What Facebook Has Done To Us

For anyone watching this video, statistically speaking, you fit into one of two camps. You’re either in the 60% who are too young to have experienced what I'm about to describe, or you’re part of the 40% and have your own variation of this story, even if you watched it play out from a distance. Regardless of which camp you sit in, we must start this story with optimism. Optimism is how we arrived at the predicament we find ourselves in today, and the best way for me to illustrate that optimism is to begin with my own experience.

I grew up in Limerick, a small city in the southwest of Ireland, and lived there until I was in my mid-twenties. By the end of my time in Limerick, I had become pretty disillusioned and restless. I was living a purely musical life, spending most of my time managing the band I was in. If I wasn't performing, I was writing or recording or traveling around the country, delivering CDs to radio stations. Yeah, CDs.

We had a distribution deal that put them in shops around the country, not yet realizing that Napster and its successors were in the late stages of strangling that model to death. But I did much more promotion than just that. I spent a lot of time building and maintaining our website using Flash and ActionScript, not yet realizing that platform dominance had made that endeavor almost completely pointless, and also that my Flash and ActionScript skills were soon to be redundant too. Even back then, I knew that social media was important, so I dedicated a lot of time to our MySpace page, which was the place to promote your band at the time. It was buggy, inundated with ads and pop-ups, and most annoyingly, it was a volcano of spam.

Scammers were free to pester you with impunity, and deleting random friend requests from people you didn't know was a daily chore. I was also heavily involved in the general music scene in Limerick. I helped put on shows, recorded other bands, and had a nice sound engineering gig too. For a time, I liked my little niche. I liked my circle of friends, and I liked working as a musician.

But I just had no money; I couldn't figure out how to bootstrap our band to get to the next stage of professionalism. With the odds feeling like they were stacked against us, we eventually went the way of most bands and agreed to split up. The obvious next step for me was to move to London, where I could make use of my design degree and get into software. Money had always been the problem, and my goal was to solve that issue to bankroll my next steps as a musician. When I made the decision to leave Limerick, I was gone within a month, and I remembered that final month really well.

I soaked up as much atmosphere as possible, saw as many shows as I could, and visited everyone I knew. I knew that this was it. I wanted to remember this time clearly. I didn't want faces to begin to fade from memory, and of course, I didn't want to be forgotten. So during my first year in London, I'd go back to visit Limerick all the time, to play solo shows and even do sound engineering occasionally.

But over time, people began to stop thinking of me as someone they could call if they needed something. My friends formed new bands, sound engineers replaced me, and before I knew it, I was a stranger in my own hometown, which was made even worse by the fact that I was an even bigger stranger in my new adopted city. It was a weird and sometimes oppressively lonely time. I wasn't yet a very experienced designer, so I sat right at the bottom of any pecking order in any company I worked in. No one knew me.

I auditioned to join a few bands, but couldn't find anyone I gelled with. So, for six months—the only six months in my life where I did this—I decided to put music on hold and try some other things and meet some people. That's when I took a brief detour and became interested in the world of amateur dramatics, joining a nearby theatre group. I'd always wondered whether my affinity for performing on stage would transfer to acting, but unfortunately, I never found out. Once the theatre director discovered I could do sound engineering, my fate as technical staff was sealed.

I'd do sound effects, audio cues, little moments of piano playing, and slight whistles for when someone's trousers fell down—you name it. I worked on a few productions, participated in competitions, made one really good lifelong friend, and attended a whole bunch of wrap parties. Along the way, I developed a slightly flirty relationship with an actress who was also Irish. I wasn't totally sure she was interested, but it seemed like it was going that way. At one particular party, when we were saying our goodbyes, she asked me for my email address, which I gave her. The next morning, I woke up to find an invitation to this thing called Facebook.

Now, when I got my first look at Facebook, I wasn't all that impressed. To my eyes, as a designer working on sophisticated web products, it looked a bit crude. What is it, another MySpace but with nothing on it? What I didn't realize was that I had just inadvertently become a very early adopter in the UK. But of course, because I was only connected to one person on it, there wasn't much to do.

I posted a message on her wall: "Hey, I'm here, yay!" and left it at that. Then, a week or two later, I received a curious email notification informing me that I had just been poked. She had poked me. So, I logged onto Facebook again and saw that I had indeed been poked. It seemed my only available option was to poke her back.

I'm not sure if this is the right thing to do. It feels a bit weird. Okay, I poked her. Unfortunately, this was the most active period in our janky dalliance, and neither of us messaged the other again. So, once more, Facebook faded out of memory, and I moved on.

The end. And then one day, a few months later, I got a new notification. A guy I'd studied with in primary school, who I'd actually forgotten existed, wanted to be my friend. I barely recognized him. The chubby boy I knew had become an adult man with a beard and everything.

That's crazy. Hey man, oh my god, how have you been? What are you doing these days? Hey Martin, it's great to see you. I'm an electrical engineer.

I live in Berlin now. Wow, it's been 15 years since I saw you. I'm working as a designer in London. Brilliant! I always remember those mad comics you used to make about our teachers.

What an interesting and totally random thing to happen. Then, about a week or two later, it happened again. This time, it was a guy I had been very close with when I studied in a brutal all-boys secondary school. We used to ditch PE class together and talk about how we would eventually get out of Ireland and become artists. I started talking to him all the time and found out that he was living in London too.

And then the notifications started coming in, multiple each week. People I'd studied with in college, people whose band I'd recorded, a guy we toured with in the Czech Republic, my old guitar teacher, etc. It was such an amazing experience. "Hey, it's so great to see you again. How have you been?"

Is your band still going? Can you send me your music? What country are you in? What job are you doing? Whatever happened to Tom?

Whatever happened to Jill? Oh, Tom and Jill are here. Send them a friend request. What followed this was what I can only describe as two weeks of total insanity—15 or 20 people per day wanting to be my friend, everyone I'd ever met who remembered my name.

Within the space of one month, I went from feeling totally adrift and isolated to suddenly being able to chat with all my friends on a daily basis. Yes, there had long been things like chat rooms and early social media sites like 6degrees.com, Makeout Club, Friendster, Friends Reunited, MySpace, or Orkut, but none of these services had come close to what Facebook had achieved. They had the right combination of privacy settings so our conversations weren't exposed for everyone to see. They engineered an experience that made us feel safe using our real names. You got almost no spam and, most importantly, the entire service was geared towards fostering communication.

When you logged in, your feed would be updated with a chronological list of what your friends had been talking about while you were away. You just scrolled back to see what you had missed. Back then, I came across all kinds of fascinating conversations. In particular, since many of my friends had totally different professions from mine, it was really instructive to read discussions they had with other professionals in the same field—programmers, civil servants, artists, academics, historians, you name it.

I remember during one election in the UK, I really enjoyed frontline dispatches from friends who were involved as either volunteers or advisors. It could sometimes be so much more instructive than traditional newspaper reporting. Of course, there would be disputes, but they took place between people who knew each other, and you were much more likely to moderate your responses simply because conversations were happening in front of people you liked and respected. There were a few people I knew who felt the need to constantly post their thoughts and conversations where I’d rather they didn't. It could be annoying, but it wasn't anything earth-shattering.

More than anything else, my memory of this period of Facebook is that it was kind of chaotic and freeform. People used it in very different ways. For example, there was a time when people would post long think pieces or blogs that would spark lengthy discussion threads. I remember once posting an intentionally terrible screenplay of a full episode of *Lost*. I was always really disappointed by the writing of that show and thought that writing my own version would be an amusing way to criticize it.

I spent about a week writing it. My intended audience was my network of connections on Facebook. When I started writing music again, I would post demos to get reactions from other musicians. It was great. I had a lot of fun on Facebook back then and would often spend countless hours on it talking with my closest friends, especially late at night on weekends when I could grab a bottle of wine, listen to some music, and forget about whatever crappy job I was working at the time.

It was my connection to my hometown, my three brothers, and my closest friends, and it was increasingly becoming my connection to my new life in London too. Over the next two or three years, as I became a more experienced designer, I managed to transition away from bad agencies and began working in software and games companies, eventually landing in a company that made games exclusively on the Facebook platform. It was now both my social life and my working life. And if that wasn't enough to make it clear how important this site had become to me, the first few conversations I ever had with a girl who would eventually become my wife took place on Facebook.

For me, it was a transformative technology. There's just no other way to describe it. The funny thing is, during the earliest years on the platform, I never really paid close attention to what was happening behind the scenes, and it would be years before I'd start following the Facebook story more closely. So, when I decided to make this video, I went back to find out as much about the early days as I could, and this is when I came to fully appreciate for the first time the sheer vastness of media coverage that exists about Facebook, which, when you take it all together, tells a fascinating story right up to the present day. The story of Facebook, from beginning to now, is the story of numerous crucial moments in the development of the internet—Rubicons that can't be uncrossed.

It's the story of the transformation of our behavior and how legal rules built for the era of TV and print are struggling to keep up. And it all starts with an explosion of positivity. In early 2004, Mark Zuckerberg launched Facebook, a site exclusively for students of Harvard University. When it proved to be a hit there, other universities emailed the team to ask if the service could also be expanded to include them. A lot of the expansion work was done by co-founder Dustin Moskovitz, as described by Mark in a lecture he gave at Harvard a little later.

Dustin was like, "Hey, I want to help out. I want to do the expansion." And that's pretty cool, dude, but you don't really know any PHP or anything like that. So that weekend, he went home, bought the book "Perl for Dummies," came back, and was like, "All right, I'm ready to go." I was like, "Dude, the site's written in PHP, not Perl."

So he picked up PHP over a few days and went to work. Dustin's contribution was nicely captured in a CBS segment about young entrepreneurs. Dustin just launched a hundred schools a couple of days ago, so he's exhausted. What's up? Dude, you're on TV!

The Facebook's initial purpose was to offer a convenient way for people at the same university to look each other up. Users could search for a person and find a picture of how they would like to portray themselves, in addition to information such as the courses they were taking, sports teams, and clubs they were a part of. However, there are many instances where people would go out, meet someone, and think, "Okay, I don't know much about this person." They would then return to their room and do what people call "Facebooking them," which involves going on Facebook to look them up, just to see the snapshot they have. In order to make its college users feel safe, privacy was of the highest importance. Additionally, users were given complete control over which parts of their profile were shown.

We don't force anyone to show anything, and we like to give people granular control over some of the more sensitive information. For example, right next to the cell phone field, there's another field that asks, "Who do you want to show this to?" We genuinely care about this issue. If people feel that their information isn't private, it negatively affects us in the long term as well. We are, of course, very sensitive to people's privacy.

By the way, just as an aside, it’s really interesting to see how poorly attended this lecture was. That wouldn’t last long. In August 2004, the venture capitalist and co-founder of PayPal, Peter Thiel, invested $500,000 in Facebook, which bought him a 10.2% stake in the company and a place on its board—an investment that was vital for the early expansion of the product. Many former employees and advisors would later report that Thiel wielded significant influence over Zuckerberg's thinking, especially during the early days. For example, Thiel is often credited with convincing Mark of the importance of having absolute control over the company.

However, in order to achieve this, there was a problem to overcome. You see, while at college, Mark had liberally handed out large stakes to a few of his co-founders, including, famously, Eduardo Saverin, who held a 30% stake in the company, which he had bought for a fairly trivial investment. To get around this, Mark incorporated a new company that effectively purchased Facebook, which allowed him to redistribute the shares, reducing Eduardo's stake to less than 10%. By doing this, Mark consolidated his position as the majority shareholder, with additional mechanisms put in place to ensure that his decisions could not be overridden—a situation that persists right up to today.

Then, another feature appeared. There's this feature called "poking," where you can go to someone's profile and poke that person. But what does it do? Nothing. It just sends them a message that says, "You've been poked."

You know, like, who cares? I mean, it's like I thought about it when I was drunk. Oh, so that's what that was about. Anyway, since the platform was now rapidly growing in registrations, a collection of new investors began bidding for a stake in the company, including The Washington Post. However, the winning investment eventually came from Jim Breyer of Accel Partners, who offered $12.7 million for a 10.7% stake in the company.

By June, the team moved operations to Palo Alto, one of the principal cities of Silicon Valley. Around this same time, Facebook expanded to serve high schools as well. An important thing to note at this point is Facebook's strict policy that registrants use their real identities. This policy was enforced by only accepting email addresses from high school domains.

This was enforced by only accepting email addresses issued by colleges or universities. As we'll see later, this policy of non-anonymity is one of the core aspects of their success. One immediate benefit, which I described earlier, was that when you have to use your real identity, it acts as a disincentive to behave badly. Another immediate benefit of this approach was that they didn't need to dedicate too many resources to content moderation. MySpace has almost a third of their staff monitoring the pictures that get uploaded for pornography.

We hardly ever have any pornography uploaded. I think a lot of the reason is that people use their real names on Facebook. When it came to hiring, they took a pragmatic dual approach. On the one hand, they began poaching high-caliber developers from other companies by offering higher salaries, primarily targeting those who worked at Google. This was one of the many tactics that would make the two competitors mortal enemies.

However, they also diversified their teams by focusing on inexperienced raw talent as well. You know, I started the company when I was 19, right? So I can't institutionally believe that experience is that important.

You know, we invest in people who we think are truly talented, even if they haven't done that specific thing before. Technically, one of Facebook's major advantages was how quickly it could process information. Other social networks like MySpace and Friendster were suffering from mounting performance problems as more users joined, increasingly struggling to cope with the amount of information being shared and consumed. Facebook's innovation was a system for storing data called the social graph, which was partly responsible for its ability to scale rapidly. It's worth describing the social graph, as it played a significant role in their rapid growth.

It's worth describing the social graph in a basic sense. You, as a user, are a single node in a network; other people are also individual nodes. If you send them a friend request and they accept, a connection is formed that describes the relationship between both of you. And it's the connection description that is crucial here. Say you make a post of some kind.

This post becomes a new object in the network. Its connection to you is that you're the publisher. If your friend comments on this post, a new connection is drawn that describes their relationship to it as a commenter. So, as you and your friends engage with and comment on each other's content, more and more nodes and connections are created that explain the relationship back and forth. This means that everything around you— all the actions you've taken on anything you've seen, and all the actions taken by anyone else that relate to you— is stored as a massive constellation of crisscrossing connection points.

This allowed the data to be decentralized so that information most relevant to you could be loaded quickly, which made Facebook far faster than its competitors, who relied on more traditional databases. These traditional databases needed to scan through numerous massive tables of data for each user, matching them against other massive tables for each type of social interaction. This process became increasingly inefficient as the platform scaled up. The scaling issue that social networks tend to encounter is that users consume orders of magnitude more data than they create, which means the database storing all this information is under constant bombardment by millions of computers trying to fetch information. This results in slower performance, especially as the overall population grows. To run this efficiently, Facebook initially used the open-source solution Memcached, which, as the name suggests, caches versions of the database on multiple servers distributed all over the world. This way, computers fetch information from Memcached servers rather than directly from the database.

If the information is missing from the cache, the Memcached server fetches it from the database and stores the result. As the years progressed, Facebook would build more sophisticated systems on top of this so that it could keep up with the rapid growth in users. This investment in scalability would prove to be one of the decisive advantages that Facebook had over Friendster and MySpace. Moreover, armed with real identities and a rich database of connections, it was possible for Facebook to make accurate friend recommendations that greatly sped up the process of adoption. Just as all of this was beginning to really lift off, in late 2005, the influential author Tim O'Reilly published a now-famous article titled "What Is Web 2.0?"

In this article, he predicted that the future power holders of the web would be the platforms that harnessed what he called the collective intelligence of the internet—in other words, those who controlled the most data. The race is on to own certain classes of core data: location, identity, calendaring of public events, product identifiers, and namespace. The winner will be the company that first reaches critical mass via user aggregation and turns that aggregated data into a system service. He may not have known at that exact moment that he was describing Facebook two years in the future. In 2005, Facebook released the ability to upload photos and also the ability to tag people in them. If someone tagged you in a photo, it would automatically be linked to your profile, making them immediately accessible to anyone you were friends with, which helped to make the site feel more alive and engaging.

They also introduced events around this time. This feature allowed users to organize social gatherings or other kinds of public events. In the early days, it was used by almost everyone I knew—gigs, meetups, house parties, you name it. It was a major aspect of how Facebook felt like the center of your social life.

Even now, thinking back to how everyone used to use events makes me feel a little nostalgic. Around this time, they changed the name of the site from "The Facebook" to just "Facebook" and bought the facebook.com domain for $200,000. 2006 was an enormous year. In September, they opened up registrations to the whole world. Now, anyone over the age of 13 who had a valid email address could join.

But this new version of Facebook that the public saw was very different because it included a revolutionary design change—something that is not hyperbolic to say would go on to change the internet and the wider world: the News Feed. Prior to the News Feed, Facebook was a static site where, if you wanted to find out what your friends or family were up to, you visited their wall, which resided on their profile page. From there, you would post messages to them and wait for a response. If you didn't do that, you missed out.

To find information, you basically went to one person's profile, then another, and then another. It was hugely inefficient. We knew we could do better. To solve this, the news feed aggregated posts and activities from all of your friends' profiles and collected them together in one continuous chronological feed. We wanted to deliver 10 million unique newspapers every single day.

It was undoubtedly a major success, despite an initial uproar from many of its users who didn't expect such a significant change. They were used to one-on-one discussions, not one-to-everyone discussions. There were some other unexpected consequences too. For example, say you just went through a breakup and changed your relationship status to single. Now, due to the newsfeed, that life event would suddenly be broadcast to all your family and friends, as well as many of your partner's family and friends too, bringing what would be, for many, an enormous amount of unwanted attention to a particularly painful episode in your life.

We woke up to hundreds of outraged people. Groups had formed in the middle of the night, such as "I Hate Facebook" and "Ruchi's the Devil." Amidst all that chaos and noise, we noticed something unusual: even though everyone claimed to hate it, engagement had doubled.

There were more page views than there ever were before. Aggregation feeds might not have been a new idea, and there were already somewhat comparable examples on other social media sites, such as Flickr. However, Facebook's application of this concept to the way we interact with our friends was massive, especially when they began applying sorting algorithms later to determine what content should appear in our feeds. In July 2006, Mark was faced with a crucial decision when the search giant Yahoo offered to buy Facebook for $1 billion—an offer which, after some soul-searching, he eventually turned down. That was really stressful because a lot of people really thought that we should sell the company.

So, billion—a offer that, after some soul-searching, he eventually turned down. After a couple of years, that was like a home run, right? And it is a home run, right? The part that was painful wasn't turning down the offer; it was the fact that after that, huge amounts of the company quit because they didn't believe in what we were doing.

2007 is when Facebook began to demonstrate to the wider world how significant it was going to be. This is the year we can look back on and see that Mark Zuckerberg was, from a business perspective, becoming a highly savvy long-term strategist. Case in point, in April, they launched the Facebook Platform. This allowed third-party developers to create applications and games that ran on Facebook, which turned out to be an incredibly valuable two-way relationship. Anyone who has ever worked in a startup can tell you that the hardest part of trying to grow a digital business is getting people to even notice you exist.

In particular, getting users to register can be a challenge. No one likes signing up for things, and to convince them, you often need to invest in marketing. So, if you were an indie developer, you either attracted investors to help with marketing or you were almost certainly out of luck. However, Facebook created a way for software developers to access users' data from the social graph, enabling them to sell their products to the largest set of connected users in the world. For game makers, this was an enormous opportunity.

If people played their game on Facebook, their friends would get notified about it in their news feeds. Free advertising. As I mentioned, I worked in a gaming company that thrived on this platform. Our access to the social graph was absolutely crucial for our survival. Facebook was no longer just a site; it was a platform with endless different kinds of applications and games that would keep people coming back and staying longer.

By 2012, Zynga, the biggest name in social gaming on the platform at the time, made up 11% of Facebook's revenue. I'll never forget this period in London, when there was investor money sloshing around everywhere looking for the next big social gaming company. I remember interviewing at one of the darlings, Playfish, which had just been bought by EA. They were so in need of people with excellent Flash skills that they made me an offer on the spot. I then used that contract to negotiate a better offer with the company that I eventually joined.

The second major change in November of 2007 was Facebook Ads. The significance of this move is one I doubt I need to expand upon all that much. It was a system that now also granted advertisers access to the rich data contained in the social graph, allowing them to target their adverts with a level of precision never seen before. Age, gender, region, interests, previous purchases—you name it. Four months later, they hired Sheryl Sandberg to take on the role of Chief Operating Officer.

Sandberg had previously been instrumental in growing Google's advertising business, and her move to Facebook was a significant coup. Facebook ads, combined with Sandberg's experience, were a major factor that led to Facebook becoming a tech giant. According to The New York Times reporters Shira Frankel and Cecilia Kang, Cheryl was the perfect foil for Zuckerberg. She oversaw all the departments that didn't interest him: policy and communication, legal, human resources, and revenue creation. Drawing on years of public speaking training and political consultants to curate her public persona, she was the palatable face of Facebook to investors and the public.

It was also in 2007 that Microsoft invested in Facebook, buying $240 million in preferred stock, thereby valuing Facebook as a $15 billion company. At the time, this investment was criticized for being too cavalier. Now, it's clear that Microsoft could see what had been predicted by Tim O'Reilly. They understood the ridiculous power that comes with having the world's largest database of real user identities, who, through their usage of the platform, were constantly feeding the system with more and more data about their interests and behaviors. Later in 2007, Facebook introduced Pages, a feature that allowed entities other than personal profiles, such as artists, brands, and public figures, to create a presence on the platform.

We'll be coming back to this one later. In 2008, they launched Facebook Connect, which allowed users to sign into third-party sites using their Facebook login details. This integration enabled other services to utilize information shared by Facebook and vice versa. For example, Spotify used this integration very effectively, allowing users to share their playlists with friends on Facebook, which in turn drove more traffic back to Spotify. In return, Facebook obtained additional data and was able to provide an even richer experience for its users.

Overall, this technology effectively made any third-party sites that supported it an extension of Facebook. Also, in 2008, they introduced a new messaging feature that provided a way for users to communicate privately. Then, in 2009, they introduced what would become the defining symbol of the platform – the Like button. The Like button almost immediately became a new indicator of social capital, letting users know how much people appreciated the content they posted. Shortly after, Likes would also be used by Facebook as a vital metric that would determine what appeared in people's News Feeds.

By June 2009, they announced publicly that Facebook was now being used by 250 million people, an incredible milestone. Finally, in 2010, rounding out the core experience we generally know today, they introduced Groups, a feature that allowed users to invite like-minded people to groups that could be collaboratively managed by their members. This enabled people to engage in group chats, which could be shielded from view if needed by making them private. And that's it. An incredible growth story over a six-year period, where each step widened the possibilities that paved the way for an even larger next step.

This is how they managed to become vital not just to my private and working life, but also to those of countless others around the world. So I hope you enjoyed my story about Facebook, how it got started, and where it ended up. I expect that for some of you young whippersnappers out there, it might serve as a bit of an education, right? If you'll allow me to elaborate for just a few more seconds, there are some great books I want to recommend that helped me a lot when putting this video together. The first, and in my opinion the best, is *Broken Code* by Jeff Horowitz of The Wall Street Journal, released at the end of 2023.

It's kind of a strange name, actually—Broken Code. It sounds vaguely negative. Anyway, then there's "An Ugly Truth" by Cecilia Kang and Shira Frenkel, which collects numerous accounts from former employees. Another is an account by Roger McNamee, who knew Zuckerberg during the early days.

It's okay. Not essential reading, but it has its moments. That one's called "Zocked." Gosh, I should actually probably read these. Hold on, let me just grab my phone.

Own Kindle, my bundle. What? Ah, right. Um, uh, yeah, right. Okay.

Uh, right. You know what? Let me just put that down. This does actually bring back some memories. Just for the sake of being thorough, I'd like to add a quick addendum to this video.

Okay. There was actually something that bothered me a little bit in the early days. It’s not a big deal; I can be finicky about these things. Right?

Okay, so Facebook was an experience that changed an awful lot during the first few years. Although many of the changes were positive, there were others that gave me pause. For example, I remember scrolling through the news feed one day and coming across a very simple, but also, to me, rather unusual statement: "Andy Jones likes Budweiser."

That's it—a picture of the Budweiser brand, with a "like" associated with my friend Andy. The funny thing was that, since I knew Andy really well, I was pretty certain he wasn't much of a Budweiser fan. I mean, you'll struggle to find anyone in the UK or Ireland who is. The other funny thing was that, since Andy was a work colleague, he just happened to be sitting across from me at the time, so I called him over. "Andy, what do you make of this?"

And I'll never forget the look of confusion and mild frustration on his face when he read the message. I don't remember his exact explanation for the post he had seen that he liked, but what I do remember was his conclusion: "And now they're using me to sell you Budweiser?" He genuinely felt embarrassed by this association, as would I. This was a feature called social ads, initially launched in 2007.

The idea was simple. Brands could post interactive updates that appeared in the side panel or newsfeed, and if you made a comment or liked the message, that engagement would then be broadcast to your friends' feeds. This was a popular method of advertising because research data showed that people were much more likely to engage with advertisements if they took the form of a recommendation from their friends. As you can imagine, this was not a popular move, largely because the change had not been implemented with any transparency. Since people were not asked to give their consent to be used as advertisements, many became a little less trusting.

And bear in mind that this was happening during a time when people were beginning to wake up to a new reality, where their personal information was being used for the purpose of targeted advertising across the Internet. To further erode trust, there were numerous privacy blunders, which people were becoming more accustomed to, even as early as 2007. One particularly concerning early example was uncovered by the privacy activist and blogger Chris Seguyán. In a nutshell, he discovered that even if your privacy settings were quite tight, allowing your personal information to be seen only by friends, it was still possible to learn deeply personal details about you using their advanced search feature. For example, if you searched for your neighbor, John Smith, and limited the search to the town in which you live, it could reveal sensitive information.

Boom! There's John Smith. Then you refine your search, this time including the filter "Religion: Buddhist." Oh, John Smith's name no longer pops up. I wonder why.

Okay, let's instead change the religion to Christian. Ah, John's name appeared again. I guess he must be Christian. After a little bit of additional research on this exploit, Segoin eventually realized that you could brute force a lot of personal information about large numbers of people, including things like the kinds of groups they were signed up for, as well as personal identifiers or hobbies they listed on their profiles. Using this information, Segoin demonstrated that one could, say, quickly isolate a group of people in a university who fit this criteria.

Christian men interested in other men, Muslim women interested in other women, students who work in local strip clubs, and students under the age of 18 who self-identify as being interested in alcohol, etc. It's not hard to see how this information could be put to very bad use.

When Segoin published this particular revelation, it was picked up by several different news publications, eventually leading to Facebook tightening up its privacy settings. You can still see a comment from one of the Facebook representatives announcing the fix in the comments section under Segoin's original blog post. Another, even earlier privacy-related outcry was caused by the News Feed itself. Remember that example I gave earlier about changing your status to single and everyone being able to see it? Many people thought that this was highly creepy, resulting in a significant outcry.

So, why did they do it? Well, let's return to that clip I showed earlier by Ruchi Sng, the then product manager of the News Feed. Even though everyone said they hated it, engagement had doubled. We may as well stamp that phrase into our memory banks right now: even though everyone said they hated it, engagement had doubled.

Now, I apologize for pointing out the obvious here, but engagement and fulfillment are not the same thing. If someone clicks on more content, that doesn't necessarily mean they're happier. The themes in this video will run through to the end. Nevertheless, the outcry was so severe that Zuckerberg published what was quickly becoming a common feature of the site: the semi-apology blog post. Calm down.

Breathe. We hear you. People are concerned that non-friends can see too much about them. We're listening to all your suggestions on how to improve the product. It's brand new and still evolving.

And we agree, stalking isn't cool, but being able to know what's going on in your friends' lives is. Okay, so if we return to 2007, believe it or not, we've still not gotten to the biggest privacy scandal of that year: Beacon. Beacon was an advertising program that third-party vendors could sign up for, which allowed your purchases on their sites to be repurposed into ads targeted right back at your friends on Facebook. So what are the consequences of this? When a Facebook user bought his wife a diamond ring online, the surprise was ruined because Beacon notified all his friends and his wife about it on Facebook.

Oh dear, the cat's out of the bag. There is an absolute uproar because users were not informed about the change at any point, nor were they asked for their consent. Beacon was opt-out. This prompted a severe online response from users and privacy advocates, as well as a lot of critical news coverage. Regardless, Zuckerberg resisted making changes to the feature, justifying it as a cool, positive way to share your tastes with others.

Is there any concern that you're turning Facebook into something much more commercial? I actually think that this makes it less commercial. I mean, what would you rather see? A banner ad from Bloomingdale's, or that one of your friends bought a scarf?

Well, I guess this shows how difficult it is for a company like yours to make money through advertising while also protecting people's real privacy and their sense of privacy. You know, it might take some work for us to get this exactly right. This is something that we think is going to be a really good thing. There have to be ads either way because we need to make money. I mean, we have 400 employees, and we have to support all of that.

And make a profit. No one bought this. Soon after, Zuckerberg then posted another apology blog entry on Facebook titled "Thoughts About Beacon." "We've made a lot of mistakes building this feature, but we've made even more with how we've handled them. We simply did a bad job with this release, and I apologize for it."

Again, many were not impressed by this apology, especially because there was no suggestion Beacon was going to be discontinued. It was merely a poorly executed effort. Then, even worse, a researcher named Stefan Berto published a blog post revealing that Beacon tracked users' activity and sent it to Facebook, even when they weren't signed into Facebook—a serious privacy breach, even for 2007. Eventually, in an attempt to put the whole fiasco to rest, Zuckerberg agreed to make Beacon opt-in. But by then, there was already a class-action lawsuit filed against them, which they eventually settled for $9.5 million in 2009.

And then, to coincide with this payout, Facebook finally agreed to discontinue Beacon entirely. In February 2009, Facebook made changes to their terms of service. These changes appeared to say that Facebook would hold onto users' data even after they deleted their accounts and that this information could be used for advertising purposes. This change was reported on by a now-defunct blog called The Consumerist and sparked enormous backlash and media attention. The pressure eventually led to Facebook reverting back to their previous terms of service.

But regardless, its effect was to broadcast what looked like a presumption of permanent ownership over users' personal data. In a similar vein, another increasingly unpopular feature from a privacy point of view was "Find Friends," a service that allowed people to use their address book of email contacts to quickly locate their friends, which was much more convenient than searching for them manually. However, what users didn't realize was that the addresses they were uploading could be stored and used to build what's referred to as a shadow profile for those not yet on the platform. Worse still, this data could potentially be shared with advertisers as well. This meant that at the time, Facebook could obtain and benefit from your personal data even if you actively chose not to be on Facebook.

And there we have it: Facebook's early privacy issues. Just a few months later, another setback for Facebook. The social networking giant reportedly revealed the sexual orientation of some users to advertisers. The problem was that if you were viewing a page or group related to a specific sexual orientation and clicked on an ad while there, advertisers could potentially deduce your sexual orientation by looking at the referrer URL, even if your profile was set to private.

But that's not all. Just last week, we learned that some of the most popular apps on Facebook were leaking users' information to advertisers. This was a bigger deal. The crux of the story is this: If one of your friends was playing a popular game, like Mafia Wars or FarmVille, even if you never opted to play the game yourself, the games could still scrape your personal information and give it away to fourth-party advertisers.

It was a problem reported to have affected tens of millions of users. So why was this possible? Well, in 2010, Facebook released a tool called the Open Graph, which granted outside developers access to data available from the social graph. In order to ensure that the Open Graph was adopted, it included a significant amount of personal information. Usernames, locations, emails, birthdays, political persuasions, relationship status, and employment history, as well as the activities, posts, and likes of one's friends, were all made accessible.

The Open Graph puts people at the center of the web. One former Facebook employee, Sandy Parakilas, warned at the time that the Open Graph could be misused for mass data harvesting by bad actors, but claimed that her objections were ignored. Just store this one away for later. Open Graph, we have not heard the last of this.

It's like déjà vu. We're always talking about Facebook and privacy. Now, aside from Beacon and the Open Graph, you might have noticed that a few of the errors I've just described were unforced errors. In fact, unforced errors basically came to typify the platform during this time. But why do they keep happening?

Well, partly it's due to the development ethos that underpinned Zuckerberg's now infamous catchphrase, "Move fast and break things," also known as "done is better than perfect." The idea is this: when trying to grow in a highly competitive landscape, it's better to get a feature released as quickly as possible so that you can collect data on how it's being used. From there, you can use that data to inform how you refine the feature and make it better. Or, if it's underperforming, you can kill it completely and move on. In this way, when you fail, you fail fast, rather than wasting tons of time and resources polishing things that are destined to be dead on arrival.

However, not all features are created equal. The consequence of a page failing to load is embarrassing for sure, but fairly trivial compared to the consequences of mishandling your users' private data. However, this distinction does not appear to have been treated with enough importance by Facebook, and there are innumerable accounts from former employees indicating that the company was a total mess when it came to stability. According to Jeff Horowitz, bonuses and promotions were doled out to employees based on how many features they shipped. Reviews incentivized employees to complete products within six months, even if it meant the finished product was only minimally viable and poorly documented.

Engineers and data scientists described living with perpetual uncertainty about where user data was being collected and stored. To Brian Boland, a former vice president in Facebook's Advertising and Partnerships division, building things is way more fun than making things secure and safe. Until there's a regulatory or press fire, you don't deal with it. Now, of course, Facebook stated at the time that privacy was of the utmost importance to them. But to rattle off an old cliché, it's what you do, not what you say, that indicates where your real priorities lie.

And with that in mind, let's begin building a profile of Facebook's actual philosophy. We've already included the phrase "even though everyone hated it, engagement had doubled." Now let's add "done is better than perfect." These are both aspects of the same general drive to grow as quickly as possible. Bear in mind that at the time, Facebook was by no means a guaranteed long-term success, and Zuckerberg was fully aware of this.

Yes, they had hit 250 million users, but as we'll see later, there was an enormous amount of growth left to do. Friendster and MySpace had already demonstrated what happened to complacent social media firms, and it was entirely possible that a new upstart could come along and knock Facebook out of the game. This fear was even expressed in an internal manifesto handed out to employees, called the "Red Book." "If we don't create the thing that kills Facebook, somebody else will." To go back to Tim O'Reilly's observation in his famous "What is Web 2.0" article, the race is on to own certain classes of core data: location, identity, calendaring of public events, etc.

The key word here is race. This was a race, and Zuckerberg understood this better than most. He was in this game to win, and the only way to win was to become so big and so dominant worldwide that you couldn't really be touched. As his co-founder, Chris Hughes, put it, "From our earliest days, in the past few days, Mark used the word domination to describe our ambitions." This leads us, neatly, to the biggest privacy scandal during Facebook's formative years—a scandal largely in response to the rise of a new upstart.

Twitter. Founded in 2006, Twitter began to explode in 2007, going from 20,000 to 60,000 tweets per day. Its model was completely different from Facebook because everything you posted was public. This made it an attractive platform for celebrities, journalists, and organizations, who could reach a wide audience without requiring a prior connection. Tweets could go viral to the world in ways that Facebook posts couldn't, and it was also more suited to breaking news and real-time coverage of live events.

At this moment in time, Twitter was a major competitor that Facebook needed to clear off the board. So, in 2008, Zuckerberg tried to acquire it. However, talks quickly broke down due to disagreements about the valuation of the company, and Twitter decided to carve out its own path. As Twitter continued to skyrocket, Zuckerberg began to realize that the very thing that had put Facebook on the map—being able to converse with your circle of friends out of view of the general public—was becoming a limiter to growth. If Facebook users would instead post their thoughts publicly, they would not only be able to blow Twitter out of the water, but they’d also become even more attractive to advertisers and third-party developers, too.

When people share more, the world becomes more open and connected. In a more open world, many of the biggest problems we face together will become easier to solve. So, in December 2009, they introduced the Privacy Transition Tool, a pop-up that appeared in front of all Facebook users. This tool contained a large and complex set of privacy controls, ostensibly created to help users tweak their privacy settings more precisely. If users wanted to skip all this detail, they were encouraged to press a button that simply accepted the recommended settings, allowing them to get on with their day. However, what was not obvious was that Facebook had changed whatever defaults people had previously set up.

Information that was supposed to be only viewable by friends had automatically been changed to a new setting: viewable by everyone. I remember this moment really well. A pop-up appeared about some new widget they had created where I could fiddle with privacy settings. That's nice, but I didn't really care too much because I was already reasonably pleased with how it worked.

And it wasn't until months later that I realized my profile was now basically public. I'll never forget how irritated I was when I finally investigated these new settings in detail and saw the new defaults. I would never have chosen those settings had I been presented with them clearly, and it was definitely one of those key moments for me where I began to regard Facebook as a suspicious and untrustworthy company. I'd argue that this specific scandal was the one that began seriously altering the general public's impression of Facebook. It was no longer a cool site that made a bunch of dumb mistakes, which could be forgiven; it was now a site that we all felt could be relied upon to intentionally deceive you for their own benefit.

The move seemed too transparently self-serving for there to be any other explanation. We are here today to urge Facebook's creator and CEO, Mark Zuckerberg, to revisit this new policy as soon as possible. We believe Facebook should reverse its policy and begin exploring guidelines so users know upfront what’s happening. Do you feel like it’s a backlash, or do you feel like you’re violating people’s privacy? You know, there were real learning points and turning points along the way in terms of building things.

If I knew what I know now, then I hope I wouldn't have made those mistakes, but I can't go back and change the past. I can only do what I think is the right thing going forward. In 2011, the FTC ultimately issued an eight-part complaint against Facebook, culminating in an order to cease these kinds of practices. The resulting settlement forced Facebook to take a number of different steps to improve their protection of users' personal data, including submitting external audits. At the time, it appeared that these steps might finally bring Facebook to heel.

But, since all this was generating so much bad press, Facebook decided to go on the offensive by hiring the public relations firm Burson Marsteller, which came up with a pretty silly scheme: deflect attention away from Facebook by trying to make other tech giants look bad. To do this, they contacted tech bloggers and tried to convince them to write negative stories about Google. Specifically, they fed them a weak story about one of Google’s features, which they claimed was misusing users' private data. To sweeten the deal, they suggested that if the bloggers wrote these stories, they would help get them published in prestigious outlets like The Washington Post or Politico. One of the bloggers they emailed was Chris Seguyen, the guy who broke the story about Facebook's search exploit.

Chris, immediately sensing that something suspicious was going on, emailed back, "Who's paying for this? Not paying me, but paying you." To which they responded, "Thanks for the prompt reply. I'm afraid I can't disclose my client yet, but all the information included in this email is publicly available. Any interest in pursuing this?"

Seguin published the email exchange, and soon after, USA Today finally linked it back to Facebook, causing them a significant amount of embarrassment since the scheme was pretty ham-fisted. They really thought it was okay to tell a journalist, "Don't worry, I'll write your story for you; no one will know." What does that make you feel like when you get this call saying, "Hey, I'm going to let you be a fake journalist for a day"? I was a little shocked that Facebook's PR firm, acting on behalf of Facebook, thought that I would go along with their pitch and attack one of their competitors, not for a legitimate privacy problem, but for something that was really not bothering anyone. This is not the last time Facebook will try to do something like this.

Anyway, this pretty much concludes what I think of as phase one of Facebook's privacy misadventure, which very much sets the stage for phase two. But for now, let's move on. So, if we stop for a second and look at our profile so far, we've addressed Facebook's goal of achieving massive growth as fast as possible, and I've mentioned their working practice mantra that causes constant sloppiness. These are major factors in their decision-making, for sure, but there's another core factor I've not yet included, which it's time to address: corporate virtue.

I've worked at many companies over the years and have witnessed, on quite a few occasions, how proposals for new directions can be framed as acts of virtue—something altruistic, not self-serving. These proposals are often accompanied by a mobilization of employees to get on board. Sometimes, when these proposals gain enough traction and achieve widespread self-delusion, they can lead to truly terrible decisions that baffle outside observers. The need to appear very virtuous is especially true for social media companies, which, because they serve people, can't justify their existence on purely self-interested grounds. They can't say, for example, "Our goal is to grow so fast that we acquire the world's data before anyone else does."

No, they need to convince those within and outside that there's a higher moral principle underpinning their actions. Facebook's statement of virtue, their mission, was to make the world more open and connected. This value is consistently posited as the ultimate good because when people are connected and opened up to a world of information, it ultimately empowers them. It's not an insane thing to say in the abstract.

Being connected to networks of friends has clearly been beneficial for hundreds of millions of people. However, when the idea of connection is positioned as the ultimate good, we begin to see problems. Later in this video, we'll discuss utterly catastrophic issues, but for now, we're still only talking about privacy. We're still in the playpen, folks. So you might be thinking, "Okay, but where's this internal mass self-delusion?"

Well, helpfully, an internal memo called "The Ugly," written by Andrew Bosworth, one of the founding members of Facebook, lays that bare. It was an attempt to boost employees who were beginning to feel the weight of external criticism. In the memo, he reminds them of their core mission: connecting people. He then boldly states that this mission is so important that it justifies many of the privacy scandals we just looked at. Maybe it costs a life by exposing someone to bullies. Maybe someone dies in a terrorist attack coordinated using our tools, and still, we connect people.

The ugly truth is that we believe in connecting people so deeply that anything that allows us to connect people more often is, de facto, good. That's why all the work we do in growth is justified, including all the questionable contact and porting practices, the subtle language that helps people stay searchable by friends, and all of the efforts we make to facilitate more communication.

Anyway, this statement of corporate virtue—connecting people is the ultimate good—has to be added to our profile. This one is the rocket fuel because it supports the first two statements. While we're on the topic of corporate virtue, things became a bit more complicated when, in 2012, Facebook became a public company, with their initial public offering being one of the most anticipated in history. Our mission isn't to be a public company; our mission is to make the world more open and connected.

From this point onwards, when Mark made decisions, there would now be shareholders expecting those decisions to deliver returns. Anyway, the statements of virtue just kept on coming. Remember that internal manifesto I mentioned a minute ago, the one they handed out to employees called the Red Book? Well, it contained a pretty prophetic statement: changing how people communicate will always change the world.

A bit silly and self-aggrandizing, this statement also contained an obvious logical flaw. You know, change isn't always good. And man, was Facebook about to change. Starting around 2011, I began to notice something quite concerning. My closer friends increasingly seemed to be routinely ignoring any of the things I posted, especially when they were think pieces or ideas about a video.

Larger topics I was interested in and wanted to get feedback on were increasingly overshadowed by my growing disappointment in how others were changing what they shared. Posts were rarely conversational anymore and were increasingly becoming one-liners, trite jokes, and toothless observations, along with memes, trivialities, or just links to YouTube videos with no commentary. At the time, I interpreted this as a general lack of interest in actual conversation and an increasing tendency to be lazy. This came to a head in 2013 when I took a career break, moving to Glasgow to study for a master's in composition.

So, once again, I found myself in a new city where I didn't know anyone, and once again, I tried to take some refuge in Facebook. But this time, it was totally different. Almost no one saw anything I posted, and I began to develop a mild hang-up about it. Sometimes, I got no response from anyone. It wasn't until I asked one of my friends straight up, "Are you not seeing the stuff I'm posting anymore?"

His response? What stuff? Incidentally, while making this video, I looked back over some of my posts from that period, which didn't take long because there are only a handful of them, and noticed an interesting one from 2008. It was a link to a YouTube video I had just released—a comic review of the notation app Sibelius. On YouTube, this was my first experience of a properly viral hit, seen by hundreds of thousands of people within the first week of its release.

On Facebook, its total, as of today, is 13 likes and 5 comments, which is high for me. All of the comments, bar one, are from immediate family members. None of the hundred or so composers or music teachers I'm connected to on Facebook saw it there; all of them saw it on YouTube. I don't know when it happened, but Facebook made a decision at some point that Martin Keery is as boring as hell.

Send him right to the back of the queue every time. But here's the thing: I wasn't the only person experiencing this. My assumptions that my friends were getting sick of me were largely wrong. At least I hope so.

And my perception that people were becoming lazier, causing them to post dumber stuff, was also off the mark. So, what was going on? Well, the main reason was that the news feed kept changing at a rapid pace. Initially, it was guided by a simple algorithm called EdgeRank, which decided what content you should see. Its purpose was to prioritize posts by the friends it deemed most likely to be important to you, judged in large part by how much you engaged with them.

Over time, though, EdgeRank was gradually replaced with more sophisticated machine learning-driven algorithms, which started to optimize for maximizing engagement and the amount of time users spent on the platform. The justification was that more time equals more good. The introduction of the like button supercharged this situation. People began to think of their posts in terms of how successful they were, and Facebook now had a powerful metric to help determine which posts should be boosted. Since my posts were only intended for a few people and therefore didn't receive much engagement compared to more widely appealing memes and one-liners, this changed my entire perception of what Facebook was for. It made me feel like I was now somewhat in competition for attention.

The most successful thing I ever posted on Facebook was a really simple observation about some workspace jargon I noticed people were beginning to use. Those who saw this post started chiming in with examples of workspace jargon they found agonizing, and the post blew up. For days, people commented on it, and it was a nice feeling for as long as it lasted. But what's the lesson here? Think like an editorial team?

Think of things to say that will draw people in and get clicks. What's the point? I just want to talk to my friends; it's too much work. However, there were plenty of people who did get drawn into this competition, and many of them started learning from the algorithm, discovering new techniques that tended to generate more engagement.

My absolute least favorite of these—and I know this is not an original observation—is what I'm going to call the "Living the High Life" post. You and your friends on a yacht, drinking champagne, glamour shots in exotic locations, and posts where you allude to how successful you are. Now, early on, I was a little bit guilty of this last one. I definitely bragged a little about things I thought were impressive until I read a blog post titled "7 Ways to Be Insufferable on Facebook" by Tim Urban, creator of Wait But Why. His main observation was that there are fundamentally two different kinds of Facebook posts: those you post for other people and those you post for yourself.

Let me give a few examples. Heading to Hawaii with the love of my life for a month of relaxation, yoga, and shopping. I need to be careful, though; it can disappear just like that. What's the purpose of this post?

Well, to craft an image of yourself so that others are jealous of you. I just got my degree—distinction, baby! This second example, when viewed through Tim Urban's critical lens, was a bit of an eye-opener for me because I recognized that I was slightly guilty of doing this. Although I hadn't specifically posted my degree results, which were excellent, I definitely posted other achievement milestones over the years.

Now, in one respect, so what? I think a legitimate life achievement should be celebrated because they don't happen every day, and it's understandable that you'd want to share them with loved ones and friends, even if you are kind of boasting a little bit. However, this blog was now forcing me to confront an awkward problem I legitimately had not considered until then: unintentional harm. What if one of my friends was having a hard time—an especially bad breakup, flunking out of university, losing their job, or facing a serious bout of depression, etc.?

And at their worst moment, they suddenly see my post about how well things are supposedly going for me, which acts as a punch in the stomach for them. The worst part is, I wouldn't know I had done it. The moment I started pulling that thread, I realized social media had a lot more serious consequences than I had first imagined. We'll get back to this topic later. There were plenty of other major social changes happening too, some of which were natural evolutions that can't be blamed on Facebook. For example, that moment when you get a friend request from your mom.

I remember jumping on a chat with my three brothers to discuss how we were going to handle this, ending with us all jointly realizing that, like, we were hardly going to reject a friend request from our mom. It turns out that most of the people we knew were also coming to the same conclusion, and bam, the older generation is all suddenly here, all suddenly commenting on our posts and browsing with mild concern through the photos we’ve been tagged in. Photos of our nights out, gigs we played in crazy goth clubs with naked people dancing around. Another thing that affected Facebook usage was the horror stories we kept reading about, where people got fired from their jobs because they didn’t realize the things they thought were private were actually public, including posts where they complained about how much they hated their boss. A famous incident of social media foot-shooting occurred when college students who had appeared in the movie *Borat* sued the filmmakers.

In the movie, they'd been seen guzzling booze and making racist remarks, which they argued was an unfair portrayal that caused them significant harm, including damage to their chances of seeking employment. As part of the filmmakers' defense, they submitted images posted publicly on the plaintiff's Facebook page that showed them engaging in similar kinds of boozy behavior. The lawsuit was ultimately unsuccessful for various reasons, but it taught the world a valuable lesson: curate your profiles carefully. It could be argued that all people had to do was edit their privacy settings to be more strict; however, Facebook's privacy settings were miserably designed and hidden away where you often wouldn't go looking.

Bad design aside, after the privacy transition tool debacle in 2009, we didn't have much faith that private information would remain private. As a result, we became much more cautious about what we shared, especially since our extended families and work colleagues were now on the platform with us. As for the Facebook team, the changes kept on coming. In 2012, to the surprise of absolutely no one, Facebook introduced ads directly into the News Feed, which were called Sponsored Stories.

Initially, they didn't perform very well and annoyed advertisers who weren't seeing the returns they expected, with some becoming completely disillusioned, stating that they were "fundamentally some of the worst performing ad units on the web." In order to fix this situation, Facebook stepped up their investment in the burgeoning field of machine learning. One team, led by a former Microsoft AI specialist, Joaquin Quinoneiro Candela, started producing some very powerful results by taking sophisticated user classifications, for example, men in their early 30s who like Bon Jovi, ice cream, and rugby, and combining them with brute force machine learning. This process started by simply aiming everything at everyone and continuously optimizing the results over time to maximize conversions. As the money started rolling in, the team continuously tested hundreds of machine learning variants to see which would make the most money. These tweaks mattered.

A successful tweak could mean additional hundreds of millions of dollars in revenue. Then, because these successes were so significant, they created a tool called FB Learner Flow, which allowed unspecialized coders around the company to launch machine learning experiments on other parts of Facebook, such as the News Feed. However, FB Learner Flow was a black box, with many engineers having no idea what the underlying machine learning technology was actually doing. One former employee referred to this as giving rocket launchers to 25-year-old engineers. This situation created a scenario where nobody really knew why or how things were appearing in people's feeds, as the rules that governed them were total abstractions only understood by the machine learning algorithms themselves, setting the stage for some major crises down the line.

Getting back to 2012, in May, another change came that had a dramatic effect on news outlets as well as innumerable other businesses that had created a Facebook presence by establishing a page to promote their services. The change was the introduction of promoted posts, which allowed anyone running a page to pay to boost their views. Bear in mind that Facebook had originally built pages to create a win-win situation. Facebook benefits from the diversity of content, which keeps people engaged for longer periods. News outlets and other businesses benefit because, if they do a good job, they can grow their audience at a rate that wasn't possible before.

Once that benefit was established, many news outfits—small, medium, and large—started dedicating their efforts and resources to growing a Facebook presence, becoming increasingly reliant on Facebook as a source of traffic and revenue. They became committed. However, promoted posts turned out to be a massive rug pull. Only two months after they were launched, the reach of organic posts, meaning posts that you don't pay for, started rapidly declining, leading to numerous accusations that Facebook had intentionally engineered this decline to force page owners to pay in order to maintain their reach. The reality, though, was a bit more complicated than this.

For example, with more and more news outlets entering the space, the competition to appear on users' feeds was becoming increasingly fierce, and those running pages could not reasonably expect stable growth forever under those conditions. However, the timing, speed, and scale of the decline of non-paid posts were quite suspicious. A TechCrunch article from around this period reported a 40% decline in average non-paid posts across the board over a duration of only four months, as well as an unconvincing response from Facebook representatives about why this was happening. The massive reduction in organic traffic sent untold numbers of indie news outlets scrambling to reevaluate their budgets or simply going out of business. On the consumer side, we saw less from the pages we had intentionally followed and more promoted posts we didn't care about, along with a slew of annoying ads that were often bafflingly irrelevant.

Incidentally, in 2011, Facebook did something similar with respect to gaming by instituting a 30% tax on all revenue earned on the platform. The social gaming company I worked for at the time couldn't recover from this tax and eventually folded. But it wasn't just indie games and news outlets; celebrities, musicians, and public figures who had built up a respectable audience were also having an increasingly hard time actually reaching their audience organically.

Moreover, Facebook had its own strange rules that dictated what people saw and what went viral, which changed constantly according to criteria no one understood. This situation persists even today. And this leads us to the trash: ever-increasing numbers of clickbait articles and spam, much of which is now being force-fed to us via our news feeds. This problem is made worse as Facebook alters its mechanics to maximize the effects of viral content. Reality.

Removing the limits on how many friends you could make in one day and eliminating restrictions on how many places you could cross-promote the same article were moves that benefited spammers. They could piggyback off posts that had already gone viral by simply copying and reposting them to as many people and in as many different spaces as they wanted. This led to not only an increase in the amount of low-quality content but also the repeated exposure to the same low-quality content. Learn unusual tricks that will help you lose weight.

This weird trick helps cure diabetes in 30 days. You'll find out how these tasty pastries contain a real ingredient that reverses diabetes. I do not know how long this video will be online. Your total investment today is now $19.99. Listen, don't even decide right now.

As social media became more prevalent, a new term began to be used in research and media coverage: brain hacking. This practice involves introducing mechanics that encourage frequent use, maximizing the potential for capturing people's attention. These mechanics include automatic emails that constantly notify you when any activity occurs, such as someone liking your post, for example.

Infinite scroll keeps loading more content, so you never get the feeling you've reached any kind of checkpoint where you can catch your breath. Videos that play automatically in your feed maximize the chance of catching your attention. And, my least favorite, annoying red notification medallions are all over the place and never go away. Even if you check your notifications and clear them, they still turn red again almost immediately to tell you something—anything—just to encourage you to click on them and lose a few more minutes of your day. Cook didn't invent half of these techniques, of course, but they employed basically all of them. In combination, they made the experience more exhausting and unbelievably less fulfilling.

So, all of this taken together—the notifications, the trash, the adverts, the insecurity about privacy, the engagement patterns, and the algorithm changes that rewarded either trivial or downright toxic posts—left me, and many others like me, to suddenly find that there wasn't very much for us on Facebook anymore. It had distanced me from my friends, who didn't see anything I posted, and vice versa. My news feed was full of random crap. I couldn't express myself honestly because I couldn't quite trust that Facebook wouldn't trick me into making it public, and even when I did want to post something, the new engagement rules that assigned a score to my efforts were pushing me toward inane one-linerism. That moment when I began my master’s in Glasgow, with the crazy darkness at 3 PM, not having made any friends yet, doing the long-distance relationship thing with my now-wife who was still working in London at the time, that's when I decided that Facebook was no longer a good place to be.

I preferred to just deal with the darkness and the loneliness on my own without Facebook wearing me down further with its relentless mission to serve me nothing but depressing, irrelevant garbage. And so, I became one of the zombie users of Facebook, logging in just to check if I had any messages and then leaving quickly. I wasn't the only one. In 2013, according to an article in The Information, the sharing of original personal posts had declined by 21%, and a year later, by a further 16%. It was now much more of a content consumption site that prioritized low-quality videos and news.

And just one last thing about change. While I was putting this video together, I decided to look back over my old timeline, starting right at the beginning. What I found was really weird and a little bit sad too. There were conversations that no longer make any sense because the people I had them with have since left the platform. So it's often just my responses.

It reads like I'm having a mental breakdown, talking to myself. This led me to a chilling realization: for me, all I have to look forward to is more of this. Facebook is eventually going to become a mortuary, a reminder of all my friends and family who've either left or died, where the evidence of our interactions is then scrubbed away like a Soviet-style rewriting of history. Oh look, another ad—deceased likes Budweiser.

That's not something I want to experience, and rather than leaving it in zombie form, I've now decided to cross the point of no return and delete my account entirely. That said, it was really 2013 when my story with Facebook ended. If this was where it all ended, the story would be one of a wasted opportunity and a major disappointment. But Facebook would roll on without me, and instead, the story would transition from disappointment to outright tragedy. So, let's begin this part of the story with a quick look at how Facebook was broadly perceived by the public in the early 2010s.

You see, even though more and more people were becoming frustrated with Facebook, many still viewed it through a positive lens, especially during the Arab Spring uprising, where activists in Tunisia and Egypt used it to organize protest events and provide the public with information that contradicted the narratives provided by official state media. The success of the uprisings, as well as the role that social media played in them, was widely reported around the world. One of the key organizers of the Egyptian protests, Wael Ghonim, praised Facebook for making the uprising possible. "You're giving Facebook a lot of credit for this?" "Yeah, for sure."

I want to meet Mark Zuckerberg one day and thank him, actually. The idea that Facebook could help people overcome censorship and produce positive change made an impression on a large number of people. Although Facebook had become something that was viewed with skepticism by civic-minded observers, the Arab Spring was an encouraging sign that convinced some highly talented leaders to overcome their skepticism and join the company, applying their skills to issues around security, privacy, protection of minors, counter-terrorism, and more. In 2010, Zuckerberg was named Time's Person of the Year. In 2012, Obama's re-election campaign once again successfully demonstrated the pioneering use of social media, especially Facebook, to reach voters.

Zuckerberg and Obama were seen together often around this time, most notably during a town hall meeting at Facebook's headquarters in Palo Alto. For sure, Obama's unique global popularity aided the perception that Facebook was a tool for good in the democratic process. Part of what makes for a healthy democracy is when you have informed and engaged citizens, and what Facebook allows us to do is ensure this isn't just a one-way conversation. Facebook's headquarters were buzzing with experts and researchers looking to understand whatever they could about how the platform affected society and the ever-changing nature of online behavior.

Much of this was done specifically to spot and avoid risks, and a lot of research was published internally for other employees to read and act upon. Unexpected issues occurred all the time, both big and small, and it was the job of various teams to identify problems and suggest solutions before they got out of hand. The only problem was that within the corporate structure, which was very much tied to improvements and core metrics, many of these civic-minded employees felt that they were second-class citizens compared to the teams tasked with achieving rapid growth. It is much easier to demonstrate that you brought in more users than it is to demonstrate that you prevented something bad from happening. To illustrate, let's return to the issue of trash content, a problem that was continuously growing throughout the early 2010s.

If you clicked on a junk science article, you would be recommended not just more from the same authors, but also related pages that peddled junk science. Facebook was so heavy-handed with these kinds of recommendations that many began to game the system by setting up multiple pages that all promoted the same type of content. This had a multiplier effect that, unbeknownst to Facebook, sent overall views for these content creators into the stratosphere. This was exacerbated by the blatant plagiarism I mentioned earlier. If you had no scruples, you could simply find a page with a large following and start copying and pasting its most viral content, as well as viral content from Twitter, YouTube, or Reddit. Viral content always has high potential to go viral again and again, so with Facebook failing to adequately enforce rules against plagiarism while simultaneously supercharging anything that went viral, you had the perfect environment for low-quality content.

Awards, no punishment. One of the best examples of this was the infamous troll farms, run in large part by teenagers in Macedonia. Through trial and error, these teenagers managed to figure out how to game Facebook's engagement metrics and were making ad money hand over fist by simply reposting whatever viral content they could lay their hands on. They amassed an enormous number of pages with high followings, contributing to a situation where the top 10 pages for popular subjects were often led by troll farm creators rather than legitimate ones. Troll farms overwhelmingly dominated the top 20 most popular Christian pages, including the most popular page of all, "Be Happy, Enjoy Life."

They also dominated the top 20 pages aimed at African Americans, including the top page, "My Baby Daddy Ainge." According to a report by data scientist Jeff Allen, our platform has given the largest voice in the Christian American community to a handful of bad actors who, based on their media production practices, have never been to church. Our platform has also given the largest voice in the African American community to a handful of bad actors who, based on their media production practices, have never had an interaction with an African American. Adding more fuel to the fire, Facebook made additional tweaks to their algorithm, giving more weight to comments and emojis like the angry face. This meant that if you saw content that made you really angry and you decided to comment on it, getting into a little argument along the way, you were more likely to see content from that same source again.

And since Facebook didn't pay enough attention to what was actually taking place on their platform, users' anger appeared to them to be simply an improvement in engagement on an engineer's analytics dashboard. Worse still, this produced a situation where tribal, polarizing content, which tended to get more likes, ended up drowning out moderate, thoughtful content. Wa'el Ghanim, the Arab Spring organizer we saw earlier, himself became very worried about these problems in later years. If you increase the tone of your posts and become more exclusionary against your opponents, you're going to get more distribution. So if I call my opponents names, my tribe is happy. This was the situation at Facebook when primaries were taking place in the run-up to the 2016 U.S. election.

More and more content about either Trump, Bernie Sanders, or Hillary Clinton started to go viral. Since trash content tended to make people argue more, highly partisan content, disinformation, or rage bait garnered more engagement. This, in turn, made it more likely to go viral and, consequently, more likely to be recycled by nihilistic Macedonian teenagers half a world away. One story claimed that Hillary Clinton was dying mid-campaign and had been replaced with a doppelganger. Another reported that the Pope was endorsing Trump for president. To make matters even worse, there was a more sinister way that Macedonian teenagers could monetize their pages. They could sell them, along with their massive followings, to third parties.

Third parties whose identities and motivations were a mystery. We'll come back to this in a moment. Other beneficiaries of Facebook's recommendation setup were radical organizations that learned they could draw large audiences by first luring them in with milder, more acceptable news pages, knowing that they would then be recommended far more extreme pages that they also owned—an efficient mass funnel for speedy radicalization. This technique was exploited successfully by all kinds of extremist organizations, including pre-COVID anti-vaxxer groups and ISIS. Highly populist politicians like Narendra Modi in India and Duterte in the Philippines also found this new variant of Facebook to be an excellent platform for amplifying hateful rhetoric in the lead-up to their elections.

It later became clear that misinformation campaigns had been successfully run in elections in Turkey and Indonesia as well. On the other hand, Facebook's targeted advertising capabilities became more prevalent in political campaigns. In 2013, a company called Aggregate IQ was established to help political campaigns understand and target specific voter segments more effectively through voter behavior analysis, demographic studies, and predictive modeling. By integrating this data with Facebook's advertising systems, they claimed to be able to direct political messages at highly specific target audiences much more effectively than had previously been possible. From 2015 to 2016, during the Brexit referendum in Britain, Aggregate IQ began working with four different Vote Leave campaigns, with the official campaign dedicating 40% of its budget to them.

The Leave campaign's most successful ads spread incorrect assertions about the cost of EU membership, as well as misleading statements about immigration and national sovereignty. When Britain ultimately voted to leave the EU, Aggregate IQ took as much credit as they could. On their website, they proudly posted a quote from the Leave campaign's chief strategist, Dominic Cummings. Without a doubt, the Vote Leave campaign owes a great deal of its success to the work of Aggregate IQ. We couldn't have done it without them.

Anyway, we've finally come to that point in the video I know you've all been waiting for, where we discuss the 2016 U.S. election, which produced arguably the most famous of Facebook's black eyes. Things are momentarily going to get a bit complicated here because there were actually four different controversial factors at play that are all related but also distinctly different. In many ways, this is the moment when all of Facebook's chickens came home to roost. The first factor is that Facebook had been developing services to sell to political campaigns, which included dedicated assistance from Facebook employees—an obvious move given the amount of money invested in advertising during election years. These services were offered to both the Clinton and Trump campaigns, but whereas the Clinton campaign largely turned down the offer, the Trump campaign took full advantage of it.

The second factor was Facebook's evolving attitude towards news. In the years leading up to the election, Zuckerberg had increasingly insisted that they were a completely neutral platform. This was largely in response to actual accusations from conservative politicians and news outlets that Facebook had a left-leaning bias. Zuckerberg's chumminess with Obama, as well as Sandberg's involvement with the Clinton campaign, only helped to fuel these accusations. With Trump ramping up anger about media bias throughout 2015 and 2016, this pressure became more pronounced.

But then, in May 2016, a scandal appeared out of nowhere in the form of a Gizmodo article titled "Former Facebook Workers: We Routinely Suppressed Conservative News." In the article, an anonymous Facebook employee described his frustrations working on a content curation team, which he said regularly removed news about conservative politicians from Facebook's trending topics section. Given that the 2016 election was only six months away, this story generated a massive outcry of conservative anger towards Facebook. In response, the trending topics team was let go, and Zuckerberg cracked down on all kinds of content curation. Then, much to the disappointment of many Facebook employees, influential conservatives including Glenn Beck and Tucker Carlson were invited to meet Zuckerberg at Facebook's campus so he could assure them of their neutrality.

After this, the whole company was put on notice. From now on, automated trending content was king. This leads us to the third factor, which I've already mentioned: the algorithm changes fueling the spread of polarizing content. This, in combination with now completely unmoderated news, meant that Americans were far more likely to see content that was trashy, unsubstantiated, crass, or simply untrue. Hillary Clinton had a secret child out of wedlock.

Hillary Clinton could coordinate a domestic terror attack on the U.S., etc. Of course, this content would then be copied and pasted by troll farms, including our pals, the Macedonian teenagers, which gave the most polarizing content second, third, fourth, and fifth lives. It wasn't all just alt-right fake news either; some fake news aimed to bolster Hillary's chances. All in all, in the three months leading up to the election, fake news articles favoring Trump were shared 30 million times on Facebook, whereas those favoring Clinton were shared 8 million times.

The number of people who saw all these shared posts was in the multiple hundreds of millions. Later, it was reported that over half of American adults who saw fake news said they believed the stories at the time. Groups also played a significant role in spreading misinformation. A friend of mine, Kat, who was studying misinformation in the lead-up to the election, joined multiple different Trump groups to keep an eye on them. Soon after, Facebook started recommending those same groups to her family and friends, pointing out, helpfully, that she was in them.

And then there’s the fourth factor: Russia. Let’s provide a little backstory here. In 2011, protests against Vladimir Putin were breaking out all over the Russian Federation, inspired in large part by Alexei Navalny, who used social media to organize protest events. This prompted Putin to launch a counter-offensive to disrupt the protesters’ ability to organize online and interfere with their communications in general. It was an investment in a new type of information warfare, where sowing confusion was the main goal.

In addition to this, he publicly accused the then Secretary of State, Hillary Clinton, of having run an influence operation aimed at the protests in Russia. You know, Russian voters deserve a full investigation of all credible reports of electoral fraud and manipulation. Later on, it was reported that a U.S. spy overheard a Russian military intelligence officer stating that they were going to get Clinton back for what she had done. Now, when we think of the 2016 election and Russia, we think of hackers, and that's certainly not off the mark.

Traditional hacking did play a significant role. In early 2015, Facebook's threat intelligence team discovered that Russian hackers were trying to gain access to the accounts of figures involved in the upcoming U.S. election. Then, a few months later, U.S. cybersecurity firms published a report claiming that hackers with ties to Russia had successfully hacked the email accounts of various senior figures in the Democratic Convention. Soon after, a threat detection analyst at Facebook named Ned Morin used the information in these reports to uncover a Facebook page called DC Leaks, which he suspected was linked to the email hack. He began to keep a close eye on DC Leaks by monitoring their activities, including reading their private instant messages.

From this, he discovered that the hackers were contacting us journalists on Facebook and leaking emails directly to them in order to embarrass the Clinton campaign. At this point, Trump would then issue one of the more astonishing statements of the campaign. "It shows how little respect they have for our country when they would hack into a major party. But it would be interesting to see. I will tell you this."

Russia, if you're listening, I hope you're able to find the 30,000 emails that are missing. I think you will probably be rewarded mightily by our press. Once Moran had confirmed his findings, he then sent reports about the problem up the chain of command, intended to eventually arrive at the desk of one of the then leaders of the company, Elliot Schrage, the VP of Global Communications and Public Policy. Schrage is an interesting individual. Most accounts I've read paint him as a domineering company man.

If I press a like button on a brand, it could pop up as a sponsored story. I might not be aware of it, and I certainly wouldn't necessarily agree to it. When you press a like button on a brand, you're expressing that you like it, but you're not saying that you endorse it.

You're asking a profound question: What is advertising? On the Facebook system, I'm affirmatively communicating that I'm associating myself with whatever I like. Schrage will pop up a few more times in this video; however, for now, it will suffice to say that the Russian hacking report intended for him did not appear to reach either Zuckerberg or Sandberg. It has been suggested numerous times that Zuckerberg's strong resistance to curating political content acted as a disincentive for top leaders to bring this kind of information to him.

Now, just in case you didn't know, I should probably mention that Trump went on to win that particular election. But even before that happened, stories were already emerging about Facebook due to the amount of lies and polarizing content that people were experiencing on the platform. Soon after the election, Zuckerberg was asked about the fake news problem. His answer made headlines around the world. "You've been getting a lot of pushback from people who feel that you distorted the way that people perceived the information during the course of the campaign because you didn't filter out enough fake stories that might have been published simply to gain traction and sell advertising."

You know, I've seen some of the stories that have been published simply to gain traction and sell advertising. I've seen some of the stories you're talking about regarding this election. I think the idea that fake news on Facebook influenced the election in any way is a pretty crazy notion. Soon after this, when Zuckerberg finally found out about the Russian hacking taking place on Facebook, he was reported to have said, "Oh f***, how did we miss this?" It's not exactly clear how he didn't know about it, and there's never been a transparent explanation as to where exactly in the chain the communication broke down.

Moreover, it was reported that Facebook dragged its feet on the idea of disclosing their findings to the public because they knew how damning the story would be. When it comes to Russian hackers trying to gain access to high-profile accounts or trying to disseminate hacked information, this is the good old-fashioned stuff. The kind of hacking that Facebook was used to and which, once identified, could be prevented. However, hacking was only the tip of the iceberg. Ned Morin and his boss, Alex Stamos, who had been trying to sound the alarm about Russian meddling for months, both believed there was much more going on than just hacking.

Later, they would tell journalists that their fears were largely dismissed by Schrage, who they believed was taking direction from Sandberg herself. Everyone seemed to want the story to go away, but it wouldn't die. Finally, an article published by Time magazine in 2017 blew the story wide open, confirming what Stamos had been warning about for months: that a Russian-backed campaign of disinformation had been employed on Twitter and Facebook in order to influence the outcome of the election. This pressured both U.S. intelligence agencies and Facebook to work together to uncover the extent of the interference, prompting Zuckerberg and Sandberg to give the thumbs up to Moran and Stamos to finally uncover what had actually taken place on their platform. And after a few weeks of searching, they got their answer.

The main culprits were a group known as the Internet Research Agency, which was run out of St. Petersburg. They weren't using some hyper-sophisticated form of hacking. No, they were using the exact same tactics as the Macedonian teenagers: troll farming. They built up numerous pages with names like "Secured Borders," all sharing the same content in order to game Facebook's engagement algorithms. They spread highly polarizing content on key voter issues, which drove people insane and made them argue.

Which Facebook then misrecognized as virtuous engagement and boosted even further. In order to siphon as many voters away from Clinton as possible, the pages favored not just Trump but also Bernie Sanders, a rather clever tactic to spread bitterness among more left-leaning voters. Worse still, Russian troll farms were using Facebook's own advertising system to buy cheap page-like ads, which, as the name suggests, encourages people to like your page. Once they do this, they'll start seeing more of your content in their feeds. It was an incredibly cheap way of scaling up so your disinformation articles and memes are much more likely to go viral organically. And to cap it all off, when anything the Russian troll farms posted went viral, you could be sure that other troll farms would plagiarize it for ad revenue, including our dear old friends, the Macedonian teenagers.

When this report was delivered to Facebook's leadership, as well as to U.S. intelligence agencies, they were appalled. It turned out that the Internet Research Agency owned 120 different pages and posted 80,000 pieces of content, reaching a total of 126 million Americans. The exact degree to which the Russian disinformation campaign contributed to Trump's election is unknown, but when this report was released to the public, it set off an unholy firestorm of bad public relations for the company. It also prompted a relatively new question in the minds of many people, a question that has dogged the company ever since: Is Facebook a net negative for the world?

Hearings on social media manipulation began in autumn 2017, which was attended by representatives from Twitter, YouTube, and Facebook. Separately, Sheryl Sandberg visited Washington to discuss steps Facebook could take to prevent such incidents in the future. Then, prepare yourselves for déjà vu: Elliot Schrage directed a PR company called Definer's Public Affairs to help navigate all the bad press. They, in turn, began publishing news articles to try and spread the blame, particularly towards Google again. They also looked for ways to discredit George Soros when he became critical of Facebook as well. But just keep that in your head a moment longer.

As for Ned Morin and his boss, Alex Stamos, they were dispirited throughout this entire ordeal. They tried to sound the alarm but had largely been ignored. However, they did have at least one moment of light. They discovered that a coordinated Russian attack was taking place to disrupt France's upcoming election, but this time they were able to tip off the French authorities in advance. This meant that when the hackers suddenly released 20,000 emails from the Emmanuel Macron campaign just two days before the election, security officials were able to immediately inform the public about where the hacks had come from, which led to news outlets advising voters not to believe what they read.

This was an example of where Facebook could do better, as long as executives like Schrage, Zuckerberg, and Sandberg allowed them to. But then, in 2018, just when we thought this issue was beginning to wind down, unbelievably, almost inconceivably, it actually managed to get worse. That's right, there's actually a fifth factor. Let's wind back for a second to understand how it happened.

Remember earlier I described numerous scandals related to the Open Graph and how it granted third-party developers far too much access to users' personal data? Well, in 2014, an academic named Alexander Kogan accessed the Open Graph to plug into an app he had developed called "This Is Your Digital Life." The app was presented as a personality quiz, and it collected personal data from users who installed it. Almost 300,000 Facebook users participated in this quiz, from which Kogan was able to extract not just their data, but also that of their friends, ballooning the overall dataset to almost 90 million people. Kogan shared this dataset with a company called Cambridge Analytica, whose trade was to create what they called psychographic profiles of voters in order to target them with personalized political messages.

Then we add a unique extra layer of data about personality, decision-making, and motivation. This creates an unparalleled, rich, and detailed view of voters and the issues they care about. We call this behavioral micro-targeting. Our techniques are working across the globe, from local elections to the U.S. 2016 presidential race. Cambridge Analytica's then-Vice President, Steve Bannon, eventually left his position to become the Chief Executive of Trump's 2016 election campaign.

Campaign. He then hired Cambridge Analytica to work for the campaign and produce targeted advertising. When The New York Times and The Observer in London broke the story that Facebook had allowed users' private data to be stolen and that this data was subsequently utilized by the Trump campaign for political advertising, the world's media went into a frenzy. It wasn't long before politicians were insisting that Zuckerberg himself testify before Congress about this. In the UK, the scandal had an interesting counterpart when it was reported that Cambridge Analytica had extremely close ties to Aggregate IQ, which, as I mentioned earlier, had been utilized by the Leave campaigns for their targeted advertising. According to former Cambridge Analytica employees Brittany Kaiser and Christopher Wylie, the connection between the two companies was so close that Cambridge Analytica considered Aggregate IQ to be one of their departments and that they shared the same tactics and technology.

This kicked off a separate firestorm in the UK about the use of stolen data to influence the Brexit referendum. For Sheryl Sandberg, this was especially devastating because, as the person ultimately in charge of public relations, it was the second time she had failed to head off a major scandal in the space of a year. However, this seems slightly unfair. Sheryl's more notable failure was how she and Elliot Schrage failed to notice the Russian hacking and disinformation campaign despite Facebook employees trying to inform them about it. The Cambridge Analytica scandal, however, was a direct result of the Open Graph, which was very much Mark's baby.

Elliot Schrage would then go on to do something even more stupid. When Sheryl Sandberg went to testify during congressional hearings, the public relations firm he had hired, Definers, started digging up dirt on the senators questioning Sandberg, spreading the information they had obtained to reporters in an attempt to portray the hearings as politically motivated and not based on genuine public concern. When this story broke, it further destroyed Facebook's credibility, spelling the end for Elliot Schrage, who had by then demonstrated such consistently bad judgment that he felt obliged to offer his resignation. This also damaged Sandberg's reputation, from which she would never fully recover. The Cambridge Analytica scandal produced unprecedented public ire towards Facebook.

The #DeleteFacebook hashtag began trending on Twitter, with high-profile celebrities like Cher announcing that they had deleted their accounts, prompting many others to follow suit. Amidst all of this, Facebook's stock also plummeted by 10%. This brings us to the significant moment when Zuckerberg finally testified before Congress for the first time. To prepare for this, Zuckerberg held mock hearings with a team of litigators who trained him on his responses and equipped him to handle tactics such as grandstanding, aggression, and unexpected questions designed to throw him off his talking points. They also took other precautions.

For example, since Mark had a history of sweating during interviews, his staff insisted that the AC be turned down low during hearings. When the day finally arrived, Mark's testimony was calm and assured. However, most privacy advocates felt that the questions addressed to him were weak. Worse still, some lawmakers displayed a notable lack of technical literacy. So, how many data categories do you store?

Does Facebook store data? Regarding the categories that you collect, Senator, can you clarify what you mean by data categories? There have been some past reports indicating that Facebook collects about 96 data categories. So, how many of those does Facebook actually store?

Do you store any? Senator, I'm not actually sure what that is referring to. Can somebody call you up and say, "I want to see John Kennedy's file?" Absolutely not. Could you, if not, not? Could you, not would you do it, could you do it?

In theory, do you have the right to use my data, assign a name to my data, and share it with somebody? I do not believe you have the right to do that. If I'm emailing within WhatsApp, does that ever inform your advertisers? No, we don't see any of the content in WhatsApp.

It's fully encrypted, right? But is there some algorithm that sends information to your ad platform? For example, if I'm emailing about Black Panther within WhatsApp, will I receive a Black Panther banner ad? You mentioned before that Facebook would always be free.

Is that still your objective? Yes. Well, if so, how do you sustain a business model in which users don't pay for your service? Senator, we run ads. I see.

Consequently, this shifted the media's focus away from Facebook's wrongdoing and instead onto how clueless the senator was. Congress wanted to teach him a thing or two, but it turns out they didn't know much at all. As a result of Zuckerberg's performance, Facebook's stock recovered significantly, increasing his net worth by around $3 billion in just 24 hours. However, the public relations damage still remains to this day, with the name Cambridge Analytica serving as a byword for Facebook's cavalier attitude toward users' data.

Before we move on, I want to mention one funny twist in this last scandal: Cambridge Analytica's supposed secret sauce, their psychographic profiling, was a totally useless service and was heavily criticized by many experts in polling data and psychology due to the lack of evidence and transparency provided. Even Alexander Kogan, the researcher who had shared the data with Cambridge Analytica in the first place, knew the approach didn't work. According to him, their profiling data was only accurate 1% of the time. Worse still, by the time the 2016 election rolled around, the stolen data they had acquired was already two years old and nowhere near as detailed as the real-time treasure trove Facebook themselves had access to. Given that Facebook already offered highly calibrated targeted advertising to politicians, the irony here is that Trump's campaign would have been far better off ditching Cambridge Analytica entirely and simply doubling down on the much more powerful targeted advertising they were already using Facebook to produce.

Had they done that, they would have wasted a lot less money. Facebook would have been spared the scandal to end all scandals, and users' personal data would have been utilized even more effectively to sway the election. From this point in the video onwards, the topics discussed are, at times, going to get much darker. Although I promise to avoid explicit or gruesome detail, I still think it's wise to issue a trigger warning here. In August 2013, Facebook announced a new service called Intranet.org, which would bring free internet access to underserved communities around the world, particularly in developing countries. They would do this by partnering with mobile operators and other technology companies across Latin America, Asia, and Africa, providing a stripped-back version of the internet that could be served at low bandwidths.

At the time, it was reported that this was the project Zuckerberg was most excited about. When people are connected, we can accomplish some pretty amazing things. We can gain access to new jobs, opportunities, and ideas. However, technology must serve the whole of society. Connectivity can't just be a privilege for a select few of the rich and powerful.

It needs to be something that everyone shares. So how did it work in practice? Well, when you accessed Internet.org, you were provided with a few free services. Most importantly, a simplified version of Facebook and Facebook Messenger, as well as Wikipedia, BBC News, and the forecasting service AccuWeather. It also included a few local services too, depending on the region.

However, Internet.org was not a fully free internet service. If you wanted to access any site outside of these core offerings, you would hit a paywall. To give an example of how all this combined, let's take Indonesia. At the time, the free search engine they provided in Indonesia was Ask.com, but if you tried to access any of the listings on Ask.com, you would be prompted to pay for data. Criticism of this idea appeared almost immediately, mainly focusing on how Internet.org broke with the principle of net neutrality, which holds that certain service providers and governments should treat all internet traffic equally, thus ensuring a level playing field.

To give some specifics, internet service providers should not grant higher speeds to certain services, as this could stifle fair competition by favoring established players over startups. Secondly, the entire internet should always be accessible. Making some parts of it paid again tips the scales in favor of the powerful, whose services can be offered for free, making it harder for smaller entities to compete. Lastly, internet service providers should never limit freedom of expression, especially regarding how individuals choose to communicate. Such barriers to access could be manipulated by those in power to discriminate against voices they do not agree with.

All in all, many pointed out that Internet.org was not free access to the internet; it was free access to Facebook’s network, a walled garden where Facebook gets to decide what people can access. For this reason, 65 NGOs from around the world published an open letter criticizing Internet.org as a poor attempt to bring the developing world online. Apart from the concerns I have already mentioned, it included strong opposition to the idea of a two-tiered internet that favored wealthier citizens. It also outlined some security concerns, indicating that the technology opened the door for users' activities to be monitored by their governments, as well as grave concern that many of its users would not understand privacy consent due to a lack of education on the topic and high levels of illiteracy among poorer populations. Some called the effort digital colonialism due to the overwhelming priority given to Western services over local ones.

Now, few believed this project was motivated purely by philanthropy. The benefit to Facebook of being installed as the core communications platform for potentially billions of additional people was obvious. It was the growth tactic to beat all growth tactics. Just to remind you one more time of that Web 2.0 quote: the race is on to own certain classes of core data, such as location, identity, and calendaring of public events. Internet.org was set up so that if anyone wanted to sell a service, blog about something, or message a friend, their only real option would be to do it on Facebook.

Thus, Facebook creates an ecosystem and, therefore, a dependency on their service that would persist even if an open internet came into existence. Now, some observers might feel that Facebook's initiative is not being treated fairly here and that these are acceptable trade-offs given the potential benefit to those who would otherwise have no internet at all. That's a reasonable position, and I will address it in just a few minutes. But for now, let's look at some of the practical problems faced by Internet.org when it was launched. The most obvious was that many of the services were only available in English.

Another, more subtle issue was that to newly online users, Facebook was largely considered to be the internet. According to Hmong Palatine, who studied the Philippines version, this misleads the public and makes them think that these sites and services are the essential tools. This would turn out to be a significant problem. A little later, Facebook rebranded Internet.org to Free Basics, which they said better communicated the service being provided. Their number one target? India.

In order to establish Free Basics in India, Mark engaged in a PR campaign, visiting local villages and cities and meeting with various politicians, including the Prime Minister, Narendra Modi. Here in India, 243 million people are connected to the internet, and more than 100 million are on Facebook already. However, there are still more than a billion people in India who cannot connect and do not have access to the same opportunities as everyone else. Unfortunately, enthusiasts began to sour on the idea when they finally saw what Free Basics really looked like. There were no other social media apps available besides Facebook, Bing was used instead of Google for search, and access was limited to only 36 sites, all of which were chosen by Facebook.

This prompted fierce resistance from privacy and net neutrality advocates around the country. What Zuckerberg means by "internet for all" is essentially Facebook for all, along with a few non-profit services thrown in to give it the appearance of philanthropy. Soon after, net neutrality became a national issue, with intense protests coming from all over the country. Ultimately, despite significant lobbying from Facebook, the Telecom Regulatory Authority of India sided with the net neutrality advocates, rejecting Free Basics outright. At the time, this was seen as a massive blow for Facebook.

But just to make that blow a little bit worse, one of Facebook's board members, Marc Andreessen, decided to get really angry about it on Twitter. Anticolonialism has been economically catastrophic for the Indian people for decades. Why stop now? Very good. I'm sure Zuckerberg and Sandberg adored that particular contribution, especially how it likens Facebook to colonialism.

Nice one. As you can imagine, this was reported everywhere. He later apologized for it, but who cares? In 2017, a damning report from Global Voices summarized Free Basics as not being particularly useful. They observed that of all the versions they tested, none met the linguistic needs of their users and that it was sorely lacking in sites for local services or independent news sources.

But behind the scenes, unbeknownst to most people, worse things were happening. In particular, Facebook seemed to be under-investing in local moderation teams. According to one former employee, “I can’t recall anyone at the company directly questioning Mark or Cheryl about whether there were safeguards in place or raising something that would qualify as a concern or warning for how Facebook would integrate into non-American cultures.” This would eventually open the door for all kinds of problematic content to be shared, with too few people working at Facebook knowing about it. And all of this together—the relentless drive for growth and speed, with insufficient attention paid to safeguarding the site to protect its users—set the stage for a truly horrific event.

Facebook's relationship with Myanmar began in 2010. When it was launched, it allowed users to access its platforms without incurring data charges, which helped it rapidly rise in popularity. In 2013, the government in Myanmar, under Thein Sein, introduced some moderate reforms to the country, including allowing mobile operators in for the first time. This, in turn, prompted a flood of cheap mobile phones into the market. These changes produced the perfect conditions for Free Basics, which was eventually introduced in 2016. To many people in Myanmar, Facebook quickly became the most trusted source of news, so powerful that a print magazine was even created for those not yet online, allowing them to read about various popular Facebook posts each week.

However, beyond the technological changes taking place, ethnic tensions in Myanmar had been escalating over the same period, with numerous leaders trying to turn the country's Buddhist majority against its Muslim population, particularly targeting the Rohingya, an ethnic minority from the northern Rakhine State. This involved spreading stories about the Rohingya's supposed misdeeds. There were warnings about new attacks that the Rohingya were planning, but which never happened. All kinds of fake science stories were spread as well. For example, a claim that Rohingya DNA was distinctly different from that of other human beings.

And this was coupled with one of the most worrying telltale signs that likened the Rohingya to vermin. In addition to all of this, there was a new problem: all of this information was spreading like wildfire on Facebook. There were several reasons why Facebook was so powerful in spreading hate messages in Myanmar. The first is that, in comparison to the heavily censored state media, Facebook was seen by the population as a source of legitimate news.

Here’s the text reformatted with correct capitalization and spelling: News that came from the real people on the ground indicated another reason: a sizable portion of the population, having only just come online, was not yet experienced enough to recognize the tactics and language of online disinformation. Asseline, as you can imagine, was Facebook's algorithms, which unwittingly promoted hate speech and disinformation. Lastly, a crucial oversight allowed for the unmitigated disaster that was about to unfold—a serious lack of content moderation. In fact, as early as 2014, Matt Schisler, a doctoral student at the University of Michigan, had been trying to alert Facebook to the situation in Myanmar, eventually being promised a direct line to one of their working groups.

However, this group would turn out to not be particularly useful, as we'll see. There were numerous acts of violence and outrage from 2014 to 2017, including a riot triggered by a viral Facebook post that falsely claimed two Muslim men were to blame for the assault of a Buddhist girl in the city of Mandalay. Many Rohingya were displaced during the violence that occurred as a consequence of that lie. Later, Schisler noticed a dangerous post beginning to spread about an elderly man who had taken pity on the displaced Rohingya and who had been photographed delivering rice to them. The posts accused this man of being a traitor, prompting calls for him to be punished.

Schistler later recalled that when he reported this to the Facebook working group, it took weeks to get a response. When that response finally came, they stated that there was nothing they could do unless the individual himself reported the comments personally, which was impossible because he didn't have a Facebook account, and Schistler had no way to contact him. In response to accusations that they weren't doing enough, Facebook developed sticker packs with statements like "Think Before You Share," which users could apply to posts and comments. Theoretically, these would help Facebook moderators spot extremely harmful content. However, it was reported that these stickers had an unintended side effect.

Here is the text reformatted with correct capitalization and spelling: Which was that the algorithm counted them as a sign of engagement. So, when applied, they actually amplified hateful messages even further. Around this time, Schisler would discover that Facebook had only one moderator working out of the Dublin office. One moderator who only spoke Burmese in a country of around 54 million people and over 100 languages, not to mention dialects. By 2015, the number of moderators was increased to only four.

Again, all Burmese speakers. Facebook's automatic detection systems regularly miss extreme hate speech due to problems recognizing Burmese script. One such post, translated to English, was the nonsensical statement, "I shouldn't have a rainbow in Myanmar." However, the original post in Burmese script was nothing like that: "All the Muslims that you see in Myanmar, none of them should be left alone."

By 2017, with 15 million people now on the platform, the Myanmar security services began a massive social media campaign to convince the public that it was time to wipe the Rohingya from the map. When it began, it was reported that Facebook had almost no idea this was happening. The leader of Myanmar's military at the time posted, "We openly declare that absolutely our country has no Rohingya race." What followed was a systematic campaign to burn the villages of the Rohingya. Official reports state that around 24,000 Rohingya were killed during the campaign, including children, with another 700,000 fleeing to dangerous and disheveled refugee camps in Bangladesh.

A UN investigation announced that it believed Facebook played a leading role in the amplification of hate speech. A report from Amnesty International stated that Facebook's dangerous algorithms, which are hardwired to drive engagement and corporate profits at all costs, actively fanned the flames of hate and contributed to mass violence. As Mohammad Shawif, a Rohingya activist, put it, "The Rohingya just dream of living in the same way as other people in this world, but you, Facebook, you destroyed our dream." Since then, Facebook has faced similar criticism for amplifying hate speech in other countries, most notably in Ethiopia. To give one example, according to an investigation by Vice in 2020, it was a central source of disinformation that led to ethnic violence in the region of Oromia, resulting in hundreds of deaths.

Warnings about tensions in Ethiopia persist even today, a situation exacerbated by the complexity of the culture, which includes around 80 different languages. In 2021, two posts appeared on Facebook condemning a man named Mehreg Amare, a 60-year-old Tigrayan professor living in northern Ethiopia. The posts wrongfully accused him of helping a paramilitary group commit atrocities during an armed conflict. These posts also included pictures of Amare's home, with multiple people commenting on the post, suggesting that he be harmed. Trusted partners in Ethiopia, tasked with notifying Facebook about such issues, stated that Facebook refused to take the posts down because it didn't want to restrict free speech.

Soon after, Amara was murdered outside his home. The posts that caused this remained up on Facebook for a further year, and his family is now suing Facebook for $2 billion. The case my clients have made is that not only does Facebook allow such content to be on the platform, but they also prioritize it and profit from it. I hold Facebook directly responsible for our father's murder. Since then, the trusted partners I just mentioned have ended their relationship with Facebook, and multiple Ethiopian observers have reported that they feel there is no point in bringing concerns like these to the company because they do not take any action.

Facebook actually has a process called "break the glass," which they use in highly dangerous situations, allowing them to take significant measures to prevent the spread of inflammatory information. However, in Ethiopia, according to Amnesty International, there is evidence suggesting that Zuckerberg personally intervened to stop mitigation measures from being applied because the measures may have interfered with the meaningful social interaction metric. There are many more stories like this, but for the sake of time, I’ll move on. Facebook has since announced a significant investment in local moderators but will also need to establish other safeguards to spot dangerous trends before they get out of hand. At the time of writing, the situation in Ethiopia is still highly uncertain, with news reports about Facebook's presence in the country popping up regularly.

This is one to keep an eye on. The original critics of Free Basics, who argued that it broke the principle of net neutrality, could hardly have predicted how bad the situation was going to get, with Facebook becoming the sole source of news and crowding out local sources. This alone is validation of their criticism, but I want to add one more thing to it. One thing to bear in mind is the idea of a company that can become so powerful that it dominates conversation in general. Facebook has made innumerable terrible mistakes throughout its history, as we've seen and as we're going to see.

And every single time they've been exposed, it was published via an external source outside of Facebook: a privacy advocate report, a tech blog, a whistleblower publishing on a different platform, an open letter, or, most commonly, a news report by The Wall Street Journal, The New York Times, The Guardian, BuzzFeed, etc. A service like Free Basics, if it can gain enough traction in a given country, automatically disadvantages external criticism of Facebook by hiding it behind a paywall and crowding it out with their own free content. If Facebook became your sole source of information, you'd never hear any criticism of Facebook. This situation is so unacceptable that I believe Free Basics should be rejected wherever it is encountered.

Okay, so to do this section justice, we need to look back at one last aspect of the platform I haven't yet discussed. Remember Tim Urban's blog, which I mentioned earlier, about the different kinds of posts he felt were self-serving and therefore insufferable? Well, in that blog, he proposed five negative motivations that lead people to post what he considered to be insufferable stuff. These motivations were image crafting, trying to make others feel jealous, craving attention, and narcissism. But it's the fifth motivation that slightly rubs me the wrong way.

Although I have to say, this was written in 2013, the article is obviously tongue-in-cheek, and I don't think Tim was intentionally being spiteful here. The fifth motivation was loneliness. Seeing a lonely person acting lonely on Facebook makes me and everyone else sad. So the person is essentially spreading their sadness, and that's a lousy thing to do. This one caught my attention because, well, it's a bit mean.

Oh, lighten up, will you? Your sadness is really killing my buzz. It's a pretty surefire way to make a lonely person feel even more lonely. Now, as I mentioned earlier in this video, sometimes, when you see content that’s intentionally jealousy-inducing—examples of how great other people's lives appear to be—it can have a destructive effect on your sense of self-worth. My wife, who works as a psychotherapist, has been noticing, along with her colleagues, an uptick in how social media plays a role in people's negative perceptions of themselves and their accomplishments as far back as 2010.

But at the time, there were obvious positives too. In 2007, an influential study investigated the potential Facebook had for developing social capital among college students. It pointed out that it could increase one's network of peers, where online relationships eventually turn into face-to-face relationships. This aspect of Facebook has not gone away. In May of this year, I went to a reunion gig for the band Cardiacs, which took place in London.

After the show, when we were all hanging around outside the venue, I learned something I thought was really sweet. A few young fans who couldn't afford to stay overnight in London have been offered beds by some of the local organizers who just wanted to help others have their first Cardiacs experience. All of this was organized on Facebook. Apart from that, ongoing research leading right up to the present day suggests that some neurodivergent people can benefit from Facebook, which, due to its structured and predictable communication system, helps to facilitate positive social interactions. This is in part because it removes some of the complex social cues involved in talking to someone face-to-face. However, going back to the early years again, the positive stories were beginning to get outnumbered by the negative ones.

For example, how Facebook was being used as a tool for bullying, as well as other concerning developments. One well-known study from 2013 found that heavy use of Facebook was strongly correlated with the decline in users' sense of well-being. Later, another widely shared study found a correlation between time spent on social media and feelings of social isolation. Both studies acknowledged the benefits I just mentioned but also agreed that something unhealthy was going on and that it seemed to be getting worse. When trying to understand why this was happening, I found it interesting to read even earlier research, back when we were all trying to figure out what social media was for and how it could change people.

And I think it's important to look at this for a second, back to first principles, as it were. To channel the sociologist Morris Rosenberg, we all have something called a self-concept, which is our understanding of how we appear to those around us. One of the core aspects of our self-concept is our identities, which we construct through a kind of validation process, a social dialectic. It starts with an identity announcement, where we put the concept of our identity out into the world by talking to someone, wearing certain clothes, attending a certain type of social gathering, etc.

Then, in return, identity placement occurs, where those around us accept or reject the identities we're trying to project, as indicated through their actions, expressions, or words. When we're talking to someone face-to-face in the real world, we're aware that the identities we're projecting need to be, to some extent, consistent with how we look, how we speak, how confident we are, as well as a host of other traits; there's a performance and appearance burden needed to pull it off. However, the identities we construct online are not restricted in this way. We can withhold unwanted information about ourselves, hide things about our appearance that we don't like, and role-play. On Facebook, which is all about projecting our identities, we can curate exactly what others see, allowing us an opportunity to validate a hoped-for possible self.

An ideal self. This is why the like button, when it was introduced in 2009, was so consequential. Let's say you post something that's very important to you and three people comment on it positively. Then, a person you know posts something else and it gets five positive comments. No biggie.

But if we add likes, you have four, and they have fourteen. Now it seems a little bit like your identity announcement was a little less validated by your peers than this other person’s. But then Facebook decides that the post with fourteen likes is rather popular, so they choose to show it to more people. This means that when you check back later, you see the tally has changed. You still have four, but this other person now has fifty-nine.

You've been obliterated. A 2010 study found that users with low self-esteem or narcissistic traits tended to be more active on Facebook, tended to post more self-promoting content, and were much more likely to manipulate their photos before posting them. You still have four likes, but this other person now has 59. You've been obliterated.

Other studies conducted experiments to understand the nature of social comparison on Facebook. In one experiment, female participants browsed the platform, noting the attractiveness of others, and reported feeling bad about their own appearance by comparison. Another study indicated that when participants were depressed, they were more likely to compare themselves negatively against others, leading to feelings of envy. All of this seemed to point to a significant problem. While Facebook appeared to be creating a refuge for people with mental health issues, it also seemed to be developing mechanisms that could exacerbate those problems.

Then there's the second major issue that was beginning to cause concern – children. In 2011, Consumer Reports released an estimate that Facebook had 7.5 million users under the age of 13, a finding that alarmed advocates of child protection. In 2012, Bill Price, the chairman of Common Sense Media, a child advocacy group, visited Facebook's headquarters to meet with Sheryl Sandberg and Elliot Schrage. Price's aim was to try and partner with Facebook to help strengthen protections for children.

However, Sandberg and Schrage used the meeting to express anger over a book that Common Sense Media had just published, titled *Talking Back to Facebook,* which gave advice to parents on how to protect their children from the worst dangers of social media. Price later recalled that when he suggested tools aimed at improving the safety and mental health of teenagers, Schrage dismissed him out of hand while also making the threat that Facebook could sever business relationships with any tech investors on Common Sense Media's board. Price later said, "I left the meeting feeling pretty depressed. It's sort of a typical story of where ultimate power corrupts people's perspectives." And then, in that same year, Facebook acquired Instagram.

The acquisition of Instagram was a big deal, causing concern among regulators who worried about the consolidation of social media power under one company. However, it was ultimately approved by the FTC a few months later. The final price tag, $1 billion, raised quite a few eyebrows, considering that Instagram earned no revenue and had only 13 employees. With the benefit of hindsight, we can see that it was an incredibly strategically valuable asset, given its emphasis on imagery rather than text, its rising popularity among teens, and its dominance on mobile devices. In its early years, Instagram was reasonably uncontroversial, with Facebook generating most of the bad headlines.

For example, in 2017, a 12-year-old girl live-streamed herself ending her own life on Facebook. A journalist for The Guardian wrote at the time that when she first saw this video, she flagged it as inappropriate content, only to receive the following response: "We reviewed the share you reported for showing someone injuring themselves and found that it doesn't violate our community standards." This video remained up for a further two weeks before they finally bowed to the pressure to remove it.

And although this isn't related to children or mental health, I have to also mention here the incident in 2019 where a white nationalist in New Zealand used Facebook to live stream his attack on a mosque, one of the worst and most savage mass shootings in history. Now, as for Instagram, its grace period began to significantly erode around 2017. A survey conducted at this time found that over one in five teenagers aged between 12 and 20 experienced bullying on Instagram. Also in 2017, a UK report on the rise of anxiety and depression among teenagers caused a stir when the research indicated that Instagram was actually the most negative of all the social media platforms. What was going on?

Well, in 2017, an event occurred that provided a brutal answer and shaped how the platform has been viewed in the UK up to the present. It began when a 14-year-old girl named Molly Russell, who was very active on Instagram, took her own life. In the inquest that followed, her phone was analyzed, revealing that over the duration of six months, she had seen more than 2,000 pieces of content related to self-harm or depression on the platform. Official reports noted that some of this content was recommended to Molly based on her activity and not specifically requested by her. The psychologist who reviewed the content Molly had been looking at reported being unable to sleep for a few days afterward. When the coroner's report finally arrived, it contained something that had never been seen before—a world first: the inclusion of social media in the official cause of death. Molly Rose Russell died from an act of self-harm while suffering from depression and the negative effects of online content.

Media reporting on this event heavily focused on the tendency of Instagram's algorithms to notice when users viewed dangerous content and then recommend even more of it back to them. Various studies emerged that appeared to support this claim. According to one study, among those who were exposed to self-harm content on Instagram, only 20.1% indicated that they had intentionally searched for this content. Indeed, the majority of users seem to encounter Instagram's self-harm content unintentionally. Other research found that hashtags concerning self-injury rose from 63,000 to over 110,000 throughout 2018.

Then, in 2019, in response to the Molly Russell case in particular, Instagram announced that it would no longer allow explicit images of self-harm on their platform. However, at the same time, a related form of content was exploding on Instagram: the glorification and promotion of what I will just call EDs, which was being spread through a variety of popular hashtags that I won't name here. Apart from highly disturbing and bleak imagery related to EDs, participants in this kind of content would often publicly set themselves incredibly dangerous target weights. As the phenomenon grew bigger, ED coaches began to emerge, ostensibly to help others hit these targets, encouraging them to join private groups where they could delve into even more horrendous detail.

After heavy media reporting on this topic for years, Instagram finally began to crack down on it. However, it wasn't plain sailing. Even by 2021, journalists were still able to demonstrate, by creating new accounts and learning some of the new hashtags, that it was still relatively simple to access this type of content. They pointed out that when you did get access, Instagram's recommendations would then kick in to promote more of it back to you. Around this time, another scandal occurred when it was reported that Instagram was recommending terms like "appetite suppressants" and "fasting" to those who had searched for eating disorder content in the past. Then, Instagram released augmented reality filters that could be applied to your selfies, which, in one case, showed you how you would look if you had plastic surgery.

Others applied lip filler or other kinds of beautification effects to users' faces. These caused a lot of backlash, with some pointing out that they promoted unrealistic body expectations and were more likely to make teenagers feel insecure about their appearance. Internally, Instagram began debating whether or not to ban these filters. We now know that this was objected to by Andrew Bosworth, the same guy who wrote the "Ugly Memo" we looked at earlier. Remember that?

The one who argued that breaching users' private data was actually an act of virtue? Well, in this case, he contended that by banning the filters, users would simply move to other apps that are less likely to be as restrained. In other words, he suggested that if we keep the filters, we must grow. Fortunately, his opinion was ignored, and Instagram began banning some of the most controversial filters in 2019. At the heart of all this lies the problem of negative social comparison, something compounded by Instagram's tendency to promote fashion and beauty content, as well as celebrity lifestyles. Instagram's internal research teams were also investigating the causes of negative social comparison at that time.

One document, published in 2020 and later leaked, highlighted several troubling areas. For example, it found that 1 in 10 people felt negative social comparison on Instagram often or always, with teenage women being the most vulnerable group. It also identified issues with the high degree of celebrity content being promoted in people's feeds. However, the impact varied by celebrity; Ariana Grande, Kylie Jenner, Katy Perry, and Jennifer Aniston were among those whose profiles tended to cause the strongest negative comparisons. Additionally, the study found that like counts significantly contributed to negative comparisons, especially regarding individuals within one's social circle.

For example, a celebrity's post needed to achieve 10,000 likes before it could cause a negative social comparison, whereas for a peer, it only took 10 likes. Now, before we move on, I need to mention one crucial thing about a lot of this research. Even though there is an enormous number of damning studies on how social media is negatively affecting mental health—so much so that one might consider this topic an open and shut case—in the last three years, a few new studies have emerged that call much of this previous research into question. One of the most respected reports from the Oxford Internet Institute criticized previous research for drawing conclusions from unreliable methods, such as questionnaires, where respondents can be influenced depending on how the questions are asked. Another problem they pointed out was the lack of longitudinal research that tracks the impact of social media over years or even decades.

One reason this is important is that social media changes so rapidly; research from even five years ago might now be hopelessly outdated. We have no way of knowing. Longitudinal research would help to address this, allowing us to determine whether mental health issues are persisting over a long period of time. As the lead researcher on one paper, Candice Odgers, put it, "there doesn't seem to be an evidence base that would explain the levels of panic and consternation around these issues."

Now, on the other hand, there has been a lot of research over the last three years that still finds significant negative correlations, although all of it agrees that the total body of research is not strong enough. The problem is that we have mountains of correlation, enough to make us very concerned, but not yet proof of causation. As you can imagine, Meta loves this research, especially the one by the Oxford Internet Institute, and they reference it constantly with statements like "there is zero proof our platforms do any harm." However, what they leave out is that this research has not been disproved either. Moreover, the lead researcher of the study they particularly favor, Professor Andrew Pleszliewski, drafted an open letter to Meta signed by a large coalition of other scholars, pleading with them to share their data so this issue could be conclusively resolved.

And if Mehta won't do it, he called on legislators to force their hand. Hardly a ringing endorsement. At the time of writing, Meta has not complied with this request. To hammer this point home, even the U.S. Surgeon General Vivek Murthy got involved this year, recently issuing a public warning for parents that included the sentence, "Social media has not been proved safe." However, if we set aside the issues of mental health, there are also numerous factual reports of issues that simply can't be ignored, such as reports of a massive rise in child grooming abuses on Instagram during COVID, as well as the eating disorders and self-harm problems I have already described.

And now, imagine people's reaction when, in 2021, BuzzFeed revealed leaked plans for something called Instagram Kids, a version of Instagram for children under the age of 13. Media coverage was, of course, blistering. We’ll return to Instagram Kids in a moment because there was a more pressing problem: children were already using Instagram, and serious questions were emerging about the number of underage accounts that actually existed on the platform. One embarrassing revelation occurred live when Adam Mosseri, the head of Instagram, was interviewing JoJo Siwa.

Jojo, are you there? Can you hear me? We're good. We're here. We're live.

Now, the offending part of this interview has almost been completely scrubbed from the internet, but I was able to eventually find a part of it. The moment when JoJo blurted out an admission that she'd been on the platform since she was a child. I know you're not supposed to have Instagram until you're 13. I did. I had an account.

Many, many five worlds. You didn't hear that. Nobody else heard that. And with all this groundwork out of the way, we finally reach the point in this chapter where we can discuss the next world—a worldwide scandal, arguably the most damning scandal of all.

To understand how it all unfolded, we need to momentarily set Instagram aside and return to Facebook. In 2019, the data scientist Frances Haugen joined Facebook's Civic Integrity team, the team whose job it was to make the platform safe. However, it didn't take long for Haugen to become disillusioned due to what she described as hostility from upper management towards recommendations from her team. One example she would later recount was that instead of removing 97% of hate speech as Facebook claimed, their system was instead calibrated to be highly permissive, removing only around 5% of hate speech in actuality. She was also appalled by the dysfunction of something called cross-check, a manual review process for potentially harmful content posted by high-profile figures such as politicians or celebrities.

Haugen would later write that Cross-Check was seriously understaffed, meaning that if a high-profile user posted something appalling, it would take up to three days for it to be reviewed. Since these posts were allowed to remain public until they were reviewed, they could go viral, reaching around 75% of the audience they were ever likely to get. The actions of the Brazilian football star Neymar were a notable illustration of how unfair the system was. After a woman had accused him of rape, he responded by posting her DMs on Facebook, which included naked pictures of her.

Even though this was a clear violation of Facebook's rules, he was protected by Crosscheck, which allowed his posts to be seen by 56 million people, provoking his fans to send a firestorm of abuse toward his accuser. Haugen eventually concluded that for any civic-minded employee, there were three options: you could either ignore the truth about how the platform worked, quit, or do your best to solve the problem, all the while knowing there isn't enough institutional investment or will to actually fix it. Her team created a detailed policy for how Facebook should handle disinformation or hate speech spread by politicians all over the world. This was seen as a particularly important policy, given that the 2020 U.S. election was on the horizon.

However, the policy was rejected by Zuckerberg, who complained that he could write a better proposal over the weekend. That proposal ended up being that Facebook will not touch speech by any politician under any circumstances. This new policy would then be formally announced by Elliot Schrage's replacement, the former Deputy Prime Minister of Great Britain, Nick Clegg. But it is not our role to intervene when politicians speak. And that's why I want to be really clear with you today.

We do not submit speeches by politicians to our independent fact-checkers, and we generally allow them on the platform, even when they would otherwise breach our normal content rules. So despite everything that happened in 2016, Facebook would not fact-check politicians' speeches in the run-up to the 2020 election. As one Politico journalist put it, while major news organizations are strengthening fact-checking and accountability, Facebook is saying, "If you are a politician who wishes to peddle in lies, distortions, and not-so-subtle racial appeals, welcome to our platform. We will not fact-check you."

You are automatically newsworthy. You are automatically exempt from scrutiny. To give one example of this in practice, after violent protests broke out in response to the shooting of George Floyd in 2020, Donald Trump created a post on Twitter and Facebook that ended with the line, "When the looting starts, the shooting starts." Twitter took it down, whereas Zuckerberg personally made the call to leave it up, despite the fact that it broke Facebook's rules about inciting violence. Not long after this, just as the team Haugen was a part of had overseen the exhausting mission of keeping Facebook safe during the 2020 U.S. election, it was announced that violence would be a concern.

Not long after this, just as the team Haugen was a part of had overseen the exhausting mission of keeping Facebook safe during the 2020 U.S. election, it was announced that they were being dissolved, with their members being scattered across the company. This prompted the leader of her team, whom she highly respected, to quit. Right after this, Haugen opened up LinkedIn to start looking around for other work and discovered a message from the journalist Jeff Horowitz, who was wondering if she'd be willing to share information with him. Over time, as they got to know each other, Haugen began contemplating whether she should go public with the information she knew about Facebook. The 2020 election will be rigged if mail-in voting is allowed to take place.

How are you handling the President of the United States to let him know, and how are you informing people who are using your platform that this is misinformation? Is it not? I certainly think that anyone who's saying that the election is going to be fraudulent is problematic, and I believe additional context needs to be added to that. The final straw for Haugen was just after the 2020 election ended when the "Stop the Steal" movement unexpectedly began to explode, aided by a technique called super-inviting. This method allowed a small number of users to send out enormous numbers of invites to bring people to "Stop the Steal" groups.

What's particularly damning about this is that, according to Haugen, Facebook had actually developed really powerful protections that would have stopped this from happening—protections which I mentioned earlier, called "break the glass measures." To illustrate how effective they were, look at this data demonstrating how break the glass measures obliterated QAnon groups in the three months leading up to the election. So, what happened? Why did break the glass measures fail to prevent the "Stop the Steal" movement?

Well, first, as I mentioned, they disbanded the Civic Integrity team at the beginning of December—the team that had been heavily involved in developing these protections in the first place. Second, because the protections caused a drop in revenue, they were simply switched off before Christmas, despite the fact that "Stop the Steal" was growing at that time and despite the large number of Facebook employees who went on vacation over Christmas, which left the company vulnerable. According to Haugen, Facebook didn't switch back on the "break the glass" measures until 24 hours after the January 6th riots occurred. What would have happened had we intervened on December 15th or December 26th or even January 1st? Unquestionably, the adversarial groups would not have grown as quickly as they did.

It was then that Hogan decided to go public with what she knew. But before she left the company, she decided to collect as much evidence as she could by trawling through Facebook's employee forum, searching through thousands of internal documents, and sharing them with Jeff Horowitz. Ultimately, this led to a series of articles published in The Wall Street Journal, all of them under the umbrella title "The Facebook Files." In the past, Facebook had always been able to plausibly claim that the harms they had caused occurred due to technical mistakes that were unintended or social developments they couldn't have foreseen. A leak was something new, leading to a simple overarching allegation.

Facebook knew about the harms they were causing. Facebook willfully chose not to fix the problems that resulted in these harms. Facebook intentionally misled the public about the nature and severity of those harms, delivering reassurances they knew to be untrue. The Facebook Files was not just one scandal; it encompassed around twelve different scandals. However, the most damning of all, the one that sparked the most outrage and demands for justice, was the article titled "Facebook Knows Instagram is Toxic for Teen Girls."

The article references a variety of leaked internal studies that suggested Instagram was aware of the harms it produced, including how it recommended dangerous content leading to depression, eating disorders, self-harm, or even suicide. The kicker is a revelation that one study, called the "Mental Health Deep Dive," was cited in a presentation given to Mark Zuckerberg himself. Yet, according to the article, the company withheld this information from the public and failed to act on its own research. Now, just as all of this was happening, Facebook, the parent company, rebranded to Meta, but the issues remain the same.

It's still about bringing people together. I'm not going to dwell on the rebrand here, except to say that from this point onwards, I'm going to be referring to Meta as the decision-making entity and not Facebook. Anyway, the Facebook files kicked off the most unbelievable storm of activity among journalists, lawmakers, and legal experts around the world, who began looking back over Meta's previous statements about mental health, many of which bore very little comparison to their own leaked studies. New congressional hearings were set up, endless new articles were published, and almost all attorneys general across the U.S. began legal proceedings. Everyone else is way skinnier and way prettier, which also hurt my own self-confidence about my body, leading me to practically starve myself for like a year.

As a consequence, Meta was then compelled to share some of its internal communications. These internal communications reveal quite a bit. From them, we've learned even more about Meta's attitude towards the toxic content problem. In particular, a lawsuit by the Tennessee Attorney General against Meta, which was published in its unredacted form in January 2020, contains a treasure trove of internal emails.

This document is incredible. First, we learned that on March 8th, 2019, the Director of Research sent Sheryl Sandberg a summary of their findings about the damage Instagram was doing, recommending that Meta invest in fixing these problems and also fund more research. On April 8th, this request was reiterated in an email sent to Sandberg, Mosseri, and Zuckerberg. Eventually, a member of Meta's finance team responded, stating that the recommended investments would not be funded. Adam Mosseri, clearly dissatisfied by this, wrote on the same day, "Unfortunately, I don't see us funding this from Instagram anytime soon."

He later wrote, "Well-being is the existential question we face, and we lack a roadmap of work that demonstrates we care about well-being." Nick Clegg chimed in too, saying that Meta needs to do more, and we're being held back by a lack of investment on the product side, which means that we're not able to make changes and innovations at the pace required. According to the document, Mark Zuckerberg did not respond to any of these emails. Then, as the Instagram story exploded, things got crazy—crazy dysfunctional.

Zuckerberg, under pressure to make a public statement about all of this, decided to deflect the problem in the strangest of ways. He was in the middle of a personal rebranding at the time, a rebranding intended to present himself as a fun-loving guy. So instead of meeting the issue head-on, he pushed the head of communications to put out a joke statement that referenced a video clip of him riding a hydrofoil. The idea of the joke was to correct a reference to his hydrofoil, which The New York Times had mistakenly called an electric surfboard. The post read, "Look, it's one thing for journalists to make false claims about my work, but it's crossing a line to say I'm riding an electric surfboard when it's clearly a hydrofoil, and I'm pumping that thing with my legs."

Nick Clegg's exasperation about all of this deserves to be read out in full. Am I missing something here? On the day a Meta representative is pulled apart by U.S. senators regarding whether we care enough about children on our services, Zuckerberg is going to post about surfboards? Maybe I've lost my sense of humor about this whole thing, but I really think this would seem to any casual observer to be pretty tone-deaf. Given the gravity of the accusations we face, if I were him, I wouldn't want to be asked, while your company is being accused of aiding and abetting teenage issues, why your only public pronouncement was a post about surfing?

The Wall Street Journal's reporting has dramatically consolidated a wider narrative that we're bad for kids, which has been brewing for some time. It now makes regulations certain, and in my view, makes launching Instagram Kids nigh impossible. I've told Zuckerberg and Sandberg this already." Defending the hydrofoil post, the head of communications responded, "I'm really eager to just do whatever he wants at this point. My spine has been surgically removed."

The lawsuit also revealed internal frustrations with Zuckerberg, who, due to his majority shareholding, was the only person in the company capable of making big decisions. As Adam Masseri put it, Zuckerberg was the only non-middle manager at Meta. Another executive put it more bluntly, referring to Meta as an absolute dictatorship. It's surreal. If you've read nothing else I've referenced in this video, read this lawsuit.

You'll find everything linked in the description. In October 2021, just over a decade after Mark Zuckerberg was Time's Person of the Year, he appeared on the cover once more. This time, the headline was "Delete Facebook." Quite a reversal. Anyway, just as the Facebook Files was midway through its run, Frances Haugen went public, announcing that she was the internal whistleblower who had shared the documents with The Wall Street Journal.

She then testified before Congress about everything she had seen while working at Facebook. The documents I have provided to Congress prove that Facebook has repeatedly misled the public about what its own research reveals regarding the safety of children and its role in spreading divisive and extreme messages. At the end of 2021, Adam Mosseri was also called to testify. By that time, the questions being posed to Mosseri and other Meta representatives had become broadly better, more specific, and more informed.

In particular, the Chairman, Richard Blumenthal, as well as Senators Marcia Blackburn and Amy Klobuchar, were always sharp and well-prepared. However, these hearings had become a common spectacle at this point, and it was becoming clear that an impasse had been reached. By now, Meta's boilerplate defense was that legislation was needed on an industry level, which is correct on paper. We'll get back to this. The lawmakers' position was that, in the absence of legislation, Meta needed to invest significantly more in making their platforms safer while also being more transparent with their data and research, which is also valid. Due to this impasse, the hearings had generally become a bit of a whipping exercise. The person in the hot seat looks penitent for a while, they get slammed for things they've done and said, they announce a few new initiatives that sound nice, and then everyone goes home.

Inevitably, this led to criticism about the ineffectiveness of lawmakers to legislate and solve these problems. With all these new leaked documents now out in the open, more and more articles kept emerging at a constant rate, alleging all kinds of new damage that Facebook and Instagram were causing. It became a negative PR avalanche. There were advertisements for human trafficking in the Gulf, facilitating sex slavery all over the world, Mexican cartels using the platform to hire hitmen and posting videos of them murdering people, and governments using the platform to crack down on pro-democracy protesters. Images and videos of abuse flooded the site. Meta was aware that most of their automated flagging systems didn't work in Ethiopia and had agreed to suppress anti-government protesters in Vietnam in return for favorable treatment from the Vietnamese government.

And on and on. As for Instagram, the story just kept getting worse, too. In November 2023, more communications were unsealed, indicating that Instagram knew it had millions of underage users but only restricted a fraction of them. Within the company, Meta's actual knowledge that millions of Instagram users are under the age of 13 is an open secret that is routinely documented and zealously protected from disclosure to the public. The outcry following all these revelations finally seemed to prompt Congress into action.

Throughout 2022 and 2023, new legislation with bipartisan support began to appear, including the Kids Online Safety Act. The purpose of this bill is to force platforms to change their design to eliminate harmful content or provide an opt-out for algorithmic recommendations altogether. To ensure this happens, the bill also proposes a duty of care, which makes platforms liable when harm is caused to minors. This led to another significant moment when, in January 2024, Zuckerberg was again called to testify before Congress to answer for Meta's failures on teen mental health.

The existing body of scientific work has not shown a causal link between using social media and young people having worse mental health outcomes. However, this time around, the stakes were higher. Not only was the Kids Online Safety Act now a looming threat, but the senators themselves were armed with much more damning material. In revenue in 2022 from advertising directed at children and teenagers, nearly $2 billion in ad profits were derived from users aged 12 and under. Instagram also displayed the following warning screen to individuals who were searching for child abuse material.

These results may contain images of child abuse. You then provided users with two choices: get resources or see results. Anyway, Mr. Zuckerberg, what were you thinking?

The basic science behind this is that when people are searching for something that is problematic, it's often helpful to direct them towards resources that could aid in getting the help they need. In what sane universe is there a link for “C” results, anyway? Child safety is an area where, quote, "we are investing heavily," end quote. We now know that statement was untrue.

We know it from an internal email. It's an email written by Nick Clegg. You know who he is. He summarized Facebook's problems. He said, quote, "We are not on track to succeed for our core well-being topics."

Nick Clegg was asking you, pleading for resources, and you rejected that request. Here's a quote from your own study: "Issues worse for one in three teen girls." Thirty-seven percent of teenage girls between 13 and 15 were exposed to unwanted nudity in a week on Instagram. You knew about it.

Who did you fire? Senator, this is why we're building all... Who did you fire? Senator, I don't think that that's... Who did you fire? I'm not going to answer that. There are families of victims here today.

Have you apologized to the victims? Would you like to apologize for what you've done to these good people? It's terrible. No one should have to go through the things that your families have suffered. This is why we invest so much and will continue to make industry-leading efforts to ensure that no one has to experience the types of suffering that your families have endured.

A month later, in February 2024, the Kids Online Safety Act gained the backing of over 60 senators, ensuring its passage through the Senate. At the time of writing, it has yet to have a companion bill that will be voted on in the House. My bet is that it ultimately won't see the light of day, but I might be wrong. It's possible that if there's enough public pressure, enough representatives might vote it through. To conclude this section, let's address the question of why legislation is so hard.

Since this is a very large topic, I'm just going to touch on one or two aspects. The trouble with trying to legislate to prevent harm caused by search engines or social media platforms is that it's probably impossible to do so without accepting negative trade-offs. This goes all the way back to the foundations of the internet. Now, I'm sorry for another mini story here, but it's impossible to understand where we are with legislation now if you don't know about this. In the early 90s, while the internet was still in its infancy, there was an online bulletin board called Prodigy, which promoted itself as a family-friendly platform. They used human moderators to filter inappropriate content.

However, a major problem occurred when one of Prodigy's users posted a scathing, defamatory criticism of the brokerage firm Stratton Oakmont, the same company depicted in *The Wolf of Wall Street*. In response, Stratton Oakmont sued Prodigy for hosting the defamatory post and won. The reason the court ruled in favor of Stratton was that they applied the same standards of liability reserved for traditional news publications. In the court's opinion, since Prodigy actively moderated its message boards, it was acting more like a publisher than a simple conduit of information and was therefore liable for the defamatory post. This was rightly seen as a destructive precedent.

How was the Internet going to flourish if companies could be bullied in this way? Precedent. How was the Internet going to flourish if companies could be bullied in this way? And how was the Internet ever going to be safe if Internet services, fearing liability, were too afraid to moderate? The existing legislation was not in line with the technology.

That much was clear to many. To fix this, Representatives Chris Cox and Ron Wyden introduced Section 230 as part of the Communications Decency Act in 1996. Section 230 is the legislation that made the internet as we know it possible. In essence, it protects services so they can actively moderate content posted by their users without being held liable for it. This protection means Meta can't be held liable even when illegal content is posted on their platforms, although they must take it down once they become aware of it.

This is why those who report on Meta and want to see them reined in tend to focus on stories where they can demonstrate intentional inaction. Now, the Kids Online Safety Act, or KOSA for short, doesn't seek to directly amend Section 230, but it certainly butts up against it in a few key ways. While Section 230 is a clear legal shield for platforms, protecting them from being held liable for content posted by their users, KOSA makes platforms liable if harm is caused to children, which could very plausibly result from content. How can this work? Well, the smart aspect of KOSA is that it seeks to avoid unintended censorship by allowing social media platforms to choose the ways they would protect kids.

One proposed approach to comply with COPPA is to alter the platform significantly if the user is underage. However, we don't know how this would play out. An accepted global standard for digital age verification does not exist. Even if it did exist, children can still trick the platform into being recognized as adults anyway. What happens then?

It could be argued that COA might force platforms to overcorrect, censoring a broader range of content for everyone, although I must say I find that a little implausible. Ultimately, it's up to each individual to decide for themselves whether this is the legislation they want to help protect children on social media platforms. My only aim here is to frame it as transparently as I can. I should also mention that this is by no means the only legislation currently being proposed either.

There are too many to try and explain in this video, so I'll just list them for those interested in researching for themselves. With that, we’ve just about covered all the important events that led up to 2024. All that's left is right now. Hello! Oh dear, I look a bit different, don't I?

Well, yeah, I got my hair cut, and you don’t know how many times I’ve recorded this bloody part. The news keeps changing, so I have to keep redoing it. Also, I really wanted this video out before the U.S. election, so hey, we’re breaking continuity. Let’s go. Anyway, congratulations, we’re still here.

You've made it to the end! I'm aware the running time is quite long, but at least you know I wasn't slacking off over the last year. And don't worry, we're not going to limp to the finish line here. The topics I'm going to discuss in this last chapter are truly remarkable. By the way, if you’ve enjoyed what you’ve seen so far and want to support my work, please consider subscribing to the streaming service Nebula, where you'll find videos of mine that aren't available on YouTube. There's a special link in the description that benefits my channel if you use it.

Alternatively, consider supporting me on Patreon instead and gain access to behind-the-scenes content, special videos, and other exclusive items. Support from these platforms is how my channel gets funded. Anyway, with all that said, let's finally bring this thing home. After digesting all this material and this enormous, convoluted history, I can't help but wonder what stage in the evolution of social media we're currently living in. Is it still early days?

Are we still essentially guinea pigs, the generations whose aggregate misfortune will help to temper a better and safer social media in the future? Or are we experiencing the early warning signs of a far worse future existence—one that needs to be averted? To help us answer this, let's look at Meta right now. At the time of writing, Meta as a business is in better shape, thanks in no small part to the rise of AI.

Because Meta had a surplus of GPUs, as well as long-standing experience in the field of machine learning, they entered the AI race very quickly. Since then, they have released multiple versions of their large language model, LLaMA, which is open source, in contrast to the closed models of OpenAI and Google. Think of LLaMA as Meta's answer to ChatGPT. However, the current rosy outlook is only a recent development. Just back in 2022, with the global economy souring, Meta's stock dropped significantly, largely due to skepticism over Zuckerberg's metaverse strategy.

To tackle this, Zuckerberg engaged in what he called the "Year of Efficiency," letting around 22% of Meta's workforce go over a very short timeframe—something most companies of that size are unable to achieve. This drastic cost-cutting, combined with their quick entrance into the AI race and the incorporation of AI features into Facebook before Google or Apple could do the same, caused Meta's stock to explode. At the time of writing, Meta is, once again, the darling of the stock market, resulting in a significant improvement in how business publications talk about them. It’s as if he reorganized an entire company in three months. It’s unbelievable what Zuckerberg did.

Regarding the metaverse strategy itself, I'm not going to dwell on this too much because the technology requires a few more breakthroughs. What I will say is that I always find it a little bit irritating when publications refer to Mark's vision of a metaverse. Because it's not his vision. Setting aside the decades of sci-fi novels about this idea, from a tech point of view, it's much more Microsoft's vision. In 2015, when I worked at Microsoft, all of these ideas—conferences and virtual spaces, arranging apps around your room, a large virtual cityscape where people can interact, etc.—were being discussed ad nauseam.

For a while, I worked on a team that built applications for the HoloLens, so I was right in the thick of all this stuff. Loads of people I knew ended up at Meta, so when I saw these ideas recycled by Zuckerberg in 2021, it felt a bit like déjà vu. It happened again when Apple recycled it in 2023, bringing absolutely nothing original to the table themselves. With that in mind, I did actually enjoy Zuckerberg's takedown of the Vision Pro. It was cathartic and also convincing. I know that some fanboys get upset whenever anyone dares to question if Apple is going to be the leader in a new category.

And that takes me to a different point: Zuckerberg's personal rebrand, which seems to have kicked into a new gear. For as long as he's been in the public eye, he has consistently been the butt of the joke. A lot of this began with the movie "The Social Network," which portrayed him as a cold-hearted dork. While I loved that movie, there are aspects of the portrayal with which I have real problems.

Some aspects are just untrue. In particular, the idea that he had no success with girlfriends and that his obsession with social media was a coping mechanism to get over it. In fact, he met his now-wife, Priscilla Chan, while he was studying at Harvard. Did you ever end up seeing the film, The Social Network? Yeah, I did.

Cool. Oh, thanks. And what did you think? It was interesting. His personal gaffes up until this point are pretty much never-ending.

Being sweaty during high-profile interviews, having a booster seat during congressional hearings, looking awkward on a hydrofoil, exhibiting a generally robotic delivery, and creating that incredibly creepy-looking metaverse video— I think it's time for my workout. Anyway, as part of his personal rebrand, Mark has become an increasingly visible figure in the world of MMA. He has done away with the tightly cropped hairstyle, allowing his hair to grow out more naturally. He has also changed the types of clothes he wears to avoid looking so stiff.

And crucially, he is generally far more natural in interviews these days. He conducts a lot of interviews, especially on YouTube. His choice of subject matter and interviewers has decidedly changed as well. In the past, up until around 2019 and 2020, it seems that the strategy for communication was to be seen engaging in significant questions about technology and its effects on society. This is clearly evident in his conversation with Yuval Harari, who really challenges him.

I know that Facebook faces a lot of criticism for encouraging people to join extremist groups. I believe it's something that can be addressed if enough energy is invested in the solution. What I worry about more is increasing inequality. Additionally, there's the concern of undermining human agency and the basic philosophical ideals of democracy, the free market, and individualism. Mark gives a decent account of himself here, but ultimately he's always on the defensive and often deflects criticisms rather than directly engaging with them. As a result, the comments under the video are almost entirely hostile towards him.

But in the following year, after the PR disaster of the Facebook Files, which, for one thing, forced them to drop Instagram Kids, there has been a very clear U-turn in their communication strategy. Now the focus is almost exclusively on the topic where he is strong – cool tech. The choice of interviewers is pretty clever too. They are all, to a person, toothless: tech podcasters, intellectual lightweights, or sympathetic bros, many of whom see the opportunity to interview Zuckerberg as a giant boost to their credibility.

Like you and I share, we're tech optimists, and I think almost all of this tech, if not all of it, ends up being a net benefit for humanity in the long term. The one thing that unifies all of these more recent interviews is that Zuckerberg is rarely, if ever, meaningfully challenged. It's also useful for Meta that the world has largely identified a new villain in Elon Musk due to his consistently ridiculous decisions with respect to Twitter, as well as a variety of other things. But I’d be remiss if I didn't mention that Mark and Priscilla pledged to give away or sell 99% of their Facebook shares within their lifetime, with the funds going to a new LLC they set up that would allocate money to various charitable efforts. This is by no means the only charitable work they've done, but it's the most eye-catching.

It's almost like Mark's intention is to continue doing what he's doing for many years to come, and when he's finally gone, he'll donate a ton of money as a way of apology for all the damage he's done and hopefully secure a positive legacy. Combined with all the good business press, the question is: will this work? Will the general public's perception of Meta and Zuckerberg improve over the coming years as new generations transition into adulthood? Will they be able to leave behind the incredible legacy of mistakes and damage? More to the point, will Meta eventually become a company that people broadly agree is a force for good?

Well, to answer the first point about public perception, never underestimate Zuckerberg's problem-solving skills. Long gone are the days when Sheryl Sandberg was wheeled out as the public spokesperson whenever the company faced trouble. Sandberg's star fell during the scandals I've already covered. She stepped down as COO in 2022 and left the board in January 2024. What's left is the new communications strategy I've just outlined.

Focus on the positive technology advancements and deflect everything else. I believe, at the time of writing, that it's working. Have you heard the term "Zuckassance"? Zuckassance! You haven't heard of that?

It's Zuck's Renaissance. As for the question of whether Meta is beginning to turn it around, it's going to be very interesting to look back on this video in five years to see how things actually play out. However, to my mind, the answer is that I'm very pessimistic. Even though there hasn't been a mega scandal in the last two years, there have been many negative developments—developments that demonstrate the same brutal drive for growth and control, as well as new issues that could connect in dangerous new ways. Starting with the smaller stuff, let's talk about Instagore, Instagram's current incarnation, which, at the time of writing, is recommending brutal, shocking content, including videos of people getting murdered or mutilated, animals and children being severely injured, fatal car crashes, electrocutions, and people falling off buildings. The list goes on.

I'd heard people talking about this new brutal trend on Instagram, but not being a regular user myself, I hadn't seen it. So, I turned to my Discord group and asked whether any of them had, and I really was not prepared for what came back. Twenty minutes later, I had to shut off my computer and try to go to sleep with horrible images banging around inside my skull. In particular, the things I saw happening to small children made me imagine my own small children in the same situation. So, why are people uploading this kind of stuff?

Well, some things don't change. For numerous pages that I've seen, there's an accompanying lengthy text promoting cars or other products, with many different accounts of identical promotions indicating they’re owned by the same people. I probably wouldn't have mentioned this if I hadn't experienced it. The problem here isn't just that the content is horrible; it’s the relentlessness of it that makes it so much worse. It creates a compounded effect where every 15 seconds you’re watching another brutal injury, another death, another animal being mistreated, another child being harmed—another death, another injury.

And then Instagram interprets your stunned, caught-in-the-headlights lack of action as engagement and just pumps out more of it. After only two or three minutes of this, you enter into a kind of elevated sense of borderline panic. It feels like your body has kicked up your adrenaline levels due to a sense of imminent danger to your person. It's horrible. It affects you.

It's been widely reported that moderators for Meta are at serious risk of developing PTSD, and I can see why. So what about data and privacy? Well, Meta is doing terribly here too. Let's take Europe. In July 2023, the EU Court of Justice ruled that Meta was in violation of our data protection law, GDPR.

By forcing users to accept targeted advertising as a condition of using Facebook or Instagram, GDPR states very clearly that users have a fundamental right to opt out of having their data shared for the purposes of advertising. Meta's actions really give me the sense that they have nothing but contempt for this regulation. So, what did they do when that ruling was handed down in July? Well, they finally created the option to allow people to opt out for a subscription fee of €251 a year. This tactic is now referred to as "pay or okay."

In April of this year, the European Data Protection Board ruled that "pay or okay" was not acceptable. It now seems Meta has run out of options and will be forced to provide a proper, free opt-out mechanism soon, although God knows what new tricks they'll come up with next that EU legislators won't be expecting. Meta's relationship with the European Union is going to continue to be very interesting, especially now that the Digital Services Act has come into full force in 2024. This legislation requires online services to swiftly remove illegal content and other harmful online activities. It also mandates much more transparency about exactly how online services moderate content.

If successfully enforced, this would tighten the screws in such a way that a lack of compliance would be much easier to prove. And if it is proved, the legislation now gives the courts the power to impose fines of up to 6% of Meta's annual global turnover. That's not a fine that could easily be brushed off. Now, in my opinion, the privacy-related story to keep an eye on, a scandal that's emerging right now, is how Meta is using the content we post on their platforms to train their AI models. In the U.S., where protections aren't yet in place, if you'd like to opt out of this, you are, frankly, out of luck.

They don't need to inform you that your data is being used to train their AI, and if you submit an objection, they're not obliged to act upon it. In the EU, where legislation prevents them from doing this, they pushed out a highly convoluted notification to its users, which I think takes the award for the worst collection of dark patterns in Meta's history. And that's a fiercely competitive pool. I tweeted about this in May, a few days after they rolled it out, and that tweet certainly did the rounds. Let me summarize.

Here's the text reformatted with correct capitalization and spelling: "See, I've worked in growth teams for years, and I know how to design and streamline an experience to maximize traffic. The rule is, the simpler the experience, the more people will make it through to the end. With every additional step you add, large percentages of people will drop out. Now, Meta did the opposite of this. It's hilariously awful."

Okay, so step one. You receive a notification with a misleading non sequitur: "We're planning new AI features for you. Learn how we use your information." Notice the absence of the terms "private," "data," or "opt-out."

And once you click on that, step two, you are shown this notice, where they explain the legal basis they are using for collecting your information: legitimate interest. Let me quote the opinion of the European Centre for Digital Rights on this. Instead of asking users for consent or opting in, Meta argues that it has a legitimate interest that overrides the fundamental right to data protection and privacy of European users. Once their data is in the system, users seem to have no option of ever having it removed.

Right to Be Forgotten. The shameful aspect here is how hidden the right to object link is, placed toward the end of the second paragraph. It seems to me that there is a rather large and more intuitive space below that could have been used instead. It's unpleasant. Also, notice the line that states if your objection is honored, it will be applied going forward.

I don't see that too often, so let's object. This leads us to step three: this big form you need to read. It's only at this stage, the stage where you are objecting, that they inform you about which of your content they plan to use to train their models. They also state, surprisingly, that they may choose to ignore your objection. So, step four: you submit your objection.

Bear in mind that needing to fill in a form is not a necessary friction. They don't need it, and I highly doubt any humans are going to read it. The one thing I am sure about is that it will turn more people away because they couldn't be bothered filling it in. Step 5: Once you've confirmed your objection, you're then told you need to check your email and grab a code they sent you.

I'd love to hear their justification for this. Step 6: You open the email they sent, which contains a code that's only valid for one hour. If you miss it, you'll need to start again. Anyway, now you copy the code.

Step seven: Enter the code and get a confirmation message. Then, comically, in the space of around eight seconds, you receive an email letting you know whether or not your objection is going to be honored. Wow, those Meta folks sure know how to work efficiently. I should mention that when I tweeted this, some of the people who responded to my tweet told me that they got an error when they objected, which is just classic Facebook. So, let's get into AI.

Now, when it comes to this subject, I’m not a deep pessimist. Yet. I'm excited about the potential of the technology, even though, like most others, there are aspects I find pretty horrible. Like, sorry to beat a dead horse here, but as someone who spent enormous portions of my youth learning to draw and exploring different kinds of styles and techniques, I feel really deeply for the visual artists of today who are particularly vulnerable to the rise of generative AI in a way I don't think is true for musicians.

It sucks if you've been using Instagram to promote your work for years and now find yourself with no recourse as that same work is used to train programs that are literally in the process of replacing you. However, when discussing Meta's AI strategy, I want to avoid leaning into hyperbolic pessimism like your garden-variety video essayist. The only observation I can make with some confidence is that, as is always the case with Meta, they provide tools for free which could potentially be very beneficial to the outside world while also locking things down in such a way that they guarantee their own long-term dominance. So that's how I'm going to frame this. On the beneficial side is Meta's choice to open-source a significant number of trained models, making them available to the public.

Apart from their flagship large language model, LLaMA, you also have Detectron, a computer vision system that can detect objects and segment them. With enough reliability, imagine how this could be utilized in, for example, healthcare. Tumor detection, glaucoma detection, assisted surgery, or any one of a thousand other possibilities. And those are just two; the list goes on.

There are many more useful models like this that could serve as incredibly powerful tools for smaller businesses and startups. It must also be noted that Meta has a history of releasing beneficial technologies to the public, such as React, React Native, and PyTorch. Where things start to get complicated is when you consider the wider strategy being employed. First, let's investigate how they're open-sourcing their AI models. Taking LLaMA as an example, if you look at the terms of the license, it's clear that this model is, in fact, only partially open source, to put it charitably.

First, the data used to train the large language model is proprietary, as is the code used for the training. This prevents others from forking the project to create their own version, one that could, for example, be much more open, allowing users to interrogate it to find out what sources it used when answering questions. That is not an open-source approach, not even close. Additionally, this raises significant suspicions about the origins of the training data. If it has been trained on our conversations on Facebook and Instagram, then the data could never be released because it would represent the biggest privacy breach in history.

And just to be clear, I'm not saying they should definitely release everything open source, but they just can't call it open source unless they do that. However, the most eye-opening part of this license is definitely the clause that if you're a company with over 700 million monthly active users, you can't use Lamr for free and must obtain a commercial license instead. Hmm, 700 million—what's that about? Well, Snapchat hit 750 million in 2023, and TikTok and WeChat passed 700 million a few years before.

And this helps us to understand the dominance aspect. Something free to the end user, labeled as open-source, is great for public relations, but it can also be a sharp-edged business strategy because it eliminates competition. Providing their AI services for free suits Facebook's advertising model while simultaneously undermining competitors in their weight class, like Google or OpenAI, who charge for access to their AI services. Making it free also reduces the risk of new tech giants emerging to compete in the future, but it can strangle small businesses too, preventing them from ever getting off the ground. Over the last year, I've spoken to a few well-known AI startups who have expressed grave concern that their entire business model could be completely snuffed out if Facebook decides to roll out a similar feature for free.

So, in summary, open-sourcing their AI models is a clear power play to broaden their domination as a tech superpower by disadvantaging other tech giants while also making it unlikely that new businesses can grow big enough to become properly competitive. And it's our data, often given without our knowledge or consent, that's empowering them to do this. It is, as I hope I've demonstrated clearly throughout this video, classic Mark Zuckerberg. And then there's the other side of the AI coin. Prepare yourselves for another variation on a classic theme here: AI content is completely out of control on Meta's platforms.

In 2024, an alarmingly creepy new trend emerged, with vast amounts of bizarre AI-generated images flooding Facebook. One example is a deeply unsettling series of images featuring young children from historically disadvantaged countries surrounded by elaborate or impossible structures made of food. In particular, the theme of overabundance of food feels really ugly and mean-spirited to me. There are many others that are overtly racist in theme, which I'm obviously not going to show. Then there's the endless series of pictures of Jesus.

Shrimp Jesus. Crab Jesus. Jesus performing feats of heroism or revealing Himself in unexpected places. There are all kinds of hyper-real, cutesy images of babies—babies with kittens.

Wide-eyed babies with miniature kittens. Babies with Jesus. Okay, okay, this one's actually kind of funny. And there are all kinds of other things. There are images of amazing stuff.

People celebrating their 120th birthday. Amazing scenes of various kinds. Wow! This woman's garter stitch technique is off the charts! We are quadruplets.

Today we turn 90. We are waiting for congratulations from you. Uh, how is her arm doing that? What's going on over here? This one is an amazing category.

Families sitting around a table that contains the preserved corpse of someone's mother. Okay, so I've kept most of these examples light, and you might be thinking, "This all seems harmless enough, Tantacruel." Ah, Tantacruel, get a grip. Well, you see, with this stuff flooding Facebook, an enormous number of these images are going viral. In the case of some of the Jesus imagery, this is due to devout Christians around the world feeling obliged to comment on them and reshare them further.

The problem is that, according to a Stanford Internet Observatory report from March 2024, a significant number of the pages that disseminate this content are either spammers attempting to direct people off-platform to content farms or are outright scamming vulnerable individuals by using the comments section to extract information from them. There is also evidence of multiple pages owned by the same individuals working together to boost virality. Furthermore, because anyone can generate and share large quantities of AI-generated images daily, the cumulative effect is that it seems the platform is being overtaken by it. Once you've seen a few, you get sucked into the AI vortex. This aspect has already been reported by several YouTubers, many of whom focus their attention on what they perceive as the real story here.

How stupid people are to believe that this stuff is real. The thing is, when it comes to scamming, I don't think it's a stretch to suggest that one of the advantages of using AI in this way is that it acts as a filter, much like the famous Nigerian Prince scam, which helps to identify the older, the more gullible, or the less tech-savvy individuals—the easy marks. But there's another worrying reason why these AI pages exist: to grow them to be as large as possible in order to sell them, almost always to unknown third parties who want those pages for unknown purposes. Imagine there was an election in a majority Christian country with a population that has historically been vulnerable to disinformation, like Ethiopia.

Well, if you've managed to create a significant following among Ethiopian Christians on Facebook through the use of seemingly harmless AI-generated Jesus imagery, there might be some folks interested in the outcome of that election who would be willing to take that page off your hands. The marketplace for buying and selling accounts is not hard to find; I mean, it's all over Reddit. If you have a Facebook page with 50k followers to sell, please contact me—serious buyers only.

Hi, is the majority of the followers in the USA? Please PM me the details. On Swapped.co, at the time of writing, you can buy an Instagram account with 1.4 million followers, 90% of whom are women, 25% of whom are in the U.S., for $5,000. Many would feel safe signing up to "Unicorns Are Cute," until... Congrats! You're now a card-carrying member of Alpha Battle Brotherhood, Iron Fist Nation!

And this leads us nicely to the thing we've all been waiting for: the beginning of AI being deployed to sway political discourse. In May 2024, Meta dismantled a network of over 500 fake Facebook accounts and 32 Instagram accounts linked to the Israeli firm Stoic. According to Meta, the accounts pretended to be owned by U.S. citizens but were all posting AI-generated comments in various discussion threads, praising Israel and criticizing U.S. campus protests over the war in Gaza. In this case, Meta, to their credit, seems to have been ready for this kind of inauthentic behavior. Now, what we've all been expecting to see is, of course, the use of AI-generated disinformation to influence the U.S. elections this year.

But because this approach is so obvious, Meta must have been preparing for it for ages. Sure enough, in August 2024, they announced the removal of tons of AI disinformation content, with the majority of abuses coming from—oh, I wonder which country it might be. So, if we're destined to see yet another major disaster in this upcoming election, I'd be surprised if AI plays a big part in it. In fact, I'm now going to boldly suggest that because Facebook will be so focused on not getting yet another black eye this time around, the social media story, if there is one, will happen over on Twitter. But that doesn't mean Facebook's doing great.

Apart from AI, we can also rely on garden-variety disinformation, the types spread by humans. Take, for example, a story that originated in Russia about Olena Zelenska, the wife of Ukrainian President Vladimir Zelensky, which falsely reported her lavish spending during a shopping spree. A lifestyle, they pointed out, made possible due to U.S. war aid. Another video circulated that claimed, falsely, to show Hunter Biden assaulting a woman in a hotel. In August 2024, two British men were jailed for spreading false information about immigrants and asylum seekers, which contributed to violent anti-immigration riots that lasted a week in the UK.

The riots themselves were set off by the lie spread around Facebook and Twitter that a man who had assaulted three children in a violent spree was himself an asylum seeker. In a move that may elevate risk a bit further, Meta has just announced their intention to shut down support for a service called CrowdTangle, a tool that enables outside observers to track the spread of disinformation on Meta's platforms. CrowdTangle has been used to prove embarrassing stories about Meta on numerous occasions. The choice to shut this service down in the run-up to the U.S. election and replace it with a new tool many feel is much more restrictive is—what can I say? The decision has been met with damning and widespread criticism.

Beyond that, the outcome of the U.S. election might be really consequential for Facebook. The initial choice between Joe Biden and Donald Trump was a pretty uncomfortable one for Zuckerberg. Biden strongly supported legislation to bring Meta to heel, whereas Trump has a massive axe to grind. He hates that Facebook banned his account and cracked down on disinformation after the January 6th riots, and he even threatened on Truth Social and in his new book that he’d imprison Zuckerberg for life if he interferes with the upcoming election. Trump has also expressed strong support for repealing Section 230, which would be a disaster for Facebook.

Although it's highly unlikely Trump would actually see that through, the addition of J.D. Vance to the ticket is a curveball because he's the protégé of Peter Thiel. Thiel himself is a major donor for Trump, which I would imagine bought him a degree of influence. To my mind, that could only really work in Zuckerberg's favor, but to what degree is anyone's guess?

Probably not that much. Kamala Harris has similar sentiments to Joe Biden, although she's somewhat less fervent about them. Even so, she's expressed support for Section 230 reform, as well as antitrust legislation and social media legislation. So, one way or the other, it's clear Zuckerberg won't have too many friends in Washington for the next four years. Additionally, there are even more reasons to be concerned about children.

In 2023, there was another massive finding by Jeff Horowitz, and I'm going to have to be super careful about how I phrase this here: the revelation that there's a collection of grown men who are now using Instagram to satisfy their interest in extremely young users. Not only that, but if you were inclined to follow these men, say it with me, Instagram's recommendations would kick into gear, recommending you more of the content they were looking at. A New York Times article from February 2024 found even more evidence of this, running a story about the experiences of some parents who reported receiving vile comments from grown men on pictures they posted that included their children, with some men making offers to pay to see more. Now with that, we've tied up just about everything, and I had been planning on ending the video here, but I'm actually not quite done. Because there's one last really strange aspect to this story I think you need to be aware of, as it's happening right now and is often directly affecting us in ways we don't see.

This was a rabbit hole I inadvertently stumbled into almost two years ago. It started in early 2023 when I published a video about dangerous online forums. In that video, I also briefly touched on Section 230, the legislation we looked at earlier. After I released that video, in order to stay informed about Section 230, I made a few contacts in the U.S. and also followed a number of different legal experts who specialize in matters related to privacy, free speech, and tech regulation. This allowed me to keep track of the latest developments by reading relevant articles, summaries, and think pieces.

In particular, I valued articles that offered deep legal analysis, and when searching for that information, I eventually stumbled across a group called the Chamber of Progress. Now, the Chamber of Progress members are staunch advocates of Section 230, and they believe it should remain unchanged, a perspective I thought it important to understand because I didn't want to be naive. As I mentioned, amending Section 230 would involve a very complex balancing of harms, and I wanted to understand those trade-offs in detail. So, with this in mind, I would occasionally read posts by members of the Chamber of Progress, especially when they were discussing ongoing lawsuits or legislation aimed at big tech. Sometimes I found their explanations very enlightening.

However, I also noticed something strange about them. For example, one of their members tended to get into vicious arguments on Twitter all the time. Additionally, some of their takes were just baffling to me. Here's an example tweet I found particularly odd: "People."

AI companies are scraping my posts to train their models, as are individuals. Here's a public photo dump of me from when I was 19, due to Twitter trends. This struck me as really childish and mean-spirited, suggesting that if you happen to be naive about privacy, you don't deserve privacy. There was another instance when the Chamber of Progress began posting criticism of the EU AI Act, the first sweeping standard for AI regulation in the world, arguing that it dealt a death blow to innovation.

The thing is, like any legislation, there's a balance to be struck between protecting people, enforcing transparency, preventing potential harms, and yes, not overly hampering innovation. In general, I expect civic-minded legal experts providing commentary on these matters to give a holistic overview to help others understand whether those balances are acceptable or not. That said, their articles did give me pause from time to time. They made me concerned that perhaps the EU had overstepped. Nevertheless, over time, I began to see them as a bit of a broken record—ideological zealots who you could always rely on to criticize any legislation related to big tech with unwavering certainty.

I also got sick of their constant bickering on social media, so I quietly unfollowed their members and moved on. Then, about six months later, while making this video, I put out a call on Twitter to interview Section 230 experts on the topic of the Kids Online Safety Act. Lo and behold, one of its members got in contact, offering to speak to me and wanting to know more about what I was up to. So, I outlined the basic purpose of my video and received no response. Not even a, "Oh, I'm a bit busy," just nothing, which I found a little bit weird. They seemed really interested at first.

Well, that got my spider sense tingling. Until that point, even though I'd followed some of its members, I was only aware of the organization itself in the most hazy of ways. So, I quickly logged onto the Chamber of Progress website to finally look at what they actually do. The Chamber of Progress is a new tech industry coalition devoted to a progressive society, economy, workforce, and consumer climate. We back public policies that will build a fairer, more inclusive country in which all people benefit from technological leaps.

Oh, that's reassuring. Then I looked at their partner section up here and, ah, they’re funded by Meta, Google, Amazon, Apple, and Midjourney. Ah. Then to verify, I checked Meta's page on their political engagement, the candidates they fund, how much they spend on lobbying, etc. And sure enough, the Chamber of Progress was listed there too.

And that's when I realized that this entire time I had been a total sucker. I had read so many articles on big tech, often with so many different threads to follow up on, that I hadn't scrutinized material from this group carefully enough. I had failed to ask the most basic of questions that you always need to ask when encountering opinions that don't quite add up: How do these people get paid? I thought I had been reading the opinions of completely independent legal scholars.

What really bothers me about this is that there's a broad spectrum of opinion out there, and it's already not a level playing field. But for Meta to pump millions into one particular opinion—the opinion that happens to serve their interests—so that it's dramatically amplified strikes me as highly unprincipled. I couldn't believe they'd managed to manipulate me from beyond their platforms. To the outside observer, the materials produced by these organizations can appear as grassroots opinions from civic-minded citizens. If the articles I'd read had been accompanied by the disclaimer "Chamber of Progress is a lobbying group that receives funding from Meta," it would have completely changed how I regarded their work.

And this is the final part of the new communications strategy I mentioned earlier. There is enormous spending on lobbyists and influence groups to try to shoot down legislation or drive popular anger about tech regulation, often by creating PR campaigns and even documentaries that attempt to undermine legislators in the eyes of the public. In 2023, Meta spent $19 million on lobbying, but 2024 looks set to dwarf that. Spending in the first half of 2024 is already significantly higher than in 2023 and also far surpasses that of any other tech giant. The Chamber of Progress is run by Adam Kovacevic.

As one tech journalist puts it, Adam Kovacevic has become a dialect quote for reporters, always happy to make big tech's case. The companies stay above the fray. For their money, the big tech companies get an organization willing to push their viewpoints in the press, sometimes with spotty disclosure. What does he mean by spotty disclosure? Well, listen to how Kovacevic is introduced in this news report.

He's the founder and CEO of Chamber of Progress, a center-left tech industry policy coalition. It aims to ensure all Americans benefit from technological advances and works to make sure the tech industry operates responsibly. Hearing that, I'd think Kovacevic is holding big tech to account. For another example, in an article he published on a site called CEPA, the Chamber of Progress is described as a center-left tech industry policy coalition promoting technology's progressive future. No mention of funding again. But if you go to the "About" section on the CEPA site and select "Supporters," boom!

Hello, Meta! Kovacevic pops up regularly when you look up legislation that seeks to rein in big tech companies, usually in the form of an article on Medium that criticizes that legislation. Sometimes this criticism is justified; sometimes it's really forced, clutching at straws. For example, take the proposed American Innovation and Choice Online Act, championed by Senator Amy Klobuchar, which seeks to limit tech giants' ability to unfairly promote their own services over those of smaller competitors. Specifically, it aims to prevent the intentional disadvantaging of smaller businesses by doing things like forcing them to pay in order to rank highly in search results. The Chamber of Progress's response to this legislation is that it would break Amazon Prime.

That's the doomsday scenario, forcing Amazon to rethink how Prime works. On all of this, I'm joined by Adam Kovacovic, founder and CEO of the Chamber of Progress. Adam, they're funded by Meta, Amazon, and a bunch of others. You forgot to mention that.

Talk to us about how significant this bill, if passed, could be. That would mean that Amazon couldn't feature Amazon Basics brand products. It probably couldn't highlight Amazon Prime two-day shipping products. Even if true, would that not be a price worth paying for a properly competitive playing field where big tech can no longer monopolize absolutely everything? Amy Klobuchar outlined the degree to which these groups are interfering with the legislative process.

Let's talk about what we're up against. I'm talking about some of the most powerful companies in the world, with armies of lobbyists and lawyers. They are everywhere—in every corner of this town, at every cocktail party, and all over this building. I tell my colleagues that they don't even realize sometimes when someone is trying to influence them. I advise them to ask the person if they are being paid by a tech company, if they sit on the board of a tech company, or if they have some affiliation with one of the big tech companies.

They are also lobbying the American people with astroturf campaigning and other dishonest PR tactics. These companies have been telling anyone who will listen that acting to protect competition in our digital markets will somehow threaten our national security or outlaw Amazon Prime—claims that have been disputed by the Department of Justice and Amazon's own law firm. I'll be honest. But they don't stop there.

Because providing what appears to me as incredibly one-sided legal analysis isn't their only approach. They also run fear-mongering ads on Facebook. Your Amazon Prime subscription is in danger. I suppose they hope to motivate their target demographic to complain to their representatives to put them under pressure. And then, of course, there's the money.

I think this is actually the best evidence of just how big, dominant, and bullying these companies are. Running ads in states where people are in tough races—how obvious can it be? They also run other kinds of cringe PR campaigns. My favorite is their 2024 gem, "Generate and Create," a campaign that woefully celebrates how generative AI is a force for creative liberation. Actually, opponents of fair use are working to shut down AI-created art through legal and legislative action.

We're fighting back, and we're being paid by Metagoogle and Midjourney to do so. How are they fighting back? Well, by creating a laughably produced YouTube video in which they drag three artists onto the stage to argue their case for them. Them.

These sad sacks then proceed to mumble unconvincingly about how generative AI is useful for their workflow, kind of. My favorite line by Kovacevic is this: "Generative AI is a net plus for creativity overall. It's expanding access to creative tools for more and more people and bypassing a lot of the traditional gatekeepers." What gatekeeping is he talking about?

The Salon de Paris? Is he talking about the admissions process for art colleges or animation studios? To whom it may concern, it has been a dream of mine to study at your institution. I am passionate, creative, and hard-working. Enclosed is my AI-generated portfolio.

Looking forward to hearing back, Timmy! They even created a rubbish website to kick off their Generate and Create campaign, and it's filled with hilarious cringe. I really do love it when legal professionals weigh in on matters related to creativity. It's so cute! We didn't hire a professional designer or need to learn Adobe Illustrator to create our campaign mascot, Artifish.

We created Artie to Midjourney, an AI tool to expand creativity, and Adobe Illustrator to create our campaign mascot, Artie Fish. We developed Artie to Midjourney as an AI tool to enhance creativity. With AI, anyone can instantly create custom designs. Well done, everyone! That is how image generation works. Get a grip!

I don't even understand why this is here. Is this meant to be inspiring? And also, "arty fish"? Please tell me the concept was AI-generated too. But the icing on the cake was a tweet from one of their more overzealous members, clearly getting a bit carried away after one too many doctrinaire Zoom calls. It's a masterpiece of nonsense, one of the most condescending, backwards pieces of mumbo-jumbo you'll ever read.

For context, it's aimed at the artist community on Twitter, and reads: "Have demonstrated their dedication to keeping marginalized folks, including those with disabilities, out of their community and marketplace. Gen.ai says, 'F*** that!' Those barriers come down now, and I'm here for it! As you can imagine, this caused absolute uproar, with an enormous number of critical responses coming from disabled artists who pointed out that, far from being marginalized, their ability to create art provided a vital way for them to earn a living and interact with others online. One of my favorite responses was this."

Double combo! The diversity and gatekeeper card played at the same time. This is why big tech pays so much for this lobbying. But that's not all. Another purpose of this ridiculous campaign is to try to sell the idea that artwork generated by AI should be protected under fair use.

Yeah, I know. So, in order to do that, they have to try and mesmerize the general public into thinking that if you tell a computer to make you art, you're an artist. And even if you sympathize with that idea, just remember that the reason they're saying it is not because they actually care about creativity or have the faintest clue what they're even talking about when it comes to creativity. They are trying to manufacture a broad consensus as a step toward protecting generative AI from claims of copyright infringement. Now, I’m not a big lover of the copyright holders of this world, but I can tell you for sure I'd rather a world where those guys are still holding the aces, rather than the entirety of Earth’s artwork being subsumed by the likes of Meta.

Another of their big pushes is to advocate that Section 230 be extended to also cover generative AI, meaning that tech firms would be completely shielded from anything the AI outputs. By the way, the original authors of Section 230, Ron Wyden and Chris Cox, do not share this opinion. According to Cox, Section 230 is about protecting users and sites for hosting and organizing users' speech, and it has nothing to do with protecting companies from the consequences of their own actions and products. According to Wyden, it isn't a particularly close call. Now, you might agree with some of the Chamber of Progress's positions— I agree with some of their positions— but I think it's completely disgraceful that Meta is willing to actively fund them with money, enabling them to hire more staff, take out ads, and waste legislators' time and our time. Big Tech should be able to hire third parties for research, advertising, or legal representation, but these relationships have to be completely transparent.

And in my opinion, given the examples I've just shown, as well as countless others I haven't, the transparency test has not been passed. Not even close. If you take all this together, that’s why I can finally conclude that I have incredibly low faith that Meta is going to turn things around. There seems to be no end to their callousness and their incredible lack of respect for their users. Every time there's an opportunity to improve, they choose the most self-serving and disappointing option, regularly making the world around them demonstrably worse.

Bear in mind that in January 2024, Mark stated, boldly, that he welcomed regulation and wanted it applied at an industry level. When I first heard him say this, I felt it was a good sign. I believed he had accepted that change was needed. But for Meta to then go and fight regulation at every turn, that's about as low as it gets. They've now gone completely through the looking glass.

They work for them, not for us, and their behavior is as objectionable as it has ever been. What else needs to be said? Meta regularly ranks in the top ten lists of the most hated companies in the world. This is really troubling because Zuckerberg answers to no one. Yes, he looks bad if shares plummet, but he can't be ousted for poor performance.

He can't be ousted by his own company on any grounds, which makes him, to put it bluntly, possibly the most influential person on the planet given his control over how we communicate. Yes, he looks bad if shares plummet, but he can't be ousted for bad performance. However, he has consistently demonstrated a pattern of terrible judgment when it comes to the negative effects on society.

Why? Who knows? It's hard to know how much he's been influenced by the techno-libertarian ideas of those like Peter Thiel, but whatever the reason, he seems to be both ideologically and intellectually incapable of acknowledging his responsibility for the damage his platforms cause. He's a top-tier developer and businessman, but unfortunately for all of us, that's all he is. As his one-time mentor, Roger McNamee, later wrote, what I did not grasp was that Zuck's ambition had no limit.

I did not appreciate that his focus on code as the solution to every problem would blind him to the human cost of Facebook's outsized success, and I never imagined that Zuck would craft a culture in which criticism and disagreement apparently had no place. His recent retreat from engaging in valuable social conversations, coupled with his attempts to actively hamper legislation, have made it clear, to me at least, that he's given up on the world and sees his critics as opponents that need to be beaten. So, what can be done? Well, there are three angles. The first and most obvious, that we've already spoken about, is legislation that erects appropriate safeguards.

First, we should force social media platforms to openly share their analytics and research so that they can be tracked and held accountable for their failings, and establish standards they have no choice but to follow. If that's not your preference, the second option is antitrust legislation, which could seek to break up Meta into smaller parts because it has simply become too big and too monopolistic, similar to what was done with Standard Oil in 1911 or AT&T in 1982. These moves have proven to be massively beneficial for fostering competition and innovation in the U.S. While this approach would not completely solve the problem of regulating social media, it would weaken Zuckerberg's iron grip, including his ability to funnel money into lobbying, set terms favorable to himself, and stifle open competition. At the very least, it would help diversify the strategies used to make social media safer and less intrusive.

And just so you know, at the time of writing, there are already three new antitrust legislation bills being actively discussed in the U.S. right now. The third way is to "kill" Meta with our feet—either by closing our accounts or by moving to a competitor. However, the choice of competitor is important here to avoid repeating the same problems. I don't think it's a stretch to suggest that most of us would be happy to leave Meta's platforms if there were a proper alternative that had the potential to achieve mass adoption and that we believed would do right by us in the long term. There are many alternative social media platforms out there run by rival tech companies or by non-profits.

It's an enormous topic, which I could dedicate an entire video to, but for now, I'll just summarize quickly. Since for-profit companies are clearly too dangerous to run a social media utility, the ideal solution would be a non-profit organization, like the Wikimedia Foundation. However, it would need to secure enough funding to be competitive because money is necessary for it to be useful, fun, and well-moderated. Apart from that, it would also need a system where its leaders are either re-elected every few years or ousted if they underperform. Then there's the experience itself.

Now, many of the existing non-profit alternatives fall into the trap of championing technical solutions to solve social problems, like federated social media using open decentralized technologies such as ActivityPub. These concepts may sound great to the tech-savvy, but they mean nothing to regular people, who only notice the usability challenges that make them confusing and annoying to use. Regular users should not need to understand how the fediverse works. They should not need to know what server they're on or how to post to other servers. It's not good enough. To compete with Meta, it's not enough to be virtuous.

It's not enough to have a cool technical solution; you have to provide an excellent experience, or there's just no point. Apart from that, any social media firm, no matter how virtuous, would eventually face awkward challenges regarding free speech. Therefore, it would need a public charter and an oversight board to address decisions that might curtail free speech. This process becomes much simpler when you don't have to worry about a hit on ad revenue.

Gosh, just thinking about it, I’d love to be involved in setting up something like this, or just joining an existing promising effort if there was enough momentum behind it. One can dream. So the final thing I’ll say, given everything we've looked at over this terrifyingly long video, is to emphasize the importance of trust. Trust is the most precious of assets, and Meta's biggest problem is that Mark Zuckerberg has lost it.

Looking to the future, we need to have solutions that we trust because it's impossible to avoid mistakes. We will never have social media that is completely safe. We will never have social media where horrific abuses, manipulation, or exploitation do not happen. However, what we can have is better people at the top, motivated by the right things, making the best of a bad situation. As I have just pointed out, we do have options here.

You can force trust by supporting safeguarding legislation. You can avoid placing too much trust in one person by promoting antitrust legislation. Alternatively, you can vote with your feet by closing your accounts or by finding a more trustworthy option. Regardless of the path you choose, if you agree that action is necessary, take whatever active steps you can to make it happen.