A Look Into Facebook's Soul — With WSJ's Jeff Horwitz and Ex-FB Exec. Brian Boland

Jeff Horwitz is the Wall Street Journal reporter who unearthed a trove of internal Facebook documents that reveal a damning disconnect between what the company says in public and its actions inside. Brian Boland is a former Facebook executive who spent more than 11 years inside the company. The two come on to discuss Horowitz's bombshell series of reports, unpacking what they tell us about Facebook, dissecting the company's responses, and looking at potential solutions. Enjoy this bonus episode! Learn more about your ad choices. Visit megaphone.fm/adchoices

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Alex Kantrowitz?

  This episode is brought to you by Genesis. What does it actually take to put Agentic AI to work across an entire enterprise? I recently spoke with Genesis chairman and CEO Tony Bates about why orchestration is emerging as a competitive advantage in customer experience and what companies need to get right in the AI era. Then I sat down with Adam Mitchell, head of enterprise business solutions at Voya Financial, to go inside a large-scale Agentic CX transformation, including the business decisions, trade-offs, and lessons along the way. You can watch both conversations now on my YouTube channel today. It's never been more challenging to get frontier compute. How do you know where to look, who to trust, or how much to pay?

• 0:43Alex Kantrowitz?

  Akash is an open cloud that provides easy access, flexibility, and peace of mind. That's how companies like Razor were able to cut their inference cost by 15 times. Visit akash.network/alex for a free consultation. That's akash.network/alex for a free consultation.

• 1:15Alex Kantrowitz?

  Hello and welcome to the Big Technology Podcast, a show for cool-headed nuanced conversation of the tech world and beyond. And today, we have an emergency podcast for you because this week in quick succession, The Wall Street Journal has published five fascinating damning stories about Facebook that it calls "The Facebook Files." They've come every day of the week and they paint a picture of Facebook that is fairly concerning and something that's worth discussing right away. And to do it, we have two amazing guests for you. The first is Jeff Horwitz. He is the journalist who obtained the documents and has been on all five of the stories. And he's going to come on and take us through his reporting a little bit. So Jeff, welcome to the show. Thanks.

• 1:59Alex Kantrowitz?

  And then we're going to take you inside the company because we have Brian Boland, who has been inside Facebook for 11 years or was inside Facebook for 11 years. As an executive there, Brian and I have gotten the chance to get to know each other over the years. So it's really great to have a chance to have him on. And not only are we going to talk about the problems, which is stuff Jeff can highlight, but we can also talk about the mentality inside the company and then potential solutions with Brian. Brian, welcome so much to the show.

• 2:26Speaker 01

  Hey, Alex. Thanks for having me.

• 2:28Alex Kantrowitz?

  Thanks for being here. Okay, so let's just quickly start by summing up the five stories that the journalist published. So the first is that Facebook has a multi-tiered system for content moderation that lets, you know, close to 6 million VIP users basically say whatever they want with little recourse, little chance their posts are going to get taken down. The second is that Instagram knows that it harms girls' mental health and has done little about it. The third is that Facebook's, Facebook made an algorithm tweak to keep the company relevant and it did so and it worked, but it did so by making people angrier on the platform and it passed it off at the time as something that it would encourage meaningful social

• 3:13Alex Kantrowitz?

  interactions and keep people, keep people's well-being intact. The fourth is that cartels and human traffickers are operating freely on the platform with little recourse. And then the fifth is that pertinent to this day is that anti-vaccine comments are dominating the discussion on Facebook, making a big chunk of the discussion far out beyond what they would in real life and Facebook is struggling to deal with. So Jeff, does that sound about right to you?

• 3:41Speaker 02

  That does sound about right.

• 3:43Alex Kantrowitz?

  So these are, these stories are pretty fascinating. They are built off of tons of documents inside Facebook. How did you get them?

• 3:51Speaker 02

  So I think, I'm not the first guy to be provided with a whole bunch of information from inside that company. And I think something's worth noting is that the information tends to get given to people like me from people who've been asked to handle societal grade problems that might be related to a new and very powerful form of technology and communication that Facebook has become the leading purveyor of. And I think the reason that this just seems to be happening over and over again is that Facebook has asked people who

• 4:34Speaker 02

  You know, or have really promising tech careers to sort of go into a different part of the company, one that is not considered to be a path to the C suite. Or, you know, necessarily even good for your career.

• 4:49Alex Kantrowitz?

  The integrity team.

• 4:49Speaker 02

  People are doing it because they really think it's important to, um, to actually like try to fix these things. And I think something, it's pretty hard if people feel like they've been asked to fix these things and they aren't being given the leeway and latitude to do it. And so I think that sort of that tension, um, becomes something that kind of makes people feel like they have a moral obligation to talk to people like me. Um, and obviously there's a whole bunch of obligations and whether it's right or wrong to leak, et cetera. People have plenty of debates on over, but I, I do think that the way these documents make it to folks like me is people have a moral crisis.

• 5:30Alex Kantrowitz?

  Hmm. Now, a big theme in your reporting is transparency. So in the name of transparency, will you release the documents that you got?

• 5:38Speaker 02

  Um, I really like that idea. Um, they are not, I am not the sole voice that gets determined that. There are other entities inside the Wall Street Journal and perhaps out that might have a say on it. Um, I would say TBD on it. I think that, that the question is a good one and a very fair one. And, um, nothing would make me happier than to let other entities have a go at this, um, either to find things that, you know, maybe we didn't or to, um, take issue with how we did our work. So I think, you know.

• 6:15Alex Kantrowitz?

  The outside entity with a say in it?

• 6:18Speaker 02

  I don't want to get into that right now. Uh huh. You know, no, no, like, like, let's put it this way. Like, like, it's not like it's, it's, you know, okay. I don't want to get into that right now. Yeah.

• 6:29Alex Kantrowitz?

  Brian, you've been inside the company. Uh, what happens when someone, you know, provides a publication with this much? I mean, this is sort of unprecedented, but they've provided the journal with all this information. Um, are they going to find the person that did this or the people that did this?

• 6:47Speaker 01

  Hard to say, you know, probably would be my guess. You know, the company prides itself on being open and transparent with the information internally. And, you know, based on Jeff's reporting today, that that may be changing, which would be in a lot of ways unfortunate. But, you know, the company takes those kinds of things seriously. So the steps this person took speaks to the level of conviction that they had around the importance of these topics. Otherwise, that that personal risk they took wouldn't be worth it.

• 7:15Alex Kantrowitz?

  Yeah. So if I were to put these stories into categories, the first one is that I would say is Facebook just being unable to manage its size. I mean, I think, you know, Jeff, there was a line from a Facebook researcher in your story about vaccines that was basically like, you know, we've created a monster and we don't know how to control it. And the three stories that sort of go to that are the fact that Facebook needed to exempt millions of people from content moderation, the fact that it couldn't effectively manage or wouldn't effectively manage the cartels and human trafficking activity on the platform, and the fact that it couldn't contain the anti-vax stuff. So is there a way to do this?

• 7:59Alex Kantrowitz?

  Or when, I mean, you know, they could have the best intentions, but if you're a platform of, I think, what is it, 2.7 billion people on any given day trying to manage a network that big, might just be physically impossible, no matter whether your heart's in it or not. So what do you think about that, Jeff?

• 8:18Speaker 02

  So I would, the first thing is I'd say I'm going to be interested in Brian's response to this question, but I don't think the tenor from people inside the company and from the documents that I've seen, isn't this stuff is hopeless. If it's hopeless, then, I mean, there's, you know, that's almost kind of like, well, you do what you can, and that's that. I think there are, you know, just saying it's too big is perhaps a little bit simplistic, right? Crosscheck, this is a manageable program. Facebook is That's the process of getting it managed. Yeah, I'm sorry. That's the special, the whitelisting and special exemptions. Crosscheck is what they renamed it because the word shielding didn't sound great. Sounded arguably a little bit too much like what it is. Yeah.

• 9:01Speaker 02

  And so I think like crosscheck and shielding and whitelisting and all of that stuff, this is manageable. Facebook, in fact, has been improving the program. The fact that it took them this long and that they weren't willing to rein it in earlier, in part because they were relying on it to get them through the 2020 election. I think raises some questions about, about sort of approach. The fact that it was so understaffed that in the review queues for VIP enforcement were so understaffed that they were in early 2020 only reviewing 10% of the material that had already been flagged because Facebook, Facebook systems basically said they believed it likely violated.

• 9:42Speaker 02

  You know, that's like, that's, that's a choice. That's like, that's inattention. It's not like woe is me, that could never be addressed. I think the same thing goes for some of the international stuff, right? I mean, like one of the, the story about the global issues just demonstrated that Facebook's flat out spending on international work is a pittance relative to their spending on, you know, the US and Europe. And even though that Facebook is very, very aware that the consequences of Facebook not getting things right are so much worse abroad. You know, that people literally die or are like sold into, you know, sex slavery. I mean, this is, you know, some pretty rough stuff and

• 10:22Speaker 02

  you know, you shouldn't need, as evidenced by the, by basically the way that one of that, that, that situation played out, which is Facebook didn't really do anything for a long time. And then when Apple threatened to kick the big blue app and IG off the App Store, suddenly just went into overdrive and took care of the problem. Again, that's like, That's solvable, right? And the fact that they then kind of went back to being inattentive after that is, again, like, solvable. It's just, that's will. I do think that there are some questions about how complex the system can get and whether Facebook, you know, I think something I ran across repeatedly in the documents

• 11:02Speaker 02

  was people just being surprised by like, oh, God, like, we just inadvertently took accident, uh, uh, uh, took action against a post from President Trump because he was not cross-checked or shielded from enforcement in this like one weird minor element. And we have these 60 different systems that are all sort of patchwork overlap and this one slipped through. I mean, this is a company that that literally couldn't prevent Mark Zuckerberg's own Q and A on misinformation.

• 11:30Alex Kantrowitz?

  From being flagged.

• 11:31Speaker 02

  From being classified as misinformation, right? So like, I'm just saying that like, I think it's one thing to talk about sort of responsibly monitoring the speech of 2.7 or 8 billion people and saying that like, well, of course that can never be done perfectly. It's another thing to just be like screwing it up because it's way too complicated and the attention isn't there.

• 11:53Alex Kantrowitz?

  Right. I mean, Brian, what do you think? Because I imagine inside Facebook, it's like, uh, almost being in this game of whack-a-mole where, you know, you might successfully address 20 problems, but 30 more will come up just because you're managing such a big, uh, network. So is this a, a true, uh, failure of will here? Or is it a product of the network size? Or is it something that I'm not thinking about?

• 12:18Speaker 01

  No, I mean, Jeff's done a great job in the reporting of highlighting how big of a problem this is, right? And how massive the scale of this issue is, that it's, it's global in nature, it's complex in the types of issues you're dealing with. So I'm incredibly sensitive to the fact that these are huge, huge issues and frankly, like, brand new issues to the world. No one's had to deal with this before. But that said, there's also a question of focus and effort, right? If you think about two other really interesting times in Facebook's history, where the leadership, and particularly Mark, focused the company on something important and got the will of the employees to move in that direction. One of them, you remember, he had that F8 talk where he talked about moving away

• 13:01Speaker 01

  from the move fast and break things to move fast with stable infra. And that signaled to the company that infrastructure is important, that the company can't scale without infra. The second famous one is the shift to mobile, where instead of having like a mobile team, which is what Facebook originally had, was they had a mobile, like one team, small team, building mobile. And he reoriented the company to say, we're going to build mobile first. And now, you look at the company, it's the largest mobile properties that exist. So it's possible to shift the attention of the employees when you tell them it's important. And I think that's one of the big missing elements, is that it's reactive safety as opposed to proactive, safety first by design. And I think that while these are huge problems,

• 13:46Speaker 01

  the company has hired some of the brightest minds in the world to work on growing the company. It also has these same bright minds who could work on these incredibly complex problems and start to make progress on fixing them. That signaling that safety is core, that understanding the impact of products in a harmful way, not just a growth way, is core, would be absolutely critical.

• 14:09Alex Kantrowitz?

  Yeah. And so, I mean, what is the rationale for not making this a priority? I mean, it's obviously exploded in some, you know, pretty dramatic ways for the company, at least this week, through Jeff's stories. It seems to erode trust among users, yet Facebook hasn't made it a priority, obviously, in the way that the others you speak of. And I'll offer one possibility. I mean... These scandals don't seem to hurt Facebook's bottom line. Jeff, you pointed this out in your first story, Brian. You worked in the ad side during some of Facebook scandals and I know what happened there.

• 14:53Alex Kantrowitz?

  I mean, the money stayed, you know, on course. It kept coming in from advertisers despite their fake boycotts and stuff. So is it really, I mean, is it, is it a money thing for Facebook? Is it, it just doesn't see the financial consequences here? Or is it, is it something else? Is it that Mark Zuckerberg, maybe this is just sort of like, I guess the other side of it is maybe this is a product thing. And if you start to rein in this stuff, you might not get the same amount of usage and engagement. So, and that would eventually lead to, to a money thing. Maybe it's not money. I don't know, Brian, what do you think?

• 15:31Speaker 01

  You know, hard to say, like, I'm not in everyone's head, but I do think that there's this ideological thing that you've seen come up time and time again. You saw it from Messeria this week, in this notion that the good is outweighing the bad. Right, is that with these massive transformative things that you have both good, you have both bad. And I felt like the leadership just firmly believes, and mind you, believes without researching it or having deep evidence of the fact, that there is way, way, way, way, way more good than the harm. The challenge with that to me is that when you're dealing with, was it 3 billion people you were talking about?

• 16:10Speaker 01

  That when you start to take percentages, even like 1% harm on a 3 billion population is just freaking massive.

• 16:17Alex Kantrowitz?

  Yeah.

• 16:18Speaker 01

  So I think this ideological thing comes to play. And I think the challenge is that a lot of the things that you would do to test improving systems lead to impacts in usage, engagement, revenue. And I just don't know whether the company would be willing to take those impacts.

• 16:35Alex Kantrowitz?

  So there is some financial risk there, combined with this optimistic view of what the products are.

• 16:41Speaker 01

  Well, yeah, anytime that you would make massive changes, and this is what kind of really popped for me in a lot of Jeff's reporting, is this notion of you being worried about sacrificing engagement or sacrificing core metrics to test some of these things. So if you were to say, look, like, we're going to make this a top company priority, we're really going to investigate our products, and we're really going to try some things to clean this stuff up. There are going to be unintended consequences of those, but some of the consequences will be on core metrics, which I just don't know that the company would do.

• 17:11Speaker 02

  I think there's a really interesting thing that sort of arises, given, look, inside the company, one of the problems for people on the integrity team has been that content quality and questions of sort of societal good are really hard to sort of justify, right? Like, is, you know, it's really hard to sort of benchmark, right? Because Facebook can tell you exactly what even the most minute change to its platform is going to do to user metrics, right? You like change the size of the comment box by two pixels, they can tell you whether or not that was a good or bad thing for those metrics. They can't tell you nearly as well about quality, and obviously, there are some inherently subjective issues here.

• 17:57Speaker 02

  And I think the really interesting thing about the current, like, we do more good than bad thing, I mean, it's not current, that's been historically the only thing, is that that claim, and so a lot of people in integrity have sort of always been frustrated because they're like, look, just like, very clearly, yes, it's subjective, but very clearly, the things we are doing here are really bad. And, you know, the response is, well, define bad and quantify it for us, you know, and then come back, right? At which point they walk off with like shaking their heads because that's impossible. The interesting thing, though, is that the company's response on this and like Adam Mosser's, you know, we do more good than bad line on Instagram, is like, the exact same concerns apply, which is like, okay, like, how are you going to quantify that?

• 18:39Speaker 02

  They are very convinced of it. There's no question that Facebook legitimately, from the bottom of its heart, at the senior levels, believes that like, basically, they are being beset by like a whole bunch of chattering idiots who aren't appreciating how Facebook has, you know, improved society. That's like absolutely how the company feels.

• 18:57Alex Kantrowitz?

  Is that, is that how this is going to be received inside Facebook, Brian? Are people going to look at these stories and be like, now the press is out to get us and, you know, ignore the noise and charge on?

• 19:09Speaker 01

  I think it's gonna be hard with these because a lot of this reporting has both internal documents and concerns from teams. It also, man, like the content in this stuff, the topics is just so, so, so awful.

• 19:26Alex Kantrowitz?

  Right. Like, I understand. Yeah, I understand how like getting, you know, believing you're doing more good than bad, like, you know, it sort of prevents you from taking some serious action to crack down on the platform. But when you have these examples, I think like you're referencing of human trafficking and cartels posting, you know, their murders on Instagram and, you know, a country being pushed toward vaccine hesitancy. It's, I just don't see how that's like something that, you know, you can keep going on and saying, "All's good."

• 19:58Speaker 01

  Yeah, I don't know how. Like, when I, when I got to the point where, where it was time for me to leave, it wasn't because, you know, I didn't have a great job or like wasn't working on amazing things. And frankly, it wasn't because I wasn't working with amazing inspirational people. Like, absolutely was. I got to the point where I was concerned about polarization in the platform, just racial polarization around the stuff around George Floyd and last summer. And then continuing to see the lack of effort to really understand different categories of polarization. That was enough for me to feel like I, I questioned whether we're doing more good than bad and I have to leave. I didn't know about all this other stuff.

• 20:35Alex Kantrowitz?

  Right.

• 20:36Speaker 01

  You know, like, this is, this is the ugliest of the ugly.

• 20:40Alex Kantrowitz?

  So how can it get solved? I mean, you know, Jeff, I saw you were on CNBC talking with John Ford about this. And he asked you, like, is it gonna cause any change? But Brian, you're, and you know, Jeff, you politely said, "Listen, all I can do is put the information out there." Brian, you've been inside the company. So I'm curious if you have any ideas about like, well, we talked about whether it's gonna change. Well, how do you solve this stuff?

• 21:03Speaker 01

  Incredibly hard things to solve, right? Like, I'm super sympathetic to the scale of these problems and they're global, right? So you have endless languages and countries to deal with in the context of these issues. And Jeff's just highlighted a handful that probably exists in a lot more places and probably are as ugly. So I think there's two things. One, you really need a signal from Mark. Like, the company follows Mark's direction. If Mark said, "You know what? This stuff is abhorrent and we are going to shift towards really understanding and ending..." frankly, if they want to claim it's a narrative, an anti-Facebook narrative, fine. Like, let's end it, improve once and for all how the platform does shape and whether it's net good or net harmful for society.

• 21:46Speaker 01

  You will never be able to measure it with some like 80/20 kind of sort of metric, but you could start to do research that a lot of these internal things would be done with external research. So, one thing is Mark's got to signal that the internal teams, like, this is important, this is a priority, and it's a goal. And like, sacrificing growth or sacrificing revenue on behalf of safety is important. Second one is transparency. You know, a lot of these issues, you could start to see if more of the public data was transparent in the way that CrowdTangle brings engagement data to the public. There's another report about troll farms out today from MIT that was super interesting. And it was the kind of thing that you would find if you had reach data. Again, that's the data that Facebook says is more important than engagement anyways.

• 22:29Speaker 01

  And here's a thing showing that for Christians and for Black Americans, the most popular pages on Facebook were troll farms, right? And transparency brings the ability to see that. And then, look, it's time for regulation. Like, regulation is the kind of thing that happens when industries grow and they're successful. And it's happened in the autos industry in the 60s. It happened with chemicals before that. It happened with building codes. Like, imagine buildings without building codes. Time and time again, we've brought in regulation to help industries do a better job of being safe. I think the regulatory approach right now is super important.

• 23:06Alex Kantrowitz?

  Right. And I guess like the key here, right, is that this stepped up, just to bring it back to business. The stepped up enforcement would be a labor cost. And, you know, if, I mean, Jeff, your story highlighting international, right? If you start to put the same amount of content moderation in every country as you have in the US, which is in Canada, where Facebook makes most of its money, it would cost a heck of a lot in labor costs. Facebook has a 26, you know, multiple price equity ratio.

• 23:38Speaker 02

  So I actually think-

• 23:39Alex Kantrowitz?

  If you hire that, you know, more people, then the multiple goes down. So yeah, go ahead. What do you think about this?

• 23:44Speaker 02

  Yeah, I actually think, look, the idea of having every country having the same level of investment in content moderation that's occurred in the US is not possible under Facebook's current system. Their current system, as we note in the global story, is very much language-based. The US has the best. After that, hello, romance languages. After that, maybe Hindi. After that, bye-bye. I mean, it's not that bad, but there are, like, a lot of the classifiers don't work. The algorithmic systems for content detection just flat out do not work or do not exist. And this is sort of like every incremental country

• 24:28Speaker 02

  or every incremental community, like linguistic community that Facebook goes into is almost definitely going to be smaller and poorer than the last, just because that's the order of operations for growth. And so they've been kind of moving further and further down the scale in terms of societal wealth and kind of ability to get attention. And with a system that is very much specific language-based. So I don't think it's like necessarily reasonable to expect that they would be able to pull this off on the 100 and whatever languages they've got that currently offer services on in the platform. And that's not to say additional languages that they just don't offer services, but people may do.

• 25:09Speaker 02

  I think that something that occurred to me a lot in the course of this reporting is how much of these problems would exist with Facebook circa 2008. And now, obviously, this is like a very much a counterfactual, I think.

• 25:26Alex Kantrowitz?

  We like this here.

• 25:27Speaker 02

  And I'm not suggesting that it's time to go back to Facebook 2008 or 2006. But I am saying that- Many of the issues highlighted, the groups in which maids were being sold, the concerns about virality and recommendation systems, these literally didn't exist in the early iterations of Facebook. And arguably, if the goal is to make it possible for people to connect each other, that goal was being carried out fairly effectively in the early days. I mean, any person in the world could, for free, log on and chat with any other person

• 26:10Speaker 02

  in the world. You know, like, "congratulation, mission accomplished." But the product has sort of continued to get increasingly complicated. And it does seem like a lot of the problems that Facebook has have come out of those iterations and sort of new features. And again, where lines are drawn isn't clear, but I know that's been a frustration to people inside the company, which is that, you know, they keep on building and they expect integrity to clean it up, whereas the previous version was safer.

• 26:41Alex Kantrowitz?

  That says so much. Last question before we go to break. Jeff, three of these stories are essentially on content moderation. There's going to be people who say that you're a whiny reporter that wants Facebook to censor. What's your response?

• 26:58Speaker 02

  Oh, I think censorship is... One of the interesting things about censorship is you have to censor things. Look, letting anyone say whatever they want to say, it works great on the open internet. There are Nazis on the open internet. My name's Horowitz, right? So, certain ethnic background. I don't mind Nazis being on the internet. They can, you know, have their own websites and Nazi it up as much as they want. I think the problem is when you start, like, basically providing tools that make small...

• 27:27Speaker 02

  Potentially militant communities more powerful? I think that that is something that perhaps raises different questions. So in some respects, I think the fact that Facebook has to moderate its platform as much as it does is because of how viral things are on the platform. So I guess I'd say like, I think there's actually like, I think this is like a really interesting thing when writing about Parler, right? Like, everyone hated Parler because, well, a lot of reasons.

• 27:58Alex Kantrowitz?

  Not everyone.

• 27:59Speaker 02

  But, but like-

• 28:00Alex Kantrowitz?

  Parler users liked it, but sorry.

• 28:01Speaker 02

  Yeah, okay, it's true. It's true. Everyone, everyone in our general profession kind of looked askance at that thing. I'll just say that. And I actually think they didn't do algorithmic content ranking. You just saw the things from the people you followed. And I think that that was a really interesting, in some way, limiting factor on how bad things could get. I'm not saying that Parler was good and that there was a ton of super, it wasn't a ton of super bad stuff on there, but I am just saying that like, even with a commitment to not having rules, if you don't have recommendation systems, it's kind of a less risky business.

• 28:44Alex Kantrowitz?

  Yeah. Okay, I do want to get into the algorithm and the recommendation system. So why don't we do that right after the break? We'll be back here on the Big Technology Podcast right after this with Jeff Horowitz and Brian Bullant. This episode is brought to you by App Point. Everyone's racing to roll out AI right now. Copilots, chatbots, agents doing real work. But here's the part nobody loves talking about. All that AI runs on your data, and most teams have no single way to see it, secure it, and prove it's under control. That's exactly what App Point does. For 25 years, they've been the trusted layer beneath the world's most demanding data, now extended across your entire AI estate.

• 29:25Alex Kantrowitz?

  Your data, your cloud, and the agents acting on your behalf. It's how more than 28,000 organizations deploy AI with confidence. So innovation scales without scaling risk. App Point, the unifying trust layer for AI. Head to apppoint.com to see how enterprises deploy AI with confidence. Learn more at avpt.co/bigtechnologypodcast. That's avpt.co/bigtechnologypodcast.

• 29:58Alex Kantrowitz?

  And we're back here on the Big Technology Podcast, talking about the Facebook files of bombshell Wall Street Journal investigation into Facebook that features tons of original documents and memos that we've been discussing. Well, there was one part of the reporting that I found most interesting, and that was this, what happened when Facebook optimized its algorithm towards meaningful social interactions. Facebook at the time told us that this was all about improving people's well-being. But the truth was that it came at a time in 2017 where, according to your reporting, Jeff, was it comments, reshares, likes, and original posts were all declining, and without that engagement, Facebook essentially becomes a shell.

• 30:43Alex Kantrowitz?

  People stopped going to it. And Jeff, from your reporting, what happened was the shift worked. People were more engaged, but they were kind of just hanging out in flame wars, you know, in the comments, going back and forth, yelling at each other about controversial stuff. And publications and politicians took note of this and started putting more outrageous stuff on the platform. So I guess like Jeff, maybe you could, first of all, you know, let me know if I got that right. And then, because, look, it's been a week, spending my mornings reading your stories.

• 31:16Speaker 02

  Trust me, it's been an even longer week over here.

• 31:18Alex Kantrowitz?

  I believe you, it's probably your whole summer. But number two, I want to ask you, and then, you know, we'll toss it over to Brian after this, but like, does Facebook's existence and getting people angry, are they kind of linked in some way? Because it is interesting that like, when they made people matter, they ended up reversing some of the downtrends that they were seeing.

• 31:37Speaker 02

  I don't know that it needs to be that tightly linked. And again, like, we are now entering dangerous reporters for, territory for a reporter because, you know, thinking about platform design, there's, there's a reason that like, no one's on, you know, Jeff Horowitz's book, right? And that said, so there were two parts with the meaningful social interaction thing, I think was interesting. One of them was to focus kind of more on content from friends and family. And, you know, that was kind of, kind of trying to take Facebook a little bit more back to its roots, and...

• 32:17Speaker 02

  Research the company did found that that was, I think, from an integrity point of view, content quality point of view, sort of unambiguously positive.

• 32:25Speaker 02

  And, you know, in fact, like, definitely something that Facebook can do when they want to, you know, if they're worried about content quality issues, is they can show people more stuff from their immediate friends. It tends to be safer and, you know, like, nobody starts to flame wars in on vacation photos, right? I don't know, we might have different parts of friends on Facebook. Possibly not, but you know, you know where I'm going. But, uh, but so the, uh, the other part, though, was that they did explicitly want to goose engagement, right? And, you know, the way you goose engagement is you simply reward content that uses engagement. And, um, they did that quite successfully. It's not like that was the first time they'd

• 33:06Speaker 02

  ever sort of, you know, done engagement based ranking. I mean, this is a company that originally started out with just like likes, you know, that was the cumulative, that was the metric, right? And, um, but that said, like, they, they did switch more heavily to this. And I think the difficulty they had first teching it as an issue, um, it was kind of a hindsight based thing of like, uh-oh, you know, like, oh, God, we did this, uh, is interesting. And, you know, I don't know, though, that like it has to be that way. There are other ways to weight things. Um, you know, you don't have to, like, I mean, one, one thing

• 33:50Speaker 02

  that was of, of interest was like, how much should the emoji, the anger emoji be worth? Cause it turned out that like, that was a very specific thing that you could optimize for as a publisher. Um, I mean, the concept of hate bait, right? Um, you know, just riling people up and it was very successful. Uh, I think, you know, one of the things from MSI that I found fascinating that my colleague Keach, uh, Keach, uh, reported out was that,

• 34:18Speaker 02

  It turns out that targeting political ads to your opponent's supporters was a strategy that some

• 34:25Alex Kantrowitz?

  What?

• 34:25Speaker 02

  Oh, my God.

• 34:26Alex Kantrowitz?

  Because they get angry.

• 34:28Speaker 02

  Yeah. And this is a new thing under the sun, right? Like, yes, politics has always been negative, but nobody ever was like, OK, sweet. Like, I'm gonna go find

• 34:37Alex Kantrowitz?

  I'm gonna deliberately piss people off.

• 34:39Speaker 02

  Yeah, like I, a liberal Democrat, am going to, you know, I'm gonna go basically buy up every slot on like Glenn Beck. Like, that was not like a thing that anyone ever thought of doing.

• 34:49Alex Kantrowitz?

  Right.

• 34:50Speaker 02

  As a, as a means of getting distribution.

• 34:52Alex Kantrowitz?

  Yeah, you're definitely, you're in a troublesome spot when you start to say, I'm, the way that I beat this system is to make lots of people who don't like me, uh, super angry. Uh, I do wonder, sorry, go ahead.

• 35:05Speaker 02

  No, no, I was just gonna say, and I think something just very interesting about how people engage with the platform is that I guess I'd say, like, we're sort of in the middle here. I think most people use Facebook because they want to use Facebook and because they think it's interesting. We use social media because it's kind of relevant to our jobs. Yes, we do compare, you know, like, look, no one ever tells me, like, Jeff, you need to go get more clicks, right? Like, but obviously getting, you know, getting attention is a good thing. Um, but then there are also, I think, entities that are just like purely dedicated to getting attention on Facebook. And there are definitely very effective ways, and those entities are the ones that are

• 35:49Speaker 02

  gonna be most gearing themselves to trying to, one, understand how the system operates and what it rewards, and, two, trying to game the living hell out of it.

• 35:56Alex Kantrowitz?

  Yeah.

• 35:57Speaker 02

  Because, I mean, everyone else is gonna like, Facebook makes an algorithm change, great, they're gonna keep on posting vacation photos and like interesting news links. However, like, like basically crap Macedonian publishers, they are gonna work hard. And like, so you expect them to be the ones that adapt much harder than everyone else. And so there's something that's like really adversarial about The kind of Facebook attention garnering community.

• 36:23Alex Kantrowitz?

  Yeah, I spoke about this with Neil Yal a couple of weeks back. We had an interesting discussion about this. Brian, you were inside. So, and you have a little bit more freedom, I think, than Jeff does as a reporter to talk about this stuff. And definitely an inside view. So what's your perspective on how meaningful social interaction has gone wrong? And is it that simple that angerous stuff will make Facebook more successful? Or is there something more nuanced that I'm missing here?

• 36:51Speaker 01

  Yeah, no, I'm gonna have to go with what the documents found, because it was the folks who are researching it and looking at stuff. But, you know, look, like, a lot of what we keep talking about here is the role of algorithms and the role of artificial intelligence in this stuff. Right? Because the new thing for mankind is the algorithmic determination of what you will see. Right? Like, here Facebook talk about, hey, look, you're just joining the town square. This is the public town square. Right? But it's really not that. Like, it is an algorithmically personally created town square, where you're lumped into a set of content based on what the algorithms are optimizing for. And I think it's very, very hard to design algorithms that you understand that they're driving certain

• 37:35Speaker 01

  quantitative things versus the metrics, which Jeff was talking about earlier. You know, like, optimistic builders are like the kind of people who work at Facebook. And optimistic builders, you don't do your product pitch deck with a big section on what could go wrong? Or what are the ill consequences of these things?

• 37:52Alex Kantrowitz?

  Yeah.

• 37:52Speaker 01

  Right? Like, what's the worst? Like, nobody imagined at Facebook, the Macedonian kind of stuff.

• 37:57Alex Kantrowitz?

  Right. Yeah, it reminded me of Amazon, where like, they write those six pages, like, saying what's gonna happen with the product that they want to build once it goes out in the wild. And, you know, they've told me that, oh, it's like writing science fiction, except they always imagine the best case scenario, and there is no room for what could go wrong.

• 38:14Speaker 01

  Yeah, and those are the kinds of folks who have built some amazingly powerful products, right? It's like, it's optimism. And I get it. And I was in it. And I was super optimistic. But there's this need for the really critical worst case scenario thinking. Right? And I think, you know, in this case, the algorithms, when you read it, the algorithms rewarded that kind of angry content.

• 38:39Speaker 01

  Wasn't the design, that wasn't like what the team was hoping for, like, hey, let's just get a whole bunch of people pissed off and riled up and like, boost engagement.

• 38:48Speaker 01

  But it does lead to just way more focus on the worst-case scenarios from the beginning. And then when you do release it, don't just see what the growth metrics look like.

• 39:00Alex Kantrowitz?

  Yeah.

• 39:01Speaker 01

  And really important from Jeff's comment earlier is don't leave it to some other team to clean up. Right?

• 39:07Alex Kantrowitz?

  Yeah.

• 39:08Speaker 01

  That's a problem when you're saying, hey, look, we'll have an integrity team who deals with all the bad stuff. And I don't even need to worry about it. I'll just build. Right? It's that making it a part of everyone's job and a part of everyone's plans to build safely. That's really important.

• 39:23Alex Kantrowitz?

  Yeah, it's a good moment to talk about Facebook culture for a minute. I mean, it's so interesting. This is so interesting to me because Facebook has a very strong feedback culture where you're encouraged to bring ideas and suggestions and tell people it's wrong to anybody across the organization. But it requires a willingness to listen. And one of the things that I wonder if the message has gotten through, and obviously the message didn't get through here, which is interesting. But compensation, where Brian, maybe you can enlighten us a little bit on this. But from my understanding, people that build the products are compensated primarily based off of growth. And if that's the case, how does this ever change?

• 40:06Speaker 01

  I mean, you're answering the question by asking it, right? Which is compensation and goals needs to change to think about and reflect on deep understanding of what the product can do from a harm standpoint and then mitigating those harms. Right? So there's, I was talking with Brian Settler a month, two months ago now. And he kind of brought this great quote of when you invent the plane, you invent the plane crash. When you invent the car, you also invent the car crash.

• 40:32Alex Kantrowitz?

  Yeah.

• 40:33Speaker 01

  But it's really being thoughtful on what the car crashes and plane crashes are here at the beginning. And you can goal teams on that, right? So it's more of a softer metric and people who pay out goals hate soft metrics. But really, you can, you can, you can, like, you do end up compensating people on how they are as a cross-functional partner, how they are as a peer, how they are as a mentor, those get paid out. So you're rating people on their ability to really think through the implications of their products is absolutely something they could go on. And then making making safety a part of every team's problem. So like if I was gold, pretend I was building a product.

• 41:14Speaker 01

  And if I was gold on the integrity issues on my product is a core part of my compensation and advancement. Even if there was another team that was primarily responsible for that, I would pay a lot more attention to it. I, as a product group leader, would be having reviews on what's the latest on integrity in my platform. I, as a leader in the partnerships organization, would be saying, hey, look, let's go through an analysis. And let's go through the latest on understanding of integrity issues around my partners. Right? But because that's not part of what you're compensating on, it's somebody else's problem, it's an afterthought.

• 41:45Alex Kantrowitz?

  Yeah. And engagement is what keeps the company going. And it seems to me that it's much easier to take half measures and then say, we're always looking to do better than take aggressive measures, like changing the way that you compensate, even if it risks some of the growth. And social media is so competitive. I mean, Facebook, you look at it, it's got Snapchat, YouTube and TikTok all, you know, trying to take its users away and threatening successfully. So it's a tough problem. And I do wonder if the company even believes that it can, from a competitive standpoint, make any changes here.

• 42:19Speaker 01

  Well, you know, I think the interesting thing is, is that, is that those other companies you mentioned have been super quiet in this conversation. And there's a reason why. And to Facebook's credit, right, to Facebook's credit, they have looked. Yeah. They've actually put some, some effort in trying to look at some of these issues and trying to understand these issues. My fear is that the other companies aren't looking and they're not researching these things. And none of them are transparent, right? So, you know, Facebook keeps talking about how we're doing more than everybody else. And that is factually true. They're doing more than everybody else, but it's still woefully inadequate. What also terrifies me, though, is what's happening on TikTok. You talk about algorithms and virality and amplification.

• 43:01Speaker 01

  That is the most amplified and viral app that you can find. And nobody knows what's going on in that platform. So, yes, these are Facebook issues. I give Facebook credit for allowing their teams to continue to do some of this work. I hope, hope, hope, hope that one of the reactions isn't to clamp down on research. Because that, that will be a reaction in turn. Somebody internally will say, we need a review board to approve all future research, guarantee it. And like that, that, that would be awful.

• 43:31Alex Kantrowitz?

  Yeah.

• 43:32Speaker 01

  But we also need to remember that these other platforms aren't looking at it and they have to.

• 43:36Speaker 02

  Yeah. I would, I would actually second that. I think, look, like, I do not have details here on what the equivalent of the cross-check slash special exemption slash whitelisting slash shielding situation looks like elsewhere, but it has been darkly intimated to me that Facebook is not the only company that has different sets of enforcement rules depending on who you are. And, and yes, I mean, I think like the idea that Facebook should, needs to take more care when dealing with high-profile accounts is of course obvious, right? Like, there's no, you know, it's not like the, the alternative to what Facebook had, which was whitelisting,

• 44:17Speaker 02

  was like, just basically taking down Donald Trump whenever anyone felt like it. Um, so there are other ways to do it. I think some of the questions are what level of reach things that are previously flagged should get while they're awaiting adjudication? How long the review queues are? Uh, and also sort of like what level of difference in outcome one expects from the overall systems, right? Like, obviously, if you just eliminate stupid errors, it's still a favor, but it's less of a favor than if, let's say, senior leadership weighs in asking for a penalty to be waived on one of the account's top influencer, or

• 45:02Speaker 02

  one of the platform's top influencers, right? So there's like, there are, there are like, I think, what I would say, clearly obvious, better and worse ways to do this from a, we can explain that to the public and no one's going to get super angry point of view. Uh, but I, you know, I, I truly do not know, and but I said, have heard some dark mumblings regarding of what that looks like elsewhere in the industry.

• 45:26Alex Kantrowitz?

  Yeah, the TikTok stuff in particular, we just had an episode about the risks of TikTok's rise. And when you start to think about it, you do start to shiver and start rooting for Facebook, because you don't know what's going to happen with that. Do you guys have five minutes to talk a little bit about the Instagram stuff? I know we're kind of out of time here. So we can end.

• 45:47Speaker 02

  I can do a couple more minutes. You're, yeah, you're going to be, you're going to be upsetting German public television.

• 45:54Alex Kantrowitz?

  Okay.

• 45:54Speaker 02

  You know, I think we can make that happen.

• 45:56Alex Kantrowitz?

  Okay. All right. Sounds good. Sorry, Germany. We'll be back after this for five more minutes.

• 46:00Speaker 03

  Now, more than ever, we need someone on our side. But Rob Whitman? Whitman's votes drove up costs, cut Medicaid, and risked 350,000 Virginians' healthcare to pay for tax breaks for himself and his donors. He's not on our side. Shannon Taylor will fight corruption, cut taxes, and bring down costs. This November, vote for Shannon Taylor. She's for us. Paid for by HMP, the House Majority PAC dot com. Not authorized by any candidate or candidate's committee.

• 46:31Alex Kantrowitz?

  And we're back here for one final segment on the Big Technology Podcast with Jeff Horowitz, Ryan Bullen. We talked a little bit about, in the second, in the second segment about when you design the plane, you think about the plane crash, and the car, you think about the car crash. It's a good segue because Instagram head Adam O'Shaughnessy talked about how, you know, cars do a lot of good, Instagram does a lot of good, and you can't just throw, you know, throw the invention out because of the fact that, you know, I guess the intimation was it might kill people. And then everybody went, you know, kind of wild dunking on the tweet, talking about how, you know, cars are regulated, you know, through the roof before, you know, the industry became safe for people. And Adam in the podcast, he said this in

• 47:16Alex Kantrowitz?

  the podcast with Peter Kafka and Recode Media, and he said, you know, they welcome regulation, although certain kinds of regulations. So.

• 47:24Alex Kantrowitz?

  What do you guys think? Is that a fair defense from Adam? He also said, I'll just throw all his points out there. He also said that, you know, we never really did research on like how fashion magazines were making girls feel back in the day. And, you know, Instagram has supportive communities as well as like this comparison culture to it. So I wonder what you think about the defense.

• 47:45Speaker 02

  Yeah, I'm going to defer to Brian here because the automobile analogy is one that he actually raised with me well before the story started running when I went and sought him out after he first surfaced in a Kevin Ruth story. So I will, I will leave this to him because I have already cribbed an analogy from him once.

• 48:07Alex Kantrowitz?

  Okay, cool. Brian, you're up.

• 48:08Speaker 01

  Yeah, no, look, and like I did find it ironic and knew that people would jump all over the regulatory side of it because the NHTSA didn't exist before there was public pressure in the 60s. And Ralph Nader's book, Unsafe Any Speed, came out and people really understood the harms. And they importantly understood that the auto manufacturers weren't going to invest in safety and they weren't taking the steps that were required. And that's, that's happened every time there's been a regulatory regime, like chemicals do good. Like we use chemicals every day, but that was an industry that wasn't regulated. We live places, we have cities, we have buildings, we have building codes and regulation that governs those, those didn't exist, but now they do because of safety issues. So there's clearly a need for regulation here.

• 48:51Speaker 01

  And, and you know, the, the defensive analogies that make me cringe are those that, you know, we never studied magazines for their body image issues. Because it dodges the question, right? And tries to get you to look somewhere else. The question is, you have evidence that you have researched that there is potential harm being done to a significant, not an insignificant, a very significant portion of your user risk. And just wouldn't you want to know, like, if I told you that the basement of your house, I was pretty sure had some toxic chemicals in it? Would you let your kids still play in there and still give them ice cream in the basement?

• 49:33Speaker 01

  Or would you be like, look, like, I just want to make sure that I have a good understanding of whether my basement has toxic chemicals or not.

• 49:39Speaker 02

  Like, hey, other houses have chemicals.

• 49:41Speaker 01

  Right, yeah, look, like, hey, you know, like, not all the kids wait on somewhere else. So that, that is like, to me, the, the comparisons to other industries and other things in time. If you had done the homework, if you had really, if I felt like you had done everything in your power to understand the impacts and that you were working deeply with regulators at the academic sense to understand the impacts, fine, like, we can do that analogy. But that analogy doesn't hold because on its own, you have a flag and you should research to your deepest ability and in partnership, whether there are these ill impacts or not.

• 50:17Alex Kantrowitz?

  Hmm, 100%. All right, Jeff, if you have one takeaway you want people to get from your reporting this week?

• 50:24Speaker 02

  Um, you know, I think I'm going to be some people who got me much smarter on the subjects that were in these stories. Like, look, I started covering Facebook two and a half years ago, and I had not previously covered tech, and I didn't use social media very much. So like, basically, all glory goes to people who have bothered to walk me through this stuff.

• 50:50Speaker 02

  I, I think that, that just goes back to what you were saying at the beginning, which is just like, is it all hopeless? You know, that definitely isn't the impression that folks have that I've spoken to, right? That there are, you know, yes, there, there are always going to be trade offs, but that perhaps the trade offs, if understood, and if others were at the table, aside from just the company, the trade offs might be different and perhaps more acceptable to a lot of people than they are. So just, you know, that's obviously getting past my capacity to.

• 51:28Speaker 02

  Speak or think or come up with ideas, but you know, it seems like there is a growing body. There's. And I think one of the things that's most interesting, there's a growing body of people who've been like talking about this stuff on Twitter who like candidly understand it all a hell of a lot better than I do. And that's because they came straight out of the company. And so I think it's like really encouraging that folks like Brian are out there. It's even more encouraging that people who had like senior roles doing integrity stuff are like just popping up and I'm hoping that like there may be a kind of new class of Facebook trained people who are, I believe, about as smart in this stuff as anybody in the world, who are going to be kind of out and about and perhaps

• 52:11Speaker 02

  maybe offering some ideas beyond just like, well, this seems kind of messed up, which is kind of where where we end up. Right. And Brian is one of those people. Final thought. There's hope.

• 52:22Speaker 01

  There's absolutely hope, you know, like the platform has done some really great things. Just I think last year the platform helped register four million people in the US to vote. Like that's a, that's a big, big deal, right? To be able to do these things. You know, I think there's two things that I'd love to see. One, I wish I, you know, I was, I pled my case internally, but I pled it quietly. And I pled it to leaders at the company. I think it's time for people internally to really speak up and rise up and use their force and their voice as employees to say enough is enough. And retaining that exceptional talent is critical for Facebook. So people speaking up and people, you know, really mandating these as priorities is important.

• 53:04Speaker 01

  And then it is time for regulation and it's time for the right regulation. And I think that's the thing that I'm most hopeful for in the US. But note, a lot of the regulation you may get in the US and in the UK and in the EU is going to focus on those countries. It's these, you know, South America, African countries that.

• 53:28Speaker 02

  I believe the phrase is rest of world, Brian. Let's just call them rest of world.

• 53:32Speaker 01

  Rest of world. You know, countries, I do work in Kenya and Uganda and that people, people rely on Facebook, right? It's the thing they get everything from. And so if we if we do regulation in the US, we do regulation in Europe, but we don't take care of Kenya or Uganda or Ethiopia, then we're leading to undue suffering around the world and we can't let that happen. But like there are paths to improve this stuff, but it's time to take it seriously and do it.

• 54:04Alex Kantrowitz?

  Yes, definitely. Facebook employees, you have your marching orders. All right, Jeff, if people want to read the series, I guess they just go to the homepage of the Wall Street Journal.

• 54:13Speaker 02

  Actually, I think, look, I don't entirely understand this, but the paywall has been lifted for people visiting through social media. If that doesn't work, try it in a different browser window. If that doesn't work, I'm sorry, I don't understand why.

• 54:27Speaker 02

  If you're a Facebook employee, of course, you can personally reach out to me and I will send you the documents personally, as well with a request for you to talk to me in the future. No, seriously, I'd actually would love to hear from people who work there, even if they just want to tell me why they think I'm an idiot, with the understanding that that will be treated as off record. That's all fun, too.

• 54:44Alex Kantrowitz?

  That's very generous, very generous. And Brian?

• 54:46Speaker 02

  That's not very, very, very self-interested, honestly.

• 54:49Alex Kantrowitz?

  Well, yeah, it's good. Hey, look, if you're able to do more of this good work, everybody's going to benefit from it. So we appreciate it. And Brian, if people want to reach out to you or learn more about the Delta Fund, how could they do that?

• 55:04Speaker 02

  Brian, you're either muted or you walked.

• 55:06Speaker 01

  Oh, sorry, I was muted myself. Man, I'm muted the whole time without doing that very 2020, 2021 air.

• 55:16Speaker 01

  Yeah, so happy for people to reach out to me. You know, I'm on Twitter, people can DM me there. And then the deltafund.org is where people can learn about some of the philanthropic work and economic investment work that we're doing, which is in a lot of places all around the world and trying to help people improve equity.

• 55:37Alex Kantrowitz?

  Amazing. Well, with apologies to German television, I want to thank both of you guys. Thank you, Jeff and Brian. What a great discussion at the last minute here on a Friday. It's been a pleasure getting a chance to speak with you both. And I just love this reporter and tech worker dynamic where we can sort of all, tech executive dynamic, where we could all sort of talk about problems and solutions together. It's what we try to do here on the Big Technology Podcast. Thanks so much for joining us.

• 56:15Speaker 03

  Now, more than ever, we need someone on our side. But Rob Whitman? Whitman's votes drove up costs, cut Medicaid, and risked 350,000 Virginians' health care to pay for tax breaks for himself and his donors. He's not on our side. Shannon Taylor will fight corruption, cut taxes, and bring down costs. This November, vote for Shannon Taylor. She's for us. Paid for by HMP, theHouseMajorityPac.com. Not authorized by any candidate or candidate's committee.

• 56:44Speaker 04

  In this life, you can't win.

• 56:46Speaker 05

  When the world gets you down.

• 56:47Speaker 04

  No!

• 56:48Speaker 05

  Escape to Pluto TV.

• 56:50Speaker 04

  This is objectively fantastic.

• 56:51Speaker 05

  Relax into the movies and shows you already know and love, all for free.

• 56:55Speaker 04

  Schwing!

• 56:56Speaker 05

  Even our news is uplifting.

• 56:57Speaker 01

  You stay classy, San Diego.

• 56:59Speaker 05

  Watch School of Rock, Crazy Ex Girlfriend, Anchorman: The Legend of Ron Burgundy, The X-Files, Wayne's World, and so much more.

• 57:05Speaker 04

  We're not worthy!

• 57:06Speaker 05

  The best things on TV really are free.

• 57:09Speaker 04

  You just have to know where to look.

• 57:10Speaker 05

  Escape to Pluto TV. Stream now, pay never.