:format(webp)/cdn.vox-cdn.com/uploads/chorus_image/image/63986374/REC_ASA_CODE19_20190610_153642_9818.0.jpg)
“This year, this week, it was unfortunate,” YouTube CEO Susan Wojcicki says. “We managed to upset everybody and we’re working really hard to try to do the right thing.”
Speaking with Recode’s Peter Kafka at the 2019 Code Conference, Wojcicki looked back on YouTube’s messy resolution to a dispute between conservative commentator Steven Crowder and Vox journalist Carlos Maza, saying YouTube made the right decision but was unintentionally “hurtful to the LGBTQ community.” More generally, she said it’s better for the platform to continue reviewing potentially controversial videos after they’re published — not before.
You can watch the interview below (on YouTube, naturally), or listen to it on our podcast Recode Media — which is available on Apple Podcasts, Spotify, Google Podcasts, Pocket Casts, and Overcast. But if you’re short on time, scroll down to find a full, lightly edited transcript of Peter’s conversation with Susan.
Peter Kafka: I’m not going to do a long wind-up here, because I have a lot of questions for my next guest. I’m delighted she’s here. Please welcome Susan Wojcicki, CEO of YouTube. Hi, Susan. They gave you a good hip-hop theme for your way in.
Susan Wojcicki: Thank you.
Thank you for coming.
Sure. Thank you for having me.
I’m really glad we get to have this conversation. I’m glad we get to do it in public, on a stage, on the record. That’s great. Let’s start here. There was a bunch of news last week. Some of it involved you. Some of it involved vox.com, where I work. There was a policy change. I think they all sort of happened at the same time. Can we just walk through what happened, and if they’re parallel tracks, or if they were connected?
Sure. So, first of all, thank you. A lot of things happened last week, and it’s great to be here and talk about what happened. But I do want to start, because I know that the decision that we made was very hurtful to the LGBTQ community, and that was not our intention at all.
Should we just set context, for anyone who was not following this? What decision this was?
Yeah. So, let me ... I’ll go into that. But I thought it was really important to be upfront about that, and to say that was not our intention, and we were really sorry about that. But, I do want to explain why we made the decision that we did, as well as give information about the other launch that we had going on.
Really, there were two different things that happened at the same time. The first one I’ll talk [about] is, we made a really significant change involving hate speech. This is something we had been working on for months, and we launched it on Wednesday of last week.
And this is a series of policy changes you’ve been rolling out for years now.
So, just to be clear ... Yeah. So, we’ve been making lots of different policy changes on YouTube. We have made about 30 changes in the last 12 months, and this past week, we made a change in how we handle hate speech. That took months and months of work, and hundreds of people we had working on that. That was a very significant launch, and a really important one.
What we did with that launch is we made a couple big changes. One of them was to make it so that if there’s a video that alleges that some race or religion or gender or group, protected group, is superior in some way, and uses that to justify discrimination or exclusion, that would now no longer be allowed on our platform. Similarly, if you had a religion or race, and they alleged that inferiority, that another group was inferior, and they used that to justify discrimination in one way. Those were changes that we made. So, examples would be like, “Race X is superior to Y, and therefore Y should be segregated.”
Is it weird to you that you had to make a rule that said, “This shouldn’t be allowed”? That this wasn’t covered either by an existing rule? That you had to tell your community, “Look. This is not acceptable”?
Well, actually, a lot of this ... We’re a global company, of course. And so, if you look at European law, there are a number of countries that have a really strong hate speech law. And so, a lot of this content had never been allowed in those countries, but had actually been allowed in the US and many other countries. And so what we had actually done with it a few years ago is we had actually had limited features, meaning that it wasn’t in the recommendations. It wasn’t monetized. It had an interstitial in front of it to say that this was content that we found offensive. And when we did that, we actually reduced the views to it by 80 percent. So, we found that it was effective, but we really wanted to take this additional step, and we made this step on Wednesday.
We also added, which is really important, a few other definitions to protected groups. So, we added caste, because YouTube has become so significant in India. Then, we also added victims of verified violent events. So, like saying the Holocaust didn’t happen, or Sandy Hook didn’t happen, also became violations of our policies. And so, this was happening on Wednesday, and we launched it on Wednesday.
There were thousands of sites that were affected. And again, this is something that we had been working on ...
This was coming already.
It was coming already. We had started briefing reporters about it in Europe over the weekend, because they’re ahead. You know, the train had left the station. And then at the same ... On Friday, there was a video. We heard the allegations from Mr. Carlos Maza, who uploaded a video on Twitter with a compilation ...
Works at vox.com.
He works at vox.com, yes. With a compilation of different video pieces from Steven Crowder’s channel, putting them together, right? And asked us to take action. Each of these videos had harassment ...
Saying, “He’s directing slurs at me, and the people who follow him are attacking me outside of YouTube, as well.”
Yes. So, he alleged that there was harassment associated with this, and we took a look at this. You know, we tweeted back and we said, “We are looking at it.” You know, Steven Crowder has a lot of videos, so it took some time for us to look at that and to really understand what happened, and where these different snippets had come from and see them in the context of the video.
Actually, one of the things I’ve learned, whenever people say, “There’s this video and it’s violative. Take it down or keep it up,” you have to actually see the video, because context really, really matters. And so, we looked through a large number of these videos, and in the end we decided that it was not violative of our policies for harassment.
So, were you looking at this yourself, personally? Vox is a relatively big site. It’s a big creator. Were you involved in this directly?
I mean, I am involved whenever we make a really important decision, because I want to be looking at it.
So, you were looking at the videos.
Well, so we have many, many different reviewers.
Mm-hmm.
They will do a review. Again, there are lots of different videos produced by Steven Crowder. He’s been a longtime YouTuber.
But in this case, did you weigh in personally? Did you look at the stuff?
I mean, yes. I do look at the videos, and I do look at the reports and the analysis. Again, I want to say there were many videos, and I looked certainly at the compilation video. It was ...
So, when the team said, “We believe this is non violative. This doesn’t violate our rules,” you agreed with that?
Well, let me explain to you why.
Mm-hmm.
Why we said that.
But you agreed?
I agreed that that was the right decision, and let me explain to you why I agreed that was the right decision. Okay? So, you know, when we got ... First of all, when we look at harassment and we think about harassment, there are a number of things that we look at. First of all, we look at the context. Of, you know, “Was this video dedicated to harassment, or was it a one-hour political video that had, say, a racial slur in it?” Those are very different kinds of videos. One that’s dedicated to harassment, and one that’s an hour-long ... So, we certainly looked at the context, and that’s really important.
We also look and see, is this a public figure? And then the third thing that we look at is, you know, is it malicious? Right? So, is it malicious with the intent to harass? And for right or for wrong right now, malicious is a high bar for us. So the challenge is, like when we get an allegation like this, and we take it incredibly seriously, and I can tell you lots of people looked at it and weighed in.
We need to enforce those policies consistently. Because if we were not to enforce it consistently, what would happen is there would be literally millions of other people saying, “Well, what about this video? What about this video? What about this video? And why aren’t all of these videos coming down?” And if you look at the content on the internet, and you look at rap songs, you look at late-night talk shows, you look at a lot of humor, you can find a lot of racial slurs that are in there, or sexist comments. And if we were to take down every single one, that would be a very significant ...
So, to stipulate that you take it seriously. I want to come back to the idea that there’s a ton of this stuff here.
Well, so what we did commit to — and really, this is I think really important — is we committed, like, “We will take a look at this, and we will work to change the policies here.” We want to be able to ... When we change a policy, we don’t want to be knee jerk. We don’t want it to be like, “Hey, I don’t like this video,” or, “This video is offensive. Take it down.” We need to have consistent policies. They need to be enforced in a consistent way. We have thousands of reviewers across the globe. We need to make sure that we’re providing consistency.
So, your team spends a bunch of time working on it. They come to you at some point and they say, “We don’t think this is violative.” You say, “We agree.” You announce that. And then a day later you say, “Actually, we do have problems with this.”
Well, so what ... Okay. So, we did announce it, and when we announced it, if you look carefully at the tweet, what we actually said at the end is, “We’re looking at other avenues.”
Mm-hmm.
That’s because we actually have two separate processes. One of which is like, “Is this content violative,” from just the purely community guidelines. But then we also have monetization guidelines, and that’s because we have a higher standard for monetization. We’re doing business with this partner. Our advertisers also have a certain expectation of what type of content they are running on.
And so, we had the first review. We said, “It doesn’t violate the community guidelines on harassment, but we’ll take a look at our harassment guidelines and commit to updating that.” Which actually had been on our plan anyway. I had actually put that in my creator letter that I had just done a few weeks ago, saying we were going to take a hard look at it. But we had been working so hard on the hate speech, and so our teams were caught up on that. But that really had been next on our list.
So, we have a higher standard for monetization, so then we did announce the monetization change. That Steven Crowder was, his monetization was suspended.
So, was that in reaction to people reacting to you not reacting?
No.
Or was that something that you were already planning to do and just hadn’t gotten around to announcing?
No. We were in the process of looking at that, and there were ... When we look at these accounts, there are many different components that we look at, and that’s actually why we put the line, “There are other avenues that we’re still looking at.” And that might have been too subtle. If I were to do it again, I would put it all into one ...
Do it in one go.
Yeah, I would do it all in one go. But we were also ...
So you said, “We’re not kicking you off, but we’re not going to help you make money on YouTube. It’ll be directly through ads.”
We’re suspending monetization.
Meaning, “We’re not going to run ads against your stuff. If you still want to sell racist coffee mugs or whatever you’re selling, that’s your business, but we’re not going to help you. We’re not going to put an ad in front of your stuff.”
Well, we said we’re not going to put an ad in front of it, but the conditions by which we will turn it on can be broader than just that. So, for example, if they’re selling merchandise and linking off of YouTube, and that is seen as racist or causing other problems, that’s something that we will discuss with the creator.
So, one more question specific to this.
Because again, we’re putting advertising there, so we need to make sure that the advertisers are going to be okay with it, and we have a higher standard. And so, we can sort of look at all different parts of that creator and what they’re doing, and basically apply that higher standard there.
So, people I work with at Vox and other people are saying the one problem we’ve got with all this, in addition to what seems like a back and forth, is that we don’t understand why you made the decision you made. There’s not enough transparency. We can’t figure out what rules he did or didn’t break.
And also, by the way, it seems clear that he did break these rules. But they’re asking for transparency, they’re asking for more understanding of what went on here in this specific case. Is that something that’s reasonable for someone to expect out of you and out of YouTube? To say, “Here’s exactly what happened. Here’s exactly what broke the rule for us. Here’s exactly why we’re demonetizing it”?
Which case are you talking about?
Well, in the case of the Crowder/Maza stuff. But for anything, right?
So, we tried to be really transparent. We communicated numerous times, including publishing a blog explaining some of the rationale for our decision. We try to be really transparent with our community, with our guidelines. We get that request actually a lot from our creators, because they want to know what’s allowed on our platform, what’s not allowed, from a monetization standpoint. And so, we do get the request for transparency, and we are working to continue to be more transparent and explain why something is a violation of our policies or not.
So, you were talking earlier. You said, “We have to take this very seriously, because if we make a ruling here, someone else is going to say, ‘Look at this rap video where they use this slang term.’”
Yes.
This, to me, seems like the actual issue you’ve got across YouTube, which is you’re doing this at scale, 2 billion users. What, it’s 500 hours of content uploaded every minute? It seems like no matter what decision you make at any particular case, someone is always going to come up and say, “What about this?” Or they’re going to say, “If this is the new rule, I’m going to figure out a way to skirt around it.” Or someone’s going to say, “By the way, you’re going to see content that you have never contemplated showing up.”
It seems like at the scale you’re working, on an open platform where anyone can put anything up, that you guys are always going to be on this treadmill, and no matter how many humans you throw at it and how much AI you train up, you can’t actually solve this problem. Do you have confidence that this is something you can actually get a handle on?
We can definitely do, and continue to improve, how we manage the platform. I see how much improvement we’ve already made. For example, if you just look a few years ago, two years ago, there were a lot of articles, a lot of concerns about how we handle violent extremism. If you talk to people today who are experts in this field, you can see that we’ve made tremendous progress. At the end of the day we’re an information company. We have the access to Google, some of the algorithms there. We have the resources to deploy. We’ve committed to, last year, we committed to having over 10,000 people who are working on controversial content, so I see how much progress that we have already made.
Like I mentioned, we’ve made all these different changes to our policy. We actually have just made changes to our recommendation algorithms as well for not violative content, but borderline content. We announced that we’ve seen a 50 percent reduction in the views coming from recommendations from that. If you combine the combination of much better policies, tighter policies, and we consult with many third parties who try to make sure that we get them right, and we’re hearing from all parties, you combine that with technology to be able to do that at scale, I think you can be in a much better place. I’m not ...
So you can get better, but can you get it to the point where we’re not seeing a story about something awful happening within YouTube on a sort of weekly basis? Can you get it where this thing is an exception to the rule?
I mean, I think there’s always going to be ... At the scale that we’re at, there are always going to be people who want to write stories, but ...
Well, there’s also people who want to put terrible things on your website. Right? You guys talk about the fact that you took down 8 million terrible pieces of content in the last quarter.
Mm-hmm.
Right? And that you’re proud of that because you were able to ... What was 75 percent of it, no human ever saw. If I ran a business where people dumping that much sludge onto my property on a quarterly basis, I would really rethink what I’m doing. It seems like, I mean, I just can’t fathom why there’s 8 million pieces of terrible things coming onto your site on a quarterly basis, but that would really upset me and worry me.
Well, it matters what’s the denominator. You gave the numerator. Right? We have a large denominator, meaning we have lots of content who’s uploaded, and lots of users, and lots of really good content. When we look at it, what all the news and the concerns and the stories have been about this fractional 1 percent. If you talk about the other 99-point-whatever that number is, that’s all really valuable content of people who are sharing valuable points of view that we haven’t heard about, educational content, addressing really important issues. I think it’s important to remember that and put that in perspective.
I say that not because we are not committed to solving the fractional 1 percent. We are very committed, and I’ve been really clear that that responsibility is my No. 1 priority. There is a lot of work for us to do. I acknowledge that, but I also know that we have tremendous tools at our fingertips that we can continue to invest in to do a better job. So, yes, while there may be something that slips through or some issue, we’re really working hard to address this, and I think we have some good tools to do so.
What if YouTube wasn’t open? What if I couldn’t upload a video that I wanted to without asking you for permission? What would that do to YouTube? What if you had some sort of barrier to entry that required me to get some kind of permission to upload something before it went up there?
I think we would lose a lot of voices. I don’t think that’s the right answer because we would lose a lot of voices and a lot of people who share content. Sometimes we hear from creators that they started sharing content like they started doing... I don’t know, you look at Sal Kahn. Right? He got started almost by accident, right, who now creates very valuable educational content. But what I think is the right answer that you’re alluding to but not quite ... I’m going to ...
I haven’t got there yet.
I’m going to improve a little bit your suggestion here, which is having more trusted tiers. In a sense, we’ve already started doing that with monetization, saying, “Look, you can’t just come onto the platform and have monetization on day one.” You have to have a certain number of views. You have to have a certain number of hours.
You have to earn your way into getting ads.
You have to be in good standing. That’s an example of where we have more of a trusted relationship with them. We did a similar thing with livestreams in terms of certain number of views that you have to have ... Sorry, subscribers that you need to have. I think this idea of not everything is automatically given to you on day one, that it’s more of a ... We have trusted tiers, and ...
But there’s still, at bottom, you can put stuff on YouTube without asking for it.
Yeah. Everyone can start.
Yep.
Everyone can start. Everyone can be their own media provider, but in order to get some of the broader distribution or to have monetization, it’s something that you work your way up to.
Is this insistence that the platform has to be open, I can think of a business reason for that. I can think of legal reasons for that. Right? Section 230 gives you legal protections, but you have to not actually prevent someone from uploading something. I can think of ideological reasons why you want the platform to be open. Which one is most important to you?
Sorry, I’m trying to remember them all. You said business reasons?
Business, legal, ideology, but you can add as ... I mean ...
I mean, they all ...
I get that more voices is better, but if having unlimited voices has a problem that is always going to trouble you ...
Look, it’s a core issue. If you want to limit it and say, “Hey, we’re only going to have a select set of people,” but what are the factors that you’re determining that? How are you deciding who is getting to be on the platform and have speech, and who’s not?
But you are deciding, fundamentally, right? You have teams, and you have software, and you weigh in. You are fundamentally making the decision at some point.
We do, but that’s after they’ve been on the platform, and we have an understanding of what they’re doing. It’s based on the content that people have uploaded. Look, we see all these benefits of openness, but we also see that that needs to be married with responsibility. You do need to have more of this responsibility in place, and you need the different ways to understand what is the content that you should be putting in recommendations, what is the content that should be promoted for different use cases. YouTube deals with a lot of different cases. We deal with entertainment, music. Why should we say some people can, new musicians or old musicians, why would we want to close something like that?
But there is a world, right, where people make those decisions all the time. Traditional media does this. The New York Times or NBC or Vox knows what’s on their website because they said, “Yes, let’s publish this. Let’s distribute this.” You guys don’t know. That would, again, make me just fundamentally nervous to be running a platform with 2 billion people on it and I don’t know what’s happening on it.
Well, we have a lot of tools. We work hard to understand what is happening on it and really work hard to enforce the work that we’re doing. I think that if you look across the work, I think you can see that we’ve made tremendous progress in a number of these areas. Again, I’m not saying we’re done, but I think you can see that we’ve made progress. If you were to fast-forward a couple years and say, “Well, what would that look like in 12 months and then another 12 months? What are all the different tools that have been built in place?” I think you’ll see that there’ll be a lot of progress from the perspective.
Your boss, Sundar, described the problem you guys are dealing with in an interview this week on Axios as a search problem. Is that the correct analogy to think about the problems you guys are facing?
Well, I think what he ... I don’t want to interpret, necessarily, what he said, but if you look at Google, Google has done ... Google was first founded on delivering information. Right? That’s been the mission of the company, to organize the world’s information. You look at our first patent, it was about page rank. Page rank is actually about figuring out who is authoritative on the web. Who are the people that you should bring up for any different query? Being part of Google, for informational queries, is a really, really important part of how we’ll build some of those systems going forward.
Again, when you talk about music or you talk about cooking or you talk about DIY, these other categories, you want to discover new people. You want to discover those fresh ideas, and you want to see a variety of content, but when you’re dealing with really sensitive areas like news or medical, you want to get those from authoritative sources. I think there’s an opportunity ... That’s where we work with Google, and we learn a lot from Google, and we think we can do a better job that way.
There’s a lot of discussion about regulation and what kind of regulation makes sense to use with tech companies, including yours. What regulation would be helpful to you, and what regulation would you not want to see at all?
Well, I agree there’s definitely more regulation in store for us. One of the groups I’ve been working, that we’ve been spending a lot of time, has been Article 13, which is the copyright directive in Europe, now named Article 17. I think what we have seen is that a lot of times regulation is really well-intended. People want to change society for the better, but it has all these unintended consequences.
What I would say is really important is for us to be able to work closely with these different providers, the different governments, and be able to explain how can we really implement it in a reasonable way, how to make sure there aren’t unintended consequences that they didn’t think about that actually might make the situation worse that they just, because they’re not running our businesses, don’t necessarily have that visibility.
There’s a push to break up some of these companies. What would happen if you were split off from Google?
I don’t know.
That’s a fair ...
I’ve been really busy this week working with all these other concerns. I mean, I don’t know. I’ll worry about it ... I don’t know. I mean, we would figure it out.
That’s a good answer. One more specific video question.
Yeah.
BuzzFeed wrote about a creator with some popularity who’d made a series of problematic videos — I hate that I just used the word problematic — offensive videos. One of them was a death threat aimed at you. You took down that video and another video. Her channel is still up, which brings me back to this sort of whack-a-mole where you guys are sort of sifting through videos as they come to your attention and saying, “This one’s a problem. This one is okay. This one threatened to kill me, but she can still stay on the site.” How do you feel about having — so her name is Soph — how do you feel about having her on YouTube?
Well, like I mentioned beforehand, every single time there is a controversial video, you really need to see the video, and you need to understand what’s happening. She has a large amount of political satire and political speech that’s in her channel. The video where she made the threat to me was struck, and we removed monetization from her channel because, like I said, we have a higher bar for monetization. That was an issue.
Look, I think there’s certainly different levels of threats in the videos and different ways that we would interpret it depending upon the context of who is the person and what they’re saying. In many cases, we would terminate the channel if we thought it was a channel that was dedicated to a number of, I don’t know, hateful, harassment, etc. But in this case, it was a large amount of political speech. Seeing it in context, we struck the video, removed monetization, but left the rest of the channel up.
There’s been a lot of threats directed at you and other YouTube executives in the last few days since you announced your policy. I assume some of that, again, is ideologic, but some of it is that you’ve got these creators who are trying to make money on your platform. I think that’s a difference between what you do and what Google and Twitter do. Generally, most people aren’t making a living on Google or Twitter, and people can legitimately do that. Then you had a YouTuber who was angry about being demonetized, I think, and came onto your campus and killed someone a couple years ago. What’s that like to manage that teeming group of people that you don’t really actually manage?
Well, first of all it’s difficult, but I will say there was something in your question that I don’t totally agree with. I think it’s really important to clarify. You talked about, you made this association of creators who could be upset, or angry, or difficult and then you alleged, right, that it ties to monetization. Again, I want to ...
They’ll often make that argument.
Well, I want to just be clear that, again, it’s this 99 fractional 1 percent problem, but I also want to point out that because we have a higher bar for monetization that we’re really ... This isn’t a business problem from that perspective. Right? We’re focused on having high-quality content available as part of our ecosystem, but we also want to have a broad range because we wanted to keep it open and enable lots of different points of view to be on the platform.
It’s certainly, I think like any time that you have a bunch of creators or people are upset, it’s difficult. Certainly, this year, this week it was unfortunate. We managed to upset everybody and we’re working really hard to try to do the right thing. It was really in our effort to be consistent that we wound up upsetting people.
It’s not an easy job. It’s a tough job, but I am encouraged by the fact that I hear so many good stories of people who have been able to pursue their passion, start a career, have a business on YouTube, talk about how they learned an instrument or learned something new they never thought they could do beforehand. And so it’s really all the good, it’s the 99 percent of good, valuable content that I’m really encouraged, that keeps me motivated and passionate about what I do.
I have more questions but I want to open it up to the audience because I think they’re going to have questions, too. I assume there’s a microphone, maybe two microphones, lights? Can you introduce yourself?
Steve Kovach: Yeah, I’ve Steve Kovach from CNBC. I was wondering what goals and metrics Sundar and your other bosses give you that you’re evaluated on and how or if those have changed over the years.
Yeah, well, so you know, I think like any company we use multiple metrics to evaluate our success. We have sort of at the top of our pyramid quality and responsibility right now. When we talk about quality we’re talking about recommendation quality, search quality, comment quality, etc. We talk about responsibility. We talk about ruling out changes to our recommendation services to make sure that we’re recommending the right content to our users. Those are the top-line metrics that we have and so we put that at the top of the pyramid because it’s really important to...
Steve Kovach: Have those changed over the years?
Excuse me, what?
Steve Kovach: Has that pyramid changed over the years?
It definitely has. It’s certainly evolved over time. There are lots of other metrics, to be fair, that we look at, like how many creators do we have on our platform, revenue, etc., how are users engaging with our videos, right, number, daily active users. So yes, there are many other metrics that we use. But we have been really focused on putting these at the top to be clear.
Steve Kovach: For a long time you guys were pushing for watch time, right? You wanted to get to a billion hours. In retrospect, was that emphasis on watch time a mistake?
We definitely ... yeah, we’ve always had lots of different metrics, to be clear, not just one. When we actually introduced the watch time metric it was the goal of ... a quality one, at the time, because what we saw is that views a lot of times cause click spam, or people clicking on content that they didn’t really want. And we thought watch time was the best way to show our people really engaged.
But we have evolved in many different ways. We also started adding satisfaction into our recommendations, so after people watch a video, were they satisfied with those videos? That’s an example of how our metrics have evolved.
Okay, we’ve got a lot of folks. We’ll try to go quick.
Okay.
Mark Mahaney: Susan, Mark Mahaney at RBC. Two questions, you pick which one you want, the near term and the long term. Near term, in the March quarter there’s a real sharp deceleration in growth, ad revenue growth at Google, some talk about product changes maybe that occurred at YouTube. Do you have anything you can share, any light you can shed on that?
Then the other question is bigger and it’s what Peter was asking you earlier, you’ve been at Google since the beginning of the beginning. Do you think the company has become too big that it really should be more, much more closely regulated than it ever was in the past? Google grew up against in a way one tech monopoly. Do you think it’s become another tech monopoly?
I’ll take the second question, given that you’ve given the two. I did just celebrate my 20-year anniversary at Google and it’s pretty amazing to see the way Google has evolved over time. I can tell you from my vantage on YouTube is if you look at all the people who are getting into video and big, well-resourced companies all getting into online video, this seems like an incredibly competitive time. I’m also often asked the question like, “What about X, Y, Z, all these other providers?” I see that competitiveness.
I also see some of the advantages. I mentioned beforehand, like if we are going to focus on some really hard computer science information problems. I think that’s what Sundar referred, like being part of Google really helps us because we’re able to use different signals and technology that they have to be able to figure out what the right way to do our recommendations and rank it.
Mark Mahaney: You need to be this big to solve the scale of the problem that you have because you’re so big?
Having enough scale does help us build systems that are world class to be able to address this. I do think there are some advantages of that and there are some technologies that we’re able to learn from Google and those techniques that really benefit us. I do see that there are some real advantages that we’re able to deliver a better product because we have some of the ... We run on shared infrastructure and there’s some technology that we can borrow or just even techniques that we can learn from on how to deliver the right information.
We’ll take two more questions real quick.
Nilay Patel: Hi, Nilay Patel from the Verge. You’re CEO of YouTube. Do you think you’ve set a fair goal for your policy team to create a single, coherent content moderation policy that covers beauty influencers, tech YouTubers, and how-to channels, and political comedians? It feels like you’ve set them up to fail.
When we think about ... First of all, we have many policies, right? One of the things I’ve learned from spending a lot of time on these different policies is that there could be certain areas that need a different set of policies, like I’ll just say news and information is a whole category in and of itself. This is actually why it hasn’t been so simple to solve because we need to be able to solve each of these different areas with a different set of experts, deep understanding, different technology.
For example, in news we have misinformation cards. We have authoritative ranking, breaking news shows, right, so we have a bunch of different product solutions. Then I think from a policy standpoint we also are going to have different policies as they appear in different verticals.
Nilay Patel: Do your creators know that they fall under different policy regimes?
Well, people know that if they create ... Look, there’s some standards like how we handle harassment or how we handle hate that applies to everyone, but what I’m trying to say is that if you’re in the news category and you’re producing misinformation we’re going to ... There could be a set of cases that really only relate to news. I’m just saying that we’re trying to understand how to cover everything from a policy standpoint. But when we identify, “Look, there’s a problem over here,” it’s specific to this area, specific to this industry, we’re also going to look and say, “Well, what policy should we change, make that will really affect this effectively?”
I’m going to change my policy. I’m going to let two more because I want to have Ina go and I’m going to let Kevin from the Times ask a question because he wrote a really good story this week.
Sure.
So, super quick, Ina?
Ina Fried: Yep, Ina Fried with Axios. You started off with an apology to the LGBTQ community but then you also said that you were involved and that you think YouTube made the right call. A lot of people don’t really feel like that’s an apology and are concerned that YouTube flags LGBT-positive content just for being LGBT as sometimes sensitive and yet slurs are allowed. I’m curious, are you really sorry for anything to the LBGTQ community or are you just sorry that they were offended?
Yeah, so first of all, I’m really personally very sorry and it was not our intent. Our goal was ... YouTube has always been a home of so many LGBTQ creators and that’s why it was so emotional and that’s why I think this really ... That’s why even though it was a hard decision, it was made harder that it came from us because it has been such an important home. Even though we made this hard, this decision which I’m going to get into a little bit more with your questions, I’m saying we’ve really, like YouTube, we have so many people on YouTube from the LGBTQ community and we’ve always wanted to support the community openly in spite of this hard issue that we’ve had right now.
I’m just saying people have gotten a lot of criticism like, “Why are you still ... Why did you change your logo to rainbows even though you made this hard decision?” It’s because as a company we really want to support this community. It’s just that from a policy standpoint we need to be consistent because if we ... Look, if we took down that content there would be so many other, so much other content that we would need to take down. We don’t want to just be knee jerk. We need to think about it in a very thoughtful way be able to speak with everyone. We’ll speak to people from the LGBTQ community, make sure that we’re incorporating that going forward in terms of how we think about harassment and then make sure that we are implementing that in a fair and consistent way going forward.
I think that it was a hard week across the board and I am truly, truly sorry for the hurt that we caused that community. It was not our intention at all. I do want to ... Not that I want to take away in any way from the hurt or the challenge of what happened this week, but I do want to let you know that many changes that we made in the hate policy which, again, is a different set of policies we think will be really beneficial for the LGBTQ community and there are a lot of videos there and a lot of ways that that community is attacked where we will be taking down those videos going forward and we will be very consistent. If we see that we will take it down. So again, thank you for your question and I really do apologize for the hurt that we caused.
Kevin really does get the last question.
Okay. Kevin.
Kevin Roose: Hi, the story that Peter mentioned was about radicalization and a 21-year-old man who says that YouTube helped to radicalize him into the far right through his recommendations. As part of that story I got to ask a lot of people about YouTube and radicalization but I never got to ask you, so I wanted to ask you, do you think YouTube is having a radicalizing effect on our politics?
Um, I mean, so, um ... We offer, as you know and you’ve researched, a broad range of opinions. We have looked in all of these different areas across the board and we see that we are offering a diversity of opinions, right? So when people go and they look up one topic — whether it’s politics or religion or knitting — we’re going to offer a variety of other content associated with it, but we have taken these radicalization concerns very seriously and that’s why at the beginning of January we introduced some new changes in terms of how we handle recommendations.
I know you’re familiar with them. You’ve referenced them, but what we do just for background for the other people is that we basically have an understanding of what’s content that’s borderline and when that content is borderline we determine that based on a series of different raters that we use that are representative of different people across the US. And when we determine that we’re able to build a set of understanding of what’s considered borderline content and then reduce our recommendations.
We have reduced by 50 percent the recommendations that we’re making of borderline content and we’re planning to roll that out to all countries, to 20 more countries this rest of the year. I think the combination of the changes that we’re making of our policies as well as the changes that we’ve made to our recommendations are going to make a really big difference.
Just to restate Kevin’s question, do you think YouTube helps radicalize people, particularly on the right?
I, I mean, I, it’s, like, you know, you’ve done lots and lots of research on this. Our view has been that we are offering a diverse set of content to our users and that users ... We’re providing that diverse set and users will choose different types of content for them to see, but we see that we’re offering a diverse set over time. All the same, we really want to be able to address it and that’s why we’ve made the changes that we have made.
We could do this double-length, but we don’t have time. Susan, thank you for coming.
Sure, thank you.
Thanks for taking our questions.
Thank you.
Recode and Vox have joined forces to uncover and explain how our digital world is changing — and changing us. Subscribe to Recode podcasts to hear Kara Swisher and Peter Kafka lead the tough conversations the technology industry needs today.
