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/354 Thanks! 1 00:00:09,160 --> 00:00:10,929 I would like to introduce to you Anita 2 00:00:10,930 --> 00:00:13,179 Gordis, who's a postdoc at the University 3 00:00:13,180 --> 00:00:15,249 of Mannheim and also working for the 4 00:00:15,250 --> 00:00:17,979 Human Rights Data Analysis Group, 5 00:00:17,980 --> 00:00:19,509 and she's going to give a talk about 6 00:00:19,510 --> 00:00:21,699 information control and state repression. 7 00:00:21,700 --> 00:00:23,289 So please give a warm welcome. 8 00:00:23,290 --> 00:00:24,290 Applause. 9 00:00:33,020 --> 00:00:34,999 Well, thank you very much. 10 00:00:35,000 --> 00:00:37,129 Yeah, my name is Anita Godus, and 11 00:00:37,130 --> 00:00:39,379 I'd like to say this is my first C.C.C. 12 00:00:39,380 --> 00:00:40,579 and I'm pretty sure it's the best 13 00:00:40,580 --> 00:00:41,599 conference I've ever been to. 14 00:00:41,600 --> 00:00:43,000 So pretty cool. 15 00:00:47,210 --> 00:00:49,279 I work at the University of Mannheim as 16 00:00:49,280 --> 00:00:51,199 a researcher and as a conflict 17 00:00:51,200 --> 00:00:52,909 researcher, I'm interested in 18 00:00:52,910 --> 00:00:55,189 understanding how governments 19 00:00:55,190 --> 00:00:57,709 use violence, where they use violence 20 00:00:57,710 --> 00:01:00,469 and what types of violence they use. 21 00:01:00,470 --> 00:01:01,759 And for the last five years, I've also 22 00:01:01,760 --> 00:01:03,739 been working for the Human Rights Data 23 00:01:03,740 --> 00:01:05,509 Analysis Group. 24 00:01:05,510 --> 00:01:07,729 And what we do is we try 25 00:01:07,730 --> 00:01:09,949 to analyze statistically 26 00:01:09,950 --> 00:01:12,049 large scale human rights violations 27 00:01:12,050 --> 00:01:14,119 and ultimately build evidence based 28 00:01:14,120 --> 00:01:15,889 arguments that can help bring the 29 00:01:15,890 --> 00:01:18,060 perpetrators to justice. 30 00:01:19,640 --> 00:01:21,829 So the talk I want to give today 31 00:01:21,830 --> 00:01:23,509 is on information control and strategic 32 00:01:23,510 --> 00:01:24,510 repression. 33 00:01:25,870 --> 00:01:27,969 So the ability to connect via 34 00:01:27,970 --> 00:01:29,829 social media has been celebrated across 35 00:01:29,830 --> 00:01:31,809 the world by human rights groups, by 36 00:01:31,810 --> 00:01:34,149 researchers and by policymakers, because 37 00:01:34,150 --> 00:01:35,709 it seems like we finally have a tool that 38 00:01:35,710 --> 00:01:37,989 can help ordinary citizens 39 00:01:37,990 --> 00:01:39,849 connect, overcome collective action 40 00:01:39,850 --> 00:01:41,769 problems and mobilize against repressive 41 00:01:41,770 --> 00:01:42,770 rulers. 42 00:01:43,260 --> 00:01:46,199 Now, for all of you sitting in this room, 43 00:01:46,200 --> 00:01:47,699 you're quite aware that this view on the 44 00:01:47,700 --> 00:01:50,219 digital revolution alone is deceptive 45 00:01:50,220 --> 00:01:52,379 and it's deceptive for for 46 00:01:52,380 --> 00:01:53,969 one important reason, and that is that 47 00:01:53,970 --> 00:01:56,099 governments principally still remain 48 00:01:56,100 --> 00:01:57,859 in control of network accessibility. 49 00:01:58,860 --> 00:02:00,239 So governments haven't just been sitting 50 00:02:00,240 --> 00:02:02,669 on the sidelines watching mass uprisings 51 00:02:02,670 --> 00:02:04,349 in their countries. 52 00:02:04,350 --> 00:02:06,749 They've been building an arsenal of tools 53 00:02:06,750 --> 00:02:09,538 to manipulate, censor 54 00:02:09,539 --> 00:02:11,729 and survey information exchange happening 55 00:02:11,730 --> 00:02:12,730 within their countries. 56 00:02:14,260 --> 00:02:15,939 Now, we know this and we've learned a lot 57 00:02:15,940 --> 00:02:17,529 about this in the last few days, and I'm 58 00:02:17,530 --> 00:02:19,839 really scared also about 59 00:02:19,840 --> 00:02:21,969 what I've learned here, but what we don't 60 00:02:21,970 --> 00:02:24,039 really know is how this control of 61 00:02:24,040 --> 00:02:26,109 the Internet informs government 62 00:02:26,110 --> 00:02:27,879 repression. So how does it actually 63 00:02:27,880 --> 00:02:30,069 inform the violence happening 64 00:02:30,070 --> 00:02:32,259 to people in countries by government? 65 00:02:35,980 --> 00:02:37,839 Right. So in the next 20 minutes, I want 66 00:02:37,840 --> 00:02:39,849 to give you a very brief theoretical 67 00:02:39,850 --> 00:02:41,919 background, then discuss what I 68 00:02:41,920 --> 00:02:43,419 mean by different types of repression 69 00:02:43,420 --> 00:02:44,439 used by governments. 70 00:02:45,730 --> 00:02:47,169 And then I want to discuss a very 71 00:02:47,170 --> 00:02:48,999 important topic, which is how we deal 72 00:02:49,000 --> 00:02:50,269 with the dark fear of violence. 73 00:02:50,270 --> 00:02:52,419 So how do we figure out who 74 00:02:52,420 --> 00:02:55,299 is being killed and not documented? 75 00:02:55,300 --> 00:02:56,439 And then lastly, I'll give you some 76 00:02:56,440 --> 00:02:57,999 results on how Internet control and 77 00:02:58,000 --> 00:02:59,679 violence seem to coincide. 78 00:03:01,060 --> 00:03:03,189 OK, so 79 00:03:03,190 --> 00:03:04,869 let's think about this from the 80 00:03:04,870 --> 00:03:06,669 perspective of government, that's what we 81 00:03:06,670 --> 00:03:09,369 try to do as far as conflict researchers, 82 00:03:09,370 --> 00:03:12,399 governments and use repressions, 83 00:03:12,400 --> 00:03:14,139 for example, during riots. 84 00:03:15,470 --> 00:03:17,389 So there are two different types of 85 00:03:17,390 --> 00:03:18,680 threats that governments can 86 00:03:19,700 --> 00:03:20,700 can 87 00:03:22,100 --> 00:03:23,779 can encounter, on the one hand, the 88 00:03:23,780 --> 00:03:25,549 hidden threats so that people who are 89 00:03:25,550 --> 00:03:28,039 organizing to overthrow a government 90 00:03:28,040 --> 00:03:30,469 to to build an opposition group 91 00:03:30,470 --> 00:03:32,119 and the government doesn't really know 92 00:03:32,120 --> 00:03:34,459 exactly who the threat is and 93 00:03:34,460 --> 00:03:35,539 where the threat is located. 94 00:03:35,540 --> 00:03:37,939 Right. So in these situations, 95 00:03:37,940 --> 00:03:40,489 state, state institutions 96 00:03:40,490 --> 00:03:42,079 need information. They need intelligence 97 00:03:42,080 --> 00:03:43,789 on where these people are, what they're 98 00:03:43,790 --> 00:03:45,829 organizing, so they can eliminate it 99 00:03:45,830 --> 00:03:47,659 before it gets even bigger. 100 00:03:49,800 --> 00:03:52,019 The other situations where the threat 101 00:03:52,020 --> 00:03:53,489 is already fully visible, the government 102 00:03:53,490 --> 00:03:55,129 knows exactly who its enemy is, 103 00:03:56,490 --> 00:03:58,079 it can already see thousands of people 104 00:03:58,080 --> 00:04:00,719 gathering on the streets, fighting for 105 00:04:00,720 --> 00:04:02,219 for change. 106 00:04:02,220 --> 00:04:04,229 And in these situations, they're likely 107 00:04:04,230 --> 00:04:05,789 to just say, OK, we're going to use any 108 00:04:05,790 --> 00:04:08,459 means necessary to crush down this 109 00:04:08,460 --> 00:04:09,460 this protest. 110 00:04:11,900 --> 00:04:13,849 So if we think about these different 111 00:04:13,850 --> 00:04:16,409 types of responses to 112 00:04:16,410 --> 00:04:18,499 to threats with within the 113 00:04:18,500 --> 00:04:20,569 state, we might think about 114 00:04:20,570 --> 00:04:22,789 how governments might 115 00:04:22,790 --> 00:04:24,740 use different digital responses as well. 116 00:04:27,110 --> 00:04:29,719 So on the one hand, they could 117 00:04:29,720 --> 00:04:31,789 survey the Internet in face of a threat, 118 00:04:31,790 --> 00:04:33,829 they can glean intelligence from people's 119 00:04:33,830 --> 00:04:35,689 Facebook, profiles from what people are 120 00:04:35,690 --> 00:04:36,949 twittering, what they're planning on 121 00:04:36,950 --> 00:04:39,199 doing from events and so on and so forth, 122 00:04:39,200 --> 00:04:41,509 from email exchanges and use that 123 00:04:41,510 --> 00:04:42,560 against opponents. 124 00:04:43,730 --> 00:04:45,739 The problem is, as soon as you let people 125 00:04:45,740 --> 00:04:47,809 generate content online, you give them 126 00:04:47,810 --> 00:04:50,459 a platform to mobilize and organize. 127 00:04:50,460 --> 00:04:52,519 So the other option would be 128 00:04:52,520 --> 00:04:54,709 to shut down the Internet, shut down 129 00:04:54,710 --> 00:04:56,629 all communication and give them no 130 00:04:58,460 --> 00:05:01,309 way to to organize virtually. 131 00:05:01,310 --> 00:05:03,289 Now, these are obviously very two very 132 00:05:03,290 --> 00:05:05,179 abstract concept concepts. 133 00:05:05,180 --> 00:05:07,219 Lots of governments, surveillance, censor 134 00:05:07,220 --> 00:05:08,989 at the same time. 135 00:05:08,990 --> 00:05:11,179 But these are kind of the abstract 136 00:05:11,180 --> 00:05:13,939 ideas that they could use. 137 00:05:13,940 --> 00:05:15,379 And what I'd like to argue is that the 138 00:05:15,380 --> 00:05:17,599 choice of digital response, either using 139 00:05:17,600 --> 00:05:19,879 surveillance or using censorship, 140 00:05:19,880 --> 00:05:22,369 is likely to inform the type 141 00:05:22,370 --> 00:05:24,409 of actual violent repression used by 142 00:05:24,410 --> 00:05:25,410 governments. 143 00:05:28,200 --> 00:05:30,359 OK, so the case 144 00:05:30,360 --> 00:05:32,789 of my research has been the ongoing 145 00:05:32,790 --> 00:05:33,989 conflict in Syria. 146 00:05:35,770 --> 00:05:37,389 The Syrian civil war has been called the 147 00:05:37,390 --> 00:05:40,149 most socially mediated conflict 148 00:05:40,150 --> 00:05:41,150 in the history. 149 00:05:42,670 --> 00:05:44,919 What we see here is the Syrian 150 00:05:44,920 --> 00:05:47,379 presidency's Instagram account. 151 00:05:47,380 --> 00:05:48,669 And I don't know about you, but I can't 152 00:05:48,670 --> 00:05:50,889 see any traces of conflict in 153 00:05:50,890 --> 00:05:52,959 this in this account. 154 00:05:52,960 --> 00:05:54,939 It looks like Assad is ruling a country 155 00:05:54,940 --> 00:05:57,189 that's united under him and 156 00:05:57,190 --> 00:05:58,899 his wife is smiling and they're having a 157 00:05:58,900 --> 00:05:59,900 good time. 158 00:06:00,700 --> 00:06:02,529 We can't see anything of death, death and 159 00:06:02,530 --> 00:06:04,179 destruction that's actually going on 160 00:06:04,180 --> 00:06:05,949 within the country. 161 00:06:05,950 --> 00:06:07,509 At the same time, we know that the Syrian 162 00:06:07,510 --> 00:06:09,519 government is using different types of 163 00:06:09,520 --> 00:06:11,499 surveillance against its population. 164 00:06:11,500 --> 00:06:13,689 It's using shut 165 00:06:13,690 --> 00:06:14,829 downs of the Internet 166 00:06:15,850 --> 00:06:17,859 and it's manipulating information as 167 00:06:17,860 --> 00:06:18,189 well. 168 00:06:18,190 --> 00:06:20,289 So so this picture is obviously 169 00:06:20,290 --> 00:06:21,290 highly deceptive 170 00:06:23,500 --> 00:06:25,719 to investigate just how deceptive it is. 171 00:06:25,720 --> 00:06:26,860 My colleagues and I 172 00:06:29,170 --> 00:06:30,849 coauthored a report that was commissioned 173 00:06:30,850 --> 00:06:33,969 by the UN Human Rights Office 174 00:06:33,970 --> 00:06:36,309 that appeared in August 2014, 175 00:06:36,310 --> 00:06:37,599 where we looked at the number of 176 00:06:37,600 --> 00:06:39,879 documented killings that 177 00:06:39,880 --> 00:06:41,229 have occurred in the first three years of 178 00:06:41,230 --> 00:06:42,129 the conflict. 179 00:06:42,130 --> 00:06:44,199 And what we find is that we can 180 00:06:44,200 --> 00:06:46,449 uniquely identify almost 181 00:06:46,450 --> 00:06:48,609 200000 individuals who've 182 00:06:48,610 --> 00:06:50,610 been killed by the regime, 183 00:06:52,660 --> 00:06:54,759 200000 people who we know 184 00:06:54,760 --> 00:06:55,659 have been killed. 185 00:06:55,660 --> 00:06:57,399 So we know that the dark figure is likely 186 00:06:57,400 --> 00:06:59,109 to be much bigger because not everyone is 187 00:06:59,110 --> 00:07:00,110 documented. 188 00:07:02,340 --> 00:07:04,139 So when we hear about refugee flows 189 00:07:04,140 --> 00:07:05,140 coming to Europe, 190 00:07:06,540 --> 00:07:08,369 we know that this is for this is 191 00:07:08,370 --> 00:07:10,019 happening for a reason. People are dying 192 00:07:10,020 --> 00:07:11,309 on the ground and they're dying every 193 00:07:11,310 --> 00:07:12,310 day. 194 00:07:14,660 --> 00:07:16,789 Right, but these people haven't all 195 00:07:16,790 --> 00:07:18,739 been killed in the same way some people 196 00:07:18,740 --> 00:07:20,749 have been killed through through 197 00:07:20,750 --> 00:07:22,129 explosions. Some people have been 198 00:07:22,130 --> 00:07:23,239 directly executed. 199 00:07:23,240 --> 00:07:25,549 And the question is, how do we understand 200 00:07:25,550 --> 00:07:27,229 the different types of repression used by 201 00:07:27,230 --> 00:07:28,230 the Syrian government? 202 00:07:29,820 --> 00:07:31,949 OK, so what we see here is 203 00:07:31,950 --> 00:07:34,079 a snapshot of our database, what we 204 00:07:34,080 --> 00:07:35,080 received were 205 00:07:37,140 --> 00:07:38,819 lists from four different human rights 206 00:07:38,820 --> 00:07:41,129 groups on individuals 207 00:07:41,130 --> 00:07:43,799 who were killed, including the location 208 00:07:43,800 --> 00:07:44,800 where they were killed, 209 00:07:46,140 --> 00:07:47,849 the date on which they were killed. 210 00:07:47,850 --> 00:07:50,129 Sometimes information on their agenda and 211 00:07:50,130 --> 00:07:51,779 their full name, which I've anonymized 212 00:07:51,780 --> 00:07:54,449 here, and we were able to match 213 00:07:54,450 --> 00:07:57,089 every individual in every group 214 00:07:57,090 --> 00:07:59,129 through this information so we could 215 00:07:59,130 --> 00:08:01,979 arrive at a number of uniquely identified 216 00:08:01,980 --> 00:08:02,980 killings. 217 00:08:03,820 --> 00:08:06,699 What you don't see in the snapshot is 218 00:08:06,700 --> 00:08:08,199 auxillary information that tells us 219 00:08:08,200 --> 00:08:10,209 something about the circumstance of every 220 00:08:10,210 --> 00:08:12,269 individual's death. 221 00:08:18,690 --> 00:08:20,999 OK, so if we think about different types 222 00:08:21,000 --> 00:08:24,029 of violence used by by governments 223 00:08:24,030 --> 00:08:25,859 in conflict with research, we often 224 00:08:25,860 --> 00:08:28,049 broadly distinguish between two different 225 00:08:28,050 --> 00:08:30,389 types of two different types 226 00:08:30,390 --> 00:08:32,009 of violence used by the government, the 227 00:08:32,010 --> 00:08:34,199 one we can we can call 228 00:08:34,200 --> 00:08:35,229 targeted violence. 229 00:08:35,230 --> 00:08:37,379 So this is violence that's used 230 00:08:37,380 --> 00:08:39,689 to eliminate individuals or groups based 231 00:08:39,690 --> 00:08:40,869 on certain characteristics. 232 00:08:40,870 --> 00:08:43,109 So examples would be a government 233 00:08:43,110 --> 00:08:45,149 target, someone for the information he or 234 00:08:45,150 --> 00:08:47,429 she has, because that person has defected 235 00:08:47,430 --> 00:08:49,710 from the army because that person is 236 00:08:50,730 --> 00:08:52,109 is leading an opposition group. 237 00:08:52,110 --> 00:08:53,549 So these are people who are eliminated 238 00:08:53,550 --> 00:08:55,439 because they they they share specific 239 00:08:55,440 --> 00:08:56,399 traits. 240 00:08:56,400 --> 00:08:57,809 Another example would be 241 00:08:58,890 --> 00:09:00,959 a government that that is an ethnic 242 00:09:00,960 --> 00:09:02,699 minority and that is trying to target 243 00:09:02,700 --> 00:09:04,559 another ethnicity. 244 00:09:04,560 --> 00:09:06,269 These would be what we call targeted 245 00:09:07,680 --> 00:09:08,680 forms of violence. 246 00:09:09,880 --> 00:09:11,829 The other form is on targeted generalized 247 00:09:11,830 --> 00:09:14,439 terror, the entire civilian population 248 00:09:14,440 --> 00:09:16,419 is seen as a threat to the government and 249 00:09:16,420 --> 00:09:18,609 anyone is attacked. 250 00:09:18,610 --> 00:09:20,289 Now, these are two different types of 251 00:09:20,290 --> 00:09:22,269 strategies that a government can use to 252 00:09:22,270 --> 00:09:24,009 suppress its population. 253 00:09:24,010 --> 00:09:26,019 And oftentimes they're both used at 254 00:09:26,020 --> 00:09:27,369 different points in time within a 255 00:09:27,370 --> 00:09:28,809 conflict. 256 00:09:28,810 --> 00:09:31,239 And Assad's army 257 00:09:31,240 --> 00:09:33,249 has been using both types in the Syrian 258 00:09:33,250 --> 00:09:34,989 conflict as well. 259 00:09:34,990 --> 00:09:35,990 OK. 260 00:09:36,720 --> 00:09:38,579 So the axillary information we get from 261 00:09:38,580 --> 00:09:40,679 these human rights groups can help us to 262 00:09:40,680 --> 00:09:42,779 understand the circumstances under which 263 00:09:42,780 --> 00:09:44,489 each individual was killed. 264 00:09:44,490 --> 00:09:47,219 So this is an example of one person 265 00:09:47,220 --> 00:09:48,959 who was recorded in three different 266 00:09:48,960 --> 00:09:51,529 sources and source A, B and C. 267 00:09:51,530 --> 00:09:53,279 And what this tells us is the first 268 00:09:53,280 --> 00:09:55,289 source tells us the person died under 269 00:09:55,290 --> 00:09:56,549 torture in the prison system. 270 00:09:56,550 --> 00:09:58,679 Right. So that gives us information that 271 00:09:58,680 --> 00:10:00,209 that person was singled out, he was 272 00:10:00,210 --> 00:10:02,610 detained or she was detained and then 273 00:10:03,960 --> 00:10:05,189 tortured. 274 00:10:05,190 --> 00:10:08,129 The second source tells us this person 275 00:10:08,130 --> 00:10:09,130 was shot. 276 00:10:09,720 --> 00:10:11,459 So this person didn't just die 277 00:10:12,570 --> 00:10:13,979 from the effects of torture. 278 00:10:13,980 --> 00:10:15,270 This person was actually shot. 279 00:10:16,840 --> 00:10:18,819 The third source tells us where that 280 00:10:18,820 --> 00:10:21,399 person was arrested and when they were 281 00:10:21,400 --> 00:10:23,019 where they were held and on what date 282 00:10:23,020 --> 00:10:24,369 they were arrested. 283 00:10:24,370 --> 00:10:26,169 So if we put this information together, 284 00:10:26,170 --> 00:10:28,479 it tells us something about the nature 285 00:10:28,480 --> 00:10:30,009 of repression used against that 286 00:10:30,010 --> 00:10:31,010 individual. 287 00:10:33,700 --> 00:10:34,720 So examples of 288 00:10:35,770 --> 00:10:37,299 of words that we find 289 00:10:39,100 --> 00:10:40,449 when we look at individuals who were 290 00:10:40,450 --> 00:10:42,159 killed in a targeted way would be people 291 00:10:42,160 --> 00:10:44,289 who were previously arrested or detained, 292 00:10:44,290 --> 00:10:46,059 who were found with their hands and legs 293 00:10:46,060 --> 00:10:48,129 tied, who were tortured, who 294 00:10:48,130 --> 00:10:50,439 were directly executed or who were named 295 00:10:50,440 --> 00:10:51,440 to be defector's. 296 00:10:53,150 --> 00:10:55,219 And examples of untalked that violence 297 00:10:55,220 --> 00:10:56,809 would be people who were killed in 298 00:10:56,810 --> 00:10:59,209 explosions, in bombings or children, 299 00:10:59,210 --> 00:11:00,210 for example. 300 00:11:02,570 --> 00:11:05,899 Now, I look at the period from mid 2013 301 00:11:05,900 --> 00:11:07,999 to mid 2014, which means 302 00:11:08,000 --> 00:11:10,579 I have over 60000 records of people 303 00:11:10,580 --> 00:11:11,989 who were killed by the government. 304 00:11:13,340 --> 00:11:15,829 And in order to classify these, according 305 00:11:15,830 --> 00:11:17,480 to these two different categories, 306 00:11:18,620 --> 00:11:21,079 I draw a sample of 2000 307 00:11:21,080 --> 00:11:23,389 and I can code them to establish 308 00:11:23,390 --> 00:11:25,579 what I would consider to be targeted 309 00:11:25,580 --> 00:11:27,499 and targeted violence. 310 00:11:27,500 --> 00:11:29,449 And then I use supervised machine 311 00:11:29,450 --> 00:11:30,349 learning to help me with the 312 00:11:30,350 --> 00:11:31,849 classification of the rest. 313 00:11:31,850 --> 00:11:33,319 And I'm happy to talk about the details 314 00:11:33,320 --> 00:11:35,019 of the method afterwards. 315 00:11:37,050 --> 00:11:39,209 OK, so the next big 316 00:11:39,210 --> 00:11:41,339 problem we have is that we 317 00:11:41,340 --> 00:11:43,409 have very good and very detailed 318 00:11:43,410 --> 00:11:45,329 information on people who were 319 00:11:45,330 --> 00:11:47,639 documented, but in every 320 00:11:47,640 --> 00:11:49,589 conflict that people who fall under the 321 00:11:49,590 --> 00:11:51,659 radar of human rights groups, no 322 00:11:51,660 --> 00:11:53,370 matter how good a job they're doing. 323 00:11:55,110 --> 00:11:57,419 So depending on the time and location, 324 00:11:57,420 --> 00:11:59,549 when someone is killed, the probability 325 00:11:59,550 --> 00:12:01,679 of he or she being documented 326 00:12:01,680 --> 00:12:03,809 by a group is likely to vary. 327 00:12:05,710 --> 00:12:07,779 And this is a problem when we're 328 00:12:07,780 --> 00:12:09,729 trying to understand patterns, because 329 00:12:09,730 --> 00:12:11,349 the patterns might be biased if we only 330 00:12:11,350 --> 00:12:12,580 look at documented violence. 331 00:12:15,150 --> 00:12:17,279 What we can do is estimate the stock 332 00:12:17,280 --> 00:12:18,899 figure using a method called multiple 333 00:12:18,900 --> 00:12:20,070 systems estimation. 334 00:12:21,090 --> 00:12:22,229 If you attended C.C.C. 335 00:12:22,230 --> 00:12:24,359 last year, you might have seen the talk 336 00:12:24,360 --> 00:12:26,789 by our founder of the Human Rights 337 00:12:26,790 --> 00:12:28,499 Then Analysis Group, Patrick Ball. 338 00:12:28,500 --> 00:12:30,359 We explained the method in much more 339 00:12:30,360 --> 00:12:32,729 detail and expanded on the case 340 00:12:32,730 --> 00:12:33,629 using Guatemala. 341 00:12:33,630 --> 00:12:34,949 So I encourage you to look at that talk. 342 00:12:34,950 --> 00:12:36,869 If you're interested in the method. 343 00:12:36,870 --> 00:12:38,399 I'm not going to go into more detail on 344 00:12:38,400 --> 00:12:40,469 it now. Importantly, this method 345 00:12:40,470 --> 00:12:42,599 can help us identify this dark 346 00:12:42,600 --> 00:12:44,819 figure. And on the one hand, 347 00:12:44,820 --> 00:12:46,589 it's important to identify the stock 348 00:12:46,590 --> 00:12:48,569 figure for the means of analysis. 349 00:12:48,570 --> 00:12:50,549 But it's also important because we really 350 00:12:50,550 --> 00:12:52,019 need to think about accounting for each 351 00:12:52,020 --> 00:12:54,419 individual who was killed by the Syrian 352 00:12:54,420 --> 00:12:56,729 regime. So there's obviously a 353 00:12:56,730 --> 00:12:58,529 much bigger importance in doing this than 354 00:12:58,530 --> 00:12:59,530 just for analysis. 355 00:13:04,110 --> 00:13:06,239 OK, so to give you an idea 356 00:13:06,240 --> 00:13:08,339 of how important it is to look at the 357 00:13:08,340 --> 00:13:10,320 dark figure of violence, this is 358 00:13:11,400 --> 00:13:14,759 a map of Syria by different governorates 359 00:13:14,760 --> 00:13:16,529 across the period of one year. 360 00:13:16,530 --> 00:13:18,899 And the darker areas show us 361 00:13:18,900 --> 00:13:20,759 where we know very little about what's 362 00:13:20,760 --> 00:13:21,929 actually happening on the ground. 363 00:13:21,930 --> 00:13:23,879 So the dark of the areas are the higher 364 00:13:23,880 --> 00:13:26,529 the percentage of undocumented killings. 365 00:13:26,530 --> 00:13:28,649 And as you can see, this high variation 366 00:13:28,650 --> 00:13:29,729 both within 367 00:13:30,930 --> 00:13:33,029 governorates across time and 368 00:13:33,030 --> 00:13:35,489 both and across different governorates. 369 00:13:35,490 --> 00:13:37,199 So if you would have only used the 370 00:13:37,200 --> 00:13:39,269 documented killings, we would 371 00:13:39,270 --> 00:13:41,669 have gotten a very different picture on 372 00:13:41,670 --> 00:13:43,109 about what's happening on the ground. 373 00:13:45,130 --> 00:13:46,659 Now, you can also imagine that someone 374 00:13:46,660 --> 00:13:47,660 who was 375 00:13:48,940 --> 00:13:51,009 intentionally detained and tortured 376 00:13:51,010 --> 00:13:52,899 by the government and then killed might 377 00:13:52,900 --> 00:13:54,309 have a different probability of being 378 00:13:54,310 --> 00:13:56,139 reported than someone who was killed in 379 00:13:56,140 --> 00:13:57,140 an explosion. 380 00:13:58,150 --> 00:14:00,219 So we need to estimate 381 00:14:00,220 --> 00:14:02,619 these different figures separately. 382 00:14:02,620 --> 00:14:04,959 And only once we have this information 383 00:14:04,960 --> 00:14:06,189 can we actually carry on with our 384 00:14:06,190 --> 00:14:07,190 analysis. 385 00:14:08,150 --> 00:14:10,999 OK, so now we have information 386 00:14:11,000 --> 00:14:13,219 on different strategies 387 00:14:13,220 --> 00:14:15,049 of violence used by the Syrian government 388 00:14:15,050 --> 00:14:16,429 across the period of one year. 389 00:14:18,400 --> 00:14:19,839 The next thing we need is information on 390 00:14:19,840 --> 00:14:22,719 Internet access in Syria, and for this, 391 00:14:22,720 --> 00:14:24,819 the Syrian Digital Security Monitor 392 00:14:24,820 --> 00:14:26,470 has been has been 393 00:14:27,670 --> 00:14:29,409 collecting information for the last one 394 00:14:29,410 --> 00:14:31,449 and a half years on the level of Internet 395 00:14:31,450 --> 00:14:33,459 accessibility within each Syrian 396 00:14:33,460 --> 00:14:34,899 district. 397 00:14:34,900 --> 00:14:36,549 I mean, they've been doing this on a 398 00:14:36,550 --> 00:14:38,109 scale from one to four. 399 00:14:38,110 --> 00:14:40,269 So either it's fully accessible, it's 400 00:14:40,270 --> 00:14:42,519 temporarily accessible, it's 401 00:14:44,470 --> 00:14:46,269 infrequently accessible, or there's no 402 00:14:46,270 --> 00:14:47,529 Internet access at all. 403 00:14:47,530 --> 00:14:49,659 And this is the information 404 00:14:49,660 --> 00:14:52,149 that they kindly shared with me that I'm 405 00:14:52,150 --> 00:14:54,219 that I'm using as a measure of 406 00:14:54,220 --> 00:14:56,619 Internet access within Syria. 407 00:14:56,620 --> 00:14:58,449 And what you can see here is how it 408 00:14:58,450 --> 00:15:00,609 varies over time across different 409 00:15:00,610 --> 00:15:02,739 Syrian governorates for 410 00:15:02,740 --> 00:15:05,109 different types of Internet access. 411 00:15:05,110 --> 00:15:07,119 Importantly, what we can see here is it's 412 00:15:07,120 --> 00:15:09,219 not just that some bomb falls and 413 00:15:09,220 --> 00:15:10,559 then all access disappeared. 414 00:15:10,560 --> 00:15:12,609 So it's not just a technical problem 415 00:15:12,610 --> 00:15:14,109 happening here, but we see, obviously, 416 00:15:14,110 --> 00:15:16,419 that the government is putting 417 00:15:16,420 --> 00:15:18,519 putting work into either providing access 418 00:15:18,520 --> 00:15:19,539 or denying access. 419 00:15:20,880 --> 00:15:23,159 OK, so what we're really interested 420 00:15:23,160 --> 00:15:25,229 in is understanding if the variation 421 00:15:25,230 --> 00:15:27,689 in this pattern affects 422 00:15:27,690 --> 00:15:29,819 or is correlated to 423 00:15:29,820 --> 00:15:32,339 the variation in the pattern of violence 424 00:15:32,340 --> 00:15:33,899 used by the Syrian government. 425 00:15:35,700 --> 00:15:37,919 So what we see here is the expected 426 00:15:37,920 --> 00:15:40,289 proportion of targeted killings given 427 00:15:40,290 --> 00:15:41,369 different levels of Internet 428 00:15:41,370 --> 00:15:42,370 accessibility. 429 00:15:43,270 --> 00:15:45,429 So where Internet is fully accessible, 430 00:15:45,430 --> 00:15:47,769 almost one quarter of all 431 00:15:47,770 --> 00:15:49,869 killings that occurred are targeted. 432 00:15:49,870 --> 00:15:51,519 So these are people who've been tortured. 433 00:15:51,520 --> 00:15:53,709 These are people who directly executed or 434 00:15:53,710 --> 00:15:54,909 these are people who were previously 435 00:15:54,910 --> 00:15:56,109 detained. 436 00:15:56,110 --> 00:15:57,819 And as we move down in the level of 437 00:15:57,820 --> 00:16:00,639 accessibility, that number decreases. 438 00:16:00,640 --> 00:16:02,829 It decreases to about eight percent where 439 00:16:02,830 --> 00:16:04,299 there's no Internet accessibility. 440 00:16:06,720 --> 00:16:08,819 Or put in words in governorates where 441 00:16:08,820 --> 00:16:11,039 the Internet is fully accessible, one 442 00:16:11,040 --> 00:16:12,630 in four or targeted killings. 443 00:16:15,060 --> 00:16:17,069 And in governorates where there's no 444 00:16:17,070 --> 00:16:19,379 Internet access, we see about less 445 00:16:19,380 --> 00:16:21,389 than one in 10 targeted killings. 446 00:16:21,390 --> 00:16:23,489 Now, I'm not making any kind of moral 447 00:16:23,490 --> 00:16:25,619 or qualitative judgment on on 448 00:16:25,620 --> 00:16:26,579 any type of killing. 449 00:16:26,580 --> 00:16:28,949 All this is saying is that the type 450 00:16:28,950 --> 00:16:30,989 of violence used by the government, by 451 00:16:30,990 --> 00:16:32,819 the government is different in 452 00:16:32,820 --> 00:16:34,829 governorates where they have Internet 453 00:16:34,830 --> 00:16:36,299 access versus ones where there's no 454 00:16:36,300 --> 00:16:37,799 Internet access. 455 00:16:37,800 --> 00:16:39,659 So the higher the Internet accessibility 456 00:16:39,660 --> 00:16:41,909 is, the more targeted 457 00:16:41,910 --> 00:16:44,129 violence the government's uses. 458 00:16:44,130 --> 00:16:46,319 And conversely, where there's almost 459 00:16:46,320 --> 00:16:48,629 no access, there's going to be much 460 00:16:48,630 --> 00:16:50,429 more indiscriminate violence used. 461 00:16:53,310 --> 00:16:55,829 OK, so in conclusion, 462 00:16:55,830 --> 00:16:57,899 I'm not making a fully causal 463 00:16:57,900 --> 00:16:59,939 argument here where I'm saying that the 464 00:16:59,940 --> 00:17:01,440 different levels of 465 00:17:02,460 --> 00:17:05,009 of accessibility directly 466 00:17:05,010 --> 00:17:06,749 cause a different type of violence. 467 00:17:06,750 --> 00:17:09,389 What I'm trying to say is that 468 00:17:09,390 --> 00:17:11,699 the choice of digital response is likely 469 00:17:11,700 --> 00:17:14,159 to inform, constrain 470 00:17:14,160 --> 00:17:16,739 or enable different types of 471 00:17:16,740 --> 00:17:18,839 of repression used by the government. 472 00:17:20,010 --> 00:17:21,598 It's likely to be part of a larger 473 00:17:21,599 --> 00:17:22,618 repressive campaign. 474 00:17:25,170 --> 00:17:27,509 OK, what we see is that it very spatially 475 00:17:27,510 --> 00:17:29,639 and temporally, so it's not just as 476 00:17:29,640 --> 00:17:32,159 if in the beginning of the conflict, 477 00:17:32,160 --> 00:17:33,239 everything is selective. 478 00:17:33,240 --> 00:17:35,039 And then as it moves over time, the 479 00:17:35,040 --> 00:17:37,169 government uses only indiscriminate 480 00:17:37,170 --> 00:17:39,239 violence. We see a big variation both 481 00:17:39,240 --> 00:17:40,240 over time and space. 482 00:17:41,840 --> 00:17:43,909 And we see that not only affect scale 483 00:17:43,910 --> 00:17:45,559 that actually affects the type or the 484 00:17:45,560 --> 00:17:47,269 nature of oppression used by governments. 485 00:17:48,590 --> 00:17:50,509 I'm happy to talk about all the methods 486 00:17:50,510 --> 00:17:52,699 and and and other 487 00:17:52,700 --> 00:17:54,619 details after the talk. 488 00:17:54,620 --> 00:17:55,909 And if you have any questions, please 489 00:17:55,910 --> 00:17:56,910 email me. Thank you. 490 00:18:06,440 --> 00:18:08,059 Thank you very much for this very 491 00:18:08,060 --> 00:18:10,309 interesting talk, so we still have a lot 492 00:18:10,310 --> 00:18:13,279 of time for questions and answers 493 00:18:13,280 --> 00:18:15,559 so people can gather up at the mikes 494 00:18:15,560 --> 00:18:17,179 and I will call you. 495 00:18:17,180 --> 00:18:19,159 Do we have a question from the I.R.S. 496 00:18:19,160 --> 00:18:20,160 or Twitter? 497 00:18:23,000 --> 00:18:24,349 Signal angel, right? 498 00:18:24,350 --> 00:18:26,539 No, not really. 499 00:18:26,540 --> 00:18:28,339 OK, then we'll go on with microphone 500 00:18:28,340 --> 00:18:30,019 number one, please. 501 00:18:30,020 --> 00:18:32,089 OK. I have actually 502 00:18:32,090 --> 00:18:33,439 two questions. 503 00:18:33,440 --> 00:18:35,789 One is in regard 504 00:18:35,790 --> 00:18:37,879 of the targeted and non targeted 505 00:18:37,880 --> 00:18:38,809 killings. 506 00:18:38,810 --> 00:18:41,450 You mentioned children as a target. 507 00:18:43,550 --> 00:18:45,859 How do you see in this perspective the 508 00:18:45,860 --> 00:18:48,439 recent Pakistani school shootings? 509 00:18:48,440 --> 00:18:50,629 I mean, I would say this was 510 00:18:50,630 --> 00:18:52,879 targeted even though not 511 00:18:52,880 --> 00:18:55,189 by the government, but by the Taliban 512 00:18:55,190 --> 00:18:57,319 and it affected children. 513 00:18:57,320 --> 00:18:59,449 So is it not 514 00:19:00,710 --> 00:19:02,779 the indicator seems to be not valid in 515 00:19:02,780 --> 00:19:04,039 all cases. 516 00:19:04,040 --> 00:19:06,259 Yes, you can agree. 517 00:19:06,260 --> 00:19:07,399 Did you have another question, though? 518 00:19:07,400 --> 00:19:08,449 Yes. OK, yeah. 519 00:19:08,450 --> 00:19:09,769 So, no, definitely. 520 00:19:09,770 --> 00:19:10,770 The the 521 00:19:11,830 --> 00:19:13,999 the the measure that I use for target 522 00:19:14,000 --> 00:19:15,289 and targeted violence is definitely 523 00:19:15,290 --> 00:19:17,029 something that situation specific. 524 00:19:17,030 --> 00:19:19,129 So so in the context of 525 00:19:19,130 --> 00:19:21,349 of of Syria, what we often see is 526 00:19:21,350 --> 00:19:23,629 that the circumstantial evidence tells us 527 00:19:23,630 --> 00:19:26,589 that these were these were, for example, 528 00:19:26,590 --> 00:19:28,249 kids who were five year old who were with 529 00:19:28,250 --> 00:19:29,989 their parents and who were bombed, for 530 00:19:29,990 --> 00:19:31,159 example. 531 00:19:31,160 --> 00:19:33,379 But we don't see very many cases 532 00:19:33,380 --> 00:19:35,209 where children are tortured, for example. 533 00:19:35,210 --> 00:19:36,949 So so I think in the Syrian case, that's 534 00:19:36,950 --> 00:19:38,599 makes sense. But you're absolutely right, 535 00:19:38,600 --> 00:19:40,669 depending on the type of 536 00:19:40,670 --> 00:19:42,859 the type of conflict we're looking at or 537 00:19:42,860 --> 00:19:45,319 non conflict, the the measure 538 00:19:45,320 --> 00:19:47,089 of the type of violence needs to be 539 00:19:47,090 --> 00:19:47,989 buried. 540 00:19:47,990 --> 00:19:50,329 And my second question is, do you 541 00:19:50,330 --> 00:19:52,489 think or do you have evidence 542 00:19:52,490 --> 00:19:54,949 that a higher number of targeted killings 543 00:19:54,950 --> 00:19:57,739 during times of full Internet access 544 00:19:57,740 --> 00:19:59,839 directly correlates to each other 545 00:19:59,840 --> 00:20:02,059 as a to do with the fact that in 546 00:20:02,060 --> 00:20:04,699 times of full Internet access, the regime 547 00:20:04,700 --> 00:20:07,219 has more and better information 548 00:20:07,220 --> 00:20:09,470 to conduct targeted killings? 549 00:20:10,550 --> 00:20:12,439 So that would be my hypothesis. 550 00:20:12,440 --> 00:20:14,689 This is not something I can directly test 551 00:20:14,690 --> 00:20:17,389 because I don't know which information 552 00:20:17,390 --> 00:20:19,189 the government is actually using, that 553 00:20:19,190 --> 00:20:20,809 it's that it's gleaning from its 554 00:20:20,810 --> 00:20:22,049 surveillance techniques. 555 00:20:22,050 --> 00:20:23,869 But I would I would argue that that is 556 00:20:23,870 --> 00:20:25,339 one of the reasons why they're probably 557 00:20:25,340 --> 00:20:27,559 more likely to directly target people 558 00:20:27,560 --> 00:20:28,669 more effectively. 559 00:20:28,670 --> 00:20:29,670 OK, thank you. 560 00:20:31,010 --> 00:20:33,349 Microphone number two, I 561 00:20:33,350 --> 00:20:35,329 just two observations, more and more 562 00:20:35,330 --> 00:20:36,289 comments than anything. 563 00:20:36,290 --> 00:20:39,019 Firstly, your construct of government 564 00:20:39,020 --> 00:20:41,119 versus civil society or society 565 00:20:41,120 --> 00:20:43,429 is kind of simple. 566 00:20:43,430 --> 00:20:45,649 Syria is a very large, porous country. 567 00:20:45,650 --> 00:20:47,869 You can enter from almost any angle 568 00:20:47,870 --> 00:20:50,269 and you have several religious 569 00:20:50,270 --> 00:20:52,099 orientations to the country. 570 00:20:52,100 --> 00:20:54,679 And so there is fractionalization 571 00:20:54,680 --> 00:20:56,779 that is not really reflected 572 00:20:56,780 --> 00:20:58,579 and it's a shifting situation. 573 00:20:58,580 --> 00:21:00,649 So if you imagine a war 574 00:21:00,650 --> 00:21:02,899 zone in which your town is shelled, 575 00:21:02,900 --> 00:21:04,369 you have no electricity, therefore you 576 00:21:04,370 --> 00:21:06,349 have no mobile phone, you have no 577 00:21:06,350 --> 00:21:08,419 Internet. So the outages 578 00:21:08,420 --> 00:21:09,859 are spurious. 579 00:21:09,860 --> 00:21:11,929 You might be accounting for 580 00:21:11,930 --> 00:21:14,149 a cut out and electricity rather than 581 00:21:14,150 --> 00:21:15,499 cut out in the network. 582 00:21:15,500 --> 00:21:17,569 So two points right to one, 583 00:21:17,570 --> 00:21:19,069 the government against people. 584 00:21:19,070 --> 00:21:20,749 So people against the government is also 585 00:21:20,750 --> 00:21:22,279 a form of violence. 586 00:21:22,280 --> 00:21:23,449 And we might be measuring the wrong 587 00:21:23,450 --> 00:21:24,450 thing. Thanks. 588 00:21:26,360 --> 00:21:28,339 Absolutely, I mean, I'm highly I mean, 589 00:21:28,340 --> 00:21:30,169 I'm using a theoretical, highly 590 00:21:30,170 --> 00:21:32,299 simplified concept, and in 591 00:21:32,300 --> 00:21:34,429 this in this specific piece of research, 592 00:21:34,430 --> 00:21:36,799 I only looked at the effect of government 593 00:21:36,800 --> 00:21:38,449 violence or basically the how government 594 00:21:38,450 --> 00:21:40,549 violence works in 595 00:21:40,550 --> 00:21:42,169 obviously Syria is a is a much more 596 00:21:42,170 --> 00:21:43,519 complicated case. 597 00:21:43,520 --> 00:21:45,649 But but if we only look at the 598 00:21:45,650 --> 00:21:48,799 side of of government against 599 00:21:48,800 --> 00:21:51,140 against the opposition, 600 00:21:52,370 --> 00:21:54,739 we can still assume that the people who 601 00:21:54,740 --> 00:21:56,389 are acting on the on the on the side of 602 00:21:56,390 --> 00:21:58,579 the government are somewhat unified 603 00:21:58,580 --> 00:22:00,289 in that they want to remain maintain the 604 00:22:00,290 --> 00:22:02,479 status quo or go back to 605 00:22:02,480 --> 00:22:04,369 the previous status quo. 606 00:22:04,370 --> 00:22:06,619 And in terms of in terms 607 00:22:06,620 --> 00:22:08,779 of measuring the wrong thing, I think 608 00:22:08,780 --> 00:22:10,999 there are probably periods within 609 00:22:11,000 --> 00:22:12,919 within the data that that that reflect 610 00:22:12,920 --> 00:22:13,920 exactly that. 611 00:22:14,570 --> 00:22:17,029 But since we see that no no governorate 612 00:22:17,030 --> 00:22:18,139 just goes off the grid. 613 00:22:18,140 --> 00:22:19,519 Right. So it's not like they're just 614 00:22:19,520 --> 00:22:20,720 bombings. And then the whole 615 00:22:22,070 --> 00:22:23,809 the whole level of accessibility goes 616 00:22:23,810 --> 00:22:24,810 down forever. 617 00:22:25,610 --> 00:22:26,909 We see that it goes back up again. 618 00:22:26,910 --> 00:22:29,509 So obviously there 619 00:22:29,510 --> 00:22:31,669 it's it's not like it's just one kind 620 00:22:31,670 --> 00:22:34,159 of uniform response, 621 00:22:34,160 --> 00:22:35,160 so. 622 00:22:36,960 --> 00:22:38,949 Thank you. Number four, please. 623 00:22:38,950 --> 00:22:41,129 I first of all, couldn't thank 624 00:22:41,130 --> 00:22:43,529 you more for putting numbers on 625 00:22:43,530 --> 00:22:45,659 this matter because I think 626 00:22:45,660 --> 00:22:47,939 it helps tremendously 627 00:22:47,940 --> 00:22:50,129 in arguing also at a political 628 00:22:50,130 --> 00:22:51,809 level, two questions. 629 00:22:51,810 --> 00:22:53,969 A, I think you 630 00:22:53,970 --> 00:22:56,249 demonstrate some very interesting 631 00:22:56,250 --> 00:22:57,909 correlations. 632 00:22:57,910 --> 00:23:00,059 Still, I 633 00:23:00,060 --> 00:23:03,209 have the feeling that the causation 634 00:23:03,210 --> 00:23:05,309 is a bit more difficult to explain. 635 00:23:05,310 --> 00:23:07,589 Could you elaborate on this and 636 00:23:07,590 --> 00:23:08,549 attach to this? 637 00:23:08,550 --> 00:23:11,189 My second question, have you made any 638 00:23:11,190 --> 00:23:13,349 temporal analysis seeing 639 00:23:13,350 --> 00:23:15,509 patterns like Internet going 640 00:23:15,510 --> 00:23:17,729 off, something happening intended going 641 00:23:17,730 --> 00:23:18,730 on? 642 00:23:19,790 --> 00:23:21,889 Yes, so so 643 00:23:21,890 --> 00:23:23,359 absolutely, as I mentioned at the end of 644 00:23:23,360 --> 00:23:25,579 my talk, I'm not using the word causation 645 00:23:25,580 --> 00:23:27,289 in the in the in the social science 646 00:23:27,290 --> 00:23:29,419 sense. In this case, what I'm 647 00:23:29,420 --> 00:23:32,269 saying is that these two these two 648 00:23:32,270 --> 00:23:33,919 phenomena, Internet control and violence, 649 00:23:33,920 --> 00:23:35,299 are likely to be part of the same 650 00:23:35,300 --> 00:23:37,459 campaign used by governments 651 00:23:37,460 --> 00:23:39,229 when they use which campaign is a very 652 00:23:39,230 --> 00:23:40,969 different question. 653 00:23:40,970 --> 00:23:42,589 So the first thing I wanted to establish 654 00:23:42,590 --> 00:23:43,849 with this was to see 655 00:23:44,900 --> 00:23:47,059 do they even Cauvery, do we see any 656 00:23:47,060 --> 00:23:49,489 kind of correlation here? 657 00:23:49,490 --> 00:23:51,589 The second question is when 658 00:23:51,590 --> 00:23:53,809 governments use what type? 659 00:23:53,810 --> 00:23:55,459 I refer to that a little bit in the 660 00:23:55,460 --> 00:23:56,359 theoretical part. 661 00:23:56,360 --> 00:23:58,879 So we're likely to see governments 662 00:23:58,880 --> 00:24:00,559 or in the case of the Syrian government, 663 00:24:00,560 --> 00:24:01,940 we're likely to see them using 664 00:24:03,070 --> 00:24:05,179 using targeted violence in areas where 665 00:24:05,180 --> 00:24:06,529 they haven't fully given up hope of 666 00:24:06,530 --> 00:24:07,819 control. Right. So where they're still 667 00:24:07,820 --> 00:24:09,889 fighting to maintain control, where they 668 00:24:09,890 --> 00:24:12,199 still have have some faint 669 00:24:12,200 --> 00:24:14,209 hope that that it'll be theirs at some 670 00:24:14,210 --> 00:24:16,519 point again. So in a sense, leaving 671 00:24:16,520 --> 00:24:18,649 the Internet on there is a 672 00:24:18,650 --> 00:24:19,650 form of 673 00:24:20,870 --> 00:24:23,029 a form of providing infrastructure for 674 00:24:23,030 --> 00:24:24,919 the people who are still supporting them 675 00:24:24,920 --> 00:24:26,749 and spying on the people who might not 676 00:24:26,750 --> 00:24:27,750 be. 677 00:24:28,070 --> 00:24:29,839 And in places where they've completely 678 00:24:29,840 --> 00:24:31,699 given up hope, where they say, OK, we're 679 00:24:31,700 --> 00:24:34,609 going to punish people here for their 680 00:24:34,610 --> 00:24:36,739 for their involvement with 681 00:24:36,740 --> 00:24:38,390 any any of the opposition groups, 682 00:24:39,860 --> 00:24:41,209 that's probably where we're going to see 683 00:24:41,210 --> 00:24:42,380 more indiscriminate violence 684 00:24:43,640 --> 00:24:45,709 in terms of temporal analysis. 685 00:24:45,710 --> 00:24:48,739 Yes, I've I've written a different 686 00:24:48,740 --> 00:24:51,799 research project on temporal analysis. 687 00:24:51,800 --> 00:24:54,229 And what we there basically see is that 688 00:24:54,230 --> 00:24:56,479 during network outfall network outages, 689 00:24:56,480 --> 00:24:57,859 as they occurred a couple of times in 690 00:24:57,860 --> 00:25:00,019 Syria, the level of violence, 691 00:25:00,020 --> 00:25:01,639 the scale of violence increases 692 00:25:01,640 --> 00:25:04,099 tremendously and significantly 693 00:25:04,100 --> 00:25:06,380 so. So there's definitely some kind of 694 00:25:08,240 --> 00:25:10,429 larger strategic idea 695 00:25:10,430 --> 00:25:11,389 going on behind that. 696 00:25:11,390 --> 00:25:13,339 I'd love to talk to someone from the 697 00:25:13,340 --> 00:25:14,749 Syrian government, but I haven't been 698 00:25:14,750 --> 00:25:16,609 able to up until now. 699 00:25:16,610 --> 00:25:17,099 Thank you. 700 00:25:17,100 --> 00:25:18,100 Yeah. 701 00:25:18,440 --> 00:25:20,329 Thank you. So, Christian from the 702 00:25:20,330 --> 00:25:21,330 Internet. 703 00:25:21,800 --> 00:25:23,039 Uh, yeah. 704 00:25:23,040 --> 00:25:25,369 Um, is any 705 00:25:25,370 --> 00:25:27,919 work planned or always done 706 00:25:27,920 --> 00:25:30,650 regarding what is happening in Ukraine? 707 00:25:33,820 --> 00:25:35,649 I haven't I haven't worked on the 708 00:25:35,650 --> 00:25:37,749 Ukrainian case, I'm 709 00:25:37,750 --> 00:25:39,549 I'm aware of other other researchers 710 00:25:39,550 --> 00:25:41,169 working on the Ukrainian case, but I 711 00:25:41,170 --> 00:25:42,909 think that's that's also going to be very 712 00:25:42,910 --> 00:25:45,099 interesting, very 713 00:25:45,100 --> 00:25:47,169 interesting example of how governments 714 00:25:47,170 --> 00:25:49,809 can strategically use the Internet 715 00:25:49,810 --> 00:25:50,949 and even more there. 716 00:25:50,950 --> 00:25:52,179 I mean, what I've what I've been looking 717 00:25:52,180 --> 00:25:54,279 at here is kind of very crudely looking 718 00:25:54,280 --> 00:25:55,959 at Internet access as a form of 719 00:25:55,960 --> 00:25:57,369 information control. 720 00:25:57,370 --> 00:25:59,259 But actual information manipulation might 721 00:25:59,260 --> 00:26:00,249 be something that would be really 722 00:26:00,250 --> 00:26:02,289 interesting. And looking at the Ukrainian 723 00:26:02,290 --> 00:26:03,290 case. 724 00:26:04,300 --> 00:26:06,579 Then microphone number one, please. 725 00:26:06,580 --> 00:26:09,489 I thanks for the talk first. 726 00:26:09,490 --> 00:26:11,649 Could that be that in areas with 727 00:26:11,650 --> 00:26:14,019 less or no Internet access, there's 728 00:26:14,020 --> 00:26:16,179 just less information getting 729 00:26:16,180 --> 00:26:18,579 out that could identify a death 730 00:26:18,580 --> 00:26:21,159 as a targeted killing so 731 00:26:21,160 --> 00:26:23,379 that the correlation isn't as 732 00:26:23,380 --> 00:26:25,569 strictly or as strong between 733 00:26:25,570 --> 00:26:26,739 those? And there's a little bit 734 00:26:26,740 --> 00:26:29,169 correlation between no Internet access 735 00:26:29,170 --> 00:26:32,019 and less information 736 00:26:32,020 --> 00:26:34,179 that identifies killings as 737 00:26:34,180 --> 00:26:35,309 such. 738 00:26:35,310 --> 00:26:37,209 Um, no, we don't find that. 739 00:26:37,210 --> 00:26:38,469 That's a very good point. And I think 740 00:26:38,470 --> 00:26:40,629 what we do see is that where 741 00:26:40,630 --> 00:26:43,089 there's no Internet access, we know 742 00:26:43,090 --> 00:26:45,369 less about the number of people killed. 743 00:26:45,370 --> 00:26:47,499 But the information that we do have so 744 00:26:47,500 --> 00:26:49,799 on recorded killings is just as good. 745 00:26:49,800 --> 00:26:51,849 So we have information on that. 746 00:26:51,850 --> 00:26:53,619 The people were, for example, killed in 747 00:26:53,620 --> 00:26:54,469 bombings. 748 00:26:54,470 --> 00:26:56,889 So the quality of the information 749 00:26:56,890 --> 00:26:58,209 stays roughly the same. 750 00:26:58,210 --> 00:27:00,309 It's just less the quality stays 751 00:27:00,310 --> 00:27:02,479 the same. The quantity sometimes 752 00:27:02,480 --> 00:27:03,129 varies. Yes. 753 00:27:03,130 --> 00:27:04,130 Thank you very much. 754 00:27:05,450 --> 00:27:06,950 Another question from the Internet. 755 00:27:08,300 --> 00:27:10,879 Uh, yeah, um, 756 00:27:10,880 --> 00:27:13,009 there was a question about this 757 00:27:13,010 --> 00:27:14,599 whole scattergories. 758 00:27:14,600 --> 00:27:16,969 Uh, somebody want to know if you can 759 00:27:16,970 --> 00:27:19,549 name explicitly some of these 760 00:27:19,550 --> 00:27:20,550 sources. 761 00:27:21,740 --> 00:27:23,899 So, yeah, so the sources we use, 762 00:27:23,900 --> 00:27:26,389 human rights groups working in Syria, 763 00:27:26,390 --> 00:27:28,759 so if you do have it, 764 00:27:28,760 --> 00:27:31,009 if you have a look at the 765 00:27:31,010 --> 00:27:32,090 report we wrote, 766 00:27:34,280 --> 00:27:35,539 where is it? Yeah. 767 00:27:35,540 --> 00:27:37,549 So the report names the groups in detail 768 00:27:37,550 --> 00:27:39,409 and therefore human rights groups that 769 00:27:39,410 --> 00:27:41,269 have been working actively on the ground 770 00:27:41,270 --> 00:27:42,739 and they've been doing absolutely amazing 771 00:27:42,740 --> 00:27:43,849 work. 772 00:27:43,850 --> 00:27:46,039 So when I talk about holes 773 00:27:46,040 --> 00:27:47,449 in the documentation, this doesn't mean 774 00:27:47,450 --> 00:27:49,009 that these human rights groups aren't 775 00:27:49,010 --> 00:27:50,629 doing great work. They're doing amazing 776 00:27:50,630 --> 00:27:52,369 work. It's just that some areas are 777 00:27:52,370 --> 00:27:53,959 almost inaccessible for them. 778 00:27:53,960 --> 00:27:55,220 So, yeah. 779 00:27:58,480 --> 00:28:00,759 OK, then, one last question for my two. 780 00:28:02,740 --> 00:28:03,740 Hello. 781 00:28:04,690 --> 00:28:07,449 My question is about the mythology, 782 00:28:07,450 --> 00:28:09,579 how you determine if 783 00:28:09,580 --> 00:28:10,869 there is Internet access. 784 00:28:10,870 --> 00:28:13,089 And another question 785 00:28:13,090 --> 00:28:14,460 you partly answered already. 786 00:28:15,490 --> 00:28:17,649 I think the proportion between targeted 787 00:28:17,650 --> 00:28:19,959 killings and non targeted killings 788 00:28:19,960 --> 00:28:22,269 regarding in relation to Internet 789 00:28:22,270 --> 00:28:24,649 access is like a proportion. 790 00:28:24,650 --> 00:28:27,249 But have you also looked at the 791 00:28:27,250 --> 00:28:30,099 absolute figures, how they change with 792 00:28:30,100 --> 00:28:31,809 having Internet access or maybe having 793 00:28:31,810 --> 00:28:33,130 some Internet access? 794 00:28:34,530 --> 00:28:36,629 Yeah, yeah, um, 795 00:28:36,630 --> 00:28:38,069 so. 796 00:28:38,070 --> 00:28:40,289 So in terms of the methodology, as 797 00:28:40,290 --> 00:28:41,489 I say, the data I use that Internet 798 00:28:41,490 --> 00:28:42,899 access wasn't something I collected 799 00:28:42,900 --> 00:28:46,439 myself, but as far as I understand, 800 00:28:46,440 --> 00:28:48,629 they've they they partly 801 00:28:48,630 --> 00:28:50,130 had people helping them 802 00:28:51,690 --> 00:28:53,789 collect collect information on the ground 803 00:28:53,790 --> 00:28:55,109 within each district. 804 00:28:55,110 --> 00:28:57,419 So they had people who were 805 00:28:57,420 --> 00:28:59,549 who were telling them every month 806 00:28:59,550 --> 00:29:01,019 we've been able to access the Internet 807 00:29:01,020 --> 00:29:02,669 via our phone. We've been able to access 808 00:29:02,670 --> 00:29:04,379 the Internet via DSL and so on. 809 00:29:04,380 --> 00:29:05,309 Like a survey. 810 00:29:05,310 --> 00:29:06,359 Yes, like a survey. 811 00:29:09,040 --> 00:29:11,349 I'd be very interested in hearing 812 00:29:11,350 --> 00:29:13,539 other ways to measure this, maybe on a on 813 00:29:13,540 --> 00:29:14,889 a more discreet scale. 814 00:29:14,890 --> 00:29:16,989 So if if people have 815 00:29:16,990 --> 00:29:18,699 ideas on that, I'd be I'd be very 816 00:29:18,700 --> 00:29:19,779 interested to hear about that. 817 00:29:21,430 --> 00:29:23,799 I haven't looked at the the differences 818 00:29:23,800 --> 00:29:25,869 in scale of violence when 819 00:29:25,870 --> 00:29:27,669 it comes to different different levels of 820 00:29:27,670 --> 00:29:28,659 NetWare accessibility. 821 00:29:28,660 --> 00:29:30,399 What I have done is looked at the 822 00:29:30,400 --> 00:29:32,079 difference in the scale of violence 823 00:29:32,080 --> 00:29:33,849 between days where the Internet is off 824 00:29:33,850 --> 00:29:35,439 and days where the Internet is on. 825 00:29:35,440 --> 00:29:37,269 And there we do see a significant 826 00:29:37,270 --> 00:29:38,799 difference. So on days where the Internet 827 00:29:38,800 --> 00:29:40,809 is off, we see a much higher level of 828 00:29:40,810 --> 00:29:42,909 violence and this is even controlling 829 00:29:42,910 --> 00:29:44,139 for documentation patterns and 830 00:29:44,140 --> 00:29:46,089 everything. So it's not just an artifact 831 00:29:46,090 --> 00:29:47,229 in the data. 832 00:29:47,230 --> 00:29:48,230 Thank you. 833 00:29:48,970 --> 00:29:50,469 Thank you very much for your questions. 834 00:29:50,470 --> 00:29:52,149 If you still have questions, you can just 835 00:29:52,150 --> 00:29:53,829 after the talk, walk up to any time to 836 00:29:53,830 --> 00:29:54,789 ask her. 837 00:29:54,790 --> 00:29:56,229 And then I would like to thank you again 838 00:29:56,230 --> 00:29:57,339 for your presentation. 839 00:29:57,340 --> 00:29:58,340 Thank you.