← The Monopoly Report
The Monopoly Report
Episode 76: John Battelle on AI, Data Ownership and the Future of the Open Internet
01:15:30
Loading summary
Transcript77 lines
- [00:01]
Alan Chappelle
This podcast is brought to you by the Build, a new podcast from the
- [00:05]
John Battelle
guys behind Sincera, Michael Sullivan and Ian Myers.
- [00:08]
Alan Chappelle
They built their company by figuring out clever solutions to a few important ad tech problems in our industry, and that's exactly what the show is about. Mike and Ian interview some of the
- [00:17]
John Battelle
smartest tech minds in the biz to hear about how they identified opportunities, solved
- [00:23]
Alan Chappelle
their hardest challenges, and grew their businesses in the process. Listen to the Build with Mike o' Sullivan wherever you get your podcasts. Welcome to the Monopoly Report the Monopoly Report is dedicated to analyzing the big issues impacting the global advertising economy. Alan I'm Alan Chappelle, outside counsel and fractional CPO to a bunch of tech companies and am the principal analyst at their Chappelle Regulatory Insider, which is a monthly report that focuses on privacy, antitrust and other regulations impacting digital media worldwide. You can find a link to a sample copy of the Chappelle Regulatory Insider in the show. Notes this week my guest is John Batap. John is a journalist, entrepreneur, author and professor. He co founded Wired in the early 1990s where he personally edited everyone from Bruce Sterling to William Gibson to John Perry Barlow. He went on to launch the Industry Standard, which chronicled the dot com era better than anyone, and then Federated Media, which was designed to build an ad business that respected independent publishers. John has written a best selling book, the Search How Google and Its Rivals Rewrote the Rules of Business and Transformed Our Culture. Today, John sits on the board of Live Ramp and Sovereign, teaches at Berkeley and runs doc, which is an intriguing org focused on truth in medicine, longevity and science. A quick aside on health I know it's not typically in my purview here, but a few years ago Jerry Harrison of Talking Heads fame got me involved in a company called Ofrix, which has developed a universal antidote to snake venoms. We are hoping that Ofyx can impact what is one of the most neglected global epidemics that currently impacts 5 million people each year. Anyway, Jerry shows up indirectly in this episode as he's the one who introduced me to John Perry Barlow. John Batal and I spend some time talking about Barlow on today's episode. I wanted John on the show for a bunch of reasons. Most of the people talking about AI today are either selling something or completely hating on it. John is way more cerebral in his approach, asking harder questions regarding whose interest does this architecture actually serve and how would we know if it served different ones? We talked about what John calls the inner monologue bloom, the fact that hundreds of millions of people are now uploading therapy grade disclosures into chatbots owned by a handful of companies and what that means for the concept of privacy. We talk about OpenAI's audacious projection of 100 billion in ad revenue by 2030 and whether that number is achievable, aspirational or simply made up. We talked about why publishers are in worse position than retailers, why Reddit selling its archive to Google looks more like a sugar hit than a strategy, and why John thinks the programmatic ad tech stack might be the most underappreciated and most underused piece of digital infrastructure ever built. We also talked about Apple blocking a piece of Claude generated code on John's own personal Mac, which led to a broader conversation about agency control and what John calls the Token Act, John's proposed federal law that he thinks could unlock more economic value than any antitrust remedy currently on the table. There's some mid-90s energy taking place in this episode, some genuine optimism, some warranted alarm, and a lot of specifics. So let's get to it. Hey, John, thanks for coming on the pod. How are you?
- [04:13]
John Battelle
I'm well, thanks. Thank you for having me, Alan.
- [04:15]
Alan Chappelle
And where is my audience finding you today?
- [04:18]
John Battelle
I live most of the time, I would say, what is that? About two thirds of the time I live on the island of Martha's Vineyard, which is off the coast of Massachusetts. I also have a small place in New York to stay connected to the rest of the world because the island is very much an island, particularly, you know, in the off season.
- [04:39]
Alan Chappelle
There's something really cool about having that, like, little getaway that you can, that you can jump to about it.
- [04:45]
John Battelle
It is for us, the getaway is New York. But when we lived in California and in New York, we, this was our getaway. But the pandemic sort of flipped that particular model on its ear and we decided to, you know, why don't we live here? Which are. The blood pressure goes down quite a bit.
- [05:04]
Alan Chappelle
So about 12 years ago I picked up a houseboat in Sausalito and actually we actually lived there for about two and a half years during the pandemic. Probably for the same reason that you got out of New York during the pandemic and that that has just been probably one of the best decisions I ever made is to pick that thing up because it's. That is an interesting community over there, my friend.
- [05:24]
John Battelle
Are you neighbors with Stuart Brand?
- [05:26]
Alan Chappelle
You know what, he and I have met. I don't know him well.
- [05:29]
John Battelle
Yeah, he was one of the original Brain Trust of the founding of. When we founded wired, Stewart was on the, what we called the Brain Trust. That's what we called our, I guess what you call it, advisory board now. But like back then we called it the Brain Trust. And we used to have these regular dinners and I had just finished reading his book on the Media Lab, which changed my life. And I couldn't believe I, you know, I got to edit Stuart Brand, you know, and William Gibson and Bruce Sterling, all these crazy people that I only read. And he told me that he lived on a houseboat, Sausalito.
- [06:02]
Alan Chappelle
I, I had a similar experience to what you're describing when I got to meet John Perry Barlo a number of years ago out through. Out there through some music friends of mine. And I ended up at some of this kind of intimate dinner at a friend's house at the top of Mount Tam. And Barlow was sort of recounting how he went from the Grateful Dead to cattle rancher to the eff and like, the two most important things of my life are music and privacy because that's how I built my career. And, boy, was that a absolutely fun dinner to be hanging out in.
- [06:32]
John Battelle
Yeah, well, Barlow was also in our brain trust and I edited everything he wrote, including a Declaration of Liberty in Cyberspace, that, you know, manifesto that he published, published it in Wired, but also on the EFF website. And I've spent a fair amount of time over the last year and will probably over the next few years, spent a fair amount of time tracing that somewhat naive and certainly optimistic brand of libertarianism that Barlow reflects. You know, he, he embodied it and Wired embodied it and how it mutated to what we have now, which is a, an aggressive, almost cancerous version of libertarianism that has taken over, you know, kind of the leading voices of the. Of Silicon Valley. You can directly trace it to the early 90s. Optimistic, you know, keep your hands off my Internet libertarianism that we espoused in the pages of Wired.
- [07:36]
Alan Chappelle
Yeah, I can remember. And I was, I was reading a lot of that stuff from the vantage point of Jupiter Research back in the, Back in the day, and I did not realize that you had edited that sort of makes sense. But I hadn't put the dots together
- [07:49]
John Battelle
that you were editing Wired from 1993 when we launched until I left in late 97. So there wasn't much that I didn't, you know, try to, you know, hit, in Barla's case, turn into prose because John wrote, you know, like a lyricist.
- [08:07]
Alan Chappelle
Yeah. So you had coined, I guess we're going back a while, but you'd coined the database of intentions back in 2003, and the goal there was to describe what Google was building from search queries. And you've since gone on to discuss AI conversations and how they represent something far more intimate. So not just breadcrumbs, but the full inner monologue. And so one way of framing this shift is, in my view, and I say this as the privacy guy, is that like exponentially more and more sensitive data seems to be now in the hand of a small group of companies. And I'm curious, how does that transition play out? Or how does it, what does it mean for the concept of data privacy?
- [08:52]
John Battelle
Well, I mean, this is, as we've probably discussed, you know, before we started this particular conversation, privacy is probably the only handle that, you know, your average Internet consumer can pick up the Internet buy and say, ooh, this is dangerous. You know what I mean? Like, privacy has been the one word that, you know, makes eyebrows go up for, you know, like my mother and my kids, right? Like, it's consistently the only thing that lawmakers, you know, use as a cudgel to talk about what's wrong with the Internet and so on and so forth. And what we've learned, and certainly what those in the valley have said, is that no one seems to care about it very much. Everyone says they care about it, but no one seems to actually act as if they care about it, right? So, you know, my kids don't like the idea that everything about them is known. And Instagram seems to know everything about them, but that doesn't stop them from using Instagram, right? So we've built over nearly four decades now a set of digital technologies that are essentially harming a fair definition of privacy, but we don't care. So what I call the inner monologue bloom, like, you know, the algae blooms that happen, you know, but there's this like the amount of information and the richness sort of, if you think about an algae bloom, like the richness of nutrients in an inner monologue. And the things that we are thinking about and talking to our computers about now that, that honestly have never been captured. It's an entirely new superset of, of digital data that, that literally didn't exist prior to AI. Now there were memoirs and I guess online digital journals, right? And, and small little point solution type versions of this data. But we're talking about an asset class at hundreds of millions of people scale that this simply didn't exist before. That is very similar to the search Query stream that bloomed in the early 2000s. This one has the potential to aggregate and inflame the current discussion around privacy. But I think that's not the right way to think about it. Throughout history, capitalist systems have looked for and enclosed free or nearly free inputs. So land, timber, coal, fossil fuels, labor, these are inputs that capitalism would prefer to have to be free. And if there are externalities in the gathering and processing of these inputs, like pollution or you know, early death for people who are being forced into indentured labor or slavery, those are externalized as cost to society. This is not a communist manifesto. This is simply just true. Oh, by the way, all of those inputs are what economists call rivalrous, right? So you know, you can use it, but once you use it, I can't use it, right? And so it's rivalous. Data is non rivalous as an asset. And you know, anyone like you or, or, or other thoughtful commentators on the Internet scene are well aware that data is non rivalous. As a matter of fact, the fact that it's non rivalous is why software companies have amazing gross margins, right? So we all know this, but we have built an architecture of a political economy that treats data as if it were rivalrous, which is very, very strange and wrong. But essentially the companies that are now the most powerful companies in the world, and certainly in the technology industry, are all enclosers of data and treat data as if it's a rivalry asset. Most of us as consumers don't think about this in these terms. Most folks don't even know what rivalrous means as a economic, you know, term of art. And that's fine and it didn't really matter. But this infrastructure that we've built, this architecture of control of data, you know, the metas and the Googles and the Amazons and the hundreds of thousands of companies that have a similar approach to meta. And Amazon and Google have built an infrastructure that treats data as rival risk. And it did not matter to us because we got such awesome things from that infrastructure, right? We got like one day, you know, turnaround on anything we ordered on Amazon, cut down to six hours, in some cases one hour in other cases, you know, we got instant information from Google, we got instant connection to our friends and, and, and to people we care about on Facebook. So we swapped that convenience for the enclosure of all of this insanely powerful and valuable new asset class that we generally understand to be data. And that was fine. We didn't think too much about that bargain, this new bloom that, that, that you raised the inner monologue Bloom has the potential to make us reconsider that bargain, I think. And it's going to take some kind of a, it's a coordination problem for you know, hundreds and hundreds of millions of people, billions soon who are using chat GPT or Claude or other, you know, similar kinds of chatbot surfaces have this ongoing inner monologue. I don't know about you, but I probably have like 10 different ones going on, on different topics. Right. So it's not just one, it's not just like, you know, my journal, you know, or for those of the listeners who might have read the recent novel Culpability, which is a, a piece of fiction that thinks through the implications of this. There's a character in there as a young teenager girl who, who's having this ongoing monologue with a chatbot. And it's part of the story. It's not just that. It's like, you know, your work, six different parts of your work, you know, six different parts of your life. My relationship with my mother, my relationship with, you know, my girlfriend or my wife or my, you know, significant other or 10 of them, who knows? It's a lot. And what I am concerned about is that the architecture, the built architecture that we currently have in technology is going to address this inner monologue bloom as just another data asset just like all the others. And that could create a significant tipping point in consumers understandings of this concept that you are so familiar with that we call privacy. Because it strikes me that it's like we're kind of watching a slow motion car crash here. OpenAI this week rolled out its ads manager. So it's self service platform, Right. And I applied immediately as an advertiser because I want to understand it, right? Not just as like I'm going to see ads and GPT, but I want to see what the other side looks like. Right. First of all, it was built by engineers who built Metaz ads products. So I think we can.
- [16:28]
Alan Chappelle
So it'll be easy to use.
- [16:29]
John Battelle
Yeah, it'll be very easy. And if they're going to get to 100 billion plus in ad revenue in four years, which would be, I don't know, that would be the most extraordinary thing ever done in history. Capitalism, probably. But if they do that, they will do that because they've recruited millions upon millions of advertisers just like search did, just like social did and because they have somehow figured out a way to not freak everyone out about these ads being integrated into these inner monologues. And I Don't know how the hell they're going to do that. I wish I had an answer for you, but you know, I don't know how they're going to do it.
- [17:17]
Alan Chappelle
I think it's kind of a made up number at this point. No, and because in order for that number to happen they would need to get, if you look at current like projections of worldwide ad spend, they would need to capture something like 4, 45% of the incremental growth over a two year period. And that would have to be at the expense of somebody, right?
- [17:42]
John Battelle
Well, it's got to be at the expense of somebody, right? I mean, and, and, and we all know who that somebody is. There's two somebodies, it's Meta and it's Google, you know, I mean all the rest are just bit players compared to that. And so they're looking at the, you know, probably saying a hundred billion doesn't sound crazy when these guys are doing a hundred billion a quarter right now. And so, you know, no big deal. And yeah, I think it is a made up number. But then again, as someone who's made up numbers for a living, I mean I've run eight different media companies, you make them up and then you hope you hit them. Right. But it's audacious. It's audacious. I think what has to happen for that to have any chance of coming to fruition is that billions of us have to trust our generative AI. And I can't see a way that that works in the current model. The current model being I'm swapping this lovely product and the convenience of this product and the value I get from this product for my data. Now that is the grand bargain has been since, you know, really the rise of advertising in the Internet and it's sort of genetic twin commerce, right? They're very related. But people didn't think about that bargain. They, they participated in it and felt fine, but they didn't. You know, every once in a while it's like, oh shit, Cambridge Analytica, you know, and everyone thinks about it for like 45 minutes and then, you know, that's it. Right? But this one, I think there's a higher possibility that other models will evolve that are preferential and, and just as good. And, and what I mean by that is if the current model, which it's not, if the current model is that all of my information that goes up into chat, GPT or Claude or you know, what have you, there's a lot of other large ones that are not in the United States. But if all of that information is essentially being treated the way that Facebook treats my information or Google treats my information or Amazon, they'll come a time when someone says to me, hey John, have you checked out X? Now that's a wrong, bad word. Why? Yeah, have you checked out why it does all the things that ChatGPT does, but you have complete control over it on your device. And not only that, it creates a, an asset that you can use to do other things. Like you give this asset to your e commerce agent and it goes off and finds you deals because it's been listening and it understands you and it knows what you want and it works only for you and it's completely controlled by you and you don't have to think hard about it. It kind of brings stuff like a dog fetching, you know, a paper and puts it, you know, your feet and says, are you interested? Right. You don't have to think, it's just right there for you. And I, when I started writing about this idea, I called these genies like I dream a genie.
- [20:51]
Alan Chappelle
Right? Right.
- [20:52]
John Battelle
And that the genie only works for one person, you. The genie doesn't work for, you know, State Farm or Google or Microsoft or Meta or chatgpt, it's, it works only for you. And that idea of a powerful technology that is bounded by my trust, to me, that I trust doesn't exist, isn't that strange? But it isn't extant in the world. And technically, from a pure like is this technically possible point of view? It should, it can, it could. Right? But, but the architecture that we built is centralized. The data goes up into the cloud and gets managed and processed and leveraged and exploited and enclosed by the current techno capitalist system we have. But wouldn't it be cool if because data is non rivalrous, you can still send it all up there if you want, fine, but you get to keep a copy and do whatever makes sense for you to do. Now most people don't want to think at all about any of the shit I'm talking about. They don't. That's fine. Some entrepreneur or thousands of them certainly see this opportunity. And I have been in touch with many of them because I write about this stuff and they reach out to me, go, I'm doing this right? And then I look at their stuff and I'm like, this is awesome, but you've got to boil like the seven Cs before this works, right? Coordination, all these different problems, right.
- [22:27]
Alan Chappelle
But how do you get consumers to care? Isn't that like, I look at a lot of those. You know, there's every once in a while somebody comes up to me and is like, I've got this great tool, it lets you control all the data so that you can then tell Company X that I want to use it. But, and some of them are kind of interesting, but I haven't seen one of them to get beyond the like 1/10 the 1 percentile of users.
- [22:50]
John Battelle
No, it's like, it's always a, like a Firefox extension. Right. It's like no one's going to change their browser and then start using extensions. And by the way, everyone's on their phone. What are you talking about? Right? So that is the problem. And look, I joined the board of a company at the time called Axiom and now called liveramp in 2012 maybe. And at the time the board was having an interesting debate about all this data that Axiom owned. And they had built this, as you well know. They had built this, you know, they were at the time a data broker and they built off the census data, this incredible sort of stockpile of data about every American in the, in the United States and, and a lot of others. And the instinct was, let's give it back to everybody. Let's let them see what, what we know about them. And in a perfect world people will care and they'll edit it and make it better and then we can become sort of a platform to enable a new economy of data sharing that's driven by personal wants and needs. Essentially the database of intentions. Right. And they had me on the board because I wrote that book, because I've been writing about that stuff. I got very excited and they wrote, they actually built this site called about the Data where you go and log in and see everything that the ad tech ecosystem knew about you. Which of course was at least half wrong, but good enough. Right.
- [24:20]
Alan Chappelle
That was sort of the problem when I, when I built, when I helped Omar Tawako build a similar thing at Blue Kai back in the day, it was such a neat thing. I wish more ad techs would, would build that kind of technology, would simplify this whole DSR nightmare that we're going through in California. But at the end of the day, you know, the, when they think that, you know, John is a, you know, 27 year old female. Right. It kind of blows up the whole thing.
- [24:45]
John Battelle
Yeah. And Omar was on the board at Libram for a while. He's awesome.
- [24:49]
Alan Chappelle
Right.
- [24:49]
John Battelle
So I got excited about it. But then I remember whiteboarding this with Like, I don't know, a couple of the executives at the company and some of the board members, like, how do we make people give a shit? That was like what I put at the top. How do we make people give a shit? And we couldn't make people. There's just no way that we as a small company could, you know, I mean, we were a billion dollar revenue company, but that was very small compared to, you know, we just could not come up with a way, like, I don't know, loyalty programs, you know. You know, fear was not going to be how people ended up giving a shit. Because, you know, we've, we've been through that many times with Wikipedia leaks in the Cambridge Analytic. All that fear es and flows, but it's not what drives people to constant different habits and it shouldn't be. That's a terrible society. It was driven by fear. So it's got to be. I pull this towards me because it's better and I like it and it makes me feel more powerful and more efficient and more something. It's gotta be more and good. It can't be bad and away, right? And we couldn't figure that out because the mechanisms, and this again was like 12 years ago, the mechanisms were so strong already for the centralization of data and the processing enclosure of it by, you know, large players. We couldn't figure it out. And I still am not entirely there, but I believe that the potential is there now with generative AI agents to have agents that work only for us. And once if you see what's happening with Vibe coding and you see what's happening kind of with the eye opening, like, holy shit, this is possible now it's running headlong into the terms of services and the privacy policies of all of the companies we work with, right? So just one tiny example. I wrote a little Claude code program to search a bunch of stuff on my computer and across a little database that I had built of all my writing that I had on my computer. So it was all on my local map. And as soon as I tried to execute it, this little piece of code that, that, that I co created with Claude, Apple wouldn't let me do it. And Apple said, oh no, we don't trust this code. And this was like essentially Apple brought the model of the App store for the iPhone onto my personal space. It was like, it's like they came into my home and said, sorry, you can't put that painting on that wall. We don't think it looks good. You know, I'm like, What the are you talking about? This is my space. I get to do what I want to do here. Right? And that's when I was very, like, direct example of. Of how the approach that we've built into our invisible contract with technology is intruding on kind of the kinds of things that Barlow used to rail about in the mid-90s. And I think we're at this really interesting tug of war between tens of thousands of entrepreneurs and vibe coders and just hackers who. Who see what's possible with agents and with AI and are trying all sorts of crazy shit with openclaw and other harnesses, right? And the built infrastructure that has, as you well know, like 20 million lines of policy code built in that do not allow automated scraping, do not allow agents to log in on behalf of individuals. All of it because they're protecting your privacy or they're protecting you from malicious code. They're protecting you from all that bad stuff that happened when the Internet happened. Right? That's Apple in a nutshell. And that's every privacy policy in terms of service for every. You know, I don't care if it's Strava Uber, like, whatever the app is you're using. They all have the same basic privacy in terms of service policies. And that infrastructure does not play well with this new world of agents, or genies, as I used to call them. And I think that's a fascinating tension, and I think things will break out that people are like, I just got to use. This is just too fucking good. I don't care if it breaks Spotify's terms of service. You know, I don't care if it breaks Amazon's terms of service. And you see Amazon, you know, fighting against this right now. And not all the actors that are pushing these breakpoints are good actors. Most of them are bad actors. But that's exactly how we find a new paradigm for engaging with powerful tools, is everyone tries to take advantage, and then we find some regulatory stasis, and we're in the midst of a great realignment right now that I don't think, I certainly hope will not end up status quo, because status quo is suppressing an incredible amount of value and innovation and growth and flourishing that could be happening outside of the kind of controlled architecture that was built over the last 30 years in the technology industry.
- [30:18]
Alan Chappelle
But who's creating the agent that's going to be able to, you know, circumvent Spotify's terms of service or. Or thwart whatever Apple's trying to put out There because if the, you know, it's not anthropic, I would think it's not open AI because those guys seem very clearly on the kind of the insidification path where, you know, the challenge on their end, I think, by the way, is, is going to be, you know, they don't have lock in yet. And your ability to really fuck with people, both advertisers and users, is just different until you have lock in.
- [30:55]
John Battelle
That's why this moment is so interesting, Al, because there isn't lock in. Switching costs are very low. In particular, they're low if you can export easily your conversation logs, which currently you can. They're even lower if you can automate that and you don't know you're doing it. In other words, if I start like, you know, a new AI chat service that starts to take off, let's say, amongst the tweens, and I'm like, hey, just press this button and we'll log into Chat GPT on your behalf and take all that stuff and put it in our service. Right. First of all, I am for that. I think that should be absolutely the law of the lamp, that data portability is machine readable. Data portability is like built into the constitution of the Internet and if you, if you read the words, the words say it is, but then those machines get stopped by the policies saying, well, no. And actually in this case, no. Right. It just gets all. That's why all of these little data locker businesses that, you know, got started over the last 20 years, none of them worked because they couldn't scale, because of this problem and many others. But this one is kind of a particularly difficult obstacle. So who's going to do it? Well, one hopes it's the equivalent of Sergey and Larry in grad school and in Michigan. You know, you hope that, that, that some people are going to, you know, and they're, I mean, it is the wooliest, wildest west right now for, you know, people who are building a gas town and you know, all the open claw stuff and like this stuff is bananas, it's impossible to follow, it's moving so fast. And I think the big guys are rightfully terrified of it. They should be. So I don't know who's going to do it. I just know that like every so often you see something happen that matters, where these forces drive a large player to acknowledge and embrace them and sort of break ranks with the oligarchy, if you will. And you know, for example, Android was a good example of that. You had this Sort of locked down, closed enclosed mobile operating system thing. And Apple was owning it and BlackBerry was owning it and Microsoft made their attempt at doing the same. Didn't work. And then Android came along and Google said, we're going open right now. Didn't end up that way down the line, 15 years down the line. But that moment pushed a whole different approach architecturally to mobile because all of a sudden there was an alternative. So I'm hoping that we see some of that start to break out.
- [33:47]
Alan Chappelle
Yeah, I'm sensing some real like mid-90s energy coming from. Yes, yes. So let me throw out one other factor because I would just love your take on this. One of the things that I'm thinking about is like, to what degree are all of these services currently being subsidized? Like what's the real cost of, you know, a single AI, you know, token? Because if it's, if it's being subsidized at like, you know, 3X, that's going to be a pretty big bill.
- [34:18]
John Battelle
We're learning. I mean, this is. Look, 100 years from now, assuming that the world still exists, there will be so many rich case studies of the economics of what's happening right now around this question of, of the actual cost of, of, of compute and network usage. The models are changing so rapidly. I mean, you saw probably at Microsoft's earning calls this week where, where they're like, oh, whoops, it's not per se, it's per se plus consumption. Right? And every software business is figuring this out in real time and changing their models. It is not easy to do that as someone who's been somewhat involved in this part of the business, the pricing and, and, and consumption side of it. But that's one big force that's happening, is people are realizing, okay, these costs, we can't keep shouldering them, you know, and use equity financing or debt financing to pay for them. Someone's got to pay and ooh, shit, that means the service has to be worth whatever that price tag is, right? So that's, that's on one side of the ledger. On the other side of the ledger are it costs this much to do it. But Moore's Law, right? The, the, the, the sort of like there's got to be more efficient ways to do it than the way we're currently doing it, right? And no one's more motivated than Google or Microsoft or Meta or Nvidia, really. I mean, at the end of the day, we will consume as much compute, as much bandwidth and as much storage as we possibly can forever, I think, I just think we will. So there's never going to be no need for that stuff. But it has to come down in price. The economics don't make sense, as you pointed out, currently the costs of all of these things don't pencil out to the value delivered. And so both things have to happen. The costs have to come down and the value has to go up and we will find some middle ground. And to my mind, one that's, that's yet another kind of tectonic force in the mix to drive compute storage and lower bandwidth by driving it to the edge of the network. Right. Which is the whole idea of the Internet distributed intelligence. Right. So this idea that we gotta build these massive data centers and we've gotta like, you know, spend trillions of dollars in capex across the top 20, 30, you know, technology companies. That's true. But we can do so much more if we use the compute that's right here and the compute of the computer that I'm staring into right now and if we understand how to intelligently distribute the load of processing, storing and communicating all of this. And again, that's the problem the Internet was built to solve. And I think we need to sort of go back to some first principles here. And I think we have enough really smart people, probably many who are listening to this, who are thinking about solutions that might change the game so we get away from this sort of centralized mainframe era thinking about how we relate to compute, data and storage. It's like to me there's, there's got to be a better way.
- [38:02]
Alan Chappelle
Okay, so we've talked a bunch about how this would impact, you know, just sort of regular users. We haven't talked yet about, you know, the impact of all this on publishers and even to a certain extent on E commerce players. Because if you have an agent who is now you've figured out how to direct them to get you, you know, the information on the war or not war, who knows, in Iran, or you've figured out how to tell that agent to get you a new pair of shoes so you don't have to go to Zappos or Amazon or whatever. How does that impact those models?
- [38:35]
John Battelle
Let's put retail to the side for a second because you're selling something people need for the most part. I'm not sure about your Instagram feed, but maybe not everything you need. But there's want there. But, but need and want. Whereas with publishers, the question of whether people need and want what you're making has been already thrown into stark relief over the last 20 years or 30 years by the Internet. So I would say it's not, you know, these models, the models of AI summarization or agents fetching summaries for you and so on my morning pages, right. These models are already under pressure and will collapse in this kind of a world. It's just, they don't work. So what replaces them is. Right. Is kind of the next. Well, this is where the architecture of trust comes in. There's, there's two pieces to it. What do the AI models that are generating the summary's trust and what do you as a consumer trust? Right. So to the first one, there are a lot of really interesting efforts going on to figure out a way to properly remunerate trusted sources of high quality information in AI models. They are early, they are blunt, they do not scale. Right? So if you look for example, at the 60 to 200 million dollars a year that Reddit is bringing in by selling its entire archive and ongoing conversation to Gemini and OpenAI and you know, everybody else, that is a business development deal. And it is blunt. You know, they don't really, Reddit doesn't really know how much value they're delivering to the models that they are giving their data to. They have no idea how much value is being generated on the other side of that equation. But 60 million sounds pretty fucking good.
- [40:38]
Alan Chappelle
Yeah.
- [40:39]
John Battelle
You know what I mean?
- [40:40]
Alan Chappelle
Like, well, so they're not running a newsroom that, that their incremental cost for creating this content is just a league different than, you know, WaPo or the Times or somebody like that Now.
- [40:51]
John Battelle
Now. So, right, so you're, you're taking my point and you're running exactly where I was hoping we'd go, which is this does not. That model, you know, a business development model, if you will, does not help or scale at all beyond maybe if I'm News Corp or the New York Times and I want to cut a deal that I feel like is fair, you know, I'm sorry, but if I'm a, you know, 40 person, you know, newsroom in Schenectady, I am not going to get the $750,000 a year I need to pay my reporters from AI right now. And the model that, that whether or not the Times wins its suit with the OpenAI or, or, or cuts a deal and gets 15, 20, $30 million a year, that doesn't help me if I'm a local news publisher, right? And it doesn't help me if I'm, you know, Tom's hardware. And it doesn't help me if I'm the industry standard, which, you know, doesn't exist anymore. But, you know, if I were running it now, Right, So what might help me? Well, a reliable, programmatic, trustworthy, economic engine of exchange that properly auctions value for discrete pieces of information. That sounds like something we already built. That's the entire ad tech fucking, like, infrastructure. It's right there. It's already built. It's just not being used for that. And so I've spent some time, you know, people who, like, love to dream and build in this space, like Bill Gross, right, with Pro Rata or Matt Prince, you know, at Cloudflare, they see it, but again, we have a coordination problem. And as long as the old model is good enough, people will keep scraping and stealing and, you know, they don't want to pay for it because their competition isn't paying for it. And so why should they add marginal costs when their competitors don't have those marginal costs? So we kind of have this coordination problem in terms of getting the local news or the small blogs paid properly for that brief moment, 2003 to 2008 maybe, where AdWords and AdSense kind of marginally added enough value to smaller sites that were high quality with good composition and good return on advertising spend for the automated system. There was a moment when that shit all kind of started to work, and then it got blown up by people who would build basically copies of those same sites, but with 70% of the quality. And then you could just hit repeat and do that, you know, all day long. And all of a sudden the Internet became flooded with shit. Well, maybe it wasn't shit. Maybe it was just not as good, right? And then it was a race to the bottom, right? And we certainly see that. I mean, I think I saw recently that 40% of Reddit comments are now AI generated. So we've got an Urus Burros on our hands, right?
- [44:00]
Alan Chappelle
LinkedIn might be 80%.
- [44:02]
John Battelle
Yeah. Hey, so now if I'm, if I'm Google and I'm looking at renewing my deal with Reddit next year, and I'm like, wait a minute, I'm consuming from the feed that I'm paying you $60 million for, I'm consuming 80% or 40% or some significant percentage is I'm eating my own tail. It's like, you know, it's an Aerosporos, maybe I don't want to pay you 60 million anymore. It just doesn't feel like a scalable deployment defensible business model for anyone in publishing to do one time, you know, deals for content. It feels like a sugar hit that might keep you going for a little while, but it's not a defensible model. And I guess right now the only defensible model is one to one direct connection between a creator of some value of information and a consumer of that information. And that's why we're seeing substacks and you know, podcasts like this. And I also don't like that as an endpoint. I think that's important and should always exist. But I also think that we need shared common truth through brand publications. Branded publications. I think that's important. So we have to figure out ways of monetizing them that does not rely on the shifting sands of a grand bargain with the technology industry.
- [45:29]
Alan Chappelle
Right, but how,
- [45:34]
John Battelle
I mean, you know, if I knew that I'd be selling it, you know, bottling it and selling it. I mean I think there's lots of examples of things that, that, that work because they're hyper local. You know, I think Bob Con's onto something with the Baltimore Banner and you just have to have a really strong local base supporting you if you're in that particular game.
- [45:56]
Alan Chappelle
I've been following Richard Gringas a little bit and I had him on the pod a while back and like I, I think he's sort of similar idea like the, he's got, you know, now, you know, that's how you can make a living doing that. You, you may not make, you know, you're not going to make an Internet, you know, mogul living, but most people who go into journalism aren't really there to make Internet mogul salaries.
- [46:20]
John Battelle
Not. And so it's really, you know, it's kind of the Lord's work and it is not easy because a model like, like Bob's, you know, it's part grant making, it's part hustling to sell ads. It's part of part the endless grind of subscription renewals and you know, and, and marketing. It is what it is. Right. But you can't make it by playing the rules of the game from five or 10 or even 15 years ago. You just can't.
- [46:55]
Alan Chappelle
So okay, so we've talked a little bit about sort of publishers now do the same principles apply or maybe the same challenge is certainly applied to the E commerce players, just any retailer. Because I feel like the publishers, every time I go to a publisher event there is some flavor of oh my God, I can't believe how screwed we are. I Don't know. We're going to be in business six months from now and it doesn't feel like that's hit the retail world.
- [47:20]
John Battelle
No, it hasn't. And I think it's because they haven't lost their marginal revenue yet. Their marginal revenue for the publishers was advertising, right? That was, you know, I mean, you paid your journalist salaries with your subscription and then your marginal revenue gave you your profit, right? So when you lose your marginal revenue and then people stop subscribing because they can get it for free on the Internet, you're double fucked. The retailers, on the other hand, they get paid when they sell a something right to someone. And that model got intermediated by the Amazons of the world, right? And so they, they got disintermediated, but still they found ways to, you know, cut their costs and figure out, you know, how to give that piece that they have to give to Amazon. They figured out a way to keep going. Right. Some of them built good DTC businesses. Many of them refused to, for, for many years because, not because they were worried about competing with Amazon, but because they didn't want their other channels to be pissed off. Right? So you don't want to, if you're making diapers, you don't want to piss off Walmart which sells like 30% of your diapers by starting a DTC brand for diapers.
- [48:36]
Alan Chappelle
Yeah.
- [48:36]
John Battelle
Meanwhile, you're watching every Tom, Dick and Harry build a DTC diaper brand and make a bunch of money and steal your customers. You're like, what do I do? Right? I guess I better get into dtc. And then it turns out that, oh shit, it's a really good thing to be in because now I have a direct relationship with my consumer and I can understand how to market to them directly and I can get them on a subscription and I can, you know, and now I've got a bunch of data which allows me to use the ad tech ecosystem much more efficiently and increase my roas and you know, blah, blah, blah, blah blah, right? And develop new products and push that down my supply chain and understand, you know, things I could do with my vendors, I couldn't. And like having that customer data is insanely valuable, right? Especially in an era of AI. So I say this to anybody who asks, if you don't have a direct relationship with your customer as a retailer, you know, get into a different business and you could get that through your channel partners. Every single one of those channel partners has a retail media network now. I mean, you name one that doesn't Right. So Target, Walmart, Best Buy, Costco, they all have them. That data is there. So the negotiation is, wait, I don't just want to buy that data, which is my customers buying my stuff. Right. I have to spend more so that I can then spend more on your retail network to convince those customers to buy my product on your network. That's just a few too many hands in the. I think at some point that rationalizes. But as we've seen over and over and over, with the renegotiation of the economic contracts that bind our society together, there's always people making way more than they should on some side of that negotiation and it gets evened out over time. Right. And with retail, I think we're in the midst, as we have been with everything else, of a pretty significant renegotiation. But they aren't in the same bind that the publishers are because they're not losing an entire marginal revenue stream. As a matter of fact, there's an opportunity to build a new one. But you need capital and you need execution risk. It's going to, it's going to succeed or fail, but there's an opportunity to do that. The problem on the publisher side is we not only lost the advertising marginal revenue, we lost the consumer demand. That's like, you know, I lost, I got disintermediated from my diapers and no one wants diapers, you know, like. Yeah, and, and so that's a bit different. You know, people might want different diapers, but they're still going to pay for them. Right. And when you can go, you know, on Google or Chat GPT or you know, to a certain extent, Claude, and say, you know, give me a rundown of what matters to me in news today and all you're paying for are maybe a $20 a month subscription maybe. And that $20, none of it is going to the, to the news. Right. That is irrational from a point of view of democracy. It's just irrational. It, it does not make sense when public knowledge of what matters in society becomes a private exchange between the encloser and, and the consumer. And the information itself is a public good. We've seen over the last 20 years really since the rise of Web2 the enclosure of public good. And sure, let's look at transportation. Public good is people can get from one place to another. So most municipalities have some mass transit of some kind that is paid for through government subsidy. And then you had the medallion taxi system and it was corrupt and awful and. But you could get a cab from one place to another. And that was a public good. And it was. That public good was supported by a regulatory framework that regulated taxing medallions. Then Uber came along and said, nope, we're going to take that public good and enclose it and turn it into data, by the way, which allows us then to build an advertising business, a food delivery business, a data segments business, a retail media. Like what? Doesn't matter. They built it all that is capitalism. Do what capitalism always does. Free inputs, externalities pushed out onto society and sort of a rapacious search for profit. At some point, I think we as a society are going to wake up and say some public goods need to be protected. And I am not in any way afraid of the idea of one of those public goods being named as journalism. However, I absolutely am certain that in the United States it will never happen. It just won't. It's just not. It's not in our DNA to think that, that journalism rises to the level of a protected public good despite the fact that we wrote it into our constitution.
- [53:51]
Alan Chappelle
Yeah, I feel like I've been waiting for that pendulum to swing back for, you know, 50 some odd years and I'm not sure we're done with it. Swinging to the right.
- [54:02]
John Battelle
No, I don't think we are. Worse I can get, I think you can get much worse. Before I just finished reading, and I think this is probably a fashionable book amongst friends that we have in common. Sven Burkett's Capitalism A Global History.
- [54:17]
Alan Chappelle
And it is, that's on my list.
- [54:19]
John Battelle
It is eyeball burning, you know, because it's just so clinical, you know, it's just like, it doesn't, it's not polemical, it's not trying to argue that capitalism is bad. It's just saying, well, we chose this and this is what it's done, you know, and, and it's, it was, you know, and he starts in like, you know, the 1100s and goes all the way to the present. And it's like, wow, yeah, here we are predictable. This is predictable. And the systems that we built, the, that we're now struggling with and that I think AI has the potential to both massively disrupt and massively lock in. Like the stuff we have now could be locked in by AI and it could be completely disrupted by AI. And we're in this wonderful moment of not being sure which one's going to be true. And it has to do with who has, in my opinion, who has agency, not just control, but agency over what data. To me, that's the fundamental question is agency.
- [55:26]
Alan Chappelle
Well, I think now might be. Well, you know what, before I even go there, I, you know, we've talked to publisher, we've gone into E Retailer. You know, my audiences are ad tech nerds, or a large part, portion of them. Like, how is all of this in your view going to be impacting the ad space and maybe in particular the data space. Because like I've got, you know, some people whispering in my ear saying that it's going to, you know, completely change the data space. And the others are saying, well, data, but, but addressability is really the place where things are going to change. What's your take?
- [55:59]
John Battelle
I mean, I agree with both. I think addressability is kind of everything. I wrote a piece 15, 10, 15 years ago claiming that the programmatic ad tech infrastructure as built and this was whatever mid 2010s was probably or possibly the greatest living artifact ever created by humanity. And the ability to in micro milliseconds negotiate a complicated transaction involving literally along the whole chain, millions of data points was just a fantastical invention that would redefine everything. But we keep just using it for ads. And that seems to me to be kind of like we're rebuilding. We have rebuilt a scene from space, similar artifact in LLMs, in sort of the chatbot surface of LLMs, and once again, what are we using it for? You know, there is no question in the ad tech space that we are seeing the power structure shift towards those who have high quality first party permission data. That's where it should shift in my opinion, and that's really important. High quality, permissioned first party. Right now, if it's permissioned in a way that's enlightened, then it becomes high quality permissioned second party data. And that's fucking powerful. Right now I'm on the board of Liveramp still, so I'm going to be biased. I think we need to unleash that data throughout the economy in as frictionless an environment as possible so everyone can access it, query it, gain insights by it, and build innovative things based on those insights. We currently are in a data economy that almost actively does the opposite of that. It is very hard to use permissioned first party high quality data because if you've got that, you enclose it and try to leverage it just for yourself. That's the model we've taught everyone over literally centuries of capitalism, to treat that as a rival risk asset that you don't share with anybody and only leverage for yourself. I guarantee you, and I'll be dead when this is proven. But I guarantee you that historians one or two generations from now will look back at us and go, what were they thinking? Like, once you allow that kind of high quality information to move to its natural fluid states throughout the economy, we will 10x to 100x the innovation and the economic growth in our economy, we just will. But we're still acting like kids who don't want to share their toys.
- [59:07]
Alan Chappelle
Well, I think there's, there's, you know, I think some of the concern there is privacy, right?
- [59:12]
John Battelle
And, and that's why it's really important that it's permissioned. You know, Yes, I, I see that smile because there's a lot of permissioning going on that people are completely unaware of. But that's why I get back to the model that we talked about earlier, where I really think that the only way that this actually changes long term is if it's permissions at the level of the individual, through an individual's trusted and solely controlled agent who's always out there checking with everybody, checking with Amazon, checking with Liveramp, checking with all the ad tech cookies and pixels and saying, hey, no, wait, you don't have my permission. Nope, you don't. You know, that seems crazy. That's all that work. That's what everyone pushed back when I made investments in early data locker companies. They're like, there's no way anyone's going to spend all the time managing all this stuff. No. That's why we have AI agents now. Right? So I just really think when we get to that point where I want to start a restaurant on this little island because I, there's a kind of a place I want to eat at that isn't here. And that's just me putting my finger in the air and going, well, I hope it works out right. But wait, in this zip code on this island, there are 25,000 people using DoorDash, using Uber, using Lyft, using Resi, using OpenTable, you know, using grocery apps, all of that data. Imagine if I could put an RFP out that was geofenced by this island and said, what kind of food do you eat? How much do you spend? How, how far are you willing to drive to come to a restaurant? What's, you know what, what's your preferred time for reservations? What's the size of your party? If I had all that information, I could start a restaurant that had a much higher chance of actually working. But I can't get access to all the spoils of that data because it's treated as rivalries. That's just dumb. It's like societally ignorant to not allow that to happen, in my opinion.
- [61:14]
Alan Chappelle
How do you prevent, okay, you're living in, I believe Massachusetts has a whole bunch of laws which specifically prohibit what you're, what you're doing, or at least make it really difficult.
- [61:24]
John Battelle
They make it really difficult.
- [61:25]
Alan Chappelle
Yeah.
- [61:26]
John Battelle
They don't specifically prohibit it because they never imagined such a possible. We live in, literally, Alan, we live in Flatland. We live in like this two dimensional world of the Internet and we think it's cool and everything, but we don't realize that there's this whole other dimension that could pop up if we just started using data in an intelligent way.
- [61:44]
Alan Chappelle
So I get the value side, but particularly with the scenario you just described, you're living in a pro choice state. If you were not, you know, as interesting as you, the restaurateur would find that data set, the government is going to find it even more interesting. And that's sort of the concern is like, and this probably not to speak for Barlow, but like that's sort of where, you know, where I think he was, one of the places he was going is like, you know, you create that data set.
- [62:11]
John Battelle
That was the whole thing with him. Yeah. And it turned out that that wasn't really a thing we needed to be concerned about. As a matter of fact, we sent Barlow into the CIA to do a piece about, hey, what are they up to right, at wired in the 90s. And he came back. I'll never forget what he said. He said, john, these guys have no idea what they're doing. Don't worry about it now. Now that's certainly no longer true when you have governments, number one, like the one we currently have, so sort of philosophically inclined to not care about things that we hold sacred. And number two, supported by Palantir, Google, Microsoft, OpenAI and so on, right? So the government is becoming a boogeyman again. But it wasn't for the last 40 years, right? So we got kind of like lazy about that as the villain, right? The villain became Zuboff's surveillance capitalism, right? That became the villain. But when things change rapidly and ultimately for the better because of fear, there's usually a lot of blood on the streets. And I'm really hoping that's not how things change because we're afraid. I hope they change because of what I'm talking about, which is because we're excited and optimistic about what we can do and we just start demanding the right to do it. I Should be able to do an RFP into an open marketplace for a geofenced amount of data from everyone who lives around here. So I can build a great restaurant, by the way, that the bank would give me a loan for because I've got all the data proving that the average bill size is $75 and the average taste wants American Italian fusion. Right. Like I should be able to prove that and get the loan and start the business and help the economy thrive. That's just one tiny example of millions of things that could be done if this data weren't enclosed by a ridiculously outdated way of thinking about economic inputs.
- [64:17]
Alan Chappelle
Fair. And just, just to finish the point, I, I think the reason that people forgot about how powerful the government was was, you know what, 1973, they outlawed the government from collecting certain data. Now they ended up just going to private, you know, to, well, to the axioms of the world initially. And then, by the way, we sold
- [64:36]
John Battelle
Axiom, you'll recall, decided we did not want to be in the data manufacturing business. Yep.
- [64:41]
Alan Chappelle
No, no, fair enough. But that's sort of the, but the challenge becomes like, I get that people want it, you know, how do we build that core doctoro coalition that he seems to be, you know, thinking about?
- [64:57]
John Battelle
And how do we build it without sort of crowning a set of villains and scolding everyone else? Right. That like, to me there's one. And I, I wrote a piece in 2018, I, I think it's called the Token Act. There's one thing we could canonize in law that I think. And I, I, I was at Columbia at the time at School of International Public affairs. Right. So I had a lot of policy people I could bounce this off of and they're all like, sounds good, you know, like no one pushed back on it. And I, Tim Wu, didn't push back on it. Right. But it was essentially this, just one simple line of code that we insert into federal law which says that every single collector of data must make that data available to its rightful co creator, the consumer, in a machine readable way. That's it. Now will there be harm from that? Yes, there will be so many gray actors who go, hey Alan, you know, just allow me to go get your data for you and I promise I won't do anything bad with it. Right, that's gonna happen. But that's, oh, that's what you're balancing when you're making these kinds of sort of the equivalent of public health decisions. Right. You know, like the net benefit of enabling hundreds of Thousands of entrepreneurs who now know that they can come to us and say, hey, do I got a deal for you? I am going to ask you to push this button and download your entire Amazon purchase history so that I can go over to Walmart and share it with them. And I bet they're going to come back and say, Holy shit, here's $1,000 if you switch to Walmart. And you're going to save $800 a year if you keep shopping with us. Because we've looked at your history and man, you're getting ripped off. Right? Like, entrepreneurs will leap on the idea that they can actually do that. So will bad actors. That was the first 10 years of the Internet. And you know what? It worked out okay. You know, like, that was the era of shareware. Yeah, every time I got a 3 1/2 inch floppy disk and stuck it in my Mac and it was like dirt dirty. And it had been like 100 people had passed it around before it got to me. There probably could have been a virus on that, but I still stuck that thing in the Mac. And by the way, I had another one that was an antivirus software. So, you know, let's get back to that energy of like, fuck it, let's try it. You know, and the bad actors, we already got lost for them, you know, them really well, like, will manage it and there will be, you know, people whose data gets deleted and misused and. But damn, the value is so much higher than the cost.
- [67:44]
Alan Chappelle
But in order to get there, it sounds like we sort of need to upend the current power structure a bit.
- [67:50]
John Battelle
Well, I think that one piece of code, as I call it, that one law, would do a lot to do it. I think if we just codify. GDPR does not codify machine readability. So who cares if you can get a copy of your data? I asked for a copy of all of my data once GDPR went into effect and I got like CD ROMs from Facebook, got like an unreadable HTML file from Twitter, you know, like, what the fuck? This stuff is not machine readable. It's purposefully useless. But if we just made it such that it was purposeful and useful, I, I think a whole new economy could blossom and it would surprise the hell out of us. How amazing it would be.
- [68:32]
Alan Chappelle
You could create an authorized agent to make requests on behalf of, you know,
- [68:37]
John Battelle
a million people for listeners to, to, to look into. It had all sorts of good and bad and gray in it is the story of plaid and, and plaid is just an example of something that the banks fought for the entire life of plaid and then realized, wait a minute, if we actually support this, our businesses will grow. Huh? You know, and like that's what I'm talking about at massive scale. Is that kind of an approach? Just let us use our data. We don't even know we want to. But believe me, entrepreneurs like me, I mean I'm kind of, I've done eight companies. I probably won't do more but like, but like I was for the past 30 years. We'll find things to do with it that will delight and, and, and, and astonish and surprise us and provide new magical moments. Like the first time you use Google or the first time you had an iPhone or the first time you chat with chat GPT. They'll, they'll, they'll keep coming. Or we could just lock down the model we built over the last 20 years and call it a day. Hang up our spurs. Because that's a boring kind of ugly world.
- [69:47]
Alan Chappelle
Yeah. Agreed. John, thank you so much for giving so much time. This has been a fantastic discussion. Where can people find you?
- [69:56]
John Battelle
I hang well, I write on my own site which is batelmedia.com so if you want to, I have a newsletter there you can sign up for is anything I write goes out on that. And then LinkedIn seems to be the place most people find me these days. I, I've left Twitter, I left Instagram. I kind of don't like the those platforms anymore. So I'm still on LinkedIn. So follow me there or ask for a message request. I'm a little bit bankrupt. I think they only let 25,000 connections and I'm like at 24,500 so I might not be able to connect directly but you can follow my work there.
- [70:27]
Alan Chappelle
Well, fantastic. John Patel, I really appreciate your time. This has been a great discussion. That was a very meaty discussion. I feel like this could have gone on for another hour or two. So a few thoughts. First, I love the framing John offered around data being a non rivalrous asset that is generally architected as if it were rivalous. He throws it out as a casual aside, but it actually renders much of the privacy debates we've had over the last 15 years as rather derivative. Every fight over cookies and consent banners and lots of the discussions around privacy enhancing technologies, all of it is downstream of a foundational economic choice that was made decades ago to treat data the same way we treated coal or timber. Enclose it, extract from it, externalize the costs. John's point is that this choice was never inevitable. And the moment we're in right now where the architecture is being rebuilt around AI, well, this is one of the rare windows where that choice could actually be revisited. I'm not sure it will be, but but I take his point that the window exists. Second, on the OpenAI $100 billion ad number, John was generous in his framing. He called it audacious, and he acknowledged that he's made up numbers himself. As a publisher, my take on that prediction is a little more cut and dry. For OpenAI to hit that figure by 2030, they would need to capture something close to 45% of the incremental growth in global digital ad spend over a two year window, and a chunk of it would need to have to come out at Google and Meta. I have a really hard time seeing that happening. In my view, listeners should be skeptical when they see those numbers cited as if they were serious forecasts. Third, how we restore balance to Platform controls John's example of Apple's gatekeeping, in this case Apple's blocking cloud generated code on John's personal MacBook was an apt illustration of where the next decade of platform control fights is headed. I can certainly see the attraction in some of the gravitational pull towards revisiting John Perry Barlow's Declaration of the Independence of Cyberspace, or even the more recent Robin Bergeron's push to rewild the Internet. My question is this, how much of what we liked about the Internet of the 90s was made possible simply because it wasn't yet a mass medium? Fourth, on publishers, the hard truth John pointed to is that nobody has solved the coordination problem at the heart of content monetization in the AI era, and the incumbents have no real incentive to do so. The Reddit deal, the News Corp. Woo and sue approach, even the New York Times litigation. These aren't long term strategies, they're closer to one offs, and they certainly don't scale to the local newsroom in Schenectady or Atlanta. There isn't currently a viable demand side of the equation, and until there is, there's not much point to the AI content marketplaces being built right now. Fifth, John's proposed Token act, the idea of creating a US Federal level requirement that any data collector must, upon request, return that data to the consumer in a machine readable format. I love the idea of unlocking access to data, but it's worth noting that this concept of data portability has long been enshrined in privacy and Data Protection law. GDPR requires controllers to provide access in machine readable format, CCPA and other state privacy laws mandated as well. In that way, this isn't so much a question of whether or not we need additional laws. Rather, it's a question about why the existing law is not enabling the type of access to data that John and other entrepreneurs are seeking. Also, wearing my privacy hat, I gently push back on the concept unless it came with a way to meaningfully address what I would characterize as the bad actor issue, not to mention my concerns about allowing the government to access this type of data at scale. There is a fascinating debate to be had regarding the degree to which having open and perhaps unlimited access to data could positively impact our society, but equally, there needs to be proper safeguards in place. One thing I really appreciated about this conversation is that John refused to settle into either the techno optimist or techno pessimist category. I came away from this conversation more convinced that the worst version of the next decade is the one where privacy gets used both to suppress innovation and to stifle competition. And I want to highlight that's the world we're living in right now. One of the goals of this podcast is to push for a better one, one discussion at a time. On that note, we have a bunch of other fantastic guests coming up on the Monopoly Report podcast over the next few weeks. Please subscribe to the show@monopoly-report.com or on Spotify, Apple, YouTube, or wherever you listen to your podcasts. And thanks for listening.