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/742 Thanks! 1 00:00:13,400 --> 00:00:15,649 A British mathematician once said 2 00:00:15,650 --> 00:00:17,149 that data is the new oil. 3 00:00:18,450 --> 00:00:20,639 Our data points are collected 4 00:00:20,640 --> 00:00:23,069 and analyzed, they're utilized 5 00:00:23,070 --> 00:00:24,070 en masse. 6 00:00:24,870 --> 00:00:27,479 Our swipes, our purchases, 7 00:00:27,480 --> 00:00:29,879 what we're interested in, how we move, 8 00:00:29,880 --> 00:00:32,339 how we hold our phone, anything 9 00:00:32,340 --> 00:00:34,949 can create material 10 00:00:34,950 --> 00:00:36,659 for analytical companies to make 11 00:00:36,660 --> 00:00:39,419 extensive profiles of us 12 00:00:39,420 --> 00:00:40,979 for the purpose of selling those. 13 00:00:42,090 --> 00:00:44,159 So in these in the so-called economy 14 00:00:44,160 --> 00:00:46,859 of surveillance, a lot of actors 15 00:00:46,860 --> 00:00:48,000 make a lot of money, 16 00:00:49,290 --> 00:00:51,539 maybe not the user, him or herself 17 00:00:51,540 --> 00:00:53,639 so much, but how 18 00:00:53,640 --> 00:00:55,499 this works in detail. 19 00:00:55,500 --> 00:00:57,689 That is going to tell us what Crystal 20 00:00:57,690 --> 00:00:58,690 today. 21 00:00:59,250 --> 00:01:02,009 Beaufoy Krista is a technologist, 22 00:01:02,010 --> 00:01:04,109 researcher and activist 23 00:01:04,110 --> 00:01:06,569 from Austria and together 24 00:01:06,570 --> 00:01:08,749 with the privacy scholar Sarah 25 00:01:08,750 --> 00:01:10,889 Speakman, he wrote Networks 26 00:01:10,890 --> 00:01:12,719 of Control. This report has been 27 00:01:12,720 --> 00:01:15,089 published this year. 28 00:01:15,090 --> 00:01:16,499 In October. 29 00:01:16,500 --> 00:01:18,809 He has also developed a 30 00:01:18,810 --> 00:01:20,909 game called Data Dealer that some 31 00:01:20,910 --> 00:01:23,429 people might know, which is about privacy 32 00:01:23,430 --> 00:01:25,349 and surveillance. 33 00:01:25,350 --> 00:01:28,049 Today he is going to tell us, 34 00:01:28,050 --> 00:01:29,639 give us a talk about corporate 35 00:01:29,640 --> 00:01:32,069 surveillance, digital tracking, 36 00:01:32,070 --> 00:01:33,989 big data and privacy. 37 00:01:33,990 --> 00:01:36,180 Please help me welcome BofI Krysta. 38 00:01:43,230 --> 00:01:44,639 Thanks a lot. 39 00:01:44,640 --> 00:01:45,539 Can you hear me? 40 00:01:45,540 --> 00:01:46,679 Yes. 41 00:01:46,680 --> 00:01:47,680 Hi, everybody. 42 00:01:48,930 --> 00:01:49,859 My name is William Kristol. 43 00:01:49,860 --> 00:01:51,959 I'm from Vienna, Austria. 44 00:01:51,960 --> 00:01:54,359 And first of all, 45 00:01:54,360 --> 00:01:56,429 I'd like to modify the title of my talk a 46 00:01:56,430 --> 00:01:58,890 bit and get rid of the big data. 47 00:02:01,040 --> 00:02:02,659 To be honest, I only use it because it 48 00:02:02,660 --> 00:02:04,729 always helps to get some people 49 00:02:04,730 --> 00:02:06,169 to listen to me. 50 00:02:06,170 --> 00:02:08,478 I mean, big data can mean everything 51 00:02:08,479 --> 00:02:10,668 and nothing, so let's just 52 00:02:10,669 --> 00:02:12,739 get rid of it, OK? 53 00:02:12,740 --> 00:02:15,379 Nevertheless, during the last few years, 54 00:02:15,380 --> 00:02:18,139 we've seen the birth of a large scale 55 00:02:18,140 --> 00:02:19,519 surveillance economy. 56 00:02:19,520 --> 00:02:21,679 Today, thousands of businesses are 57 00:02:21,680 --> 00:02:24,019 monitoring, profiling, categorizing, 58 00:02:24,020 --> 00:02:26,149 raiding and affecting 59 00:02:26,150 --> 00:02:28,819 the lives of billions across platforms, 60 00:02:28,820 --> 00:02:31,549 devices and live contexts. 61 00:02:31,550 --> 00:02:33,649 In my presentation, I'll talk about how 62 00:02:33,650 --> 00:02:35,449 networks of these companies are 63 00:02:35,450 --> 00:02:38,569 collecting, analyzing and utilizing 64 00:02:38,570 --> 00:02:40,909 our data, often hardly 65 00:02:40,910 --> 00:02:43,099 without effectively informed 66 00:02:43,100 --> 00:02:45,469 consent and even largely 67 00:02:45,470 --> 00:02:46,849 without our knowledge. 68 00:02:46,850 --> 00:02:48,799 I will also talk about how personal data 69 00:02:48,800 --> 00:02:50,989 analytics is already being used 70 00:02:50,990 --> 00:02:52,879 in fields like marketing, retail, 71 00:02:52,880 --> 00:02:55,399 insurance, finance, health care, 72 00:02:55,400 --> 00:02:57,529 employment and so on to 73 00:02:57,530 --> 00:02:59,479 make decisions and people. 74 00:03:00,500 --> 00:03:03,259 I'll take you on a small tour through 75 00:03:03,260 --> 00:03:05,569 examples, many examples 76 00:03:05,570 --> 00:03:07,489 of corporate practices. 77 00:03:07,490 --> 00:03:10,069 And I'll also address the question 78 00:03:10,070 --> 00:03:12,229 what could possibly go 79 00:03:12,230 --> 00:03:13,230 wrong? 80 00:03:15,710 --> 00:03:18,229 But beware, I will use some more bullshit 81 00:03:18,230 --> 00:03:21,019 terms because I have to 82 00:03:21,020 --> 00:03:23,359 if you want to discuss all this marketing 83 00:03:23,360 --> 00:03:25,519 data sphere and so on, 84 00:03:25,520 --> 00:03:27,739 we unfortunately have to dove into some 85 00:03:27,740 --> 00:03:29,929 marketing data bullshit 86 00:03:29,930 --> 00:03:32,029 terms as 87 00:03:32,030 --> 00:03:34,369 well. I'm sorry. And I also will use 88 00:03:34,370 --> 00:03:36,860 the big data term again. 89 00:03:38,210 --> 00:03:40,339 And this is likely, many of you 90 00:03:40,340 --> 00:03:43,069 probably know it, a browser extension 91 00:03:43,070 --> 00:03:45,349 which shows who is watching us 92 00:03:45,350 --> 00:03:47,179 when we are surfing the web. 93 00:03:47,180 --> 00:03:49,489 I visited five websites here, 94 00:03:49,490 --> 00:03:52,279 a large health website, weather site, 95 00:03:52,280 --> 00:03:54,859 dictionary dot com, the New York Times 96 00:03:54,860 --> 00:03:57,079 and what magazine in 97 00:03:57,080 --> 00:03:58,879 the background these five sites connected 98 00:03:58,880 --> 00:04:01,489 to one hundred and eighteen 99 00:04:01,490 --> 00:04:03,739 other third party services and told 100 00:04:03,740 --> 00:04:06,349 them about my visit, 101 00:04:06,350 --> 00:04:08,569 only five websites, and it was recorded 102 00:04:08,570 --> 00:04:11,030 by more than 100 companies. 103 00:04:12,380 --> 00:04:13,909 Let's have a look at it a bit more in 104 00:04:13,910 --> 00:04:15,109 detail. 105 00:04:15,110 --> 00:04:16,909 The five websites I visited are 106 00:04:16,910 --> 00:04:19,939 represented by the circus, 107 00:04:19,940 --> 00:04:22,189 the third parties by this 108 00:04:22,190 --> 00:04:25,279 many small Trangie icons 109 00:04:25,280 --> 00:04:26,569 you can see. 110 00:04:26,570 --> 00:04:28,789 So why do these five 111 00:04:28,790 --> 00:04:31,009 websites transmit information 112 00:04:31,010 --> 00:04:34,099 about clicks to other companies? 113 00:04:34,100 --> 00:04:36,439 Because they actively put some small 114 00:04:36,440 --> 00:04:39,079 pieces of code into their websites. 115 00:04:40,190 --> 00:04:42,709 And who are these third party companies 116 00:04:42,710 --> 00:04:45,259 represented by the small Trangie 117 00:04:45,260 --> 00:04:47,779 icons, ad networks, 118 00:04:47,780 --> 00:04:49,999 analytics services, consumer 119 00:04:50,000 --> 00:04:52,129 data brokers, and, of course, Google 120 00:04:52,130 --> 00:04:55,459 and Facebook and many others? 121 00:04:55,460 --> 00:04:58,069 As you can see, some of the third parties 122 00:04:58,070 --> 00:05:01,429 are connected to two or more websites 123 00:05:01,430 --> 00:05:02,329 visited. 124 00:05:02,330 --> 00:05:04,459 That means they're able to monitor 125 00:05:04,460 --> 00:05:06,559 and track me across several 126 00:05:06,560 --> 00:05:07,759 websites. 127 00:05:07,760 --> 00:05:09,859 Now, imagine what happens when I'm 128 00:05:09,860 --> 00:05:11,600 surfing the web for a day, 129 00:05:12,800 --> 00:05:14,570 a week or a month. 130 00:05:15,800 --> 00:05:18,649 This is how thousands of companies 131 00:05:18,650 --> 00:05:20,059 and companies are able to compile 132 00:05:20,060 --> 00:05:22,459 profiles of other online behavior. 133 00:05:22,460 --> 00:05:24,589 I know there are some 134 00:05:24,590 --> 00:05:27,139 tools to block some of these trackers, 135 00:05:27,140 --> 00:05:28,969 but most people outside the privacy 136 00:05:28,970 --> 00:05:31,699 bubble are still very surprised 137 00:05:31,700 --> 00:05:33,619 when they see this. 138 00:05:33,620 --> 00:05:35,479 And beside of that, some of these anti 139 00:05:35,480 --> 00:05:37,399 tracking tools became part of the 140 00:05:37,400 --> 00:05:39,679 tracking ecosystem as well. 141 00:05:40,820 --> 00:05:43,039 Of course, it's not only surfing the web, 142 00:05:43,040 --> 00:05:45,349 there's also our smartphone, 143 00:05:45,350 --> 00:05:47,449 a powerful small computer, 144 00:05:47,450 --> 00:05:49,909 even if it's a bit broken, even 145 00:05:49,910 --> 00:05:52,249 now and then, like this 146 00:05:52,250 --> 00:05:54,379 one, containing lists of our 147 00:05:54,380 --> 00:05:56,539 contacts, our friends, our calls, our 148 00:05:56,540 --> 00:05:58,669 messages, it tracks our movements, lots 149 00:05:58,670 --> 00:06:01,699 of very private information. 150 00:06:01,700 --> 00:06:03,979 And who is able to access this data? 151 00:06:03,980 --> 00:06:06,079 Many apps if we give them permission 152 00:06:06,080 --> 00:06:07,279 to access it. 153 00:06:07,280 --> 00:06:09,469 And what's the business model of many 154 00:06:09,470 --> 00:06:11,659 app developers collecting and 155 00:06:11,660 --> 00:06:14,569 selling our personal information? 156 00:06:14,570 --> 00:06:17,059 The interesting thing is that most people 157 00:06:17,060 --> 00:06:19,759 would not hand over detailed information 158 00:06:19,760 --> 00:06:21,799 about all the contacts, addresses and 159 00:06:21,800 --> 00:06:24,109 movements to other people 160 00:06:24,110 --> 00:06:25,110 they don't know. 161 00:06:25,880 --> 00:06:28,159 At the same time, most people don't 162 00:06:28,160 --> 00:06:30,289 seem to care to 163 00:06:30,290 --> 00:06:33,139 hand it over to thousands of companies 164 00:06:33,140 --> 00:06:34,140 they don't know. 165 00:06:34,760 --> 00:06:36,469 Back to the five websites I visited. 166 00:06:36,470 --> 00:06:39,649 Let's have a look at some of the 118 167 00:06:39,650 --> 00:06:41,839 third parties which received information 168 00:06:41,840 --> 00:06:43,609 about me surfing the Web. 169 00:06:43,610 --> 00:06:45,789 Let me introduce you to look 170 00:06:45,790 --> 00:06:47,959 at this and look 171 00:06:47,960 --> 00:06:49,999 at these two domains. 172 00:06:50,000 --> 00:06:52,009 Tracked my visits to the weather website 173 00:06:52,010 --> 00:06:53,329 and to W magazine. 174 00:06:53,330 --> 00:06:55,849 I'm sure you've seen at this 175 00:06:55,850 --> 00:06:58,099 already they offer website providers 176 00:06:58,100 --> 00:06:59,489 to show these tiny little. 177 00:06:59,490 --> 00:07:02,009 Social sharing patterns, modeling, 178 00:07:02,010 --> 00:07:04,259 looking similar to that and blue 179 00:07:04,260 --> 00:07:06,389 sky, you 180 00:07:06,390 --> 00:07:08,639 won't see it on the website you visited, 181 00:07:08,640 --> 00:07:11,549 this is a typical data collection service 182 00:07:11,550 --> 00:07:14,279 and both at this and blue sky 183 00:07:14,280 --> 00:07:15,769 belong to orecchiette, 184 00:07:17,370 --> 00:07:19,979 one of the world's largest database 185 00:07:19,980 --> 00:07:21,539 and business of the providers. 186 00:07:21,540 --> 00:07:23,609 Irakly hasn't been known as a consumer 187 00:07:23,610 --> 00:07:25,739 data broker until recently, 188 00:07:25,740 --> 00:07:27,839 but in the last years they acquired 189 00:07:27,840 --> 00:07:30,059 several data companies, Blue 190 00:07:30,060 --> 00:07:32,559 Sky and Online Data Marketplace. 191 00:07:32,560 --> 00:07:35,219 They are Logic's, which has partnerships 192 00:07:35,220 --> 00:07:37,529 with stores who offer membership 193 00:07:37,530 --> 00:07:39,869 or loyalty cards, collecting 194 00:07:39,870 --> 00:07:42,689 purchase data from 1500 195 00:07:42,690 --> 00:07:45,209 large retailers, and 196 00:07:45,210 --> 00:07:47,069 their logic's is able to link those 197 00:07:47,070 --> 00:07:49,859 purchases to the digital world. 198 00:07:49,860 --> 00:07:52,019 And Oracle acquired at this, 199 00:07:52,020 --> 00:07:54,479 which is harvesting behavioral data 200 00:07:54,480 --> 00:07:56,909 about website visitors are more than 15 201 00:07:56,910 --> 00:07:59,819 million different websites. 202 00:07:59,820 --> 00:08:02,429 Orica has now integrated these companies 203 00:08:02,430 --> 00:08:04,929 into its data cloud. 204 00:08:04,930 --> 00:08:06,209 Very nice name. 205 00:08:07,560 --> 00:08:09,959 This is a nice slide about it here. 206 00:08:09,960 --> 00:08:12,359 The company explains that the record 207 00:08:12,360 --> 00:08:14,579 that they record, what consumers 208 00:08:14,580 --> 00:08:16,799 do, what consumers say and 209 00:08:16,800 --> 00:08:17,850 what consumers buy. 210 00:08:19,260 --> 00:08:20,260 Great threat. 211 00:08:21,630 --> 00:08:23,819 According to Iraq's own statements, the 212 00:08:23,820 --> 00:08:26,039 aggregate three billion user 213 00:08:26,040 --> 00:08:28,229 profiles from 214 00:08:28,230 --> 00:08:30,269 the 15 million websites which have which 215 00:08:30,270 --> 00:08:32,699 have at this installed, but also 216 00:08:32,700 --> 00:08:34,949 700 million social 217 00:08:34,950 --> 00:08:36,629 messages daily. 218 00:08:38,220 --> 00:08:40,408 They claim to do that and 219 00:08:40,409 --> 00:08:42,538 billions of purchases in order 220 00:08:42,539 --> 00:08:45,209 to target people, personalized content 221 00:08:45,210 --> 00:08:47,279 and to measure how people 222 00:08:47,280 --> 00:08:49,349 are interacting across platforms, 223 00:08:49,350 --> 00:08:51,719 channels and devices 224 00:08:51,720 --> 00:08:54,209 from online mobile email, 225 00:08:54,210 --> 00:08:56,759 social media to TV radio, 226 00:08:56,760 --> 00:08:59,339 direct mail and even in-store. 227 00:08:59,340 --> 00:09:01,559 They also try to monitor and link 228 00:09:01,560 --> 00:09:03,869 offline purchases. 229 00:09:03,870 --> 00:09:06,509 And in the center you can see 230 00:09:06,510 --> 00:09:08,609 the Iraqi identity 231 00:09:08,610 --> 00:09:10,739 graph, which allows them to link 232 00:09:10,740 --> 00:09:12,899 and match user profiles across 233 00:09:12,900 --> 00:09:15,059 platforms, devices and company 234 00:09:15,060 --> 00:09:17,249 databases in order to 235 00:09:17,250 --> 00:09:19,649 create one addressable 236 00:09:19,650 --> 00:09:23,009 consumer profile as they right 237 00:09:23,010 --> 00:09:25,109 to identify customers 238 00:09:25,110 --> 00:09:27,179 everywhere, to 239 00:09:27,180 --> 00:09:30,239 unify addressable identities 240 00:09:30,240 --> 00:09:32,009 and so on and so on. 241 00:09:32,010 --> 00:09:34,109 Here can use here you can see the 242 00:09:34,110 --> 00:09:36,489 different ideas from the cookie 243 00:09:36,490 --> 00:09:38,849 idea into email at Deposal and 244 00:09:38,850 --> 00:09:40,979 Mobile, the set 245 00:09:40,980 --> 00:09:43,139 up idea. This is the 246 00:09:43,140 --> 00:09:45,269 set top boxes for TVs. 247 00:09:45,270 --> 00:09:48,089 So they also try to link that information 248 00:09:48,090 --> 00:09:50,969 and probably also from other devices 249 00:09:50,970 --> 00:09:52,379 and platforms. 250 00:09:52,380 --> 00:09:54,479 Irakly also provides a wide range of 251 00:09:54,480 --> 00:09:56,459 data from partners. 252 00:09:56,460 --> 00:09:59,399 This is already a data directory. 253 00:09:59,400 --> 00:10:01,589 Quite interesting document already 254 00:10:01,590 --> 00:10:03,749 provides shares and combines 255 00:10:03,750 --> 00:10:06,149 data from data brokers like Axium 256 00:10:06,150 --> 00:10:09,299 Inter-Group LIFESTAR, 257 00:10:09,300 --> 00:10:11,399 also from Experian, TransUnion and 258 00:10:11,400 --> 00:10:14,279 Equifax. These are the three big 259 00:10:14,280 --> 00:10:16,439 credit reporting agencies 260 00:10:16,440 --> 00:10:19,019 in the US and many others. 261 00:10:19,020 --> 00:10:21,299 They also provide data from credit 262 00:10:21,300 --> 00:10:23,519 card companies like Visa and 263 00:10:23,520 --> 00:10:25,919 MasterCard and of course, Oriente, 264 00:10:25,920 --> 00:10:28,169 also partners with Google and 265 00:10:28,170 --> 00:10:29,939 Facebook data. 266 00:10:29,940 --> 00:10:31,829 Large Data Logic's, one of these 267 00:10:31,830 --> 00:10:34,049 companies already acquired, was 268 00:10:34,050 --> 00:10:36,179 even one of the first data brokers 269 00:10:36,180 --> 00:10:38,129 which started to partner with Facebook 270 00:10:38,130 --> 00:10:40,469 back in 2012. 271 00:10:40,470 --> 00:10:42,569 They connected their offline data about 272 00:10:42,570 --> 00:10:44,609 purchases, financial behavior, credit 273 00:10:44,610 --> 00:10:47,129 cards and home value, net worth income. 274 00:10:47,130 --> 00:10:49,499 And along with Facebook's 275 00:10:49,500 --> 00:10:51,629 rich user profile 276 00:10:51,630 --> 00:10:53,819 information, of course, platforms 277 00:10:53,820 --> 00:10:56,039 like Facebook or Facebook, as 278 00:10:56,040 --> 00:10:57,040 I'd like to call it, 279 00:10:58,230 --> 00:11:00,809 is collecting vast amounts of information 280 00:11:00,810 --> 00:11:03,059 itself about the everyday 281 00:11:03,060 --> 00:11:05,309 lives of one point eight billion 282 00:11:05,310 --> 00:11:07,529 people and still one point two billion 283 00:11:07,530 --> 00:11:09,489 people use it every day. 284 00:11:09,490 --> 00:11:11,969 Facebook puts these nearly two 285 00:11:11,970 --> 00:11:13,859 billion people into thousands of 286 00:11:13,860 --> 00:11:16,109 categories and lets advertisers use 287 00:11:16,110 --> 00:11:18,509 these categories in order to include 288 00:11:18,510 --> 00:11:20,729 or exclude people from ads 289 00:11:20,730 --> 00:11:22,979 on an individual level. 290 00:11:22,980 --> 00:11:25,229 A few weeks ago, the US based nonprofit 291 00:11:25,230 --> 00:11:27,389 ProPublica was able to purchase a 292 00:11:27,390 --> 00:11:29,459 housing it on Facebook 293 00:11:29,460 --> 00:11:31,949 that wanted to address people who are 294 00:11:31,950 --> 00:11:33,240 likely to move 295 00:11:34,800 --> 00:11:36,600 and are interested in houses 296 00:11:37,650 --> 00:11:38,969 and so on. 297 00:11:38,970 --> 00:11:41,189 But if you look at the bottom, 298 00:11:41,190 --> 00:11:43,229 you can also exclude people who met 299 00:11:43,230 --> 00:11:45,419 specific criteria. 300 00:11:45,420 --> 00:11:47,489 In this case, people who are categories 301 00:11:47,490 --> 00:11:50,159 categorized as Hispanics, African 302 00:11:50,160 --> 00:11:52,229 or Asian Americans are 303 00:11:52,230 --> 00:11:54,359 excluded from seeing 304 00:11:54,360 --> 00:11:55,349 the EBT. 305 00:11:55,350 --> 00:11:57,509 Beside of that, this is probably 306 00:11:57,510 --> 00:11:59,159 illegal discrimination. 307 00:11:59,160 --> 00:12:01,559 The US, how do they know 308 00:12:01,560 --> 00:12:03,749 about the ethnic affinity, how they call 309 00:12:03,750 --> 00:12:06,029 it of someone? 310 00:12:06,030 --> 00:12:09,189 And are these classifications accurate? 311 00:12:09,190 --> 00:12:11,279 Basically, we don't know, 312 00:12:11,280 --> 00:12:13,229 but maybe they use similar methods. 313 00:12:13,230 --> 00:12:15,539 As in this Berkeley study, an academic 314 00:12:15,540 --> 00:12:17,879 paper, the researchers 315 00:12:17,880 --> 00:12:20,069 try to predict these 316 00:12:20,070 --> 00:12:22,439 private attributes of users just based 317 00:12:22,440 --> 00:12:25,019 on the Facebook likes. 318 00:12:25,020 --> 00:12:27,000 And those were the results 319 00:12:28,260 --> 00:12:29,909 they were able to successfully predict 320 00:12:29,910 --> 00:12:32,129 ethnicity, sexual orientation, 321 00:12:32,130 --> 00:12:34,349 political and religious views 322 00:12:34,350 --> 00:12:35,639 just based on data. 323 00:12:35,640 --> 00:12:37,769 About 170 324 00:12:37,770 --> 00:12:39,539 Facebook likes per user. 325 00:12:41,950 --> 00:12:44,049 The study had 60000 participants, and 326 00:12:44,050 --> 00:12:46,209 as you can see, ethnicity can be 327 00:12:46,210 --> 00:12:48,009 predicted quite accurately. 328 00:12:49,210 --> 00:12:51,339 In Europe, Facebook doesn't offer 329 00:12:51,340 --> 00:12:53,319 to directly categorize people according 330 00:12:53,320 --> 00:12:54,969 to their ethnicity, but they still 331 00:12:54,970 --> 00:12:56,319 support people in many different 332 00:12:56,320 --> 00:12:57,519 categories. 333 00:12:57,520 --> 00:12:59,439 For example, you could purchase an ad on 334 00:12:59,440 --> 00:13:01,689 Facebook targeted on people 335 00:13:01,690 --> 00:13:03,529 living in Norway. 336 00:13:03,530 --> 00:13:04,989 My last talk was in Oslo. 337 00:13:04,990 --> 00:13:06,549 So this is not where 338 00:13:08,680 --> 00:13:10,869 25 to 40 years old speaking 339 00:13:10,870 --> 00:13:12,879 a specific kind of Norwegian language and 340 00:13:12,880 --> 00:13:15,009 then excludes people 341 00:13:15,010 --> 00:13:17,379 who are interested in 342 00:13:17,380 --> 00:13:19,599 Arabic language Islam, 343 00:13:19,600 --> 00:13:22,029 multiple sclerosis, online gambling, 344 00:13:22,030 --> 00:13:24,249 personality disorder, plus size 345 00:13:24,250 --> 00:13:26,319 clothing, stomach 346 00:13:26,320 --> 00:13:28,689 cancer, trade union, 347 00:13:28,690 --> 00:13:30,519 wheelchair and so on. 348 00:13:30,520 --> 00:13:33,189 These are basically protected 349 00:13:33,190 --> 00:13:35,529 categories of data 350 00:13:35,530 --> 00:13:36,530 in Europe. 351 00:13:37,750 --> 00:13:39,939 So ethnic profiling isn't directly 352 00:13:39,940 --> 00:13:42,159 available in Europe, but you can still 353 00:13:42,160 --> 00:13:43,160 use proxies. 354 00:13:44,510 --> 00:13:45,969 However, relax. 355 00:13:45,970 --> 00:13:48,069 After massive criticism and 356 00:13:48,070 --> 00:13:50,229 media coverage, Facebook told us in 357 00:13:50,230 --> 00:13:52,089 November that it will build a system to 358 00:13:52,090 --> 00:13:54,579 prevent advertisers from bank credit, 359 00:13:54,580 --> 00:13:56,709 housing or employment ads that 360 00:13:56,710 --> 00:13:58,779 exclude viewers by 361 00:13:58,780 --> 00:14:00,519 race. So take it easy. 362 00:14:00,520 --> 00:14:02,799 Read then 363 00:14:02,800 --> 00:14:04,539 wait only about ads. 364 00:14:04,540 --> 00:14:06,059 Who cares? 365 00:14:06,060 --> 00:14:08,269 Also, a few weeks ago, when the large 366 00:14:08,270 --> 00:14:10,809 UK insurer ADM announced to introduce 367 00:14:10,810 --> 00:14:13,299 to Price car insurance 368 00:14:13,300 --> 00:14:15,669 based on Facebook posts, Facebook quickly 369 00:14:15,670 --> 00:14:18,099 reacted and blocked the insurers 370 00:14:18,100 --> 00:14:21,309 app Perfect Red. 371 00:14:21,310 --> 00:14:22,779 On the other hand, Facebook itself 372 00:14:22,780 --> 00:14:24,909 registered a patent about credit 373 00:14:24,910 --> 00:14:27,070 scoring based on Facebook data. 374 00:14:28,750 --> 00:14:31,119 The right. When an individual applies 375 00:14:31,120 --> 00:14:32,919 for a loan, the lender examines the 376 00:14:32,920 --> 00:14:35,509 credit ratings of members of individual 377 00:14:35,510 --> 00:14:36,909 social network. 378 00:14:36,910 --> 00:14:38,499 So from your friends, 379 00:14:39,850 --> 00:14:41,919 if the average credit rating of 380 00:14:41,920 --> 00:14:44,019 these members is at least a minimum 381 00:14:44,020 --> 00:14:46,029 credit score, the lender continues to 382 00:14:46,030 --> 00:14:47,919 process the loan application. 383 00:14:47,920 --> 00:14:49,809 Otherwise, the loan application is 384 00:14:49,810 --> 00:14:50,810 rejected. 385 00:14:51,970 --> 00:14:53,919 It's Facebook planning to provide credit 386 00:14:53,920 --> 00:14:55,899 scores based on data about our Facebook 387 00:14:55,900 --> 00:14:56,900 friends. 388 00:14:57,570 --> 00:14:59,789 We don't know Face will probably 389 00:14:59,790 --> 00:15:01,829 say it's just a patent. 390 00:15:01,830 --> 00:15:03,929 We won't do it, but 391 00:15:03,930 --> 00:15:05,220 can we trust this company? 392 00:15:08,450 --> 00:15:10,819 In 2014, when Facebook acquired WhatsApp, 393 00:15:10,820 --> 00:15:12,499 they said, don't worry, your WhatsApp 394 00:15:12,500 --> 00:15:14,149 data is safe, we won't share it with 395 00:15:14,150 --> 00:15:16,259 Facebook 2016, 396 00:15:16,260 --> 00:15:17,809 the company announced to start sharing 397 00:15:17,810 --> 00:15:18,980 WhatsApp data with Facebook. 398 00:15:20,240 --> 00:15:22,219 After continued criticism, Facebook told 399 00:15:22,220 --> 00:15:24,379 the Europeans that 400 00:15:24,380 --> 00:15:26,479 it won't combine Facebook and WhatsApp 401 00:15:26,480 --> 00:15:28,309 that data for now, 402 00:15:30,380 --> 00:15:31,399 whatever it is, I see it. 403 00:15:31,400 --> 00:15:33,289 Like many data companies, Facebook mostly 404 00:15:33,290 --> 00:15:35,329 acts in a way like two steps forward, one 405 00:15:35,330 --> 00:15:38,059 step back and so on. 406 00:15:38,060 --> 00:15:39,769 We'll see. Of course, it's not only the 407 00:15:39,770 --> 00:15:41,929 web, smartphone apps and 408 00:15:41,930 --> 00:15:43,669 Facebook. There are many ways of how 409 00:15:43,670 --> 00:15:45,229 personal data is being collected and 410 00:15:45,230 --> 00:15:46,669 utilized today. 411 00:15:46,670 --> 00:15:48,650 Let's have a look at visual DNA. 412 00:15:49,850 --> 00:15:52,579 They use data from online quizzes 413 00:15:52,580 --> 00:15:54,319 to create personality profiles about 414 00:15:54,320 --> 00:15:55,219 users. 415 00:15:55,220 --> 00:15:57,199 They say that the quizzes have been taken 416 00:15:57,200 --> 00:16:00,439 by 40 million people already 417 00:16:00,440 --> 00:16:02,959 altogether. Visual DNA provides 418 00:16:02,960 --> 00:16:05,359 digital profiles, about 500 million 419 00:16:05,360 --> 00:16:07,459 consumers, for example, for marketing 420 00:16:07,460 --> 00:16:10,129 and online targeting purposes. 421 00:16:10,130 --> 00:16:12,469 But also and that's remarkable 422 00:16:12,470 --> 00:16:14,539 for credit scoring and 423 00:16:14,540 --> 00:16:15,739 risk assessment. 424 00:16:15,740 --> 00:16:18,169 Therefore, we DNA partners with 425 00:16:18,170 --> 00:16:20,449 MasterCard, the credit reporting 426 00:16:20,450 --> 00:16:22,549 agency Experian and 427 00:16:22,550 --> 00:16:24,679 ADM, the large UK 428 00:16:24,680 --> 00:16:25,849 insurer. 429 00:16:25,850 --> 00:16:27,769 You remember me from before? 430 00:16:30,380 --> 00:16:32,239 This is remarkable because they cross the 431 00:16:32,240 --> 00:16:34,369 line between the context of marketing 432 00:16:34,370 --> 00:16:36,919 data on the one hand and risk management 433 00:16:36,920 --> 00:16:39,049 data, credit scoring on the other 434 00:16:39,050 --> 00:16:40,009 hand. 435 00:16:40,010 --> 00:16:41,569 And if you remember or against data 436 00:16:41,570 --> 00:16:43,849 directory, IRAKLY also offers 437 00:16:43,850 --> 00:16:46,010 data from visual Vicini. 438 00:16:47,840 --> 00:16:49,939 As you can see, although we've got 439 00:16:49,940 --> 00:16:52,069 some large players, we're talking about a 440 00:16:52,070 --> 00:16:54,379 landscape of many interlinked 441 00:16:54,380 --> 00:16:56,569 commercial databases about people. 442 00:16:56,570 --> 00:16:58,729 This is what I call networks 443 00:16:58,730 --> 00:17:01,009 of corporate surveillance. 444 00:17:01,010 --> 00:17:04,368 For example, have a look at segment, 445 00:17:04,369 --> 00:17:06,469 a company which says Collect all 446 00:17:06,470 --> 00:17:08,088 of your customer data and send it 447 00:17:08,089 --> 00:17:09,089 anywhere. 448 00:17:10,430 --> 00:17:12,139 It provides tools for website and app 449 00:17:12,140 --> 00:17:14,239 developers to easily collect data from 450 00:17:14,240 --> 00:17:16,669 their users and then automatically 451 00:17:16,670 --> 00:17:18,949 send it to more than 150 other 452 00:17:18,950 --> 00:17:21,019 companies from 453 00:17:21,020 --> 00:17:23,179 ad networks, data brokers and analytics 454 00:17:23,180 --> 00:17:25,399 providers to CRM system and even 455 00:17:25,400 --> 00:17:27,200 fraud detection services. 456 00:17:28,640 --> 00:17:31,459 This is only one third of the services 457 00:17:31,460 --> 00:17:33,229 that companies can automatically sent 458 00:17:33,230 --> 00:17:35,629 their user on customer data to a nice 459 00:17:35,630 --> 00:17:37,519 logo wall. 460 00:17:37,520 --> 00:17:39,679 And by the way, this one, 461 00:17:39,680 --> 00:17:42,199 Eloqua, also belongs to Oracle. 462 00:17:42,200 --> 00:17:44,779 So networks of corporate surveillance. 463 00:17:44,780 --> 00:17:46,939 But which types of data about people 464 00:17:46,940 --> 00:17:48,709 do companies collect and trade? 465 00:17:48,710 --> 00:17:50,899 One way to group it 466 00:17:50,900 --> 00:17:53,089 would be on how personal data is being 467 00:17:53,090 --> 00:17:54,709 obtained. 468 00:17:54,710 --> 00:17:56,899 First, volunteer data, which 469 00:17:56,900 --> 00:17:59,149 is created and explicitly shared 470 00:17:59,150 --> 00:18:01,490 by individuals, at least in theory, 471 00:18:02,780 --> 00:18:04,429 for example, address information in 472 00:18:04,430 --> 00:18:05,430 online form 473 00:18:06,560 --> 00:18:09,169 or maybe the qu'est data from visual DNA. 474 00:18:09,170 --> 00:18:11,269 But I'm not sure if people were really 475 00:18:11,270 --> 00:18:13,129 aware what will happen with that 476 00:18:13,130 --> 00:18:14,089 information. 477 00:18:14,090 --> 00:18:17,059 Second, observed data 478 00:18:17,060 --> 00:18:19,219 consumers generated passively 479 00:18:19,220 --> 00:18:21,559 after it gets recorded completely without 480 00:18:21,560 --> 00:18:22,999 their knowledge. 481 00:18:23,000 --> 00:18:24,889 For example, when a company tracks the 482 00:18:24,890 --> 00:18:27,079 receipt, the receipts, the websites 483 00:18:27,080 --> 00:18:29,659 we visit and finally 484 00:18:29,660 --> 00:18:31,759 inferred data, which is based on 485 00:18:31,760 --> 00:18:34,009 the analysis of volunteered 486 00:18:34,010 --> 00:18:36,569 our observed information 487 00:18:36,570 --> 00:18:38,149 another way to group it, based on the 488 00:18:38,150 --> 00:18:39,680 contents of the data 489 00:18:41,000 --> 00:18:43,279 companies directly or indirectly collect 490 00:18:43,280 --> 00:18:46,729 financial information about people, 491 00:18:46,730 --> 00:18:48,829 for example, about their income or credit 492 00:18:48,830 --> 00:18:50,119 ratings. 493 00:18:50,120 --> 00:18:53,029 Of course, contact data, demographic 494 00:18:53,030 --> 00:18:55,159 attributes like age, gender or 495 00:18:55,160 --> 00:18:57,979 ethnicity, transactional data 496 00:18:57,980 --> 00:19:00,079 such as purchases and the price is 497 00:19:00,080 --> 00:19:02,479 paid, contractual information 498 00:19:02,480 --> 00:19:04,579 like service details and history, for 499 00:19:04,580 --> 00:19:07,369 example, by utility and mobile 500 00:19:07,370 --> 00:19:09,169 network provider. 501 00:19:09,170 --> 00:19:11,359 And by the way, this categorization 502 00:19:11,360 --> 00:19:13,279 is from a data broker point of view. 503 00:19:14,450 --> 00:19:16,729 Information collected also includes 504 00:19:16,730 --> 00:19:18,949 location data, not only from 505 00:19:18,950 --> 00:19:21,259 mobile devices, behavioral 506 00:19:21,260 --> 00:19:23,629 data like the websites visited, rap 507 00:19:23,630 --> 00:19:26,389 usage, technical identifiers 508 00:19:26,390 --> 00:19:28,879 such as IP addresses or device 509 00:19:28,880 --> 00:19:31,519 I.D., and not least 510 00:19:31,520 --> 00:19:33,559 communication contents like social media 511 00:19:33,560 --> 00:19:35,899 posts or email, text and social 512 00:19:35,900 --> 00:19:36,919 relationships. 513 00:19:36,920 --> 00:19:39,139 Information about your contacts and 514 00:19:39,140 --> 00:19:41,239 your friends to just take one 515 00:19:41,240 --> 00:19:43,519 of these, especially location data, 516 00:19:43,520 --> 00:19:45,709 is very popular in 517 00:19:45,710 --> 00:19:47,089 corporate surveillance. 518 00:19:47,090 --> 00:19:49,399 Even mobile network providers 519 00:19:49,400 --> 00:19:52,309 are increasingly selling insights 520 00:19:52,310 --> 00:19:55,399 into their location databases. 521 00:19:55,400 --> 00:19:57,919 This chart is from an industry report 522 00:19:57,920 --> 00:20:00,289 which carefully investigates which ways 523 00:20:00,290 --> 00:20:02,389 of utilizing location data 524 00:20:02,390 --> 00:20:04,699 could be the best for a mobile 525 00:20:04,700 --> 00:20:05,750 network provider, 526 00:20:07,340 --> 00:20:10,489 maybe based on apps and chips, location 527 00:20:10,490 --> 00:20:13,519 or on network cell location. 528 00:20:13,520 --> 00:20:15,979 What about in their location tracking 529 00:20:15,980 --> 00:20:18,559 or even use the emergency 530 00:20:18,560 --> 00:20:19,990 services location? 531 00:20:21,990 --> 00:20:23,699 Look at that, they even think about 532 00:20:23,700 --> 00:20:26,369 selling the emergency location, 533 00:20:27,510 --> 00:20:29,159 which which which would have many 534 00:20:29,160 --> 00:20:31,019 advantages if you look at the green 535 00:20:31,020 --> 00:20:32,400 fields, for example. 536 00:20:33,600 --> 00:20:36,629 No need to install anything on the device 537 00:20:36,630 --> 00:20:39,089 to get the emergency location, 538 00:20:39,090 --> 00:20:40,090 correct? 539 00:20:40,590 --> 00:20:41,549 Yeah. 540 00:20:41,550 --> 00:20:43,019 However, I don't want to explore this in 541 00:20:43,020 --> 00:20:45,239 detail now. It's also a very small and 542 00:20:45,240 --> 00:20:46,289 no. 543 00:20:46,290 --> 00:20:48,599 Instead, I want to show you a part 544 00:20:48,600 --> 00:20:51,419 of this marketing video of factorial, 545 00:20:51,420 --> 00:20:53,579 a company providing a product 546 00:20:53,580 --> 00:20:56,189 which is called observation 547 00:20:56,190 --> 00:20:57,190 graph. 548 00:20:57,970 --> 00:21:00,269 They've got such nice 549 00:21:00,270 --> 00:21:01,459 names of the products, 550 00:21:02,760 --> 00:21:05,499 they are combining geographic information 551 00:21:05,500 --> 00:21:08,489 and meta metadata about events in places 552 00:21:08,490 --> 00:21:11,129 with mobile location data of users. 553 00:21:11,130 --> 00:21:12,419 So here we go. 554 00:21:12,420 --> 00:21:14,279 This observation graph technology is a 555 00:21:14,280 --> 00:21:16,229 revolutionary way of understanding the 556 00:21:16,230 --> 00:21:18,119 real world behavior of mobile users. 557 00:21:21,170 --> 00:21:22,789 With products powered by observation 558 00:21:22,790 --> 00:21:24,709 graph, you can deliver intelligent mobile 559 00:21:24,710 --> 00:21:26,299 experiences. 560 00:21:26,300 --> 00:21:28,669 Observation Graph is built on factual, 561 00:21:28,670 --> 00:21:30,949 proprietary Global Places data, 562 00:21:30,950 --> 00:21:33,079 which includes over 90 million local 563 00:21:33,080 --> 00:21:35,479 businesses and points of interest in 50 564 00:21:35,480 --> 00:21:37,609 countries, and is integrated into leading 565 00:21:37,610 --> 00:21:40,369 applications such as Apple Maps, 566 00:21:40,370 --> 00:21:42,349 Facebook places and Microsoft. 567 00:21:42,350 --> 00:21:44,569 Bing observation graph also 568 00:21:44,570 --> 00:21:46,939 uses demographic data, event 569 00:21:46,940 --> 00:21:49,159 data and other geographic data 570 00:21:49,160 --> 00:21:51,230 to fully understand the physical world. 571 00:21:52,700 --> 00:21:54,799 All of this data, combined with signals 572 00:21:54,800 --> 00:21:57,049 from mobile devices, enables factual 573 00:21:57,050 --> 00:21:59,390 to catalog real world user behavior 574 00:22:01,910 --> 00:22:04,399 each day. Observation graph generates 575 00:22:04,400 --> 00:22:06,499 billions of discrete observations 576 00:22:06,500 --> 00:22:07,500 globally. 577 00:22:16,860 --> 00:22:18,869 Observation that powers products that 578 00:22:18,870 --> 00:22:21,269 enable advertisers to create highly 579 00:22:21,270 --> 00:22:23,669 accurate mobile audiences by describing 580 00:22:23,670 --> 00:22:26,219 specific real world behaviors. 581 00:22:26,220 --> 00:22:27,220 OK, enough 582 00:22:28,470 --> 00:22:30,779 user I.D. nine, eight, seven, one, 583 00:22:30,780 --> 00:22:32,669 two, three activity. 584 00:22:32,670 --> 00:22:34,769 Bocking sounds 585 00:22:34,770 --> 00:22:36,359 so innocent, right? 586 00:22:37,770 --> 00:22:39,690 What about activity protesting? 587 00:22:41,190 --> 00:22:44,139 I guess the Eastern Germany 588 00:22:44,140 --> 00:22:45,749 Chairman Stassi would have been really 589 00:22:45,750 --> 00:22:47,190 happy to have it to looked at 590 00:22:48,870 --> 00:22:51,029 as you may be. US police departments 591 00:22:51,030 --> 00:22:53,579 have already used data feeds containing 592 00:22:53,580 --> 00:22:56,129 location data from marketing analytics 593 00:22:56,130 --> 00:22:59,249 companies to track protesters. 594 00:22:59,250 --> 00:23:01,499 But location data is just one of many 595 00:23:01,500 --> 00:23:03,059 different types of personal data 596 00:23:03,060 --> 00:23:04,169 companies are collecting. 597 00:23:04,170 --> 00:23:06,629 Another way to group the data collected 598 00:23:06,630 --> 00:23:08,759 by companies is to distinguish 599 00:23:08,760 --> 00:23:11,069 between first party and third 600 00:23:11,070 --> 00:23:13,469 party data, while first 601 00:23:13,470 --> 00:23:15,569 party data is collected by businesses 602 00:23:15,570 --> 00:23:17,519 which have a direct relationship with 603 00:23:17,520 --> 00:23:18,659 consumers. 604 00:23:18,660 --> 00:23:20,369 For example, the shop you've bought 605 00:23:20,370 --> 00:23:22,949 something your mobile network provider 606 00:23:22,950 --> 00:23:24,959 are the app you've installed on your 607 00:23:24,960 --> 00:23:25,919 phone. 608 00:23:25,920 --> 00:23:28,259 Third party data is either purchase 609 00:23:28,260 --> 00:23:31,019 or licensed from a first party, 610 00:23:31,020 --> 00:23:33,449 or it is collected from publicly 611 00:23:33,450 --> 00:23:35,519 available sources. 612 00:23:35,520 --> 00:23:37,589 However, as we've seen, third party data 613 00:23:37,590 --> 00:23:40,409 collection can also be invisibly embedded 614 00:23:40,410 --> 00:23:42,719 into websites, mobile apps 615 00:23:42,720 --> 00:23:45,659 and other first party contexts. 616 00:23:45,660 --> 00:23:47,459 Finally, data being collected by 617 00:23:47,460 --> 00:23:49,589 companies could also be grouped into 618 00:23:49,590 --> 00:23:52,169 actual and modeled 619 00:23:52,170 --> 00:23:54,389 data, where actual data 620 00:23:54,390 --> 00:23:56,459 is straightforward information about 621 00:23:56,460 --> 00:23:57,689 individuals. 622 00:23:57,690 --> 00:23:59,819 For example, the postal address, 623 00:23:59,820 --> 00:24:02,309 the date of birth, or the fact 624 00:24:02,310 --> 00:24:04,709 that they've been at a specific location, 625 00:24:04,710 --> 00:24:07,829 or that they've bought a specific product 626 00:24:07,830 --> 00:24:09,899 and model. Data, on the 627 00:24:09,900 --> 00:24:12,059 other hand, results from 628 00:24:12,060 --> 00:24:14,219 drawing inferences about personal 629 00:24:14,220 --> 00:24:17,249 attributes or predicted behavior 630 00:24:17,250 --> 00:24:19,409 in marketing. For example, at your 631 00:24:19,410 --> 00:24:21,569 segment, segmenting 632 00:24:21,570 --> 00:24:23,249 people into groups with shared 633 00:24:23,250 --> 00:24:25,680 characteristics and likely behaviors 634 00:24:26,910 --> 00:24:30,569 already dates back to the 1970s. 635 00:24:30,570 --> 00:24:33,599 For example, segments could be 636 00:24:33,600 --> 00:24:34,499 segments. 637 00:24:34,500 --> 00:24:36,659 Data brokers could be rich posers who 638 00:24:36,660 --> 00:24:39,359 will likely buy an expensive car 639 00:24:39,360 --> 00:24:41,459 or old woman who will 640 00:24:41,460 --> 00:24:42,899 likely donate for a loan. 641 00:24:42,900 --> 00:24:45,269 Some animals are 642 00:24:45,270 --> 00:24:47,849 simply valuable customers 643 00:24:47,850 --> 00:24:50,579 on the one hand, and waste 644 00:24:50,580 --> 00:24:52,289 and the other. Hence, data brokers have 645 00:24:52,290 --> 00:24:54,719 really used the waste 646 00:24:54,720 --> 00:24:57,479 to categorize people in financial 647 00:24:57,480 --> 00:24:58,480 difficulties. 648 00:24:59,490 --> 00:25:02,309 But while earlier consumer segmentation 649 00:25:02,310 --> 00:25:03,839 was mainly based on large scale 650 00:25:03,840 --> 00:25:06,149 information like census data 651 00:25:06,150 --> 00:25:08,429 or surveys with small 652 00:25:08,430 --> 00:25:10,949 sample sizes, today, segmenting 653 00:25:10,950 --> 00:25:13,079 systems can use detailed 654 00:25:13,080 --> 00:25:15,269 individual level information about 655 00:25:15,270 --> 00:25:17,849 billions of consumers in real time 656 00:25:17,850 --> 00:25:20,099 and then are somehow related. 657 00:25:20,100 --> 00:25:22,439 Concept is scoring, which 658 00:25:22,440 --> 00:25:24,779 emerged more on the risk side 659 00:25:24,780 --> 00:25:25,799 of personal data. 660 00:25:25,800 --> 00:25:27,599 Business credit scoring has been around 661 00:25:27,600 --> 00:25:28,889 for decades. 662 00:25:28,890 --> 00:25:31,169 A credit score is a number 663 00:25:31,170 --> 00:25:32,759 which claims to describe your 664 00:25:32,760 --> 00:25:34,829 creditworthiness or to 665 00:25:34,830 --> 00:25:37,109 predict predict a person's 666 00:25:37,110 --> 00:25:39,539 future payment behavior 667 00:25:39,540 --> 00:25:41,699 on an individual level. 668 00:25:41,700 --> 00:25:43,859 The FICO score, still one 669 00:25:43,860 --> 00:25:45,569 of the most important credit scores in 670 00:25:45,570 --> 00:25:47,909 the US, is based on the consumer's 671 00:25:47,910 --> 00:25:50,639 payment history, the amount owed, 672 00:25:50,640 --> 00:25:52,319 the length of the credit history and 673 00:25:52,320 --> 00:25:53,729 other information. 674 00:25:53,730 --> 00:25:56,159 Even though this is a rather conservative 675 00:25:56,160 --> 00:25:58,259 mix of data compared to 676 00:25:58,260 --> 00:26:00,509 many other of today's scoring 677 00:26:00,510 --> 00:26:03,089 machines, credit scoring can still 678 00:26:03,090 --> 00:26:05,459 destroy lives 679 00:26:05,460 --> 00:26:07,019 in the US and in other countries. 680 00:26:07,020 --> 00:26:09,389 A bad credit score can not only mean 681 00:26:09,390 --> 00:26:11,519 that the person doesn't get access 682 00:26:11,520 --> 00:26:13,859 to financial services, but also 683 00:26:13,860 --> 00:26:17,039 that this person doesn't get an apartment 684 00:26:17,040 --> 00:26:18,510 or a shop anymore. 685 00:26:19,860 --> 00:26:21,719 So when somebody who is already in 686 00:26:21,720 --> 00:26:23,709 financial trouble doesn't get a job 687 00:26:23,710 --> 00:26:25,889 anymore, situation can get 688 00:26:25,890 --> 00:26:27,359 even worse. 689 00:26:27,360 --> 00:26:29,909 This way, a bad credit score can become 690 00:26:29,910 --> 00:26:32,699 a self-fulfilling prophecy. 691 00:26:32,700 --> 00:26:34,919 In addition, a credit score 692 00:26:34,920 --> 00:26:37,379 may be based on flawed data 693 00:26:37,380 --> 00:26:39,809 in someone's credit report or unflawed 694 00:26:39,810 --> 00:26:42,029 prediction algorithms. 695 00:26:42,030 --> 00:26:44,339 The latter are typically completely 696 00:26:44,340 --> 00:26:45,299 secret. 697 00:26:45,300 --> 00:26:47,339 Today's current products are often based 698 00:26:47,340 --> 00:26:49,439 on a much wider range of data, 699 00:26:49,440 --> 00:26:51,809 and they're used in many other contexts, 700 00:26:51,810 --> 00:26:53,279 not only in the field of personal 701 00:26:53,280 --> 00:26:54,280 finance. 702 00:26:54,900 --> 00:26:57,209 If you look at trust in online fraud 703 00:26:57,210 --> 00:26:59,489 detection company, they 704 00:26:59,490 --> 00:27:01,079 seem to be rather open minded. 705 00:27:01,080 --> 00:27:02,909 On the website you can see which kinds of 706 00:27:02,910 --> 00:27:05,219 data they process to keep 707 00:27:05,220 --> 00:27:07,379 fraud out while letting good customers 708 00:27:07,380 --> 00:27:08,380 through. 709 00:27:10,080 --> 00:27:11,819 What they're doing is a kind of fraud 710 00:27:11,820 --> 00:27:14,039 scoring, which can decide about 711 00:27:14,040 --> 00:27:15,139 which pay. 712 00:27:15,140 --> 00:27:16,999 Shipping options you get in an online 713 00:27:17,000 --> 00:27:19,129 shop or whether you're accepted 714 00:27:19,130 --> 00:27:22,519 or not as a customer in general, 715 00:27:22,520 --> 00:27:23,899 they use many different types of 716 00:27:23,900 --> 00:27:26,119 information phone numbers, email, 717 00:27:26,120 --> 00:27:28,129 postal addresses, browsing antibodies, 718 00:27:28,130 --> 00:27:29,779 fingerprints, credit checks. 719 00:27:29,780 --> 00:27:32,149 Transaction is the IP address is mobile 720 00:27:32,150 --> 00:27:34,309 carrier details, cell location and much 721 00:27:34,310 --> 00:27:36,439 more of its 722 00:27:36,440 --> 00:27:38,359 parent company, TransUnion. 723 00:27:38,360 --> 00:27:40,279 One of the three large credit reporting 724 00:27:40,280 --> 00:27:42,259 agencies in the US has data on one 725 00:27:42,260 --> 00:27:44,419 billion consumers globally 726 00:27:44,420 --> 00:27:46,729 obtained from 90000 data 727 00:27:46,730 --> 00:27:47,730 sources. 728 00:27:48,930 --> 00:27:51,059 Frascati, another example of a company 729 00:27:51,060 --> 00:27:53,159 which provides scoring based on personal 730 00:27:53,160 --> 00:27:55,979 data completely out of context to signify 731 00:27:55,980 --> 00:27:58,139 a US company which calculates credit 732 00:27:58,140 --> 00:28:00,419 scores for individuals from phone 733 00:28:00,420 --> 00:28:02,279 call records. 734 00:28:02,280 --> 00:28:05,039 They explain to allies called durations 735 00:28:05,040 --> 00:28:07,259 of the time calls are made, 736 00:28:07,260 --> 00:28:09,329 who is frequently called and so 737 00:28:09,330 --> 00:28:11,429 on. And they say four weeks of 738 00:28:11,430 --> 00:28:13,739 calling history is enough for 739 00:28:13,740 --> 00:28:15,899 us to predict the creditworthiness 740 00:28:15,900 --> 00:28:16,900 of someone. 741 00:28:17,760 --> 00:28:19,829 The partners include large mobile 742 00:28:19,830 --> 00:28:21,750 network providers like Telefonica 743 00:28:22,980 --> 00:28:24,929 and Equifax, one of the three largest 744 00:28:24,930 --> 00:28:27,089 consumer credit reporting 745 00:28:27,090 --> 00:28:29,249 agencies in the US. 746 00:28:29,250 --> 00:28:31,439 But how does Signifier calculate their 747 00:28:31,440 --> 00:28:33,929 credit scores from phone call records? 748 00:28:33,930 --> 00:28:36,079 The answer is we don't 749 00:28:36,080 --> 00:28:38,279 know, but maybe they use 750 00:28:38,280 --> 00:28:39,269 similar methods. 751 00:28:39,270 --> 00:28:41,579 As in the following academic 752 00:28:41,580 --> 00:28:43,439 study, researchers found that they are 753 00:28:43,440 --> 00:28:45,899 able to predict someone's personality 754 00:28:45,900 --> 00:28:47,999 just from someone's smartphone 755 00:28:48,000 --> 00:28:50,699 metadata like call dates, frequencies 756 00:28:50,700 --> 00:28:52,019 and durations. 757 00:28:52,020 --> 00:28:54,389 They use offline personality, 758 00:28:54,390 --> 00:28:56,579 personality questionnaires recorded 759 00:28:56,580 --> 00:28:58,709 to participants phone data 760 00:28:58,710 --> 00:29:02,249 and found statistical correlations. 761 00:29:02,250 --> 00:29:04,319 The prediction accuracy is far from 762 00:29:04,320 --> 00:29:06,509 perfect, but still significantly 763 00:29:06,510 --> 00:29:07,589 above chance. 764 00:29:07,590 --> 00:29:08,969 Michael J. 765 00:29:08,970 --> 00:29:11,099 So maybe companies like Signifier just 766 00:29:11,100 --> 00:29:13,259 use similar methods. 767 00:29:13,260 --> 00:29:14,260 We simply don't know. 768 00:29:15,870 --> 00:29:17,699 So what is this all about? 769 00:29:17,700 --> 00:29:19,859 It's all about data mining involving 770 00:29:19,860 --> 00:29:22,319 methods from mathematics and statistics 771 00:29:22,320 --> 00:29:23,820 to machine learning. 772 00:29:24,900 --> 00:29:27,029 Machine learning algorithms learn 773 00:29:27,030 --> 00:29:28,319 from existing data. 774 00:29:28,320 --> 00:29:30,419 They get trained to find correlations 775 00:29:30,420 --> 00:29:32,579 in large datasets, and they're able to 776 00:29:32,580 --> 00:29:35,339 more or less find connections 777 00:29:35,340 --> 00:29:37,679 between variables where human 778 00:29:37,680 --> 00:29:40,259 beings really give up. 779 00:29:40,260 --> 00:29:42,329 Another example in credit scoring. 780 00:29:42,330 --> 00:29:44,399 This is Test Finance, a company which 781 00:29:44,400 --> 00:29:46,589 has been founded by Google's 782 00:29:46,590 --> 00:29:49,229 former chief information officer. 783 00:29:49,230 --> 00:29:50,819 They're combining thousands of data 784 00:29:50,820 --> 00:29:53,129 elements to calculate credit scores 785 00:29:53,130 --> 00:29:55,289 about consumers, which data 786 00:29:55,290 --> 00:29:57,569 from how people use smartphones 787 00:29:57,570 --> 00:30:00,209 and social networks to dispelling 788 00:30:00,210 --> 00:30:02,279 somebody uses in an online loan 789 00:30:02,280 --> 00:30:03,480 application form. 790 00:30:04,530 --> 00:30:06,659 In 2016, Test Finance 791 00:30:06,660 --> 00:30:09,059 announced a partnership with Baidu, 792 00:30:09,060 --> 00:30:12,389 China's largest website search provider 793 00:30:12,390 --> 00:30:14,619 like Google in China, test 794 00:30:14,620 --> 00:30:17,009 finances that by Rich uses 795 00:30:17,010 --> 00:30:19,289 search data will be valuable for loan 796 00:30:19,290 --> 00:30:21,359 underwriting and assessing 797 00:30:21,360 --> 00:30:23,820 credit risk and a website this state. 798 00:30:25,020 --> 00:30:27,329 We believe that more data is always 799 00:30:27,330 --> 00:30:28,330 better, 800 00:30:29,550 --> 00:30:31,649 and the founder once said all data 801 00:30:31,650 --> 00:30:33,599 is credit data. We just don't know how to 802 00:30:33,600 --> 00:30:34,600 use it yet. 803 00:30:35,760 --> 00:30:38,189 All data, there's lots of data about 804 00:30:38,190 --> 00:30:39,869 us out there today. 805 00:30:39,870 --> 00:30:41,759 Virtually everything we do is recorded, 806 00:30:41,760 --> 00:30:44,039 monitored, attract in some way. 807 00:30:44,040 --> 00:30:46,139 All kinds of collected data end up 808 00:30:46,140 --> 00:30:49,559 in huge clusters of databases, 809 00:30:49,560 --> 00:30:52,019 data mining technologies, how to find 810 00:30:52,020 --> 00:30:54,239 the relevant information in these 811 00:30:54,240 --> 00:30:56,879 massive amounts of data. 812 00:30:56,880 --> 00:30:59,369 And then there's one US company 813 00:30:59,370 --> 00:31:01,019 which is always mentioned when it comes 814 00:31:01,020 --> 00:31:02,699 to consumer data brokers. 815 00:31:02,700 --> 00:31:04,829 It's called Acxiom, who 816 00:31:04,830 --> 00:31:06,059 has heard of Acxiom already. 817 00:31:07,990 --> 00:31:09,009 So quite some 818 00:31:10,220 --> 00:31:12,279 extreme is one of the largest of these 819 00:31:12,280 --> 00:31:14,469 companies and says, is that up to three? 820 00:31:14,470 --> 00:31:16,749 It has up to 3000 attributes 821 00:31:16,750 --> 00:31:18,520 and 700 million people. 822 00:31:19,570 --> 00:31:21,279 For example, they have the credit 823 00:31:21,280 --> 00:31:23,559 history, driving history, 824 00:31:23,560 --> 00:31:25,839 criminal history, residential history, 825 00:31:25,840 --> 00:31:28,119 employment history, education history, 826 00:31:29,230 --> 00:31:32,139 information about income 827 00:31:32,140 --> 00:31:34,900 and so on, purchase behavior. 828 00:31:35,920 --> 00:31:38,379 They don't collect information 829 00:31:38,380 --> 00:31:40,719 about illnesses of people, but about 830 00:31:40,720 --> 00:31:41,829 health interests. 831 00:31:45,430 --> 00:31:48,009 By the way, this is a marketing 832 00:31:48,010 --> 00:31:50,379 new form of information has sought 833 00:31:50,380 --> 00:31:51,969 enlightenment. I think we don't need to 834 00:31:51,970 --> 00:31:54,279 sound. What I really like is how they use 835 00:31:54,280 --> 00:31:55,539 the X in the name. 836 00:31:55,540 --> 00:31:57,249 Now it's yeah, it's appearing. 837 00:31:57,250 --> 00:31:59,439 And now just wait a few 838 00:31:59,440 --> 00:32:00,440 seconds. 839 00:32:03,060 --> 00:32:04,380 It's the slow targeted. 840 00:32:07,620 --> 00:32:09,689 Oh, my God, XM is a 841 00:32:09,690 --> 00:32:11,879 kind of old school data broker. 842 00:32:11,880 --> 00:32:14,339 They started 40 years ago 843 00:32:14,340 --> 00:32:16,529 or longer with sending personalized 844 00:32:16,530 --> 00:32:18,659 letters for the Democratic Party 845 00:32:18,660 --> 00:32:20,640 in the US during elections 846 00:32:21,840 --> 00:32:23,909 later. They also sold voter profiles to 847 00:32:23,910 --> 00:32:26,099 the Republicans and 848 00:32:26,100 --> 00:32:28,259 they became a consumer data change 849 00:32:28,260 --> 00:32:30,629 for all business fields since 850 00:32:30,630 --> 00:32:31,889 2014. 851 00:32:31,890 --> 00:32:34,049 They've also got data partnerships with 852 00:32:34,050 --> 00:32:36,150 Kouji, Facebook and Twitter 853 00:32:37,220 --> 00:32:38,699 and other large data brokers. 854 00:32:38,700 --> 00:32:41,429 LexisNexis Risk Solutions, 855 00:32:41,430 --> 00:32:43,679 also a very nice name. 856 00:32:43,680 --> 00:32:45,719 Coming up, more from the risk side of 857 00:32:45,720 --> 00:32:46,829 personal data business. 858 00:32:46,830 --> 00:32:48,659 They've got a similar amount of profiles 859 00:32:48,660 --> 00:32:51,269 on consumers and the really impressive 860 00:32:51,270 --> 00:32:53,609 range of offers on the website. 861 00:32:53,610 --> 00:32:55,829 They're not only selling data on problem 862 00:32:55,830 --> 00:32:58,259 renters, they also 863 00:32:58,260 --> 00:33:00,509 offer their employment 864 00:33:00,510 --> 00:33:02,369 screening enterprise edition. 865 00:33:03,720 --> 00:33:06,029 And they've got offers for health care 866 00:33:06,030 --> 00:33:08,609 here. They say social network analytics 867 00:33:08,610 --> 00:33:11,219 reveal hidden relationships 868 00:33:11,220 --> 00:33:12,299 and their rates right. 869 00:33:12,300 --> 00:33:14,519 Somewhere on the website, we help 870 00:33:14,520 --> 00:33:16,529 predict the likelihood that the consumer 871 00:33:16,530 --> 00:33:18,569 will become delinquent in the next 18 872 00:33:18,570 --> 00:33:19,570 months. 873 00:33:20,850 --> 00:33:22,979 And of course, they also have offers for 874 00:33:22,980 --> 00:33:24,449 governmental agencies and law 875 00:33:24,450 --> 00:33:25,439 enforcement. 876 00:33:25,440 --> 00:33:28,079 Look at this nice black helicopter 877 00:33:28,080 --> 00:33:29,190 or whatever this is. 878 00:33:30,570 --> 00:33:32,909 But at the same time, they also provide 879 00:33:32,910 --> 00:33:34,949 some marketing solutions. 880 00:33:34,950 --> 00:33:37,109 And this is a general trend, the 881 00:33:37,110 --> 00:33:38,819 same data, the same analytics. 882 00:33:38,820 --> 00:33:40,949 Technology is more and more used in 883 00:33:40,950 --> 00:33:43,379 completely different contexts, 884 00:33:43,380 --> 00:33:45,659 from marketing to banking, insurance 885 00:33:45,660 --> 00:33:47,699 and even law enforcement. 886 00:33:47,700 --> 00:33:50,009 Take Pelletiere, this Silicon 887 00:33:50,010 --> 00:33:51,989 Valley data mining company founded by 888 00:33:51,990 --> 00:33:54,899 Peter Thiel, the first Facebook investor, 889 00:33:54,900 --> 00:33:56,969 co-founder of PayPal and currently a 890 00:33:56,970 --> 00:33:58,979 member of Donald Trump's Trump's 891 00:33:58,980 --> 00:34:00,149 Transition Team, 892 00:34:01,290 --> 00:34:02,789 Pinetti. It provides products for 893 00:34:02,790 --> 00:34:04,889 companies in the fields of health care 894 00:34:04,890 --> 00:34:07,169 insurance finance 895 00:34:07,170 --> 00:34:09,388 like kind of big data analytics. 896 00:34:09,389 --> 00:34:11,519 The software is based on people's 897 00:34:11,520 --> 00:34:13,799 fraud detection algorithms, and 898 00:34:13,800 --> 00:34:15,749 they've got partnerships with the German 899 00:34:15,750 --> 00:34:18,749 business software provider SAP, 900 00:34:18,750 --> 00:34:20,819 the US Department of Defense and with 901 00:34:20,820 --> 00:34:23,399 the CIA, or 902 00:34:23,400 --> 00:34:25,619 Take FCL Group, a company 903 00:34:25,620 --> 00:34:27,689 that sees itself as working on the 904 00:34:27,690 --> 00:34:31,109 forefront of behavioral change. 905 00:34:31,110 --> 00:34:33,928 They provide data driven marketing 906 00:34:33,929 --> 00:34:36,209 for commercial purposes, 907 00:34:36,210 --> 00:34:38,459 but also information operations for 908 00:34:38,460 --> 00:34:40,619 defense and intelligence. 909 00:34:40,620 --> 00:34:43,079 And at the same time, they see themselves 910 00:34:43,080 --> 00:34:45,599 as a global election management 911 00:34:45,600 --> 00:34:46,749 company. 912 00:34:46,750 --> 00:34:49,169 Uh, FCL groups US 913 00:34:49,170 --> 00:34:51,809 branch is called Cambridge Analytica. 914 00:34:51,810 --> 00:34:53,729 You probably heard of it. 915 00:34:53,730 --> 00:34:55,859 Of them, they claim to have a national 916 00:34:55,860 --> 00:34:58,199 database of 220 million 917 00:34:58,200 --> 00:35:00,359 US citizens containing five 918 00:35:00,360 --> 00:35:02,519 thousand data points about every 919 00:35:02,520 --> 00:35:04,679 person, according to The 920 00:35:04,680 --> 00:35:08,159 Guardian. The company has also harvested 921 00:35:08,160 --> 00:35:10,230 data on millions of Facebook users. 922 00:35:11,250 --> 00:35:13,379 But what are they doing with 923 00:35:13,380 --> 00:35:14,549 this data? 924 00:35:14,550 --> 00:35:16,259 This sort people into different 925 00:35:16,260 --> 00:35:18,389 categories, for example, alongside 926 00:35:18,390 --> 00:35:20,579 the political views on issues like pro 927 00:35:20,580 --> 00:35:23,019 life, environment, gun rights, 928 00:35:23,020 --> 00:35:25,619 national security or immigration 929 00:35:25,620 --> 00:35:28,199 in order to target and address 930 00:35:28,200 --> 00:35:30,449 people differently. 931 00:35:30,450 --> 00:35:32,579 This way, they can communicate different 932 00:35:32,580 --> 00:35:34,769 messages to different small 933 00:35:34,770 --> 00:35:37,139 groups of people according to their 934 00:35:37,140 --> 00:35:39,239 political views, and based on vast 935 00:35:39,240 --> 00:35:41,339 amounts of personal information 936 00:35:41,340 --> 00:35:43,050 on an individual level. 937 00:35:44,220 --> 00:35:46,739 And by the way, uh, Cambridge Analytica 938 00:35:46,740 --> 00:35:49,559 also contributed to the 939 00:35:49,560 --> 00:35:51,899 Brexit campaign and also to Donald 940 00:35:51,900 --> 00:35:53,699 Trump's election campaign. 941 00:35:53,700 --> 00:35:55,859 And Peter Thiel 942 00:35:55,860 --> 00:35:58,949 is on the board of Cambridge Analytica. 943 00:35:58,950 --> 00:36:00,239 OK, voter targeting. 944 00:36:00,240 --> 00:36:02,819 There are many other companies like that 945 00:36:02,820 --> 00:36:05,009 also for the Democratic Party. 946 00:36:05,010 --> 00:36:07,229 And there are many other fields where 947 00:36:07,230 --> 00:36:09,449 personal data analytics is being 948 00:36:09,450 --> 00:36:11,939 used. For example, 949 00:36:11,940 --> 00:36:14,159 cheesiness, health, healthcare, 950 00:36:14,160 --> 00:36:16,439 a US company calculating individual 951 00:36:16,440 --> 00:36:18,569 health risks from a wide range of 952 00:36:18,570 --> 00:36:20,519 patient data, including from canonic 953 00:36:20,520 --> 00:36:23,129 genomics, medical records, 954 00:36:23,130 --> 00:36:25,769 lab data, mobile devices 955 00:36:25,770 --> 00:36:28,019 and also consumer behavior. 956 00:36:29,310 --> 00:36:32,429 They offer to identify people likely 957 00:36:32,430 --> 00:36:34,649 not to participate in 958 00:36:34,650 --> 00:36:36,749 interventions to 959 00:36:36,750 --> 00:36:38,909 predict progression, progression 960 00:36:38,910 --> 00:36:41,009 of illnesses and intervention 961 00:36:41,010 --> 00:36:43,259 outcomes, and to rank 962 00:36:43,260 --> 00:36:45,419 patients by how much 963 00:36:45,420 --> 00:36:48,089 return of investment the insurer 964 00:36:48,090 --> 00:36:49,619 can expect. 965 00:36:49,620 --> 00:36:51,779 If it targets patients with 966 00:36:51,780 --> 00:36:54,149 particular interventions, 967 00:36:54,150 --> 00:36:56,219 it may be to just exclude 968 00:36:56,220 --> 00:36:57,479 the hopeless cases. 969 00:36:58,590 --> 00:36:59,590 I'm sure not 970 00:37:01,020 --> 00:37:02,639 OK with health care being. 971 00:37:02,640 --> 00:37:04,199 Insurance, election campaigns, law 972 00:37:04,200 --> 00:37:06,359 enforcement, and 973 00:37:06,360 --> 00:37:08,609 not every field I mentioned is already 974 00:37:08,610 --> 00:37:10,799 completely connected to the consumer data 975 00:37:10,800 --> 00:37:12,899 ecosystem, but they're working 976 00:37:12,900 --> 00:37:13,900 on it. 977 00:37:15,660 --> 00:37:17,759 What if our data would 978 00:37:17,760 --> 00:37:20,310 be only used for marketing? 979 00:37:22,260 --> 00:37:24,389 Based on my research, marketing has has 980 00:37:24,390 --> 00:37:27,059 been it still is the major driver 981 00:37:27,060 --> 00:37:30,149 for pervasive corporate surveillance. 982 00:37:30,150 --> 00:37:32,639 In 2007, Apple introduced 983 00:37:32,640 --> 00:37:35,009 a smartphone and Facebook app, just 984 00:37:35,010 --> 00:37:37,199 30 million users. 985 00:37:37,200 --> 00:37:39,899 Also in 2007, online advertisers 986 00:37:39,900 --> 00:37:42,179 started to use individual level 987 00:37:42,180 --> 00:37:44,309 data to profile and target 988 00:37:44,310 --> 00:37:45,310 users. 989 00:37:45,810 --> 00:37:47,909 Four years later, it was around 990 00:37:47,910 --> 00:37:49,979 100 relevant companies in the 991 00:37:49,980 --> 00:37:52,349 field of so-called marketing technology, 992 00:37:52,350 --> 00:37:54,449 or ad tech, which 993 00:37:54,450 --> 00:37:56,339 are largely, largely based on personal 994 00:37:56,340 --> 00:37:59,399 data collection and profiling. 995 00:37:59,400 --> 00:38:02,429 And 2012, it was already 350 996 00:38:02,430 --> 00:38:04,739 companies, then 997 00:38:04,740 --> 00:38:08,069 thousands, then 2000. 998 00:38:08,070 --> 00:38:10,619 Now in 2016, we've got nearly 999 00:38:10,620 --> 00:38:12,779 4000 relevant companies in 1000 00:38:12,780 --> 00:38:14,549 marketing technologies. 1001 00:38:14,550 --> 00:38:16,889 The logos are quite small 1002 00:38:16,890 --> 00:38:17,890 here. 1003 00:38:18,300 --> 00:38:21,479 So today, less than 10 years after 2007, 1004 00:38:21,480 --> 00:38:23,309 we've got thousands of online platforms 1005 00:38:23,310 --> 00:38:25,469 etcs, app developers, analytics 1006 00:38:25,470 --> 00:38:27,329 companies, data brokers and many other 1007 00:38:27,330 --> 00:38:29,759 kinds of companies which are constantly 1008 00:38:29,760 --> 00:38:32,519 tracking, profiling, categorizing, 1009 00:38:32,520 --> 00:38:34,719 rating and scoring as 1010 00:38:34,720 --> 00:38:35,699 a time. 1011 00:38:35,700 --> 00:38:38,369 Have a look at TellApart, a so-called 1012 00:38:38,370 --> 00:38:41,249 predictive marketing platform. 1013 00:38:41,250 --> 00:38:43,769 Their slogan is Turn Our Silence 1014 00:38:43,770 --> 00:38:44,879 into your sales. 1015 00:38:47,510 --> 00:38:49,129 They claim to provide a so-called 1016 00:38:49,130 --> 00:38:51,679 TellApart identity network, 1017 00:38:51,680 --> 00:38:53,809 which incorporates massive amounts 1018 00:38:53,810 --> 00:38:55,969 of data from both online and offline 1019 00:38:55,970 --> 00:38:58,189 sources to create a TellApart 1020 00:38:58,190 --> 00:39:00,469 identity, Keat 1021 00:39:00,470 --> 00:39:02,509 for each and every shopper, and then they 1022 00:39:02,510 --> 00:39:04,999 calculate a customer value 1023 00:39:05,000 --> 00:39:06,000 score. 1024 00:39:06,820 --> 00:39:09,579 For each shopper and product combination, 1025 00:39:09,580 --> 00:39:11,709 a company, a compilation 1026 00:39:11,710 --> 00:39:14,109 of the likelihood to purchase 1027 00:39:14,110 --> 00:39:16,449 predicted order size and customer 1028 00:39:16,450 --> 00:39:18,010 lifetime value. 1029 00:39:19,360 --> 00:39:21,519 A kind of score about 1030 00:39:21,520 --> 00:39:24,549 how profitable or valuable 1031 00:39:24,550 --> 00:39:26,679 or not valuable a customer 1032 00:39:26,680 --> 00:39:28,759 is. As a result, 1033 00:39:28,760 --> 00:39:30,939 some customers get personalized offers 1034 00:39:30,940 --> 00:39:33,009 based on their online and even offline 1035 00:39:33,010 --> 00:39:34,269 behavior. 1036 00:39:34,270 --> 00:39:36,909 Uh, last year, TellApart 1037 00:39:36,910 --> 00:39:39,669 was acquired by Twitter for about 1038 00:39:39,670 --> 00:39:41,110 five hundred million dollar 1039 00:39:43,180 --> 00:39:45,459 personalized pricing based on real time 1040 00:39:45,460 --> 00:39:46,449 customer values. 1041 00:39:46,450 --> 00:39:48,669 Course and the like 1042 00:39:48,670 --> 00:39:51,459 could soon be everywhere. 1043 00:39:51,460 --> 00:39:53,749 Large global online shops already show 1044 00:39:53,750 --> 00:39:55,869 differently priced products for different 1045 00:39:55,870 --> 00:39:57,969 users or even the same products 1046 00:39:57,970 --> 00:40:00,489 at different prices based on people's 1047 00:40:00,490 --> 00:40:03,009 online behavior, location data 1048 00:40:03,010 --> 00:40:05,799 or the devices they use. 1049 00:40:05,800 --> 00:40:07,899 A research paper from 2012 1050 00:40:07,900 --> 00:40:10,209 already showed that personalized prices 1051 00:40:10,210 --> 00:40:12,399 differed up to one hundred and 1052 00:40:12,400 --> 00:40:14,259 sixty six percent. 1053 00:40:14,260 --> 00:40:16,629 The problem is that it's difficult, 1054 00:40:16,630 --> 00:40:19,119 if not impossible, to prove price 1055 00:40:19,120 --> 00:40:21,369 discrimination based on individual 1056 00:40:21,370 --> 00:40:23,250 attributes or user behavior. 1057 00:40:24,490 --> 00:40:26,199 For individuals, it's completely 1058 00:40:26,200 --> 00:40:27,310 nontransparent, 1059 00:40:28,450 --> 00:40:31,329 and maybe personalized pricing will soon 1060 00:40:31,330 --> 00:40:33,699 not only happen in online shops or travel 1061 00:40:33,700 --> 00:40:36,249 websites, but could appear 1062 00:40:36,250 --> 00:40:38,739 in the form of many smaller 1063 00:40:38,740 --> 00:40:40,989 or bigger differences in how consumers 1064 00:40:40,990 --> 00:40:43,959 get offers, discounts and personalized 1065 00:40:43,960 --> 00:40:46,119 communication from companies in 1066 00:40:46,120 --> 00:40:47,120 general. 1067 00:40:48,010 --> 00:40:49,719 It clearly won't be a problem for 1068 00:40:49,720 --> 00:40:51,879 somebody to miss a single at or 1069 00:40:51,880 --> 00:40:54,039 discount. Often we are even 1070 00:40:54,040 --> 00:40:56,979 happy to miss those, 1071 00:40:56,980 --> 00:40:59,109 but it could be a problem on 1072 00:40:59,110 --> 00:41:01,149 a structural level. 1073 00:41:01,150 --> 00:41:03,309 Imagine someone experiences many of 1074 00:41:03,310 --> 00:41:05,559 these small disadvantages, 1075 00:41:05,560 --> 00:41:07,659 disadvantages every day, 1076 00:41:07,660 --> 00:41:09,849 perhaps without being aware 1077 00:41:09,850 --> 00:41:11,039 of it. 1078 00:41:11,040 --> 00:41:13,419 Michael Fertik, US privacy advocates 1079 00:41:13,420 --> 00:41:15,579 said the rich see 1080 00:41:15,580 --> 00:41:17,709 already a different Internet than 1081 00:41:17,710 --> 00:41:20,709 the poor based on personalization, 1082 00:41:20,710 --> 00:41:22,839 based on digital records about their 1083 00:41:22,840 --> 00:41:25,869 lives on a much more general 1084 00:41:25,870 --> 00:41:26,870 level. 1085 00:41:27,550 --> 00:41:29,679 Personalization based on our data is now 1086 00:41:29,680 --> 00:41:31,359 nearly everywhere and often it's a good 1087 00:41:31,360 --> 00:41:32,259 thing. 1088 00:41:32,260 --> 00:41:34,329 But it's nontransparent and bears 1089 00:41:34,330 --> 00:41:36,429 the risk of reinforcing 1090 00:41:36,430 --> 00:41:38,679 discrimination or even increasing 1091 00:41:38,680 --> 00:41:39,799 it. 1092 00:41:39,800 --> 00:41:41,949 And the more invasive the data 1093 00:41:41,950 --> 00:41:43,959 sharing between companies and the 1094 00:41:43,960 --> 00:41:46,089 decontextualized usage of 1095 00:41:46,090 --> 00:41:48,399 the record information gets, 1096 00:41:48,400 --> 00:41:51,369 the more Apakan nontransparent 1097 00:41:51,370 --> 00:41:53,799 the whole system becomes. 1098 00:41:53,800 --> 00:41:56,139 A key concern for me is that the data 1099 00:41:56,140 --> 00:41:58,389 companies are increasingly 1100 00:41:58,390 --> 00:42:00,909 using unique identifiers 1101 00:42:00,910 --> 00:42:02,950 across different companies, 1102 00:42:04,870 --> 00:42:06,939 devices and contexts which they pretend 1103 00:42:06,940 --> 00:42:08,019 to be anonymous, 1104 00:42:09,310 --> 00:42:11,529 merely identifiers derived from email 1105 00:42:11,530 --> 00:42:14,439 addresses and phone numbers. 1106 00:42:14,440 --> 00:42:16,419 This way, a person can be recognized 1107 00:42:16,420 --> 00:42:18,489 again as the same person as soon 1108 00:42:18,490 --> 00:42:20,949 as he or she clicks swipe spies 1109 00:42:20,950 --> 00:42:23,889 or does some other recorded interaction 1110 00:42:23,890 --> 00:42:26,289 any time across 1111 00:42:26,290 --> 00:42:27,790 networks of data companies. 1112 00:42:28,930 --> 00:42:31,089 And in fact, those identifiers 1113 00:42:31,090 --> 00:42:33,579 are not anonymous at all. 1114 00:42:33,580 --> 00:42:34,810 Their site on 1115 00:42:36,640 --> 00:42:38,949 Facebook also uses those 1116 00:42:40,060 --> 00:42:42,279 identifiers and calls them 1117 00:42:42,280 --> 00:42:43,929 the identified, I think. 1118 00:42:45,400 --> 00:42:47,529 In addition, we've got the identifiers 1119 00:42:47,530 --> 00:42:49,659 from Google, Apple and Microsoft on 1120 00:42:49,660 --> 00:42:51,999 the mobile mobile devices 1121 00:42:52,000 --> 00:42:54,099 and so on, which 1122 00:42:54,100 --> 00:42:56,309 are more and more replacing the hardware 1123 00:42:56,310 --> 00:42:57,429 device. It is. 1124 00:42:57,430 --> 00:42:59,709 And several companies have introduced 1125 00:42:59,710 --> 00:43:02,289 their own persistent identifiers, 1126 00:43:02,290 --> 00:43:04,959 unique identifiers for people, 1127 00:43:04,960 --> 00:43:07,029 for example, Oracle, Axiom, 1128 00:43:07,030 --> 00:43:09,249 VeriSign, Experian and 1129 00:43:09,250 --> 00:43:11,379 many others to 1130 00:43:11,380 --> 00:43:13,809 match and link names, email 1131 00:43:13,810 --> 00:43:15,879 addresses and postal addresses, 1132 00:43:15,880 --> 00:43:18,369 phone numbers, kookie these 1133 00:43:18,370 --> 00:43:20,529 devices, these set top box 1134 00:43:20,530 --> 00:43:22,749 I.D. and many more 1135 00:43:22,750 --> 00:43:25,329 across the data providers 1136 00:43:25,330 --> 00:43:27,429 and clients, companies and 1137 00:43:27,430 --> 00:43:29,679 both Oracle and Axium run 1138 00:43:29,680 --> 00:43:32,879 so-called data management platforms. 1139 00:43:32,880 --> 00:43:35,139 A data management platform is a kind 1140 00:43:35,140 --> 00:43:38,319 of Real-Time online data marketplace. 1141 00:43:38,320 --> 00:43:40,479 It acts as a central hub 1142 00:43:40,480 --> 00:43:42,909 used to aggregate, integrate, 1143 00:43:42,910 --> 00:43:45,159 manage and deploy disparate 1144 00:43:45,160 --> 00:43:46,419 sources of data. 1145 00:43:46,420 --> 00:43:48,879 Is a consulting company summarized 1146 00:43:48,880 --> 00:43:50,019 it. 1147 00:43:50,020 --> 00:43:52,149 A DMP offers 1148 00:43:52,150 --> 00:43:54,609 clients to important data, 1149 00:43:54,610 --> 00:43:56,799 for example, their customer 1150 00:43:56,800 --> 00:43:59,079 relation, customer database, 1151 00:43:59,080 --> 00:44:01,239 email addresses, purchases and so 1152 00:44:01,240 --> 00:44:03,549 on, then match 1153 00:44:03,550 --> 00:44:05,269 customer IDs. 1154 00:44:05,270 --> 00:44:06,279 Uh. 1155 00:44:06,280 --> 00:44:08,949 New data by putting text and the websites 1156 00:44:08,950 --> 00:44:11,319 in their e-mail, newsletters 1157 00:44:11,320 --> 00:44:13,599 and so on, they 1158 00:44:13,600 --> 00:44:16,659 offer access to other data vendors, 1159 00:44:16,660 --> 00:44:19,479 sometimes to many other data sources, 1160 00:44:19,480 --> 00:44:22,119 and they offer to analyze and categorize 1161 00:44:22,120 --> 00:44:23,229 people. 1162 00:44:23,230 --> 00:44:25,899 The marketing guys call these 1163 00:44:25,900 --> 00:44:28,629 segments or audiences. 1164 00:44:28,630 --> 00:44:30,580 I would say it's categorizing people. 1165 00:44:33,630 --> 00:44:36,059 They also offered to find similar people, 1166 00:44:36,060 --> 00:44:39,239 then the people that 1167 00:44:39,240 --> 00:44:41,519 the customers of a company so called look 1168 00:44:41,520 --> 00:44:44,129 alikes and 1169 00:44:44,130 --> 00:44:46,170 they offered to send instructions 1170 00:44:47,460 --> 00:44:49,859 who to target and which device, 1171 00:44:49,860 --> 00:44:52,949 how to personalize content or messages 1172 00:44:52,950 --> 00:44:56,009 or ads or websites or whatever. 1173 00:44:56,010 --> 00:44:57,010 And sometimes 1174 00:44:58,440 --> 00:45:00,569 apps also allow to categorize and 1175 00:45:00,570 --> 00:45:02,819 enrich the company's own customer 1176 00:45:02,820 --> 00:45:04,139 databases. 1177 00:45:04,140 --> 00:45:05,939 This way, companies can sort the 1178 00:45:05,940 --> 00:45:08,219 customers into valuable and non 1179 00:45:08,220 --> 00:45:10,349 valuable customers and 1180 00:45:10,350 --> 00:45:12,419 into risky and non risky 1181 00:45:12,420 --> 00:45:14,130 people and so on. 1182 00:45:15,180 --> 00:45:17,549 Examples of companies running the piece 1183 00:45:17,550 --> 00:45:19,709 include Acxiom and Oracle, but also 1184 00:45:19,710 --> 00:45:20,879 Salesforce. 1185 00:45:20,880 --> 00:45:22,769 Maybe somebody heard of it. 1186 00:45:22,770 --> 00:45:24,300 A CRM vendor 1187 00:45:25,740 --> 00:45:28,259 and Adobe also runs the data management 1188 00:45:28,260 --> 00:45:29,729 platform. 1189 00:45:29,730 --> 00:45:31,829 Another example is Lotame 1190 00:45:31,830 --> 00:45:34,139 Me, which provides access 1191 00:45:34,140 --> 00:45:36,569 to, um, 1192 00:45:36,570 --> 00:45:38,699 three billion cookies and two billion 1193 00:45:38,700 --> 00:45:40,679 mobile device I.D. 1194 00:45:40,680 --> 00:45:42,779 And by the way, they also tricked me 1195 00:45:42,780 --> 00:45:45,119 when I was visiting those five websites 1196 00:45:45,120 --> 00:45:46,349 I showed at the beginning of my 1197 00:45:46,350 --> 00:45:47,350 presentation. 1198 00:45:49,200 --> 00:45:50,939 It's a very interesting domain name, 1199 00:45:50,940 --> 00:45:53,400 crowd control, dot net, 1200 00:45:54,900 --> 00:45:55,979 wonderful. 1201 00:45:55,980 --> 00:45:58,469 But let me explain it on its website. 1202 00:45:58,470 --> 00:46:00,509 It's your data you have to read to 1203 00:46:00,510 --> 00:46:02,729 control it, share it and use it how 1204 00:46:02,730 --> 00:46:03,959 you see fit. 1205 00:46:03,960 --> 00:46:05,699 Sounds nice, right? 1206 00:46:06,750 --> 00:46:08,669 But of course, they don't speak to 1207 00:46:08,670 --> 00:46:09,899 consumers. 1208 00:46:09,900 --> 00:46:11,820 They speak to their corporate clients. 1209 00:46:13,710 --> 00:46:15,599 I really love that. 1210 00:46:15,600 --> 00:46:17,789 I think it's a good representation 1211 00:46:17,790 --> 00:46:19,679 of today's information technology 1212 00:46:19,680 --> 00:46:20,969 landscape. 1213 00:46:20,970 --> 00:46:23,039 While we as individuals became more 1214 00:46:23,040 --> 00:46:25,619 and more transparent, practices 1215 00:46:25,620 --> 00:46:27,779 of corporate surveillance remain 1216 00:46:27,780 --> 00:46:29,609 largely obscure. 1217 00:46:29,610 --> 00:46:31,379 Today, each of our interactions is 1218 00:46:31,380 --> 00:46:33,329 monitored allies and assessed by a 1219 00:46:33,330 --> 00:46:34,979 network of machines and software 1220 00:46:34,980 --> 00:46:37,379 algorithms operated by companies 1221 00:46:37,380 --> 00:46:39,599 we rarely have ever heard of. 1222 00:46:39,600 --> 00:46:42,209 It's not balanced at all 1223 00:46:42,210 --> 00:46:44,429 without our knowledge and often without 1224 00:46:44,430 --> 00:46:46,619 our consent, our interests, 1225 00:46:46,620 --> 00:46:49,439 weaknesses, illnesses, successes, 1226 00:46:49,440 --> 00:46:51,899 secrets and purchasing power 1227 00:46:51,900 --> 00:46:52,979 as a weight. 1228 00:46:52,980 --> 00:46:55,079 And companies increasingly use 1229 00:46:55,080 --> 00:46:57,299 the collected data about our 1230 00:46:57,300 --> 00:47:00,179 lives to make decisions and us 1231 00:47:00,180 --> 00:47:03,599 from which ads and products we see 1232 00:47:03,600 --> 00:47:05,789 which discounts in prices we get, 1233 00:47:05,790 --> 00:47:08,209 how long we have to wait when calling 1234 00:47:08,210 --> 00:47:10,349 on the phone hotline, which payment 1235 00:47:10,350 --> 00:47:13,019 methods we get to massive decisions 1236 00:47:13,020 --> 00:47:14,909 in the fields of personal finance, 1237 00:47:14,910 --> 00:47:17,879 insurance, housing, employment 1238 00:47:17,880 --> 00:47:19,319 and health care. 1239 00:47:19,320 --> 00:47:21,959 While we as individuals and human beings 1240 00:47:21,960 --> 00:47:24,179 and as a society mostly don't even 1241 00:47:24,180 --> 00:47:26,399 know who is tracking us, 1242 00:47:26,400 --> 00:47:28,529 how our data is being used, and how this 1243 00:47:28,530 --> 00:47:30,539 could impact our future lives. 1244 00:47:30,540 --> 00:47:32,759 The companies took control 1245 00:47:32,760 --> 00:47:34,259 of our data. 1246 00:47:34,260 --> 00:47:35,960 So what has to be done? 1247 00:47:37,840 --> 00:47:39,579 One option could look like this. 1248 00:47:49,050 --> 00:47:51,119 No, I love 1249 00:47:51,120 --> 00:47:52,289 information technology 1250 00:47:54,540 --> 00:47:55,949 during the last few years, I tried to 1251 00:47:55,950 --> 00:47:57,449 find out how to beat this challenge. 1252 00:47:57,450 --> 00:47:58,890 My first try looked at that. 1253 00:48:00,090 --> 00:48:02,519 I was asking, worried about your privacy, 1254 00:48:02,520 --> 00:48:04,829 forget it, turned the tables 1255 00:48:04,830 --> 00:48:06,749 and find out all the details about your 1256 00:48:06,750 --> 00:48:08,279 friends, your neighbors and the rest of 1257 00:48:08,280 --> 00:48:09,280 the world. 1258 00:48:10,350 --> 00:48:12,119 Together with a small team from Vienna, I 1259 00:48:12,120 --> 00:48:14,639 created a data dealer, an online game, 1260 00:48:14,640 --> 00:48:17,339 browser game and even a Facebook game 1261 00:48:17,340 --> 00:48:18,839 about collecting and selling personal 1262 00:48:18,840 --> 00:48:20,309 information. You can still play it 1263 00:48:20,310 --> 00:48:22,499 online, and 1264 00:48:22,500 --> 00:48:25,559 I promise that we won't sell your data. 1265 00:48:25,560 --> 00:48:26,560 Trust me. 1266 00:48:36,230 --> 00:48:38,119 Since then, I did lots of research about 1267 00:48:38,120 --> 00:48:40,399 today's personal data ecosystem, I wrote 1268 00:48:40,400 --> 00:48:42,349 newspaper articles, contributed to 1269 00:48:42,350 --> 00:48:44,479 documentaries about this show tracking, 1270 00:48:44,480 --> 00:48:46,489 and recently published an extensive 1271 00:48:46,490 --> 00:48:48,169 report about these issues. 1272 00:48:48,170 --> 00:48:50,519 It's called Networks of Control. 1273 00:48:50,520 --> 00:48:52,699 A report on corporate surveillance this 1274 00:48:52,700 --> 00:48:55,399 year, tracking big data 1275 00:48:55,400 --> 00:48:56,689 and privacy. 1276 00:48:56,690 --> 00:48:58,759 You can download a PDF version 1277 00:48:58,760 --> 00:48:59,760 for free, 1278 00:49:00,890 --> 00:49:02,839 but also by it is a printed book. 1279 00:49:04,190 --> 00:49:06,499 I wrote it together with Sara Speakman, 1280 00:49:06,500 --> 00:49:08,389 privacy researcher and university 1281 00:49:08,390 --> 00:49:11,119 professor, also from Vienna. 1282 00:49:11,120 --> 00:49:12,949 In our report, we basically tried to 1283 00:49:12,950 --> 00:49:15,049 explain how today's companies are 1284 00:49:15,050 --> 00:49:16,609 collecting, analyzing and selling 1285 00:49:16,610 --> 00:49:18,959 information about our lives 1286 00:49:18,960 --> 00:49:21,289 and 160 pages with 1287 00:49:21,290 --> 00:49:23,029 900 references. 1288 00:49:23,030 --> 00:49:24,979 I've been working on that for a long time 1289 00:49:24,980 --> 00:49:27,229 without any budget. 1290 00:49:27,230 --> 00:49:29,359 So if you want to help, please read 1291 00:49:29,360 --> 00:49:31,589 it, spread it 1292 00:49:31,590 --> 00:49:32,929 or even read about it. 1293 00:49:34,190 --> 00:49:36,289 It was published in October, 1294 00:49:36,290 --> 00:49:38,479 but I guess it could still need some more 1295 00:49:38,480 --> 00:49:39,480 attention. 1296 00:49:41,330 --> 00:49:43,459 Yeah, and of course, I'll continue with 1297 00:49:43,460 --> 00:49:45,619 my research on these issues and 1298 00:49:45,620 --> 00:49:47,869 will soon start with another project 1299 00:49:47,870 --> 00:49:49,939 working on a prototype for an 1300 00:49:49,940 --> 00:49:52,519 online research platform called Tracking 1301 00:49:52,520 --> 00:49:53,929 the Trackers. 1302 00:49:53,930 --> 00:49:56,239 It should become a comprehensive online 1303 00:49:56,240 --> 00:49:58,369 knowledge base about the topics I've been 1304 00:49:58,370 --> 00:50:00,199 talking about today. 1305 00:50:00,200 --> 00:50:02,179 The main target groups for this will 1306 00:50:02,180 --> 00:50:04,729 include academics, journalists, activists 1307 00:50:04,730 --> 00:50:06,019 and policy makers. 1308 00:50:06,020 --> 00:50:08,359 So it's planned to be a research tool 1309 00:50:08,360 --> 00:50:10,610 for kind of expert stakeholders 1310 00:50:12,020 --> 00:50:13,219 in the medium term. 1311 00:50:13,220 --> 00:50:16,339 It could also evolve into a collaborative 1312 00:50:16,340 --> 00:50:17,839 community platform. 1313 00:50:17,840 --> 00:50:20,479 We'll see if you're interested in this 1314 00:50:20,480 --> 00:50:21,499 ad anyhow. 1315 00:50:21,500 --> 00:50:23,929 If you're interested in collaboration, 1316 00:50:23,930 --> 00:50:26,809 please say hello to me after my talk, 1317 00:50:26,810 --> 00:50:29,359 OK? This is what I will do. 1318 00:50:29,360 --> 00:50:31,009 But I still didn't really answer the 1319 00:50:31,010 --> 00:50:31,999 question. 1320 00:50:32,000 --> 00:50:33,230 What has to be done 1321 00:50:34,500 --> 00:50:36,679 in general? I think what we should 1322 00:50:36,680 --> 00:50:39,109 not do is things like blaming 1323 00:50:39,110 --> 00:50:41,929 people because they're using Facebook 1324 00:50:41,930 --> 00:50:43,849 and telling them they're guilty and it's 1325 00:50:43,850 --> 00:50:45,949 their own fault that their personal 1326 00:50:45,950 --> 00:50:48,729 information is being abused and so on. 1327 00:50:48,730 --> 00:50:50,119 Of course, everybody should use 1328 00:50:50,120 --> 00:50:52,429 alternatives to the dominant services 1329 00:50:52,430 --> 00:50:54,559 and apps and browser extensions 1330 00:50:54,560 --> 00:50:56,539 to avoid some of the most invasive 1331 00:50:56,540 --> 00:50:58,369 tracking and so on. 1332 00:50:58,370 --> 00:51:00,589 But I think there is no easy way to 1333 00:51:00,590 --> 00:51:02,809 completely opt out of today's 1334 00:51:02,810 --> 00:51:05,149 surveillance economy at an individual 1335 00:51:05,150 --> 00:51:07,339 level without opting 1336 00:51:07,340 --> 00:51:09,959 out of too much from modern life. 1337 00:51:09,960 --> 00:51:12,499 For example, if a non-technical person 1338 00:51:12,500 --> 00:51:14,629 wants to use a normal state of the 1339 00:51:14,630 --> 00:51:17,089 art phone today, this person 1340 00:51:17,090 --> 00:51:19,669 has no other choice than choosing between 1341 00:51:19,670 --> 00:51:21,709 using it with a Google, Apple or 1342 00:51:21,710 --> 00:51:23,989 Microsoft account is a disaster. 1343 00:51:25,970 --> 00:51:27,529 It's not an individual problem. 1344 00:51:27,530 --> 00:51:30,079 We have to solve this on a societal 1345 00:51:30,080 --> 00:51:31,999 level, I think. 1346 00:51:32,000 --> 00:51:33,409 And that's why I'd like to present a 1347 00:51:33,410 --> 00:51:36,019 quick summary of basic policy 1348 00:51:36,020 --> 00:51:38,119 recommendations which resulted from 1349 00:51:38,120 --> 00:51:39,120 my research. 1350 00:51:40,940 --> 00:51:42,799 I think the most urgent challenge is to 1351 00:51:42,800 --> 00:51:44,899 make corporate data collection and 1352 00:51:44,900 --> 00:51:47,599 utilization more transparent. 1353 00:51:47,600 --> 00:51:50,029 This could happen by supporting research, 1354 00:51:50,030 --> 00:51:52,339 also by developing developing technical 1355 00:51:52,340 --> 00:51:54,409 tools to examine 1356 00:51:54,410 --> 00:51:57,049 the black boxes from the outside. 1357 00:51:57,050 --> 00:51:59,659 But, of course, also by regulation, 1358 00:51:59,660 --> 00:52:02,329 which quickly leads me to 1359 00:52:02,330 --> 00:52:04,579 the never ending story of the European 1360 00:52:04,580 --> 00:52:05,959 data protection reform. 1361 00:52:05,960 --> 00:52:08,749 I hope that European 1362 00:52:08,750 --> 00:52:11,119 data protection regulation to GDP 1363 00:52:11,120 --> 00:52:13,489 are and the EU Privacy Directive 1364 00:52:13,490 --> 00:52:15,919 will make things at least a bit better 1365 00:52:15,920 --> 00:52:17,989 from 2008, 1366 00:52:17,990 --> 00:52:19,459 if not much better. 1367 00:52:19,460 --> 00:52:21,529 We'll have to carefully watch how 1368 00:52:21,530 --> 00:52:23,839 it will work on a practical level 1369 00:52:23,840 --> 00:52:25,999 and if necessary, we'll 1370 00:52:26,000 --> 00:52:27,139 have to further update it. 1371 00:52:28,790 --> 00:52:30,589 In addition, I think that also other 1372 00:52:30,590 --> 00:52:32,839 fields of law, such consumers such 1373 00:52:32,840 --> 00:52:34,159 as consumer protection, 1374 00:52:34,160 --> 00:52:36,799 anti-discrimination and also competition 1375 00:52:36,800 --> 00:52:39,169 law could help to rebalance 1376 00:52:39,170 --> 00:52:41,599 the current information technology 1377 00:52:41,600 --> 00:52:42,649 landscape. 1378 00:52:42,650 --> 00:52:44,659 But even if we get the best regulation, I 1379 00:52:44,660 --> 00:52:46,789 am afraid that this won't 1380 00:52:46,790 --> 00:52:48,589 be enough in the moment. 1381 00:52:48,590 --> 00:52:50,809 Companies are not only in control of 1382 00:52:50,810 --> 00:52:52,969 our personal data, they're even trying to 1383 00:52:52,970 --> 00:52:55,429 shape our future information 1384 00:52:55,430 --> 00:52:58,219 society at all. 1385 00:52:58,220 --> 00:53:00,109 And during the last 10 years, they were 1386 00:53:00,110 --> 00:53:02,299 successful without much 1387 00:53:02,300 --> 00:53:04,879 democratic debate or discussion 1388 00:53:04,880 --> 00:53:05,989 or whatever. 1389 00:53:07,260 --> 00:53:08,819 They've got billions and billions of 1390 00:53:08,820 --> 00:53:10,739 financial resources and they're moving 1391 00:53:10,740 --> 00:53:12,270 ahead very fast. 1392 00:53:13,290 --> 00:53:15,209 I think we need much more support of 1393 00:53:15,210 --> 00:53:18,269 decentralized privacy where technology 1394 00:53:18,270 --> 00:53:20,369 or maybe even a completely 1395 00:53:20,370 --> 00:53:22,829 new industrial policy and billions 1396 00:53:22,830 --> 00:53:25,409 for open source components and frameworks 1397 00:53:25,410 --> 00:53:27,629 which help creating a different kind of 1398 00:53:27,630 --> 00:53:30,089 innovation, which is respecting 1399 00:53:30,090 --> 00:53:32,159 our privacy, for example, and the 1400 00:53:32,160 --> 00:53:34,319 European Union level. 1401 00:53:34,320 --> 00:53:36,719 And not least there is already 1402 00:53:36,720 --> 00:53:38,789 a large group of people committed 1403 00:53:38,790 --> 00:53:40,889 for this kind of a different Internet, 1404 00:53:40,890 --> 00:53:43,469 which is not dominated by corporations 1405 00:53:43,470 --> 00:53:46,229 building centralized services, 1406 00:53:46,230 --> 00:53:47,300 for example, here, 1407 00:53:48,450 --> 00:53:49,679 the Congress. 1408 00:53:49,680 --> 00:53:51,899 This is why I think it's also crucial 1409 00:53:51,900 --> 00:53:54,989 to make digital civil society 1410 00:53:54,990 --> 00:53:56,549 much stronger. 1411 00:53:56,550 --> 00:53:58,379 I mean, many organizations and 1412 00:53:58,380 --> 00:54:00,629 individuals are doing amazing 1413 00:54:00,630 --> 00:54:03,419 work, fighting hard to get some 50000 1414 00:54:03,420 --> 00:54:04,420 years. 1415 00:54:05,740 --> 00:54:08,229 That's not clever from a society 1416 00:54:08,230 --> 00:54:10,329 point of view, I would say. 1417 00:54:10,330 --> 00:54:12,699 And yes, we need much better and 1418 00:54:12,700 --> 00:54:14,799 much better level of digital 1419 00:54:14,800 --> 00:54:16,089 literacy. 1420 00:54:16,090 --> 00:54:18,279 We need well informed citizens 1421 00:54:18,280 --> 00:54:20,589 for a Democratic Future Information 1422 00:54:20,590 --> 00:54:23,229 Society. And I'm not talking about 1423 00:54:23,230 --> 00:54:25,299 just knowing how to use Microsoft 1424 00:54:25,300 --> 00:54:26,300 Word. 1425 00:54:27,710 --> 00:54:29,629 It's about better knowledge of what the 1426 00:54:29,630 --> 00:54:31,699 digital age really means for 1427 00:54:31,700 --> 00:54:33,829 us as individuals and 1428 00:54:33,830 --> 00:54:34,830 as a society 1429 00:54:35,900 --> 00:54:36,979 beside of this. 1430 00:54:36,980 --> 00:54:39,679 I think the worst thing that could happen 1431 00:54:39,680 --> 00:54:42,289 would be if we get desperate or even 1432 00:54:42,290 --> 00:54:44,449 cynical, like the 1433 00:54:44,450 --> 00:54:46,459 NSA is collecting every piece of data 1434 00:54:46,460 --> 00:54:48,649 anyway and everybody's using Google. 1435 00:54:48,650 --> 00:54:50,749 So fuck it, 1436 00:54:50,750 --> 00:54:51,750 why should we care? 1437 00:54:52,850 --> 00:54:54,799 Actually, quite the opposite. 1438 00:54:54,800 --> 00:54:56,989 And that's why I'd like to finish my 1439 00:54:56,990 --> 00:54:59,209 talk like always 1440 00:54:59,210 --> 00:55:01,309 with a major recommendation 1441 00:55:01,310 --> 00:55:03,289 by a major guy. 1442 00:55:03,290 --> 00:55:07,009 Google's Eric Schmidt said in 2013, 1443 00:55:07,010 --> 00:55:08,809 you have to fight for your privacy or you 1444 00:55:08,810 --> 00:55:09,810 will lose it. 1445 00:55:10,720 --> 00:55:13,269 Friendly advice or a serious 1446 00:55:13,270 --> 00:55:14,270 threat? 1447 00:55:15,760 --> 00:55:17,349 I don't know. 1448 00:55:17,350 --> 00:55:18,350 Thanks a lot. 1449 00:55:36,650 --> 00:55:38,479 Thank you so much. We do have time for 1450 00:55:38,480 --> 00:55:40,579 questions, so if you do 1451 00:55:40,580 --> 00:55:42,679 have questions, please go to 1452 00:55:42,680 --> 00:55:45,169 the for one of the four microphones. 1453 00:55:45,170 --> 00:55:47,389 For those of you who are leaving 1454 00:55:47,390 --> 00:55:49,549 right now, the room, please do 1455 00:55:49,550 --> 00:55:51,799 it as quietly as possible so we can not 1456 00:55:51,800 --> 00:55:53,959 get interrupted or disrupted in the 1457 00:55:53,960 --> 00:55:54,960 rest of us. Talk. 1458 00:55:56,510 --> 00:55:57,529 Let's just take a moment. 1459 00:55:57,530 --> 00:56:00,349 OK, we have a question on the microphone 1460 00:56:00,350 --> 00:56:01,550 to my left. 1461 00:56:03,320 --> 00:56:05,679 So know any any kind of studies 1462 00:56:05,680 --> 00:56:08,059 made how we know how this all just 1463 00:56:08,060 --> 00:56:10,529 makes cell phones itself so good? 1464 00:56:10,530 --> 00:56:12,959 Neera, can you speak into the phone? 1465 00:56:12,960 --> 00:56:15,019 Yeah. Yeah. So do know how these 1466 00:56:15,020 --> 00:56:16,309 tracking technologies influence 1467 00:56:16,310 --> 00:56:18,769 self-censorship of how people 1468 00:56:18,770 --> 00:56:20,929 start to change their behavior 1469 00:56:20,930 --> 00:56:23,419 online and restrict 1470 00:56:23,420 --> 00:56:25,669 their communications in 1471 00:56:25,670 --> 00:56:28,129 order not to leak someone's private 1472 00:56:28,130 --> 00:56:30,619 world view to to the trackers? 1473 00:56:30,620 --> 00:56:33,469 Yes, I didn't talk about that now, 1474 00:56:33,470 --> 00:56:36,019 but this is a crucial point, 1475 00:56:36,020 --> 00:56:37,999 this kind of chilling effect. 1476 00:56:38,000 --> 00:56:40,069 When you know that you're 1477 00:56:40,070 --> 00:56:42,169 constantly being watched, 1478 00:56:42,170 --> 00:56:44,300 then you will behave differently. 1479 00:56:46,730 --> 00:56:49,099 We know that since a long time, 1480 00:56:49,100 --> 00:56:50,959 there are many studies about that. 1481 00:56:50,960 --> 00:56:53,389 And I think the crucial 1482 00:56:53,390 --> 00:56:54,390 point today 1483 00:56:56,210 --> 00:56:58,279 is that in many cases we 1484 00:56:58,280 --> 00:56:59,659 really don't know. 1485 00:56:59,660 --> 00:57:01,909 For example, we see some 1486 00:57:01,910 --> 00:57:04,219 ads and we think 1487 00:57:04,220 --> 00:57:06,799 it was that because of 1488 00:57:06,800 --> 00:57:08,599 that interaction, that website with it, 1489 00:57:08,600 --> 00:57:09,600 or was it not 1490 00:57:10,940 --> 00:57:13,459 did they check my location or 1491 00:57:13,460 --> 00:57:14,539 didn't they do it? 1492 00:57:14,540 --> 00:57:17,719 So I think this really 1493 00:57:17,720 --> 00:57:19,819 leads to an uncomfortable feeling. 1494 00:57:19,820 --> 00:57:21,919 And in general, the 1495 00:57:21,920 --> 00:57:23,629 situation is a disaster because we know 1496 00:57:23,630 --> 00:57:25,399 the governmental surveillance, which is 1497 00:57:25,400 --> 00:57:28,249 also accessing the corporate databases, 1498 00:57:28,250 --> 00:57:30,529 is omnipresent. 1499 00:57:30,530 --> 00:57:32,899 So, yeah, 1500 00:57:32,900 --> 00:57:34,219 I think this is a crucial point. 1501 00:57:34,220 --> 00:57:36,289 And I also addressed that a bit more 1502 00:57:36,290 --> 00:57:38,119 in detail in my report. 1503 00:57:39,620 --> 00:57:42,589 We have a question from the Internet. 1504 00:57:42,590 --> 00:57:43,639 Yes, we do. 1505 00:57:43,640 --> 00:57:45,979 There is the rather cynical 1506 00:57:45,980 --> 00:57:48,349 point of how do they categorize people 1507 00:57:48,350 --> 00:57:50,899 that are trying to hide their data 1508 00:57:50,900 --> 00:57:53,689 or mostly white male 1509 00:57:53,690 --> 00:57:54,920 earning above average? 1510 00:57:56,750 --> 00:57:58,340 Is that not just an actual data point? 1511 00:58:00,590 --> 00:58:01,590 Again, sorry, 1512 00:58:03,350 --> 00:58:05,809 people who are hiding 1513 00:58:05,810 --> 00:58:06,810 their data. 1514 00:58:08,220 --> 00:58:09,749 Can you repeat the question people that 1515 00:58:09,750 --> 00:58:12,119 are trying to opt out of 1516 00:58:12,120 --> 00:58:14,339 the cookies and everything? 1517 00:58:14,340 --> 00:58:16,469 Yes, I think people who don't 1518 00:58:16,470 --> 00:58:19,260 participate, sometimes 1519 00:58:20,970 --> 00:58:21,970 they are. 1520 00:58:22,830 --> 00:58:26,039 They consider this suspicious 1521 00:58:26,040 --> 00:58:27,119 from the beginning. 1522 00:58:27,120 --> 00:58:28,799 Also from the marketing, from fraud 1523 00:58:28,800 --> 00:58:30,599 detection companies and from many other 1524 00:58:30,600 --> 00:58:32,369 companies, I think this is also a 1525 00:58:32,370 --> 00:58:33,629 problem. 1526 00:58:33,630 --> 00:58:35,759 So we don't really have the choice to 1527 00:58:35,760 --> 00:58:38,159 just solve this on an individual 1528 00:58:38,160 --> 00:58:40,469 level, to just use encryption 1529 00:58:40,470 --> 00:58:42,599 and browser extensions 1530 00:58:42,600 --> 00:58:44,549 and digital self defense and so on. 1531 00:58:44,550 --> 00:58:46,319 That's why I'm talking about that. 1532 00:58:46,320 --> 00:58:48,869 We need really a societal 1533 00:58:48,870 --> 00:58:50,659 solution for this. 1534 00:58:52,110 --> 00:58:54,029 And then I take a question from the 1535 00:58:54,030 --> 00:58:56,169 microphone in the back on my left. 1536 00:58:56,170 --> 00:58:58,769 It's it's you. Yeah, I 1537 00:58:58,770 --> 00:58:59,759 am. 1538 00:58:59,760 --> 00:59:01,800 Do you have any individual tools to 1539 00:59:03,780 --> 00:59:05,909 support, like Ghostery or 1540 00:59:05,910 --> 00:59:08,009 start beads and stuff 1541 00:59:08,010 --> 00:59:10,199 like that? That will help because 1542 00:59:10,200 --> 00:59:12,419 there is a tool that tries 1543 00:59:12,420 --> 00:59:14,779 to block trackers and that 1544 00:59:14,780 --> 00:59:17,069 is an alternative search engine that 1545 00:59:17,070 --> 00:59:19,859 does a search through Google but 1546 00:59:19,860 --> 00:59:21,929 claims to hide your 1547 00:59:21,930 --> 00:59:23,219 personal data. 1548 00:59:23,220 --> 00:59:25,229 Am I the first part of the question? 1549 00:59:25,230 --> 00:59:27,119 And the second is, do you really trust 1550 00:59:27,120 --> 00:59:29,129 governments that governments could 1551 00:59:29,130 --> 00:59:31,289 actually impose rules on how 1552 00:59:31,290 --> 00:59:34,289 data is used, that what 1553 00:59:34,290 --> 00:59:36,749 governments with laws could 1554 00:59:36,750 --> 00:59:39,599 actually provide 1555 00:59:39,600 --> 00:59:41,249 a good framework? 1556 00:59:41,250 --> 00:59:43,259 Yeah, the first question, of course, I 1557 00:59:43,260 --> 00:59:45,389 would strongly support recommend to 1558 00:59:45,390 --> 00:59:48,239 use stop or search engines, 1559 00:59:48,240 --> 00:59:50,549 which are not based on individual 1560 00:59:50,550 --> 00:59:52,649 profiling and tracking, 1561 00:59:52,650 --> 00:59:54,959 of course, but it's 1562 00:59:54,960 --> 00:59:56,369 getting a bit more difficult when it 1563 00:59:56,370 --> 00:59:58,439 comes to browsing tangents like Ghostery 1564 00:59:58,440 --> 01:00:00,569 because they started 1565 01:00:00,570 --> 01:00:03,209 to to participate 1566 01:00:03,210 --> 01:00:05,339 in a tracking ecosystem 1567 01:00:05,340 --> 01:00:06,929 now itself. 1568 01:00:06,930 --> 01:00:09,149 So I'm not sure if 1569 01:00:09,150 --> 01:00:11,669 we can trust companies like 1570 01:00:11,670 --> 01:00:12,569 Ghostery. 1571 01:00:12,570 --> 01:00:14,039 I think we cannot. 1572 01:00:15,300 --> 01:00:17,999 Also Adblock Plus, 1573 01:00:18,000 --> 01:00:20,129 it's the same story, so it's really 1574 01:00:20,130 --> 01:00:21,179 difficult. 1575 01:00:21,180 --> 01:00:22,169 Currently, we have U. 1576 01:00:22,170 --> 01:00:24,479 Block Origin and privacy 1577 01:00:24,480 --> 01:00:26,519 BETCHER from the EFF, which are working 1578 01:00:26,520 --> 01:00:28,979 quite well. But yeah, 1579 01:00:28,980 --> 01:00:31,079 they are not the solution for for the 1580 01:00:31,080 --> 01:00:32,489 whole problem. 1581 01:00:32,490 --> 01:00:33,569 We should use it. 1582 01:00:33,570 --> 01:00:35,639 It's good, but they are not the solution. 1583 01:00:35,640 --> 01:00:37,739 The second thing is, I 1584 01:00:37,740 --> 01:00:38,849 don't know. 1585 01:00:38,850 --> 01:00:40,829 I know many people don't trust 1586 01:00:40,830 --> 01:00:42,869 governments and I also don't trust 1587 01:00:42,870 --> 01:00:44,249 governments. 1588 01:00:44,250 --> 01:00:46,589 But I don't think if we are talking 1589 01:00:46,590 --> 01:00:48,179 about governmental surveillance, 1590 01:00:49,260 --> 01:00:51,689 law enforcement, intelligence authorities 1591 01:00:51,690 --> 01:00:53,789 on the one hand and on the other hand, we 1592 01:00:53,790 --> 01:00:55,919 have some we still have something 1593 01:00:55,920 --> 01:00:59,399 like a democratic 1594 01:00:59,400 --> 01:01:00,960 parliamentary system 1595 01:01:02,220 --> 01:01:04,499 with a balance of power. 1596 01:01:04,500 --> 01:01:06,959 And if we have 1597 01:01:06,960 --> 01:01:09,179 there are interest groups, different 1598 01:01:09,180 --> 01:01:11,939 interest groups in in governmental 1599 01:01:11,940 --> 01:01:14,519 sectors and also on the European level. 1600 01:01:14,520 --> 01:01:16,859 So these are not the same 1601 01:01:16,860 --> 01:01:18,959 people who 1602 01:01:18,960 --> 01:01:19,960 are 1603 01:01:21,360 --> 01:01:23,729 fighting for better data protection 1604 01:01:23,730 --> 01:01:25,349 regulation than the people who are 1605 01:01:25,350 --> 01:01:28,199 running the government, the intelligence 1606 01:01:28,200 --> 01:01:30,509 and the data collection, law 1607 01:01:30,510 --> 01:01:33,149 enforcement and intelligence agencies. 1608 01:01:33,150 --> 01:01:35,789 So I think, yes, it's important 1609 01:01:35,790 --> 01:01:36,790 to use 1610 01:01:38,430 --> 01:01:40,679 the regulation and law 1611 01:01:40,680 --> 01:01:43,259 approach to address 1612 01:01:43,260 --> 01:01:44,489 all these problems. 1613 01:01:44,490 --> 01:01:46,469 We want to make it. 1614 01:01:46,470 --> 01:01:49,019 We thought, OK, 1615 01:01:49,020 --> 01:01:50,069 time is up. 1616 01:01:50,070 --> 01:01:52,079 Unfortunately for those who still have 1617 01:01:52,080 --> 01:01:54,059 questions, I'm sure you can reach Vafa 1618 01:01:54,060 --> 01:01:54,979 afterwards. Yeah. 1619 01:01:54,980 --> 01:01:55,979 Yes. 1620 01:01:55,980 --> 01:01:56,980 Thank you so much, Paul.