Brandon Silverman, ex-Crowdtangle/Facebook: “legislation is coming”

photo of Brandon Silverman
Reimagining the Internet
Reimagining the Internet
Brandon Silverman, ex-Crowdtangle/Facebook: "legislation is coming"
/

Brandon Silverman developed a news analytics tool called Crowdtangle that was so good at tracking popular stories on Facebook that the company acquired it. Today on Reimagining the Internet, Brandon tells us how he’s working to regulate his former employer.

Transcript

Ethan Zuckerman:

Hey, everybody. Welcome back to Reimagining The Internet. I remain Ethan Zuckerman. I am here today with my friend, Brandon Silverman. Brandon, at the moment, is Oakland resident, dad, walker of dogs, but in part, we’re having him on because he started CrowdTangle, a really remarkable tool that we was acquired by Facebook in November 2016. He left Facebook last year in 2021.

Brandon’s had a really interesting path in the tech world. He came in through the Center For Progressive Leadership, which is a nonprofit organization aimed at training a new generation of political leaders. His interest in building online communities helped him found a company called OpenPage Labs, which was interested in helping progressive nonprofits use Facebook for activism and organizing. And that tool, in turn, turned into CrowdTangle and I think, arguably, changed Brandon’s life. Brandon, welcome. Nice to have you here.

Brandon Silverman:

Thank you. Super excited to be here.

Ethan Zuckerman:

So Brandon, what is and what was CrowdTangle? Because this is a product that has really had a couple of different phases of its existence. What was CrowdTangle when you initially built it, and what problem were you trying to solve?

Brandon Silverman:

Yeah, great question. Well, I mean, the truth is, like a lot of startups, our original idea was not a great one, or at least not the one we thought, or at least we hoped it was going to be. But the basic idea was I had been a community organizer for the first part of my post-college career, and one of the things my work had led me to do was run these fellowship and leadership training programs. We brought together these really amazing leaders and helped connect them to each other, but also get lots of skills and training and mentorship. And did that for about five or six years before a light bulb went off, and what we realized we were really doing all that work was building community, and it was really the relationships between all the people we were bringing in our programs that was by far the most impactful thing we were doing and that they were leaving with.

And so we began to realize that actually one of the most important things we could do in all of our work was create a space where those communities could live on well past when they went through the various programs we were running. And this was right around the time when the internet was beginning to be a much more obvious place of political organizing. Howard Dean campaign had raised tons of dollars for the first time and really raised everyone’s awareness about the role it could play in campaigns, but nobody had quite figured out how to build a space for people who cared to about out the same issues to come together and stay in touch with each other to communicate, but also to organize. And we ran into that same problem right around the same time, but it seemed like obviously the internet could help fix this in some way.

And so when we started CrowdTangle, the original idea, actually, was could we create some space online where people could come together, stay in touch with each other, share news, organize, and even get into some core organizing stuff around leadership developments and taking actions online, et cetera. And the other really key part of whole story was that for several years, people had been trying to solve that problem, and the very first iteration of that was every nonprofit, at least that I knew, tried to build their own Facebook. And actually, there was only one version that ever really worked, and that was myBO.com. That was a Barack Obama campaign, their first time. It was actually, I think, maybe the only truly scaled, successful bespoke political social network.

Ethan Zuckerman:

Although, they had a certain advantage in having Barack Obama, a very active ground organizing game, and the idea that all the conversations had to happen there. It’s putting your thumb on the scale just a little bit.

Brandon Silverman:

Totally. It turns out, not replicable if you don’t have all those characteristics. So the other thing that happened is the progressive, political, nonprofit world is seeing all these things. They still have a need to organize their members, but are seeing that doing your loan doesn’t work, these out-of-the-box tools don’t work, and right around that time Facebook launched the Open Graph. And so you suddenly now, and by the way, if you’re an organizer, one of the core, first principles organizers is go to where people are. Don’t ask them to come to you.

Ethan Zuckerman:

So people are adopting Facebook in droves. Facebook, at this point, is super interested in having people build apps on top of their platform. Open Graph was a pretty unique moment in Facebook’s development. What was Open Graph and why was it so important for CrowdTangle?

Brandon Silverman:

I mean, if you were a developer, I mean, it was a dream come true. It was suddenly this massive trove of APIs where you could go and take different components of the entire Facebook app and build on top of them. What we started doing was saying, “Hey, we want to go find all of the groups and pages and events that are all connected to some issue,” and let’s say it’s like, at the time, the Occupy movement or Sierra Club or something, we could go and find all of them, put them into our own custom front-end interface, and build functionality and tools and all sort of things on top of them. And so we said, “We’re not going to ask people to do anything new. We’re just going to take all the existing activity on Facebook and allow leaders to organize it based on either an organization or a topic.

In the beginning, the idea is, “We need to make a little money.” So mainly the idea is we would sell this tool to a Sierra Club or an AFL-CIO or whoever. They would then use it to organize all of their Facebook activity in one place. But eventually, we would then pivot to a more consumer facing version where if you wanted to go see everything happening in the labor movement or racial or economic or climate justice, whatever it was, we could start to build these bespoke more issue based versions. But around that time, another thing happened, it was also Facebook’s decision, is it was around 2012, 2013 when basically Facebook made a much more active decision to put third-party content into the newsfeed. And what I mean by third-party content is not just friends and family, but news and celebrities and sports, to put page content into it.

And so what suddenly happened is if you were a publisher around the world, you’re Washington Post or Times, Buzzfeed, you used to get. I don’t know, let’s say you had 10 million monthly visitors. Suddenly, Facebook turned on the fire hose, and you would wake up in June and you would have 30 million monthly visitors that month, and 20 of them would come from Facebook.com. And so the whole publishing world, just suddenly their traffic just started skyrocketing because Facebook just suddenly started to deliver just massive amounts of traffic to the entire publishing world.

And I think for us, one of the very serendipitous things that fell into our lap was when you’re designing for nonprofits and newsrooms, they’re understaffed and they’re insanely busy, and you have to figure out to how to deliver value to them the minute they open up the tool. And if they don’t, they just don’t fucking have time to … Sorry, I don’t mean to curse. They don’t have time to wade through data. And so one of the things that meant, it just really forced us to not be afraid to be opinionated about what we thought would be really valuable.

Ethan Zuckerman:

CrowdTangle is an accidental analytics extension, and it wasn’t about track your brand, it wasn’t about, “Here’s all the possible data within it.” It really was, first and foremost, about this question of what’s doing well on Facebook? What’s getting a lot of attention? What are the big movers and shakers in the environment? Which is something that the other tools don’t necessarily give you access to. So clearly, many people adopted CrowdTangle. We used it quite a bit in our own research, along with the Media Cloud tools. And in many ways, your acquisition was not what happens with most corporate acquisitions, which is something very clever gets shut down and mothballed and never exists again. Instead, you actually had a tool that continued to grow and expand within Facebook.

I think one of the ways many people learned about CrowdTangle was when people started using CrowdTangle, not just to interrogate what news content was doing well, but really to ask the question of, “What were people looking at on Facebook?” And maybe one of the more famous of this is Kevin Roose, who’s a columnist for the New York Times, who has appeared on the show here, and who somewhat legendarily started posting a Twitter feed every day of the top 10 or top 20 stories according to CrowdTangle, which had a very strong right-leaning association to them. So figures like Dan Bongino, who was, I think, invisible to people outside of conservative ecosystems, would consistently show up in the top 10 stories that were coming across on CrowdTangle. What was that experience like? You suddenly became a tool for insight into Facebook, which didn’t necessarily portray Facebook’s community in a particularly flattering light.

Brandon Silverman:

Yeah. So, I mean, there’s a lot of elements to this. I think one thing I always point out the point out, I think, is [inaudible 00:09:41] was like, it’s not flattering depending on your politics and that Dan Bongino is an incredibly popular figure in conservative media. He’s a very popular radio talk show host. He’s a very popular guest on TV, and there’s a very significant portion of this country that looks at that list and goes, “Oh, those are important and reliable media entities.” So I think it’s important, one, just to remember that we’re in a very polarized, divided country.

Ethan Zuckerman:

Right. Right. And the fact that I find Dan Bongino somewhat shocking might actually reflect that Facebook has actually done an amazing job of reflecting the US population as a whole, whereas Twitter has not, necessarily.

Brandon Silverman:

Yeah. So I think one is a really important piece. Is there some way you can look at that list and take the politics out of it? I do think there’s an unflattering version of that list, regardless of politics. I think those are voices that tend to be more sensational, that are definitely polarizing. So I think that’s one piece. I think the second is there are realities about the American media ecosystem that are asymmetric and can result … if you’re attempting to do a top 10 list, could result in list like this.

And some of the ones that I’ve talked about in some other places, but one is, demographically speaking, the right in this country is more similar, is a more homogenous demographic than the left. The right tends to be, at least people who identify as Republican, tend to be predominantly white, predominantly Christian. On the left, it’s a much more heterogeneous coalition-based party and system. And so what that means is you can have entities that speak to a much larger percentage of the right than you can to left.

So for instance, Franklin Graham can be somebody who 80% of Republicans are like, “He’s speaking to me,” versus on the left, it’s like, Rachel Maddow might speak to older TV watchers in metro areas of the democratic party, but maybe not-

Ethan Zuckerman:

But she isn’t necessarily speaking to Alexandria Ocasio-Cortez supporters in the Bronx. Right.

Brandon Silverman:

Or even the Police Union in Philadelphia. It’s democratic.

Ethan Zuckerman:

Right, right, right. Sure.

Brandon Silverman:

It has very different … The natural media ecosystem, I think, the demographics of both parties also make it more likely that one side is more concentrated than the other. Also, on the left, the left reads a lot of mainstream news. So the left is divided both between the overt partisan stuff, and also generally trusting the New York Times and CNN and others, whereas on the right, they don’t trust any of the mainstream. So they’re much more heavily drawn to the much more overt partisan political side.

I also think there’s demographics of the app. I think it’s not a huge shock that I think the blue app itself is getting older and that plays out in it. I also think a really important thing to note is that this is one metric and not a comprehensive look at the entire ecosystem on the platform. And in fact, when I looked at it, it was definitely not a right-wing eco-chamber. It was not as if you are a Democrat and you’re on there, and you’re being flooded with Dan Bongino. And in fact, it’s complex, and what this looks at is news being shared by pages, and it tells you that there is a very active audience and community that likes those pages on the right and engages them on a really regular basis. I also think there’s a degree to which if you are specifically trying to troll the other side, you’re going to get the other side to respond sometimes, too.

Ethan Zuckerman:

So the reason Kevin Roose decided to do it, I think was legitimate, which is that there have been many complaints on the political right that Facebook is biased against right-wing voices, that it was taking down right-wing speech. This is a pretty common complaint about Facebook, no matter what your politics are, and you don’t even have to be in the US. You can find Israelis saying that the platform is biased against Israelis, you’ll find Palestinians saying it’s biased against Palestinians.

One of the big things that I think many of us would like to research about the platform is those take-downs. Who is actually getting taken down? But in some ways, CrowdTangle found itself in the middle of this political fight, which was this was a very effective way for Roose, who leans perhaps to the left, and is certainly with the New York Times, which would be seen as leaning to the left in this, to say, “What are you guys complaining about? Actually, there are these right wing figures who are the most viral sensations on Facebook. How can you possibly complain about a tilted playing field?”

Now that may not be accurate for all the reasons you just said. If you have a more homogenous population on the right, if you have an older, whiter user base, whereas you have a younger blacker user base on Twitter, so on and so forth, all of those things could be explaining why those voices are out there. So Roose ends up making the case that perhaps this is why Facebook ends up disinvesting, constraining the team associated with CrowdTangle. And my goal here is not to get you to talk about the inside of what happened at your end of the time with Facebook. But I am curious whether you felt like there was pressure around Facebook being concerned about CrowdTangle providing a window on these very large political questions about how the platform was being used. Was that something that you felt pressure around?

Brandon Silverman:

Listen, I think there’s a real debate happening inside Silicon Valley, and especially inside social media platforms, around transparency. And they all want to be transparent, but I sometimes compare it to affordable housing. It’s like everybody’s for it until it shows up in your backyard and it gets real. And then a lot of opinions can change. But when it comes to certain types of transparency, there’s a real divide, I think, inside Silicon Valley, but also inside these companies. And I think one of the best ways you can look at it is that you have companies like Facebook that there are ways in which they have leaned into transparency more aggressively than other platforms.

There are some companies that have done almost nothing. The companies that have done almost nothing have avoided a ton of the blow-back, and it was absolutely one of the realities of our experience over the final year or so, that there were questions on, “Is it worth it?” And that there’s also another dynamic of that is that you have … There’s definitely a point of view inside these companies, too, that’s like, “Because we don’t have to worry about issues around what it means to make potentially privacy sensitive data available to groups that are out outside and yada, yada, we have a better ability to interrogate and interpret and analyze all of this than anybody on the outside ever could.” And so any of the voices and criticism on the outside is just a distraction from us doing what we think is the most important and relevant work.

So where I have come down on this is that based on the experience we had at CrowdTangle, I came to just really believe deeply that there are gaps in this work that can only be filled by outside experts and organizations and civil society, that until there is some sort of new model in which platforms have content moderator and policy experts in every single community around the world, you are going to run into challenges around language and dialects and local nuance and staying up to date on what the political situation is, that all affect what it means to have and manage an information ecosystem somewhere. And that I don’t see that ever happening.

And so in the meantime, what I think the alternative is, is finding ways to engage and collaborate and work with as many communities as you can around the world, to have them be a part of the process. That is almost a governance question. And in order to do that, you have to be willing to share data. And then also, by the way, not only is it a better solution to the scale of all these platforms, but it just over the long term helps you build a better product more actively than if you do it by yourself.

Ethan Zuckerman:

So we introduced you tongue firmly in cheek as an Oakland dad and dog walker, in part, because you’re a semi-retired guy at the moment. But in truth, you’re enormously active around these questions about platform transparency. You’ve been working fairly hard on a piece of legislation called PATA. What is PATA? What’s the status of it, and what’s the key idea you’re trying to get into this bill?

Brandon Silverman:

Yeah. And for what it’s worth, I’ve actually spent more time in Europe over the last several weeks because of a piece of legislation coming out from there. But PATA is a bill, it was actually how I got started in this work. I think it was a couple weeks after I had left Facebook that I got a ping from some senate offices saying they were basically interested in trying to find ways to make a version of the CrowdTangle model, something would be required of all major platforms.

What that bill is, is basically, in some ways, a fairly sweeping piece of legislation that basically takes I think almost every idea around transparency, and specifically data sharing transparency, and puts them into one piece of legislation. So it includes everything from Nate Persily, who used to be a member of Social Science One, but basically put out this piece of legislation around super privacy sensitive datasets and how to provide access to them, do clean rooms. So it includes a mechanism for how to make very sensitive datasets accessible to outside researchers. There’s a second piece around Safe Harbor and scraping that is a really critical component and important.

Ethan Zuckerman:

That I’ve been fairly involved with advocating for and arguing for, and we might talk about that at the end of this. But yes.

Brandon Silverman:

Totally. I mean, and your both work and also your direct feedback to me helped me shape my understanding of it as well. And then, third, basically it you want to go look at the actual piece of legislation, there’s something called Section 12, which has a bunch of, essentially, rule making authorities granted to a new agency that was sit under the FCC that would do everything from mandate specific types of report and aggregated statistics that would have to be reported, provide more transparency around ads, would have a CrowdTangle-esque solution where particularly meaningful and large accounts would have to be much more transparent and much more filterable and searchable and have APIs and interfaces so that some version of real-time data would be available for, particularly, public newsworthy and public interest content accounts. And then there’s some algorithmic transparency as well.

So basically there’s a bunch of pieces in there, and it came out of the gate with bipartisan support. But I think the big question is how much bipartisan support can it get, including how many more Republicans are willing to sign up to support it and sponsor it? And then there’s also just a timing issue. With the Supreme Court nomination, with Ukraine, we are running out of time. I’m not a political expert, but I think there’s a sense that some of these issues are not going to be as doable if there’s a turnover on parties in the House or the Senate.

Ethan Zuckerman:

I think it’s fair to call it a piece of model legislation in the sense that it’s got everything that people like you and me, people who are working on platform transparency, have put up as ideas to try to get to stick to the wall. So my big idea for years on this is we have the right to scrape datasets and create indices of them. It’s how we get this thing called Google. You don’t want to give up the right for it. And in fact, for research uses, for journalistic uses, you probably want to put it under a Safe Harbor and say very explicitly, “These uses are exempt from CFAA and other relevant legislation.”

I think you’re absolutely right that having the ability to see the top thousand, the top 10,000 most popular things that are reaching hundreds of thousands, maybe millions of people, is utterly essential for actually understanding how these platforms are informing us. And I think this idea that it shouldn’t just be Facebook, which in some ways has taken extra hits because it did have a good window carved into it, and Twitter, which has always had a good window carved into it because of the API, compared to YouTube, which is very, very hard to study at scale because it’s video content rather than text content, or TikTok, which is just a completely sealed chamber, it would be really fascinating to have that coming out on a regular basis.

Do you think that’s possible? And what do you think we’d learn from it? If this passes, either under the current legislative session or if that chunk of the model legislation survives, what do you think we would learn from having that insight, not just into Facebook, but into platforms as a whole?

Brandon Silverman:

Yeah. So I think in some ways the experiences we had over the last two or three years, I think give us some … At least when I say we at CrowdTangle, I think for me, gave me some insight to what you could get out of this work, especially if it was available to platforms and even more robust and comprehensive level than what we’re doing. But I think a couple things. So I think one is during major world events, during crises, wars, humanitarian crises, natural disasters, there’s an enormous amount of incredibly important information that is being disseminated through social media platforms. I mean, you watched this during Ukraine. The whole world was looking to social media to get a sense of what was happening. And lawmakers and world leaders and governments were making decisions based on what they were seeing. Public opinion was being shaped.

But a lot of these platforms are just black boxes. And so was the viral video about The Ghost of Kyiv, was that a real video? Was it not? Was it actually the biggest video of the day, or was there other videos way bigger? We’ve seen anecdotal evidence of disinformation campaigns on TikTok that were parroting Russian propaganda. How many accounts actually said the same thing? Who was the first one who did it? Are they still doing it? So there’s all these ways in which our information ecosystems are sitting behind these black boxes, and the world, I think, needs to have a better window into seeing what’s happening, because it is shaping how the world responds to these incredible events.

And we saw micro versions of that in other places where there was a typhoon that hit the Philippines in 2016, and one of the biggest news outlets in the country had CrowdTangle dashboards that they were putting on broadcast television, the whole country, every night, that were showing where the disaster help was, what the latest news was from some of the relief efforts and all these different things. So I just think that at some level these platforms have turned into the public town squares. Instead, we can’t actually see what’s happening. And so I think … yeah.

Ethan Zuckerman:

Right. I mean, the word that I end up using is the digital public sphere.

Brandon Silverman:

Yeah, exactly.

Ethan Zuckerman:

Stolen from [inaudible 00:25:43] and people before that. And of course, that’s the interesting piece of it. In an actual town square, you can get a sense for what people are actually talking about. We now have these town squares that have very little accountability. They are, if not entirely opaque, the windows carved into them have very particular distortions to them. And I think you did a great job explaining the ways in which the CrowdTangle window was and wasn’t an accurate picture of what’s going on within Facebook.

The response, in some ways, has to be, “Let’s carve a whole bunch more windows.” It is, I think, also simultaneously worth acknowledging that Facebook is not, in fact, the worst in all of this, that pretty much all of these platforms don’t have a way of looking at what’s going on inside of them and are reluctant to give you that way of looking at what’s going on inside of them. There’s also a certain legitimacy to that. When you have that list of, “Here are the 200 videos doing the best on TikTok today,” that’s essentially a way of reverse engineering the algorithm and trying to figure out, “What is it that I need to do to make my way to the top of TikTok?” which is not just effective for activists. It’s also effective for spammers and all sorts of other bad actors.

As someone who’s been in one of these organizations, what would you tell … I assume Zuckerberg has heard whatever he’s going to hear from you, but if someone else were to call you up, if TikTok were to call you up and say, “How should we get out in front of this? How should we be proactive about this? What should we try to do?” Or is that the wrong answer, and the answer is, “We should be trying to do it through legislation instead”?

Brandon Silverman:

Yeah. I mean, I think part of my answer is it doesn’t matter what you do or don’t want to do. Some of this is going to happen, and legislation is coming. It might not all pass, but enough trust has been lost in some of this among lawmakers that some of this is going to pass. And so it’s coming, period. Second, I think one of the questions around this stuff is just a short term versus long term, that you might take short term hits in some of this stuff. But in the end, transparency helps you build a better product and serve your users better, and that it’s a pain in the ass, but when the outside world is holding you accountable, it makes your product better and it makes you serve your users better.

But then also I think that there’s a degree to which I think at least some of these platforms never meant to be the public to house squares and never wanted the responsibility that came with it, and that one path forward for them is to find ways to collaborate more with the rest of civil society and create mechanisms, to create society-wide debates and discussions around what our information ecosystems should look like. And if it’s just going to be them deciding, they are appropriately always going to be held to an enormously high standard, and when they screw something up, they’re powerful enough that those screw-ups matter. And so creating a governance model that involves more of civil society around this stuff I think is going to help build more trust and build more accountability and get away from this universe in which social media is to blame for everything.

And then also, maybe, I could make a bunch of arguments. I think the other version of this is I do think one of the discussions around, at least the platforms that have chosen to host news, is that there’s definitely a libertarian element to some of the ethos around this stuff that they believe in free speech and they also believe the idea of a marketplace of ideas. But you can’t have a marketplace of ideas if people can’t see what the ideas are.

Ethan Zuckerman:

Right. It’s not a free market at that point. Yeah.

Brandon Silverman:

No. So also, even just in terms of your first principles and philosophical processes and stuff, make it an open, transparent ecosystem, in part, just to build the ecosystem you want, which is a world in which people are negotiating and debating and confronting ideas in order to get to a better understanding of the human condition over time.

Ethan Zuckerman:

And just one other thing for my edification, what’s the European law that you’re looking at?

Brandon Silverman:

So there’s a bill called the Digital Services Act, and it has a bunch of different components inside of it that touch upon data sharing and transparency, and it is going to pass. Unlike some of the US stuff, the DSA is passing, period. And it’s really a question of just how far the data sharing requirements go. And in some cases there have been some really promising stuff. So for instance, it is most likely going to end up with the budget funding and the commitment from all the signatories between the platforms to abide by an independent agency that will vet a lot of this stuff, and there are also discussions about how to make that agency a Trans-Atlantic agency, that the US also allows to be the primary vetting body.

And then there’s a bunch of discussions, there’s some last minute haggling around trying to get some language in there around public data. So there’ll be a piece for me going up in Politico, I think today or tomorrow, making this case. And there’s what’s called a political trilog when the EU, the European Commission, and French leadership all get together. And so the goal is to get this in front of them before some of the final language is locked in. But there’s a lot of proxy stuff in there.

Ethan Zuckerman:

Who’s drafting? Who are the main drafters of the DSA?

Brandon Silverman:

It’s a wide swath of what is called civil society organizations. So much of it is very hush-hush, behind the scenes. But I think it’s safe to say there’s several hundred civil society organizations, as well as the European Commission, is the regulatory body primarily overseeing it. There are other agencies like EDMO and ERGA and other things that are helping as well. The platforms are all involved. So it’s a combination of different entities inside those.

Ethan Zuckerman:

Got it.

Brandon Silverman:

They do they drafting. They then take the versions, get feedback. It’s insanely bureaucratic, and I probably only even understand five percent of it.

Ethan Zuckerman:

No, no, no. You understand more than I do, but that’s helpful.

Brandon Silverman:

Yeah.

Ethan Zuckerman:

Well, Brandon Silverman, it’s just been a pleasure having you on here. What an incredible journey from where you started with this project to, now, the work that you’re trying to do to handle some of these issues of transparency on the legislative side in the US, in Europe. Thank you for all you’re doing and thanks for being with us today.

Brandon Silverman:

Thank you so much. It was a pleasure and your writing and thinking have influenced me a lot over the years. So I’m just honored to be here, too.