The Atlantic’s CEO on AI, Billionaire Owners, and Moonlighting as a Creator

19:09Nicholas Thompson:Make sure to check out Bird's Eye View on YouTube and wherever you get your podcasts.

19:39Nicholas Thompson:is increasingly showing less loyalty to traditional political parties, many now more likely to identify as independent. So what is going on with the kids? I think the biggest misconception about Gen Z's politics right now is that all of a sudden they're all socialists. That is just not the case. They are embracing candidates who are offering new, bold ideas in the absence of those ideas from establishment Democrats. This week on America Actually, Gen Z researcher Rachel Jambaza joins us to separate Gen Z fact versus fiction. It's not rocket science. And this is, you know, I keep saying like, young voters aren't that complicated after all.

20:14It's pretty simple.

20:15Nicholas Thompson:Catch us every Saturday on YouTube or wherever you get your podcasts.

20:24And we're back. Now it's the AI time to talk about AI. You guys were one of the first publishers to sign a deal with OpenAI. And there was a lot of hand-wringing about whether this was a good idea or a bad idea. It's what, a couple of years into it now?

20:38Nicholas Thompson:A couple of years, yeah. So how's that going? They're paying you for access to your content. Right. So I prefer them paying us than not paying us, right? And many of the AI companies don't pay. They have, you know, we set up a whole bunch of boundaries in the contract, right? How much can you summarize? Can you quote an author? Can you try to get your model to write like an author, right? Like these things that we care about because we don't want AI to, we want it to be harder for AI to replace what we do. They've been great on that. We wanted to have a seat at the table as they talk through the role of journalism in their models.

21:11Nicholas Thompson:Absolutely. They've been amazing partners at that. That's been incredibly important. The one thing that we haven't seen, and I think it's been a surprise to both parties, one reason to do the deal was my expectation, and I said this publicly many times, that AI was going to replace traditional search and that we wanted to make sure that there was lots of traffic directed back to us. When I go to ChatGPT or Claude, they're going to be my new search engine. Yep. And there are links in ChatGPT, not in Claude. You probably don't click on a lot of Atlantic links, and we don't see a lot of traffic. That is what we hear throughout the industry.

21:50Nicholas Thompson:And that has surprised me. Now, is that a fault of our deal? No, of course not. But it has surprised me from when we signed the deal. Do you think that's about the way they've designed the product? or simply just the fact that the product gives people what they want, which is information, and in most cases they're not interested to click through? Or is it something where if the links were bigger or more prominent or there was some incentive to click through, you'd see the rates go up? You certainly could change it in ways that would make the rates go up, but the core problem is that people go to it for a different use case.

22:21Nicholas Thompson:The Google use case was people were just trained to go, oh, I want to find this thing. Okay, now I'll leave and I'll go and I'll click to a site. And with Claw, ChatGPT, no, you're in a conversation with the bot. And Google, meanwhile, has been sending less and less referral traffic at the same time. So I hear. Yeah. Can I say something interesting about that, though? The most interesting data point to me, and I don't know whether this will hold, is that the number of our Google traffic, like everybody else's and everybody else you've interviewed, is going down, right? Because you ask a question and Google answers it instead of sending you a blue link.

22:55Nicholas Thompson:the number of subscribers we get from Google traffic has gone up. Now, mysterious, right? And what's clearly happening is that Google is sending fewer people who aren't going to subscribe. They're sending you more qualified leads. Qualified leads. Now, are they doing that deliberately? Is it because of our content mix? Is it because something of the way that we've tagged our stories? Like, there are all kinds of different hypotheses. I have one, which seems super basic, right? Which is anyone who now has worked hard enough to get to you from Google or anywhere else must really want to consume your content.

23:28It's taking it seriously. They're not just looking to see what time the Super Bowl is on.

23:32Nicholas Thompson:But it's not that the numbers are flat or that they're declining. It's that they're actually up, right? And so, yes, that is a huge part of it, right? They are going in and they, like, really want to read the Atlantic and they have lots of other options and they've chosen the Atlantic. And maybe they feel a little bit of affinity since they've gone through, you know, the various different Google options to get there. But it is quite interesting. Now, I don't know if that will hold, right? It might not hold, right? And, you know, as more and more traffic goes into Google's AI answers and less is directed externally, maybe this will be like a little blip where Nick is, you know, content with what's going on.

24:06What would you need to either continue to work with them when this deal re-ups or to say, no, we tried it and we got some benefit, but there's no long term. Yeah.

24:15Nicholas Thompson:Yeah, there just has to be a fair exchange of value. So what are you giving us and what are we giving you? And like, is it fair? And if it's not, well, then we'll block you from the site or we'll sue you, right? I mean, that's the deal. So they would have to give us something of value and you can, you know, there are lots of things they can give you a value, right? They can like give you the old deal. They gave us dollars, which are really nice. They gave us credits, which like weren't that useful at the beginning. And then we're incredibly useful, right? Like, Codex credits are awesome. And so, you know, you figure out what are they giving you a value and what are you giving them and is it a fair exchange?

24:54The two obvious critiques of all these deals were you are training your replacement. You are teaching ChatGPT how to replace you. It seems like a logical product these LLMs are going to make is a news product. Right. Whether they call it a news product or not, what do I need to know today? They'll go get it. it'll make it even less likely that people will end up on an Atlantic or a New York Times because they're going to get a news delivered to them and they won't have any desire or need to go out somewhere else. Right.

25:23Nicholas Thompson:So those are two. On the first one, I will say that yes, that is of course a concern. And that's why we put in specific clause about like mimicking writers replacing them. I'll also say that let's say you wanted to create an AI bot to replace Nick Thompson, right? Some of like my stories are in the Atlantic or on Wired and having a licensing agreement with the Atlantic or Wired would make it easier to create a replacement Nick Thompson. But the vast majority is, there's screenshots of my stories. There's like Reddit excerpts of my stories. There's books, there's excerpts from the books, there's reviews of the books that then license it.

26:01Nicholas Thompson:There've been stories that I've published in publications that don't have licensing deals that then syndicated it to Yahoo. And then it's been scraped from Yahoo, right? Like Like there have been stories that have been screenshot and put into Instapaper and archive, right? So like if you opt out of the contract, you aren't actually slowing down that process. The train is going. The train is going. So yes, it's important. It's important on principle, but the train is going. Now, on the second question, will it make it more likely they create a news replacement product? My view is that if you work with the companies and you have close relationships with the companies, you are likely to get a better news product than if you don't, right?

26:38Nicholas Thompson:And if the whole publishing industry says, no, you know what, we're going to fight you, I don't know if that actually leads to them either being less likely to make a competitive product or making a less harmful competitive product. I think there's real value to talking to these people. But there's also this other interesting, there's also like an interesting ethical component, right? Which is if you believe, as I do, that AI is going to become kind of like the substrate of future knowledge, I would like the Atlantic to be in it, right? Now, I'd like other things more, right? I'd like us to have a future business.

27:14Nicholas Thompson:I care deeply about that, right? I'd like our writers to be able to make a living. But that is an interesting – like there's an ethical component on the plus side of this very complicated scale that you weigh with every company. Beyond the mechanics of your deal with OpenAI and how publishers are going to work with them, what is AI going to do – let's put the positive frame on it. What can or is AI doing for The Atlantic? What could they do for media beyond better Google? So much, right? I mean, so AI is unquestionably the scariest thing, right? And it's terrifying and it could completely wipe us up.

27:53Nicholas Thompson:No question about it. On the other hand, it's also awesome, right? So what did I do today? Well, first thing I did is I went into my bot with Glean, which is a company we use which sort of adjudicates AI use and allows us to sort of partition access and make sure that different people are able to access the right data sets. It's like an interface between the bots and your data, right? And I went into Glean and I ask, it's the same thing I ask every day, which is, okay, I have an agent and a bot. Identify everything that's happening in our product and engineering team and identify everything that needs a CEO level decision, right?

28:26Nicholas Thompson:I ran another proprietary bot that I can't tell you about that looks for specific errors in a specific section and finds them in a really interesting way that allows me to help correct things that have gone wrong, right? So AI is helping me run the business better, both like specifically and – That sounds pretty technical and not even something that maybe you as CEO should be dealing with. I mean you like it, right? You're technically minded. You get a kick out of it. I don't just get a kick out of it. So my view – so I'm running like agent farms constantly, right, to like experiment with things and build things.

28:58Nicholas Thompson:Yeah, you're that kind of guy. Right? But because I like it and because it helps me figure out where – it gives me a hypothesis on where things are going. So we made a deal with Parallel, right? So Parallel is this middleman, this company run by Parag Khanna, who used to be the CEO of Twitter, that sort of serves data to agents. And so if you want to make a PowerPoint, maybe you call up Parallel and you say, hey, Parallel, I want to do a PowerPoint on the different flavored drinks that we have on this table here and analyzing, I don't know, taste profiles. So Parallel goes out, searches the web, agents go out, take the information, feed it back to you.

29:35So they have a really interesting system where they use what's called Shapley math.

29:39Nicholas Thompson:They analyze like what sources contributed to your little, you know. Right, and you're supposed to get rewarded based on usage. You get rewarded with micropayments. We like got excited about this because I was running all these agent farms and had realized like, I don't know, six or eight months ago that like the vast majority of traffic coming out of my computer was no longer human. It was my agent's running request. And I was like, God, we need to build an Atlantic business model for that. That led us to parallel, right? So I believe in experimenting with AI, both because it's interesting. I think it can make me more efficient and because I think it can help us make better business decisions.

30:11Nicholas Thompson:So back to what AI can do for a media company. A, like every media company, it can make you be more efficient, right? Like you can go through everything during your day. And a lot of it you can do better with AI. and you should do all that and you should figure out the right ways to do it and you gotta make sure that you don't feed all your secrets to open AI and Claude, but be smart about it and do that. There are all kinds of business opportunities. You can be smart and you can get ahead and you can partner with a company like Parallel and get preferential business terms and be early on in ways that are very valuable to you.

30:47Nicholas Thompson:I'm not a reporter. If I were a reporter, I would set up all kinds of interesting agent farms to go and like track down different mysteries. When I wrote my book, right? So I wrote a book about running. I used AI never to write. I think it's unethical to use it to write. I think it's dumb, right? You need to be worried about cognitive offloading. But it's an amazing editor, right? And you can use it in all kinds of interesting ways. Not as good as human editor, but you can use it in all kinds of interesting ways to improve what you write and how you think. So I'm all in on that stuff. I suspect that legacy organizations over the next three to five years will transform in ways that are interesting, and there'll be new AI startups.

31:28Nicholas Thompson:The other reason, of course, for our business is that people don't like AI, and they don't like AI for lots of good reasons. And so that also limits what you can do. When you say people don't like it, are you talking about your audience or are you talking about the people who work for you? Well, both, but no one wants to get news from an AI, which is good for us. No one wants to be told that they're getting news from an AI. Correct, right? If I put a story out, if somebody puts a story, you start a new publication, right? Kafka News, right? And you say, this is written by Peter Kafka. Everyone's like, yeah.

31:59Nicholas Thompson:And then you're like, oh, but also like half the story is written by AI. Like everybody's going to cancel. Like nobody wants it. We don't like, the world doesn't like AI. Do you think that's a rational response? Yeah, I actually kind of do. I do. I think that the amount, like I've said AI is awesome. I've said I use AI all the time. I talk about AI all the time. The amount that it's going to harm our world is massive, and I totally understand the backlash against data centers. I think it would be better for America if there are more data centers here because you get more control over AI. But I can see why people don't like it.

32:37It's funny. There's a data center backlash, but there's also people who get really angry when they see the local deli use what's obviously an AI thing to create terrible pictures of a panini that they're not going to make. And I'm like, that's just a guy running a deli. He spent two seconds thinking about the image. You got to just lay off there. Yeah, and it depends.

32:57Nicholas Thompson:So part of the reason is it used to be sometimes it's bad panini images, but when it's really good panini images, I think we'll still be mad if we know it's AI. Yeah, I think it's when you're like, that is not a panini. That seems to be a piece of plastic and another piece of plastic. I mean, there was an interesting study that came out a couple months ago, and it was about advertising. And you basically take a product, a generic product, and create ads for it. and you then show them to consumers. And the one they like the most is the one made with AI, right? Even more than the one made with humans working with AI, and certainly more than the one made just by humans.

33:25Nicholas Thompson:But then you label them properly, and the thing they like is the one made by humans, right? And so that's interesting. It creates an incentive for deception, right? Particularly when AI gets really good. So one difference I've had with a lot of people who critique AI and don't like AI is that there's a line of critique of AI that it's not good and won't be good, right? That it's like fundamentally flawed, that because the way it's made, you'll always have these horrible hallucinations. It's a fancy autocomplete. It's a fancy autocomplete. It's a statistic parrot, all of that stuff. And that was true for a blip in time.

34:00Nicholas Thompson:It's just going to get better and better and better and better and better. And so that Panini image that you can make fun of because it looks disgusting, it's actually going to be an amazing Panini image. And the way you can identify AI writing on LinkedIn right now, it's super obvious because everyone writes the same way, they'll correct for that. They'll correct for that. And you will no longer have that like three-part structure. Yeah. Another big theme I hear when I talk to people who do what you do, they say, we're very interested in the creator economy and we're trying to figure out some way to involve creators with our business, which is this old standard way of doing business.

34:32And we want to figure out how to work with these other people who have audiences doing interesting things. You did a version of this. I think I wrote about it. I know I wrote about it. I think I would have broken the news. Like in 2021, you were hiring people to write newsletters, and that was really a version of what people are talking about today. That didn't work. Well. Or you stopped doing it. We did stop doing it. So why did you stop doing it, and how is that informing how you're thinking about that question today?

34:58Nicholas Thompson:Yeah. So what we did is we started what we call the subscriber newsletter program. We were obviously Substack was set a – it's always been a good moment. The early days of Substack excitement. Yeah, and there's lots of, you know, there were a lot of, we needed to make sure that we, I wanted to and we wanted to on the business side and on the editorial side, make sure that we had a mechanism to retain the kind of writer who might leave for Substack and to attract the kind of writer who might go to Substack from another publication. And so what can Substack offer you? Well, it gives you complete freedom.

35:31Nicholas Thompson:You can publish what you want, when you want. All the money. All the money, right? So what if we give you some freedom and some money, right? Like you still get Atlantic editing. You still have to like comply with Atlantic standards. You can't have total freedom. And we'll give you a guaranteed salary plus upside as opposed to only upside. Because the problem with Substack is, you know, it's all or nothing, right? You don't sell subscriptions. You get nothing. And so we're like, maybe we can find a middle zone. And so when we hired these writers, we created – you get a base salary that's less than if you were a standard Atlantic staff writer.

36:03Nicholas Thompson:But you have upside that allows you to get to more. and we all share in success we're all aligned and you get Atlantic editing and then it just maybe we didn't set the numbers right maybe the writers we recruited in weren't right for that model but what's nice is that we brought in a bunch of really interesting people many of whom became Atlanta regular Atlantic staff writers once we realized the economics didn't work you know Charlie Wurzel right came in that way Tom Nichols right like came in that way all these like wonderful people came in that way. And so the program didn't work, but am I glad we did it?

36:38Nicholas Thompson:Yes, absolutely. I'll be right back, but first a word from a sponsor.

36:44Do you hear that?

36:45Nicholas Thompson:That sound, right? That means that summer's officially here. It means that grown adults just sprint into the street for a frozen dessert shaped like a cartoon. But this summer, Mint Mobile has a better treat. Every plan, including unlimited, is$15 a month. And unlike ice cream, it won't drip down your wrist or look nothing like the picture. Does anyone have any cash? Give it a try at mintmobile.com slash switch. Upfront payment of$45 for three months, $90 for six months, or$180 for 12-month plan required, $15 per month equivalent, taxes and fees extra. New customer offer for initial plan term only greater than 50 gigabytes may slow when network is busy.

37:13See terms. Every living cell in every living thing is related. We're all part of the same tree of life. Now, scientists are trying to make a new tree of life, building their own cells out of non-living chemical parts. And hope that it comes to life. To create new technologies that will help our economy, that will help the climate. But they have to navigate the risks. There is the potential in what we do to be mad scientists.

37:46Nicholas Thompson:If you have a cell that can be engineered to do whatever you want, you could make something harmful. Suddenly, we're talking about something which could be a catastrophic threat to life on this planet. So do you fear us more than the people who made nuclear weapons? Life from Scratch is a new three-part series from Unexplainable. Listen wherever you get your podcasts or watch us on Netflix. You've DM'd me at 2 a.m. asking if you should panic about your 401k. You've paid an advisor$1 ,000 for one question. You've Googled, what is a rollover, for the 10th time and still aren't sure. On this week's episode of Net Worth and Chill, I'm revealing my biggest secret.

38:31I finally built something to help address your finance woes. It's called Ask Dolly, short for dollar, and it's about to become your newest financial companion. Ask Dolly diagnoses your financial strengths and weaknesses, answers your toughest money questions, and makes economic news headlines actually relatable and connects you to real human certified financial planners so anyone can make the right financial move regardless of their bank balance. No jargon, no judgment. No, you don't have enough assets for us. I'm breaking down exactly how Ask Dolly works and why I built her on this week's episode of Net Worth and Chill.

39:06You're not gonna wanna miss this. Listen wherever you get your podcasts or on youtube.com slash yourrichbff.

39:19And we're back. It's another, I mean, when I ask publishers about this and think about it myself, I understand why you'd want to hire people and why people would want to go work for you or another publication. Not everyone wants to be on their own. Most people don't. Yeah. I struggle with if you are successful in some way on your own, Substack, whatever it is, what would you get out of aligning with an Atlantic? Some money, you'll say you're getting more exposure to a different audience, but it just seems like they're a freelance contributor, and that's a pretty old model, and it's not that big of a deal.

39:54Nicholas Thompson:Well, it depends, right? So, you know, if somebody has like 20 ,000 paid subscribers on Substack, right, so they're getting, what would that be? A lot of money. That's a lot of money, right? It's$2.5 million a year. They're not coming to us. They're not going to come take a job as an Atlantic staff writer. But let's say they've got, I don't know, 10 ,000 unpaid subscribers and like 500 ,000 of them are paid. Well, sorry, 500 of them are paid. Well, maybe then the economics of having an Atlantic salary, you know, does make sense. And then remember the other things you get when you join the Atlantic.

40:26Nicholas Thompson:Like some people don't like having editors. I love being edited, right? You know, when I wrote my book, I like sent it to so many people, went through so many drafts, so much great feedback. you know editors are amazing fact checkers amazing copy editors amazing lawyers amazing right I love some people some people say all these people are preventing me from doing what I want which is to publish this thing yeah and at the Atlantic no they're not they're helping you you know barely slowing you down and they're making sure it's accurate but yes some people don't want to be part of that. So you can imagine somebody who really, and also we promote through the social media channels of the Atlantic, which are not insignificant.

41:12Nicholas Thompson:And we give you a desk with air conditioning, right? And so there are a lot of advantages to coming to work at a place like the Atlantic, even if you want to be a solo creator. I also, this is, okay, so this, I may turn this around and ask you. I worry about creators being out-competed by AI because there may be an AI version of the Atlantic that competes with us, right? There are 100 % going to be AI creators, right? Particularly video creators. And I worry about this, you know, so I do a daily video on LinkedIn. I've done it for eight years. I talk about tech policy. Like how long until there's someone who can - Someone makes a version of you that is good enough.

41:54Nicholas Thompson:It's like it's not even a version of me. It's not a fake Nick Thompson. It's just like someone like you. Smarter, more handsome, like not hard with like basic video editing tools to do that. Today it wouldn't work, but we're assuming that the next number of months or years it will. Right. And days maybe at the rate AI is improving. And so I actually think that the solo – if I were a solo creator, the only way I made money right now – So I have this like tiny little Nick Thompson creator wing, which I've run for a long time because it's a good way to learn about this stuff. I'd be terrified, right, about the coming competition and like what it's going to be like.

42:32Nicholas Thompson:LinkedIn, you might have a space because LinkedIn has a real identity layer. Yeah, I just think that applies to everyone who makes content. And what you're kind of saying is the creators are easier to replace than the Atlantics of the world. They might be. What I fear is that the tide is going to come out and we're all going to find that collectively the stuff we make that we think is great is way less valuable. And all the people who say, but our brand means a lot and people will come to us, that will be true for a very small – this is my fear – a very small slice of people. And the rest of it is like, no, it's fine.

43:08I used to subscribe to The Atlantic, but I get enough stuff now. I'm fine. I'm not missing it.

43:12Nicholas Thompson:It might be lights out everywhere, but A, I don't think so. And B, I do think that there are advantages that brands like The Atlantic have or brands like Peter Kafka. Also, it's interesting. If you think about audio, one-on-one conversations, no one wants to hear you interview an AI or me be interviewed by an AI. I keep reading and hearing about successful podcasters who are making AI versions of themselves. And I keep thinking, I must be missing something. There's no way people actually want to consume that, but maybe I'm just delusional. No, I don't think they do. I do think that I think AI will obviously will compete for time.

43:51Nicholas Thompson:One thing I could have done is like taken the transcripts of your previous, I don't know, 10 interviews, uploaded them to Notebook LM and said, hey, give me a conversation that will help me understand what Kafka is going to ask me. Weirdly, podcast transcripts, at least through ChatGPT, total blind spot right now. I cannot, unless maybe you can tell me afterwards, but I cannot get chat to do that exact thing. Well, there are different, well, A, because there's some licensing restrictions. That's what I assume. And it's like sometimes hard to extract them and they don't want to get sued. Particle has just created a product that does that.

44:20So, yeah. We'll talk more about it. You're talking about all the layers of support you get when you work with or for the Atlantic. You guys ran this sensational story about Kash Patel this spring that describes him essentially as incompetent, paranoid, drunk. And not surprisingly, when you publish that, he threatened to sue you and has now sued you. When a story like that is coming up, and again, that's all Jeff Goldberg and his team, is there a point in which they go, hey, we are going to publish this thing that is going to bring a lot of attention and maybe legal risk for us? Do they float that to you at some point?

44:56Do they give you a heads up? I can tell you exactly.

45:00Nicholas Thompson:I got a heads up while I was running home about five minutes before it published. So I don't know. Yeah, no. So fully, like we just published a thing. We're about to publish a thing in five minutes that will likely result in the FBI director suing us. Yeah, but, you know, I completely trust Jeff Goldberg and our legal team and all the writers he hires and the editors he has and the fact checkers we have. And of course. And then do you have to tell LaRene Powell Jobs, hey, guess what we just did? Well, there'll be board conversations about it, but she doesn't need to know in advance either. You know, there can be situations that are, you know, if there's a story that could lead to increased hacking threats, right, or they could lead to an office raid, right, then you have like maybe a little more conversation.

45:47Nicholas Thompson:But I never know what the stories are in advance. And you don't think you don't think a highly critical piece about a very paranoid FBI director couldn't lead to something like an office, right? It didn't. Yeah. I mean, look, the story. Well, it's a very it's a very well reported story with a lot of great editing and, you know, very good legal review. You mentioned the fact we both mentioned the fact that you do your own podcast. Do you make your own LinkedIn videos? You say you have one point six million followers on LinkedIn. I see you on TikTok because that's where my brain is at. Yeah, I just started on TikTok.

46:21You produce these. They are made by you, CEO of The Atlantic. There are sponsors attached to them many times. They are not editorial products. Jeffrey Goldberg had nothing to do with them. Have you guys had a talk about, well, we have some kind of Atlantic content here that Jeffrey Goldberg makes, and there's other Atlantic content that the CEO makes. In our minds, they're separate. But if you're a consumer, it's all just stuff coming from the Atlantic.

46:48Nicholas Thompson:Well, the stuff the CEO makes is labeled, and it says this has not come from the Atlantic's editorial arm, right? I just wonder if that means any – one, if that means anything to a normal person, and two, how you guys think about that. I mean, we make a lot of money on the stuff I do. And, you know, Jeff likes when we make money because you know what we do with the money we make? We hire more reporters. So it does not go into your pocket when you have a – You know what Jeff likes? He likes hiring more reporters. So it's different, right? I don't know if there are other – there's some CEOs who do similar things, but there's nothing quite like this thing that I have where I have a – it's very small.

47:22I mean, usually when you hear about a CEO publishing something, it's like, hey, the editor of – the owner of the LA Times is running his own op-eds on the homepage. And everyone goes, that's pretty gross. We don't like that. That's why you shouldn't do it. We also tolerate it because he owns it, and so that's his right to do it. But it's not against the law. But it's also a little creepy. Or when A.G. Salzberger wants to write a big column, he doesn't publish it in the New York Times. He publishes it in another – in part because he doesn't want to see – he doesn't want the Times to be viewed as his personal sub box.

47:57Nicholas Thompson:Yeah. I mean, in thinking he's writing – I think that's right. I think that's the right thing to do. So we keep a line between what I do and the editorial side. It's all very clear what is mine. and because neither side knows what the other is doing, there's not a, it's not like, I just interviewed Richard Socher, which I can say because the podcast will air when this podcast comes out, right? You know, the editorial team of The Atlantic might be preparing a hit piece on Richard Socher. It might be preparing a puff profile. They have no idea that I just interviewed him and that it's going to air on Wednesday.

48:34Nicholas Thompson:And so it's totally independent. What's the biggest thing, most interesting thing you've learned about independent content creation being an independent content creator who was also a CEO? Oh yeah, this is, this is so interesting. Um, and this is partly why I started doing it because you learn, I mean, the first lesson I learned was

48:56Nicholas Thompson:different content. This is, it's an obvious lesson. And so I'm trying to figure out a way to say it that doesn't seem like I'm, like I'm was just born yesterday, but different content works really well in different platforms and really badly. Like when I started making those videos, I was like, wow, you know, I'm much bigger on Twitter than I am on LinkedIn. Like, they'll work on Twitter. Let's go, right? But you post one of these videos and I like break apart an AI paper on Twitter and A, people are scrolling too quick and B, they're too mean. So the comments aren't like, oh, that's interesting.

49:21Nicholas Thompson:I learned something about this math problem. They're like, you know, Nick, go eat a hamburger. Your cheeks look too skinny, right? And so, you know, and there's no - You're very gaunt. Yeah, right? Like, which is - Healthy, healthily gaunt. That's the look I go for, healthy gaunt. But anyway, the conversation on Twitter, like never worked. Sometimes I'll post them, but it didn't really work. They didn't really work on TikTok at first and it's now starting to do it. YouTube is now starting to work. But you learn a lot about how the different algorithms work. You learn about the feedback mechanisms.

49:51Nicholas Thompson:You learn about the conversations. Like a lot of how, one of the things I noticed was how much better the conversations were on LinkedIn. So, okay, so what about the conversational architecture of LinkedIn makes the conversations better? Well, okay, it's partly that it's tied to your real name and your real identity, right? So it's not tied to anonymity. Okay. So that's interesting conversational architecture question. And so you start to learn all these different things. And then you also learn a little bit about how to do it and the kind of like, okay, so what is it about my personality that helps?

50:21Nicholas Thompson:What is it about my personality that doesn't help? What about the way? You just, you have to learn by doing, right? You can't learn to drive a car by reading a book about how to drive a car. You have to actually drive it. The idea that different stuff works on different platforms we get. The thing I am always thinking about is how, you know, you can learn that. This is the way you do a TikTok video. This is the way you do a LinkedIn. And then LinkedIn can change their algorithm or what they want to do. And so you're ultimately always sort of working for the distributors. Yeah. And that's a good thing to learn viscerally because it feels different emotionally when you learn it.

50:53Yeah. And so I'm wondering how you think about that. Do you think that's the same as a publisher also has to do the same stuff because you guys are dependent on the chat GPTs and Google's the world? Or is there a fundamental difference of the way the Atlantic thinks about putting out its stuff versus Nick Thompson?

51:10Nicholas Thompson:Okay. It's a really interesting question. So I have very little say over any part of the Atlantic's. Obviously, as we talked about before, I don't know what they're going to publish. I don't know when they're going to publish. And I have very little say even over how we promote it. But I have a very strong philosophy from having been the editor of Wired and having worked at the New Yorker, which is what a publisher should do is to not think for one second about what the algorithms want, what they need, what's trending when you're creating the content. Yeah, I'm laughing because I have yet to work at a place like that in the internet era, but go on.

51:44Nicholas Thompson:So look, but I'm going to differ. I differ from people on all sides of this debate and it's a debate I've had a lot over the years. So not for one second, think about what's trending. You should not have a trending topics tag. You should not be looking at any of that. Then when the story is done and you've closed it and you've closed the zipper, you have to fight like hell on every platform to get readers. And the problem in a lot of publishers, the ones that don't look at the algorithms, also then don't fight like hell, right? And the ones who do fight like hell are optimizing from the beginning.

52:11Nicholas Thompson:So I have a very different strategy. And that's the way I think publishers make good stuff, then figure out how to distribute it. And like, which seems basic, But I think most of the world, most publishers today have been trained not to do that. What's the thing that you're going to do best at the next couple of years in your role? Where are you going to really excel? Where am I going to add? Yeah, you personally. What value are you going to add? I hope that I am able to help identify points where we have leverage to increase subscriptions. And so there are all these different, like, changes you can make.

52:45Nicholas Thompson:And I've said before we did hundreds of tests, but like whether you put down the paywall, where you put the paywall, like how you deal with paid marketing, right? So we have a new tool where we take stories that have crossed a threshold in subscription propensity. We then feed them into AI. We then feed that into Facebook. We then use Facebook's AI. We market it, right? We think we have a story that is going to convert, and then you've got a bunch of mechanisms to – No, no. We don't even think. Like the data tells us that this story, and it may be a story from 1976 that is converting today because who knows why, right?

53:20Nicholas Thompson:So an automated system. So put money and effort into this thing. Right. Figuring out tools like that that you can make 1%. If you can make like 1 % improvement in your dunning rate, the credit card failure rate, like you do all this stuff, you can make the brand really successful. So I want to do that. I want to identify where AI is changing our business at a core, and I want the Atlantic to be ahead of that. I think this deal with Parallel is pretty interesting. We knew the world is moving to a post-Google search era, so what do you do? Okay, we have a whole strategy, right? Both to get as much as we can out of it, also to migrate people to your app, more live events, all that stuff.

54:01Nicholas Thompson:Started from a while ago. We now know that most of the web is going to be agents, right? So, okay, so how do we make money in that? How are we going to make advertising revenue in that? I don't know that yet. How are we going to make subscription revenue? Okay, well, are micropayments going to be part of it? I want to get those big questions right. And so what will make like the next couple of years successful to me are if, you know, advertising continues to grow and then we make the big correct decisions on AI and then we continue to improve the sort of the consumer paywall business. Those are good plans.

54:27It has taken me eight plus years to have a second conversation with you. I think we should do one more frequently. I would love to do it more frequently. Thank you, Peter. Thanks again to Nick Thompson. Thanks again to Charlotte Silver, who produces and edits this show. Thanks to our sponsors who bring this show to you for free. Thanks to you guys listening. See you soon.

54:57Nicholas Thompson:Do you hear that? That sound, right? That means that summer's officially here. It means that grown adults just sprint into the street for a frozen dessert shaped like a cartoon. But this summer, Mint Mobile has a better treat. Every plan, including unlimited, is$15 a month. And unlike ice cream, it won't drip down your wrist or look nothing like the picture. Does anyone have any cash? Give it a try at mintmobile.com slash switch. Upfront payment of$45 for three months,$90 for six months, or$180 for 12-month plan required. $15 per month equivalent. Taxes and fees extra. New customer offer for initial plan term only.

55:24Greater than 50 gigabytes may slow when network is busy. See terms.