1 00:00:00,040 --> 00:00:03,640 We all know that icky feeling when your privacy is violated, 2 00:00:04,000 --> 00:00:07,240 people feel like their phone is spying on them, and all these 3 00:00:07,240 --> 00:00:09,600 things are happening, right? A lot of times when we talk 4 00:00:09,600 --> 00:00:11,920 about privacy, there are a lot of technologies that could 5 00:00:11,920 --> 00:00:15,880 expose us to this privacy race. We heard a lot in news data 6 00:00:15,880 --> 00:00:18,440 breaches, data leaks, people's privacy. 7 00:00:18,960 --> 00:00:21,120 Cambridge Analytica. Catherine Jarmel is the 8 00:00:21,120 --> 00:00:24,480 Principal data Scientist at ThoughtWorks and the author of 9 00:00:24,480 --> 00:00:27,320 Practical Data Privacy. If you don't know what data you 10 00:00:27,320 --> 00:00:30,880 have, if you don't know where it lives, then you kind of end up 11 00:00:30,880 --> 00:00:33,960 in a place where privacy is not possible. 12 00:00:34,280 --> 00:00:36,360 You don't know what you're going to use the data for. 13 00:00:36,560 --> 00:00:39,760 You're probably just amassing large cloud computing fees for 14 00:00:39,760 --> 00:00:42,640 no reason. Why can't we hold a higher bar 15 00:00:42,840 --> 00:00:47,040 for organizations to create secure and private by design 16 00:00:47,040 --> 00:00:49,880 services? Maybe can help explain what is 17 00:00:49,880 --> 00:00:52,600 privacy by design? Privacy by design is actually a 18 00:00:52,600 --> 00:00:55,920 quite old concept. There's seven design principles. 19 00:00:56,360 --> 00:00:59,480 You see a lot of software thinking allowing user choice, 20 00:00:59,560 --> 00:01:04,200 allowing transparency, but also by default building things that 21 00:01:04,239 --> 00:01:07,320 respect the user. And a lot of people associate 22 00:01:07,440 --> 00:01:10,920 privacy with PII, personally identifiable information. 23 00:01:11,240 --> 00:01:14,640 Do you think that privacy relates just to PII or is it 24 00:01:14,640 --> 00:01:19,200 more than that? Privacy is a lot about having 25 00:01:19,200 --> 00:01:37,000 the. Hey guys, welcome back to 26 00:01:37,000 --> 00:01:39,320 another new episode of the Technical General Podcast. 27 00:01:39,320 --> 00:01:42,720 Today we are going to cover a topic that is becoming trendy in 28 00:01:42,720 --> 00:01:44,680 most parts of the world, data privacy. 29 00:01:44,960 --> 00:01:48,360 I have Catherine German here. She is the author of a book 30 00:01:48,360 --> 00:01:52,520 titled Practical Data Privacy, Enhancing Privacy and Security 31 00:01:52,520 --> 00:01:55,200 in Data. So Catherine has a lot of 32 00:01:55,200 --> 00:01:57,960 background expertise in machine learning and data scientists. 33 00:01:57,960 --> 00:02:00,840 But today we'll try to cover the topics in more like journalist 34 00:02:00,840 --> 00:02:04,880 term and maybe we can discuss the importance of data privacy 35 00:02:04,880 --> 00:02:07,200 and maybe some potential risks that you need to be aware of 36 00:02:07,200 --> 00:02:09,160 whenever you build your projects or products. 37 00:02:09,280 --> 00:02:10,759 So, Catherine, welcome to the show. 38 00:02:11,680 --> 00:02:13,880 Thanks so much, Henry. I'm excited to be here. 39 00:02:14,680 --> 00:02:17,640 Right, Catherine, I'd love to ask my guests to maybe share 40 00:02:17,640 --> 00:02:20,320 some turning points in your career that you think we all can 41 00:02:20,320 --> 00:02:23,120 learn from that? I think one of the turning 42 00:02:23,120 --> 00:02:27,240 points in my career, the early ones, is I started my job in 43 00:02:27,240 --> 00:02:31,200 technology as a data journalist, or what we would today called 44 00:02:31,200 --> 00:02:34,520 data journalists. Kind of like how do we use data 45 00:02:34,520 --> 00:02:38,360 and interactives to tell stories and to support investigative 46 00:02:38,360 --> 00:02:41,240 journalism. I was at the Washington Post 47 00:02:41,240 --> 00:02:43,720 doing that. There was certainly a large 48 00:02:43,720 --> 00:02:47,720 turning point where I left kind of the field of media and 49 00:02:47,720 --> 00:02:50,040 journalism and I went into startups. 50 00:02:50,320 --> 00:02:53,520 And at the time I went into startups that were focused on 51 00:02:53,720 --> 00:02:57,040 media companies as a customer base. 52 00:02:57,440 --> 00:03:00,600 And that led me into what I would today call large scale 53 00:03:00,800 --> 00:03:04,840 natural language processing or very early models compared to 54 00:03:04,840 --> 00:03:08,640 what we now call AI models. But still a lot of exposure to 55 00:03:08,640 --> 00:03:12,560 how do we do language processing as scale and a lot of exposure 56 00:03:12,560 --> 00:03:15,480 to thinking through things like parallel computation. 57 00:03:15,840 --> 00:03:18,760 Those were the days of Hadoop and all these other types of 58 00:03:18,760 --> 00:03:22,160 problems that came with how do we process large scale 59 00:03:22,160 --> 00:03:26,360 documents, stores and use them for some sort of other service. 60 00:03:27,040 --> 00:03:32,760 And then around 2014 is when I moved from Los Angeles, CA, 61 00:03:32,760 --> 00:03:36,680 where I was based to Berlin, Germany, where I now live since 62 00:03:36,680 --> 00:03:40,600 then. And this was also the time when 63 00:03:40,640 --> 00:03:44,880 NLP, So the space of language processing was moving from more 64 00:03:44,880 --> 00:03:47,960 simple model designs or what we would today maybe call more 65 00:03:47,960 --> 00:03:52,680 simple model designs into deep learning, which basically fuels 66 00:03:52,840 --> 00:03:55,160 a lot of the architectures that we talked about today when we 67 00:03:55,160 --> 00:03:59,440 talk about AI. And I basically took a year off 68 00:03:59,440 --> 00:04:04,840 of working and just studied deep learning at that time to kind of 69 00:04:04,840 --> 00:04:07,240 update. I already knew a lot of the math 70 00:04:07,240 --> 00:04:09,960 behind it because thankfully I always loved math. 71 00:04:10,520 --> 00:04:14,040 But to really upskill myself from how do we think of the 72 00:04:14,040 --> 00:04:16,160 problem from a statistical point of view? 73 00:04:16,160 --> 00:04:18,880 Statistical learning at the end of the day, deep learning is 74 00:04:18,880 --> 00:04:21,760 also statistical learning. But how do we move from that 75 00:04:21,760 --> 00:04:28,120 space to maybe a more kind of notion of applying concepts of 76 00:04:28,120 --> 00:04:31,320 calculus and linear algebra to a learning model? 77 00:04:31,800 --> 00:04:34,640 And that was a good idea because, yeah, kind of deep 78 00:04:34,640 --> 00:04:39,160 learning then ate most of the world of machine learning. 79 00:04:39,720 --> 00:04:43,920 Then the final career change was maybe about 3 or 4 years after 80 00:04:43,920 --> 00:04:48,080 that of doing deep learning. I started thinking or became 81 00:04:48,080 --> 00:04:51,440 enmeshed in the problem of kind of how do we think about what we 82 00:04:51,440 --> 00:04:55,240 would today called trustworthy and responsible AI development? 83 00:04:55,760 --> 00:04:59,560 Then most of the time we use the concept of ethical machine 84 00:04:59,560 --> 00:05:03,680 learning, and in looking into that problem, I became very 85 00:05:03,680 --> 00:05:08,040 interested in data privacy and data security of machine 86 00:05:08,040 --> 00:05:11,520 learning systems and models themselves. 87 00:05:11,960 --> 00:05:15,800 And that kind of fueled what eventually led to writing the 88 00:05:15,800 --> 00:05:19,920 book and obviously my current work, which is as a specialist 89 00:05:19,920 --> 00:05:22,200 in privacy and security machine learning models. 90 00:05:23,040 --> 00:05:24,600 Well, thank you for sharing your story. 91 00:05:24,600 --> 00:05:27,640 I find it quite interesting the journey that you had, right? 92 00:05:27,920 --> 00:05:31,200 And I think it's very unique that you took a career break 93 00:05:31,200 --> 00:05:34,000 simply to study deep learning and all that, right? 94 00:05:34,280 --> 00:05:37,320 So most people maybe took a break to do something else, you 95 00:05:37,320 --> 00:05:39,600 know, beyond work or some did go study. 96 00:05:39,600 --> 00:05:42,800 But I think it's still pretty rare to go deep into a certain 97 00:05:42,800 --> 00:05:46,040 area and just to study and do maybe more things to get expert 98 00:05:46,040 --> 00:05:48,520 on that. This episode is brought to you 99 00:05:48,520 --> 00:05:52,720 by Swim dot IO and I'm excited to have its CTO and Co founder 100 00:05:52,920 --> 00:05:56,360 Omar Rosenbaum with me today to tell you more about SWIM. 101 00:05:56,440 --> 00:05:59,040 Hi Henry, very nice to meet you and thank you for having me. 102 00:05:59,360 --> 00:06:02,160 So tell us a little bit more, what is swim dot IO? 103 00:06:02,320 --> 00:06:06,040 At Swim, we want to help companies understand their code 104 00:06:06,040 --> 00:06:09,000 bases. We combine static code analysis 105 00:06:09,040 --> 00:06:12,960 with generative AI to create comprehensive documents that 106 00:06:12,960 --> 00:06:16,120 help you navigate the code base. As an engineer myself, I 107 00:06:16,120 --> 00:06:19,960 wouldn't want them 10 years to spend so much time understanding 108 00:06:19,960 --> 00:06:22,400 existing code. I would want them to spend time 109 00:06:22,600 --> 00:06:26,240 creating and building new stuff. When you have code that has 110 00:06:26,240 --> 00:06:30,840 accumulated over decades, and especially in legacy languages 111 00:06:31,160 --> 00:06:36,640 that not many people are adapted nowadays, then the problem is 112 00:06:36,640 --> 00:06:40,080 even bigger. Swim dot IO is specializing into 113 00:06:40,080 --> 00:06:42,600 helping mainframe developers to understand their code base. 114 00:06:42,680 --> 00:06:45,640 Why mainframes? We actually didn't start there. 115 00:06:45,720 --> 00:06:51,320 COBOL had been by some people obsolete for a few years, and I 116 00:06:51,320 --> 00:06:54,720 discovered that it's not really obsolete, not at all. 117 00:06:54,720 --> 00:06:58,840 There are more than 800 billion lines of COBOL code that are in 118 00:06:58,840 --> 00:07:01,680 production and they drive lots of the business in the world. 119 00:07:02,360 --> 00:07:07,800 And we got more and more requests from customers to help 120 00:07:07,800 --> 00:07:10,880 them understand the legacy code visas that was written decades 121 00:07:10,880 --> 00:07:14,360 ago and got accumulated over a very long period of time. 122 00:07:14,560 --> 00:07:17,920 So from your customers so far, what are the some of the success 123 00:07:17,920 --> 00:07:21,320 stories that you can share? So we worked with an analyst who 124 00:07:21,560 --> 00:07:25,960 shared with us that it took them a year to document a single 125 00:07:25,960 --> 00:07:29,240 mainframe application, and using SWIM they were able to document 126 00:07:29,240 --> 00:07:31,800 a similar application in a matter of hours. 127 00:07:32,280 --> 00:07:36,160 So saving that amount of time enables them to focus on other 128 00:07:36,400 --> 00:07:39,280 tasks. Thanks Amir for sharing with us 129 00:07:39,280 --> 00:07:42,040 about SWIM today. To learn more about SWIM, check 130 00:07:42,040 --> 00:07:44,120 out their website at swim dot IO. 131 00:07:45,440 --> 00:07:48,520 So today's topic we are going to cover about data privacy. 132 00:07:48,520 --> 00:07:52,920 So I think in a lots of parts of the world, this topic or this 133 00:07:52,920 --> 00:07:56,520 kind of thing has become like the mainstream topics, right? 134 00:07:56,840 --> 00:08:01,200 We heard a lot in news, data breaches, you know, data leaks, 135 00:08:01,600 --> 00:08:05,080 people's privacy, you know, the Cambridge Analytica kind of 136 00:08:05,080 --> 00:08:07,360 thing, also as part of the privacy thing. 137 00:08:07,720 --> 00:08:10,880 So maybe if you can give us an overview first, what is the 138 00:08:10,880 --> 00:08:15,200 current landscape of the data privacy thing and what should 139 00:08:15,200 --> 00:08:17,360 people know about it? Yeah. 140 00:08:17,360 --> 00:08:22,080 I mean, a lot of times I try to Orient people first with the 141 00:08:22,080 --> 00:08:27,600 idea of kind of personal or we can think of like sociological 142 00:08:27,600 --> 00:08:31,840 privacy because sometimes people hear data privacy and they just 143 00:08:31,840 --> 00:08:35,440 think, oh, that's for lawyers. Like that's those people are 144 00:08:35,440 --> 00:08:37,360 responsible. They stuttered privacy law. 145 00:08:37,679 --> 00:08:39,640 Those people are really, really important in the field of 146 00:08:39,640 --> 00:08:41,320 privacy. I'm not trying to diminish the 147 00:08:41,320 --> 00:08:45,760 importance of law and regulation in the field of privacy, but I 148 00:08:45,760 --> 00:08:49,640 think we can all relate to privacy because each one of us 149 00:08:50,080 --> 00:08:53,960 probably has some sort of personal or what we might call 150 00:08:53,960 --> 00:08:57,000 individual understanding of our own privacy. 151 00:08:57,520 --> 00:09:02,040 And often that's informed by whatever cultural influences we 152 00:09:02,040 --> 00:09:06,360 grew up around whatever society. Like where did privacy play or 153 00:09:06,360 --> 00:09:11,160 where did trust play in whatever societal bonds we kind of grew 154 00:09:11,160 --> 00:09:13,920 up with and we're socialized with or maybe even where we live 155 00:09:13,920 --> 00:09:16,200 now. So like, for example, now I live 156 00:09:16,200 --> 00:09:20,040 in Germany has a very different relationship with privacy than 157 00:09:20,280 --> 00:09:24,560 where I come from, California. And so we can also be influenced 158 00:09:24,560 --> 00:09:29,600 by kind of shifts or changes along our life that allow us to 159 00:09:29,600 --> 00:09:34,280 see the problem differently. But essentially privacy is a lot 160 00:09:34,280 --> 00:09:40,280 about having the autonomy and the control to decide who I want 161 00:09:40,280 --> 00:09:44,840 to show up as and what I want to share with whom under what 162 00:09:44,840 --> 00:09:48,680 circumstances or contexts. And I think we all know that 163 00:09:48,680 --> 00:09:52,800 there's been times probably in our lives when information that 164 00:09:52,800 --> 00:09:56,160 we didn't want to share with a particular person or with groups 165 00:09:56,160 --> 00:10:00,280 of persons got out some which way, sometimes through 166 00:10:00,280 --> 00:10:02,920 technology, sometimes through a person, maybe both. 167 00:10:03,560 --> 00:10:07,520 And we all know that icky feeling when your privacy is 168 00:10:07,520 --> 00:10:11,440 violated. And that's how it can also teach 169 00:10:11,440 --> 00:10:15,440 us how much it is about trust, how much it is about context, 170 00:10:15,720 --> 00:10:18,960 how much it is about us having a choice. 171 00:10:18,960 --> 00:10:23,760 And I think that you can see kind of the social constructs in 172 00:10:23,760 --> 00:10:26,040 a lot of the regulation that you then read. 173 00:10:26,040 --> 00:10:29,840 Because then of course, we have regulatory or what I would call 174 00:10:29,840 --> 00:10:32,640 like judicial legal understanding of privacy. 175 00:10:32,960 --> 00:10:36,320 And then we have a technical understanding of privacy. 176 00:10:36,320 --> 00:10:40,280 And kind of the best part is when all of those work together, 177 00:10:40,280 --> 00:10:43,400 when the technology is reflecting not only what we 178 00:10:43,400 --> 00:10:47,400 legally need to do, but also what we socially and culturally 179 00:10:47,400 --> 00:10:51,640 understand, and knowing that that can shift depending on all 180 00:10:51,640 --> 00:10:55,200 sorts of different things down to the individual level. 181 00:10:56,080 --> 00:10:57,960 I like the three things that you mentioned, right? 182 00:10:57,960 --> 00:11:00,960 First, it may be relates to the context and society, right? 183 00:11:00,960 --> 00:11:04,360 So privacy in one particular area might be different and 184 00:11:04,440 --> 00:11:07,240 maybe the norm of the culture in some other parts of the world. 185 00:11:07,560 --> 00:11:10,760 I also like the trusting because like most of the time when we 186 00:11:10,760 --> 00:11:13,320 use software, right, the concept of trust is a little bit 187 00:11:13,320 --> 00:11:14,520 abstract. It's vague, right? 188 00:11:14,600 --> 00:11:18,200 I know that I submit my details, but that's not necessary a trust 189 00:11:18,200 --> 00:11:20,720 being created or established when I use the software. 190 00:11:21,120 --> 00:11:22,880 And the last one is about choice, right? 191 00:11:22,880 --> 00:11:26,880 I do have a choice to actually maybe take out my data, delete 192 00:11:26,880 --> 00:11:28,360 my data or whatever that is, right? 193 00:11:28,360 --> 00:11:31,400 So I think those are the key terms that I just picked from 194 00:11:31,400 --> 00:11:34,120 you what you explained just now. And a lot of people actually 195 00:11:34,120 --> 00:11:36,880 associate privacy with PII, right? 196 00:11:36,960 --> 00:11:39,000 Personally identifiable information. 197 00:11:39,280 --> 00:11:43,040 Do you think that privacy relates just to PII or is it 198 00:11:43,040 --> 00:11:46,360 more than that? Yeah, I mean, PII is a great 199 00:11:46,360 --> 00:11:50,680 place to start. I don't wanna PII bash here, but 200 00:11:51,000 --> 00:11:57,040 I think PII is a very small view of data that we might think of 201 00:11:57,040 --> 00:11:59,840 as what I like to call sensitive data. 202 00:12:00,400 --> 00:12:03,480 Within the field of sensitive data, of course we have PII. 203 00:12:03,480 --> 00:12:05,880 That's things just for people who don't know that term. 204 00:12:05,880 --> 00:12:09,520 That's things like e-mail addresses, birth dates, names, 205 00:12:09,920 --> 00:12:13,040 things that we would say might be unique to an individual and 206 00:12:13,040 --> 00:12:16,440 certainly in combination are unique to an individual. 207 00:12:17,000 --> 00:12:20,680 And then we might have what I tend to call person related 208 00:12:20,680 --> 00:12:23,040 data. Not everybody calls it that. 209 00:12:23,040 --> 00:12:24,840 Some people want to call personal data. 210 00:12:24,840 --> 00:12:28,160 Some people call it, yeah, data related to persons. 211 00:12:28,160 --> 00:12:30,600 Anyways, there's all sorts of terms for it, especially when 212 00:12:30,600 --> 00:12:33,560 you get into the legal realm. But I think of person related 213 00:12:33,560 --> 00:12:38,080 data as like things like what I click on, what I buy can be 214 00:12:38,080 --> 00:12:42,200 person related, What I enter into search terms can be person 215 00:12:42,200 --> 00:12:44,520 related. Because when we think of all 216 00:12:44,520 --> 00:12:48,200 these things in combination with each other, probably when we 217 00:12:48,200 --> 00:12:52,680 combine them, those can also point to uniquely identifiable 218 00:12:52,680 --> 00:12:56,400 characteristics of a person. And you already mentioned 219 00:12:56,400 --> 00:12:59,200 Cambridge Analytica. Some of the original research 220 00:12:59,200 --> 00:13:03,120 was focused on can we use things like Facebook likes or very 221 00:13:03,120 --> 00:13:07,760 short Facebook surveys or those combinations to decide how this 222 00:13:07,760 --> 00:13:11,520 person might vote, whether they're easily influenced in 223 00:13:11,520 --> 00:13:14,280 voting, and these types of characteristics which we 224 00:13:14,280 --> 00:13:16,400 probably would think of as sensitive. 225 00:13:17,120 --> 00:13:20,120 And then we also have other sensitive data that isn't really 226 00:13:20,120 --> 00:13:24,840 person related that we might call proprietary or confidential 227 00:13:24,840 --> 00:13:28,680 data. And a lot of times we don't 228 00:13:28,680 --> 00:13:30,720 think of that when I think we think of privacy, because 229 00:13:30,720 --> 00:13:33,240 obviously it doesn't really have to do with privacy. 230 00:13:33,520 --> 00:13:38,560 And yet the types of protections that we use for privacy can also 231 00:13:38,560 --> 00:13:42,720 be applied to things like corporate secrets, other 232 00:13:42,720 --> 00:13:46,000 proprietary confidential information that we also don't 233 00:13:46,000 --> 00:13:48,760 want to share outside of a particular context. 234 00:13:49,040 --> 00:13:52,680 And so these all kind of can fall under some idea of data 235 00:13:52,680 --> 00:13:56,760 that we might need to protect in a different way than other data. 236 00:13:57,880 --> 00:13:58,840 Nice. Thanks for all the 237 00:13:58,840 --> 00:14:01,600 classifications of different person related, right? 238 00:14:01,600 --> 00:14:04,200 It could be the sensitive, highly sensitive ones, it could 239 00:14:04,200 --> 00:14:07,080 be just the person related, it could be like confidential data. 240 00:14:07,120 --> 00:14:09,160 So I think that totally makes sense, right? 241 00:14:09,480 --> 00:14:12,120 So not necessarily just like person related, it could be 242 00:14:12,120 --> 00:14:15,360 entity related. And I think a lot of times when 243 00:14:15,360 --> 00:14:17,240 we talk about privacy these days, there are a lot of 244 00:14:17,240 --> 00:14:20,960 technologies that could expose us to this privacy risk, right? 245 00:14:21,280 --> 00:14:23,760 I mean, just to mention a few things like social media, 246 00:14:23,760 --> 00:14:25,760 definitely. And then you have the web 247 00:14:25,760 --> 00:14:28,680 technologies like cookies, you know, when you browse a certain 248 00:14:28,680 --> 00:14:31,160 things, right? It tends to follow you from one 249 00:14:31,160 --> 00:14:34,360 page, one website to the other. You have the connected device. 250 00:14:34,440 --> 00:14:37,360 You know, we talk about, you know, this Alexa or, you know, 251 00:14:37,440 --> 00:14:41,520 the Google Home thing, right? And lately also AI, right? 252 00:14:41,520 --> 00:14:44,440 So people just, you know, ask AI about certain stuff. 253 00:14:44,440 --> 00:14:47,480 Sometimes it could take your data away, maybe looking at 254 00:14:47,560 --> 00:14:50,360 those technologies. What will be your advice for 255 00:14:50,360 --> 00:14:52,520 people when they use those technologies, right? 256 00:14:52,520 --> 00:14:54,400 Because it's such a ubiquitous thing now. 257 00:14:54,760 --> 00:14:58,360 What should people care about in terms of their privacy? 258 00:14:59,320 --> 00:15:02,920 Yeah, that's, that's a hard question because obviously 259 00:15:02,920 --> 00:15:05,960 everybody might be different. And again, like we said, like 260 00:15:05,960 --> 00:15:10,080 there's also different levels of trust that people might have 261 00:15:10,080 --> 00:15:14,040 with different services, right. So you might really, really like 262 00:15:14,040 --> 00:15:17,520 Copilot or whatever it is that you use everyday and you might 263 00:15:17,520 --> 00:15:21,280 decide, you know what the usefulness of this is enough for 264 00:15:21,280 --> 00:15:25,480 whatever trade off that I have. At the end of the day, a lot of 265 00:15:25,480 --> 00:15:29,840 my work is focused around informing organizations what 266 00:15:29,840 --> 00:15:34,360 they should be doing to better protect the people's privacy 267 00:15:34,360 --> 00:15:36,680 that use their services and products. 268 00:15:36,680 --> 00:15:41,000 And I think we run into this problem and I think this happens 269 00:15:41,000 --> 00:15:45,240 a lot also in information security or cybersecurity where 270 00:15:45,240 --> 00:15:50,120 we kind of decide that is a user problem and that like it's your 271 00:15:50,120 --> 00:15:54,360 job to figure out like what you think is the most secure e-mail 272 00:15:54,360 --> 00:15:57,400 service and what is like the best thing. 273 00:15:57,400 --> 00:16:01,120 And like, oh, you're bad if you use WhatsApp because Whatsapp's 274 00:16:01,120 --> 00:16:03,680 not secure, which is not necessarily true, right? 275 00:16:04,000 --> 00:16:07,920 But the this kind of culture, I call it kind of like blaming the 276 00:16:07,920 --> 00:16:13,400 victim type of culture, where why is it that I, if I want to 277 00:16:13,400 --> 00:16:18,360 talk to a person and they only use one particular messaging 278 00:16:18,360 --> 00:16:23,320 application, why is it my job to make sure to try to inform 279 00:16:23,320 --> 00:16:26,560 people what is the most secure or not most secure? 280 00:16:26,560 --> 00:16:32,160 And this why can't we hold a higher bar for organizations to 281 00:16:32,160 --> 00:16:37,840 create secure and private by design services so that people 282 00:16:37,840 --> 00:16:39,560 can choose whatever one they like. 283 00:16:39,560 --> 00:16:41,120 If they like the colors, I don't care. 284 00:16:41,120 --> 00:16:44,280 They like the interface, they like whatever product they like. 285 00:16:44,280 --> 00:16:46,840 If they want to buy a package deal from a certain cloud 286 00:16:46,840 --> 00:16:50,800 provider, If we're looking at privacy more holistically, it's 287 00:16:50,800 --> 00:16:54,040 less about like you must use this software. 288 00:16:54,360 --> 00:16:59,080 Then we need to hold software companies and product companies 289 00:16:59,080 --> 00:17:03,560 to a higher standard so that people can choose whatever it is 290 00:17:03,560 --> 00:17:07,560 they like, which I'm like very pro-choice on that. 291 00:17:07,560 --> 00:17:12,800 One of obviously there's lots of good advice out there how to 292 00:17:12,800 --> 00:17:16,480 inform yourself, doing things like reading through the boring 293 00:17:16,480 --> 00:17:20,280 privacy policies and figuring out how to make it a fun game 294 00:17:20,280 --> 00:17:22,839 for yourself. There's plenty of good advice to 295 00:17:22,839 --> 00:17:25,800 inform yourself for how do you make those choices? 296 00:17:26,200 --> 00:17:32,040 But I think it almost focuses like the blame on the individual 297 00:17:32,040 --> 00:17:34,800 who has to use things. And at the end of the day, I 298 00:17:34,800 --> 00:17:38,640 want people to, yeah, be able to connect and use technology in 299 00:17:38,640 --> 00:17:44,480 cool ways and not have this very privileged burden of reading 300 00:17:44,480 --> 00:17:47,440 through every single privacy notice to figure out which best 301 00:17:47,440 --> 00:17:49,280 aligns. And, oh, by the way, then it 302 00:17:49,280 --> 00:17:51,960 gets updated six months later. They got to read through and 303 00:17:51,960 --> 00:17:56,720 compare all of them over again. Mozilla has a great resource, by 304 00:17:56,720 --> 00:18:00,400 the way, for people who might be looking for what criteria might 305 00:18:00,400 --> 00:18:03,640 I use called Privacy Not Included. 306 00:18:04,160 --> 00:18:07,920 And that reviews things like I think they did one on autos, 307 00:18:07,920 --> 00:18:11,600 like automobiles. I think they did one on like 308 00:18:11,600 --> 00:18:14,360 connect to devices, home assistance like you mentioned, 309 00:18:14,760 --> 00:18:16,960 and a few other things. And then at least you can look 310 00:18:16,960 --> 00:18:20,320 like what criteria did they use and maybe start to build your 311 00:18:20,320 --> 00:18:24,720 own criteria to inform your choices should you have the time 312 00:18:24,720 --> 00:18:28,160 and energy to do so. And there's no shame if you 313 00:18:28,160 --> 00:18:31,160 don't because at the end of the day, we got to hold the orgs 314 00:18:31,160 --> 00:18:33,200 accountable. Right. 315 00:18:33,200 --> 00:18:36,040 So very interesting the way you explain that, right. 316 00:18:36,040 --> 00:18:39,280 So it used to be, you know, like it's a user's problem, right? 317 00:18:39,280 --> 00:18:42,920 So the one who is responsible to take care about the data that 318 00:18:42,920 --> 00:18:45,400 you share. And I think we can read all this 319 00:18:45,400 --> 00:18:48,800 privacy policy and all that sometimes in most of in, I don't 320 00:18:48,800 --> 00:18:51,600 know most of it, but many software, but we don't have a 321 00:18:51,600 --> 00:18:54,120 choice. We simply just to have to Czech 322 00:18:54,400 --> 00:18:57,840 agree and you know, like, you know, whatever language that 323 00:18:57,840 --> 00:18:59,480 they provide, we just agree, right. 324 00:18:59,480 --> 00:19:01,720 Otherwise we can't use anything from the software at all. 325 00:19:02,040 --> 00:19:05,480 So I think that's a very yeah, that's a very good segue now to 326 00:19:05,480 --> 00:19:09,320 actually move into towards the organizations or the teams who 327 00:19:09,320 --> 00:19:10,920 build the software and the products, right. 328 00:19:11,240 --> 00:19:15,280 So the first thing is I came from some, you know, belief, 329 00:19:15,280 --> 00:19:18,000 probably right last time, especially in the era of big 330 00:19:18,000 --> 00:19:21,800 data, we have to collect as many data as possible, you know, from 331 00:19:21,800 --> 00:19:25,880 the users so that we can derive insights, we can analyse things. 332 00:19:26,280 --> 00:19:29,160 And now this seems to be clashing, you know, with data 333 00:19:29,160 --> 00:19:32,240 privacy. So in this era, these days, what 334 00:19:32,240 --> 00:19:35,320 would you advise people in terms of balance, right, about 335 00:19:35,320 --> 00:19:38,640 collecting as many data as possible versus, you know, 336 00:19:39,000 --> 00:19:43,720 treating privacy more seriously? I think First off, I want to say 337 00:19:43,720 --> 00:19:46,440 like, I'm not trying to get anybody to lose their job. 338 00:19:46,560 --> 00:19:50,680 So if any of the advice I gave you think would make you lose 339 00:19:50,680 --> 00:19:54,720 your job, don't do it right. So there's some mini disclaimer. 340 00:19:56,240 --> 00:19:59,480 But I think I think as you mentioned, I think there's, 341 00:19:59,680 --> 00:20:05,200 there's starting to be I think a more popular shift into even 342 00:20:05,200 --> 00:20:09,800 just consumers, so to speak, the average person thinking more 343 00:20:09,800 --> 00:20:13,360 about privacy. Namely because I think this 344 00:20:13,360 --> 00:20:17,960 trend, as you mentioned, of the big data era of just collecting, 345 00:20:17,960 --> 00:20:22,000 you know, at will and doing all sorts of things that at the end 346 00:20:22,000 --> 00:20:24,960 of the day make people feel creeped out, you know, make 347 00:20:24,960 --> 00:20:28,800 people feel like their phone is spying on them and all these 348 00:20:28,800 --> 00:20:31,360 things are happening, right? And in some cases that might be 349 00:20:31,360 --> 00:20:34,760 true, and in other cases it might be a fact that we have so 350 00:20:34,760 --> 00:20:39,200 much data that we can actually infer sensitive things about 351 00:20:39,200 --> 00:20:43,080 people without their knowledge and also without our intention, 352 00:20:43,200 --> 00:20:44,480 right? When we think about 353 00:20:44,920 --> 00:20:48,560 recommendation algorithm development, search response 354 00:20:48,560 --> 00:20:52,680 algorithm development, so search algorithms and content 355 00:20:52,680 --> 00:20:56,360 algorithms, when we think about a lot of this, there can be 356 00:20:56,360 --> 00:20:58,840 latent, what we call latent variables. 357 00:20:58,840 --> 00:21:01,920 And what I mean by that is, you know, because you follow and you 358 00:21:01,920 --> 00:21:06,360 like these five things, we kind of have learned your gender or 359 00:21:06,360 --> 00:21:09,800 your age or your family status or these other things. 360 00:21:10,160 --> 00:21:12,520 We didn't explicitly try to learn them, but we kind of 361 00:21:12,520 --> 00:21:16,120 learned them by amassing, as you say, so much data and then 362 00:21:16,120 --> 00:21:19,840 putting an algorithm on top and not controlling for anything 363 00:21:19,840 --> 00:21:22,480 like privacy. And then you get super creepy 364 00:21:22,480 --> 00:21:26,720 ads and you're like convinced your phone is spying on you and 365 00:21:26,800 --> 00:21:29,280 then you must hide it or you know what I mean? 366 00:21:29,280 --> 00:21:32,000 Like, I don't want people to be afraid of, I don't think any of 367 00:21:32,000 --> 00:21:33,920 us want people to be afraid of technology. 368 00:21:33,920 --> 00:21:37,320 I also don't personally want to be afraid to have a mobile 369 00:21:37,720 --> 00:21:39,920 phone. It's a nice thing to have. 370 00:21:40,400 --> 00:21:45,320 And so when I think about like this shift that's happening, 371 00:21:45,600 --> 00:21:48,480 maybe one of the things that you can start talking about in an 372 00:21:48,480 --> 00:21:52,080 organization that may or may not already have a culture around 373 00:21:52,080 --> 00:21:56,760 thinking about privacy is how do we reason about the trust that 374 00:21:56,760 --> 00:21:59,800 we're creating? And I think this is great for 375 00:21:59,800 --> 00:22:04,000 tech leads, as you aptly point out, and product people, because 376 00:22:04,000 --> 00:22:06,720 at the end of the day, when you're like a technical lead or 377 00:22:06,720 --> 00:22:10,920 a principal or a senior person in the type of architecture 378 00:22:10,920 --> 00:22:15,360 decisions, product decisions, data decisions that get made and 379 00:22:15,360 --> 00:22:19,360 you're a top product person, those people can have a real 380 00:22:19,360 --> 00:22:23,120 conversation about, do we want to talk with our customers about 381 00:22:23,120 --> 00:22:25,560 this problem of privacy? Do we want to talk with our 382 00:22:25,560 --> 00:22:28,600 customers about what's the mental model that they think 383 00:22:28,840 --> 00:22:32,480 about how our service works? What parts of this is up for 384 00:22:32,480 --> 00:22:34,360 debate? What parts of this is not? 385 00:22:34,360 --> 00:22:37,880 Because there are like indeed businesses that by default, 386 00:22:38,720 --> 00:22:42,520 probably privacy is not going to play a huge role, right? 387 00:22:43,080 --> 00:22:45,840 But you can at least start that conversation. 388 00:22:46,320 --> 00:22:49,320 And what we often talk about in the field of technical privacy 389 00:22:49,320 --> 00:22:53,680 is balancing between the most privacy we can offer, which 390 00:22:53,680 --> 00:22:58,000 would be collecting no data and utility, right? 391 00:22:58,000 --> 00:23:01,800 Is what we think about privacy, utility balance or trade-offs 392 00:23:02,240 --> 00:23:04,880 and utility. There we might think of as this 393 00:23:04,880 --> 00:23:08,880 old idea of we're going to collect every single mouse 394 00:23:08,880 --> 00:23:14,600 movement or single character you type in in the hopes of getting 395 00:23:14,600 --> 00:23:18,560 some sort of insight. Which I would also argue as a 396 00:23:18,560 --> 00:23:21,280 data person, if you don't know what you're going to use the 397 00:23:21,280 --> 00:23:25,440 data for, you're probably just amassing large cloud computing 398 00:23:25,440 --> 00:23:29,200 fees for no reason. And so it's better to have 399 00:23:29,200 --> 00:23:33,640 smaller experiments with less data and then to grow it over 400 00:23:33,640 --> 00:23:35,760 time. Again, if you will lose your job 401 00:23:35,760 --> 00:23:39,760 for doing that, I'm sorry. And maybe you can find a new job 402 00:23:39,760 --> 00:23:43,520 one day. Right. 403 00:23:43,840 --> 00:23:46,320 So I think you mentioned something very important, right? 404 00:23:46,440 --> 00:23:49,880 And many, many privacy experts also say the same thing, right? 405 00:23:49,920 --> 00:23:53,320 If you don't know what you collect that data for, then 406 00:23:53,320 --> 00:23:55,440 don't do it, right? Don't store it, don't do it, 407 00:23:55,440 --> 00:23:58,240 don't ask for it, right? So I think that's also very, 408 00:23:58,240 --> 00:24:01,440 very important, right, for everyone who builds software 409 00:24:01,440 --> 00:24:03,480 these days, right? Especially the technologies, 410 00:24:03,480 --> 00:24:05,200 right? And also when you build AI 411 00:24:05,200 --> 00:24:08,040 model, right? I mean, last time we also simply 412 00:24:08,040 --> 00:24:11,720 say for machine learning, it needs as many variables as much 413 00:24:11,720 --> 00:24:14,000 as possible, right? So that it can train the model 414 00:24:14,000 --> 00:24:15,560 better. But again, like, if you don't 415 00:24:15,560 --> 00:24:18,320 know how you will want to use the data, maybe it's better not 416 00:24:18,320 --> 00:24:20,360 to collect them. And this comes back to this 417 00:24:20,360 --> 00:24:23,840 thing called privacy by design. I think it's mentioned a lot of 418 00:24:23,840 --> 00:24:26,160 times, right? In security, there's a secure by 419 00:24:26,160 --> 00:24:27,880 design. In privacy now there's the 420 00:24:27,920 --> 00:24:30,880 privacy by design. So maybe it can help explain 421 00:24:30,880 --> 00:24:33,600 what is privacy by design and how we should use it. 422 00:24:34,480 --> 00:24:38,320 Yeah, Privacy by Design is actually like a quite old 423 00:24:38,320 --> 00:24:41,760 concept, which is pretty cool. I think it was developed first 424 00:24:41,760 --> 00:24:46,840 in the early 2000s by a researcher in Canada, Kabukin, I 425 00:24:46,840 --> 00:24:48,440 don't know if I pronounced the name right. 426 00:24:48,440 --> 00:24:53,880 Anyways, and she developed these principles as part of research. 427 00:24:53,880 --> 00:24:56,840 I think she was working on as somebody that both understood 428 00:24:56,840 --> 00:25:01,800 some of the privacy regulatory aspects and then also the 429 00:25:01,800 --> 00:25:04,240 software aspects. And when you read the original 430 00:25:04,240 --> 00:25:07,280 privacy by design, I think there's seven design principles. 431 00:25:07,840 --> 00:25:11,360 You see a lot of software thinking this is like really pre 432 00:25:11,360 --> 00:25:16,440 data, pre algorithms as like a normal part of software design. 433 00:25:16,840 --> 00:25:19,560 And a lot of it talks about the same types of themes we're 434 00:25:19,560 --> 00:25:23,920 talking about now, allowing user choice, allowing transparency, 435 00:25:24,280 --> 00:25:29,200 but also by default building things that respect the user, 436 00:25:29,640 --> 00:25:33,360 that respects the fact that privacy does have all these, you 437 00:25:33,360 --> 00:25:38,160 know, is informed by context. And so if I put something in a 438 00:25:38,160 --> 00:25:41,320 form, it doesn't necessarily mean I want that to be used to 439 00:25:41,320 --> 00:25:45,320 train an algorithm, right? That we understand like how the 440 00:25:45,320 --> 00:25:47,960 user is imagining what it is they're doing. 441 00:25:48,280 --> 00:25:51,080 And that we're building technology that hopefully 442 00:25:51,080 --> 00:25:54,680 mirrors that or if not, makes it more obvious what is happening. 443 00:25:55,520 --> 00:26:00,120 And then security principles, so basic, not only application 444 00:26:00,120 --> 00:26:02,480 security, but also infrastructure security 445 00:26:02,480 --> 00:26:07,480 principles so that we don't end up exposing something that we've 446 00:26:07,480 --> 00:26:11,080 collected and trusts that we are even using responsibly. 447 00:26:11,320 --> 00:26:16,040 But then open the system up for potential attacks or other ways 448 00:26:16,040 --> 00:26:20,240 of exfiltrating that data and using it for another purpose. 449 00:26:20,800 --> 00:26:23,720 And this is where I think we can take privacy by design. 450 00:26:23,720 --> 00:26:26,080 We can use that to architect and build our systems. 451 00:26:26,400 --> 00:26:30,040 And then we can also evolve that thinking for new things like 452 00:26:30,320 --> 00:26:33,960 building large scale data systems, algorithmic systems. 453 00:26:34,360 --> 00:26:38,280 And then we have a huge problem now around third party data 454 00:26:38,280 --> 00:26:42,480 systems of the, you know, which I don't think by any means 455 00:26:42,600 --> 00:26:45,520 private by design. And so when we think of third 456 00:26:45,520 --> 00:26:49,560 party, it can either be services we're using and we are the 457 00:26:49,560 --> 00:26:53,880 intermediary then between our users who maybe we have a direct 458 00:26:53,880 --> 00:26:56,200 relationship with and direct trust with. 459 00:26:56,560 --> 00:26:59,880 And then we want to use a third party system, which is fine as 460 00:26:59,880 --> 00:27:03,160 long as we're really clear and as long as we're actually 461 00:27:03,160 --> 00:27:06,680 reviewing the third party vendors that we use for things 462 00:27:06,680 --> 00:27:10,840 like privacy or the other way around where we're actually a 463 00:27:10,840 --> 00:27:14,800 third party vendor, we have 0 access to the users. 464 00:27:14,800 --> 00:27:17,600 And this is a lot of us that work in the B to B space. 465 00:27:17,760 --> 00:27:21,240 We don't actually, I mean, usually the companies that we're 466 00:27:21,240 --> 00:27:24,360 interacting with, they might have customers or they might 467 00:27:24,360 --> 00:27:26,640 even have customers who have customers, right? 468 00:27:26,640 --> 00:27:30,760 And so how many layers until we get to a human who can make a 469 00:27:30,760 --> 00:27:33,880 choice about whether they want their data used that way or not. 470 00:27:34,200 --> 00:27:38,120 And do we have a good review process to make sure that that 471 00:27:38,200 --> 00:27:43,120 kind of chain of trust is verified and that it's mentally 472 00:27:43,120 --> 00:27:45,720 sound both the way that we're dealing with these trust 473 00:27:45,720 --> 00:27:49,440 relationships and also how we're eventually communicating all the 474 00:27:49,440 --> 00:27:53,560 way down to a person who can say, you know what, I'd actually 475 00:27:53,560 --> 00:27:57,560 rather not if I could opt out. I'd rather use this service, but 476 00:27:57,560 --> 00:28:00,960 without the interface to ChatGPT, for example. 477 00:28:01,280 --> 00:28:04,120 And I think we kind of see this in some of the new products that 478 00:28:04,120 --> 00:28:09,280 come out that there's a little bit more transparency of, hey, 479 00:28:09,280 --> 00:28:12,720 we built it this way, here's the services we're using. 480 00:28:13,000 --> 00:28:16,040 Do you want to use this particular feature or not? 481 00:28:16,160 --> 00:28:19,240 At least? I see that a lot in the way that 482 00:28:19,240 --> 00:28:23,080 Apple is choosing to design some pieces of Apple intelligence and 483 00:28:23,080 --> 00:28:25,640 so forth. Is trying to have that 484 00:28:25,640 --> 00:28:28,400 conversation in the open. Does it always go well? 485 00:28:28,400 --> 00:28:31,840 Maybe not, but at least trying to start that conversation with 486 00:28:31,840 --> 00:28:36,240 users in the open about how they can better understand and 487 00:28:36,240 --> 00:28:39,560 control the way that the data flows through the systems. 488 00:28:39,560 --> 00:28:43,000 And I would call that like today's updated version of 489 00:28:43,000 --> 00:28:44,760 Thinking through Privacy by design. 490 00:28:45,880 --> 00:28:49,400 So after you mentioned about, you know, third parties and how 491 00:28:49,400 --> 00:28:52,880 data their data intermediaries or even people these days, we 492 00:28:52,880 --> 00:28:55,400 use a lot of SAS, right? I mean software product teams, 493 00:28:55,400 --> 00:28:57,200 right? We use a lot of SAS, you know, 494 00:28:57,200 --> 00:29:00,120 maybe manage database or serverless database and things 495 00:29:00,120 --> 00:29:01,960 like that. So that kind of like implies the 496 00:29:01,960 --> 00:29:05,280 data can be stored in most of other places and you know, 497 00:29:05,280 --> 00:29:08,400 beyond our premise, right? So I think this also speaks 498 00:29:08,400 --> 00:29:11,160 very, very true. The fact that sometimes if your 499 00:29:11,160 --> 00:29:14,560 organization grow really large, it's very difficult to actually 500 00:29:14,560 --> 00:29:17,920 know what data have you collected, where they reside, 501 00:29:18,640 --> 00:29:21,920 the retention, you know, who people get access to, right? 502 00:29:22,120 --> 00:29:24,800 So tell us about the importance of this, you know, data 503 00:29:24,800 --> 00:29:27,880 governance, because in many parts of conversation I was in 504 00:29:27,880 --> 00:29:31,280 about data privacy, I think the first step is always to know the 505 00:29:31,280 --> 00:29:34,200 inventory of your data, right? Govern that particular 506 00:29:34,200 --> 00:29:36,400 inventory. So maybe tell us the importance 507 00:29:36,400 --> 00:29:38,320 of data governance. Yeah. 508 00:29:38,320 --> 00:29:41,800 And I think data governance is such a critical even outside of 509 00:29:41,800 --> 00:29:44,880 privacy. But you said it perfectly. 510 00:29:44,880 --> 00:29:47,720 I mean, if you don't know what data you have, if you don't know 511 00:29:47,720 --> 00:29:51,160 where it lives, if you don't know what pathways it goes 512 00:29:51,160 --> 00:29:53,400 through, like when we talk about data engineering, if you don't 513 00:29:53,400 --> 00:29:56,440 know where it's being processed, how it's being processed, by 514 00:29:56,440 --> 00:30:01,760 whom along the way, then you kind of end up in a place where 515 00:30:01,760 --> 00:30:05,800 privacy is not possible to some degree, right? 516 00:30:06,120 --> 00:30:10,800 Hopefully you've kind of, again, I'll re mention having strong 517 00:30:10,880 --> 00:30:15,160 practice of vendor review and vendor assessment because I 518 00:30:15,160 --> 00:30:19,720 think like thinking through risk always means can we create a 519 00:30:19,720 --> 00:30:23,840 reasonable way that we as an organization want to think about 520 00:30:23,840 --> 00:30:26,520 what third parties we work with, what intermediates we work with, 521 00:30:26,520 --> 00:30:28,360 what SAS as you point out we work with. 522 00:30:28,840 --> 00:30:31,960 And there's no going to be no magical right choice, but there 523 00:30:31,960 --> 00:30:35,480 needs to be a commitment to some sort of choice and perhaps some 524 00:30:35,480 --> 00:30:38,720 thinking through of why it is that you choose one vendor over 525 00:30:38,720 --> 00:30:42,160 another that relates to principles like privacy and data 526 00:30:42,160 --> 00:30:45,480 security. But then beyond that, if we 527 00:30:45,480 --> 00:30:48,080 don't understand the data flows that are going through our 528 00:30:48,080 --> 00:30:52,400 systems, if we don't have data properly tagged, like some 529 00:30:52,400 --> 00:30:56,080 people might use tagging, other people use categorization, other 530 00:30:56,080 --> 00:30:58,960 people use all sorts of things. But if we don't have a 531 00:30:58,960 --> 00:31:03,120 reasonable way of organizing the quote UN quote, data catalogs 532 00:31:03,120 --> 00:31:07,360 that we have and data stores that we have, then we also don't 533 00:31:07,360 --> 00:31:10,520 have any reasonable way to think through retention schedules. 534 00:31:10,920 --> 00:31:14,840 What is a retention schedule? It means that we say that we 535 00:31:14,840 --> 00:31:19,080 collect data for a certain purpose for a given timeline. 536 00:31:19,960 --> 00:31:23,680 Maybe we collect purchase data for every customer, right? 537 00:31:23,680 --> 00:31:25,480 They go to our site, they buy something. 538 00:31:25,840 --> 00:31:28,080 Or maybe we hold other people's purchase date. 539 00:31:28,080 --> 00:31:31,040 Doesn't matter. The purchase data exists, right? 540 00:31:31,240 --> 00:31:34,640 And maybe we say we retain purchase data for as long as 541 00:31:34,640 --> 00:31:39,120 you're a customer or for up to two years after you close your 542 00:31:39,120 --> 00:31:42,760 account or something like this. Well, how do we even enforce 543 00:31:42,760 --> 00:31:44,680 that? If we haven't bothered to like 544 00:31:44,960 --> 00:31:48,560 connect, then somebody has to write like some really nasty SQL 545 00:31:48,560 --> 00:31:51,000 query. It may or may not work if 546 00:31:51,000 --> 00:31:55,360 somebody did bad data entry. And this all relates to better 547 00:31:55,360 --> 00:31:58,520 data governance, which relates to things like data cataloging, 548 00:31:58,760 --> 00:32:03,760 but also things like thinking through data quality metrics and 549 00:32:03,760 --> 00:32:07,040 how do we ensure that we're actually getting value out of 550 00:32:07,040 --> 00:32:10,840 the data we collect. And what I will say is if you 551 00:32:10,840 --> 00:32:14,000 improve trust relationships with your customers, you will get 552 00:32:14,000 --> 00:32:17,400 better data quality. Like that's by default proven 553 00:32:17,400 --> 00:32:20,800 time and time again. And so if you're just spamming, 554 00:32:21,040 --> 00:32:24,520 collecting whatever you can, especially through some sort of 555 00:32:24,520 --> 00:32:28,120 intermediary, you're probably getting pretty low quality data. 556 00:32:28,480 --> 00:32:32,680 If the signal that you're trying to measure is something like is 557 00:32:32,680 --> 00:32:37,320 this customer interested in ABC and you have a good relationship 558 00:32:37,320 --> 00:32:40,920 with a customer, then you could just directly ask them, right? 559 00:32:41,160 --> 00:32:44,280 And then you have both high data quality of high trust, you've 560 00:32:44,280 --> 00:32:48,080 created a good privacy and security relationship with your 561 00:32:48,080 --> 00:32:51,440 customer, and you know that the data that you're storing is 562 00:32:51,440 --> 00:32:55,160 actually accurate. So I think in the end, these all 563 00:32:55,160 --> 00:32:58,280 end up supporting each other. And I just want to say it is not 564 00:32:58,280 --> 00:33:03,320 a trivial undertaking to do data governance at scale and most 565 00:33:03,320 --> 00:33:06,160 organizations are not doing it very well. 566 00:33:06,160 --> 00:33:09,320 They're just kind of starting right now, I think, for a lot of 567 00:33:09,320 --> 00:33:13,880 organizations. So it's OK to decide, OK, We 568 00:33:13,880 --> 00:33:16,960 want our next few years to be a goal of having better data 569 00:33:16,960 --> 00:33:21,760 governance in which we can also have better privacy and data 570 00:33:21,760 --> 00:33:23,760 quality. Yeah. 571 00:33:23,760 --> 00:33:26,400 So I could imagine companies that have been around for years, 572 00:33:26,400 --> 00:33:29,920 right, having production data serving lots of customers. 573 00:33:30,120 --> 00:33:32,600 Definitely it will be a challenge, right, Especially, 574 00:33:33,000 --> 00:33:34,880 you know, these days also people work remotely. 575 00:33:34,880 --> 00:33:37,640 They work, you know, using Internet, you know, something 576 00:33:37,640 --> 00:33:41,440 got downloaded from the system into the device, sent over to 577 00:33:41,440 --> 00:33:44,560 WhatsApp or something like that. Or you have data pipelines, you 578 00:33:44,560 --> 00:33:47,440 know, data warehouse where data moves from one side to the 579 00:33:47,440 --> 00:33:49,120 other. I think it's really, really 580 00:33:49,120 --> 00:33:50,880 challenging. I would, I would just empathize. 581 00:33:51,040 --> 00:33:54,320 I also have this kind of problem as well to catalog knowing the 582 00:33:54,320 --> 00:33:57,880 lineage, right govern, whether it's sensitive, not sensitive 583 00:33:57,960 --> 00:34:00,680 parts of it, whether it can be shared with many others people 584 00:34:00,680 --> 00:34:02,400 or not. But I think it's definitely a 585 00:34:02,400 --> 00:34:04,320 challenge, but I think it will take some time. 586 00:34:04,400 --> 00:34:07,720 If we build awareness, I'm sure maybe we can improve the data 587 00:34:07,720 --> 00:34:10,560 governance that we have. So maybe let's go into the 588 00:34:10,560 --> 00:34:12,800 techniques. How can we actually improve our 589 00:34:12,800 --> 00:34:15,760 privacy practice, right? I think the first that is always 590 00:34:15,840 --> 00:34:18,639 mentioned is about, you know, pseudonymization or 591 00:34:18,639 --> 00:34:20,639 anonymization or data masking, right? 592 00:34:21,040 --> 00:34:23,840 So tell us maybe for people who are not familiar with these 593 00:34:23,840 --> 00:34:26,199 terms, what are those techniques, right? 594 00:34:26,320 --> 00:34:28,520 And how can we apply that practically? 595 00:34:29,320 --> 00:34:32,480 Yeah. I mean, I want to add a mention 596 00:34:32,480 --> 00:34:36,639 here that is maybe kind of also helps bridge data governance to 597 00:34:36,639 --> 00:34:39,080 like actually implementing what we might call as like 598 00:34:39,080 --> 00:34:43,520 mitigations or controls. And that is, these are not 599 00:34:43,520 --> 00:34:47,800 decisions to be made in a vacuum of like what controls get 600 00:34:47,800 --> 00:34:51,600 applied where or how do we decide what we want to do with 601 00:34:51,600 --> 00:34:53,719 what data? How do we even decide what 602 00:34:54,000 --> 00:34:58,800 sensitivity a data has? Hopefully your organization has 603 00:34:58,800 --> 00:35:04,160 the ability to create a data governance board or practice or 604 00:35:04,160 --> 00:35:08,240 even a data privacy practice where there could be multiple 605 00:35:08,240 --> 00:35:10,600 disciplines of people in the room. 606 00:35:10,680 --> 00:35:13,200 Because I think you have to have people that are business 607 00:35:13,200 --> 00:35:16,360 informed that know what are the business goals of why we're 608 00:35:16,360 --> 00:35:20,440 trying to do this thing. Data informed, software and 609 00:35:20,440 --> 00:35:25,360 tech, tech informed, and then of course, regulatory informed and 610 00:35:25,360 --> 00:35:27,720 privacy informed. So you can also hire privacy 611 00:35:27,720 --> 00:35:31,760 professionals who are not legal experts, but instead focus on 612 00:35:31,760 --> 00:35:33,760 the topic of like privacy by design. 613 00:35:34,040 --> 00:35:36,520 So if you have these people together, maybe also somebody 614 00:35:36,520 --> 00:35:40,440 from Infosec, you can start to say, you know what we think the 615 00:35:40,440 --> 00:35:44,160 biggest challenges for our data governance or our data privacy 616 00:35:44,160 --> 00:35:48,280 or our data security are all three is these top three, some 617 00:35:48,280 --> 00:35:50,120 new systems that we want to develop. 618 00:35:50,520 --> 00:35:53,560 So we're going to put that on a kind of our collective road map 619 00:35:53,560 --> 00:35:56,320 for the year and we're going to prioritize those. 620 00:35:56,840 --> 00:36:00,200 And I think then you start thinking once you can understand 621 00:36:00,200 --> 00:36:03,360 what's the risk space that you have, So what are the biggest 622 00:36:03,360 --> 00:36:07,640 problems that you have with the data privacy, then you can start 623 00:36:07,640 --> 00:36:09,960 thinking about, oh, do we need to student mise? 624 00:36:09,960 --> 00:36:13,520 Do we need to anonymise? Do we need to run some other 625 00:36:13,520 --> 00:36:16,440 type of setup? And then we get into the fun, 626 00:36:16,440 --> 00:36:20,320 fun bits, which I would all fall under kind of these types of 627 00:36:20,320 --> 00:36:24,760 privacy controls or start to go into what we would call privacy 628 00:36:24,760 --> 00:36:28,040 enhancing technologies, which are different technologies that 629 00:36:28,040 --> 00:36:31,600 we can use to help meet these goals of privacy at our 630 00:36:31,600 --> 00:36:35,040 organization. And most of us have seen 631 00:36:35,040 --> 00:36:39,440 pseudonymization at some point in time that can take many 632 00:36:39,440 --> 00:36:42,640 forms. It can be simple masking, like 633 00:36:42,640 --> 00:36:46,200 maybe you've used one of the popular libraries is Microsoft 634 00:36:46,200 --> 00:36:52,600 Presidio to try to identify things like PII and mask them. 635 00:36:53,080 --> 00:36:56,960 Masking to take several forms. You probably already use some 636 00:36:56,960 --> 00:37:01,440 hashing mechanisms or something like this, maybe a one way hash 637 00:37:01,440 --> 00:37:04,120 or maybe a two way like a reversible hash. 638 00:37:04,400 --> 00:37:07,200 These can be ways to pseudomize information. 639 00:37:07,680 --> 00:37:10,960 You've probably already have already used redaction. 640 00:37:11,240 --> 00:37:15,800 So just simply like removing certain types of sensitive data 641 00:37:15,800 --> 00:37:21,520 from let's say reporting tool or ABI dashboard to say, OK, now 642 00:37:21,520 --> 00:37:26,120 this is ready for organization wide consumption or this is 643 00:37:26,120 --> 00:37:28,920 ready for our marketing report or whatever it is. 644 00:37:28,920 --> 00:37:32,440 So like the data is scheduled for release publicly or at least 645 00:37:32,600 --> 00:37:38,200 semi publicly within an org and any of these things are small 646 00:37:38,200 --> 00:37:42,480 privacy mitigations that you do. And then those can lead up to 647 00:37:42,480 --> 00:37:47,680 more serious privacy mitigations like anonymization or even 648 00:37:47,680 --> 00:37:51,160 thinking through things like local first data processing, 649 00:37:51,160 --> 00:37:54,720 distributed data or encrypted computation. 650 00:37:54,720 --> 00:37:58,280 So gets very space aged the further you want to go. 651 00:37:58,600 --> 00:38:01,320 And those only fit certain types of problems right. 652 00:38:01,320 --> 00:38:04,640 So first you have to shape the problem so you can know as every 653 00:38:04,640 --> 00:38:06,160 developer knows it, not to say it. 654 00:38:06,360 --> 00:38:08,760 If you don't know what problem you're trying to do, then just 655 00:38:08,760 --> 00:38:11,840 putting a technology in it is unfortunately not going to solve 656 00:38:11,840 --> 00:38:14,720 the problem. Yeah, that's a very, very good 657 00:38:14,720 --> 00:38:16,760 advice, right. So I like the in the beginning 658 00:38:16,760 --> 00:38:19,680 you mentioned that it's not something one like one 659 00:38:19,680 --> 00:38:22,960 particular team or you know, person to decide, right? 660 00:38:22,960 --> 00:38:25,800 So it's like a multi departmental kind of effort, 661 00:38:25,800 --> 00:38:27,920 right? And there are many departments 662 00:38:27,920 --> 00:38:30,520 that should be involved, right? You mentioned a few that are 663 00:38:30,560 --> 00:38:33,080 very important, right? It could be the privacy, you 664 00:38:33,080 --> 00:38:35,400 know, department, the technology, definitely the 665 00:38:35,400 --> 00:38:38,320 product, right? And maybe the legal side in four 666 00:38:38,320 --> 00:38:40,800 SEC, definitely data and security is kind of like 667 00:38:40,800 --> 00:38:43,080 interrelated with each other and there could be many other 668 00:38:43,080 --> 00:38:44,960 parties, right? And especially in the 669 00:38:45,040 --> 00:38:48,160 organization, I think it's very important to have this so-called 670 00:38:48,160 --> 00:38:50,560 decision making process to actually identify whether 671 00:38:50,560 --> 00:38:52,240 something is sensitive or not sensitive. 672 00:38:52,840 --> 00:38:55,280 So I think during the explanation just now, you 673 00:38:55,280 --> 00:38:58,160 mentioned this term that is quite trendy these days, privacy 674 00:38:58,160 --> 00:39:02,160 enhancing technologies. I think still tech we all love, 675 00:39:02,160 --> 00:39:04,840 you know, to know, like what are the technologies out there so 676 00:39:04,840 --> 00:39:07,160 that we can kind of like apply and try it. 677 00:39:07,360 --> 00:39:10,480 So what are some of the privacy enhancing technologies that are, 678 00:39:10,560 --> 00:39:13,280 you know, like quite modern these days that people are using 679 00:39:13,280 --> 00:39:17,040 to solve this privacy thing? I think there's been some really 680 00:39:17,040 --> 00:39:20,280 cool developments, I would say, over the past 10 years. 681 00:39:20,760 --> 00:39:25,040 And I think the ones I'm most excited about are also the ones 682 00:39:25,040 --> 00:39:28,600 that show up in the book that I know you're now a bit familiar 683 00:39:28,600 --> 00:39:33,200 with, which is technologies like differential privacy, which is a 684 00:39:33,200 --> 00:39:37,280 way that we can reason technically about privacy. 685 00:39:37,600 --> 00:39:42,320 I like to call it for people the idea of measurable privacy or 686 00:39:42,320 --> 00:39:46,040 rigorous privacy, because we're using a scientific way of 687 00:39:46,040 --> 00:39:49,000 thinking about or trying to shape the problem and then 688 00:39:49,000 --> 00:39:53,080 trying to decide that. And that works well for things 689 00:39:53,080 --> 00:39:56,520 like if you have to release data publicly, if you want to think 690 00:39:56,520 --> 00:39:58,560 through something like anonymization. 691 00:39:59,080 --> 00:40:02,320 And it can work in many different types of use cases, 692 00:40:02,320 --> 00:40:06,560 but it can never work if you're trying to say customer A wants 693 00:40:06,560 --> 00:40:09,560 B. It has to work in some form of 694 00:40:09,560 --> 00:40:12,680 aggregation of an idea or a person. 695 00:40:13,120 --> 00:40:16,920 But of course, when we think of anonymization, we cannot ever 696 00:40:16,920 --> 00:40:20,080 say that we release something anonymized that could be tracked 697 00:40:20,080 --> 00:40:23,800 back to one individual, because then it's not anonymized, right, 698 00:40:23,800 --> 00:40:27,000 By definition. So I think differential privacy 699 00:40:27,000 --> 00:40:30,720 is really cool technology, lots of different approaches, lots of 700 00:40:30,720 --> 00:40:34,640 different algorithms to think through, and also an increasing 701 00:40:34,640 --> 00:40:37,480 number of really cool open source libraries that allow you 702 00:40:37,480 --> 00:40:40,600 to think through differential privacy from a development 703 00:40:40,640 --> 00:40:44,200 context. Then probably the next most 704 00:40:44,200 --> 00:40:48,560 famous one of recent years has been Federated learning or 705 00:40:48,560 --> 00:40:52,000 Federated analytics. Sometimes I like to call it more 706 00:40:52,000 --> 00:40:55,840 distributed learning, and it also relates to what I would 707 00:40:55,840 --> 00:40:58,640 call the field of local first software. 708 00:40:59,000 --> 00:41:02,200 I don't know if you've already had local first software experts 709 00:41:02,200 --> 00:41:06,800 on your show, but Martin Klippmann and who's famous for 710 00:41:06,800 --> 00:41:08,720 the book? Nice. 711 00:41:09,880 --> 00:41:11,360 Martin Klippmann's book is very famous. 712 00:41:11,880 --> 00:41:16,080 Sorry. Data in intensive applications. 713 00:41:16,080 --> 00:41:18,760 Something like that. Yes, yes, yes. 714 00:41:18,880 --> 00:41:24,360 Thank you. And he and a group of folks that 715 00:41:24,360 --> 00:41:27,120 have kind of started this movement, I think it probably 716 00:41:27,120 --> 00:41:29,720 started before them, but have popularized this movement of 717 00:41:29,720 --> 00:41:34,120 local first data, which is more thinking through kind of exactly 718 00:41:34,120 --> 00:41:38,520 what we were talking about before, which is if we can now 719 00:41:38,520 --> 00:41:41,720 we have devices, we have edge devices, we also have edge 720 00:41:41,720 --> 00:41:46,680 compute like mobile devices. And certainly within whatever 721 00:41:46,680 --> 00:41:49,560 AWS cluster you run, you have plenty of compute. 722 00:41:50,160 --> 00:41:56,600 So can we push processing as far down to the edge that we need? 723 00:41:56,840 --> 00:42:00,200 And therefore can we push the data as far down to the edge as 724 00:42:00,200 --> 00:42:03,640 we need? So this might mean what we call 725 00:42:03,640 --> 00:42:07,480 cross silo learning or cross silo analytics, which says 726 00:42:07,760 --> 00:42:11,320 you're a multinational company or you're in some sort of B to B 727 00:42:11,320 --> 00:42:15,760 situation and you allow every company to at least house their 728 00:42:15,760 --> 00:42:20,000 data within their own, you know, realm premise, so to speak. 729 00:42:20,000 --> 00:42:22,560 So within their own boundaries of their cloud system or 730 00:42:22,560 --> 00:42:24,000 whatever it is that they're using. 731 00:42:24,360 --> 00:42:28,200 And they don't actually exchange data, they exchange some sort of 732 00:42:28,200 --> 00:42:31,240 analysis or output. So this could be machine 733 00:42:31,240 --> 00:42:33,360 learning processing. This could just be simple 734 00:42:33,360 --> 00:42:36,280 analytics, but we kind of try to push it there. 735 00:42:36,720 --> 00:42:40,160 Or it could go all the way down to I'm building a mobile 736 00:42:40,160 --> 00:42:43,680 application, I'm going to keep all of the data local and I'm 737 00:42:43,680 --> 00:42:46,040 only going to send certain artifacts. 738 00:42:46,080 --> 00:42:49,120 Am I going to push that back up to some sort of centralized 739 00:42:49,120 --> 00:42:52,840 compute or some sort of redundancy compute, Right. 740 00:42:53,160 --> 00:42:56,480 And depending on how that works, you could even offer things like 741 00:42:56,480 --> 00:42:59,280 end to end encryption and then other types of things. 742 00:42:59,720 --> 00:43:03,000 Why would you go to this bother? Will you go to this bother? 743 00:43:03,000 --> 00:43:07,360 Because A, you're greatly reducing your exposure risk, 744 00:43:07,480 --> 00:43:09,200 right? You'd actually don't have 745 00:43:09,200 --> 00:43:12,240 centralized data. You just have some centralized 746 00:43:12,240 --> 00:43:15,440 insights. So how attractive is it to try 747 00:43:15,440 --> 00:43:19,640 to hack your system much less so the less data that you have. 748 00:43:19,920 --> 00:43:23,560 And B, if you can get the same insights, you're also saving a 749 00:43:23,560 --> 00:43:26,920 huge amount on cloud compute because again, if you're like, 750 00:43:27,160 --> 00:43:30,880 if every ping, every two seconds I open it, you're just checking 751 00:43:30,880 --> 00:43:34,680 whether I'm still at my house. Like, cool. 752 00:43:35,000 --> 00:43:37,680 I mean, not so cool. I don't think so, But you know, 753 00:43:37,760 --> 00:43:43,440 you're wasting a whole bunch of network compute and storage to 754 00:43:43,680 --> 00:43:47,840 just find out that, guess what, between the hours of, you know, 755 00:43:47,840 --> 00:43:51,120 9:00 PM and whatever time in the morning, I'm at my house, right? 756 00:43:51,360 --> 00:43:54,440 When in aggregate analytics, he probably could have already 757 00:43:54,440 --> 00:43:57,480 figured out that is, you know, 80% of your user base or 758 00:43:57,480 --> 00:43:59,320 something like this. I don't really know why you need 759 00:43:59,320 --> 00:44:02,280 to know that, but I just give you an example, a rough example 760 00:44:02,800 --> 00:44:06,800 of how differently we could engineer, architect our systems 761 00:44:07,040 --> 00:44:09,680 if we thought through again this question of what days you 762 00:44:09,680 --> 00:44:13,000 actually need. And can I do that in some sort 763 00:44:13,000 --> 00:44:17,280 of privacy respecting manner so that most of the data stays far 764 00:44:17,280 --> 00:44:20,640 away from some sort of centralized setup and certainly 765 00:44:20,640 --> 00:44:24,280 far away from some sort of data sharing setup where companies 766 00:44:24,280 --> 00:44:27,280 are just exchanging sensitive data at scale. 767 00:44:28,040 --> 00:44:31,360 And then that leads to the third type of privacy enhancing 768 00:44:31,360 --> 00:44:33,800 technology I'm personally really excited about, which is 769 00:44:33,800 --> 00:44:37,320 encrypted computation or people might have heard about that 770 00:44:37,320 --> 00:44:41,880 under terms like homomorphic encryption or multi secure multi 771 00:44:41,880 --> 00:44:45,960 party computation. And this essentially enables a 772 00:44:45,960 --> 00:44:50,920 whole bunch of use cases based on cryptography, which allows 773 00:44:50,920 --> 00:44:54,960 you to compute insights on data without decrypting it. 774 00:44:55,360 --> 00:44:58,200 Of course, then eventually somebody will decrypt it. 775 00:44:58,200 --> 00:45:01,840 So the insight gets decrypted and used at some point in time, 776 00:45:02,440 --> 00:45:07,480 but the actual processing of the data can be done in an encrypted 777 00:45:07,480 --> 00:45:10,960 state. And that provides a quality that 778 00:45:10,960 --> 00:45:14,800 we like to call secrecy, which is different than privacy. 779 00:45:15,200 --> 00:45:19,360 But secrecy we can use to say, you know what, in our cloud 780 00:45:19,360 --> 00:45:24,480 compute we only ever computed in encrypted and then we move back 781 00:45:24,480 --> 00:45:28,160 to your device where you personally unencrypted it. 782 00:45:28,400 --> 00:45:33,560 And so by the design that we have offered, we actually never 783 00:45:33,560 --> 00:45:37,600 saw the data in an unencrypted state except once it was on 784 00:45:37,600 --> 00:45:41,280 device or in a data sharing landscape. 785 00:45:41,520 --> 00:45:45,560 We actually only ever process data with another company by 786 00:45:45,560 --> 00:45:49,720 keeping that data encrypted so that neither company could learn 787 00:45:49,720 --> 00:45:52,600 anything other than the final released results. 788 00:45:52,600 --> 00:45:56,720 So any intermediary results, anything that we think might 789 00:45:56,720 --> 00:46:01,280 leak an individual level privacy, we can kind of cover up 790 00:46:01,280 --> 00:46:04,280 with a secrecy blanket. And then at the end, when we 791 00:46:04,280 --> 00:46:08,040 decrypt and we pull off that secrecy blanket, we've already 792 00:46:08,040 --> 00:46:12,680 ensured that the end result has enough privacy that we can 793 00:46:12,680 --> 00:46:16,360 remove the secrecy. I hope that was understandable. 794 00:46:16,600 --> 00:46:20,480 You tell me. Well, I think definitely sounds 795 00:46:20,480 --> 00:46:23,080 really cool, all these advanced technologies, I may not be 796 00:46:23,080 --> 00:46:25,880 exposed to many of those technologies, but it sounds 797 00:46:25,880 --> 00:46:28,440 really promising, right? Especially I think the local 798 00:46:28,440 --> 00:46:31,120 first concept thing will be very important, right? 799 00:46:31,120 --> 00:46:34,280 Especially if you don't want to get exposed to many sensitive 800 00:46:34,280 --> 00:46:38,200 data at your organization. Encryption, I think still kind 801 00:46:38,200 --> 00:46:41,520 of like the go to kind of like strategy for securing the data, 802 00:46:41,520 --> 00:46:43,440 right? It could be encryption at rest, 803 00:46:43,440 --> 00:46:46,160 it could be in transit, right? It could be just like what you 804 00:46:46,160 --> 00:46:48,600 mentioned, maybe just now homomorphic encryption and 805 00:46:48,600 --> 00:46:51,200 things like that. These days cloud provider also 806 00:46:51,200 --> 00:46:54,560 come up with, I don't know, like secure chip, you know, where you 807 00:46:54,560 --> 00:46:58,240 can actually do more secure computing on the virtual 808 00:46:58,240 --> 00:47:00,560 machine. So definitely a space that maybe 809 00:47:00,560 --> 00:47:02,680 for those people who are interested, right, you can 810 00:47:02,960 --> 00:47:06,760 follow the technology's trends. So you mentioned about all this, 811 00:47:06,760 --> 00:47:09,120 right? Definitely 1 aspect for 812 00:47:09,120 --> 00:47:13,000 engineers or product teams, it makes it really, really more 813 00:47:13,000 --> 00:47:16,040 complicated to actually, you know, come up with the solution, 814 00:47:16,040 --> 00:47:18,720 right? How can we actually, you know, 815 00:47:19,120 --> 00:47:21,960 talk to the stakeholders that we need to, you know, improve our 816 00:47:21,960 --> 00:47:24,920 solution by implementing all this complex thing that might 817 00:47:24,920 --> 00:47:27,320 increase the effort and also the amount of resources that we 818 00:47:27,320 --> 00:47:29,480 need, you know, with all these technologies, with the amount of 819 00:47:29,480 --> 00:47:31,200 development effort and things like that. 820 00:47:32,080 --> 00:47:35,160 Yeah, yeah, yeah, yeah. One thing that I think that your 821 00:47:35,160 --> 00:47:39,560 podcast has done well so far is like talk about lean principles, 822 00:47:39,600 --> 00:47:43,600 talk about creating buy in overtime, developing like a 823 00:47:43,600 --> 00:47:47,040 practice and eventually developing platforms, right? 824 00:47:47,040 --> 00:47:49,760 Because I think that's kind of the evolution that you have to 825 00:47:49,760 --> 00:47:53,160 work with. If you're starting with even 826 00:47:53,160 --> 00:47:56,280 just basic data governance, you're not going to tomorrow 827 00:47:56,280 --> 00:47:59,480 deploy homomorphic encryption, like that's not going to happen 828 00:47:59,480 --> 00:48:02,160 for you. So what I think you have to 829 00:48:02,160 --> 00:48:04,600 start with is like you have to start with some of the building 830 00:48:04,600 --> 00:48:07,680 blocks. But try, I think from a kind of 831 00:48:07,680 --> 00:48:12,040 a lean product thinking, try to think through what types of 832 00:48:12,040 --> 00:48:15,320 these more advanced technologies actually fit the problems that 833 00:48:15,320 --> 00:48:17,880 we have. You know, maybe it is exploring 834 00:48:17,880 --> 00:48:21,480 local first development or something and then pulling in 835 00:48:21,600 --> 00:48:26,720 very small use cases, very small product features or maybe even 836 00:48:26,720 --> 00:48:30,480 new product launches and saying, you know what, we're going to 837 00:48:30,800 --> 00:48:35,600 cordon off an extra 20% of the planned development time to 838 00:48:35,600 --> 00:48:39,040 think through an experiment. Could we use this new technology 839 00:48:39,040 --> 00:48:44,600 to help expedite future products so that they are launched in a 840 00:48:44,600 --> 00:48:48,520 private by design or secure by default type of mentality? 841 00:48:49,320 --> 00:48:52,880 And what you're doing there is a, you're giving people who are 842 00:48:52,880 --> 00:48:56,440 interested in the topic enough space so that they can actually 843 00:48:56,440 --> 00:49:00,280 start exploring, right? And as you know from any new 844 00:49:00,360 --> 00:49:03,800 development technology, new type of thinking around development, 845 00:49:04,040 --> 00:49:06,520 if you give people a little bit of space and they already have 846 00:49:06,520 --> 00:49:09,800 some interest or motivation, they can actually go pretty far. 847 00:49:10,080 --> 00:49:12,640 And you're also developing this culture of learning and 848 00:49:12,640 --> 00:49:14,720 experimentation, which we all need. 849 00:49:15,280 --> 00:49:18,680 And then B, you're also like boxing it enough that it doesn't 850 00:49:18,680 --> 00:49:23,360 become a blocker so that there's some sort of backup technology 851 00:49:23,360 --> 00:49:28,000 that you've always used that's there and that you're kind of 852 00:49:28,400 --> 00:49:32,000 giving, you know, you're growing the experience and knowledge. 853 00:49:32,240 --> 00:49:35,960 And I think as soon as you start focusing on the process rather 854 00:49:35,960 --> 00:49:40,360 than the outcomes, then you're going to see the rewards happen 855 00:49:40,360 --> 00:49:42,920 over time. And you asked like, how do we 856 00:49:42,920 --> 00:49:44,680 explain to the business stakeholders? 857 00:49:44,680 --> 00:49:48,440 Well, maybe you don't even yourself have to explain. 858 00:49:48,440 --> 00:49:52,840 It will become evident by allowing time in the process. 859 00:49:52,840 --> 00:49:57,000 For that, you can say as simple as we need to stay up to date. 860 00:49:57,000 --> 00:49:59,960 We need to modernize the way that we think about privacy. 861 00:50:00,400 --> 00:50:04,240 And I think this is going to pay off in the long run by allowing 862 00:50:04,240 --> 00:50:07,960 us to build platforms that allow us to do privacy at scale, which 863 00:50:07,960 --> 00:50:11,360 is the end goal, right? But along that way, we're going 864 00:50:11,360 --> 00:50:15,080 to add some extra time and cushion so that we can build in 865 00:50:15,080 --> 00:50:18,320 these new exciting technologies that I promise you, you will be 866 00:50:18,320 --> 00:50:21,320 able to talk about in some press release one day, right? 867 00:50:21,640 --> 00:50:24,280 And show that we're forward thinking and show that we're 868 00:50:24,320 --> 00:50:28,200 kind of advancing this space. But to start out and allow that. 869 00:50:28,200 --> 00:50:31,960 And I think the advocates that will come out of allowing some 870 00:50:31,960 --> 00:50:35,360 of your developers or your data people or other technologists at 871 00:50:35,360 --> 00:50:39,080 your org to learn more deeply about this stuff is going to pay 872 00:50:39,080 --> 00:50:41,800 off in and of itself. Those people will end up being 873 00:50:41,800 --> 00:50:45,880 the advocates that educate the other people in the org about 874 00:50:45,880 --> 00:50:48,920 how this stuff works. And they will find, I promise 875 00:50:48,920 --> 00:50:52,640 you, they will find clever ways of explaining it if you give 876 00:50:52,640 --> 00:50:54,640 them the time and space to be able to learn. 877 00:50:55,720 --> 00:50:57,480 Yeah. So I think that's a very, you 878 00:50:57,480 --> 00:51:00,160 know, important advice, right? So start small, right? 879 00:51:00,320 --> 00:51:02,880 Focus on the process, not necessarily just the outcome. 880 00:51:03,360 --> 00:51:06,040 And I think everyone can build awareness, I think within the 881 00:51:06,040 --> 00:51:08,840 team, within the organization, right, about the privacy, right, 882 00:51:08,840 --> 00:51:12,440 treats privacy more seriously. Obviously, one easy way to 883 00:51:12,440 --> 00:51:14,920 explain to stakeholders is about regulations, right? 884 00:51:14,920 --> 00:51:18,160 I'm sure in Europe it's easier to actually mention about 885 00:51:18,160 --> 00:51:21,280 privacy because there's the GDPR that is very strict, right? 886 00:51:21,680 --> 00:51:24,560 In some parts of the world, they are starting also to adopt these 887 00:51:24,560 --> 00:51:28,440 kind of stringent rules about personal data law, personal data 888 00:51:28,440 --> 00:51:31,800 protection and things like that. So let's go to that legal 889 00:51:31,800 --> 00:51:34,040 aspect. So what do you see the trends 890 00:51:34,040 --> 00:51:36,760 these days? Especially what I know is that 891 00:51:36,760 --> 00:51:39,640 many countries now starting to come up with this kind of law, 892 00:51:39,640 --> 00:51:42,400 this kind of GDPR like regulations. 893 00:51:42,840 --> 00:51:46,320 So is this something that we are going to see going forward, like 894 00:51:46,320 --> 00:51:49,440 all countries will have their own kind of like regulations? 895 00:51:49,440 --> 00:51:52,960 And I think it will be, it will become very complicated if let's 896 00:51:52,960 --> 00:51:55,560 say you build a product that works in multiple jurisdiction, 897 00:51:55,560 --> 00:51:57,200 right? So maybe tell us a little bit 898 00:51:57,200 --> 00:51:59,200 your insights on this. Yeah. 899 00:51:59,200 --> 00:52:02,160 I mean, I think it's already pretty complicated. 900 00:52:02,760 --> 00:52:05,920 It's probably gonna get more rather than less. 901 00:52:05,920 --> 00:52:10,600 I mean, that's why I think a lot of times when I talk about this, 902 00:52:10,600 --> 00:52:13,000 especially with high level stakeholders, like when we're 903 00:52:13,000 --> 00:52:17,920 talking with C level of some sort of multinational or company 904 00:52:17,920 --> 00:52:21,080 that wants to become multinational, we have to think 905 00:52:21,080 --> 00:52:25,000 about it as like future proofing the data strategy, right? 906 00:52:25,000 --> 00:52:30,720 Because if we just kind of developed data or AI strategy 907 00:52:31,200 --> 00:52:35,480 within what we know today, we're not going to be prepared for the 908 00:52:35,480 --> 00:52:38,240 future. And even with what we know today 909 00:52:38,240 --> 00:52:43,000 is we have a very fragmented legal jurisdiction set up around 910 00:52:43,000 --> 00:52:45,800 almost everything data, but certainly data privacy. 911 00:52:46,040 --> 00:52:48,200 But that also applies to data security. 912 00:52:48,440 --> 00:52:51,160 It also applies to data governance protections that 913 00:52:51,160 --> 00:52:53,000 different organizations are looking at. 914 00:52:53,400 --> 00:52:56,200 And one of the big trends that unfortunately I don't see going 915 00:52:56,200 --> 00:53:00,560 away anytime soon is the data sovereignty laws as well. 916 00:53:01,200 --> 00:53:04,720 And these data sovereignty laws are kind of around the 917 00:53:04,720 --> 00:53:06,880 jurisdictional control of the data. 918 00:53:06,880 --> 00:53:11,000 So for example, that data cannot leave a certain nation state or 919 00:53:11,000 --> 00:53:15,680 a group of nation states or that data cannot travel across these 920 00:53:15,680 --> 00:53:20,240 untrusted nation states. And if you already work in the 921 00:53:20,240 --> 00:53:23,040 government or you work in government adjacent, you 922 00:53:23,040 --> 00:53:25,680 probably already know this pain and you're like, you're not 923 00:53:25,680 --> 00:53:28,880 saying anything new, Catherine. Well, now the pain is coming to 924 00:53:28,880 --> 00:53:32,480 you, whether or not you're government adjacent, whether or 925 00:53:32,480 --> 00:53:36,240 not you deal with kind of highly secure systems or critical 926 00:53:36,240 --> 00:53:39,480 infrastructure. I think it's kind of ballooning 927 00:53:39,480 --> 00:53:42,960 out to other areas. And again, at the end of the 928 00:53:42,960 --> 00:53:46,480 day, a lot of this is a policy problem internally at an 929 00:53:46,480 --> 00:53:50,640 organization to decide, OK, what's our stance towards 930 00:53:50,640 --> 00:53:53,160 regulation? Where do we exist? 931 00:53:53,320 --> 00:53:57,120 Like what's important to us? And then obviously the legal 932 00:53:57,120 --> 00:54:00,760 experts, not the technology experts, get to decide, OK, 933 00:54:00,760 --> 00:54:04,760 here's our risk posture. Here's like the policies that we 934 00:54:04,760 --> 00:54:08,360 want to see related to what we know about the data, how the 935 00:54:08,360 --> 00:54:11,120 data stored, where it's stored, what is used for. 936 00:54:11,560 --> 00:54:15,760 And then we kind of get to step in as technologists at kind of 937 00:54:15,760 --> 00:54:21,760 like that Olicy and principles level, I would say and say this 938 00:54:21,760 --> 00:54:23,560 is possible, this is not possible. 939 00:54:23,560 --> 00:54:26,400 This is like a good idea. This is not such a good idea 940 00:54:26,400 --> 00:54:30,360 because and then we can talk about things like architectures, 941 00:54:30,960 --> 00:54:33,840 clouds that we use, infrastructure questions 942 00:54:34,080 --> 00:54:36,960 alongside cool stuff like privacy technologies. 943 00:54:36,960 --> 00:54:41,000 And I would say there's a non trivial number of legal 944 00:54:41,000 --> 00:54:44,600 professionals and privacy professionals that want to also 945 00:54:44,600 --> 00:54:48,000 learn more about technology or might even already know a lot 946 00:54:48,000 --> 00:54:50,760 about technology. And you if you build those 947 00:54:50,760 --> 00:54:55,640 bridges, you can have a really useful conversation around what 948 00:54:55,640 --> 00:54:59,520 is this actually look like. And yeah, I think that can be 949 00:55:00,000 --> 00:55:02,560 probably the most healthy approach rather than just 950 00:55:02,560 --> 00:55:05,320 avoiding the topic because it's scary and you don't want to talk 951 00:55:05,320 --> 00:55:07,160 with the lawyers because they might say no. 952 00:55:07,520 --> 00:55:10,320 And then five years down the line they're like, why do we 953 00:55:10,320 --> 00:55:14,440 architect it this way? And there's like no answer other 954 00:55:14,440 --> 00:55:16,680 than we didn't have the conversation early enough. 955 00:55:17,560 --> 00:55:20,600 Yeah, I think you mentioned something really, really, you 956 00:55:20,600 --> 00:55:23,520 know, very important, right to have now probably legal team, 957 00:55:23,520 --> 00:55:25,640 legal aspect when you build the products, right? 958 00:55:25,640 --> 00:55:28,440 When you architect the solution, when you architect your data 959 00:55:28,440 --> 00:55:31,680 storage and things like that, I think we are used to discuss 960 00:55:31,680 --> 00:55:34,400 about whenever we choose a cloud provider, right? 961 00:55:34,400 --> 00:55:36,480 It's the data center in the particular country. 962 00:55:36,480 --> 00:55:37,880 Now I think that's not enough, right? 963 00:55:37,880 --> 00:55:41,520 Because the data transfer, you have to also look at like if 964 00:55:41,520 --> 00:55:44,400 let's say you use a SAS provider that resides in another country, 965 00:55:44,400 --> 00:55:48,200 another jurisdiction, maybe the data sovereignty law doesn't 966 00:55:48,200 --> 00:55:50,760 allow you to do that, right? And maybe you can have a 967 00:55:50,760 --> 00:55:52,760 trouble, right, when transferring data over. 968 00:55:53,040 --> 00:55:56,640 So I think a lot of countries also implement this differently. 969 00:55:56,960 --> 00:55:58,560 I think it's very hard to keep up. 970 00:55:58,600 --> 00:56:01,040 And that's why probably the legal team is also still the 971 00:56:01,040 --> 00:56:04,840 best person to kind of like help advise the team to actually how 972 00:56:05,000 --> 00:56:07,600 to implement the better solution about this privacy. 973 00:56:08,040 --> 00:56:11,960 So Catherine, we have spoken a lot about privacy technologies 974 00:56:11,960 --> 00:56:14,040 and all that. Is there something else that you 975 00:56:14,040 --> 00:56:16,840 think the listeners here should know about data privacy? 976 00:56:17,720 --> 00:56:21,480 I think lately I get a lot of questions around AI and data 977 00:56:21,480 --> 00:56:25,480 privacy. So I just wanted to note that 978 00:56:25,480 --> 00:56:28,000 there's a lot of new developments and thinking 979 00:56:28,000 --> 00:56:32,960 through, A, how do we build machine learning or AI systems 980 00:56:32,960 --> 00:56:37,400 that can utilize private data without problems? 981 00:56:37,400 --> 00:56:41,640 But B, how do we, if we don't build the AI models ourselves, 982 00:56:41,640 --> 00:56:45,040 of which most organizations still don't at this point in 983 00:56:45,040 --> 00:56:49,520 time, how do we actually build kind of a protective blanket 984 00:56:49,520 --> 00:56:54,360 around these interfaces to, again, these like third party AI 985 00:56:54,360 --> 00:56:59,640 systems or AIAPIS? And I just want to suggest out 986 00:56:59,640 --> 00:57:02,560 there, obviously, I think my book is a great resource. 987 00:57:03,240 --> 00:57:04,920 I also have a newsletter on the topic. 988 00:57:04,920 --> 00:57:09,080 But to just kind of start to inform yourself, because what 989 00:57:09,080 --> 00:57:13,080 I'm seeing in the circles that I'm in is that this is a growing 990 00:57:13,080 --> 00:57:18,560 trend and a growing problem. And I think that it will come. 991 00:57:19,400 --> 00:57:21,800 It will come for everyone. It will come for people that are 992 00:57:21,800 --> 00:57:25,360 not machine learning experts. Soon enough. 993 00:57:25,360 --> 00:57:28,480 If it's not already a problem that you're facing where you're 994 00:57:28,480 --> 00:57:32,440 sitting there with some other team and you want to use a third 995 00:57:32,440 --> 00:57:36,040 party API and all of a sudden this questionnaire comes in from 996 00:57:36,040 --> 00:57:38,720 the legal department or how is it going to look? 997 00:57:38,720 --> 00:57:41,960 Or how do you actually build that with privacy or with 998 00:57:41,960 --> 00:57:44,440 security, right? If you're using proprietary 999 00:57:44,440 --> 00:57:49,480 documents, let's say you want to build like a rag system and you 1000 00:57:49,480 --> 00:57:52,000 want to have access to a bunch of sensitive documents that the 1001 00:57:52,000 --> 00:57:55,240 company has, like how do we make sure that we build that 1002 00:57:55,240 --> 00:57:59,600 responsibly? So I just want to point to the 1003 00:57:59,600 --> 00:58:04,000 fact that this is like a growing field of practice and you don't 1004 00:58:04,000 --> 00:58:08,280 have to learn it like to the greatest depth, but to just kind 1005 00:58:08,280 --> 00:58:13,280 of start to, when it pleases you, start to inform yourself 1006 00:58:13,280 --> 00:58:17,160 lightly like, hmm, like what might that look like if we built 1007 00:58:17,160 --> 00:58:20,240 this rag system with privacy or security? 1008 00:58:20,480 --> 00:58:24,240 How could that look different? Do we have the capabilities to 1009 00:58:24,240 --> 00:58:26,840 do that now? If not, what capabilities would 1010 00:58:26,840 --> 00:58:31,120 we need to grow and to just, you know, your tech leads, right to 1011 00:58:31,120 --> 00:58:36,160 just start to allow that topic to percolate in your head a bit 1012 00:58:36,160 --> 00:58:40,280 because I think it's a growing importance, I think in how we 1013 00:58:40,280 --> 00:58:44,200 think about integrating machine learning into normal work flows. 1014 00:58:45,080 --> 00:58:48,120 Yeah, I could imagine many teams now are kind of like scrambling 1015 00:58:48,120 --> 00:58:51,040 to implement some sort of AI into their systems, right? 1016 00:58:51,040 --> 00:58:54,720 And especially when you do that, sometimes we don't think further 1017 00:58:54,720 --> 00:58:56,640 ahead, right? Especially if you're dealing 1018 00:58:56,640 --> 00:58:58,480 with, I don't know, like customer support, you know, 1019 00:58:58,480 --> 00:59:01,640 customer confidential data. You think just by implementing 1020 00:59:01,640 --> 00:59:04,560 AI it's going to make life easier, but you are not sure 1021 00:59:04,560 --> 00:59:06,120 that they will expose certain things. 1022 00:59:06,440 --> 00:59:09,800 Also, when you use some third party systems these days, you 1023 00:59:09,800 --> 00:59:13,600 know, when, whenever these companies adopt AI, sometimes 1024 00:59:13,800 --> 00:59:17,240 the choice by default is to actually enable AI training, 1025 00:59:17,240 --> 00:59:19,520 right, using your data sometimes. 1026 00:59:20,040 --> 00:59:22,520 So be aware of that. And also like, I think those 1027 00:59:22,520 --> 00:59:25,080 things need to be more transparent and explicit so that 1028 00:59:25,080 --> 00:59:27,800 we can build trust with the third party, the systems that we 1029 00:59:27,800 --> 00:59:29,840 use. So Catherine, I think it's been 1030 00:59:29,840 --> 00:59:32,240 a great conversation. I learned a lot about data 1031 00:59:32,240 --> 00:59:35,760 privacy, although it's kind of scares me a little bit on how to 1032 00:59:35,760 --> 00:59:38,720 build a more private by design system. 1033 00:59:39,680 --> 00:59:42,080 Unfortunately, we reached the end of our conversation. 1034 00:59:42,280 --> 00:59:44,880 So one thing I would like to ask you to end the conversation is 1035 00:59:44,880 --> 00:59:47,000 what I call the three technical leadership system. 1036 00:59:47,320 --> 00:59:49,520 So you can think of them just like an advice. 1037 00:59:49,800 --> 00:59:52,560 So maybe if you can share your version to the listeners here. 1038 00:59:53,360 --> 00:59:57,080 Yeah, I mean, I hope maybe I've sparked some people who might be 1039 00:59:57,080 --> 00:59:59,800 interested in learning more about data privacy, data 1040 00:59:59,800 --> 01:00:03,040 security topics. But the number one piece of 1041 01:00:03,040 --> 01:00:06,000 advice is like, even if you're not inspired to learn. 1042 01:00:06,440 --> 01:00:10,080 I ask that you empower people on your team or in your 1043 01:00:10,080 --> 01:00:14,120 organization to learn and that's because as you can probably tell 1044 01:00:14,120 --> 01:00:20,400 from my advice today, the field is actually quite deep and very 1045 01:00:20,400 --> 01:00:24,480 intensive and it's not necessarily a field that a lot 1046 01:00:24,480 --> 01:00:28,680 of people specialize in because they studied it right. 1047 01:00:28,680 --> 01:00:33,120 So you going to want to open doors for people to grow and to 1048 01:00:33,120 --> 01:00:36,640 learn and to foster kind of that ability as a specialized, my 1049 01:00:36,640 --> 01:00:41,240 culture of learning and for you all as tech leads, like if it's 1050 01:00:41,240 --> 01:00:44,320 coming from you, it's going to be much more powerful than like 1051 01:00:44,520 --> 01:00:47,680 a junior who just joined the team and who's like, oh, wow, 1052 01:00:47,680 --> 01:00:49,320 cool. Like I learned about 1053 01:00:49,320 --> 01:00:51,880 differential privacy in college, because sometimes they do that 1054 01:00:51,880 --> 01:00:54,480 Dow like I want to try out this thing. 1055 01:00:54,480 --> 01:00:57,400 Like if you can give that person space, if if you can kind of 1056 01:00:57,400 --> 01:01:00,400 shield them and give them some ability to learn, they're going 1057 01:01:00,400 --> 01:01:03,960 to level up the entire org eventually, right? 1058 01:01:03,960 --> 01:01:07,480 And so certainly level up your team, but eventually this kind 1059 01:01:07,480 --> 01:01:11,000 of spreads and creates these cultural changes that you need. 1060 01:01:11,360 --> 01:01:15,120 So foster that learning, you know, support your people and 1061 01:01:15,120 --> 01:01:17,480 learning. Create those spaces for 1062 01:01:17,480 --> 01:01:20,200 learning. Try to give some back pressure 1063 01:01:20,200 --> 01:01:24,280 to delivery deadlines to allow for some of that learning 1064 01:01:24,280 --> 01:01:27,000 cushion. Another thing that I want to 1065 01:01:27,000 --> 01:01:32,280 share is I don't think all risk conversations have to be scary. 1066 01:01:33,320 --> 01:01:37,280 I think it's totally natural. And like I, I empathize so much 1067 01:01:37,600 --> 01:01:42,680 with it being scary because it is a big job and it is like a 1068 01:01:42,680 --> 01:01:46,560 serious job to talk about the risk of people's data and that 1069 01:01:46,560 --> 01:01:50,280 trust relationship. And yet at the same time, if we 1070 01:01:50,280 --> 01:01:55,080 can normalize conversations around risk, if we can kind of 1071 01:01:55,080 --> 01:02:00,160 allow for some of that space to have like scary topics, be part 1072 01:02:00,160 --> 01:02:03,120 of our Rd. mapping, be part of our planning, where we can 1073 01:02:03,120 --> 01:02:06,240 actually kind of just regularly do privacy risk reviews, 1074 01:02:06,240 --> 01:02:09,280 security risk reviews, auditing reviews. 1075 01:02:09,320 --> 01:02:12,760 If we can kind of normalize that as a normal thing, like we do 1076 01:02:12,760 --> 01:02:16,680 with like testing for bugs or figuring out if something is 1077 01:02:16,680 --> 01:02:21,120 deployed properly, then we also make it like more tangible. 1078 01:02:21,120 --> 01:02:24,880 And I think that it becomes less scary by default and we make it 1079 01:02:24,880 --> 01:02:27,040 less uncertain. And I think most of the 1080 01:02:27,040 --> 01:02:29,880 scariness comes from this uncertainty problem. 1081 01:02:30,320 --> 01:02:32,640 So to try to focus on that, right? 1082 01:02:33,320 --> 01:02:36,080 And then finally, like that relates to both of these relate 1083 01:02:36,080 --> 01:02:39,280 to like creating a team culture and hopefully an organization 1084 01:02:39,280 --> 01:02:44,040 culture of psychological safety. And I think by knowing that 1085 01:02:44,040 --> 01:02:46,440 we're not going to get everything 100% right all the 1086 01:02:46,440 --> 01:02:50,480 time, we can foster both learning, we can foster risk 1087 01:02:50,480 --> 01:02:53,160 discussions. And we can certainly build much, 1088 01:02:53,160 --> 01:02:57,080 much more private and secure systems by allowing us to have 1089 01:02:57,080 --> 01:03:01,080 real conversations, by allowing us to like escalate problems 1090 01:03:01,080 --> 01:03:05,560 when they exist, when we see them, and by empowering people 1091 01:03:05,560 --> 01:03:09,800 to kind of have reporting up and down about what do we think 1092 01:03:09,800 --> 01:03:11,880 about problems like privacy and security. 1093 01:03:11,880 --> 01:03:15,400 So up the hierarchical chain of the Oregon down. 1094 01:03:15,720 --> 01:03:19,840 And by fostering this kind of safety conversation around these 1095 01:03:19,840 --> 01:03:23,800 types of problems, I think we end up creating by default 1096 01:03:23,800 --> 01:03:27,000 better software systems products in terms of privacy and 1097 01:03:27,000 --> 01:03:29,400 security. Wow, very lovely. 1098 01:03:29,400 --> 01:03:31,480 And you know, it all interrelated with each other, 1099 01:03:31,480 --> 01:03:32,960 right? I especially love the second 1100 01:03:32,960 --> 01:03:35,200 one. You know, not all risk needs to 1101 01:03:35,200 --> 01:03:37,640 be scary, right? Sometimes it just because of the 1102 01:03:37,640 --> 01:03:40,600 uncertainty, like we don't know about that particular risk. 1103 01:03:40,960 --> 01:03:43,680 But I think if you dive deep, if you learn right and you allow 1104 01:03:43,680 --> 01:03:46,840 people to experiment and try to solve the problem, psychological 1105 01:03:46,840 --> 01:03:49,480 safety here. So I think maybe we can all 1106 01:03:49,480 --> 01:03:51,760 improve together. So thanks for that beautiful 1107 01:03:51,760 --> 01:03:54,200 wisdom. So, Catherine, for people who 1108 01:03:54,200 --> 01:03:56,680 love to talk more about privacy, is there a place where they can 1109 01:03:56,680 --> 01:03:59,600 find you online or maybe resources where they can learn 1110 01:03:59,600 --> 01:04:03,360 more about privacy? Yeah, I mean, I have a 1111 01:04:03,360 --> 01:04:05,560 newsletter, it's called probably Private. 1112 01:04:05,560 --> 01:04:08,280 So if you're more of like a e-mail newsletter person, you 1113 01:04:08,280 --> 01:04:11,480 can find me there. I'm starting kind of to produce 1114 01:04:11,480 --> 01:04:13,800 also some YouTube videos. They're mainly around machine 1115 01:04:13,800 --> 01:04:16,560 learning systems and privacy and machine learning systems. 1116 01:04:16,960 --> 01:04:21,160 So you can find me there. My book is mainly written for 1117 01:04:21,200 --> 01:04:25,320 data people, but I hope or I heard some feedback from also 1118 01:04:25,320 --> 01:04:29,520 software and architects that it can be useful parts of my book. 1119 01:04:29,920 --> 01:04:34,520 And then there's such an amazing wealth of resources online 1120 01:04:34,520 --> 01:04:37,720 around privacy. I would even like even just 1121 01:04:37,720 --> 01:04:41,240 start with your cloud provider, like your main cloud provider or 1122 01:04:41,240 --> 01:04:43,720 a few of your main service providers if you have major 1123 01:04:43,720 --> 01:04:47,960 services and just start looking around of like what settings do 1124 01:04:47,960 --> 01:04:51,280 they have around privacy? What do they make available to 1125 01:04:51,280 --> 01:04:53,200 you? What knobs can you turn? 1126 01:04:53,960 --> 01:04:56,960 Even just starting there or even having like somebody on your 1127 01:04:56,960 --> 01:04:59,640 team, like the art, the cloud architecture, cloud engineer on 1128 01:04:59,640 --> 01:05:04,000 your team focus on that could end up paying dividends later of 1129 01:05:04,000 --> 01:05:07,240 just starting to know, hey, it'd be really easy for us to turn 1130 01:05:07,240 --> 01:05:09,560 on, you know, this or turn off this. 1131 01:05:09,840 --> 01:05:12,240 And that would provide better privacy. 1132 01:05:12,240 --> 01:05:15,040 So I think those are some easy ways to get started and make 1133 01:05:15,040 --> 01:05:18,000 friends with privacy people at your org. 1134 01:05:18,400 --> 01:05:21,480 Be nice to them. Just reach out, set up a 1 to 1135 01:05:21,480 --> 01:05:27,360 115 minute coffee because they could be huge resources if you 1136 01:05:27,360 --> 01:05:29,520 develop that or maybe you already have the relationship, 1137 01:05:29,520 --> 01:05:32,240 but if you can develop that relationship, that can be a 1138 01:05:32,240 --> 01:05:35,800 quick person that you can just double check your thinking with 1139 01:05:35,800 --> 01:05:39,120 or think out loud with private, if they have a legal background, 1140 01:05:39,120 --> 01:05:41,280 they're never going to say yes or no, which is fine. 1141 01:05:41,360 --> 01:05:44,560 That's what they're trained to do, good job for them, but 1142 01:05:44,560 --> 01:05:46,480 they're going to give you advice, they're going to give 1143 01:05:46,480 --> 01:05:48,640 you steering ideas, they're going to give you guiding 1144 01:05:48,640 --> 01:05:51,920 questions is going to help you. So be friendly with them. 1145 01:05:52,680 --> 01:05:54,880 Well, thanks for the plug. I'm sure if there are legal 1146 01:05:54,880 --> 01:05:57,280 people listening to this, they will feel happy. 1147 01:05:58,160 --> 01:06:01,280 So I think, yeah, make friends with the legal compliance team, 1148 01:06:01,280 --> 01:06:04,160 security team, right? So those people not there to 1149 01:06:04,160 --> 01:06:07,040 make the job harder for you, but actually they help you and the 1150 01:06:07,040 --> 01:06:09,840 organization to improve. So, Catherine, I love this 1151 01:06:09,840 --> 01:06:11,400 conversation. Thank you so much for your time. 1152 01:06:11,400 --> 01:06:14,400 I hope the listeners here learn a lot about the data privacy 1153 01:06:14,400 --> 01:06:15,680 today. So thanks again. 1154 01:06:16,160 --> 01:06:17,040 Thanks so much, Henry.