The Declining Half-Life Of Social Media — With Eugene Wei

0:00A veteran of Amazon, Hulu, and Oculus, who's one of tech's leading thinkers, joins us to discuss social media's transformation and where the technology world is heading next. All that coming up right after this.

0:40realize your AI agents are a massive blind spot. Learn how Okta's identity security fabric can help you secure the next generation of identities, including your AI agents. Visit Okta.com. That's O-K-T-A dot com. Capital One's tech team isn't just talking about multi-agentic AI. They already deployed one. It's called Chat Concierge, and it's simplifying car shopping. Using self-reflection and layered reasoning with live API checks. It doesn't just help buyers find a car they love. It helps schedule a test drive, get pre-approved for financing, and estimate trade in value. Advanced, intuitive, and deployed.

1:21That's how they stack. That's technology at Capital One. LinkedIn presents.

1:31Welcome to Big Technology Podcast, a show for cool-headed. He wants conversation of the tech world and beyond. Eugene Wei is our guest today. He's a tech product expert, formerly the head of video at Oculus and SVP of product and marketing at Hulu, who's also made a lengthy stop at Amazon. Eugene's loosely examined and explained the dynamics of tech consumer products in a way that fully reveals how they work. I've been reading his work for so long, and I'm so thrilled to speak with him today. Eugene, welcome to the show. Thanks for having me, Alex. It's a pleasure. Thanks for being here. Let's just start off with the social media landscape right now.

2:09First question on my mind, why would any platform have anything other than a TikTok style algorithm? And if there's no argument against, don't we just have everything merge into each other and just kind of look exactly the same? I think there's some nuance to it. Certainly, I think one of the big realizations for Western social media when TikTok came on the scene was that if you could dial in an algorithm, kind of manage the signal to noise in a feed, you could really scale that. And it might be something that's more efficient at doing that job than another approach. But I think the issue is that depending on the medium, depending on the interface, a TikTok style algorithm might work well.

3:01It might not. And you really have to take it on a case by case basis. TikTok is a very particular style of media. These short, full screen videos, often with music playing in the background. That interface is really well suited to forcing you to make a judgment on every video that you see. And really what TikTok is interested in is what really interests you. If you're not interested in a video or if you're negative on a video, it's kind of all the same to them. What really matters is positive signal. And they can measure that by looking at how long you watch a video, how many times you rewatch the video, whether you like it, whether you share it, whether you follow that person.

3:44They get very clean signal on all of that. A lot of other social media services that I've tried to copy TikTok aren't short video based or they don't have full screen, you know, pieces of content. They have a scrolling interface where you can scroll past multiple pieces of content at once. So it's very hard to get the same type of clean signal. Like if you scroll past five tweets. You know, and you don't like any of them. Did you not like them? Did you not care about them? Or were they just like tweets that you're interested in, but you didn't care to like them? It's very hard to tell. And these algorithms are very, very dependent on getting a very clean signal in, in order to make really good judgments.

4:26So I think, you know, it depends. It's kind of like how if you look at music, I think if you look at algorithmic music recommendations, while they aren't perfect, they've gotten to a really good place. You know, if you go on Spotify and you click on a song and you say, go to the song radio and just make a playlist based on this single song, you'll get something that's pretty reasonable. Right. Discover Weekly also. Amazing. Right. Something like that. But on the other hand, if you take podcasts and you look at podcast recommendations, which are also, you know, an audio medium, it's much harder to dial in something.

5:05And it's the same if you compare TikTok to all the big streaming media services like the Netflix, the Hulu, the Disney Plus. I find long form video recommendations to be less dialed in than short video recommendations. I think there are reasons for all of these things. So my long answer is really just, you know, it depends. Right. That makes sense. So let's talk a little bit about the positive versus negative signal. I mean, when you don't have all that positive signal in terms of like time that you hover on top of something, your your theory here is that you need to collect some more negative signal where people say what they don't want.

5:44Or how does that play in? You want both. Because that's one of the things that you said about Twitter is they're not collecting enough negative signal as they've moved to this like more TikTok style algorithm. them. Yeah, basically, if you think about what the net or what the Twitter algorithm can see, they can see if you like a tweet, they can probably see if you bookmark it, if you share it. So there are some positive actions that they might see. But, you know, if you found it mildly annoying, and you just scroll past it, they don't see that negative signal. If it was something just like that just didn't interest you and scroll past maybe, you know, were you interested?

6:24Were It's just very hard. It's kind of a murky read on how you feel. And so I really think getting a good sense of your sentiment is key to these types of systems. If you can get it, then it's great because the algorithm doesn't even need to be that sophisticated. Just the sheer volume of data will do something pretty good. Like I think people think people used to think TikTok's algorithm was something magical. It's not really. It's actually, if you ask anyone in machine learning, TikTok's algorithm's approach is actually just very intuitive. It's what anyone would do. But it was their ability to get that really clean read on positive and negative sentiment that made the algorithm work.

7:11So I think you need both. You know, a lot of Chinese companies, the Chinese government, they're very influenced by this field of cybernetics. And cybernetics systems are all about this idea that in one of these systems, every output from one part of the system becomes an input for another part of the process. And so you can think of these social media services as cybernetic systems, where you're trying to get like every user action to be this output that becomes an input into your algorithm to try to figure out, oh, okay, what is this person interested in? You know, we were talking about, you asked me about why not everybody moved to a TikTok style algorithm.

7:55And there definitely was a challenge for Western social media in that it was all social graph based in the beginning. It's like, who do you follow? Who do you friend? It turns out that who you follow and friend isn't a perfect approximation for what you're interested in. Like you have a lot of friends that are just interested in different things than you are. And so we would see that in our, you know, know, Facebook feeds or, you know, our Instagram feeds, you're like, oh, okay. Yes, I do know that person. They are a friend, but a lot of their content isn't, you know, match for your, your own interests.

8:30So TikTok's algorithm, you know, promised a way to solve for that, like correct for that mismatch. But again, there's nuance in how you implement it. Exactly. But it does feel like everybody's tried to implement that TikTok style algorithm. And I'm like hearing you talk about it, I'm like, wow, like there is nuance and it's not really that suitable for some types of content yet. So many have, I mean, and you're in your Facebook feed, for instance, you're starting to see if you're still using it, right. You're still starting, you're starting to see suggested content in your Instagram feed. I mean, suggested content is all like throughout that.

9:06The Twitter feed is now fairly like, it doesn't really care anymore about who you follow. It's all about what the algorithm thinks is interesting. They, it seems like there was a memo inside all these companies that said, do TikTok, do it now. It doesn't matter the cost. I mean, were they that hasty and ignorant of what this, you know, about how their products were different and decided to implement it anyway? And does that mean we just kind of will eventually like all end up gravitating towards like short form videos and having those fill the feed? I mean, if you think about it, that's what Instagram looks like now.

9:43A lot of Facebook looks like that now. Even Twitter is starting to look like that now. Threads is definitely looking like that now for me. What do you think? Well, I think my sense is that probably short video was just underutilized in the West. I don't think it's perfect for everything. Not everyone is good at creating short video. Not everything needs to be short video. But certainly if you look at China, which, you know, I used to go back to China every year pre-pandemic to talk to tech executives there, visit with different companies, see what was going on. And it always felt like you were living further in the future in China with the way they use the internet and their smartphones.

10:24And, you know, they just had short video everywhere because their internet really took off in the era when everybody had a smartphone with a video camera on it. whereas the Western internet took off in an era when we had laptops and desktops and no cameras attached to them and so there's some amount of sort of like legacy uh you know inertia in the West and now we're sort of catching up to oh okay for video you know Snapchat TikTok maybe we've been under utilizing video in every sphere so I do think you know we'll continue to see short video gain market share just because of that. But I ultimately don't think everything needs to be short video and it won't end up that way.

11:11Right. Another thing I've been wondering is why people continue to post on social media at all. I mean, it seems like the half-life of a social network or a social media company is much shorter than it used to be, or used to think it was at least. I mean, look at all these that have gone by the wayside. There's Clubhouse and Vine, Facebook Blue seems to be, at least for me, dormant. Twitter's teetering on the edge. Your theory here is that people go onto these networks to build status, but that status is so short-lived. So do you think people will eventually just get fatigued and stop moving to new social networks and this stuff just becomes a moment in time?

11:50I think some of that fatigue has already registered in the retreat to group chats for a lot of people. which is more a form of social networking than social media. I think when we made that transition from social networking to social media, we caused a huge shift in the way people use these services. We went from thinking about social apps as ways to connect to people we knew to apps where we had to compete with other people to get distribution in these feeds. And that made us all into content creators and media figures. And that, as you know, you're a media figure too. Anyone who gets into that business, the creation business, it's very exhausting.

12:36And I think some of the retreating to group chats is just a sign that a lot of people got tired of that. As for new social apps, I think it's challenging partially because we have these powerful incumbents that all exist now. It's not like trying to build a social app back in the early 2000s when there were very few apps that were big and powerful. You didn't have these like massive incumbents that you had to compete with. And we know that users typically, you know, there's inertia to these things. They will stick with the service that they've already got followers on, where they've already built out a graph and they know how to use it.

13:16You know, for a new service to come along and really displace that, They have to do more than just offer kind of one very narrow experience. And it's tricky because, you know, it's hard enough to get initial product market fit for any social app. But now the challenge is, I think, as soon as you get any amount of traction, you have to build out even more features just to keep pace with the incumbents. So the game has gotten harder. But, you know, in the end, I would always say that people, you know, humans are social creatures. We do want to connect with each other. Even with all these social apps now, you see all these statistics about loneliness being on the rise in the West.

14:01And so I think it's still a problem that we can do better on, that we can solve in other ways. And just because we haven't quite reached that point yet doesn't mean it's not a big problem. So that's kind of how I view it. And speaking about it as a solution to loneliness, I mean, one of the things that struck me in your writing is that you think there's real community on social media. I mean, here's just one of the lines from a recent piece that you wrote. the machine learning algorithms that have been crucial to scaling our largest social media feeds are among the most enormous social institutions in human history.

14:41I mean, for them to be enormous social institutions that you have to believe that this is like real and meaningful connection that's happening here. Can you expand on that a bit? I mean, ultimately, when you think about what social institutions are, they are ones that kind of really determine how we relate to who. And, you know, there's no arguing that these algorithms that operate the feeds and social media today have a huge say in who we encounter on our phones every day and how we interact with them. That's just the structural nature of it. And so there's both good and bad. I think there are parts of social media that are very toxic and not healthy.

15:26But also, I'd have to say that through Twitter, I've met some of the best friends I have now, some people who have been really important to me both personally and professionally. And that probably wouldn't have happened as easily in a previous era. Just the fact that I'm talking to you is a function of probably the fact that I I was able to write and put things out on the internet that other people shared and read and saw. And in a previous era, it's not clear that I would have had that opportunity. So it's always a mix of the good and the bad. Social media isn't purely black and white. But I do think we have to continue to look for ways to make it better for the people who use them, for society, for all of that.

16:17I feel like in Western social media, we're maybe stuck at some local maximum right now, partially because the business models and everything have constrained everything and everybody's copying everybody else. But I don't think we're done building. Yep. You know, when I speak with you, Eugene, when I read your writing, and this is kind of a surprising conclusion because I, you know, I'm reporting on this stuff. I try all these apps, but it seems to me, and, you know, maybe I'm going to read the same facts that you've presented and come to a different conclusion. But it does seem to me that we're going to have, you know, social networks where professional content, you know, content creators performed.

16:54We'll have group chats where like normal people talk to themselves. And it will be like the same exact ecosystem. That's what I'm saying. Like maybe social media was a blip, right? And we fall back to the same ecosystem that we've always had, which is that like people talk about entertainment and then networks program entertainment, except instead of like, you know, the local watering hole talking about what was on TV. You are in the group chats talking about what you saw on TikTok. I mean, why, I guess, you know, why does social media, there's something in the middle where normal people are posting and stuff like that.

17:27Like, you know, why is it a historical necessity to exist? I don't know. I mean, I think there, the internet by its very nature is this two-way medium. So there's always some amount of user participation. it may just be that we have less of it moving forward. I don't think it fully goes away because now that it's possible, like some amount of people are going to be creators and probably a lot of those people weren't going to be creators in a, in a previous area where a previous era where a couple of gatekeepers were going to determine who got distribution through mass media. So I think the share of social media relative to the previous area of mass media, it's going to be higher, but maybe relative to the peak of social media, a lot of people will just decide, you know what?

18:15Wasn't healthy, wasn't great. And we'll opt out and, you know, we'll settle back to something which is like higher than before, but not as high as it was at its peak. Because that's the thing, right? Like you look at threads and it almost felt like we were back in the last decade when it first emerged. It was like a ready built network and people were on there and you had your typical, right? People posting for status. but then it like you see that the retention rates are terrible and i think it's my my theory on that is that it's a combination of people being like i'm not freaking doing this again like i already did this on your other networks you're not going to fool me again because ultimately like i shared the best stuff with my with my networks anyway and then the people who like continue at it are you know are people who are professional creators with some exceptions but that is i mean did you read it any differently?

19:08No, I mean, it's interesting because threads doesn't exist in a complete vacuum. Like a lot of people coming to threads came from Twitter. Some people were coming from Instagram. So they kind of, they're not complete social media novices anymore. And the interface is familiar to anyone who's used Twitter. So it's this weird mishmash. You know, No, I always say that social media like creates these communities that are the intersection of, you know, what is the medium itself and the interface? What are the features? And then who are the people doing it? Like the community itself, because they piggybacked off of the Instagram graph and they drove so many people to threads in such a short period.

19:56it was always going to be a very challenging, I think, opening two weeks because you have a lot of people who weren't used to interacting with each other through this particular medium, suddenly all trying to figure it out at the same time. And it felt a little bit just chaotic. And, you know, because we're such social media veterans now, our patience for waiting for a network to slowly turn into something is just less than it used to be. We have things we can fall back on if we're not into threads. So I understand why threads went this direction. They boosted a lot of famous people initially.

20:34They had the algorithm so you wouldn't get an empty feed. They piggybacked off of the Instagram graphs. They didn't want the cold start problem, but they created a different problem for themselves, which is if you throw a bunch of people into a thing and they try it out and they don't understand what it is or what they're going to use it for these days, they're going to, they're going to bail out pretty quickly. And then it's not clear that it's easy to get them back anymore. Um, it might've, you know, Instagram notifications. Hey, you missed a thread. That's what we're looking at. Right. And part of it is that I wonder sometimes if threads, like, I wonder how much the team actually has a very specific vision for what threads is going to be and how much it was more just like, well, Twitter's struggling.

21:24We might as well just throw a clone up and see if we can weaken them and steal some traffic. I always tend to think that it's a little dangerous when you don't have an internal intuition or compass about like, okay, why are we doing this? What is it going to be? And if you can't articulate that cleanly, it probably means you're going to be a bit lost. And, you know, with social media, you can't afford to be lost for too long. Because once people churn, it's very hard to get them back. So I don't know, you see that with all the other services that are basically, they're all trying to be Twitter, just minus Elon.

22:02Yeah. That's fine. But, you know, the chances that you'll get everybody to migrate over from Twitter and create that sense of community again, it's just, it's very difficult. It takes a while. And I just don't know if there's enough time for these apps. Like I think, like you said, with the half-life, part of it is just the fact that we aren't first-time social media users anymore. So you get a very narrow window in which to try to figure it out. and it's a complicated problem. Yeah, definitely. So where do you, I mean, I think that, well, let me ask you this, where's Threads going? I don't know.

22:47Like from what they've said publicly, you know, it does seem like they're trying to be Twitter, but minus some of the, what they consider, I guess, toxic discourse around news and politics. um they want to be kind of like twitter con like public conversations for other areas that are more positive in their vibes and have a chance of surviving well i mean you know because they're attached to the instagram graph and they have ability to like push people that way they're certainly going to have a longer a runway than some you know other startups we just don't have the resources to drive that much awareness um but i think it's an open question as to whether micro blogging is even the right interface for those types of conversations that they envision.

23:41Like, you know, my Instagram, there's a lot of travel and food and sports, you know, it's like whatever, like communities you're interested in. There's, you know, architecture, there's arts and a whole bunch of things. And actually, I just find Instagram to be a better medium for that. Like, I'm not sure micro blogging is the right interface for a lot of that. And I'm not sure that really famous people, you know, one of the complaints that I think Adam, who runs Instagram had said was, well, the interface of Instagram isn't good for conversation, you know, because it's so dominated by the piece of media that's being posted.

24:22And so they went with a thread style interface, which puts the conversation at the forefront. But you know, if you're JLo or Ariana Grande or whoever. I mean, they're not really going to converse with their fans. It's just like impossible. They're broadcast celebrities. And a lot of them, you see them trying the little prompts in threads like, hey, what's your morning routine to get yourself into the day or something like that. And that does drive a lot of responses because they're famous, but they're not going to go in and talk to all the people who respond. And so I wonder if threads isn't going after a market that doesn't really exist at scale.

25:07Yeah. I mean, I think that's probably what's happening. So then where do you think this leaves Twitter? I mean, I think you and I both agree that it's been the product changes really that have been the trouble at Twitter. Maybe that's a layoff of all the salespeople has been the revenue problem, but the product changes has kind of have opened the door more than like Elon's tweets although some of those are just really regrettable but like it really has been the fact that the home feed is no longer what it used to be and i i mean we had a debate about this last week but i personally think that the very the re-verification process has been a debacle as well what's your perspective yeah uh i don't think twitter is going anywhere i just think it'll be like my my sense is that there still isn't a real replacement for twitter and especially if threads doesn't want to do politics politics and news and that type of stuff i mean that's that's twitter's sweet spot it feels like and i don't think the other twitter alternatives they're all so small i don't think any of those subscale clones will really get any traction and so i think twitter will continue to exist but you know a lot of long-time twitter users have bailed and And I don't think most of them are coming back.

26:21It'll be fine. It'll continue to be this kind of medium-sized social network that can't really grow much larger than it is. And so I think it'll be hard to recoup the purchase price. Right. And I think you just have to write that off. That's not going to happen. You can bring some of the advertisers back. It's a very attractive demographic of people to advertise to. but it's always been challenging for Twitter because they don't have as much personal information about you. You know, it's kind of a pseudonymous network. And so the targeting isn't going to be quite as good as it is on Instagram or Facebook or something like that.

27:04And I've also found the ads really just random in my Twitter feed. I don't know, like, how they choose these ads for me. Yeah. And so, but I don't think it'll go away partially just because, you know, it took so many years to build that community and and we're seeing just how hard it is to port that community somewhere else and you know for all of us to find each other again on some other clone it's so interesting because like as like we've hit these like late social media era you've like written so much about groups and you've studied a lot about groups dynamic you have this guy olson i think it's a guy olson it talks about yeah yeah the organizations um you know when they when they take collective action, at least for large groups, that they often take a long time to emerge.

27:49But once established, they usually survive until there is a social upheaval or some other form of violence or instability. And I'm just like, wait, the guy that I read, help me make sense of what's going on with social media. Like he's talking about end times for groups. Like we've really reached sort of like a kind of dark period of what's happening with social networks where they're like, we're at the fall apart stage. Well, yeah, there's some extent, but also that inertia works in both directions. So, you know, I find a lot of services that I use on the internet now to be worse than they were at their peak, you know, like Google search results or Amazon search results or, you know, my Twitter home feed.