Get access to Protocol
Launching on June 23.
Mozilla loves a good open letter. At particularly poignant moments for the tech industry, the company has a longstanding practice of loudly calling on important companies to do the right thing.
Most recently, ahead of the election, Mozilla published an open letter — on its website and as a full-page ad in the Washington Post — saying that Twitter and Facebook need to turn off some of their important recommendations until the election is over. "Facebook and Twitter still include features that could allow disinformation about voting and election results to go viral," said the letter, signed by Mozilla and 6,000 internet users. "This could escalate quickly and threaten the integrity of the U.S. election."
Ashley Boyd, the VP of advocacy at the Mozilla Foundation, spearheaded the effort, and oversees much of Mozilla's work trying to make the internet better, from all those open letters to the Unfck The Internet program. On this week's Source Code Podcast, she talked about why she's pushing Twitter and Facebook, how she picks Mozilla's battles, and the ways a bizarre 2020 could change the internet forever, even after life goes back to normal. If there's even such a thing.
Below are excerpts from the interview, edited for length and clarity.
I want to start with the open letter. I find the process behind that sort of fascinating, because you sit down, you say, "OK, we're about to have this very consequential election, how do we pick a couple of battles? What should it be about?" What is that decision-making process like?
It's great that you recognize that it's about picking battles, because that is really the art, I think, in this work. Making sure that what you're doing is additive, and not disruptive. And I will say that when we were thinking about what to really highlight as it relates to the upcoming election, we were really asking ourselves: What's going to be most impactful? What can be done responsibly in the weeks prior to the election?
I don't think it's helpful to introduce and call for big changes that we know, practically speaking, would take quite a lot of effort to actually implement. So we were looking for big changes that we think were missing ingredients of what the platforms had already instituted. So we've seen the companies make a lot of changes, even up to these last weeks, and they're all incrementally important. But introducing a new idea that's detailed, and working at the edges, didn't feel right to us. So that's why we looked at a big action, that would be addressing the need for scalable solutions. And that would reduce bias and concerns about bias. And that's where we landed on these two asks — around Facebook groups and Twitter trending topics.
And the thing they have in common is that they're about recommendations, right? We get really hung up on specific policy decisions about specific kinds of content. Why do you see recommendations as the biggest version of this thing to talk about?
Well, recommendations are an instrumental way that the companies themselves are unintentionally recommending viral disinformation. So it's the place where they're active in this ecosystem. In the case of Twitter trends, it packages a set of content that then makes it a trend. Maybe it would have been a trend otherwise, but it definitely is the company setting forth this concept that this is A Thing. And we know that as information is coming at us quickly, even these moments where you sort of carve out what is real or what is a trend are very important to the larger conversation and ultimately, the election.
The Facebook group question is similar but slightly different. We've actually seen Facebook take action to stop health group recommendations. They said they did that because they thought it was critical that people get authoritative information about health concerns. And we feel like the election is basically exactly the same. Particularly in this critical time period, pre- and post-election, it's critical that people get authoritative information about the voting process, about the voting outcomes, and where things stand. And so that's a place where we think the intervention by the companies is in line with what they've done before and just truly impactful at this time period.
Does watching them do things like shut down health recommendations kind of make you tear your hair out? I've talked to people who see these small steps on the sides of these organizations and it's like, OK, you know the problem. You know that what you're recommending is a challenge, that not only is it about the actual content, it's about who's seeing it and when and how. You're just seeing it in this sort of very narrow, specific way.
Yes. It's both encouraging and infuriating. It's encouraging to know that they understand how this is working in that context and have taken action, so I take it as an incremental step in the right direction. It also gives me hope that they know structurally, on their end, how to do it quickly. So when we're putting out a call for pausing Facebook recommendations in the U.S. entirely, it gives me confidence that they've internally worked through the system about how to do that well and quickly. But it is frustrating to see a combination of small steps at a time when we think there need to be bold, very big steps taken. So I think it's frustrating from that standpoint.
It's also frustrating because of the lack of transparency about what's working, and what the impact is of these steps. It leaves us in the dark about what to push for more. Maybe it's a small set of things that actually, quite narrowly applied, really have a big impact. But we don't know that because the companies aren't sharing freely and working with researchers to document those successes. Or those dead ends! like those are equally important to understand what works and doesn't work.
As an advocate, I have no interest in proposing and pressing for a solution that's not effective. That's counter to my practice, and what the world needs. But we really are operating in a black box unless we set up a different relationship with the companies around transparency in third party research.
Was there a version of the thinking that you guys were doing that was obviously too big, where you sit down and you say, "Let's make them turn off all the algorithms!" That clearly isn't going to happen. So what's the too-big version of the solution here?
Yeah, I think the too-big version is "turn these off forevermore." I don't know if that's actually a good idea. I think there are good examples where trending topics have been helpful in some contexts. I'd like to work that out, and think about different use cases where it could be successfully shifted: You see Twitter already trying to create more contextualization and trends. I think we were interested in calling for a pause — I use that word on purpose — for a certain set of time, to help them get a better understanding of what might be fruitful and important changes in these features long term. But we weren't ready to say that the company should disable them forevermore.
I think, particularly as concern about the post-election period has grown, you know, one of the things we really want to emphasize is it's not too late to take action. Doing this after the election could also be very helpful. One thing that we talked with Facebook about is this notion of new groups that could spike after the election based on concerns about the results or the process, and their system relies on user reports and AI looking at the content. But if you have a new group that doesn't have a lot of content, it's difficult for their system to flag a potentially problematic group to not recommend. So it's a very specific case to this post-election period, that isn't an ongoing problem but is something that we think is something to pay attention to in this particular case.
How do you balance those things? Because we're in the middle of so many sort of theoretically temporary, but very complicated things, whether it's the election or the pandemic. How much time do you spend thinking about what we do today or this week, and then turn off when things go back to ... whatever normal looks like? Versus what should we be doing now that sticks around forever, even as the world keeps changing?
This is a really good question. In one sense, it's the Before Times thinking that this is just a time period that's particularly problematic. That's false. That's definitely a game that I play with myself to try to feel some control over this time period.
There are so many elections globally next year, so this is all a run up. There's a lot of focus and speculation and attention on the U.S. election, but we have to think about these conversations as being relevant for next year and all the elections to come. And of course on COVID and all the misinformation about COVID, and the disinformation, this does show how perennial the problems are.
So my recommendation and our thinking is, after we get over this particularly challenging and high-focus time around the election, we really need to come back to some fundamentals. Things feel chaotic and a little bit patchwork. And actually our platform policy tracker about election policies by the platforms really shows that there's just a bunch of different approaches. Different looks at how to think about political ads, different looks at how to deal with political figures. It's all a mishmash. It's fine for there to be variance, but it does feel a little chaotic and difficult to get a handle on.
So we're really looking at some fundamentals about limiting the spread of disinformation, what's most effective, providing advertising transparency across the board, not just on political ads. [And] empowering consumers: The companies have put some features in the hands of consumers to report disinformation to better control their privacy, but we have to get more concrete about what that looks like, and looks like done well.
And then back to this issue about supporting third party researchers and looking at what works. We think that combination will really help provide a little bit of stability as we lurch from crisis to crisis. Those are the fundamental tentpoles of what we need to be doing to have a sane and effective approach to this work.
I feel like this is the second or third election in a row when I've thought, "this is going to be the time when the two candidates for president actually talk about tech in a really substantive way." And it just keeps not happening. And you would think it would be this year! There's the antitrust stuff, China stuff is out there, TikTok was the biggest story in the world for a while, and yet it has been kind of absent from the discussions of the politicians who are running for president. Why do you think that is?
This year, I think it's because of such huge and impactful issues around us. Around COVID, particularly, and all of the economic and personal toll there. But, while I don't hold myself personally responsible, I hold us collectively responsible for failing to create really compelling, crisp narratives about why this is important. We are not translating the kind of nerdy technical details and policy details into the sort of top-level storytelling about why people should care.
And ultimately, I think our political leaders are more like the general public. Not because they're not engaged and interested in the issues, but when I think about trying to reach policymakers and really get them excited, and feeling like tech issues are critical to their political leadership, I'm speaking to them more like a regular user. Because it's a complex and difficult issue to break down. So I think that that's a goal of mine, and many other people: to really get to that kind of core narrative-building challenge. We haven't gotten there yet.
All right, so 2024. That's going to be the election where we actually talk about this stuff.
Yeah, that's our goal! We're going to be back here, and we're going to say, "Isn't it amazing to hear political leaders talk with such nuance and passion about all these issues that we care about?"