928 up a graph of 928 00:34:13,929 --> 00:34:14,929 that. 929 00:34:15,360 --> 00:34:17,559 Um, of course, there's also 930 00:34:17,560 --> 00:34:19,448 the other side. They not only collect, 931 00:34:19,449 --> 00:34:20,619 they also target 932 00:34:21,909 --> 00:34:24,039 and nodes of the graphs they have 933 00:34:24,040 --> 00:34:25,089 created. 934 00:34:25,090 --> 00:34:26,738 And this is another strand of the 935 00:34:26,739 --> 00:34:28,869 doctrinal debate that has been going 936 00:34:28,870 --> 00:34:31,299 on in the past 10 years because 937 00:34:31,300 --> 00:34:33,069 the initial arment of drones were the 938 00:34:33,070 --> 00:34:34,839 Hellfire missiles, which were anti-tank 939 00:34:34,840 --> 00:34:36,849 missiles developed in the 80s, which have 940 00:34:36,850 --> 00:34:38,979 quite a strong warhead where you 941 00:34:38,980 --> 00:34:41,169 have the problem of 942 00:34:41,170 --> 00:34:42,789 collateral damage. 943 00:34:42,790 --> 00:34:45,249 And what has basically 944 00:34:45,250 --> 00:34:47,439 happened since 2005 is a race 945 00:34:47,440 --> 00:34:49,779 to find, um, ammunitions 946 00:34:49,780 --> 00:34:51,519 which allow you to target individuals. 947 00:34:51,520 --> 00:34:53,619 So that's called discrete effects. 948 00:34:53,620 --> 00:34:54,759 Right. 949 00:34:54,760 --> 00:34:57,069 This is, um, 950 00:34:57,070 --> 00:34:59,529 so personal targets, 951 00:34:59,530 --> 00:35:01,449 the fact that his personal is plus less 952 00:35:01,450 --> 00:35:03,039 than leave them also you I don't know 953 00:35:03,040 --> 00:35:05,109 what less lethal means if something 954 00:35:05,110 --> 00:35:06,110 like that hits you. 955 00:35:07,120 --> 00:35:09,339 And I mean, this is just one example. 956 00:35:09,340 --> 00:35:11,889 There are many competing systems. 957 00:35:11,890 --> 00:35:14,139 Um, this this call this is called 958 00:35:14,140 --> 00:35:15,249 Viper Strike. 959 00:35:15,250 --> 00:35:17,979 And they all come with this nice little, 960 00:35:17,980 --> 00:35:20,049 uh, circles, which 961 00:35:20,050 --> 00:35:21,729 tell you the blast areas of different 962 00:35:21,730 --> 00:35:23,499 ammunitions. And basically, as you can 963 00:35:23,500 --> 00:35:25,899 see, it's it's relatively 964 00:35:25,900 --> 00:35:28,179 small compared to, uh, 965 00:35:28,180 --> 00:35:30,129 to a Hellfire missile or other 966 00:35:30,130 --> 00:35:31,089 ammunitions. 967 00:35:31,090 --> 00:35:33,399 So this is basically what shaping can be. 968 00:35:33,400 --> 00:35:35,769 There are two ways to shape a graph. 969 00:35:35,770 --> 00:35:37,449 Uh, kinetic and non kinetic. 970 00:35:37,450 --> 00:35:38,649 Non kinetic would be the food 971 00:35:38,650 --> 00:35:39,999 distribution shown in the 972 00:35:40,000 --> 00:35:42,159 counterinsurgency manual or 973 00:35:42,160 --> 00:35:44,559 the snatch and grab and kinetic 974 00:35:44,560 --> 00:35:46,719 shaping is basically the 975 00:35:46,720 --> 00:35:49,059 euphemism for killing people with tools 976 00:35:49,060 --> 00:35:50,499 like that. 977 00:35:50,500 --> 00:35:52,569 And I think what we have seen, if 978 00:35:52,570 --> 00:35:54,759 you think about it in a doctrinal context 979 00:35:54,760 --> 00:35:56,499 and also in the history of aerial 980 00:35:56,500 --> 00:35:58,629 warfare, is that 981 00:35:58,630 --> 00:36:00,369 we have a shift for weapons of mass 982 00:36:00,370 --> 00:36:02,949 destructions to weapons of individual 983 00:36:02,950 --> 00:36:05,019 destruction. This is basically they went 984 00:36:05,020 --> 00:36:07,149 into Iraq to find weapons of 985 00:36:07,150 --> 00:36:08,859 mass destruction. They came out of it 986 00:36:08,860 --> 00:36:10,599 with weapons of individual destruction. 987 00:36:10,600 --> 00:36:12,159 This is, I think, what has happened. 988 00:36:12,160 --> 00:36:13,239 Right. 989 00:36:13,240 --> 00:36:14,379 So to come to an end. 990 00:36:19,730 --> 00:36:20,730 Thank you, 991 00:36:21,920 --> 00:36:24,139 I think and you know, 992 00:36:24,140 --> 00:36:25,999 this is ongoing research and basically 993 00:36:26,000 --> 00:36:27,589 presenting in my toolbox and the ideas 994 00:36:27,590 --> 00:36:29,149 I'm developing right now, and this, of 995 00:36:29,150 --> 00:36:31,039 course, needs to be fleshed out. 996 00:36:31,040 --> 00:36:33,259 But what has happened since 997 00:36:33,260 --> 00:36:35,210 2000 may be a little bit earlier, 998 00:36:36,350 --> 00:36:38,719 which at least is what 999 00:36:38,720 --> 00:36:40,909 Binnie's what 1000 00:36:40,910 --> 00:36:42,319 knew would indicate. What many said would 1001 00:36:42,320 --> 00:36:44,689 indicate is that Graff's have emerged 1002 00:36:44,690 --> 00:36:47,209 as strategic or geostrategic assets 1003 00:36:47,210 --> 00:36:49,669 and that there is a race by governmental 1004 00:36:49,670 --> 00:36:51,799 but also business entities to 1005 00:36:51,800 --> 00:36:53,899 basically gref all the world. 1006 00:36:53,900 --> 00:36:55,009 Who owns the biggest gref. 1007 00:36:55,010 --> 00:36:57,199 That's important, right? Facebook has 1008 00:36:57,200 --> 00:36:59,029 one point two or one point four billion 1009 00:36:59,030 --> 00:37:00,030 people graphed out. 1010 00:37:01,490 --> 00:37:03,319 Also, there's this problem that they blur 1011 00:37:03,320 --> 00:37:04,639 the distinction between maps and 1012 00:37:04,640 --> 00:37:06,649 territory because the graph is not just a 1013 00:37:06,650 --> 00:37:07,789 map, right? 1014 00:37:07,790 --> 00:37:09,679 It's not just a visual representation. 1015 00:37:09,680 --> 00:37:11,749 It's actually a dynamic representation 1016 00:37:11,750 --> 00:37:13,849 of the activity, of the 1017 00:37:13,850 --> 00:37:15,529 communication, of the relations of the 1018 00:37:15,530 --> 00:37:16,530 individuals, 1019 00:37:17,660 --> 00:37:20,209 which you, of course, already shape 1020 00:37:20,210 --> 00:37:22,459 in the moment you observe it, 1021 00:37:22,460 --> 00:37:24,889 but which provides a different 1022 00:37:24,890 --> 00:37:26,989 level of actionable intelligence than 1023 00:37:26,990 --> 00:37:28,079 a map would do. 1024 00:37:28,080 --> 00:37:29,239 Right. 1025 00:37:29,240 --> 00:37:31,279 And last not least, is really difficult 1026 00:37:31,280 --> 00:37:32,689 to obfuscate graphs. 1027 00:37:32,690 --> 00:37:34,249 It's really difficult to get away from 1028 00:37:34,250 --> 00:37:36,289 that. You can encrypt your communication, 1029 00:37:36,290 --> 00:37:38,779 but you still would be visible as nodes 1030 00:37:38,780 --> 00:37:40,099 who communicate. 1031 00:37:40,100 --> 00:37:42,139 And would Binney and other people also 1032 00:37:42,140 --> 00:37:43,909 Snowden always highlight and that's 1033 00:37:43,910 --> 00:37:45,919 basically the metadata aspect, that 1034 00:37:45,920 --> 00:37:48,679 aspect. That's that's what graphs do. 1035 00:37:48,680 --> 00:37:50,809 They ingest metadata and 1036 00:37:50,810 --> 00:37:53,059 give you an idea who is connected 1037 00:37:53,060 --> 00:37:55,389 to whom in 1038 00:37:55,390 --> 00:37:57,499 in what form of relatedness, how close, 1039 00:37:57,500 --> 00:37:58,500 how far and so forth. 1040 00:37:59,570 --> 00:38:02,029 So basically, it's very difficult 1041 00:38:02,030 --> 00:38:03,050 to to get 1042 00:38:04,790 --> 00:38:07,249 to stay out of Graff's or 1043 00:38:07,250 --> 00:38:09,469 to use to develop tools which 1044 00:38:09,470 --> 00:38:10,470 will actually 1045 00:38:12,850 --> 00:38:14,929 create false leads or any kind of other, 1046 00:38:14,930 --> 00:38:17,089 uh, uh, wrong impression of you 1047 00:38:17,090 --> 00:38:18,169 in a graph. 1048 00:38:18,170 --> 00:38:20,419 So to wrap this up, I think 1049 00:38:20,420 --> 00:38:23,269 we need to think about, um, 1050 00:38:23,270 --> 00:38:25,429 who owns graphs and 1051 00:38:25,430 --> 00:38:27,049 what does that mean for us and how can we 1052 00:38:27,050 --> 00:38:28,280 deal with that? Thank you. 1053 00:38:41,170 --> 00:38:44,259 We have 20 minutes for questions, 1054 00:38:44,260 --> 00:38:46,509 so if you have questions, please go 1055 00:38:46,510 --> 00:38:49,569 to one of the four microphones 1056 00:38:49,570 --> 00:38:51,550 in the aisles. 1057 00:38:56,280 --> 00:38:57,989 We're going to start with a question from 1058 00:38:57,990 --> 00:38:59,669 the Internet where everyone else is 1059 00:38:59,670 --> 00:39:01,649 getting ready for the microphones, the 1060 00:39:01,650 --> 00:39:02,809 Internet, please. 1061 00:39:02,810 --> 00:39:04,679 We got only one question here. 1062 00:39:04,680 --> 00:39:06,889 Could you tell us the name of the YouTube 1063 00:39:06,890 --> 00:39:09,239 movie about human talent graphics 1064 00:39:09,240 --> 00:39:10,589 software? 1065 00:39:10,590 --> 00:39:12,779 OK, so what is the 1066 00:39:12,780 --> 00:39:14,849 volunteer demo 1067 00:39:14,850 --> 00:39:16,079 on counter-insurgency? 1068 00:39:17,250 --> 00:39:19,529 And I actually 1069 00:39:19,530 --> 00:39:20,939 forgot the name of the Human Terrain 1070 00:39:20,940 --> 00:39:22,439 movie, just Google Human Terrain and 1071 00:39:22,440 --> 00:39:24,809 movie. You'll find it at Solli IMDB. 1072 00:39:24,810 --> 00:39:26,459 I don't know if the whole movie is 1073 00:39:26,460 --> 00:39:28,079 available on YouTube. 1074 00:39:28,080 --> 00:39:29,080 Thank you. 1075 00:39:30,120 --> 00:39:32,429 OK, then in the front 1076 00:39:32,430 --> 00:39:34,529 on the right, what is 1077 00:39:34,530 --> 00:39:35,550 that left side. 1078 00:39:37,500 --> 00:39:38,879 First of all, thank you for your talk. 1079 00:39:38,880 --> 00:39:41,609 And I have a question regarding 1080 00:39:41,610 --> 00:39:42,610 the 1081 00:39:43,890 --> 00:39:46,559 the creation of sea grass, 1082 00:39:46,560 --> 00:39:48,179 because if you create those kind of 1083 00:39:48,180 --> 00:39:49,529 crafts, you have certainly have some kind 1084 00:39:49,530 --> 00:39:50,759 of measurement error because it's very 1085 00:39:50,760 --> 00:39:52,859 hard to to, uh, to 1086 00:39:52,860 --> 00:39:54,189 create these graphs. 1087 00:39:54,190 --> 00:39:56,309 And most contrived measures 1088 00:39:56,310 --> 00:39:58,169 are very sensitive to those kind of 1089 00:39:58,170 --> 00:40:00,569 measurement errors. So my question is, 1090 00:40:00,570 --> 00:40:03,029 do you know about any kind of research 1091 00:40:03,030 --> 00:40:05,559 that has focused on the question 1092 00:40:05,560 --> 00:40:07,739 of, um, how noisy 1093 00:40:07,740 --> 00:40:09,539 these grass are and what kind of 1094 00:40:09,540 --> 00:40:11,489 implications of there? 1095 00:40:11,490 --> 00:40:12,539 Because if you make some kind of 1096 00:40:12,540 --> 00:40:15,059 coexistence based on those graphs, maybe 1097 00:40:15,060 --> 00:40:16,649 you killed the wrong guy. 1098 00:40:16,650 --> 00:40:17,670 Yeah. Yeah. Well, I'm, 1099 00:40:18,930 --> 00:40:21,269 uh, I'm I'm not familiar 1100 00:40:21,270 --> 00:40:22,769 with that with with the academic 1101 00:40:22,770 --> 00:40:25,229 literature on that. But that's that's 1102 00:40:25,230 --> 00:40:26,609 actually questions that, for example, 1103 00:40:26,610 --> 00:40:28,769 BINYA gets asked and doesn't really 1104 00:40:28,770 --> 00:40:30,809 answer. Right. If you look in the NSA, 1105 00:40:30,810 --> 00:40:31,810 doesn't 1106 00:40:32,970 --> 00:40:35,219 he? He remains 1107 00:40:35,220 --> 00:40:37,349 fairly unclear in 1108 00:40:37,350 --> 00:40:38,249 that context. 1109 00:40:38,250 --> 00:40:40,559 If you look at, for example, the MIT 1110 00:40:40,560 --> 00:40:42,389 Network Research Center, you will find 1111 00:40:42,390 --> 00:40:43,889 papers that address that question 1112 00:40:43,890 --> 00:40:46,079 because, um, one 1113 00:40:46,080 --> 00:40:48,359 of the things you will see also in the 1114 00:40:48,360 --> 00:40:50,429 in the drawn papers by The Intercept 1115 00:40:50,430 --> 00:40:52,889 is that basically they try to 1116 00:40:52,890 --> 00:40:55,019 either optimize 1117 00:40:55,020 --> 00:40:57,599 the decision making so the responsibility 1118 00:40:57,600 --> 00:41:00,359 can be delegated to a machine 1119 00:41:00,360 --> 00:41:02,429 or else put it on as many 1120 00:41:02,430 --> 00:41:04,439 shoulders as possible. 1121 00:41:04,440 --> 00:41:06,599 So there is a review process of a review 1122 00:41:06,600 --> 00:41:08,759 process of a review process involved 1123 00:41:08,760 --> 00:41:11,129 in those decision making, uh, 1124 00:41:11,130 --> 00:41:13,319 decision making on on on who to 1125 00:41:13,320 --> 00:41:15,419 target. And that's how I think they 1126 00:41:15,420 --> 00:41:16,769 try to deal with that. 1127 00:41:16,770 --> 00:41:18,959 But the technical side, of course, 1128 00:41:18,960 --> 00:41:21,269 is probably really 1129 00:41:21,270 --> 00:41:23,609 complicated. I'm not a technical expert 1130 00:41:23,610 --> 00:41:26,159 on on on these questions. 1131 00:41:26,160 --> 00:41:27,749 Thank you so much. 1132 00:41:27,750 --> 00:41:30,839 OK, then, question over here. 1133 00:41:30,840 --> 00:41:32,279 Thank you very much. I just wanted to 1134 00:41:32,280 --> 00:41:34,349 make sure that I got this shaping thing 1135 00:41:34,350 --> 00:41:36,479 right, because does that mean 1136 00:41:36,480 --> 00:41:38,609 that people are arrested or even 1137 00:41:38,610 --> 00:41:40,919 killed just because they happen to be 1138 00:41:40,920 --> 00:41:42,189 a node in the graph? 1139 00:41:42,190 --> 00:41:44,369 Yeah, that makes the graph 1140 00:41:44,370 --> 00:41:46,109 more central, but it's removed not 1141 00:41:46,110 --> 00:41:48,899 because they did something bad or are 1142 00:41:48,900 --> 00:41:50,879 suspected terrorists, but just because of 1143 00:41:50,880 --> 00:41:52,979 the information you can gain from 1144 00:41:52,980 --> 00:41:54,449 removing that person from the graph. 1145 00:41:54,450 --> 00:41:56,069 Is that what's shaping essentially? 1146 00:41:56,070 --> 00:41:59,099 Yes, that's so 1147 00:41:59,100 --> 00:42:00,089 I don't know. 1148 00:42:00,090 --> 00:42:02,369 And it's probably 1149 00:42:02,370 --> 00:42:04,469 without a security clearance 1150 00:42:04,470 --> 00:42:06,659 and possible to get the information, if 1151 00:42:06,660 --> 00:42:08,669 greedy, fragile, which is the algorithm I 1152 00:42:08,670 --> 00:42:10,379 showed there is actually deployed 1153 00:42:10,380 --> 00:42:11,669 somewhere. Right. 1154 00:42:11,670 --> 00:42:14,099 But the doctrinal papers all basically 1155 00:42:14,100 --> 00:42:17,159 say you have to identify 1156 00:42:17,160 --> 00:42:19,739 persons, you know, that 1157 00:42:19,740 --> 00:42:21,389 have a certain influence within the 1158 00:42:21,390 --> 00:42:23,669 network as depicted by certain 1159 00:42:23,670 --> 00:42:25,919 centrality measures, and either take 1160 00:42:25,920 --> 00:42:27,989 them to extract information from them or 1161 00:42:27,990 --> 00:42:29,819 take them in order to see what happens in 1162 00:42:29,820 --> 00:42:31,019 the graph. 1163 00:42:31,020 --> 00:42:32,339 But that's that's exactly what they're 1164 00:42:32,340 --> 00:42:33,630 doing. That is what shaping does. 1165 00:42:36,060 --> 00:42:37,079 OK, thank you. 1166 00:42:37,080 --> 00:42:39,119 Then we have a question on the side, on 1167 00:42:39,120 --> 00:42:40,120 the microphone, the front. 1168 00:42:41,580 --> 00:42:43,199 I have two questions for you. 1169 00:42:43,200 --> 00:42:44,429 First of all, things representing this 1170 00:42:44,430 --> 00:42:46,529 talk from American's perspective, it's 1171 00:42:46,530 --> 00:42:48,209 pretty fascinating to see how this is 1172 00:42:48,210 --> 00:42:50,729 ingested in a European context. 1173 00:42:50,730 --> 00:42:52,109 First of all, did you look at the 1174 00:42:52,110 --> 00:42:54,299 technologies used on mounted 1175 00:42:54,300 --> 00:42:56,669 on these two poles, selectors 1176 00:42:56,670 --> 00:42:58,409 off the ground? 1177 00:42:58,410 --> 00:43:00,809 And have you looked at the application 1178 00:43:00,810 --> 00:43:03,030 of biometrics onto these 1179 00:43:04,290 --> 00:43:05,519 onto these technologies? 1180 00:43:05,520 --> 00:43:07,139 And then the second question, that's all 1181 00:43:07,140 --> 00:43:08,219 one question. 1182 00:43:08,220 --> 00:43:11,249 The second question is about prediction. 1183 00:43:11,250 --> 00:43:12,749 Pelletiere and other technologies 1184 00:43:12,750 --> 00:43:14,909 develop. Other similar platforms have 1185 00:43:14,910 --> 00:43:16,409 predictive elements that are built into 1186 00:43:16,410 --> 00:43:17,699 their algorithms. 1187 00:43:17,700 --> 00:43:20,339 Have you looked at how those are used to 1188 00:43:20,340 --> 00:43:22,769 target individuals within those networks 1189 00:43:22,770 --> 00:43:24,629 for extraction or whatever? 1190 00:43:24,630 --> 00:43:28,409 I didn't get the part of the on the UVs. 1191 00:43:28,410 --> 00:43:30,359 Did you study the technologies used to 1192 00:43:30,360 --> 00:43:32,999 pull selectors off the ground, DRT boxes 1193 00:43:33,000 --> 00:43:33,749 and so forth? 1194 00:43:33,750 --> 00:43:34,989 No, I didn't. 1195 00:43:34,990 --> 00:43:36,239 So I didn't really look into the 1196 00:43:36,240 --> 00:43:38,309 technical details. So to give you. 1197 00:43:38,310 --> 00:43:40,229 So in order to give you an idea who you 1198 00:43:40,230 --> 00:43:42,780 know what I do, basically, I'm interested 1199 00:43:43,890 --> 00:43:45,719 in the media change of statehood. 1200 00:43:45,720 --> 00:43:48,089 So how we move from 1201 00:43:48,090 --> 00:43:50,429 a from a state that was basically 1202 00:43:50,430 --> 00:43:52,529 built on on paper administrations to 1203 00:43:52,530 --> 00:43:54,959 digital administrations and how new tools 1204 00:43:54,960 --> 00:43:57,299 come into the picture and shape 1205 00:43:57,300 --> 00:43:58,300 or 1206 00:44:00,120 --> 00:44:02,729 change, how 1207 00:44:02,730 --> 00:44:04,799 a nation state, for example, deals 1208 00:44:04,800 --> 00:44:07,229 with something like individual identity 1209 00:44:07,230 --> 00:44:09,059 with borders or the notion of a 1210 00:44:09,060 --> 00:44:10,019 territory. 1211 00:44:10,020 --> 00:44:12,119 So I'm looking at that level and 1212 00:44:12,120 --> 00:44:14,849 I'm not an expert on the on the 1213 00:44:14,850 --> 00:44:16,949 technical, you 1214 00:44:16,950 --> 00:44:18,179 know, details here. 1215 00:44:19,650 --> 00:44:21,869 But a look 1216 00:44:21,870 --> 00:44:23,929 into the Snowden files. 1217 00:44:23,930 --> 00:44:26,099 There's there's a couple of catalogs that 1218 00:44:26,100 --> 00:44:27,689 will give you some information on the 1219 00:44:27,690 --> 00:44:29,759 tools that they used 1220 00:44:29,760 --> 00:44:31,919 to acquire the data and, you 1221 00:44:31,920 --> 00:44:33,989 know, capture phone signals and so forth. 1222 00:44:33,990 --> 00:44:35,879 I don't know anything about biometrics in 1223 00:44:35,880 --> 00:44:37,909 the context of these. 1224 00:44:37,910 --> 00:44:40,229 I don't think they're deployed there. 1225 00:44:40,230 --> 00:44:42,409 To what extent? Because biometrical 1226 00:44:43,500 --> 00:44:46,109 technologies just happened to have too 1227 00:44:46,110 --> 00:44:48,089 high false positive or false negative 1228 00:44:48,090 --> 00:44:49,739 identification rates. 1229 00:44:49,740 --> 00:44:51,659 And I think what they still do is 1230 00:44:51,660 --> 00:44:54,659 basically have individuals sit there 1231 00:44:54,660 --> 00:44:56,969 and use humans 1232 00:44:56,970 --> 00:44:57,970 as 1233 00:44:59,310 --> 00:45:01,360 tools for identification, like. 1234 00:45:03,190 --> 00:45:05,289 OK, then we go back over here to 1235 00:45:05,290 --> 00:45:06,760 this microphone, please. 1236 00:45:07,930 --> 00:45:09,029 I'm Henrik. 1237 00:45:09,030 --> 00:45:10,119 I'm a reporter. 1238 00:45:10,120 --> 00:45:12,309 I've written a number of stories based on 1239 00:45:12,310 --> 00:45:15,459 classified documents and stuff. 1240 00:45:15,460 --> 00:45:17,379 And I can understand your frustration in 1241 00:45:17,380 --> 00:45:18,849 not having access to a lot of this 1242 00:45:18,850 --> 00:45:19,779 material. 1243 00:45:19,780 --> 00:45:21,519 And I was wondering, can you talk a 1244 00:45:21,520 --> 00:45:23,679 little bit more about classified 1245 00:45:23,680 --> 00:45:26,079 research and which use 1246 00:45:26,080 --> 00:45:28,219 which tradeoffs you as a researcher see 1247 00:45:28,220 --> 00:45:29,409 in going into that work? 1248 00:45:29,410 --> 00:45:31,299 There's stuff going on at Heilbrunn in 1249 00:45:31,300 --> 00:45:32,529 the U.K. between the U.S. 1250 00:45:32,530 --> 00:45:34,539 HQ and a university. 1251 00:45:34,540 --> 00:45:36,489 There's West Point in the US. 1252 00:45:36,490 --> 00:45:38,619 But this is stuff that you as someone 1253 00:45:38,620 --> 00:45:41,199 publicly facing cannot access. 1254 00:45:41,200 --> 00:45:42,639 That's the first question. The second one 1255 00:45:42,640 --> 00:45:45,609 is the relationship between the graph 1256 00:45:45,610 --> 00:45:47,769 work done in the commercial world and 1257 00:45:47,770 --> 00:45:49,239 what your impression is of what's going 1258 00:45:49,240 --> 00:45:51,009 on in the secret world. 1259 00:45:51,010 --> 00:45:53,109 The strength ratio is one much 1260 00:45:53,110 --> 00:45:55,260 further than the other one cetera. 1261 00:45:57,100 --> 00:45:59,169 Well, um, as for the 1262 00:45:59,170 --> 00:46:00,879 classified research, I mean, that's 1263 00:46:00,880 --> 00:46:03,339 basically something that I have to accept 1264 00:46:03,340 --> 00:46:05,409 that I, you know, work on the 1265 00:46:05,410 --> 00:46:07,119 basis of leaks and 1266 00:46:08,200 --> 00:46:10,539 publications by academics 1267 00:46:10,540 --> 00:46:12,499 in the military context. 1268 00:46:12,500 --> 00:46:14,859 Um, and one way to 1269 00:46:14,860 --> 00:46:17,049 to deal with that is to 1270 00:46:17,050 --> 00:46:19,599 basically stay in the historical realm. 1271 00:46:19,600 --> 00:46:22,059 So, you know, look back in time 1272 00:46:22,060 --> 00:46:24,279 and try to puzzle puzzle out what 1273 00:46:24,280 --> 00:46:25,509 happened in the 80s and 90s. 1274 00:46:25,510 --> 00:46:27,069 You can talk much about present day 1275 00:46:27,070 --> 00:46:28,659 developments because you just don't know. 1276 00:46:28,660 --> 00:46:31,209 You can, you know, make educated guess 1277 00:46:31,210 --> 00:46:33,549 guesses as best as for the relation 1278 00:46:33,550 --> 00:46:36,519 between business and government, uh, 1279 00:46:36,520 --> 00:46:38,709 graphing. Uh, I don't know what 1280 00:46:38,710 --> 00:46:40,929 I think and what is fairly evident. 1281 00:46:40,930 --> 00:46:43,839 If you, uh, if you look at the documents, 1282 00:46:43,840 --> 00:46:46,329 um, and also academic debates 1283 00:46:46,330 --> 00:46:49,059 in the first part of of of, um, 1284 00:46:49,060 --> 00:46:51,279 uh, 2000, you 1285 00:46:51,280 --> 00:46:52,719 will find there's a lot of going back and 1286 00:46:52,720 --> 00:46:53,769 forth. Right. 1287 00:46:53,770 --> 00:46:55,699 I think that's just, you know, it's 1288 00:46:55,700 --> 00:46:57,279 there's not a conspiracy or anything. 1289 00:46:57,280 --> 00:46:59,139 It's just a discussion, a discourse that 1290 00:46:59,140 --> 00:47:01,359 emerged where people from academia talk 1291 00:47:01,360 --> 00:47:03,219 to people from military, talk to people 1292 00:47:03,220 --> 00:47:05,679 in, uh, 1293 00:47:05,680 --> 00:47:08,469 um, in the NSA and other 1294 00:47:08,470 --> 00:47:09,470 secret services. 1295 00:47:11,230 --> 00:47:13,299 And where 1296 00:47:13,300 --> 00:47:15,549 this these ideas basically 1297 00:47:15,550 --> 00:47:17,289 were floating around and some people made 1298 00:47:17,290 --> 00:47:19,449 a business out of it, uh, facing 1299 00:47:19,450 --> 00:47:20,169 consumers. 1300 00:47:20,170 --> 00:47:21,999 Other made a business out of it, facing 1301 00:47:22,000 --> 00:47:24,309 the government and others made a business 1302 00:47:24,310 --> 00:47:26,679 out of it, you know, uh, 1303 00:47:26,680 --> 00:47:28,779 facing not consumers, 1304 00:47:28,780 --> 00:47:31,269 but, um, uh, 1305 00:47:31,270 --> 00:47:32,889 businesses. Right. 1306 00:47:32,890 --> 00:47:35,139 So I think that's that's that's for sure. 1307 00:47:35,140 --> 00:47:37,239 I'm not I have no idea 1308 00:47:37,240 --> 00:47:39,309 how good or bad the 1309 00:47:39,310 --> 00:47:42,369 graphing graphing capabilities of, 1310 00:47:42,370 --> 00:47:45,189 uh, GHQ or 1311 00:47:45,190 --> 00:47:46,359 NSA is. 1312 00:47:46,360 --> 00:47:48,549 Uh, but it's pretty certain that they use 1313 00:47:48,550 --> 00:47:49,629 Pawlenty, for example. 1314 00:47:49,630 --> 00:47:52,119 And Pawlenty is something 1315 00:47:52,120 --> 00:47:53,679 like a software package that you can 1316 00:47:53,680 --> 00:47:55,989 deploy fairly quickly and does 1317 00:47:55,990 --> 00:47:56,889 what you need. 1318 00:47:56,890 --> 00:47:59,529 So this, uh, um, 1319 00:47:59,530 --> 00:48:01,299 the commercial sector seems to be pretty 1320 00:48:01,300 --> 00:48:02,800 strong, uh, in this. 1321 00:48:05,000 --> 00:48:06,919 And over here in the microphone to the 1322 00:48:06,920 --> 00:48:07,920 left. 1323 00:48:08,810 --> 00:48:10,879 Hello, thank you for your talk. 1324 00:48:10,880 --> 00:48:13,129 I've got one more 1325 00:48:13,130 --> 00:48:14,679 of a remarkable question. 1326 00:48:15,890 --> 00:48:19,879 You were questioning whether a biometric 1327 00:48:19,880 --> 00:48:22,099 identification actually might 1328 00:48:22,100 --> 00:48:23,420 be useful, you know? 1329 00:48:25,420 --> 00:48:27,819 I've got one idea what 1330 00:48:27,820 --> 00:48:30,069 actually I fear might 1331 00:48:30,070 --> 00:48:32,440 work, and that is 1332 00:48:33,460 --> 00:48:36,009 movement pattern detection, 1333 00:48:36,010 --> 00:48:38,139 because there definitely is 1334 00:48:38,140 --> 00:48:40,359 a lot of research also 1335 00:48:40,360 --> 00:48:41,979 in the public about it. 1336 00:48:41,980 --> 00:48:44,199 And it even works 1337 00:48:44,200 --> 00:48:46,299 on the basis of people 1338 00:48:46,300 --> 00:48:48,639 that only are like five pixels tall. 1339 00:48:48,640 --> 00:48:50,139 Yeah, yeah. 1340 00:48:50,140 --> 00:48:52,389 So, um, yeah, 1341 00:48:52,390 --> 00:48:54,639 what you will find is talk 1342 00:48:54,640 --> 00:48:57,339 about pattern of life exploitation 1343 00:48:57,340 --> 00:48:59,799 and pattern of life is basically 1344 00:48:59,800 --> 00:49:02,229 mapping out who you relate to, but also 1345 00:49:02,230 --> 00:49:05,559 where you, where you spend, uh, 1346 00:49:05,560 --> 00:49:06,909 time doing a day during night. 1347 00:49:06,910 --> 00:49:09,369 So what is your better known place. 1348 00:49:09,370 --> 00:49:10,389 Right. 1349 00:49:10,390 --> 00:49:12,189 Where do you eat and so forth. 1350 00:49:12,190 --> 00:49:14,409 And part of this targeting 1351 00:49:14,410 --> 00:49:16,839 process, identification process 1352 00:49:16,840 --> 00:49:18,939 is basically developing 1353 00:49:18,940 --> 00:49:20,649 a pattern of life signature for an 1354 00:49:20,650 --> 00:49:22,899 individual. And that, of course, 1355 00:49:22,900 --> 00:49:24,789 is also used for prediction. 1356 00:49:24,790 --> 00:49:26,859 For example, where will this individual 1357 00:49:26,860 --> 00:49:29,109 be tomorrow at noon time 1358 00:49:29,110 --> 00:49:30,249 and so forth? 1359 00:49:30,250 --> 00:49:32,839 That definitely is part of this process. 1360 00:49:32,840 --> 00:49:35,289 Um, other biometrical 1361 00:49:35,290 --> 00:49:38,079 means, like gate detection 1362 00:49:38,080 --> 00:49:40,479 or facial recognition and so forth, uh, 1363 00:49:40,480 --> 00:49:42,549 I don't think will be used in drone 1364 00:49:42,550 --> 00:49:44,739 context. If you if you there's another 1365 00:49:44,740 --> 00:49:46,899 field, this ideally, um, 1366 00:49:46,900 --> 00:49:49,839 counter ideas or, uh, improvised 1367 00:49:49,840 --> 00:49:52,089 explosive devices, they use quite 1368 00:49:52,090 --> 00:49:53,379 a lot of biometrics there, but that's 1369 00:49:53,380 --> 00:49:55,299 fingerprinting and getting a printing on 1370 00:49:55,300 --> 00:49:58,159 the ground, scanning on the ground. 1371 00:49:58,160 --> 00:50:00,429 Yeah, I definitely agree that 1372 00:50:00,430 --> 00:50:02,619 those things will not 1373 00:50:02,620 --> 00:50:03,909 be available for drones. 1374 00:50:03,910 --> 00:50:06,009 But I mean, my fear 1375 00:50:06,010 --> 00:50:08,139 is that things like 1376 00:50:08,140 --> 00:50:10,209 your walking pattern are just things 1377 00:50:10,210 --> 00:50:12,339 that you probably have no chance of 1378 00:50:12,340 --> 00:50:13,539 escaping to. 1379 00:50:15,300 --> 00:50:16,619 Thank you. 1380 00:50:16,620 --> 00:50:18,899 Um, quick question in between, 1381 00:50:18,900 --> 00:50:21,149 has the Internet any 1382 00:50:21,150 --> 00:50:23,349 on the Internet is silent? 1383 00:50:23,350 --> 00:50:24,750 Oh, 1384 00:50:26,250 --> 00:50:28,199 OK. Then we go back over here to the 1385 00:50:28,200 --> 00:50:30,269 site. Please, please speak 1386 00:50:30,270 --> 00:50:31,589 directly into the microphone. 1387 00:50:31,590 --> 00:50:32,579 Sure. 1388 00:50:32,580 --> 00:50:34,349 I have like two quick observations on 1389 00:50:34,350 --> 00:50:35,519 this. One of them is that this actually 1390 00:50:35,520 --> 00:50:37,559 kind of apparently a little open source 1391 00:50:37,560 --> 00:50:38,489 ecosystem forming. 1392 00:50:38,490 --> 00:50:40,649 So a few years back and it's a 1393 00:50:40,650 --> 00:50:41,939 kind of open source. 1394 00:50:41,940 --> 00:50:43,889 One of the databases accumulate and 1395 00:50:43,890 --> 00:50:44,999 that's been picked up by a lot of 1396 00:50:45,000 --> 00:50:46,699 academia. They've rebranded it, I think, 1397 00:50:46,700 --> 00:50:48,779 before, but they're kind of using 1398 00:50:48,780 --> 00:50:50,369 the NSA could quite happily. 1399 00:50:50,370 --> 00:50:52,499 And then a few weeks back, you 1400 00:50:52,500 --> 00:50:53,819 kind of layered on top with their own 1401 00:50:53,820 --> 00:50:55,499 open source database thing. 1402 00:50:55,500 --> 00:50:56,759 And it's kind of interesting to see that 1403 00:50:56,760 --> 00:50:58,859 they're actually open sourcing the stuff. 1404 00:50:58,860 --> 00:51:00,479 The other thing is kind of I wonder 1405 00:51:00,480 --> 00:51:01,739 whether you've seen this as well. 1406 00:51:01,740 --> 00:51:03,479 I think there's a thing about most big 1407 00:51:03,480 --> 00:51:04,739 data and it seems to apply to these 1408 00:51:04,740 --> 00:51:06,539 national security things as well, where 1409 00:51:06,540 --> 00:51:07,769 you talk about it quite a lot. 1410 00:51:07,770 --> 00:51:09,479 But then in the end, you come out and 1411 00:51:09,480 --> 00:51:11,099 come down in like day to day business to 1412 00:51:11,100 --> 00:51:12,149 very simplistic things. 1413 00:51:12,150 --> 00:51:13,919 Right. So you talk about all these grauwe 1414 00:51:13,920 --> 00:51:15,239 and behavioral patterns, things. 1415 00:51:15,240 --> 00:51:16,289 And then if you look at the actual 1416 00:51:16,290 --> 00:51:17,519 documentation for something like 1417 00:51:17,520 --> 00:51:19,589 XKeyscore, it turns out it's basically 1418 00:51:19,590 --> 00:51:20,669 a version of GRAP. Right. 1419 00:51:20,670 --> 00:51:22,589 So it's basically like very, very 1420 00:51:22,590 --> 00:51:23,519 simplistic metric. 1421 00:51:23,520 --> 00:51:25,679 And I wonder whether how much of the kind 1422 00:51:25,680 --> 00:51:27,779 of have you seen any indications 1423 00:51:27,780 --> 00:51:29,880 of how much of this is actually kind of 1424 00:51:31,550 --> 00:51:33,719 your head in the clouds 1425 00:51:33,720 --> 00:51:35,459 kind of thinking that then in practice 1426 00:51:35,460 --> 00:51:36,519 looks completely different? 1427 00:51:36,520 --> 00:51:37,520 Hmm. 1428 00:51:37,920 --> 00:51:40,109 To let a question, you know, I don't 1429 00:51:40,110 --> 00:51:41,939 know, because you can't really look 1430 00:51:41,940 --> 00:51:44,280 inside these projects. 1431 00:51:46,200 --> 00:51:48,509 But my guess would be it's fairly 1432 00:51:48,510 --> 00:51:50,399 simple. It's just a scale. 1433 00:51:50,400 --> 00:51:51,719 Right, which is not simple. 1434 00:51:51,720 --> 00:51:54,209 It's the it's accessing 1435 00:51:54,210 --> 00:51:56,429 lots of lots of data points, 1436 00:51:56,430 --> 00:51:58,769 ingesting the data and being able to 1437 00:51:58,770 --> 00:52:00,839 to store it indefinitely and search 1438 00:52:00,840 --> 00:52:01,769 it. 1439 00:52:01,770 --> 00:52:03,419 Benjamin Britten, who comes out with a 1440 00:52:03,420 --> 00:52:05,729 book called The Black Stack, 1441 00:52:05,730 --> 00:52:07,139 called this the Big. 1442 00:52:07,140 --> 00:52:09,299 So basically in the past decade, they 1443 00:52:09,300 --> 00:52:11,159 could haul in all the data of the world 1444 00:52:11,160 --> 00:52:13,289 and, um, that, you know, 1445 00:52:13,290 --> 00:52:16,229 provides, of course, a strategic resource 1446 00:52:16,230 --> 00:52:18,449 which other entities might not 1447 00:52:18,450 --> 00:52:19,769 have access to. 1448 00:52:19,770 --> 00:52:21,150 As for a Kumalo, 1449 00:52:22,500 --> 00:52:24,449 yeah, I found it interesting, too. 1450 00:52:24,450 --> 00:52:26,129 I think basically I mean, if you look at 1451 00:52:26,130 --> 00:52:27,419 Hecke News and all those websites, you 1452 00:52:27,420 --> 00:52:30,749 will find a lot of debates about, um, 1453 00:52:30,750 --> 00:52:33,359 you know, uh, the software infrastructure 1454 00:52:33,360 --> 00:52:35,429 of the stacks of the of the big five 1455 00:52:35,430 --> 00:52:37,589 companies, Apple, Amazon, 1456 00:52:37,590 --> 00:52:39,389 Microsoft. 1457 00:52:39,390 --> 00:52:40,390 Facebook, 1458 00:52:41,580 --> 00:52:43,769 and I wouldn't wonder 1459 00:52:43,770 --> 00:52:46,499 if they basically, if there's a lot 1460 00:52:46,500 --> 00:52:48,509 of knowledge and also tools going back 1461 00:52:48,510 --> 00:52:50,969 and forth, you know, it's it's 1462 00:52:50,970 --> 00:52:53,289 probably look, if you would have a chart, 1463 00:52:53,290 --> 00:52:55,110 you know, that shows you how 1464 00:52:56,220 --> 00:52:58,139 the stack looks and the NSA probably 1465 00:52:58,140 --> 00:53:00,809 doesn't look much different from Amazon 1466 00:53:00,810 --> 00:53:02,559 and maybe Amazon even runs parts of it. 1467 00:53:02,560 --> 00:53:04,919 So and that's what I mean by military, 1468 00:53:04,920 --> 00:53:07,579 academic, industrial complex. 1469 00:53:07,580 --> 00:53:09,479 It is just knowledge floating around. 1470 00:53:09,480 --> 00:53:11,879 It's know how that is available 1471 00:53:11,880 --> 00:53:13,739 and that can be applied, you know, 1472 00:53:13,740 --> 00:53:15,089 academically, it can be employed in a 1473 00:53:15,090 --> 00:53:16,649 business context, can be employed in this 1474 00:53:16,650 --> 00:53:17,539 context. 1475 00:53:17,540 --> 00:53:19,709 It's just a discourse that 1476 00:53:19,710 --> 00:53:22,019 that emerged once so much data 1477 00:53:22,020 --> 00:53:24,249 was around. And the necessity to graph 1478 00:53:24,250 --> 00:53:26,760 that out turned out to be valuable. 1479 00:53:29,080 --> 00:53:31,359 OK, thank you, and then we have 1480 00:53:31,360 --> 00:53:33,369 one more question on the left side, 1481 00:53:33,370 --> 00:53:34,179 please. 1482 00:53:34,180 --> 00:53:36,219 You told us about shaping of social 1483 00:53:36,220 --> 00:53:37,690 graphs. The military does 1484 00:53:38,890 --> 00:53:41,359 is removal of not the only operation 1485 00:53:41,360 --> 00:53:43,599 that was done there or is there also 1486 00:53:43,600 --> 00:53:46,059 research on inserting nodes or 1487 00:53:46,060 --> 00:53:48,159 taking over nodes for 1488 00:53:48,160 --> 00:53:50,769 food distribution is not removing 1489 00:53:50,770 --> 00:53:51,339 nodes. 1490 00:53:51,340 --> 00:53:53,559 It's giving nodes something and 1491 00:53:53,560 --> 00:53:55,629 an opportunity to see 1492 00:53:55,630 --> 00:53:56,799 where the food ends up. 1493 00:53:56,800 --> 00:53:57,939 Who talks to whom. 1494 00:53:57,940 --> 00:53:59,349 What do the people tell you 1495 00:54:00,490 --> 00:54:02,469 when when you give them food? 1496 00:54:02,470 --> 00:54:05,049 And I mean, basically targeted 1497 00:54:05,050 --> 00:54:07,359 ads is shaping, right? 1498 00:54:07,360 --> 00:54:10,299 I mean, if Facebook 1499 00:54:10,300 --> 00:54:12,399 helps you out some something on your 1500 00:54:12,400 --> 00:54:13,899 timeline, you click on it. 1501 00:54:13,900 --> 00:54:16,029 It has successfully shaped the graph. 1502 00:54:16,030 --> 00:54:17,369 Right, so 1503 00:54:18,550 --> 00:54:20,679 that, you know, in a doctor, 1504 00:54:20,680 --> 00:54:22,179 in a military doctor, in the context, of 1505 00:54:22,180 --> 00:54:24,009 course, there's there is an option 1506 00:54:24,010 --> 00:54:26,529 available that that includes 1507 00:54:26,530 --> 00:54:29,289 removing nodes. But the knowledge, 1508 00:54:29,290 --> 00:54:30,290 um. 1509 00:54:31,460 --> 00:54:34,409 It goes far beyond that, right? 1510 00:54:34,410 --> 00:54:36,599 The possibilities then just 1511 00:54:36,600 --> 00:54:37,600 removing an old. 1512 00:54:40,010 --> 00:54:42,139 OK, we have a question 1513 00:54:42,140 --> 00:54:44,359 on this side, on the right side, please. 1514 00:54:44,360 --> 00:54:46,609 It's not really a question, just some 1515 00:54:46,610 --> 00:54:49,069 thoughts regarding the previous question. 1516 00:54:49,070 --> 00:54:51,259 So Kumalo is actually a fork of each 1517 00:54:51,260 --> 00:54:53,479 base, which is an open source project. 1518 00:54:53,480 --> 00:54:56,179 I totally agree that it is 1519 00:54:56,180 --> 00:54:58,129 to some degree different. 1520 00:54:58,130 --> 00:55:00,409 But in the end, the core is the same 1521 00:55:00,410 --> 00:55:03,229 technology. OK, and also recently 1522 00:55:03,230 --> 00:55:05,329 I noticed that GCE HQ 1523 00:55:05,330 --> 00:55:08,089 has released an open source 1524 00:55:08,090 --> 00:55:10,309 graph to, let's say, 1525 00:55:10,310 --> 00:55:11,599 distributed graph two. 1526 00:55:11,600 --> 00:55:13,429 So it's the same of it. 1527 00:55:13,430 --> 00:55:15,329 I have it on my cell phone. 1528 00:55:15,330 --> 00:55:17,509 Sorry, but 1529 00:55:17,510 --> 00:55:19,669 what I wanted to say is that the industry 1530 00:55:19,670 --> 00:55:21,259 actually has tools that are really 1531 00:55:21,260 --> 00:55:23,419 similar, or at least we can 1532 00:55:23,420 --> 00:55:25,489 think so towards also 1533 00:55:25,490 --> 00:55:26,599 the other people are using. 1534 00:55:28,490 --> 00:55:29,490 Thank you. 1535 00:55:29,900 --> 00:55:31,279 Thank you. 1536 00:55:31,280 --> 00:55:33,340 Are there any more questions? 1537 00:55:34,620 --> 00:55:36,749 Then please go to the microphone, no 1538 00:55:38,460 --> 00:55:41,099 internet is silent, OK? 1539 00:55:41,100 --> 00:55:42,839 Thank you so much for the talk. 1540 00:55:42,840 --> 00:55:43,840 Thank you.