WEBVTT 00:00:00.000 --> 00:00:02.960 Welcome to The Bait. Good to be here. You know, 00:00:03.000 --> 00:00:06.919 there's a very specific visceral feeling that 00:00:06.919 --> 00:00:09.380 I think we all get when we open a certain app 00:00:09.380 --> 00:00:13.439 on our phones. It's this mix of professional 00:00:13.439 --> 00:00:17.219 obligation, a little bit of anxiety, and let's 00:00:17.219 --> 00:00:19.960 be honest, a significant amount of cringe. We 00:00:19.960 --> 00:00:22.100 are talking about the only social network that 00:00:22.100 --> 00:00:24.420 you essentially have to be on if you want to 00:00:24.420 --> 00:00:26.559 participate in the modern white -collar economy. 00:00:27.079 --> 00:00:37.579 We are, of course, talking about LinkedIn. That 00:00:37.579 --> 00:00:41.060 is the perfect analogy. But, you know, to simply 00:00:41.060 --> 00:00:43.280 call it a social network almost feels like calling 00:00:43.280 --> 00:00:45.960 a skyscraper a house. It's just structurally 00:00:45.960 --> 00:00:48.799 different. I mean, since Microsoft acquired the 00:00:48.799 --> 00:00:52.560 platform for $26 .2 billion back in 2016, it 00:00:52.560 --> 00:00:56.250 has solidified itself as the... Well, the dominant 00:00:56.250 --> 00:00:58.689 global infrastructure for professional life. 00:00:59.270 --> 00:01:02.130 Dominant is certainly the right word. We're looking 00:01:02.130 --> 00:01:05.790 at a platform that, as of November 2023, claimed 00:01:05.790 --> 00:01:09.370 over a billion members. It's a behemoth. It generated 00:01:09.370 --> 00:01:13.810 $17 .8 billion in revenue in 2025. But I think 00:01:13.810 --> 00:01:16.230 we need to look past the financial success and 00:01:16.230 --> 00:01:18.650 ask what this dominance actually does to the 00:01:18.650 --> 00:01:20.730 workforce. The question isn't whether LinkedIn 00:01:20.730 --> 00:01:23.430 is successful for Microsoft. The question is, 00:01:23.680 --> 00:01:26.200 What is this dominance doing to us, to the way 00:01:26.200 --> 00:01:28.340 we work, and in the way we present ourselves? 00:01:28.680 --> 00:01:31.079 Exactly. And that brings us to the core tension 00:01:31.079 --> 00:01:34.359 we're debating today. Is LinkedIn a democratizing 00:01:34.359 --> 00:01:38.219 utility that creates transparency and efficiency 00:01:38.219 --> 00:01:41.659 in the global labor market? Or, as I suspect 00:01:41.659 --> 00:01:44.079 you'll argue, is it a discriminatory gatekeeper 00:01:44.079 --> 00:01:47.319 that commodifies professional identity and actually 00:01:47.319 --> 00:01:50.599 exacerbates inequality? That is the question. 00:01:51.340 --> 00:01:54.319 I'm taking the position that LinkedIn acts as 00:01:54.319 --> 00:01:56.739 a mechanism of surveillance and stratification. 00:01:57.060 --> 00:01:59.980 I mean, through algorithmic filtering and corporate 00:01:59.980 --> 00:02:03.099 compliance, specifically around censorship and 00:02:03.099 --> 00:02:06.400 privacy, it fundamentally undermines the interests 00:02:06.400 --> 00:02:09.539 of the actual workers it claims to serve. It 00:02:09.539 --> 00:02:11.879 masquerades as a community, but it functions 00:02:11.879 --> 00:02:14.240 like a marketplace where the workers are the 00:02:14.240 --> 00:02:17.830 inventory. And I, of course, am taking the opposing 00:02:17.830 --> 00:02:21.090 view. I believe LinkedIn is a vital engine for 00:02:21.090 --> 00:02:23.650 economic opportunity. By efficiently connecting 00:02:23.650 --> 00:02:26.629 skills with needs through its economic graph 00:02:26.629 --> 00:02:29.650 and allowing individuals to, you know, control 00:02:29.650 --> 00:02:32.009 their professional narrative, it reduces the 00:02:32.009 --> 00:02:34.030 friction that has historically plagued the job 00:02:34.030 --> 00:02:36.990 market. It's about visibility. Well, let's get 00:02:36.990 --> 00:02:40.169 into it. Why don't you lay out your case? Why 00:02:40.169 --> 00:02:42.729 is this corporate behemoth actually the hero 00:02:42.729 --> 00:02:45.539 of the story? Gladly. My argument really rests 00:02:45.539 --> 00:02:48.020 on the evolution of the platform. You have to 00:02:48.020 --> 00:02:50.539 remember what job hunting used to be. It was 00:02:50.539 --> 00:02:53.180 the old boys club. It was who you know in the 00:02:53.180 --> 00:02:56.240 most literal exclusionary sense. Or, you know, 00:02:56.240 --> 00:02:58.319 it was sending a paper resume into a black hole 00:02:58.319 --> 00:03:00.939 and just praying someone read it. LinkedIn changed 00:03:00.939 --> 00:03:04.500 that by digitizing the reputation economy. The 00:03:04.500 --> 00:03:07.199 guiding vision here is what former CEO Jeff Weiner 00:03:07.199 --> 00:03:10.639 called the economic graph. That always sounded 00:03:10.639 --> 00:03:13.530 like a marketing buzzword to me. It sounds like 00:03:13.530 --> 00:03:15.770 one, I'll grant you, but the ambition is actually 00:03:15.770 --> 00:03:19.050 staggering. The goal was to map every job, every 00:03:19.050 --> 00:03:21.550 skill, every company, and every professional 00:03:21.550 --> 00:03:24.090 globally. The idea is that if you can map the 00:03:24.090 --> 00:03:26.110 economy digitally, you can make it more efficient. 00:03:26.189 --> 00:03:28.729 You can remove the geographic and social barriers. 00:03:28.969 --> 00:03:32.289 You replace the country club handshake with a 00:03:32.289 --> 00:03:34.969 searchable, data -driven marketplace. That sounds 00:03:34.969 --> 00:03:38.069 very noble in theory. A perfect map of human 00:03:38.069 --> 00:03:41.300 capital. It is noble in practice, too. I mean, 00:03:41.319 --> 00:03:44.060 look at the utility of features like Open for 00:03:44.060 --> 00:03:46.939 Business, which launched globally in 2019. This 00:03:46.939 --> 00:03:48.960 allowed freelancers and small business owners 00:03:48.960 --> 00:03:51.139 to be discovered by clients they never would 00:03:51.139 --> 00:03:53.919 have met otherwise. Or consider the Apply with 00:03:53.919 --> 00:03:56.620 LinkedIn plugin, launched way back in 2011. It 00:03:56.620 --> 00:03:59.139 standardized the digital resume, letting potential 00:03:59.139 --> 00:04:01.479 employees apply for positions using their profile 00:04:01.479 --> 00:04:04.199 data instantly. It empowers personal branding. 00:04:04.439 --> 00:04:07.199 As the expert Sandra Long has noted, the platform 00:04:07.199 --> 00:04:09.699 allows users to actively manage their unique 00:04:09.699 --> 00:04:12.020 value and position themselves to be found by 00:04:12.020 --> 00:04:14.280 search engines. It puts the control of the career 00:04:14.280 --> 00:04:15.960 narrative back in the hands of the individual. 00:04:16.519 --> 00:04:19.579 I see why you think that, but let me offer a 00:04:19.579 --> 00:04:22.740 different perspective. You call it personal branding, 00:04:22.860 --> 00:04:25.600 but I call it coerced performance. The platform 00:04:25.600 --> 00:04:28.480 forces workers into this, this performative, 00:04:28.480 --> 00:04:30.920 public -facing persona that benefits recruiters 00:04:30.920 --> 00:04:33.980 far more than it benefits job seekers. It creates 00:04:33.980 --> 00:04:36.300 a dynamic where you aren't a person, you are 00:04:36.300 --> 00:04:39.420 a bundle of keywords to be optimized. This leads 00:04:39.420 --> 00:04:41.860 to what sociologist Ofer Sharon calls the filtration 00:04:41.860 --> 00:04:45.500 effect. The filtration effect. Yes. It's where 00:04:45.500 --> 00:04:48.699 merit takes a backseat to how well you can game 00:04:48.699 --> 00:04:52.019 the profile optimization algorithms. If you don't 00:04:52.019 --> 00:04:54.139 use the specific buzzwords that the algorithm 00:04:54.139 --> 00:04:56.860 favors, you are, for all intents and purposes, 00:04:57.100 --> 00:04:59.220 invisible. You're filtered out before a human 00:04:59.220 --> 00:05:01.920 ever sees you. It forces everyone into this hyper 00:05:01.920 --> 00:05:05.180 -positive, crushing -it persona that feels incredibly 00:05:05.180 --> 00:05:08.180 artificial. I think gaming is a harsh word for 00:05:08.180 --> 00:05:11.560 optimizing. Clarity is important. I'm sorry, 00:05:11.620 --> 00:05:13.920 but I just don't buy that. Let me tell you why. 00:05:14.379 --> 00:05:17.220 The efficiency you praise comes at a massive 00:05:17.220 --> 00:05:20.899 cost to privacy and fairness. This company has 00:05:20.899 --> 00:05:23.459 a history of aggressive data practices that treat 00:05:23.459 --> 00:05:26.560 user connections as corporate property. Do you 00:05:26.560 --> 00:05:29.399 remember the Perkins lawsuit? Vaguely. That was 00:05:29.399 --> 00:05:32.310 about emails, correct? It was a hijacking. LinkedIn 00:05:32.310 --> 00:05:35.610 settled for $13 million because they were accessing 00:05:35.610 --> 00:05:38.709 user email address books to send spam invitations 00:05:38.709 --> 00:05:40.990 that look like they came personally from the 00:05:40.990 --> 00:05:44.810 user. Friend X wants you to join LinkedIn. They 00:05:44.810 --> 00:05:47.389 were exploiting personal relationships just to 00:05:47.389 --> 00:05:50.389 grow their network. That's not empowering. That's 00:05:50.389 --> 00:05:53.149 exploitation. And we can't ignore the monopoly 00:05:53.149 --> 00:05:55.750 status because they're the only game in town 00:05:55.750 --> 00:05:58.850 they can enforce arbitrary standards. Look at 00:05:58.850 --> 00:06:01.449 the quiet removal of the diversity, equity, and 00:06:01.449 --> 00:06:05.089 inclusion pages in January 2025, or the removal 00:06:05.089 --> 00:06:07.470 of protections for transgender users regarding 00:06:07.470 --> 00:06:10.029 misgendering and deadnaming later that year. 00:06:10.170 --> 00:06:13.129 It proves they prioritize corporate comfort and, 00:06:13.170 --> 00:06:16.069 frankly, political expediency over user inclusion. 00:06:16.430 --> 00:06:18.370 I want to circle back to the filtration effect 00:06:18.370 --> 00:06:20.730 you mentioned, because I think it misinterprets 00:06:20.730 --> 00:06:22.889 the technological sophistication at play here. 00:06:23.050 --> 00:06:25.709 This isn't about excluding people arbitrarily. 00:06:25.730 --> 00:06:29.199 It's about matching. In 2014, LinkedIn announced 00:06:29.199 --> 00:06:32.279 its Glean search architecture. This is the engine 00:06:32.279 --> 00:06:34.339 under the hood. How does that actually work for 00:06:34.339 --> 00:06:36.319 the person sitting at home looking for a job, 00:06:36.339 --> 00:06:38.620 though? Well, it allows for incredibly granular 00:06:38.620 --> 00:06:41.319 filtering. A recruiter could search for engineers 00:06:41.319 --> 00:06:44.300 with Hadoop experience in Brazil. Now, Hadoop 00:06:44.300 --> 00:06:47.040 is a specific software framework. Before this 00:06:47.040 --> 00:06:49.399 technology, finding that specific person in a 00:06:49.399 --> 00:06:51.379 specific region might have taken a month of phone 00:06:51.379 --> 00:06:53.600 calls and headhunters. Now it takes seconds. 00:06:54.060 --> 00:06:56.040 But does that granularity help the worker, or 00:06:56.040 --> 00:06:57.899 does it just help the boss who wants a specific 00:06:57.899 --> 00:07:00.800 cog for the machine? It helps both by reducing 00:07:00.800 --> 00:07:04.220 what economists call information frictions. If 00:07:04.220 --> 00:07:06.740 I have a skill and you have a need, we should 00:07:06.740 --> 00:07:08.939 be able to find each other. And there's empirical 00:07:08.939 --> 00:07:12.660 evidence for this. A 2019 working paper by Wheeler 00:07:12.660 --> 00:07:15.399 and others, it focused on job seekers in South 00:07:15.399 --> 00:07:18.240 Africa, found that LinkedIn training increased 00:07:18.240 --> 00:07:22.560 employment by approximately 10%. 10 %? Yes, 10%. 00:07:22.560 --> 00:07:25.279 In the world of labor economics, that is a massive 00:07:25.279 --> 00:07:27.899 number. That is a significant, tangible impact 00:07:27.899 --> 00:07:30.500 on people's livelihoods. It didn't just help 00:07:30.500 --> 00:07:33.240 them network, it helped them get hired. Furthermore, 00:07:33.579 --> 00:07:36.360 you have lists like top companies and top startups 00:07:36.360 --> 00:07:39.360 that guide talent toward high growth areas, efficiently 00:07:39.360 --> 00:07:41.879 allocating human capital where it is most needed. 00:07:42.199 --> 00:07:44.259 That's a compelling argument, and I won't argue 00:07:44.259 --> 00:07:45.939 against someone getting a job in South Africa. 00:07:46.279 --> 00:07:48.639 But have you considered the broader sociological 00:07:48.639 --> 00:07:51.980 impact? You cite one study on training, but let's 00:07:51.980 --> 00:07:54.319 look at the 2019 study by McDonald, Damarin, 00:07:54.379 --> 00:07:56.800 and others. They found that the platform creates 00:07:56.800 --> 00:08:00.800 a bifurcated winner -takes -all job market. In 00:08:00.800 --> 00:08:03.620 what sense? A winner -takes -all market usually 00:08:03.620 --> 00:08:06.279 implies something like a celebrity culture? It's 00:08:06.279 --> 00:08:08.560 worse than that. They found that recruiters use 00:08:08.560 --> 00:08:11.540 LinkedIn primarily to poach passive, high -skilled 00:08:11.540 --> 00:08:14.240 workers, people who are already employed and 00:08:14.240 --> 00:08:17.180 aren't even looking. Those are the winners. Meanwhile, 00:08:17.459 --> 00:08:20.259 active job seekers, particularly those in low 00:08:20.259 --> 00:08:22.939 or general skilled positions, are relegated to 00:08:22.939 --> 00:08:25.579 these hyper -competitive job boards. It creates 00:08:25.579 --> 00:08:28.379 a two -tier system. If you are on LinkedIn actively 00:08:28.379 --> 00:08:31.040 looking for a job, you're already signaling that 00:08:31.040 --> 00:08:33.860 you might be damaged goods compared to the passive 00:08:33.860 --> 00:08:36.740 candidate. The economic graph isn't a map of 00:08:36.740 --> 00:08:39.360 opportunity for everyone. It's a tool for managers 00:08:39.360 --> 00:08:42.460 to consolidate power. This echoes concerns raised 00:08:42.460 --> 00:08:45.500 in the ACM CIG Group Bulletin as far back as 00:08:45.500 --> 00:08:48.360 2001, warning that a lack of design focus for 00:08:48.360 --> 00:08:50.759 blue -collar workers would just concentrate power 00:08:50.759 --> 00:08:53.200 in the hands of management. I would frame that 00:08:53.200 --> 00:08:55.480 differently. The platform rewards high -value 00:08:55.480 --> 00:08:57.860 skills, which in turn encourages upskilling. 00:08:57.980 --> 00:09:01.000 It pushes the workforce to improve. But the platform 00:09:01.000 --> 00:09:04.200 actively discourages the very networking behaviors 00:09:04.200 --> 00:09:08.200 it claims to support. Look at the 2018 analysis 00:09:08.200 --> 00:09:12.539 by Ri, Huang, and Tan regarding referrals. Logic 00:09:12.539 --> 00:09:14.940 suggests, and your whole argument relies on this, 00:09:15.039 --> 00:09:17.259 that connecting with an employee at a target 00:09:17.259 --> 00:09:19.620 company would help you get a job. Right. That 00:09:19.620 --> 00:09:22.320 is networking 101. Well, they found that connecting 00:09:22.320 --> 00:09:24.559 with employees actually decreased the likelihood 00:09:24.559 --> 00:09:27.940 of a referral. That seems completely counterintuitive. 00:09:27.960 --> 00:09:30.259 Why would I connect with someone if not to help 00:09:30.259 --> 00:09:33.080 them? They called it competition self -protection. 00:09:33.779 --> 00:09:35.759 Employees viewed the connection not as a friend 00:09:35.759 --> 00:09:38.580 but as a threat, a competitor for their own standing 00:09:38.580 --> 00:09:41.179 or resources within the company. So they blocked 00:09:41.179 --> 00:09:44.019 them out or just refused to refer them. The structure 00:09:44.019 --> 00:09:46.259 of the platform pits workers against each other 00:09:46.259 --> 00:09:48.799 instead of fostering a genuine community. It 00:09:48.799 --> 00:09:51.659 turns your potential allies into rivals. I think 00:09:51.659 --> 00:09:54.279 that study reveals more about human nature than 00:09:54.279 --> 00:09:57.679 it does a platform flaw. Insecurity exists in 00:09:57.679 --> 00:10:01.200 every office, with or without an app. But let's 00:10:01.200 --> 00:10:03.460 go back to the idea of personal branding, which 00:10:03.460 --> 00:10:07.019 you dismissed as coerced performance. In the 00:10:07.019 --> 00:10:09.700 modern era, a personal brand is a necessity. 00:10:09.860 --> 00:10:13.529 It's not a vanity project. Career coach Pamela 00:10:13.529 --> 00:10:16.370 Green describes the personal brand as an emotional 00:10:16.370 --> 00:10:19.490 experience you want people to have. LinkedIn 00:10:19.490 --> 00:10:22.470 provides the canvas for that. A canvas with very 00:10:22.470 --> 00:10:25.269 strict lines and only corporate approved colors. 00:10:25.509 --> 00:10:29.490 I disagree. The platform has evolved. Originally, 00:10:29.490 --> 00:10:31.669 it was for people who already knew each other. 00:10:31.960 --> 00:10:35.600 It was a closed network, but after the 2017 interface 00:10:35.600 --> 00:10:38.600 update, it encouraged broader networking, allowing 00:10:38.600 --> 00:10:42.080 up to 30 ,000 connections. It democratized content 00:10:42.080 --> 00:10:45.120 creation through the publishing platform. Sure, 00:10:45.259 --> 00:10:47.379 you have the influencers program with people 00:10:47.379 --> 00:10:50.440 like Bill Gates or Ariana Huffington, but anyone 00:10:50.440 --> 00:10:53.320 can demonstrate expertise. You can write an article 00:10:53.320 --> 00:10:55.480 about supply chain logistics and have it read 00:10:55.480 --> 00:10:57.980 by thousands of people in your industry. That 00:10:57.980 --> 00:11:00.919 just wasn't possible before. But this democratization 00:11:00.919 --> 00:11:03.779 is an illusion because of the pressure to conform. 00:11:04.159 --> 00:11:06.919 I want to go back to Afra Sharon's research on 00:11:06.919 --> 00:11:09.740 the filtration effect. The platform forces users 00:11:09.740 --> 00:11:11.720 to construct these very specific narratives. 00:11:12.100 --> 00:11:14.980 If you have diverse interests or a nonlinear 00:11:14.980 --> 00:11:17.399 career path, maybe you took time off to care 00:11:17.399 --> 00:11:20.679 for a parent or you traveled, the algorithm punishes 00:11:20.679 --> 00:11:24.039 you. You become a risky anomaly. You're invisible. 00:11:24.379 --> 00:11:26.799 You are pressured to conform to the logic of 00:11:26.799 --> 00:11:30.039 the site just to be seen. But that logic is simply 00:11:30.039 --> 00:11:32.639 clarity. It helps people understand what you 00:11:32.639 --> 00:11:35.799 do. It's coercion. Look at the profile strength 00:11:35.799 --> 00:11:39.179 meter. You might see it as a helpful nudge. I 00:11:39.179 --> 00:11:42.379 see it as a tool to extract more data. Add your 00:11:42.379 --> 00:11:45.840 school to reach all -star status. It is gamification 00:11:45.840 --> 00:11:48.759 designed to make you surrender more and more 00:11:48.759 --> 00:11:52.059 privacy. And this branding you speak of is incredibly 00:11:52.059 --> 00:11:55.399 fragile. It is subject to corporate whims. In 00:11:55.399 --> 00:11:57.960 2022, there were reports of accounts being removed 00:11:57.960 --> 00:12:00.379 without any prior notice. And I mentioned this 00:12:00.379 --> 00:12:03.279 earlier, but the 2025 policy shift, removing 00:12:03.279 --> 00:12:05.980 protections against misgendering and debt naming, 00:12:06.159 --> 00:12:09.019 is a perfect example. You can spend years building 00:12:09.019 --> 00:12:11.019 a brand and the platform can suddenly decide 00:12:11.019 --> 00:12:14.240 that your very identity is no longer protected, 00:12:14.600 --> 00:12:16.820 making the environment hostile for transgender 00:12:16.820 --> 00:12:20.559 users. That is an empowerment. That is profound 00:12:20.559 --> 00:12:23.970 vulnerability. I admit that content moderation 00:12:23.970 --> 00:12:28.370 and policy are difficult areas, and I won't defend 00:12:28.370 --> 00:12:31.289 every single decision they make. But let's talk 00:12:31.289 --> 00:12:34.509 about the data itself. You view data extraction 00:12:34.509 --> 00:12:37.730 as inherently predatory, but I see it as a value 00:12:37.730 --> 00:12:41.409 exchange. LinkedIn uses data, specifically what 00:12:41.409 --> 00:12:43.909 they call the knowledge graph, to train machine 00:12:43.909 --> 00:12:46.929 learning models. These models infer properties 00:12:46.929 --> 00:12:49.769 to help users find relevant updates and connections. 00:12:50.480 --> 00:12:53.759 This isn't data for data's sake. How does that 00:12:53.759 --> 00:12:56.519 actually help the user, though? It powers things 00:12:56.519 --> 00:12:58.940 like the LinkedIn Learning Career Hub, which 00:12:58.940 --> 00:13:02.759 launched in October 2025. That tool uses the 00:13:02.759 --> 00:13:05.779 data to identify skill gaps. It basically says, 00:13:05.940 --> 00:13:08.840 OK, you want to be a product manager? Here are 00:13:08.840 --> 00:13:10.919 the three skills you are missing, and here is 00:13:10.919 --> 00:13:13.960 the training to get them. That is a direct reinvestment 00:13:13.960 --> 00:13:16.639 of data into user utility. It helps you level 00:13:16.639 --> 00:13:19.769 up. I just come at it from a different way. I 00:13:19.769 --> 00:13:22.450 see a timeline of security failures that suggest 00:13:22.450 --> 00:13:24.990 they simply cannot be trusted with that data, 00:13:25.129 --> 00:13:28.169 no matter how helpful they claim to be. Every 00:13:28.169 --> 00:13:30.190 major tech company has faced security challenges. 00:13:30.389 --> 00:13:33.289 It's an arms race against hackers. But LinkedIn's 00:13:33.289 --> 00:13:38.269 record is egregious. In 2012, 6 .4 million password 00:13:38.269 --> 00:13:41.070 hashes were stolen because they weren't salted, 00:13:41.230 --> 00:13:44.549 a basic cryptographic standard. Then in 2016, 00:13:44.909 --> 00:13:48.789 117 million credentials went up for sale. In 00:13:48.789 --> 00:13:52.129 2021, we saw scraping incidents affecting hundreds 00:13:52.129 --> 00:13:55.009 of millions of records. And it's not just hackers. 00:13:55.129 --> 00:13:57.870 It's the company's own features. They implemented 00:13:57.870 --> 00:14:00.409 two -factor authentication in 2013 to address 00:14:00.409 --> 00:14:03.009 those early issues. They learned. After the horse 00:14:03.009 --> 00:14:05.519 had bolted. And what about the intro feature 00:14:05.519 --> 00:14:09.240 for iPhone in 2013? Security firm Bishop Fox 00:14:09.240 --> 00:14:12.139 characterized it as resembling a man in the middle 00:14:12.139 --> 00:14:14.860 attack because it rerouted your emails through 00:14:14.860 --> 00:14:18.120 LinkedIn servers. They were literally intercepting 00:14:18.120 --> 00:14:20.139 your private emails to inject their content. 00:14:20.399 --> 00:14:23.279 That is a fundamental violation of security hygiene. 00:14:23.659 --> 00:14:26.419 And as recently as 2024, they had to suspend 00:14:26.419 --> 00:14:29.360 the use of UK user data for AI training because 00:14:29.360 --> 00:14:31.100 the information commissioner's office pushed 00:14:31.100 --> 00:14:34.240 back. They just opted users in quietly. They 00:14:34.240 --> 00:14:36.240 only seem to stop overreaching when they're caught. 00:14:36.480 --> 00:14:39.340 Security is an ongoing battle, and I do believe 00:14:39.340 --> 00:14:40.840 they have hardened on the platform significantly 00:14:40.840 --> 00:14:43.600 since those early days. But I want to pivot to 00:14:43.600 --> 00:14:46.059 the global nature of this infrastructure. We 00:14:46.059 --> 00:14:48.820 live in a complex geopolitical world. LinkedIn 00:14:48.820 --> 00:14:51.279 has had to navigate that, particularly in markets 00:14:51.279 --> 00:14:54.120 like China. Navigate is a very polite way of 00:14:54.120 --> 00:14:56.879 saying collaborate with censorship. It's about 00:14:56.879 --> 00:15:00.799 being pragmatic. In 2014, Jeff Weiner acknowledged 00:15:00.799 --> 00:15:04.139 they would have to censor content in China. His 00:15:04.139 --> 00:15:06.019 argument was that the benefits of connecting 00:15:06.019 --> 00:15:08.740 professionals in China to the world outweighed 00:15:08.740 --> 00:15:11.639 the concerns. He believed in engagement. And 00:15:11.639 --> 00:15:14.299 when compliance became impossible due to increasing 00:15:14.299 --> 00:15:16.820 requirements from the Chinese government, they 00:15:16.820 --> 00:15:19.240 did the responsible thing. They shut down the 00:15:19.240 --> 00:15:22.379 social aspects in China in 2021 and eventually 00:15:22.379 --> 00:15:26.139 the local job app in 2023. They tried to engage, 00:15:26.279 --> 00:15:28.419 and when it became untenable, they withdrew. 00:15:28.559 --> 00:15:31.669 But they were complicit for years. In September 00:15:31.669 --> 00:15:34.870 2021, right before they pulled out, they blocked 00:15:34.870 --> 00:15:37.570 the profiles of U .S. journalists in China, people 00:15:37.570 --> 00:15:40.250 like Bethany Allen Ibrahimian, because they mentioned 00:15:40.250 --> 00:15:43.330 sensitive topics. They were actively enforcing 00:15:43.330 --> 00:15:45.950 the Great Firewall. But it goes beyond censorship 00:15:45.950 --> 00:15:48.889 to outright espionage. Platforms can't always 00:15:48.889 --> 00:15:50.970 control who uses them for nefarious purposes. 00:15:51.350 --> 00:15:54.669 But LinkedIn is uniquely vulnerable. French and 00:15:54.669 --> 00:15:57.919 German intelligence, the BFV, have issued specific 00:15:57.919 --> 00:16:00.980 warnings about Chinese spies using fake LinkedIn 00:16:00.980 --> 00:16:04.500 profiles to target thousands of officials. They 00:16:04.500 --> 00:16:06.720 pose as headhunters or think tank researchers. 00:16:07.059 --> 00:16:09.820 And it's not just China. The Edward Snowden leaks 00:16:09.820 --> 00:16:13.720 revealed Operation Socialist. Operation Socialist? 00:16:13.799 --> 00:16:17.360 Yes. It was an operation where GCHQ, that's British 00:16:17.360 --> 00:16:21.139 intelligence, used a fake LinkedIn page to infiltrate 00:16:21.139 --> 00:16:24.090 the Belgian telecoms giant Bellicom. They tricked 00:16:24.090 --> 00:16:26.269 employees into clicking a link that looked like 00:16:26.269 --> 00:16:29.230 a LinkedIn profile to gain access to the network. 00:16:29.490 --> 00:16:32.450 The platform is a vector for state surveillance 00:16:32.450 --> 00:16:34.950 and manipulation. It is not a neutral meeting 00:16:34.950 --> 00:16:37.549 place. It is a hunting ground for intelligence 00:16:37.549 --> 00:16:40.799 agencies. Because we all trust that professional 00:16:40.799 --> 00:16:43.940 veneer. I'm not convinced by that line of reasoning 00:16:43.940 --> 00:16:46.600 because you're blaming the tool for the actions 00:16:46.600 --> 00:16:49.679 of bad actors. You can use a telephone to plan 00:16:49.679 --> 00:16:52.879 a crime. We don't ban telephones. Any communication 00:16:52.879 --> 00:16:55.480 network can be exploited by intelligence agencies. 00:16:55.659 --> 00:16:58.120 That doesn't negate the value it provides to 00:16:58.120 --> 00:17:00.539 the billion people who aren't spies. But the 00:17:00.539 --> 00:17:03.340 structure of the tool, the reliance on trust, 00:17:03.600 --> 00:17:06.720 this gated access approach, makes the deception 00:17:06.720 --> 00:17:10.200 so much easier. We lower our guard on LinkedIn 00:17:10.200 --> 00:17:12.680 in a way we just don't on Twitter or Facebook. 00:17:12.940 --> 00:17:15.240 We have covered a lot of ground, from the granular 00:17:15.240 --> 00:17:17.660 algorithms of the economic graph to international 00:17:17.660 --> 00:17:21.799 espionage. I want to summarize my position. Despite 00:17:21.799 --> 00:17:24.299 the valid criticisms regarding data handling 00:17:24.299 --> 00:17:28.099 and the complexities of global operations, LinkedIn 00:17:28.099 --> 00:17:31.240 remains the premier digital platform for professional 00:17:31.240 --> 00:17:34.460 existence. The ability to build a network of 00:17:34.460 --> 00:17:37.200 weak ties, those second and third degree connections, 00:17:37.849 --> 00:17:40.490 provides unprecedented transparency and mobility. 00:17:41.190 --> 00:17:43.829 The flaws you point out are solvable technical 00:17:43.829 --> 00:17:46.529 or policy issues, but the fundamental utility 00:17:46.529 --> 00:17:49.130 of a global searchable labor market is indispensable. 00:17:49.529 --> 00:17:51.849 We can't go back to the days of paper resumes 00:17:51.849 --> 00:17:54.910 and local job boards. And I will maintain that 00:17:54.910 --> 00:17:58.210 while we cannot go back, we must look critically 00:17:58.210 --> 00:18:01.750 at what we have built. LinkedIn has monopolized 00:18:01.750 --> 00:18:04.789 professional reputation. It forces workers into 00:18:04.789 --> 00:18:08.190 a system that commodifies their data. and enforces 00:18:08.190 --> 00:18:10.569 a performative conformity that stifles diversity, 00:18:10.930 --> 00:18:13.950 and the academic findings are clear, it exacerbates 00:18:13.950 --> 00:18:15.990 the divide between the high -skilled winners 00:18:15.990 --> 00:18:19.250 and everyone else. It is a system designed to 00:18:19.250 --> 00:18:22.089 help recruiters extract value, not to help workers 00:18:22.089 --> 00:18:25.089 find meaning. As the old adage goes, if you aren't 00:18:25.089 --> 00:18:27.309 paying for the product, you are the product. 00:18:27.470 --> 00:18:30.230 And on LinkedIn, the worker is the product being 00:18:30.230 --> 00:18:33.470 sold. It is certainly a topic that touches every 00:18:33.470 --> 00:18:35.940 one of our professional lives. Thank you for 00:18:35.940 --> 00:18:38.900 this rigorous exchange. Thank you. It's a conversation 00:18:38.900 --> 00:18:41.119 we need to keep having. And to our listeners, 00:18:41.299 --> 00:18:43.579 the material contains even more depth on these 00:18:43.579 --> 00:18:46.079 issues, including the specific studies we cited. 00:18:46.460 --> 00:18:48.819 We encourage you to dig in and draw your own 00:18:48.819 --> 00:18:51.039 conclusions about the role this platform plays 00:18:51.039 --> 00:18:53.599 in your working life. Thanks for listening to 00:18:53.599 --> 00:18:53.940 the debate.