As I sat down with Mark Zuckerberg at his office last week for an hour-long conversation, I was wondering if he’d be less enthusiastic about his metaverse vision than he was a year ago.
For one, Facebook is in a very different place than when he rebranded it to Meta. Its user growth is slower than ever and its revenue is declining for the first time. Meta’s stock price has fallen a staggering 60 percent this year. Low-priority projects throughout the company are being canceled, and there’s a hiring freeze in place. Meanwhile, it is losing billions a year building hardware and software for the metaverse, and managers are having trouble convincing even their own employees to use its flagship VR platform, Horizon Worlds.
During our interview, Zuckerberg acknowledged that the long-term nature of his metaverse bet — he doesn’t expect it to make meaningful money for several more years — “sets up for a trough of disillusionment.” But it was clear that he still sees VR and AR headsets as the next major computing platform and that he wants to be the one driving that shift. “Because it’s going to enable, I think, the ultimate expression of what we set out to do: building social software.”
Sitting between us during the interview was the Meta Quest Pro, a $1,499 mixed reality headset Zuckerberg unveiled today at the company’s annual Connect conference. It’s primarily designed for VR diehards and people who Zuckerberg thinks will want to get work done in the metaverse. As he put it, “Basically, this is a step toward all 200 million of those people who get new PCs every year starting to instead do some of the work in VR.”
The following transcript has been lightly edited for clarity.
Mark, thanks for doing this. We have a lot to talk about, including this thing sitting here between us, this new device you’ve got.
Yeah, that’s a good one.
There’s a lot of ground to cover. I want to start where you and I left off a year ago when we had this conversation, which was about the rebrand to Meta. It has been a year now. The world was very different a year ago. I think your company was in a very different spot, and the world was in a different spot. Looking back now, I would be curious to hear you reflect on doing the rebrand when you did. Did it achieve what you hoped it would?
That is a good place to start. I’ll just start off by saying that about a year ago this time, you gave us a lot of trouble by breaking the news of this rebrand.
Yeah, sorry about that.
You were doing your job, and I respect that.
I appreciate that.
That was just about the biggest thing of the year that we were trying so hard to keep under wraps, but somehow you got that one, so touché. I know there’s a bunch to reflect on here. First, this is a long-term journey. The internal conversations we had actually always expected that the initial moment of the rebrand was going to be quite negative and that, over time, we would just sort of build this out. Just because there were all these questions around what the future vision is that we’re building and all this stuff is ahead, right? I mean, this is the first version of the work VR device line that we’re shipping, and it’s not going to be until later this decade, when we’re on V4 and V5, that this stuff really starts to get fully mature.
So I just figured there’s so much story here left to fill out and this is a long-term thing. So, I’d say the initial response to the rebrand dramatically surpassed my expectations about what could be achieved in terms of people hearing the vision that we wanted to put out there. And I think almost overnight, in the first few months, you had all these other companies jumping in and talking about how they wanted to do stuff in the metaverse, too. I think it really popularized that term and that vision in a way that just dramatically exceeded what I’d actually expected. You look surprised.
Well, I look surprised because it exceeded what I expected as well, the way that metaverse became a meme in the business world and really everywhere. Now, there are chief metaverse officers at companies, right?
It sounds like you didn’t expect for it to catch on quite like that.
No, I assumed that we were going to have to just continue building out the roadmap and that maybe sometime, like five years from now, things would start to click and people would start to understand what we were putting together. Instead, I think a lot of people really got a lot of the vision for what it is. Obviously, it’s still this high-level concept and there’s a ton of stuff that still needs to get built, but I think it caught people’s excitement just as a long-term hope for what we want to build more than I’d actually thought was possible. And that, I think, actually poses different opportunities and challenges. On the one hand, a lot of folks are really excited about working on it. On the other hand, I do think it just sets up for a trough of disillusionment at some point because it is a vision that’s far out. We’re obviously investing a lot, and the products are working really well.
I mean, Quest 2 is working really well, and I think a lot of the research that we have in mixed reality, augmented reality, neural interfaces, all these, I think these are leading across the industry. So I’m really excited about what we’re doing. But it’s not like this stuff is going to be fully mature in a year or even two or three years. It’s going to take a long time to build out the next computing platform. So, we’ll see. That’s been generally good. But then the other piece that I’ve reflected on a lot is, as you’ve said, the world is in a pretty different place now. The market is in a different place.
Your valuation. I mean, it’s dramatically different.
Well, us and pretty much everyone else. Inflation has gone up; interest rates have gone up. When interest rates go up, that weighs on all equity prices. But there are some people who are maybe a little bit overly financial about thinking through strategy who would say, “When you see things like that, then you need to dial back on long-term projects.” I do think you’re seeing that across a lot of the industry. So, there’s a part of me that thinks that it actually would’ve been a lot harder, and probably wouldn’t have been received as well, to have announced this vision this year than last year given where the world is. So the question is, well, how do you feel about that given that people might be less excited about it now?
Listen to Decoder, a show hosted by The Verge’s Nilay Patel about big ideas — and other problems. Subscribe here!
I wake up in the morning and I’m pretty happy that we did it. You could either say, “Hey, it would be looked at somewhat more negatively today or it would be harder to do today, so are you psyched that you did this thing a year ago?” And yeah, because I think it would’ve been harder to mobilize around this now, and this is what we’re here to do for the next decade — or however long it takes: to build out this next generation of computing that’s going to be fundamentally more focused on people and delivering this sense of presence so that you feel like you’re right there with another person. So yeah, I’d say I’m pretty positive about how it’s gone so far.
But also, I don’t know, I feel like I’ve been doing this for long enough to understand that everything that goes well brings its own challenges in different ways, too. So I think you just deal with what’s in front of you. But I’m focused on this over the long term. I mean you’ve gotten some sense of the roadmap, and we’re here today to talk about the next step in the journey. And in some ways, it’s the beginning of a big part of the journey because work is a big part of computing.
There are 200 million people who get new PCs every year, primarily for work. I do think that, as we develop the Quest Pro line and continue building it out, you’re going to be able to do pretty much everything and more you can do on PCs on VR. So I think that this is just another big vector for developing the next computing platform. Basically, this is a step toward all 200 million of those people who get new PCs every year starting to instead do some of the work in VR, in addition to all the folks who are gaming, hanging out socially, etc.
Well, let’s talk about the Quest Pro because I got to try it at your research center in Redmond recently, and it’s very different from the existing Quest line. I think there are two big things that people will probably have not experienced until they try it. It’ll be the first time they see face tracking and mixed reality, and I wanted to talk about those two points with you. Maybe starting with mixed reality, which isn’t VR in the traditional sense — you’re actually mixing video of the world around you with VR. Why is this something that needs to exist? What does mixed reality represent on the continuum of where you’ve gone with VR to date?
Just first, for background, mixed reality is basically where you see the physical world around you and then you can overlay digital objects. So you can think about virtual reality as when the system is basically painting every pixel so you’re in a fully immersive world; you’re in a completely different place. Over the long term, you’ll have augmented reality, which are glasses. Something like what you’re wearing now is basically the target of what we would like to get to. I don’t know if you’ll be able to get much smaller than that because there are a lot of electronics to cram in there, right? All the silicon, the projector, the wave guides to display the holograms, the cameras to basically make sure that all the objects in the holograms are locked in the right place in the world, the speakers, and the batteries — so there’s a lot of stuff to fit into those glasses. But you’ll get that.
When you have glasses like what you’re wearing now, you’ll see the actual photons from the world, things around you, and then you’ll overlay holograms just in that place. So mixed reality is sort of this in between, where it’s a VR device where basically every pixel that you’re seeing in your vision is rendered by the graphics pipeline in the device. But it does this thing called passthrough where you have cameras on the outside, an array of cameras, because your eyes see in stereo. That’s how we get 3D. So it’s not just one camera. It’s important that we get the different perspectives. And it can basically pass that through in high resolution and in color. And then, if given the screen, it can either print what the photons are that it’s getting from the outside or it can overlay digital objects. So you can be sitting at a desk and have your perfect workstation up with three huge monitors, but you can see your physical keyboard in front of you and your physical mouse. So you can control the digital monitors that aren’t actually there.
I tried this last week, and I will say, the monitor thing is compelling. What I noticed was that the keyboard itself was still a little fuzzy, and I didn’t feel like I could see the keys super well to feel confident typing.
Yeah. I mean, I think that all the stuff will get better over time. There’s some kind of tracking augmentation that we can do for certain keyboards, too, so yours may just not have had that. But generally, you can get a sense of where this is going. This is a V1 device. So it’s not like the perfect incarnation of this. I mean, just like Quest 1 to Quest 2 is this huge jump and there’s many times more sales, I do think we’ll keep on building this out. But I think this is the best mixed reality that anyone has built so far.
“This is the best mixed reality that anyone has built so far.”
And this, I think, is enough to introduce the concept to the world, show where it’s going, and get the development ecosystem starting to go. So you’ll get people building use cases for work, whether it’s the desktop or solar productivity example. You’ll be able to have hybrid meetings where instead of workrooms that we have today where you’re in VR and you can see people’s avatars, now you’ll be able to have hybrid meetings where some people can physically be there and you’ll see them, but then other people will just show up in VR and you’ll see their avatars. So that’ll be pretty sweet.
There are a lot of mixed reality use cases, and I think that will show out over time. The other component of this is face tracking and being able to see your face movements and your eye movements. I see the value in the experience. I did a demo where I was with one of your employees in Workrooms, and it did feel more visceral, being able to see how our faces reacted in real time. So I understand the use case of it completely. I’m curious how you thought about building that into the product from a privacy perspective because, obviously, there are going to be concerns about face tracking.
Sure. First, let’s talk about what it is and why it’s valuable. Well, I can answer your question quickly — the privacy side. The face sensing data stays on the device, and we don’t send the raw data to apps. And people basically have to opt in if they want an app to be able to know where they’re looking — the eye tracking — or their face expression.
Importantly, you don’t have the raw data, either.
No, it’s on the device and it’s encrypted, and then it basically gets thrown away as soon as it’s processed. So, from a privacy perspective, I actually think that that’s been…
You think that’s solid?
Yeah, I mean, we’ve also had people come and audit it. I’m sure over time we’ll add more capabilities, and we’ll need to keep thinking through this. So security is never a thing that’s done. But it’s something that we’ve thought through very carefully given the sensitivity around it. But I do think it’s worth just talking about why this is such a big deal. So mixed reality, I think, is clearly a big deal because it’s this bridge between virtual reality, which you can build today, and augmented reality, which you want but it’s still a few years away from really being able to get built. So this starts to bring that experience in, even if it’s in a VR form factor.
The face expressions are critical because it gets to why we’re in this at all. Which is, we’re really focused on the potential of VR and AR to deliver this authentic sense of presence. No other technology can do this. It’s like, when you’re on your phone or if you’re on a Zoom call, it’s nice to be able to see the person. You can pick up some context around them, but your brain is under no illusion that you are there with them. If anything, you’re trying to convince yourself that you’re having a closer interaction. You obviously know you’re in a different room and all that. The magic of VR, for people who have experienced this, is that it basically immediately convinces your mind that you are present in another place and with the people who are there.
And when you see avatars, even if they’re expressive avatars that aren’t yet photorealistic, it feels very rich and present when you’re there — in a lot of ways even more so than what you would get on a Zoom call today, where obviously people show up in a photorealistic way, but there’s just so much that it doesn’t feel like you’re actually present. Whereas, even if you have this expressive, somewhat cartoon avatar next to you, actually you feel like you’re there next to each other, even if you’re thousands of miles apart if they’re on the other coast or something like that.
That, to you, is compelling in and of its own right for this technology — to where you think that’s going to be a reason people gravitate toward this technology?
“We’re a company where everyone here wakes up in the morning and thinks about how we’re going to help people connect and communicate.”
To me, that is the primary value of it: basically the ability to feel and deliver this sense of presence. I think this human sense of presence is such a profound and magical thing. We’re a company where everyone here wakes up in the morning and thinks about how we’re going to help people connect and communicate. You can’t deliver that kind of sense of presence on any of the platforms that we’ve had the opportunity to build on yet. So we built on web, on PC, and on mobile. There are a lot of good things about all those platforms, but if you think about what the ultimate expression of social technology is, you’re not going to get it on a phone. So that’s why we’re investing so much money and so many of our top people in trying to invent and accelerate the development of this next platform. Because it’s going to enable, I think, the ultimate expression of what we set out to do: building social software.
So then the question is, okay, what are all the things that we just need to burn down on the list of things that get in the way of feeling like you’re even more immersed and present with other people? And one of them, obviously, is realistic expressions. So I think that this is going to be one of the defining characteristics of this product. And hopefully, a lot more that we do going forward is the accurate face expressions and ability to make eye contact, which is also really powerful and also something that you can’t really do on video calls today. It’s like, if you try to look at someone’s eyes, you’re not looking at the camera, you know?
So there are all these weird issues that break the sense of presence. But in order to do that, it’s a pretty big tradeoff in the design because you’re putting a bunch of different sensors in there, which consume a lot of the CPU on the device and the silicon power budget that you have basically processing the input from these sensors in order to make it so that, when you’re in VR and mixed reality and eventually augmented reality, your representation of yourself will have realistic expressions. I’m interested to see what happens, but I think other folks in the space… Sony’s coming out with a new headset this year. I mean, this isn’t like a thing that I think that they’re prioritizing.
I think Apple’s headset is going to look and work a lot like this actually.
Well, we’ll see. I mean, I don’t know. It’s been very hard for us to have any sense of what they’re doing, so I find it best to just…
I think we’ll know soon.
Well, we’ll see. Yeah, that’ll be interesting, too. So I think this really gets to the mission of what we’re doing around this.
I wanted to talk about the price of the Quest Pro. It’s $1499, which is a lot more expensive than a Quest 2 at $399. You’ve been very clear that you want to make these devices as cheap as possible to get them into the hands of as many people as possible. I think in 2017 you said you want 1 billion people in VR, that was your goal.
Yep, well, it’s a good start.
I think the Quest 2 has done over 10 million sales to date. Does that sound accurate to you?
I mean I’m not…
Are those the estimates?
We haven’t shared any numbers.
Why not? Why not just share?
It’s a good question. I think we tend to not share numbers until things are a lot bigger.
A lot bigger. So you’re waiting for a certain number?
I don’t actually have a number in mind, but I just…
Okay. Just when it feels right?
I’m not sure that there is any particular number.
Well, the estimates are that you’ve done over 10 million with Quest. I’m sure it’s higher than that. You’re obviously far from a billion. Who is this device for at this price point? The Quest 2 is considered a gaming device, and there’s a lot of social stuff starting to happen: there’s fitness with Supernatural. Who is the target customer for this?
So, there are really two sets of folks. One is just people who want the best VR device that anyone has made. I think, if you want that, this is it. It is better than the Quest 2. It’s a lot more expensive, so it won’t be for everyone, but there’s some group of people who want that. The second is people who want basically a device that’s for productivity. When I think about the market, I think that there are going to be two basic different tiers and price segments. I know there’s going to be a consumer-oriented segment that is maybe $300, $400, $500 devices that people widely can afford — in the price [range] of an Xbox or a PlayStation. A lot of the use cases there will be entertainment-focused, whether it’s gaming, social and hanging out with people, or things like fitness. And that list of use cases will just continue growing. But it’s been pretty cool to see how that’s expanded so far. If you think about how you use computers, there’s also clearly a market for people who want to pay or are willing to pay $1,500, $2,000 — high-end professionals for their workstations.
That’s who you imagine being the clients for this?
If I could give all of our engineers a device and have them be 3 percent more productive, I’d give them a $1,500 device, for sure.
Well, yeah, I mean, for this and for the future of the Pro line overall. I do think that there’s going to be a market for people who want to get… The people who are really interested in VR being able to be their primary workstation over time. I think there’s going to be a market around that, and people who are high-end professionals there, you’re already paying thousands of dollars for your workstations. So I think that’s pretty clear that the ability to get more technology into there to make that even better, you do it. If I could give all of our engineers a device and have them be 3 percent more productive, I’d give them a $1,500 device, for sure. So, in terms of the market segmentation, that’s what we expect to happen. There’s also this advantage in developing both of them, which is that we can introduce new technology.
First in this one.
In this, before we can get it into the price point for the consumer one. And, being able to work on it and develop it actually helps us get it into the consumer one faster and better. By the time that it is in the consumer one, we already have a developer ecosystem and content around it because, even if fewer people are buying the Quest 1 that’s more of a high-end device, it’ll be enough to get the developer ecosystem going. So, of course we’re working on more devices in the consumer line, too. There will be a Quest 3 at some point — not this year. But I would love to get some of these features into future devices, whether it’s Quest 3, Quest 4, and the fact that we’re building Quest Pro and have that and people can start building for mixed reality and all that is, I think, a pretty big advantage on that, too.
You’ve been pretty open that, on the Quest 2, you are not making money on the hardware. Are you making money on this on a unit basis?
I mean, I’d have to look. There are lots of different ways to basically do the accounting on this, as I have learned.
Right. I guess what I meant was, is this a profit-generating device for you?
No. I think the strategy overall is not to make money on the hardware but to make it so that it can help develop the ecosystem. And then, over time, the business model will be based on software and services. So that remains the approach.
I wasn’t sure because you invest so much in hardware. You have so many people working on this and you’re spending so much money on hardware. I wasn’t sure if you had landed on a hardware margin business or not.
No, no, that’s what… I think it probably depends on how exactly you account for it. So, if you’re just saying what are the materials that go into the device, maybe we’re charging a little bit more for the device than the materials that go into it. But if you account for all the R&D and everything, then no way. But no, the strategy is not… I mean, we’re not trying to have premium device prices and make a profit on that. Our whole approach as a company is to get as many people as possible to be able to access these tools, and then, over time, you build a better ecosystem that way.
“Our whole approach as a company is to get as many people as possible to be able to access these tools, and then, over time, you build a better ecosystem that way.”
I think that this is a pretty deep part of our philosophy around this overall. We also want to help build the open ecosystem around all of this. So rather than being insular and trying to do everything ourselves, a big part of the theme for this year’s Connect is all of the partnerships that we have. The partnership with Microsoft, which is going to be fundamental.
Which I’m going to talk about.
We’re not an enterprise company. So making it so that this can basically succeed with enterprises.
Let’s talk about the Microsoft partnership for a minute because it’s a big, sweeping partnership. You have Satya Nadella speaking at Meta’s Connect conference, and he’s talking about a partnership across all their services. You have Teams, Azure, Windows. What’s in it for you and what’s in it for Microsoft in this dynamic? It’s unusual to see companies this large partner in such a broad way.
Yeah. I’m not actually sure how unusual it is, but I agree it’s a very big deal for the development of this because I think both companies are building important pieces of technology for the next generation of computing, and I think we’ll just be able to unlock more together. And one of the things that I’m really excited about… I mean Teams is obviously great, the Microsoft 365 announcements, you can basically have a Windows PC in the cloud, and as part of your virtual workstation, you can just stream that. So on the virtual desk that we were talking about a while ago, you can have three huge monitors that are basically streaming things from your Windows PC in the cloud. But also, the announcement around Intune and Azure Active Directory — which are basically the tools that Microsoft sells to enterprises so the CIOs know that everything on the device is going to be secure for that enterprise (in an enterprise environment) and only the people who are supposed to have access to things get it — that’s a pretty big deal.
I mean, Microsoft has just been building this stuff for decades at this point. So, well, we’re building some of the basic tools around this. And maybe in a decade from now, we’ll be in a somewhat different place, even though I don’t think we’re ever going to be primarily an enterprise company. It really jump-starts this, if we’re selling this as a work device. Work doesn’t just mean enterprise. I don’t know if you would consider your job within an enterprise, but you’re clearly a high-end professional. But enterprises are a big part of this. There are a lot of people who work at very big companies, and Microsoft is clearly going to help jump-start that and also be able to help sell it as part of the solution.
That was going to be my next question. So they potentially send customers to you and they benefit because Azure grows as your ecosystem grows?
Well, I think you can imagine something where a company goes to Microsoft and asks how they can empower their employees in the metaverse. Microsoft has, among other things, the option to put all the Microsoft suite of services on a Quest Pro and make it easy for enterprises to adopt that. I think that’s pretty compelling for both Microsoft and us — and the enterprises that now have a turnkey solution to use all of the Microsoft software that they’re used to inside their enterprise. So that’s going to be pretty powerful. At the same time, I mean, it’s not just Microsoft. We also announced a bunch of software that Adobe is bringing. I mean, they do a lot of high-end creative work.
You have Autodesk as well.
Autodesk. Accenture, which I think, in this industry, they’re not necessarily seen as a massive technology company necessarily, but they do a ton of integration and are one of the big technology integrators and creators that do all the last-mile work for all these companies. They’re an incredibly important player in the ecosystem. Whatever industry you’re in, if you want to help train your employees or help people troubleshoot, whether they’re in a factory or on an oil rig or something, you want that software to basically be able to work not just in virtual reality for training but mixed reality. I mean, that’s awesome. That way, you can see the environment around you. You can overlay the training modules on it. Someone has to build that. Accenture is basically a great company that’s trusted by all these other enterprises to do that.
So I think it’s this suite and set of partnerships that lays out that our philosophy on this is that we’re not trying to do this all ourselves. I think that this actually gets to a more general philosophy about computing that I think is going to be pretty important over the next 10 years. Which, as we see this play out, is that in each generation of computing that I’ve seen so far — PCs, mobile — there’s basically an open ecosystem and there’s a closed ecosystem. So in PCs, it was Windows and Mac. In mobile, it was Android and iPhone. In the closed ecosystem, very tightly integrated, relatively insular, a lot of the value basically just flows toward the closed ecosystem over time.
Listener, he’s talking about Apple, but yes.
Well, yeah, Macintosh and iPhone. In the open ecosystem, basically you have much broader partnerships. So Microsoft didn’t build the chips; they didn’t build the PCs; they didn’t build the App Store. It was all this key stuff that was developed around the ecosystem. Similar to Android. And that’s basically what we hope to build here — is the open ecosystem for the next generation of computing around virtual and augmented reality in the metaverse more broadly, which means that there are going to need to be all these partnerships.
And one of the interesting things that I just think about in the history of computing is that it really isn’t predetermined which type of ecosystem ends up succeeding more. In PCs, I think you’d say that Windows during the ’90s and 2000s especially was really the primary ecosystem in computing. The open ecosystem was winning. In mobile, I think you’d probably have to say that iOS is the winning one, even though there are technically a bunch more Androids than iPhones.
From a profit perspective.
I mean, I think Apple is something like 80 percent of the profits, and in countries like the US, I think that they have 60 percent market share and growing. So I think the closed ecosystem has really won in mobile. But I think, from that mix over time, it’s not predetermined that one model has to win out over the other. And I think we’re going to get a reset in the next generation of computing. So our goal and how we approach this is not just to help build the open ecosystem in partnership with all these other companies but to make sure that, in this generation of computing, the open ecosystem wins again.
Yeah, I think that’s interesting. Is it fair to say you’re taking more of an Android approach than an Apple approach here? You do a lot of custom silicon work, you do a lot of hardware, you build the hardware, you build the software, you build services.
Yeah, I think that we’re still early in the story, so I think that there are pieces of this that we’ve had to build just because there’s no ecosystem yet. But our goal is to basically be able to spread that out over time. In the future, do I expect that great companies like Samsung are going to be building VR devices? Of course they are. And would I love to work with them? Yeah, of course, at the right time. So I think that we’ll need to figure out how exactly that would work. But we’re very early in the ecosystem. I think Quest 2 is really the first mainstream VR device. And I think, before Quest 2, most of these other companies weren’t even taking it that seriously. And now, I think people are more open to it, and there are more interesting conversations happening.
Yeah. I have this theory that when Apple comes out with this headset they’ve been working on, I think everyone will see this dynamic that you’re talking about playing out a lot more clearly. I think right now, you’re out there leading, and there’s going to be more entrants that come in, specifically Apple with their approach. And I’m actually wondering, while we’re on that topic, Apple’s done quite a number on you on mobile with that ad tracking prompt. You said it cost you like $10 billion or something like that in the last earnings call.
I see what they’re doing in headsets. They obviously haven’t acknowledged it, but it’s coming. Do you think those two are related at all? Because what I see is they’ve really hurt your ads business. Their ads business is also growing pretty dramatically at the same time, and they’re about to compete with you in headsets. Do you see those things as connected?
It’s really hard to know. So it’s hard for me to…
You have to have an idea, though.
I don’t know. I think to some degree, it’s hard to know what documents or conversations they have over there that either connect or don’t these different parts of the strategy, or whatever. I mean, it’s certainly plausible that they see this competition in the future and want to hinder us. I think one thing that’s been pretty clear is that their motives in doing the things that they’re doing are not as altruistic as they claim them to be. I’m sure that they believe, at some level, in the things that they’re doing and think that they’re good for their customers, but it can’t just be a coincidence that it also aligns very well with their strategy. It’s hard for me to go too deep on this because, I mean, I don’t work at Apple. I don’t know them that well. And, at the end of the day, I can’t really control what they do.
Right. And they may not let you on their headset. They may not let you do your apps on their headset. That may be the dynamic that plays out. I guess where I’m trying to connect the dots here is, do you feel a necessity to build this stuff because of how the platform dynamics have shaped out on mobile?
That’s not the main reason, but I think it certainly is one consideration. My belief in the notion of the metaverse in these platforms dates back to before I started Facebook. I told the story last year about how I remember when I was in middle school, and I used to be in math class and would just not be paying attention to my teacher and would just be writing code in my notebook or ideas for things that I wanted to go code or build when I went home from school that day. And one of the things that I was really excited about was this idea of a social environment where you could be present and immersive in a 3D environment around that. At the time, I was a middle school kid. I didn’t necessarily have the math background or the computing power available to build a bunch of that stuff. Pretty much my whole life, I’ve been interested in the intersection of technology and how people relate to each other.
In college, I studied psychology and computer science. This has obviously been the full… the history of the company has been building this social software. So I think that more of the motivation for this… It’s this long-standing notion that basically this will unlock, in my view, the ultimate expression of social technology. The ability to be present and feel like you’re there with another person no matter where you actually are. I don’t know. I mean, just think about this. It’s like we’re doing this podcast live in person. You mentioned the other day that you hadn’t done a ton of in-person stuff since COVID started, but we don’t have the technology yet to basically do something like this and have it feel… There’s some reason why you want to do the podcast in person. There’s a connection that you have in person. There’s a reason why you didn’t want to do this over Zoom.
Sure. We’ve done it over Zoom. It’s not the same.
So I just think that when this gets developed, I really think the ability to have this conversation where, in the future, I’m just a hologram sitting here — if we can’t actually be here together if I’m in another country around the world — or you’re a hologram but it feels like we’re physically there together, I just think that there’s a real magic to that. That’s going to enable really great experiences.
And you can get some of that in VR today. But unfortunately, there’s this dynamic that we’ve seen that things that feel really neat and present when you’re in VR, they don’t yet translate that well to 2D. So if you take a video of us sitting there in workrooms, we might feel like, “Oh, wow, this is actually pretty amazing.” It’s like we feel like we’re right there. But then you put it in 2D and you put a video online…
It collapses it.
It’s just like, “Oh, that just looks flat or doesn’t look that interesting.” Which I think is part of the reason why we’re not doing this in VR right now, even though I think we very well could. But I think we’ll get there over the next few years, too, in terms of the graphical fidelity and photorealism around more of this stuff.
To hear you talk about it, I mean, you’re obviously so passionate about it. You have such a deep conviction about it. Do you feel the outside people still doubting this strategy and doubting that this is actually going to be a thing at the scale that you seem to just have deep conviction that it will be? I think there are a lot of people that still don’t (a) understand what the metaverse even means but (b) also see the appeal of why they should even try one of these headsets. Do you just think it’s going to be slow, gradual — more people and network effects will come in over time?
Yeah, yeah. It’ll start slow and then it’ll get faster and faster and faster.
So you’re not seeing some aha, killer app moment that really catches on like wildfire? You really see this as being a gradual thing throughout this decade?
Well, I think if you’re trying to build something at the scale of billions of people, that doesn’t happen overnight.
Right? So I mean…
I mean, Facebook grew really fast.
It took eight years to reach a billion people.
I mean, yeah, in hindsight, when you say it that way.
I actually think I saw someone saying that this week, or today, is the 10-year anniversary of us reaching a billion people, which I think is kind of an interesting thing.
“If you’re trying to build something at the scale of billions of people, that doesn’t happen overnight.”
But no, it took a while, and it’s a lot easier to grow software on top of a platform that someone has already built than to basically ship atoms around. So yeah, I think this happens more in step functions, like you’re saying, but it’s like all these little S-curves that add up to getting there over time.
I don’t know. I enjoy being doubted. It’s…
How do you still enjoy it? Why do you enjoy being doubted?
I don’t know. If too many people get or think that what you’re doing is obviously going to happen, then, I don’t know. I just think it gets a little comfortable.
So you feed off the hate a little bit?
Hate is different from doubt.
Right. So you feed off the haters, then?
Well, I actually think one of the difficult things about running one of these companies, or just being a public person on the internet now, is separating out people who are constructive about trying to build something versus people who are just haters. There are a lot of people who are not trying to help anything.
You probably have more haters than just about anyone in town.
I mean, I think once you reach a certain scale, I think you get saturated.
You probably have to separate that out for your own sanity, too, I’d have to imagine.
But also just like, you’re trying to find a useful signal. I mean, if you tune out everyone who thinks that you’re not doing something right, then you’re going to miss a lot of really valuable signals to do stuff better than you’re doing it today. So you want to not ignore critique, but at the same time, I just think that there are a lot of people who actually aren’t trying to help and aren’t trying to make things better. I think it’s a continual struggle for, I think, a lot of public people on the internet, not just me.
Yeah. I wanted to talk about what’s happening on the the social media side a little bit. Because you’ve got this big change happening with this Discovery Engine approach that you’re undertaking in Feed. I mean, the elephant in the room is TikTok. And what they introduce to the world, where Facebook and Instagram were built on these friends and follower models, is showing you content from them, mostly. You call it “connected” internally versus “unconnected.”
TikTok was like, “We’re just going to show you what we think our AI thinks you’ll like.” And it turns out, people really like that. And you’re currently rearchitecting the Feed to be more like TikTok’s For You page in the sense that it’s going to be content that you maybe necessarily didn’t even know you wanted to see.
And you said that this is a huge AI challenge. It’s a huge technical undertaking. I think you’ve said internally, it’ll take a while for you to be where TikTok is in terms of how good…
Very technical term.
I’m wondering, what is the actual AI challenge that you’re currently having the teams go through to make this discovery engine as compelling as you want it to be?
The basic thing is, when we got started with News Feed, we were years away from having the technology to be able to do the content understanding on each post in the system, understand what they’re about, understand what you care about, and then be able to recommend you, in real time basically with very low latency, content that you might be interested in from across tens of millions or hundreds of millions of posts that are out there.
If you think about it, it’s actually a lot easier of a problem to basically take the several hundred things that your friends and the accounts that you follow have posted today and just rank them in the order that you might want to see them. And we’re not necessarily recommending you content; you’ve chosen to follow those people. We’re just making sure that, okay, if your cousin has a baby, you’re not going to miss that. Whereas if someone who just posts a ton of stuff and you always ignore it posts again, that maybe that’s a little further down or something.
I think that there are really two major innovations recently. And I think you’re right that TikTok really showed that a bunch of these things were possible. One is the emergence of really shortform video. So when YouTube first came about, people called YouTube shortform compared to TV shows. Now, YouTube is kind of longform.
Which is crazy, and we’re old. But I think that partially what TikTok has shown is that basically there is a medium, which is even more shortform video, that’s really powerful. Which I think builds on some of the stories formats that we’ve seen over the years. But I mean, that’s a thing. That’s one trend that I think is just pretty clear. Video is becoming bigger across the apps: I think it’s something like 50 percent or more of the time spent on Facebook now is watching video.
You’ve been calling out this video trend for years. That’s a known thing. I think the ranking thing and the focusing on unconnected is the big unlock.
Yeah. That was what I’m going to get to. The AI technology to now not just be able to rank the content that you’re following from friends but also really be able to actually do a very good job of basically recommending content from the whole corpus of content that’s out there and making that be good, that’s something that I think has only really started being possible in the last few years. The thing is, that doesn’t just apply to video. So while TikTok might just be doing that for video, Facebook and Instagram are a lot of formats. So on Facebook, there are photos, text, links, news, groups, longform video, Stories — all these different kinds of things. On Instagram, there aren’t all of those formats, but you have photo, longform video, Stories, hashtags — a bunch of different things.
The same basic discovery engine technology that makes it so that you can understand what a person might be interested in, understand the meaning of all these posts, and then match that up so that way you’re showing a person something that they might be interested in even if they never expressed an interest in that thing directly, that’s going to apply to all these different formats. So I think that that’s a really interesting problem and a really interesting opportunity that we uniquely have to be able to go build. Because, over the history of our company, we’ve had a number of competitors that focus on a single format for something. And one of the things that I think we’ve done well is taking on the challenge of blending the different formats together into a single feed that basically can make it so that you can get all of the different content that you’re interested in. Because it’s not all going to be video that you want to watch with sound on.
And you clearly can’t just flip a switch on this, changing Feed into this Discovery Engine. It sounds very complex.
Yeah, but also, it’s not a binary thing. So to your point about flipping a switch, we don’t have to. What’s basically going to happen is that, over the next year or two, what you’re going to see is just that we’ll start showing more recommended content in the Feed, and we’ll know that we’re doing a good job because the content in the beginning is going to displace some other content, and either displacing that content is going to lead to negative feedback from people and lead to people connecting with each other less in all the metrics that we focus on or it will actually lead to people connecting more and being more satisfied with the product.
So, in the beginning, we’ll start off with, whether it’s 10 percent, 12 percent of the content that will be recommended. But I actually think we’ll get to a future in the next couple of years, where, I don’t know, you might have 30 percent, 40 percent of the content is recommended. The people who you care about are always going to be really important to us. So that’s not going away. You’ll always really want to get that content, too. And I think that that’ll be an important differentiator for our services: being able to do that in addition to the recommended content. But I do think the amount of recommended content will ramp up.
I have another kind of Feed question, but while we’re on TikTok, you were very early on saying [with] TikTok, there are problems with the Chinese ownership. [That] we should be concerned about this. You gave a speech a few years ago about this. Now everyone’s…
The speech wasn’t about that.
The speech was about…
No, but you talked about it in that Georgetown speech.
The speech, I think, was broadly about how I felt like some of the calls to censor more content were getting to a zone that I felt was dangerous and too much. And I’m not a complete absolutist on this. I think that there are a lot of things that are problematic that need to be moderated and dealt with. But I also think that there’s a line, and I think we need to make sure we hold that line well.
Well, where I was going to go is that I think that the TikTok fear has only grown stronger. It’s being talked about in DC all the time.
What do you think the US government should do about TikTok? Are you in favor of a ban? Are you in favor of a spinoff? I think you said that a ban would be problematic at one point. But what’s your view on that? Is this an area that the government should get into? Obviously, it would help you competitively. But do you have other concerns about it?
I don’t know what the solution is, even though I do think it’s a real question. I think, in the US, one of the things that is interesting is that there’s such a clear distinction between the private and public sectors, and the companies here operate independently, and I think people understandably get upset if the integration is too much or if there isn’t good separation there. But when I travel overseas, one of the things that’s surprising to me is, in most other countries around the world, those sectors are blended so much they almost don’t believe that America operates the way it does — and China more so than any other country that I’ve been to. It is very integrated. The notion that an American company wouldn’t just obviously be working with the American government on every single thing is completely foreign there. Which I think does speak at least to how they’re used to operating. So I don’t know what that means. I think that that’s a thing to be aware of.
This sounds like you haven’t formed a real opinion on what should happen.
I try to spend my time on things that I can make an impact on.
Well, so on the Feed, I haven’t heard you talk about this or reflect on this. The last era of the Feed was this thing you called Meaningful Social Interactions, and you were really prioritizing content that your close friends commented on and engaged with. Discovery Engine feels like a departure from that — and that you’re introducing content that’s not from your friends into the Feed. Is that a fair distinction? Do you feel like we’re…
So I think it’s a different loop. The way that people interacted in the past was someone would post something, and then there would be a lot of comments in Feed.
I think the way that social services have evolved is that most of your real interactions at this point are in messaging. So the way that Feed is primarily creating value at this point is showing people content that, then, you go find and send to your friends in messaging and have real interactions in messaging. So from that perspective, it used to matter more who posted the content that you saw in Feed because, if you’re commenting on it in line, you’re interacting with the person who posted it. Now, I think it still matters in the sense that you want to know what’s going on in your friends’ lives, so if you’re not getting updates from them in Feed… There’s a set of people who maybe aren’t your closest friends — and maybe it’s your second or third ring of people that you care about, but they’re not going to personally message you to update you on everything — that you still want to know what’s going on with. Feed is important for that.
But increasingly, what we’re seeing in the flywheel around discovery engine is, whether it’s content from a creator or content from a friend, you see something interesting, you send it to a group of friends, or a friend, and then you’re interacting there. That actually does facilitate real interactions between people. But I don’t know, I’m sure you’ve seen this in your own use of the services.
I mean messaging is really where most of the…
I think everyone has this feeling about how their feeds have changed. And I guess what I’m trying to get from you is, do you feel like the MSI — the Meaningful Social Interactions era — what’s your take on how it went? Because I think you said at the time, “If it works, it’s going to lead to less time spent on Facebook, and I’m okay with that. I want it to be time well spent.” Now, you’re shifting into video, which is a lot of consumption and a lot of viewing.
Yeah, well, so there are a few different things there. The meaningful social interactions shift that we made was when we started seeing this trend that basically people were starting to comment less. At the same time, people were watching more video, and longer-form video especially was displacing a bunch of content from friends. And people were actually writing in and saying, “Hey, I’m missing stuff for my friends because I’m watching all these videos.” So we were like, “Okay. Two things that we want to do. One is we want to make sure you’re not missing content from your friends because that’s what we’re here to do. Yeah, you can find some entertaining videos here, and you can do that in other places, too, but you’re not going to find that content from your friends in other places.” So we want to make sure we’re doing that well.
So that was the comment that I made around, even if we show less video and time spent goes down, which it did, that still seems like a good thing over time because we’re here to help you interact. We’re constantly evolving the algorithm and the way that Feed works. But at the time, we wanted to make sure there weren’t changes that we were making that were basically leading to this decline in commenting in line. So we wanted to make sure that we were building Feed and optimizing for people interacting, not just viewing content and passively consuming it.
Do you feel like the Discovery Engine will get people to consume more passively? Is that a concern for you?
Well, no, because of the thing that I’m saying, which is that most of the meaningful interactions at this point are shifting to messaging. What we see is, if we build Feed in a way that isn’t necessarily trying to optimize the amount of time in Feed but the engine that’s driving thinks, “If I show you a piece of content and you think it’s interesting and let’s say you don’t do anything in Feed but you send it to a friend and then have a message thread there,” like if that’s good…
You can goal toward that.
Yeah. And I think that is good. So I think that the initial push toward, I think what is technically in the press called the MSI change — which, from my perspective, was more of this sort of directional shift in Feed that we’ve evolved toward and continue pushing on — no, I think that reflects the values of the company. It’s like, of course we want to have Feed lead to more interactions between people and not just passive consumption of content. But I think you also want to make sure that we’re being dynamic about what are the ways that people are actually interacting.
I would love to know what you think of BeReal. They’re growing very fast. Have you tried it? What do you think about what they’re doing — focusing on close friends?
I mean, I think it’s interesting. Yeah.
Interesting? Is it something that you think your products could use more of? Or do you feel like that’s more the messaging component?
So, my basic sense of the ways that social media is evolving is basically there is a kind of whimsical, fun element of it and then there’s a professionalized element of it. So if you look at the whimsical, fun element, it’s basically this constant pursuit of finding new things that feel authentic, whether it was initially just being able to update your profile at college, being able to have a status update, being able to take an ephemeral photo or post a story. It’s like these things, I think, in the beginning, they feel so fun and whimsical, and people feel like they can be authentic doing it. But then over time, you get used to it, and then it gets a little bit more professionalized, and then you need…
The brands come in, etc.
Yeah. And then you need a… So I think that there’s just a constant need for innovation of new things like this. And I think what they’re doing is an interesting example of it. But I think what’s going to be challenging is that those things have a time limit. And then the companies that continue doing well don’t just do one but basically are able to build a bunch or implement a bunch. So that’ll be an interesting thing to see. But I think what they’re on to is certainly interesting.
The other direction that I think we’re also seeing more and more is just the professionalization. And that’s basically, I mean, the other way to talk about this is the creator economy. Which is, I mean, creators, a lot of them really look at it as it can be very business-focused. You’re trying to create content that you think is awesome and connect with an audience that you care about and express their values. So it’s not cold, and it’s this important thing. But at the end of the day, a lot of the creators also care about, “Where am I actually going to be able to most effectively reach the people in my community? Where am I going to get the most engagement and have the highest-quality engagement? And then, ultimately, how am I going to be able to support myself and make a good living doing this?”
And I think as those tools get built up more, the professionalization and the creator economy around social media is growing to be a bigger and bigger part of what social media is. And that obviously has a pretty big flywheel with the discovery engine, too, because without a creator economy that is robust, you don’t have a lot of content to recommend. So you need that base of good content in order to have the discovery engine go well, too.
Last question. It looks like Elon [Musk] might actually be buying Twitter, after all. Any advice?
Do you think Twitter’s going to be better off? I mean this has been such a wild saga. I’m really curious what you make of it.
I don’t know. I think this is another one of these things that it’s really unclear how it’ll actually turn out. So obviously, it’s out there, and I think it’s interesting as a saga, like you’re saying, but I think even at this point, it’s not actually clear what’s going to happen.
All right. Mark, thanks for doing this.
Yeah, happy to.
Decoder with Nilay Patel /
A podcast from The Verge about big ideas and other problems.