The Exchanges

Q what are the things where it would matter and which of those components do you need in order to do that for the, you know, and there's the purity test versus the reality of you make this product that's different because it uses crypto. Which bits of crypto does it have to use? Does it have to use? Every aspect of that. There's all very kind of vague questions, sorry.

A But they're good vague questions in that what I, the sense that I'm getting right now, which I think is totally fine, is that most of us don't know what we're actually trying to build when you talk with people in crypto and whether it's like trying to create a DAO, whatever it is that people are wanting, the desire that people have, we're not, I don't think we know exactly what problems we're trying to solve and I think that's okay because I think we might come across things that we probably weren't thinking of in the first place. Similar to the way that you just explained the whole story of open source is I think people had an idea of what they were trying to solve. And in the end, they actually ended up solving maybe different problems in different ways, which is exciting and great, but that's the, I, I, or maybe I'm wrong. Maybe I'm not talking to the right people, but the sense I'm definitely getting in crypto is what are you, what are we trying to build and what we're trying to solve for here?

AI assessment note: “most of us don't know what we're actually trying to build”

Q And why? And how will that work? And we're not quite at that stage yet. Like there is no OpenOffice and there's no Linux. We're like, we're three steps before that. But those are the sort of architectural questions to what might the questions be.

A It's like the architectural and the infrastructure questions, and it's fascinating to me because I am definitely someone who starts with the infrastructure and then thinks about what's possible based on the infrastructure that you have. But I also know that a lot of people think to your, to what you've just explained of just like, what could it be? Let's think really big and we'll get down to the weeds and the infrastructure. Like that's the last thing we need to think of. Um, Which becomes fascinating, because I always wonder, like, why are people interested in web three right now? Is it because they want to think of a new way to build that infrastructure, or are they more interested in that projection of what could be and what all of this could build down the line?

AI assessment note: “I am definitely someone who starts with the infrastructure and then thinks about what's possible”

Q general purpose software and it's general purpose tool that anyone can just use. I always feel like go watch the office and imagine saying, I'm just going to give co-pilot to Dwight and Fat Keith and David Brent. And so you don't need to use our accounting tools anymore. You can just use ChatGP until you do the invoicing. Are you sure? Do you really think that's a good idea?

A It's interesting that you bring this up. I was In the middle of, um, I don't know if I should actually share this, but I was in the middle of renewing my O-one visa, and obviously my lawyer is an immigration lawyer, not a business lawyer, because I have a single member LLC. She was saying, it'll be good for you to get a letter from your own LLC saying that you'll be employed by that LLC. So I was like, wait, you want me to write a letter on behalf of Tony Karen Brown saying that I'm going to hire Tony Karen Brown as a creator. That makes no sense. And she was like, well, I've seen a couple of clients do that. Anyway, I went to chat GPT, asked a few things, but what was interesting is chat GPT gave me a couple of really good answers of how I should structure this. But the last bit, I just found it, but the last bit of that chat GPT answer was really interesting to me. It's like, would you help, would you like help drafting an authorization for you? And I was like, I hadn't even thought of that. I went to chat GPT to help me understand the problem. Not in my wildest dreams that I think, oh wait, ChatGPT can also draft me the legal authorization that I will then use for my O-one visa application. And so I was just like, yeah, please go ahead, draft me an authorization, which it did. And then, you know, the next answer was like, who's the best person to sign this and what are the l…

AI assessment note: “Went back to my lawyer and I was like, well, does this work? And she was like, oh, that's great.”

Q happening where people will, and maybe it's to your point, the value that's attached to a product or the celebrity that's attached to it, we then will go out and buy the product and then have that consideration about why did I buy this? What is it for? So I'm just wondering if you've seen anything about that, that purchasing funnel and what shifts are happening in the landscape here.

A Yeah, I, so I'm a bit of a contrarian on this. I think that most people, Demography is made up nonsense that isn't true. And I think it's, it's often done to sell consulting services and to make marketing seem like they're creating generations. The truth is, I don't really believe that with one exception, which I'll come back to, that there's much difference between Gen X and Gen Y. I don't think there's a ton of difference between the millennials and Gen Y. Like, it's just, Being young has different meaning in one context, which is in the pre- and post-digital era. That's the only demarcation, in my view, that's real. If you were born after the iPhone was created, and you were born into that, so all you know is digital first, then that, clearly there's a demarcation line there where you behave quite differently. Your social norms, your values are a little bit different. Your health issues are a little bit different. Like, to me, that is the one, because I, because I think if you looked at demographic changes over time, you, if you shrink the graph long enough, short enough, you'll see blips. But if you look at it long enough, it's pretty similar until you see that demarcation point of the truly, what I would call the mobile, the mobile side of the digital age. And that is a, to me, a massive cliff in demography. So I think a lot of this stuff, like, we have conversations here …

AI assessment note: “I'm a bit of a contrarian on this. I think that most people, Demography is made up”

Q I mean, I. Can you pick that apart just a little bit, like the difference between those two investments for people who might not live in this space? Like, what, what are, like, what are the fundamental differences between those two types of investments?

A Okay, so, so sort of basic venture capital is you're right out at one end of the risk reward curve, and so you expect half of your deals to fail, you expect about five to 10% of your deals to make more than 10 times your money, and that's what produces the returns for the fund overall. And so every deal you do has to be capable of being the one that goes to 10, which is actually 20 or 30 or 40 times x. So everything you do kind of has to be a crazy idea that's very high risk and very high reward. This is very basic Mechanics of venture for the last 30 or 40 years. This is how the industry works. Um, and sort of classic dumb criticism of venture. It's kind of like, you know, it's like the guy who, the person who walks into an art gallery and looks at a Jackson Pollock and says, my kid could do that. This is not speaking truth to power. This is just being dumb. And looking at a venture deal and saying, this looks stupid. This is not speaking truth to power. This is just being dumb. It's supposed to look stupid.

AI assessment note: “basic venture capital is you're right out at one end of the risk reward curve”

Q So the radio, so anyone, so the other teams can hear the radio. Is that a legal requirement as well?

A It used to be, they redesigned their whole, and now we're digging into things that I know a little less about, but. They were redesigning their whole comm structure and the whole, how the radio was functioning. And back in the day, it wasn't official, but you could easily listen in to your other teams radios, but the FIA has to, which is the governing body for formula one has to be able to listen into all of the communications that are happening. And so now all of the other teams, so the moment that the race starts, um, basically the comms is open to all, um, until the moment they go back into the garage where public communication then becomes private again. Um, and also someone else asked me a question, and this was when I was writing, um, my guide to remote working, which funnily enough, I wrote it in January. Um, so it couldn't have been more timely, but someone had asked the question when they debrief, so they have the pre-free practice sessions and then they go back into their, their rooms and they start discussing the race and the track and the tires and what worked, what didn't. They all have their headphones on. Even if they're sat with each other in the room, everyone is listening to each other via headphones. And the reason for that is there might be only like I think it's around 60 crew members during the race weekend on the pit with them. But back home at HQ, there'…

AI assessment note: “the FIA has to... be able to listen into all of the communications”

Q My god, do they have an archive of libraries here? To learn from.

A Oh, they have an awful lot. There's an awful lot of label content out there. Yes. Uh, so that's, you know, that, you know, we've talked a lot about music, like make me a song that sounds like this, make me a video clip that sounds like this, make me a, an X that sounds like Y. Um, we talked about generative, a lot of people talked about generative search this time last year, but generative content in general, Generative stuff. Amazon is overrun. You know, if you saw this link, if you search, search Amazon for, um, I'm sorry, that would break open AI terms of service. And you just got huge numbers of marketplace products where literally the name of the product is, I'm sorry, I can't name a product for you. It's just kind of lazy spammers firing stuff, like the fire hosing stuff into Amazon's APIs, listings API. That there'll be a lot of that. Um, lots of, lots of interesting new problems. That takes us a fair way away from the, the underlying question. Like what's our, what's our AI strategy? There is a plot. There's a piece of this, which is it's a new platform. So you evaluate it the way you evaluate cloud and mobile and SAS and everything else. Like what do you buy? What do you build? What's internal? What's external? What are your capabilities? Does it allow for this or for that or for that you, you break it apart into, you know, 10 slides with, Three or four bullet points o…

AI assessment note: “Oh, they have an awful lot. There's an awful lot of label content out there. Yes.”

Q AI, like give me the best thumbnail that's going to attract users. And this is what I need it to look like. And so to your point, it's not that far behind, but I was, I don't know why I was in this space of just like, Does the value of content change? Like, what does that then look like as it becomes easier and easier to create quality content?

A Well, so, yeah, um, I mean, they're going, so it was sort of, what's the friendly word said, the opposite of talking isn't listening, it's waiting. So, um, I think the, the sort of thought occurred to me as you were, as you were talking that, um, that like my sort of frame, sort of classic, like talking point I have, things I've written about productivity, um, Is to start with Jack Lemmon in the apartment, which is his 1960 Billy Wilder film, and he's a clerk in an insurance company. And everybody in the building has got this giant electromechanical adding machine that looks like kind of the bastard offspring of a typewriter and a cash register and a telephone and a typewriter and a, um, and basically everyone in the building is a cell in a spreadsheet, right? Literally, almost literally that's what's happening is, you know, you've got. 3000 people in this And so, like, we've had these processes over and over again where, like, this whole class of human activity gets automated, and that's been going on since the early 19th century. Um, and like at a minimum, this is another of those. It's another of the shifts from, from adding machines to mainframes. And suddenly you've got this huge number of people who are basically doing mathematics. I mean, they're using the machine, but they're kind of doing mathematics by hand. Suddenly that whole class of work just disappears. And if yo…

AI assessment note: “my sort of frame, sort of classic, like talking point I have, things I've written”

Q you have the removal of all the social constraint, and the peer pressure on the influence of the village, the church, and the family, and the community, and in the city there's no church, no family, no community, and you have, particularly in the 18th century, you have this explosion of, sort of, revolutionary mobs, revolutionary movements, And what do you do about that? How do you think about that?

A Your city's analogy is always fascinating, because I can't help but think about God, we're doomed. Like, there's no way out of this apart from if you take it all down and start from scratch. Um, but you, but your Google piece was fascinating. I found myself in a room in Washington DC with people from the RNC, the DNC, Google, Facebook, I think, um, and other social, um, platforms, but also other campaigning platforms. And there was a big debate about how Google was politically biased, um, towards one political party, except that both the RNC and the DNC can decide which one. So the RNC was saying, no, no, they're biased against us. And the DNC was saying, no, no, they're biased against us. And so it was fascinating thinking about, wait, are you saying that we have platforms like Google, Facebook, and we've seen this argument a lot that face, you know, Twitter is biased against the Republicans and Facebook is biased again, you name it. And so there's just this thing of reminding, no, in my opinion, These platforms are not biased against one political party or another. They're here to make money. Why do we seem to forget that? So they're obviously going to build the best platform that actually enables more money to be made. So to your point about, well, what if we stopped sharing on Facebook? My mind went to, would that help with their business model? Would that hinder that? And …

AI assessment note: “Your city's analogy is always fascinating... but your Google piece was fascinating.”

Q sense of, not as a generic term for crypto, but if you built a website on a blockchain and it was interconnected with other websites on a blockchain, then you'd sort of have that. If you have an LLM, how does that work? Maybe that's another website use case, except you can't, you can't, except crypto that an LLM can't actually tell you which particular website drove the trash in.

A You know, what's interesting is I've just realized that when I do research for whatever it is, There's always, and I imagine it's the same for you. You've got a notes doc somewhere or a Google doc somewhere, and you're, you're inputting all the links of all the sources and everything that you found. And you might be copying, you know, a certain paragraph of an article and then you add the link to it. And I've just realized that if I'm doing any kind of search and chat GPT, I don't think I've ever copied anything out of it in terms of what I would normally do with sources. A, there's no sources that they're giving me. But you know how when you start doing research on any other topic and you're aggregating all of the information left, right, and center, then I'm inputting all of that information into another separate document. Not something that I would ever do. Not something that I've done with ChatGPT. The data, like everything I'm given just stays there.

AI assessment note: “A, there's no sources that they're giving me.”

Q seen it with Substack also, this desire to want to build a writer's community. I think that's also a thing is you, you know, what does it mean when you start building a community of people and you start deciding who is part of that community? You're also taking an editorial choice there as well. Um, which I understand the benefit of the community, but it starts blowing the line.

A I mean, I think the other interesting thing we're seeing here is payment, which we mentioned at the beginning. And, you know, I considered ages ago having a payroll on my website. And I concluded that I didn't feel like I wrote enough to be able to justify paying, and very few people have managed to make a website paid. What Ben Thompson did, and he started his newsletter roughly the same time as mine, back in sort of, 2013 or so, um, he started paid at the beginning. At that time, it was, you know, a nightmare to get it all to work. There was no subset there. Um, and I was just about to join AC's Team Z, so they, That was like, Mark Andreessen said, Andreessen Horowitz is a media company that monetizes through venture capital. So that was my, that was my revenue model was get Andreessen Horowitz to pay me lots of money to write stuff on the internet.

AI assessment note: “I think the other interesting thing we're seeing here is payment”

Q make that happen, then you can, because that's the whole pitch is that that's what you should do. I mean, Airbnb is sort of the remote world company, you know, it is so you can be anywhere company, at least if you're single, um, and don't have children or commitments, um, then, then it seems very obvious that you can do that. Um, how well that will work in practice?

A Um, Yeah, to see. But look, the big difference for me, let's look, reading again, the commitment that they put out there in the world is, A, they were super decisive, and it wasn't wishy-washy. I think so many companies are, we're gonna sit and wait and see and do a hybrid model, and that hybrid model generally never works, but it was very much, you know, it was simplicity, very, very clear in its simplicity. We have five bullet points, and this is what we're gonna do, and this is what we're gonna focus on. Um, but I did think you, it was interesting going back, like, there's maybe something interesting here, just going back to, You know, when did this hybrid model start the remote? When did we start thinking about working remotely? When did we start doing hot desking? Like, it's just interesting to think that none of this is new, but, and we still haven't figured it out 20 years later.

AI assessment note: “they were super decisive, and it wasn't wishy-washy”

Q Richard Scarry car. You know, it's like a sausage with, with a wheel in each corner. And there's a guy sitting in the front in the middle with a leather helmet and a little windscreen. And he's going ridiculously fast in this thing. But like until like the mid sixties, I suppose they didn't have all the wings and stuff that you see now. And so how did that happen?

A I don't know the exact history of how all of that happened, but it was obvious. I think it's, and I guess it's human nature. Like, as the cars went fast and faster, there was this thing of, okay, how much more money can we pour into this? How can we get the aerodynamics that they're closer to the ground? There was a point I remember, I can't remember the exact dates, and I apologize for this, but there was, they had wings that were stacked up really high, and they realized that actually they were deadly, um, because they would just come off, and you could easily chop someone's head off with the wing just flying off. Of the top of the carts, but they were really high up on the back of the cars, like, like on, like wings on stilts. It looked ridiculous, but I think this is one thing that's fascinating with F-One is it's trial and error. They've tried and tweaked, and I think there's something like every 20 minutes in F-One history, a tweak has been made on a car, and so they're constantly tweaking and changing, um, how you make these cars go faster, and they've tried things. I mean, one of the things that's interesting in what In the sort of trial and error era is thinking about the pit stops, which has now become, you know, an incredible part of an F one race strategy of, and now the pit stops have become less than, what is it? Two seconds. I think Red Bull and Williams beat, be…

AI assessment note: “I don't know the exact history of how all of that happened, but”

Q maybe, maybe Elon Musk's X, maybe a couple of other things, maybe even Microsoft will have its own model. Fine. What does this mean to the rest of us? This is commodity technology. Um, what else happens? And so then the two second buckets I had were, okay, what are these models actually good for? And how do you deploy them? How do you build companies and products with them?

A Can I ask before we go into the two buckets, has it always been the case with this sort of, um, big tech evolutions that we can bucket what we're seeing in terms of, there seems to be like two buckets of inquiry, the future trajectory, the questions that we have that we actually don't have the answers to, which is where is this going? How is it going to be used? And then all the questions about the current approaches, like, how is everyone using it right now? How are we employ, you know, how are we addressing this? How are we deploying this? How are we managing these technologies? Is it always true that we've got this like present bucket and then the future trajectory, or is this unique to AI?

AI assessment note: “Can I ask before we go into the two buckets”

Q But it goes back to that question, why would one define it?

A Yeah, there's a, there's a spread of diffusion here, but there's also a, um, I mean, another thing in the Deloitte study I mentioned earlier was, say, last year and this year, they asked about hallucinations, and they said, do you believe that generative AI always produces accurate answers? And the question is still, like, a third of people say yes, even if they've tried it. Which is the problem that it looks so fluent, which again, we've, we've talked about before. The, the, the natural language generation is perfect, so it will make perfectly, it will make grammatically correct and fluent prose that says stuff that isn't correct, but the, the prose is so good that you don't, if you don't ask it about something you know about, you won't realize this. And I still see, like, waves of this on, on, on social media, of people saying, well, I asked Google this question, and it couldn't tell me, and I asked Anthropic or ChatGPT, Or Gemini, and it gave me the answer. And then you have, like, 10 people underneath saying, that's not the correct answer.

AI assessment note: “another thing in the Deloitte study I mentioned earlier was”

Q Does it replace your, yeah, high, high-end watch?

A Fitness was, yeah, fitness was the thing that worked. Um, and it's the same here. There's 10 different things that they're suggesting. And so to kind of use case point, so they show, um, they show a social, um, video call thing that looks interesting. They show, they're using, it has all these sensors on the outside there, so they're saying, well, it can record a video of your kids, and the one thing that a lot of people thought was a false note was they've got this video of this guy wearing the headset playing with his kids recording video. I don't know if I will do that. That feels weird. On the other hand, recording video feels weird. We do lots of stuff that feels weird. So maybe, maybe not. But if some of these things work and some of them don't, that doesn't matter. That's the point. They also showed it for productivity. You can have all these big screens. I'm slightly skeptical about that. I don't think the future of productivity is seeing more columns in Excel. I think it's having an AI that builds the model for you. Um, but you know, you could, You can quibble too much. The point is that sort of almost kind of what they're showing is it's a blank canvas. It's a developer platform. It's an app platform. You can make it whatever you want. You can make lots of things in it.

AI assessment note: “Fitness was, yeah, fitness was the thing that worked. Um, and it's the same here.”

Q generalize into other things? So on the one hand, you've got lots of people looking at that imagery and saying, what do we do with this? This is amazing. What's this for? And then the other question is, but what are the implications of that beyond images and text? What else could you do with it? So there you are. That's, that's the five minute. What the hell is this?

A What, what, what are we talking about? What's the journey here? But I think it's interesting to look at. I, and I don't know if this is just me personally, but I've definitely had an easier time imagining the different use cases With the natural language descriptions versus the imagery. Like it was, it was fun to play around with the chat GPT and just like start asking these questions and asking for descriptions and understanding. And I can definitely see the argument why people are saying, oh, this is an alternative right now to Google. And that's just one tiny example. The thing I wanted to chat with you about, which I kind of knew, but I hadn't completely wrapped my head around and realized is a lot of this AI evolution seems to hinge on cleaner data and Better accessible or better data, more accessible data. And I hadn't realized that there was this whole thing about synthetic data. Um, and I just read that there was actually a, which is kind of insane to think about, but there's a lack of good data in this space that's stifling the innovation. So we need to go out and create more and better data for this innovation. What can, is there anything interesting there to chat about or that you can like add a bit more?

AI assessment note: “The thing I wanted to chat with you about, which I kind of knew”

Q can pick it up, start it. Same with the Mac. You don't need to read a lot of papers. You get it. And when I'm listening to you talk, it's like, I, There's so much room for error and mistakes for people to try and get into crypto. So what does that look like for you to safeguard and ensure that people have an experience that isn't riddled with scams?

A So there was, when I was, when I was in my first year at university, there was a, there was a mathematician on the same block who had Linux PC, um, was running Linux on his PC. And this was at the moment when sort of Linux was very, the internet was new and exciting and Linux was new and exciting. And it was just kind of reached the point that it really worked. And So this is 1995, and, um, he typed, he wanted to delete something, and so he typed, and I'm going to get the order wrong, he typed star dot star delete DEL space star dot star space star, or maybe he should have, the other way around, but anyway, this is spaces in the wrong point, and if he got it right, he'd have deleted two files, but he didn't, so he deleted, he wiped his entire hard disk in about two seconds, like, including Nunox itself, so the whole thing was wiped, including the operating system, so it's now just one new memory. You could do that with your million dollars of Ethereum really, really easily. And you read these sort of postmortems of people saying, oh, well, you know, obviously I was an idiot because I didn't notice the second sub call. No. Like, come on. This is not a key. And the point is you won't get the user adoption until you have thrown that away, but the point is if you, and the dumbest fucking thing in talking about this is people who write these long things saying it's slow and it doesn…

AI assessment note: “the point is you won't get the user adoption until you have thrown that away”

Q would always give is you've got this retailer who knows everything about the physical store, but they don't know what shopping cart abandonment rate is. Does that mean, is that, but is that a tech thing? Or is that like saying, you know, you built, you know, here you are, you were a department store, now all of your outlets are in malls. Do you understand how car parks work?

A It's the story that always fascinated me with people interested in F one of always trying to understand why Red Bull was in F one, because if you're Mercedes, it makes sense. If you're Ferrari, it makes sense. If you're McLaren, it makes sense. You're trying, this is an advertising opportunity to sell your supercars on the road. And the fascinating obviously story with Red Bull is they don't even know the production company. They don't even own the production that make the facility that makes the cans. They are purely a content-based marketing opportunity. And they pull of their marketing budget into Formula One to run these cars to basically send more cans of this stuff. It is just a fascinating world when you think of it. I'm like, what are you?

AI assessment note: “It's the story that always fascinated me with people interested in F one”

Q you going to get me users, or am I bringing you my users? And if I'm bringing you my users, that's fine if I'm Jesse Watts' name, or Noah Watts' name, or Bill Bishop, or on the other hand, Glenn Greenwald. If you've got your audience and you take it to Substack, that's fine. And Substack isn't giving you more users. Ok, why are you paying 10% in that case?

A The business model for me is really interesting as well, because, and I agree with it. Like I like the idea of the business model, but I also had this moment of, I just started a newsletter. I'm not planning on going paid anytime soon. Hence, they are never really going to care about me because I'm not bringing them any money. So they're never going to help me grow my audience. And why should they? And I remember talking with someone at Substack at the start of the year of, are you, are you actually just the software or do you want to be something more? Because if you're just the software, then I don't think you ever have to worry that much about content moderation. But the moment that you're a little bit more, in my opinion, than just the software, then the content moderation piece is a huge part of your business.

AI assessment note: “if you're just the software, then I don't think you ever have to worry”

Q think about this. And we're in the middle of a sort of a moral panic of thinking, oh my god, there's all this terrible stuff being said online, and it's skewing elections, and it's getting people to not have vaccinations, and all sorts of stuff. And how do we think about what the complexities of there might be, and whether there's a solution, or what a solution might look like?

A I've got a question for you there, which is, do you think, so yes, we've connected everyone. Yes, we're trying to figure out content moderation, and we're probably realizing that at scale, it's going to be extremely complicated. How much do you think the political aspects of this has made it that it's so much top of the agenda for absolutely everyone? Because it feels like there isn't a week or isn't a day that goes by without, whether it's journalists, academics, you name it, talking about content moderation and how this is the most important thing and this should be prioritized. But how much do you think that the role that politics has played in it and the role that politicians have played, um, has made it such a priority?

AI assessment note: “I've got a question for you there, which is, do you think”

Q That's always what I wonder when we're like, 90, 95% there. Is it because that last five percent is actually incredibly difficult to get, you know, to get accurate?

A Yeah. I'm Given, given that the digital stuff is all perfect, I'm not sure how much that's because the digital stuff has less texture, how much that's because of the problems of the, of the quality of the video pass through versus, I'm not sure how much it's screen versus the video specific. That's what I'm saying. But we can look, we can, we can look into that. Um, so that's kind of one point. The other point is, yes, it's heavy. And the battery, frankly, is not a big deal. You slip it into your pocket. It's like you've got another iPhone in your pocket. It's a nice long cord, whatever. Um, iFixit worked out that actually if you adjust for that, that, that, that, that, that this device without the battery, there's just a bit that you, well, your head doesn't have a battery, it's actually, we're slightly lighter than the Quest Pro, which does include a battery, although the battery's at the back of your head.

AI assessment note: “The other point is, yes, it's heavy. And the battery, frankly, is not”

Q question is, you know, if for the sake of argument, Mesa had bought Twitter and we'd squared all the questions about switching the network, but you know, if Mesa had bought Twitter and made it run and put Adam Mosseri in charge of running it, what would the user base get to? Would the user base get to three hundred million or 500, or would it get to a billion?

A You know, what's also interesting. It was in, it was fascinating watching people on threads get angry at people who clearly came from Instagram sharing pictures and they were just like, this is a words app. This is for words, not pictures. And you're just like, well, again, to your point, there's maybe a different audience that's going to come in. The one thing that I really liked is I love sharing tweets, obviously from Twitter onto my Instagram stories. And I, it took me a while to realize that even though there's a very easy button on Twitter to share on Instagram, it shares the picture and the caption of the tweet, but you can't actually click on it to link back to the tweet, which is kind of defeats the purpose and the point. Which again is something that is done very seamlessly with threads to Instagram. You can have that same opportunity of sharing the threads to the Instagram story and it links back.

AI assessment note: “You know, what's also interesting. It was in, it was fascinating watching people on threads”

Q Where do we draw the line? What are we comfortable with?

A And there's a sort of another macro thing in this. I mean, it's just picking up on, on the reference I made earlier that like the rent, the returns, the shipping, the ads, the marketing, those all sort of lower into one. Um, You know, who is it? Where is the money coming from to buy that thirty one billion dollars of advertising on Amazon? Well, it's mostly coming from trade marketing. It's not coming from TV ads. So, but it's also coming from stuff that wasn't in internet advertising, but wasn't also being categorized as advertising at all. But the kind of the macro point there is, um, How is it that you would build a brand in this environment? Do you build a store? Do you start by creating stores? I mean, if you think about the way you'd have done this in the nineties, if you're, you're creating a new organically from the bottom up, creating a new store as an entrepreneur, you start with a store or five stores or 10 stores or however you think about scaling it. And you wouldn't do TV advertising in conjunction with the first opening your first store. You might, five years later, you might have got big enough to start thinking about TV advertising, depending on what kind of a country you're in. And, but, you know.

AI assessment note: “And there's a sort of another macro thing in this.”

Q You'd mentioned, Jeremy, that cookies wasn't fit for purpose at some point, like, when we're looking at this wider model. So let's talk a bit about that, because that feels interesting, of what does a better advertising model even look like as we look at this, this whole scenario?

A The cookie was something that was invented, obviously, you know, way before most of this was actually considered, and so what you've got, what you've got is this sort of Uh, this sort of pyramid of stuff that's been built on the cookie and obviously, and now it's just not fit for the purpose anymore for what you, what you wanted today. So the, this concept of actually getting rid of the, or, or, you know, this modification of, of what Google wants to do, what Apple wants to do, um, or has done, um, is, is, uh, it's, I'm not actually sure personally whether that's what the consumer wants, the consumer wants. What I'm, what I'm convinced of is I, I, I don't think people don't want advertising, or don't want to be tracked. I, I personally think they want, just want, they want better advertising. They want it to actually work.

AI assessment note: “I personally think they want, just want, they want better advertising.”

Q the extreme of thinking that you thought That some information that you saw was correct, and you go, wait, hold on a minute, where did this come from? How did this get here? Um, and where does that responsibility lie? Is that responsibility lie with the platforms, or does it lie with us as users to become, you know? More informed and more educated on how to digest this content.

A I mean, this is, you know, going and following Ukrainian journalists, you know, I follow people who post pictures that I would be happy for my son to see, and then there are some of the people they follow, like you scroll up the post and suddenly you've got pictures of burnt corpses. And they mark, generally they mark them as this image is sensitive. Do you want to see this? Um, should they be blocking those? I don't think so. But then what does it mean to say blocking? Blocking for who? Is it blocking or not amplifying? I mean, this is like Alex Damos's point. There's a huge difference between blocking and not amplifying. Um, there's another, just, just shifting gear completely before we go too far down, you know, rabbit hole of non-competence. Um, the other thing I thought was kind of interesting was, um, Yeah, the story I saw this morning about Yandex. Um, so, a narrative that I've sort of, I think, point the narrative, a point that I've made a few times is that, you know, China is this sort of, um, you know, the Chinese consumer internet is its world of itself with its own bunch of companies, and they evolved in a kind of a different way. And so instead of having a Chinese Google and a Chinese Amazon and a Chinese Facebook, you've got sort of half a dozen conglomerates, you know, Tencent and Alibaba are much bigger than all the others, but then you've got a whole bunch of c…

AI assessment note: “just shifting gear completely before we go too far down, you know, rabbit hole”