Would a World Without Algorithms Be Better? — With Kyle Chayka

0:00Let's talk about how algorithms are flattening culture, and whether there's a better way to solve the problem, with a New Yorker writer whose new book on the subject comes out this week. All that and more coming up right after this.

0:42LinkedIn Presents.

1:09Welcome to the show. Thanks for having me. Thanks for being here. I've been a reader of your work for, I can't even remember. I mean, it must be more than a decade at this point. I've always found you to be an extremely astute observer of what's going on online and more and more about how that reflects offline. So we're going to talk about all that this week. And why don't we begin with, I mean, you're writing a story or a book about algorithms and how they filter our world. I mean, obviously at this point, you're kind of throwing your hands up and being like, we don't like what the algorithms are doing.

1:43What's your criticism of algorithms? Well, I think, I mean, to begin with, I do think algorithms have been necessary for the Internet as it exists at this point. Like since the advent of social media, social networks have gotten bigger and bigger. We produce more and more content. like users generate content at an extremely fast pace, extremely high volume. So you need some way of sorting it. You need a mechanism to figure out what you might actually want to see and sort that out from what you just don't care about or what you could easily miss. But I think at this point, we're facing this inflection point in social media and on the internet in general, where it's harder to like be outside of the algorithmic feed than it is to be within it so these these feeds and recommendations have just taken over so much of our experience online that i think that's kind of choking new experiences and making it harder to like find new or original or creative stuff okay so let's just get into it because i have some counterpoints i want to bring up to you first of all like it's definitely true that we're moving more towards algorithm and less toward human signal right we're going from a place where like we are not uh in the follow model anymore as much as we are the algorithm recommending and this is specifically for content right so it's less about who you design to follow on facebook or twitter and more about what tiktok shows you but that being said these algorithms do like they are designed to find new stuff that you might be interested in.

3:25The Tic Tac algorithm, for instance, looks for things that, of course, that you have watched and you're more likely to watch and stay with. But also every now and again, it will feed in a video of some interest that it's not sure whether or like you will want and see how you respond. And if you do like it, then it will send more. And that's how you end up in different rabbit holes. So there's an argument to be made here that algorithms actually do expose you to lots of new stuff you might not have otherwise. I'm sure you've thought about this when you wrote the book? What's your take on that? I think they do highlight new stuff.

4:00Otherwise, they wouldn't work at all, right? Like you don't just want the same thing over and over and over again. There's a variable in there, whether it's Spotify or X or TikTok, that's maybe marked surprise or a difference. I'm sure that's actually how it's labeled. Surprise. Yes, let's twist the surprise knob a little more. and that would be like an experience of randomness right the tiktok for you feed doesn't know if you're going to like this new video of a dog doing something funny but you probably will um so it feeds that to you sees if you engage then delivers more if you do engage with it delivers less if you don't so i think it it's nice as like a discovery engine in some ways but i think the overall effect I mean on one level I think there's not enough surprise like the surprise knob is not turned up enough and we have no way of turning it up ourselves I can't tell TikTok to give me weirder stuff directly I can't shake Instagram's idea that I only want to see cheap European property and so I'm kind of stuck in these like reflections of my own desires that I actually am bored of feed?

5:15Oh, yes. Yeah, like every Italian stone village I've seen. That's the view into my subconscious. So I suppose algorithms absolutely deliver us new things. I think they desensitize us from following those leads once we do find something new, because we just expect that there's another new thing around the corner, another new thing. And I wish they don't even and talk back a bit more. But we do talk back though in many ways. Like if we hover on a video on TikTok, we're sending that signal. And in fact, like you - But that's a passive signal. I would disagree. I think it could be active, right? Like I think the experience of using a TikTok right now is like you almost are aware of how the algorithm is interpreting.

6:02So like if you don't want to be recommended stuff as opposed to, first of all, I think you can say, show me less. But also if you want to see more, you know at this point to hover and maybe i'm giving it too much credit but you know to hover on that video and so you'll get more of it but again maybe i feel like exaggerating what's extremely online power users are more aware of how the algorithm works and how to like tune the algorithmic feed to what they want so to talk back in that way of sending signals yeah but i worry that kind of the the mainstream user or the person who is not so aware of the influence of an algorithmic recommendation just kind of like stays within that that feed that's just still recommending kind of more content that may be engaging but it's not ultimately leading you to more interesting places so are you basically saying that the algorithms are not working the way that they are designed to because let's just talk about it right like it shows you that surprised to see if you're interested.

7:04So like, for instance, if I ended up with a Tic Tac video of property in Italy, that would be a surprise for me. But if I like sent signal, sent Tic Tac the signals, which I don't even have to think about, right. If I watched that video through and it would send me more. So it sort of automatically follows those leads. Yeah. It's the sense of like surveillance, like you are communicating through your actions, what you're interested in or not. I mean, if the algorithms work the right way, yeah, for sure. Yeah, and I mean, I use TikTok a lot. I feel like I have that sense of how it reads your subconscious almost and directs you down these various rabbit holes.

7:47But I think, I mean, that model of surveillance and the passivity of those signals, I don't know. I don't think that ultimately is the only way we need to interact with culture. I think I would like there to be something more beyond the passive signals, to have people seek things out more, to have people even be able to follow artists more closely. I think another consequence of these feeds is that you might watch a musician's TikTok video for a while because you're interested in it and you might get more, but you kind of expect it to come back to you without you having to try at all. So it builds these kind of passive consumption habits.

8:34And that's like, I think it's kind of unhealthy for us, even though it's entertaining and hypnotizing, but it's really unhealthy for the artists who are not making as much money, who are not getting as much attention as they might, if they had a more direct audience that wasn't so mediated. That is, that is absolutely true. So we do now end up, instead of following a single artist that we love, we can follow a genre. Actually, yesterday, in my surprise box on Instagram, I got this guy banking these pipes and making beautiful music of it. And I actually had a moment where I was like, should I follow this person?

9:09And then I was like, no, but I think that Instagram probably knows I played it like 60 times because I'm writing and I have this bing, bing, bing, just music going on in the background. I'm like, they're going to bring it back. but I don't have any incentive to follow the artist. And that's true that we now end up in a much more, you know, amorphous area of interest as opposed to much more intentional. And that does change things. There's like an ambient consumption of stuff that you're interested in rather than like a super engaged, like in Spotify particularly, I think they talk a lot about lean in versus lean out consumption, like lean in consumption being when you're truly paying attention, when you're actually like listening to the song that's in your ears and thinking about it.

9:53And that's leaning in, being, being attentive. And then there's lean out consumption, which is just letting stuff play, experiencing it ambiently, putting it in the background, essentially. And I worry that all of these algorithmic feeds kind of lead to more lean out consumption than they do lean in consumption. So you have this pipe banging guy and you're like, oh, cool. I like a little bit of this. I like to hear it. So if I listen to it, it's going to come back to me, you know, like a piece of driftwood on the tide or something. I'm not going to take that next step to follow the guy. You're not buying his album of pipe banging.

10:33I did. I did share him on my WhatsApp channel. His name is Jake Chapman. His, uh, his, uh, handle is chappy, Mike chappy milkshake. So if anybody's interested in what I'm hearing, It's actually terrific music. So maybe I will follow after this, but you're right. We don't, we're not in the moment where we are following this stuff intentionally. And that leads to like a sort of a dissociation between our interests and the stuff that we follow. But okay, I think that this is kind of clear, like how this conversation was going to go. But for the sake of argument, I'll stand on the side of the algorithms and we'll let you stand on the side of algorithms are maybe not all they're cracked up to be.

11:10From that vantage point, and I would say that maybe they really do work in pointing us to the stuff that, that they've basically, what they do, and I think you've mentioned this, is they kind of strip out like a lot of the middle class of content creation. And they point us to a much smaller amount of, of stuff that is really well liked. And the argument that would, I would make on the behalf of the argument, on behalf of the algorithms is that a lot of that middle class was actually really dull and uninteresting. And I'm not talking about professional content creators, but I'm just saying like, let's take the Facebook newsfeed, right?

11:48So the Facebook newsfeed, most of the stuff posted within it is pretty boring and non-engaging. And what it does is it takes all these posts from these thousands of different inputs and shows you, you know, maybe here, these are the three or four like really interesting things that you will engage with. You will learn from you will find interesting. Now, of course, there's been problems with that because it's pointed you to the things that will most outrage you and most make you want to overthrow government or whatever. But when it works the way it's intended, it actually sifts through so much of the boring, dull stuff and actually gets you to interesting things.

12:23And as a consumer, that's maybe something to celebrate. For sure. I mean, in the Facebook case, Facebook made this decision to collect all kinds of content into its feed, right? It's not just friends and family, it's articles, it's videos, it's random meme group page content. So they piled everything into one platform. And because you have all that different kinds of content, because you have such a high volume of it, suddenly need to sort it in a much more aggressive way. uh so in that context the algorithmic feed really works because it's giving you the stuff that other people engaged with that other people commented on um but i think it's also possible that maybe we just didn't need that fire hose of everything at once to begin with like i think you really need that filtering and that sorting when you have an undifferentiated mass of all kinds of stuff and you need to like pick through it because it's not all very high quality it's a lot of mediocrity um if you rewind to like a bygone era of say facebook in 2011 or instagram in 2013 or something you didn't need so much sorting because it really was just the people who you were in direct contact with you were following a much smaller number of people the feeds felt quieter in a way um so i think the algorithmic the rise of algorithmic recommendations has followed this expansion of of digital platforms into all types of content and i kind of have a desire to go back to that earlier stage where maybe we didn't need so much sorting because things were more built for a specific purpose and stuck to that purpose rather than like every platform doing everything at once yeah this was a debate that i had with uh Frances Haugen, the whistleblower at Facebook, who didn't really like the interpretation of one of the papers that I had, I guess, found in her document dump.

14:24I didn't like my interpretation, which was that if you took the, Facebook did this study and found that if you took the newsfeed algorithm out and just did reverse chronological stuff, you had much more spam and content that people didn't like. And it's just a disaster. And therefore we need the algorithm. And what she was like, and I think this is like a pretty good point in retrospect, which is that if you take the algorithm out, you're not just left with the experience of the platform absent the algorithm. You're left with the platform that needs design changes to make it so that you could use it and enjoy it without this recommendation algorithm.

14:59And what my argument did was leave out the fact that Facebook would then make changes to its platform, potentially good changes, that leave it less reliant on the algorithm and more encouraging maybe, like you said, smaller communities as opposed to this big firehose. Yeah, I mean, I think the idea that every platform should do everything has led us in a not great direction as far as user experience. And I think, I mean, if you think about some non-algorithmic spaces online right now, like a Discord server or maybe a linear Reddit forum, just like the most recent posts, those work because they're not huge, because they're more intimate, because they're based on a specific context.

15:48And that's not how Facebook or Instagram or X or TikTok work at this point. So I do think you kind of have to rethink the entire mode of how we consume things online in order to build things that are not so algorithmic and encourage more intentional consumption. And that sort of goes to the second main thing in your book that I wanted to press you on, which was that if you don't have the algorithms, what do you have? So you can potentially redesign certain products to end up with smaller communities and smaller experiences, and potentially that solves some of the problem of filtering the signal from the noise.

16:29But you're still going to have this big society with all this stuff out there. And so what do you do with it? Who then determines what you might find interesting? Now, of course, part of that is going to be you. You'll have the agency. But of course, in the vacuum, we're going to have tastemakers. and I kind of hate tastemakers. I find them to, first of all, be there's, there's like tastemakers are so elite and often like follow elite tastes, but the real thing, and you know, maybe this is just me because I have about as much basic taste as possible. So I'm kind of a sucker for filter world, both the book, I guess, and the algorithms.

17:11But what tastemakers often do is tell us like how, how things don't measure up to the bar and how things are bad and why that music is bad and that art is bad and that movie that got mass appeal is bad and like i think they just sometimes try a little bit too hard to disparage things as opposed to appreciate the good for when it's good for so many people what do you think i mean the the link between taste and elitism i find so annoying like problematic i think it's like i don't know when to trace it back to like was it the 80s was it the 90s but there's this idea that when you have taste it is pretentious or to say that you have taste is pretentious and is elitist i think to me this is just talking about taste philosophically i suppose everyone has taste in stuff like you have things you like other people have things that they like less or more than you do.

18:13And it's not a problem to think about that. It's not inherently problematic to be like, I really enjoy this thing, whether it's a sports team or noise music or minimalist classical composition from 1973. Everyone can like all of those things. So I suppose rather than encouraging the return of elitist gatekeepers and tastemakers, I would rather encourage everyone to become their own tastemaker. I think in a way I see that happening on Butterbox, the film review social network, where it kind of encourages all sorts of people, everyone on the site to post their own little reviews to think about what they like about a film to rate it.

19:07I suppose that's the kind of tastemaking that I want to see more of. And that's also an example of a non-algorithmic platform that's still fun and cool. But is it chicken and egg a little bit? Because then how do you find the stuff that you're interested in? No, it's very true. But how did people find it before the internet like babes they read magazines they read newspapers they talked to their friends about what records they bought in the record shop which was probably helped along by a human being like i think the human being as a cultivator of culture and as like a accelerator of that connection of of one recommendation passing to another person is good.

19:52Like I, I'm a big fan of recommendations. Yeah. Yeah. Word of mouth. Like you should tell your friends what you like and why you like it. And that's not an inherently elitist act. Yeah, totally. No, I think that you're right. That's more of the grassroots tastemaking is like a great way to, um, sort of replace some of those or hedge against some of the elitist tastemaking. Like when I travel and we're going to get into travel in the second half. But like, I'll of course like read what's on TripAdvisor and stuff like that and Google Maps. But my favorite way to find out, and I don't think this makes me special, again, going back to the basic tastes, but my favorite way to find out what's interesting around the place is just to go and ask the people that live there and are familiar with the places themselves because they'll know better than the internet, you know, 10 times out of 10.

20:39Right. I mean, everyone, like you have that friend who goes to Mexico City fairly often and has like the Google Doc of their own personal recommendations of where to go. And that's an example of human tastemaking. And that doesn't feel pretentious or browbeating or something. It's just like, here's what this one person liked in this place. And maybe if you follow some of these suggestions, you'll find some cool stuff, you'll find more stuff, you'll figure out your own opinion of these things. And I think, I mean, you see different modes of this happening online. now too. I mean, Substack and email newsletters are a great bottom-up grassroots kind of way of spreading ideas and talking about stuff.

21:26And I mean, at least so far, email is relatively unsorted. I still get a lot of a fairly good linear feed of the stuff I want to see in my Gmail inbox. Yeah. Although one thing that you wrote about really resonated with me when it comes to sending email, you write about this thing called, maybe we can go deeper into it, called algorithmic anxiety, where anytime maybe you try to communicate with someone or is it putting a piece of content out, you sort of feel some anxiety about how the algorithm is interpreting it and whether you did everything the algorithm asked for. And then I was just like, well, you know what?

22:02I'm not subject to the algorithms because I do an email newsletter. And then I'm like, no, actually I have the same exact thing. Have I been filtered into spam and my updates versus the primary inbox? I do have algorithmic anxiety. So can you talk a little bit about what that is as a as a thing? Yeah, it was this really interesting term that was only coined in 2018, surprisingly, by this academic Shagun Javer. And he was kind of embedded at Airbnb as a sociologist. and he took on this project of interviewing Airbnb hosts about how they felt about Airbnb search engine, the rankings of their listings.

22:42And what he found was that the hosts basically had no idea how the platform was working. They had all of these kind of folktale methods of gaming the search engine and trying to provoke the algorithm into paying attention to them. Like they would change their calendar a bunch of times. They would log onto the website more times a day, like they felt like these little tricks were giving them an edge when in reality, it probably was not those variables. So I think the origin of that algorithmic anxiety is like the mismatch between how the algorithmic feed or recommendations affects us and how little we understand or can control them.

23:24Like in the context of a newsletter, you are wondering where your email is getting sorted into and if you're actually reaching your readers. I mean, when I put out a tweet or an ex post or whatever we're calling them, I worry, like I think through, okay, how can I get the algorithmic feeds attention? What can I do to sustain it? Am I doing the right thing? And that's like a source of anxiety, in part because I measured the success of the tweet based on likes, which is my own problem. I think everybody on Twitter does that. It's the scorekeeping. Yeah, exactly. Okay, so let me ask you this before we go to break.

24:07When people write a book like this, they have a goal in mind sometimes. Like you want to write a book and you want it to do something. I mean, obviously you want it to sell. But beyond that, like maybe it could have an impact. So is your intention here, is it to get the builders of tech to like remove algorithms, to get the users of tech to think about their relationship with algorithms in a different way? Like, what are you trying to do here? I was really trying to provoke this conversation. Okay, here we are. Which is like, I think a lot of users are not aware of just how much algorithmic recommendations shape what they experience online.

24:46And we all underestimate how much they influence all of the culture that's around us. So I wanted to kind of shake, like grab you by the shoulders and shake you and be like, look at how these are influencing everything at once. And I wanted to like not to be so hopeless as all that and like offer some optimism, both in the form of like showing how regulation is happening in the EU that is mandating algorithmic transparency and other things and how new platforms are working online. And there are like digital alternatives to these consumption habits that we've developed. what would you say to builders of products that i mean we have a lot of builders listening here they build they use algorithms if they hit the brakes or i think i mean for me as a user of this technology myself like i've experienced all of these feeds i've been as online as anyone over the past decade i think we've reached a tipping point People are tired of just how algorithmic these feeds have become, the kind of proportion of recommendations to things that you've chosen to follow.

26:02And I think users are thirsty for some sense of human to human connection that has become or gone missing in some of these larger platforms. And I mean, that used to happen on Facebook and Instagram. It can happen again. And on TikTok, there are kind of dregs of human interaction when you like a comment or something. But ultimately, I think I want to see more apps and more experiences that are about cultivating one person, communicating with one person, users' ability to follow a single voice in a sustainable way. I'm just like a... I think I want tech builders to have more of a sense of culture that's not just about popularity.

26:47that's not just about how many faves or how much engagement something gets. I think obscurity is an important value just as much as popularity is. Wow, yeah. I guess the question is, how would you code that into an algorithm? That is a challenge. That is a challenge. Yes, it is. You need a surprise knob. Maybe there's two sides of the same coin. Yeah, maybe you do turn up the surprise knob or the other thing that you could do is potentially allow for that more human input for a broader signal. like maybe, I guess the follow signal has gone away for a reason, right? It just wasn't strong enough, but maybe finding ways to have people tune their experience, their experiences could be good.

27:27All right. We're here with Kyle Chayka. He's the author of Filter World, How Algorithms Flatten Culture. I've gotten a chance to read it, enjoy it. In the second half, we're going to talk a little bit. So we've talked a little bit about online. In the second half, we're going to talk a little bit about how algorithms shape offline experiences. So plenty more coming up right after this. Did you know your credit card points and miles can lose value to inflation? Credit card companies often reduce the redemption value of your points and miles. Now imagine a credit card with rewards that can grow in value.

27:57With the Gemini credit card, you can earn Bitcoin or one of over 50 other cryptos instantly with no annual fee. Every swipe at the store or gas pump earns you instant rewards deposited straight to your account. Plus sign up now for a$200 Bitcoin bonus to kickstart your rewards. Visit Gemini.com slash card today. Check out the link in the description for more information on rates. Again, if you're looking to invest in Bitcoin but don't know where to start, the Gemini credit card makes it easy. The Gemini credit card is issued by WebBank. In order to qualify for the$200 crypto intro bonus, you must spend$3 ,000 in your first 90 days.

28:37Some exclusions apply to instant rewards in which rewards are deposited when the transaction posts. This content is not investment advice and trading crypto involves risk. The Gemini credit card cannot be used to make gambling related purchases.

29:11and wherever you find your podcasts.

29:25And we're back here on Big Technology Podcast with Kyle Jaca. He's the author of Filter World, How Algorithms Flatten Culture. He also writes for The New Yorker, writes about tech for The New Yorker, covers AI, a lot of social media, really good stuff. Okay, so another thing that I read in the book that I found to be pretty interesting is that algorithms are shaping the physical world, right? That I don't fully know exactly how it works, but maybe things that do really well on Google Maps have a similar type of feel and vibe to them, and therefore they beget other physical places that look exactly the same.

30:01Can you talk a little bit more about that? That's really interesting to me. For sure. I mean, I think we all feel how algorithmic recommendations shape the physical world. Like one example in the book is that Google Maps, for example, when passing by my wife's hometown of Westport, Connecticut, if the highway is too busy leaving New York, Google Maps will route all of the traffic through downtown Westport, which is a tiny town. It cannot handle many cars. There are many traffic lights. But it completely warps the quality of this downtown from being something that's for locals, like relatively slow to this massive line of cars that's just circumventing I-95 or whatever.