The Wisdom Wall

“This is like the great fallacy of enterprise software that you can just sell to the users, and it's like it only ever worked for Slack and one other company. Maybe Notion, but even Notion has capped out.”

“Well, so I was looking at, I looked at Metaverse, sorry, at Fediverse, and saw, this looks like an open source claim of Twitter, and open source claims of commercial products never work. It's the only people who use them are people who care that it's open source, not people who care about the product, for the product.”

“And as the error rate goes down, you get this philosophical conversation, which is what gets you to AI risk, where people are saying, look, you throw another hundred X more compute and 200 X more data at it, and the error rate goes to 90 X percent, 99.98%. At a certain point, it doesn't matter if it has, doesn't have…”

“And so this idea, which is also a web three idea, which is that you could sort of have a decentralized social network and have lots of different clients all plugging onto the same database is kind of an illusion because what's going into the database is what the product is. And that's determined by the client, not the…”

“WhatsApp doesn't have an ad model. WhatsApp has massive problems with harmful content. They have massive problems with rumors in India, for example, and people forming mobs and killing each other based on rumors on WhatsApp. They don't have advertising. They don't have an algorithm. So stop fucking telling me it's…”

“there was this whole story, kind of, 1015 years ago in the Valley, that like, Enterprise software will become grassroots. It'll be bottom up, bottom up enterprise, bottom up sales. You won't have to go through the sales process. You won't go to CIO. You won't have to wait. They go through an 18 month sales cycle. The…”

“I don't feel like the reason it's hard to compete with X or Y or Z company is how long it would take you to write the code to replicate the product. It's almost never the problem. It's the problem. The hard part is everything else. It's working out what the product should actually be, and what the code should be doing,…”

“There are no network effects in building the models yet. There may be in the future, but at the moment we don't know.”

“what that's saying is the head of product isn't setting the product strategy and has no idea what the product is going to be in a month's time. Because there's going to be stuff coming out of the research lab that they don't know about, that the research lab doesn't know about, but they certainly don't know about.…”

“when the technology is basically undifferentiated, the product is undifferentiated, and you've got radically different market shares, that tends to be an unstable situation.”

“There's a recurring delusion throughout the history of software that firstly, people always think there's going to be a general purpose abstraction layer that will just do everything. And secondly, people always think, um, that everyone who doesn't write code will write code. If we just made it a little bit easier,…”

“There's no network effect that we know of yet. And so unlike previous, um, kinds of software platform shift, there's not any Obvious strategy in building the models where you can make yours better than everybody else's by doing that, following that plan. Well, you can hope for a breakthrough that no one else…”

“given that there's no network effects, or don't, or not yet any network effects in LLMs, that feels very fragile. Like, you know, if you remember, you know, I cannot, you know, remember, like, everyone used Yahoo, and then everyone switched to Google, because it actually was better. Everyone used MySpace, and we did…”

“Just because if you make something much cheaper, people will probably use much more of it up to a point. Although interestingly, there's a debate now as to whether mobile data traffic is actually slowing down. Um, but that doesn't mean you'll be able to charge much more money for it.”

“Every big company's got a trial. No one's really using it very much, or not many people are using it very much. There's a small number of fields where you had this immediate obvious use case, like it's great for coding, it's great for marketing, it's great for brainstorming.”

“this is, you know, classic disruption that the new thing tends to be crap at the stuff that was important for the old thing, but then it does something different that's new. And if you test an LLM by asking it to do complex logic problems or masquerade problems, or do kind of highly specific data retrieval, it's kind…”

“The general web search engine is a natural monopoly because you need a lot of money, and then once you spend the money, no one else can do it. So it's like a water company or phone company.”

“It seems to me like the Occam's razor is, if you can't guarantee it's right, then it's features. It has to be wrapped in something that can manage or control or shape expectations or something around what it is that this is doing, because you can't just use it for anything.”

“ask yourself why it is it's a lot easier to go to a company and say, here's a slide maker, than to go to a company and say, well, look, ChatGPT is just multimodal, so just ask what you want. Tell it the slides you want. No, you want UI. You want product around it.”

“the real Policy problem with TikTok is not that it has personal data, it kind of doesn't really, it's that it's deciding what you see.”

“if you think that you're going to solve this by hiring more women or more black people, you're kind of not understanding what's going on.”

“There's a difference when you do, something can be the same theoretically, just at greatest in principle, but a much greater scale and, and practical and accessible and everything else. And in fact, making something much greater scale and much more accessible, but the same in principle actually can be a difference in…”

“The problem, newspapers problem is not that Google and Facebook have a monopoly. The newspaper's problem is that the newspapers used to have a monopoly and they don't anymore.”

“One of the things that occurred to me is this is a sort of destruction test of the whole thesis, because if basketball doesn't work on this screen, even that, I mean, it's, yes, it's four grand. Yes. It's heavy. If you put the basketball on and after five minutes, you're like, this is shit, then we're done. Yeah. It's…”

“this, this stuff does, yes, this stuff happens for the sake of argument, but it does not happen in three months or three, it takes three years. The deployment for that is going to take, yes, the, the actually turning it into software, um, that's not just ChatGPT.”

“if you make something cheaper and easier and more accessible, then quite often you have more people doing it, not fewer. If suddenly anybody can produce a beautiful, can produce, um, can make a poster with fonts, that doesn't mean you have fewer graphic designers. It means you have more.”

“Whereas with AGI, we have no theoretical model for what AGI actually is, and we also actually don't have a theoretical model for how LLMs work. Other than at a very, very high, high level. So we don't actually know what will happen if we give it three X more data and four X more compute”

“Whereas for AR, it either looks like it's there or it looks, or it doesn't. That's the thesis. And so it has to have that base level. And so Apple is basically saying that all the stuff that's on the market now just isn't worth selling.”

“with a feature phone, internet was one icon on your telephone. Whereas on a smartphone, the telephone is one icon on your, on your computer. So kind of whole thing gets kind of pulled inside out. And that's sort of the AR VR thesis here that Apple has is no metaverse is one of Apple meta's metaverse is one of the 20…”

“this is a sort of a destruction test on VR. If this doesn't make VR Work. Never mind the price. Then that's it. VR's done. ... Like if this doesn't, when this gets to 500 bugs, if it doesn't work, VR is done because this is the device. There is no, there's nothing better that this is the device. So if it doesn't work,…”

“and you could say, well, if that gets fixed, then we've got AGI, but actually I didn't think that's true because one way to fix it might be that it could just say, I don't know. The problem at the moment is it, it might not know and give you make something up, or it might know and make something up. And that's…”

“Um, but in general, machine learning has become a commodity very quickly, um, which is sort of the absurdity of people saying, well, we have to ban face recognition, or you can, you can ban it, but that won't work. Um, cause it's just a commodity technology.”

“And it's not, now they actually are using the content and not sending the traffic to the newspaper. And so that whole newspaper argument that you are using our content goes from being dishonest nonsense to being actually true.”

“I mean, this is, this is kind of, you know, like the kind of the, like the first five minute reaction to GPT, 3.5 is this is a catastrophe for search advertising, which you kind of pull back and think, well, it's already not actually, you know, if I get, I can go to chat GPT and I can say, make me an itinerary for…”

“the open source fallacy should also give people a lot of caution, which is, The difference between what technologies think matters and what users would like and the problems that users have. And, you know, I don't have access to the source code of my phone is not a user problem. I am standing on the street at 10…”

“if I've now looked at millions of pictures of wizards and I draw something, do I have to pay them? Like, no. Um, if I copy your thing, do I have to pay you? Well, yes.”

“you can't just say, well, this is what copyright law says. And this is a derivative work, and that's fair use. That's not really what's, that doesn't, that doesn't map very well to what's happening here. It's kind of like taking live performance rights from the early 19th century and applying that to radio. Well,…”

“If you're worried about a Chinese app that could have access to device on the phone, then you either just ban all Chinese apps, or you try for a systemic, scalable solution across, like, well, you have privacy laws, you have rules that all apps have to follow, you have rules that all companies have to follow. That's…”

“this is effectively a way for Amazon to squeeze more margin. Out of its suppliers, um, as much as it is advertising”

“And clearly you should be more, if you look at it and you say, well, it makes cats, but the cats have four, have three legs. If you say, well, yeah, it can make some text, but the text isn't very good and it's got mistakes in it. You're frankly, you're an idiot. You're not really realizing what the kind of the…”

“Um, but the other side of it is, um, that most of what we were doing remained in three D, in two D. Text is two D. Pictures are two D. Video is two D. Most of what we do is two D. Um, and so the shift in character to move to three D is actually much bigger than the shift. You know, maybe what will happen, but it is a…”

“I don't think the direction of travel of software is bigger screens. I don't think the direction, or more fundamentally, is not about showing you more information. It's not about having more rows in the spreadsheet. It's having less rows in the spreadsheet and not having a spreadsheet and having machine learning do it…”

“Yeah, you're going to reach a point where, like, if you didn't want to be part of Adobe, then you shouldn't have taken the fucking twenty billion dollars. Which was always the problem with WhatsApp. Like, and yes, there were all sorts of things about broken promises and blah, blah, blah, but it's like, if you didn't…”

“The challenge is like, there's no network effect there. It's a budget effect. It's not like Facebook where everyone, it's not that, it's not like Facebook where everyone has to use WhatsApp because everyone's on WhatsApp.”

“in the end, it's still about the TV shows or the books or the music or the clothes. And I think that, that sense, it's a kind of a, kind of a counter argument to the sort of software reach the world thesis.”

“sometimes what happens is that the software is just a commodity, and it's the other part of the product that's the thing. It's, it's, it's the clothes, it's the clothes, or the TV, or the light, that's the thing. The product is no longer technology, the technology is just some, some part buried inside it.”

“with banking, you can just come and be the bank. And from the consumer point of view, it's relatively straightforward as a consumer experience for that to be your whole banking experience. Whereas from healthcare, it can't be because there's all these different things. Like you, as a consumer, you've got different…”

“It's like, you know, if I take a photograph, is the camera now the artist and not me? Well, no, it's me that decided what the picture was, and where to point it, and how to take it, and which of the 2000 pictures you took is, is the one. And, you know, I was talking about this on Twitter with somebody today, and I…”

“there is no theory of what that would be, like, a secure encryption means no one can read it, but the police can read it, ok so it's not secure.”

“the more that you try and make things open and standardised, And interoperable and portable, the more you're actually describing cable TV, not the internet. Because the whole fucking point of the internet is actually how little of it is standardized, as much as how much, depending on which layer in the stack you're…”

“as a conflict between interoperability and privacy, because if you say, you know, you can say, make it easy to explain. You know, the competition regulator says, make it easy to export data, and the privacy regulator says, make it hard to export data, because actually most of the data isn't your data. You know, if…”

“what the tech industry always wanted to do was to turn TV into YouTube or Spotify. That is to say, or what YouTube or Spotify then became, which is you've got this single universal library of everything, and you pay one fee, that's a reasonable amount of money, and not like 500 bucks a month, but you, and, and you get…”

“you're asking for a recommendation algorithm that will drive all the readers to you, but you're also asking to own your list, and I think that's kind of a contradiction in terms, because if what you're saying is, I want somebody to open the app and have the app suggest stuff to them, but I also want for me to own and…”

“Cryptocurrencies for want of a better term are not fundamentally about replacing gold and fiat currency and being a store medium of storage. That's just sort of the one of the early and obvious applications. What you're really seeing here is a way to run an open source computer.”

“And so there's a whole bunch of reasons why you tend to get simplifying layers and abstracting layers and aggregating layers. Um, you know, most people don't open a command line on their PC. They don't want to. They just want to click on the icon that says internet.”

“NFTs aren't about art either, just as crypto isn't about money. Um, the point, you know, what we're still doing is trying to work out what the building blocks are. And so the NFT might be another building block within how a crypto social system or publishing system might work.”

“Tech doesn't work like that. It hasn't worked like that since the seventies, maybe the eighties, but the days when like big companies could say, right, we're going to build the future. Like that's not how the future gets built.”

“The stock price going up does not help the company, except that it makes them easier for them to raise capital by selling stock and issuing stock options. But the price going up itself doesn't tell you that they sold more product. Whereas a token activity in the token is both people speculating and actual real use.”

“Telcos aren't a technology business. They're infrastructure business. Netflix isn't, isn't a software company. It's a TV company. On the other hand, Airbnb, but Airbnb, but Airbnb isn't a hotel company. It is a software company.”

“There's a very standard presumption in Silicon Valley that software companies can learn other industries faster than those other industries can learn software. And that's generally true, but. Well, it's generally true, but it might not matter. I think that's the kind of the crucial point. Your software might be…”

“a lot of like consumer tech now is as much pop culture as it is tech. You know, is Snapchat tech or is that pop culture? You know, that, the inside, whatever you think the answer was, it wasn't software engineering. It was some insight about how people wanted to feel.”

“I think Apple sees privacy as sort of analogous to malware. It thinks that this is an architectural problem on the computer. It controls the computer. They make the operating system. It's their job to fix it in the same way that it was sort of Microsoft's job to stop viruses. It's Apple's job to stop websites, apps,…”

“privacy is a strategic asset. Um, you know, there's an old line in Silicon Valley that everyone is giving somebody else's business model away for free. Apple doesn't have, it doesn't have a social network, doesn't have a search engine, doesn't really have a, it doesn't really have an advertising business or a very…”

“So the aggregate of all of this is basically the strong get stronger, the big get bigger. The advertising moves to the big, strong, established sites. Small sites, new interim sites, less targeted sites makes less money. Content moves behind paywalls.”

“if you say you can't buy reach, what you're actually saying is you can only have organic reach. So Trump has got however many followers he's had, so he's kind of fine. So, so when you say you can't do political advertising, but you can have organic reach, which means organic advertising, you're saying celebrities and…”

“competition conflicts, almost everything that we want to do here. Competition conflicts with privacy conflicts with everything else. You pick one.”

“The big strategic competition in these markets is not someone who's in your market that fits an antitrust lawyer's definition of your market. The competitive threat is from somebody who changes the whole fucking business into something else, or offers your customers a completely different way of doing the same thing.”

“Tesla, this is the argument. Is it a software company or a car company? If it's a car company, it's screwed. If it's a software company, everyone else is screwed.”

“Um, and so for Facebook, paradoxically, VR needs to not be about games.”

“the world in which all the ideas came from within 50 miles of Silicon Valley, well, that's kind of over.”

“If you wanted to make an Instagram competitor, I would like to see everything that you've posted and everything that you'd like, and I also want to see everyone you follow and everything they like, because that's what Instagram uses to give me recommendations. Oh, but that's not my data. That's their data. So how does…”

“The trouble is, as I mentioned, we've kind of run out of people now. So there's not like another step change to come in that, in that kind of process.”

“I would actually argue that Instagram is about people you choose to follow, and mostly people you know, whereas TikTok is about people you don't know. It's a completely different graph, and it's a recommendation-based graph, not a people you've met-based graph. And so actually, that's why I think Facebook bolting reels…”

“if Instagram was a separate company, that wouldn't have any effect on teenage girls looking at self-harm content on Instagram. If YouTube was a separate company, that wouldn't have any effect on the, um, prevalence of jihadi content. That's not a competition problem. It's something else.”

“if Instagram was a separate company, that wouldn't make it easier to compete with them. Because the network effects that make it hard to compete with Instagram are part of Instagram. They don't come from it, from it being owned by Facebook. The same thing with YouTube. Like, the reason it's hard to make another YouTube…”

“the enforcement is so disconnected from the original decision that, like, nobody who's actually Nobody who's making those decisions understands that in 10 years' time they'll be defined as a monopoly. I mean, they're arguing about it now. You know, there's people in Silicon Valley saying, what are you talking about?…”

“A couple of years ago, there was this whole thing in India of lynch mobs forming based on wild rumors that were spreading on WhatsApp, which was super interesting because WhatsApp is not, doesn't have an algorithm and doesn't have advertising. And so like those kind of common reflexive, well, it's the algorithm and…”

“the creation of bad stuff is fundamentally infinitely scalable, because it's people. It's four billion people saying, retweeting stuff that fits their prejudices. Um, and the content moderation or the virus scanning is fundamentally unscalable, because you're one at a timing it. You're picking one thing after another.…”

“the first 25 years of the consumer internet was basically American, and the big companies, you know, the agenda, and therefore the kind of the cultural attitudes and the legal attitudes to how they should work were all American. And you can, again, you can generalize a bit too much, you know, clearly everyone has their…”

“part of the point of regulation is that it's expensive and it helps incumbents. You know, if you, Facebook has whatever it is, 30, 40,000 human moderators. If you create a law that says if you have a social network, a human being has to look at every post, well, that means you can't create a new social network.…”

“Instead of spending all of your money on rent, you spend all your money on Instagram, unless you have some other way of getting, unless you have some other way of getting to people. But if you're just selling a commodity product, you know, in a commodity Channel. Guess what? Like, you're not going to make any money.”

“the fixed costs and the barriers to entry are gone. So you don't need to rent the store and get up on the ground of inventory necessarily. You don't need the kind of, the kind of, the initial expenditure might have changed completely. It doesn't mean that getting to a hundred grand of revenue is going to be easy. In…”

“if you're going to do Instacart, say, You can launch in San Francisco, and then you can pick 10 more, and then 20 more, and then 10 more, and then 30 more, and you can kind of go, go out step by step by step across the country. Whereas in Europe, you couldn't do like London, and then Paris, Barcelona, and Milan, and…”

“I sometimes describe things like, um, web summit and four years from now, it's being like gay pride. Which is to say, um, if you live in Stanford, you don't need an event to tell you. If you grew up in Berkeley or Stanford, you don't need an event to tell you about startups. But if you didn't, and if you grew up in a…”

“if you're a great accountant, or a great video editor, or a great graphic designer, or a great salesperson, or a great lawyer, those are different skills to being the person who's going to work out the right way to make a really great piece of legal discovery software, or a really great piece of sales enablement, a…”

“You're not the right person to see all of those problems. You're also not the right person to get everybody to use this. And so there's just this kind of huge gap between we give everyone a tool that can make stuff, and therefore everyone is going to make new tools.”

“writing the code isn't the hard part. Writing the hard part is knowing what the code should be doing. It's seeing the problem and knowing the right way to do it. It's the opinion.”

“The first of them is, well, clearly this is an order of magnitude, maybe several orders of magnitude, cheaper to make any given piece of software. A piece of software that you might have thought of 10 years ago that has nothing to do with AI, you can use AI to make that thing much more quickly, much more cheaply.”

“The underlying point, though, is I think the hard part of making software is almost never writing the code.”

“the fields where AI is actually works best right now are the fields where people are really angry about it and sure it's useless. ... Whereas the fields where it doesn't, really doesn't quite work, or needs an awful lot of other stuff around it, The fields where people are sure that it works right now.”

“memory is a feature and a stickiness, not network effect.”

“If you're using this all day, every day, you're not a typical user. It's like the paradox. Normally you would say if you're only using this once a week, you don't, you shouldn't be working on it. You should only be working on it if you're using it every day. The paradox here would be like, you should only be working on…”

“All AI questions have one of two answers. The answer is either no one knows, like how much better will the models get, for example, um, you know, what happens to, you know, does inference cost collapse and we no longer need any more in video chips? No one knows. And the other side of the question is, okay, it'll work…”

“It's my, it's my point about, about what do you mean by, when you say AI, what's the right level of abstraction here? Should you be talking about regulating AI, or should you be talking about, we don't want people using face recognition to decide who can get into a conference hall, which is not an AI question, that's…”

“These models are very good at automating certain kinds of things, but not others. They get stuff wrong, or rather what they do is they say, what would an answer to this question probably look like? That's about the best way I can come up with formulating it right now. And there are some classes of the question where…”

“There's this whole sort of thesis of bottom up enterprise software sales that you don't need to sell to the CIO. You make this amazing cool tool and the users will discover it and use it. And what actually happens is that most that that always works for like the first five percent of the market, but most users in most…”

“And so just being, like, 75% as good as Google costs hundreds of millions of dollars or billions of dollars just to have the whole thing. It's like, it's, it's, it's, there's a, there's a fixed cost to indexing the web.”

“Because a million dollars gets you developing, it does not get you deploying, even from, even at Accenture's rates.”

“So you need to have a use case where either it doesn't matter if it's not a hundred percent right, which is, you know, Give me 50 ideas for slogan for a new brand of toothpaste. Like there's not, you're not, there's not like a complete. There's no right answer or wrong answer to that. There's only better or worse…”

“And there's been like two things in the history of enterprise software. This has worked. I mean, one of them is Slack and I forget what the other one is. And what happens is you get to like five or 10% of your tan, and then you run out of the kind of people who will go out and look for cool new software. And after…”

“the EU is trying to say, look, let's just have like a kind of general, generalized rules that will sort of try and capture lots of different questions at once. Um, the problem is, of course, as soon as you get into how does Apple actually implement this in the app store. You discover, no, it's not simple, and you are…”

“So it's sort of easy to say, oh, well, you have to publish what your training data is, but that's not Going to produce something that's, that's not going to produce something that's easy for you to audit.”

“You re-run all of the arguments we had around content moderation and social media of what should people be allowed to say. You either say more or less anything, which is Elon Musk's idea, or you end up with a long list of, of, of criteria, which is where Facebook has had to go. But it's not going to be easy either way.”

“tech has basically not been a problem for the book industry. Um, you know, obviously lots has changed, and they've got Amazon as this one giant customer, but we did not move to a model in which everybody gets all the books they want, and all the books are digital, and the publishing model breaks up, and like the whole…”

“with all of these subscription models, you're capping your whales. Um, so with music, you had a small number of people who were paying a hundred dollars a month or more, whatever it was, and now they can only, for CDs, now they can get everything for 10 dollars a month for Spotify, but there's way huge number of other…”

“And what I wrote last summer was that Meta basically has the right price and is working towards the right device. And Apple has almost the right device and is working back towards the right price.”

“The right way to present text and pictures and data is in, is in two D.”

“ChatGPT is like a giant job creation scheme for lawyers and consultancies. Because the lawyers have to work out who we're suing, and what all the new contracts look like, and the consultancies, and people actually have to go and deploy all of this stuff.”

“I think regulating AI, quote unquote, is the wrong level of abstraction. To me, that's like saying, well, look, you know, FTX did all this fraud with spreadsheets, so clearly we need to regulate spreadsheets.”

“The problem with the Fediverse model is so much of the experience design has to be in the client. And if it's distributed and there's 50 clients, that doesn't work in the same way.”

“if you're trying to build a competitor to Instagram, I wouldn't want Tony Karen Brown's posts. What I want is who's liked Tony Karen's posts, Tony's posts, and what else did they like? And who liked their posts, and who else liked that? There's like these sort of concentric circles of interconnected, of signal, um,…”

“If you're building a recommendation system that says anybody can publish anything they like, and there's no gatekeeping, and there's no algorithms, guess what? You've got 10,000 things in your feed every day, and you can't find anything. It's like you're trying to build a perpetual motion machine.”

“The other half of it is the reason that you're downranked in the newsfeed is because people don't care, A, people don't care about you that much, B, they followed 50 of your competitors, and B, they have a hundred friends. And there's no solution to that.”

“the way I always used to talk about machine learning is it gives you infinite interns. You can get an intern to do something. Anything you can get an intern to do, machine learning can do for you, you know.”

“And so this point is a command line again. And this is Bill Gates's point that GUIs replace command lines. Well, you get, you've just given me a command line now. That's what, that's what a prompt is.”

“I mean, there's a bunch of use cases where it doesn't matter how good the model is, a chatbot is just not the right interface.”

“once you've won a space that has strong network effects, then the threat is not from something that's the same, but 10% better. The threat is from something that either solves the underlying problem in some completely different way or complete or unbundles you piece by piece.”

“trending topics and search are the two ways things that tend to drive dog piles. And tend to drive everyone rushing to scream at somebody.”

“speaking from a business perspective, news is a lot more trouble than it's worth. It's not actually particularly good for business because it's a huge amount of aggravation and toxicity and anger and problem. And it doesn't actually drive a lot of, it certainly doesn't drive any money and it doesn't really drive a lot…”

“All people who think that Elon Musk is brilliant, and if you point out that any particular problem or mention him in anything but less than sycophantic terms, they're all going to stay, they're not going to go to Threads. All the Nazis are going to go, I'll stay on Mastodon. All the Trotskyists are going to stay on…”

“In that sense, it's actually, you could kind of compare it to the Macintosh, which launches in 1984 and was over 7000 dollars in today's money. And most people still didn't know why they needed it. And I think Apple only sold like half a million in the first year or two, maybe less.”

“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.”

“to actually kind of make any questions about what companies are going to happen and how we're going to use this, you always have to kind of park the AGI conversation, because if it happens, then all bets are off.”

“The challenge is, um, as I always, I've always been saying, if you use this as a general search, then the error rates might matter and you can't tell.”

“as soon as you move away from make this shot better, you're probably moving away from a prompt box, and you're probably moving to buttons, and switches, and options, and palettes, and lists of possibilities, and a screen that says, okay, here are six different colour casts you could use, which one of them you want.”

“to me, this is part of the challenge of ChatTPT versus Google is that, um, Google says here are 10 links that might be the answer. Whereas ChatGPT says this is the answer, and it's, this is the answer. But it's, it's not, you could argue that what it should be saying is like, here is a synthesis of some stuff I read on…”

“anybody can buy a camera, but that doesn't make you a photographer. You know, anybody can fire up mid journey. That doesn't mean you're going to, you know, doesn't mean you're going to produce something great in the way that somebody who has the right sensibility is getting to produce something great. So I don't like…”

“the nature of art is that you can't copy it. If you can copy it, it's not art.”

“The theory that all of this kind of is on a track to AGI in a sense is that you can create AGI out of the collective Knowledge of humanity, which is sort of what GPT is trying to do. It's just to give it, give it all the texts that there is on the internet and like intelligence pops out of that.”

“It's making a pattern without any structural understanding of what the actual thing actually is.”

“This is a two by two matrix, which is, you know, does the error matter and can you spot the error?”

“the incumbent's always trying to make the new thing a feature. Um, and so that's very obviously what's happening with Microsoft and Google, including this in PowerPoint and office and of course in search and everything else. Um, but what, what, what actually happens is, is then people come along and make something new.…”

“if you're going to, you know, build a new system that will just tell you what the story was without linking or giving you credit, that's very different from, um, you know, what Google and Facebook do, which is you search for a new story and they tell you where to read it, and the newspaper is under the impression that…”

“you can't, you know, you cannot deterministically control what it says. So you can kind of put crude keyword filters on. You can count, you can put kind of crude, crude keyword filters sort of on top, but you can't guarantee that it's not going to you talk about some stuff that you don't want it to talk about. I mean,…”

“I don't get the impression that TikTok has that same mechanic of, like, I can build a permanent audience there, and that it's more that you are more dependent on the river, on being thrown up to the river, because, you know, TikTok, YouTube gives you all these thumbnails, and you can click which ones you want, whereas…”

“The purpose of the app store is to sell iPhones. I mean, it has revenue, and it has a cost base, but it also sells iPhones.”

“And if you're going to pay to be there, then the odds are that there's an ROI on that, and the reason there's an ROI is that If you pay to be there, people buy it. And, you know, that is actually a kind of a quality filter.”

“the invention is how to scale, but most of what it actually does is stuff that retailers were doing in 1950 or 1900, or even like 1860, 18, 70. Like, private label goods was always a hilarious thing. I mean, that, the freak out around private label goods has now sort of mostly receded. I think most people have kind of…”

“And then everything else, like everything that you pay below the gross margin line, um, is how is it that you get the product to the consumer? All of that is now sort of up for grabs. You know, in the past you couldn't say, should we open a city? Should we open shops in that city or just advertise there? Or should we…”

“And the point is, there's sort of some domains where, like, a 90% accuracy produces a fantastic result, and you don't see or care the last 10%, and there's other domains where the last 10% is a problem.”

“And so this is why you have GitHub saying, well, we're going to call this Copilot, and like, it's writing code for you, but you're, you're a coder, you can look at it and see. You can look at it and check. And so what are the domains where, um, You can tell it's wrong. Versus what are the domains where being wrong is a…”

“TikTok is not Different to YouTube in any sort of fundamental product sense. It's different in UX.”

“and the same thing with, with, with machine learning models, this is what we've always been saying. You know, get right back to kind of, I don't have any structural understanding of what it is. You know, they may recognize the difference, they may be able to tell a cat from a dog, 93.2% of the time, but they don't know…”

“There are two different classes. So one, which is, it's wrong and you don't realize that it's wrong. And the other is that it's right and you don't want it to be right. And you might not be able to tell in either case.”

“So, you know, you do a database or a financial model, you know, you've made a series of logical steps and it will follow those logical steps. Whereas with, um, with machine learning systems, you're saying do something that's sort of like that, and it will be sort of like that. Um, and so the, but, but the point is…”

“Well, it's, it's a programming language. You are feeding the statistical engine logical steps. You are saying double plus this, minus that, Conclude this, de-emphasize that, emphasize this, and you've ended up with a whole paragraph of instructions, which is actually a formula. You've given it an Excel formula,…”

“Very often what's happening is the acquisitions are being used in order to make yourself more competitive and to gain a competitive lever and to allow you to, you know, and to make the industry more competitive, not less competitive.”

“the competitive threat in tech is generally not someone doing the same thing as you, but better. Or, or, or, or let me kind of qualify that. Each market sort of goes for an S-curve, and there's like an early period where everyone's fighting it out to win. So this is like the, at the time when there were 10 different PC…”

“What you're, what you're actually getting here is say to me everything you know about this. Which is, it's actually, it's actually bullshitting.”

“I think there's a, you can, one can dismiss very quickly the concept, oh my god, this is like the end of art. No, no more than Photoshop was. Um, it's another form. It's another thing that you can do.”

“And I just think there's like, you know, rather like Shin, there's going to be dozens and dozens and dozens of these companies around the sort of the higher level point is everything about the channel and the go to market and the media filters through which a brand would get to market have been blown apart. Everything…”

“It's as though you open the internet, you get your TCP, it takes you a week to get your TCP IP working, you've found something you can dial up to to connect, it's taken your week of work, you've got it working, you open a command line, you ping a known IP address, it pings back. Okay, now what? That's sort of what it's…”

“If you trade FX, you will lose all your money. There are people who work inside, you know, very large investment banks and hedge funds, who make fractions of a tiny fraction of a percent Um, arbitraging FX spreads. Nobody else makes any money on this, and why should you?”

“That a smartphone is actually much less immersive than a PC. You're not sitting in front of it for three hours a day, just looking at that screen. Instead, it's something that you do in 5:10 seconds, you know, that you can pick up and put down. That it's with you all the time, but you can put it in your pocket all the…”

“Most people don't design cars for a living. Most people do not work in three D. Still pretty niche. Salesforce or email, or like most people are not actually editing, doing work that involves a three D object.”

“that's not how technology gets built. And it is sort of, and there's two problems with this. The first of them is like, you're not engaging with the complexity of why there are gatekeepers and, you know, why this stuff works like this. And all the reasons why, the reasons why the world is as it is will not just…”

“And the thing about, talking about this with the internet, like the internet is completely interoperable, At the lower levels of the stack with IP and HTTP and SSL and so on. But once you're in the browser, like if you're building Amazon, you don't need to worry about how Instagram works. Like there isn't one, you…”

“And so even if you think that VR and AR are going to work and are going to be the universal thing, think of anything you're doing now as pure experiment, as pure experiment, pure learning, pure interest, pure interest, with no expectation that this is going to be meaningful.”

“there's a whole class of content where human, human touched originality is not particularly important. Um, whether it's, you know, tourism pictures or pictures of sunsets or pictures of, um, you know, interiors or pictures of fast cars, or, you know, there's a huge amount of very generic content that gets huge traffic.”

“You could make a shitload more Mozart. You wouldn't, that wouldn't get you Beethoven. And you might stumble on Beethoven, but you would know, it wouldn't know it was Beethoven to do more of that. And it would be buried in the other billion things that you've done.”

“it does less than Photoshop or Illustrator on one axis, but it has this whole bunch of new stuff on another axis and so pulls this whole new market in.”

“Facebook has struggled a lot more to get reels working because that's actually a different graph. It's not just another thing you would share with your friends. It's a whole other graph of different people.”

“it's really difficult to add that real time, multiple people working on it, multiple people collaborating at the same time to like a 50 meg binary file that was never intended for that and software that was never built for that. And so you'd almost kind of have to go off and build like a whole other thing, at which…”

“the TAM for the collaboration around the video edit was actually much bigger than the TAM for the video edit.”

“all questions that matter for Netflix are TV industry questions. What rights? What shows? What happens as the incumbents pull their shows off Netflix? What budget should they have? What stars? What kind of, they're all TV questions. There's no technology questions here.”

“You know, as a VC, if you didn't look at the crazy maniac who incinerated a bunch of money, but also built something really valuable, if you didn't look at his second thing, you're kind of not doing your job.”

“it's not Internet of Things, it's things connected to the internet. And they don't actually need to talk, they don't actually need to talk to each other or communicate. Some of them, in some cases, that might be useful, but most of the time it kind of isn't.”

“The hardware part mattered much more than the software part, and it's a classic case of what Clay Christensen called sustaining innovation, which is you get absorbed into these in the existing industry.”

“It's not about putting medicine in boxes and shipping it to you. Even the medicine part isn't about, isn't about the putting it in the boxes part. It's the, what are you paying? And why does insulin cost a thousand dollars in America and 50 bucks everywhere else on earth? Well, that's not a parcel. That's not a…”

“And of course this didn't really work because That level of primitive was actually better done inside AWS or inside Azure rather than as a standalone startup. And B, if you're like Caterpillar or like Chase, you can't Chase, like you can't buy natural language processing. You need several levels more of abstraction.”

“all of the systems that we're looking at now are no more intelligent than a calculator. You know, a calculator can do mass much quicker than you, but we don't think it's intelligent. It doesn't know what the mass is.”

“Is this is the privacy company that thinks they've gone through all this elaborate, complicated work to do it in a way that's private, and they get this huge backlash, which says to me, like, we don't have a coherent sense of what we mean by privacy, and so we've thrown away the old way of doing online advertising…”

“There's another point, which is one I've made in a different context, which is, yes, but as a brand or retailer, now you're looking at your ad budget as interchangeable with your rent. So you're asking the question, should we rent more stores, or should we run more ads, or do free returns, or cut the prices, or do…”

“Um, and so I think that, you know, there's sort of a grounding point, you know, if in five to 10 years, we are all using consumer services or business services or, you know, infrastructure or something that's all running on blockchains, which is basically what we mean by web three. Like you can build Instagram on a…”

“I mean, it's like, but it's like, it's like saying you can connect your Tinder with your LinkedIn, with your Instagram. They're not the same thing.”

“You can take some on lower level components back and forth, or high level components, but you're never going to have all of them completely interoperable, because then they wouldn't, well, then they wouldn't be different things.”

“The challenge is at this point, we're sort of designing social networks in 10 years time. Which kind of comes back to my, you know, the more we spend time we spend saying, so this is how I think it's all going to work. The more you're like, we have no idea how this is going to work. So maybe don't worry about it quite…”

“In the sense that sort of they become the same, they become aspects of the same problem where they used to be separate parts of the company. Um, so, I mean, the way I always phrase this is that in the past, you couldn't say, should we open stores in that state or just advertise there? And now you do. Now it's all one…”

“there's a sort of diminishing, not just diminishing return, but almost kind of a negative return to scale, that after a certain point, adding more TV shows, actually make, does, certainly doesn't make the experience better, and might even make it worse. Because now you just don't know that it exists, and you'd never…”

“Because if you're trying to solve the blank canvas thing, that, I don't think you could do that at the protocol level. I think you need to do that at the UI level. So you need everyone using the same UI.”

“And when you have hundreds of millions or billions of people using your product, Like, you need content moderation because otherwise just group dynamics means it turned into a chat. It just turns into a world of shit. And people, anybody who's really worked in a social network understands this.”

“that the reason why you've got this furious innovation in messaging apps, and so many of them that can do different things and behave in different ways, is because they don't have to be interoperable. And if they had to be interoperable, then you couldn't do half this stuff, because you'd have to wait for everyone else…”

“Apple is now, you know, has a credit card and the sort of talk that it's building a bunch of sort of financial infrastructure. It's not doing that because it wants to overturn the existing market structure of banking. It's doing that for Apple reasons.”

“in a web three model, you as a, you know, you would own, you would own both who subscribed to you, and you would also, in a sense, own what you subscribed to, because it would, It's sort of transparent and composable. You know, any bit of it is addressable. You can point it any bit at the stack and own that bit.”

“And yes, the people who followed me on the old system will see me on new one, but the other half of it is a new one has got to have billions of users because Otherwise, in a sense, why am I going there? If not, because it's, you know, question A might be I don't like the content policy or I don't like the monetization…”

“most of what Instagram knows about you isn't your, your data. It's who liked you and what else they liked and who liked those people and what else are the things those posted and which, which of your pictures they saw and didn't like, um, or didn't press like on and what other pictures were in their feed. And so…”

“what you see when the Amazon screen loads is not just a database output, there's an awful lot of work in that, and so, sorry, I'm being very incoherent here, but like, I don't think you can separate the client from the data like that. Certainly in, on the web, in, in, in, in, to continue this point, it's not really…”

“the visible complexity goes up in a kind of a bell curve. And then you reach a point by like, you know, you get GUIs and Windows and Mac and so on that the visible complexity starts going down. Even as though you carry on getting more and more plumbing underneath. And same with the internet. And sort of where Bitcoin…”

“Well, the day, this is a dangerous thing, because this is what you've got with COVID, of people saying, go off and do your own research. No, you do actually need the intermediary layer.”

“a big piece of not knowing where this is from is, you know, there's a huge amount of selectivity in the images we see. So we see, we're getting images from the Ukrainians. We're not getting images from the Russians. Partly because the Russians wanted to pretend that nothing was happening. Partly because the only…”

“It would allow you to build, not just to build a product in a different way, but build a different kind of product, which are two different important points”

“open source was hugely important in creating technology infrastructure and plumbing of every kind. And it was a much better organizational model for writing that software. It was not a great organizational model for writing end user software for consumers.”

“And crypto is basically sort of SaaS meets open source. So it's all sort of running publicly in the cloud. And so you can make different applications with this. You can do these incentive structures and open governance structures and portability that you can't do with a database.”

“there's a kind of a dumb criticism is to say the fact that we don't have email and the web and so on yet for, for blockchain tells us we never will, which is just dumb. And there's, oh, well it happened in the first 10 years for the internet and we've been 10 years in blockchain. Like, why is that the timeline? Maybe…”

“The TAM is not the advertising budget. The TAM is the whole thing, you know. Now, the counter to that is, of course, it works both ways, because your advertising spend is competing with rent, not just the other way around.”

“Technology isn't doing DTC soap companies. It's soap people and marketing people do DTC soap companies.”

“And in the same way that like the second 50 years of the car industry wasn't about cars, it was about Walmart and McDonald's and suburbia and freeways. And, you know, the fact that everyone had a car.”

“Netflix, all the questions are TV questions, like what shows, what light structures, what talent, how do they compete with Disney, what's HBO going to do, you know, what's their share of production in the UK, what happens if France starts regulating them, not that anyone cares about the French market, but, you know,…”

“Google has to have a bunch of compliance people. Okay. That doesn't change anything for anyone in Silicon Valley. It's just compliance people. It's just cost. It's just money on the EPS.”

“there's such a separation between the stuff that tech is obsessed with building and the tech that people outside of tech are obsessed by, which is Shopify. Like, what's the company that everyone outside tech is obsessed by? Shopify, not Naked Ape or Bored Ape or whatever it's called. Like, that's not what the center…”

“that retail that that retail tv and so on um advertising and now being completely overturned by stuff that people in tech were obsessed by 15 years ago”

“you get a whiteboard and you write VR, AR, social, crypto, place, tradable goods, creator economy, and you draw a box around this and above the box you write metaverse. Like, okay, but that doesn't mean it exists or you can buy it or invest in it. It's just a way of thinking about a whole bunch of stuff that's already…”

“The interesting thing is, of course, the abstracting and the simplifying means it gets more centralized, almost inherently.”

“vast majority of people in tech, this doesn't really affect how they build their business, or it will be something like GDPR, which is just kind of a pain in the ass, but it doesn't actually change what your business is or how it works.”

“even where you don't have, like, you know, conscious purpose behind this, you know, by default, any Western company will have assumptions as to how stuff works that will not necessarily be, and tend to raise the standard of behavior in a country that doesn't have those assumptions, because that's just how it works.”

“what they've sort of done is modularized, modularized the way that they roll out the hardware and the software in such a way that they can get that Predictable, repeatable thing happening every year without delaying.”

“Because I'm investing in this game, and this is a thing that we're going to own for the next seven to 10 years, which is venture capital's investing horizon. Whereas if it's, you know, if it's, if, whereas if games work more like movies, and I'm going to invest in this movie, and like, that doesn't work for the…”

“And so this is why WhatsApp and Skype basically had no effect on telecoms industry or, or had a very, um, second order effect on the telecoms industry because they had to change their pricing models.”

“once everything calms down, and you build sewers, and you build, have a police department, and you have a social safety net, cities are great. But you do kind of need to build the sewers.”

“Okay, well, that's fine, but there's 20 Chinese companies that are already making them, and there's nothing you can do about it. So you're just going to have to get, think about what that means. In the same way, like anything could be fake, Google can't stop that happening, so we're just going to have to get used to…”

“there is that sort of slightly interesting sort of actually quite narrow view in Silicon Valley of what you can and can't copy. Um, and also it's only goes on some axes, even in Silicon Valley. So no one would say like, you know, if you're doing a two-sided marketplace for oilfield workers, instead of doing it for taxi…”

“the sort of text structure in The rest of the world is you tend to have vertical winners. So Google won search, and there's a couple of other people trying to do search. Amazon won sort of generalized e-commerce, not market-specific e-commerce, won general e-commerce. Facebook won the newsfeed-based thing, and then…”

“never invest in anything called virtual, because it's always going to be like a crap copy of something in the world rather than like a real new thing.”

“the problem is, like, all search grows until you need curation, and then all curation grows until you need search.”

“having privacy, being the privacy company is a huge strategic asset for a large class of the kind of car products that they want to build, because those are products where privacy is actually much more material than it is even on the iPhone.”

“the money that was going to newspapers, a lot of it just isn't spent at all, or it's spent income not on Google and Facebook, but in other places at a 10th or a 100th of the price.”

“I sometimes say that the tech industry has Tourette's and instead of people chatting swear words, they shout like bots or voice or Autonomous cars. Yes. Clubhouse. Clubhouse. Clubhouse. Um, and certainly that's, that's, that's the word of the moment.”

“right now Shopify is, is they're at the tool level. And they're moving, they have a network in the apps, but they have to move up to have a network on the customer side so that, you know, there's a, that creates that much more reason why as a vendor you would want to use that.”

“Clubhouse needs to build a network, otherwise they can't sustain 10%, and of course Twitter has just bought a newsletter company to try and do that for you. Um, and charge five percent instead of 10%.”

“The new thing about TikTok is the swiping and the recommendation. So it's the, it's, it's the, you could call that a network, but it's the mechanics that you build around how you show people what the video is. As opposed to YouTube, which is a basically a very desktop, very like 2005 of course, of course, because it's…”

“unfair comparisons are the only kind that matter. I may have said this when we talked about the, um, talked about charts, because the fair comparisons, they sort of imply that you're playing the same game. It's the unfair comparisons that fuck you up.”

“the theory of Amazon is that everything is an API. Everything is autonomous. Your financial metrics are visible internally. Your business metrics are accessible through an API. So when Germany does shoes, they hire five people, The German shoe team plugs into the logistics API and the e-commerce API, and they don't…”

“you should split e-commerce by the logistics model, not by the product category. That is to say, instead of saying books are totally different from shoes, you should say parcels are different from stuff you have to go and collect, or stuff that has to come to your door in a refrigerated truck, which are also different…”

“there's this cliche that e-commerce has infinite shelf space, and for Amazon it's more that there's one shelf that has infinite length. Everything has to fit on that one shelf. You can't have different kinds of shelves for different things. It all has to be the same shelf.”

“Shopify is like, like the standing rebuke to the idea that Amazon swallows everything. It's a completely different model.”

“the first 50 years of tech, of cars or aircraft were about making car companies and working out what cars are and working out what aircraft are, and the second 50 years was everything that happened once everyone had a car, and everything that happened once air travel was cheapment. Universal. And maybe that's where…”

“Sometimes a hit is just a hit. Sometimes a hit is not the harbinger of a fundamentally new industry structure.”

“It's not Facebook can turn the Facebook messenger into a universal account application that subsumes everything else on the store. It's yes, if you were to work out the right way of adding three things to Uber, that might work.”

“a lot of social media is basically pop culture. I mean, Cameo is pop culture. Is Cameo a tech company? Well, not, not in any meaningful sense. Yes, they've got to execute the tech well, and the site can't crash and the app has to be good. But Cameo is based on an insight into pop culture.”

“there's no such thing as data, because you have to ask, well, what do you mean when you say data? And there's not one answer to that. There's, like, 50 answers to that, and then there's no such thing as data. There's 50 things.”

“Siemens has lots of telemetry data on wind turbines, and General Electric has lots of data, telemetry data on aircraft engines, and Transport for London has lots of touch-in, touch-out data on Oyster cards, and those are all different things, and you can't kind of put them all in a pot and get something else out of…”

“So you have this data that's valuable only in the aggregates of millions of people. Not valuable to the individual user or meaningful in any way to the individual user.”

“I send you a message. Is that your data or mine? It's both, motherfucker. You send me a GDPR. Yeah, someone sends a GDPR request. You have to delete all that data. Yeah, but that's not your data. It's also my data.”

“And so they've always been very, they've always been actually been very asset light in almost all of this stuff. It's only, it's A, retail, and B, chips. Those are the places that they push deep into assets. Because that was where you could deliver something fundamentally different, as opposed to you just doing what…”

“there's a big difference between technology that's hard and technology that's differentiated. And so I think right now what we're seeing in TV streaming TV is that video compression and distribution and getting great streaming video onto your screen is hard, but it's not differentiated.”

“There is vastly more third-party software from vastly more people being installed than ever before on this closed sealed box walled garden.”

“Well, it's a private corporation that controls how billions of people talk to each other. If they decide that billions of people can't say that, that is ipse fact to a free speech question. Doesn't matter that they're not the government or not.”

“Now you could argue that's, that means that you should just break up Facebook, but that just relocates the problem because it's now it's five people collectively who are making that decision instead of one, but the decision is still there.”

“Now, a really obvious one that you see in technology is we want more competition, and we also want more privacy. Okay, so you want to make a new social network, then the obvious competitive tool would be export your friend list and everything they're interested in. Ah, you just described Cambridge Analytica.”

“Controlling Windows doesn't give them dominance of the tech industry, and so what happens is, like, and it's the analogy that I use here, you've built your castle, and it's absolutely impregnable, and sometimes people capture your castle like they Tunnel under the tunnel. They, you know, they, they break down the first…”

“US regulation tends to proceed in criminal cases. They tend to look, they tend to say, did you break the Sherman Antitrust Act of 16, 21, or whenever the hell they passed it? And yes or no, and if yes, then we will do this thing. And 20 years ago, they tried to do this to Microsoft, and in fact, part of the problem of…”

“Yes, people will buy fashion online, but not with a grid of dresses. You need some, you need something about service, experience, discovery, recommendation, all the other sides of what retail are, retail can be.”

“all lists grow until you need search, and all search grows until you need a list.”

“you know the counterfeit is, is really good because it's coming from the same factory as the genuine ones. It's just running it at night as well as a day, so it's not actually a counterfeit. It's only a counterfeit if you're an accountant at Nike. Otherwise it's a shoe. It's a Nike shoe.”

“And then you have Europe where the markets have historically tended to be too small to create global giants. And they've also been easier for global giants to get in. And so in each country you have sort of one or two or three domestic champions that have won that market.”

“we're just seeing systemically more billion dollar, five billion dollar, ten billion dollar software companies because Because the market opportunity is now not selling to 50 CIOs of big companies. It's selling to every CIO and every company and every consumer. And so there will just be vastly more software, vastly…”

“clearly the deployment of mainframes and data processing and spreadsheets and ERPs and so on Did kind of completely change the economics of your business, and that came with a whole bunch of challenges, most of which were really internal management challenges, not operational challenges, not engineering challenges.”

“We'll all get some cost savings out of it and some efficiency gains that will get competed away in price because we're all doing it. And at the end of the day, as long as you don't screw it up, this doesn't change the competitive dynamics. Very much.”

“Jesus Christ, it takes three to six months to get somebody at an enterprise to reply to your inquiry about selling them some software. You know, the enterprise sales cycle is what, 18 months? At best? You know, you want to add a new thing and plug it into SAP and Stripe and Workday and your payment system, uh, we're…”

“all the questions I get are going to fit into two categories. The question, the answer is either no one knows, or how did it work last time?”

“There's a point where you do it in Excel, and there's a point where you buy a thing, because you want to know that somebody's thought about how it should work, and somebody is responsible for it, and will fix it, and there will be bug fixes and security.”

“Slack had a network effect in that, um, One team would use it, and then someone was working with that team, so they would use Slack, and then their team would use it, and then they, so it would kind of spread organically. There were very, very few pieces of enterprise software where that's actually happened, and where…”

“there is a stage in the S-curve in the development of a new, big new technology Where, like, at the early part of the curve, like, it's suddenly very obvious that this is going to be huge, and everyone's very excited, but you don't know how any of it's going to work, and you have to know that you don't know how any of…”

“Because that's kind of what happened with cloud. For example, there's only three hyperscalers, but, and yet like most of the applications are built by other people and most of the value is captured by people who run on AWS.”

“So the more interesting comparison or the more relevant comparison is mobile because mobile does have marginal cost. Like if your mobile traffic doubles, you kind of have to go out and build, not exactly double, but kind of double the infrastructure.”

“power isn't sophistication or complexity or doing clever things or being impressive. Um, power is the ability to make people do something that they don't want to do.”

“you go through a platform shift, and the first step of the platform shift is you do the old stuff, but more.”

“Because what generative AI does is it says this is what a good answer probably looks like. This is what most people would probably say. This is how most people would probably make it.”

“there are some cases where internet is not more efficient, like groceries, which is why Walmart is still a giant business, because it's actually more efficient to drive to the supermarket Than it is to deliver all of that product with the cold chain, and the stuff that bruises, and everything else, and deliver that to…”

“The whole point of a stack is that it's abstracted. Like Cisco didn't get much say in what the websites were, even though it might all have been running on Cisco readers.”

“You have no idea what McKinsey does or what the clients are paying for. They're not paying for the slides.”

“if you've got a piece of software, all you're, all you're really doing is wrapping up SQL and putting a car logo on it instead of a bike logo on it so that it's car software instead of bike software, then yeah, you're screwed. That's what happened with cloud too.”

“So if you're going to make your own experiences, then you are competing with the whole of Silicon Valley and the entire tech industry in trying to invent new experiences on this technology.”

“every big company now has had, like, 10 AI presentations. They've had one from Microsoft and Google, they've had the one from their ad agency, they've had the one from Bain or BCG or McKinsey, they maybe had one from OpenAI, they've had stuff from Accenture, and they probably deployed a bunch of stuff, and they've…”

“It's the entrepreneur's job to work out what the technology is for. And generally it takes 10 years of entrepreneurs failing before they work out what the technology is.”

“are you willing to take, um, You know, a 20% cut in model quality for a 95% cut in price is a very, very steep cost curve here.”

“the incumbents always try to make the new thing a feature.”

“things go from AI to smart To just software. There's a period in between where it's smart, and then it's just software.”

“it's hard to tell a computer how I recognize a cat, and that's the same category of problem as it's hard to tell a computer why I like that suitcase, or why I like that hotel, or why that's a bad flight to book.”

“influences unbundling for magazines and TV, but mostly for magazines. And of course, LLM then, you know, ChatGPT goes off and reads the whole web and synthesizes it for you, so it's bundling all the influences back up to the top again”

“the UK and the EU now have laws requiring, that basically require you to have what most people would consider like best practice content moderation, which is basically you've got to do what Facebook is kind of doing anyway. And applying it to country companies that operate in their territory in a way that effectively…”

“It might be giving, it might have internally created a category and given those categories of people, lower risk and lower, lower insurance prices, and those might be women, but it might, it wouldn't have a concept of women and you might not know if it's doing that. So you've got these kinds of challenges of what…”

“these systems are basically pattern matching pattern recognition systems at a very, very simplistic level. There might be stuff in the pattern that you want it to use. There might be stuff in the pattern. There might be pattern that you didn't know was there. There might be patterns that are important and useful. There…”

“And if there are bugs in the computer that are ruining people's lives, then that's an institutional problem, not an engineering problem. I mean, it's an engineering problem too, but the backstop has to be a willingness to consider that the computer might be wrong.”

“What they can do, the intervention point is in the distribution. It's not in the creation. Can't stop people making those images. You can think about how they go viral, how you distribute them, what the, what the, the source is.”

“You don't need to make a fake image. The fake is in the, the fake is in the caption. There's not a shortage of, there's not a fake is in the. Yeah. What is it that I'm saying? What is it that I'm seeing here? So, you know, if you want to make a picture of an airstrike in Gaza and say, oh my God, look, the Israelis have…”

“what is generative AI trying to do? Which is, is not actually trying to reproduce any individual thing in the training data. So, you know, when we did image recognition models, 10 years ago, you give it a billion pictures of cats. The purpose is not that it can recognize any individual cat picture. The purpose is it…”

“Newspapers uniquely, I think, you had this revenue model that was completely unrelated, bolted onto the news, which is classified ads, which people think means, um, like individual consumers, but it actually means like car dealerships and real estate agents and so on. Um, so you had this advertising model that had…”

“Sky realized that they could pay massively more than anyone had ever thought reasonable to pay for football, and put that on a, this new channel, which is a subscription satellite channel, also subscription cable, but pay free, premium subscription TV service, and you could use that to pay for these, These rights. And…”

“So that, that random access, the ability to access anything that's ever been created instantly has huge value for music. Almost zero value, effectively zero value for books.”

“I said, in a sense, all AI questions come into, fall into two categories. The answer is either how does every other platform shift work, or we have no idea.”

“it's not the customer's job to know what the technology is for. It's not the customer's job to have be given a ChatGPT account and told, there you are, go off and reimagine how your company does its backup of its processes.”

“there's a classic economics formulation that, you know, efficiency gains tend to be competed away. So everybody, if you and all your competitors deploy this, is the end result just that you'll have roughly the same margins, roughly the same business.”

“we are certainly having at least one clear major platform shift is now underway, um, around how we think about what generative machine learning is, just as in a sense, at a kind of a bare minimum, we also had that 10 years ago, as we thought, what is this machine learning thing? Except that it's happening much quicker,…”

“we don't have any kind of equivalent mathematics or any kind of equivalent chart that we could draw that says, well, AI is there, and we're here, and LLMs are in this place, and this is how fast it's going, and if it grows at the current rate, it will hit human level intelligence on the 17th of May,…”

“we haven't got to the point that we've got a mental, even if you think it's a platform shift, we haven't got to a point that we've got like a kind of a useful mental model of what would, what is this? Whereas for the last wave of machine learning, we're now kind of very clearly at the point where you can say, okay,…”

“The analogy I used to use for the last wave of machine learning was to say, this gives you infinite interns. So you want to listen to every call coming into the call center and tell me when the customer's angry. You want to look at every credit card transaction and tell me the weird transactions. You want to look at…”

“It's all about understanding how the car industry works. It's not really a software question. Um, and that's what happens once technology has kind of matured, and it's kind of passed on, passed on the torch to the industries that it's overturned.”

“once you've got a fixed standard, and everyone has sort of signed up to it, it's very difficult to innovate on top of that.”

“when you're using LLM, it's a kind of call and response model. You type the prompt, you press go, you wait, And something comes back. I compared it to playing battleships. You know, you call out your move and you find out if you hit anything and then you wait and you start again, as opposed to, as opposed to working on…”

“the obvious place you can see this now is with Hollywood and streaming, where yes, Netflix is a participant, but Netflix is a participant as a TV company, And all the questions are TV questions or maybe questions of rights and production and, you know, what writers should get paid and how production should work.…”

“And I was going to say ChatGPT has that sort of question, which is there's a whole bunch of stuff that theoretically you could ask it to do, but actually you want user interface and you want buttons and you listed, you want the list of the five previous queries and you want this parameter, that parameter. You want the…”

“Paradox is theoretically it can do anything, but you don't know what you can do. So you want a list. Theoretically, um, um, you can ask it for whatever you want, but actually there's the five things that you found that work for you.”

“What I tried to do is build something, build what you could say is like a reasoning engine based on having an enormous amount of material that's the output of human intelligence.”

“And it says we don't have an Excel yet. You've got the PC as this sort of substrate. Suddenly everyone has a computer and it's only, it's amazingly cheap.”

“with chat GPT, the lateral language generation is so good that you can't see the model behind it. It isn't as good as the text. Like the prose generation is perfect. The grammar is perfect. So that hides the fact that the understanding behind it isn't quite there. Whereas when it's making images, it's easier to see, it…”

“if you want to build a big, complicated thing, there is not going to be just one button that says go that produces a very specific, complicated thing you had in mind.”

“And so whether it's that or whether you were following all of your friends, that doesn't necessarily matter very well. To the kind of text that you might want to see, because it, you know, when you follow people in a text-based format, you might want to get people, see people talking about stuff, and that may well be…”

“When Quora first started, there was this moment when you'd have, like, really important, influential people chatting on it all the time. So you'd have, like, Steve Case telling stories about shipping DVDs and CDs for AOL in the nineties. And then, but then as it grew, and you got lots and lots of noise, they all left…”

“images are in this seem, you know, like a much easier thing to do than conversation. Like I'm going to share some cool pictures. That's a lot easier than, than, than getting discussion and insight and stuff going.”

“This is sort of the dream VR AR device from five years time. But today, which is, or rather next year, which is sort of why it's five, three and a half grand and not 500 bucks.”

“they've kind of done what they did with the watch. And what they did with the iPad, which is make a list of 20 possible things that might work, try all of them, and then later on we'll double down on what, what works.”

“So the default for Oculus and for every other VR device, and that's partly because of where they came from and what was possible 10 years ago, the default for them is you put on the headset and you're somewhere else. Whereas the default for this is you put it on and you're still in the same place, but now stuff can…”

“It's going to be really tough to make something that's based around hand gestures and eye tracking if you don't have either of them and you've never really experienced either of them.”

“There's kind of comparison with the iPad, the iPhone here, which is that I used to say the iPhone is just a piece of glass and it can become anything. It's part of getting rid of the buttons is it became totally unspecific. It's just a piece of glass that could be anything. And again, Apple is kind of thinking about…”

“Well, so the sort of meta observation we were chatting about earlier is that this is moving massively faster than most recent things, most recent things in tech. Cause no one needs, you don't need to buy a new device. You don't need to learn any programming language. Like, well, the whole stack is kind of there and…”

“But the idea that like it might cost, it's like it actually has more, a consumer internet company has real tangible marginal cost every time a consumer does something. That's kind of a new thing.”

“the more narrow the product get and the more expert the user, then the more visible the app, the errors.”

“you can try and stop it, but people will find ways of getting around it because it's not just doing a binary database lookup. So you might have a keyword filter on Napalm and people will come up with a way of describing Napalm without using the word Napalm.”

“asking and asking chat GPT for a source. That's not what it is or how it works. Um, because one answer is the source is the internet. Two or 300,000 things. The second answer is that when you've said, please give me a source, what, what these systems do is they give you things that look like answers.”

“The idea that it actually costs money to press enter on a consumer internet site, like posting a pic, you're actually like a Facebook if it'd cost money to do a post. And you had to wait 30 seconds for it to appear. Well, that's kind of what ChatGPT is now. Like it costs money to ask a question and it takes 30 seconds…”

“So, you know, if, if VR and AR work, VR and AR might or might not work, that has nothing to do with whether Generative Machine Learning works, and very little, a little bit, and very little to do with whether Web three works.”

“the kind of creative energy around, um, generative machine learning sort of comes from the fact that, like, you've got all the stat ready, so you're basically free-riding on top of, not free-riding, but you're sitting on top of cloud, the web, broadband networks, fast compute, all this data, all this cloud computing,…”

“Whereas it seems to me the kind of creative energy around crypto was sort of, like, it was more like the creative energy of the internet in, like, the late eighties and early nineties. In that there was no HTTP. There was no SSL. There was no SMTP. Yes, there was SMTP came earlier, but it's as though you didn't have…”

“it used to be that when you did a TV show, if it became a hit, then it would go into syndication, and so you had kind of a lottery ticket, and this is why Jerry Seinfeld is worth half a billion dollars, because it's all the syndication money, but a Netflix show doesn't have syndication. It's a one-off payment. They own…”

“the point is that, that images are both much higher bandwidth, but also much higher noise. And so the mistakes can be kind of hidden in the noise that, like, the clouds aren't quite right. If you're not a meteorologist, you don't know, and you don't, and you also, and you don't know. Um, but also with text, um, because…”

“creation of the stack with the enthusiasm for, for, for, for crypto, whereas creation at the top of the stack is the enthusiasm for, for, for, um, for machine learning.”

“People can come up with phrases to circumvent, um, keyword censorship, and of course very quickly the censors catch up to it, and so it never lasts, never, never lasts for very long.”

“there was an observation for a long time that Pinterest was sort of like invisible in Silicon Valley, because basically like 80 or 90% of the user pay, it's in New York. Is it New York? I think it's in New York. Yeah. 90% of the users are women. And so like, and, and it's all about visual collections and stuff. And so…”

“Clearly both Amazon and YouTube have unbundled Google in that sense that you go to Amazon for product and you go to YouTube for not just for quote unquote video, but for, for a lot of other stuff that might previously have been text.”

“And the point is the recommendation flow breaks the context.”

“It used to be that, that, that cheap was high. Cheap has gone from high to low. Best has gone from low to high. And the story, the hypothesis here is that people used to use the internet for a price comparison. You already knew what you wanted. You went to Google and searched for cheap X. Over time that's gone down and…”

“But if you go to ChatGPT and you ask it a question, it gives you one answer, and it presents that in natural language, which looks, which creates this impression of certainty. This is this phrase, it's a confident bullshitter. It might have exactly the same degree of certainty as the Google search engine. It's just the…”

“What the four images and the 10 links are doing is communicating uncertainty, whereas the one answer that's in text presents itself as the, as the certainty far more than you get with, um, than you get with the others, and”

“Will I be able to tell the difference running Keynote and a web browser and email and Excel? No, I will not be able to tell. Would I be able to tell running Stable Dive Fusion? Oh, suddenly I get that image back in 10 seconds instead of 20 seconds. Oh no, that, oh, now I can tell. Now there's a reason to upgrade. So…”

“just shifting the UX and the platform is often kind of a discontinuity that creates new companies. And so you have that when you go to cloud, you get Salesforce. Um, when you go to mobile, you get not just Instagram, but also Snapchat, and then TikTok”

“generative ML, which is basically running the model backwards. Um, Maybe you weren't, you've taught it what the pattern is, now it can make more of the pattern.”

“Google started, is the internet was a library, Google was the index, and you could go and you could find the thing you wanted. If you go to the librarian and say, what's a good book about this? And the librarian would tell you, and you would go and you would read the book. That's the link that Google gives you. But…”

“the fundamental competitive threat comes from something that sits outside the market definition.”

“the point is, the new thing that looks completely different happens every five or 10 years, not every 50 years, the way it does in other industries.”

“It doesn't know that. It know, it, it, it is collectively reading back to you the entire internet, or it is collectively reading back to you every single photograph that it's seen, or it is collectively reading back to you like the collective consciousness of, of everything, of all the texts, all the data you've given…”

“with Stable Diffusion or GPT-III, It can only know which ones are good by reference to the existing corpus, and it can make something new, but it doesn't, a new original, a new and original and striking, but it wouldn't know that it was good without the human input.”

“if we just give you the answer, then you're not clicking on an ad. And there's always that tension of like, what's the real estate here? Do we show you the answer when you click on it? If I said, oh, what's the best right to do this? You just, you give me a link with an affiliate fee. How does this work? How do you,…”

“in the past as a retailer or brand, you've had all these different budgets. So there was rent and there was shipping and there was advertising and marketing and pricing and returns and so on. And now these are all one budget. Do I put, do we put the money into Instagram or into shipping or into pricing? Um, they all,…”

“generically what machine learning did was it let you solve problems that are hard for people to explain, but easy for people to do. Um, the problem with, with these generative networks is of course, this is prompt engineering is it's easy for you to imagine, but very hard for you to describe.”

“And with, um, AlphaGo, the scoring, because of the scoring inherent to Go, AlphaGo could generate lots of new moves and instantly see which ones are good. So it could iterate its way through that learning process to get to really, really good moves. And so then you had go players looking at this and saying it's…”

“what tends to happen is that firstly, firstly, so there's two bits to this. Firstly, the new tool tends to be bad at the things that the old tool is valued for, but add some completely new kind of capability. That turns out to be structurally extremely difficult to add to the old tool.”

“when you get a new tool, what you tend to do is start out by forcing the tool to fit the old way that you were doing the job. And then over time, you change the way you do the job in order to fit the new tool.”

“there's no switching cost in, you know, no one cares if this call is Teams or Zoom or Meet. Or hangouts or whatever it's called today. No one cares or knows you just click the link. Whereas switching, switching between different production tools is a big deal.”

“And meanwhile, the DTC brands sink or swim based on Retailer margins and consumer product margins, not on having a website.”

“Shein, which is, which is a sort of, you know, would be the Netflix paradigm in the clothing fast fashion. It's the hero. They are using this new channel to create the product basically close, but in a new way at a new volume with a new model, just as Netflix is creating TV shows, but in a new way and a new model.”

“If the deal was obvious and straightforward, and everyone agreed on it, everyone would be doing it, and it wouldn't make 50 times your money. And so the deal, all the deals that produce good returns for venture firms are the ones that people argue about, and the ones that look kind of dumb.”

“What's happened repeatedly is that they scaled to a certain point and did not have software margins, did not have network effects, became very capital intensive, had all the economics of a new CPG business, or a new apparel business, or a new clothing business, but that doesn't mean that some of them didn't create half…”

“Like, it's, it's all about a very, very small narrow margin, and that's also the story for Amazon, and it's a story for grocery delivery, it's a story for all these one-hour grocery things, it's a story for restaurant delivery, it's a story for dark kitchens, um, it's a story for Uber, um, and it's a story for, for…”

“it becomes a scaffold, it becomes a scaffolding question, not a technology question. Yes! You know, it's, it's, it's now a rigging question and a, and a, and a roadie question. It is not a technology problem at all.”

“And in generality, what can happen is that the regulatory agency can basically say that we don't like the structure of this market, so we're going to change it. And they don't have to prove that you broke the law. This is kind of the crucial point that in the U S the kind of the unit of decision is crime. This is…”

“they won't do your industry unless they can just kind of turn it into a trivial marginal, no marginal cost piece of software that bolts onto their industry.”

“It's a feel, feel good. And B, it makes it harder for you to churn away from Apple and move to Android. And so it's not the core, it's not the core skill, which is, would be like designing chips, but it's sort of adjacent to supporting the business.”

“What's actually happening here is that you had a UI that was built for your employees because you would open a bank account and a process and a set of systems built for your employees. And suddenly you're exposing that to your users, to your customers. And so what was an internal B to B system is suddenly got a user…”

“And we're now kind of going through this again, I think, as you look at things like Dalai, or Dali, or however you pronounce it, and Imogen, and GPT-III, of again, here's something that looks like a breakthrough. Here are all these amazing demos, and the demos are kind of cool. But it's not quite clear what that would…”

“The image sensor has become a universal input. Um, it's become, no, most images are Collected by image sensors will never be looked at by people. They'll be used as inputs to computing systems of some kind.”

“problems that are where you can tell the computer the rules, and you can give it a score so it knows whether it's doing well, which is what you got with AlphaGo. It would be very difficult to do that writing a novel, because there's no set of rules, and there's no, there's no automatic way that it could give you It…”

“Yes, I think they believe it, but they're also positioning themselves as a privacy company because they're selling car products where privacy is a big deal. The watch, the credit card, and of course, if you're wearing a pair of glasses that's got a pair, a cluster of cameras in it, then privacy being the privacy…”

“The sort of natural reaction of a lot of people in tech, and of course, particularly the people in tech who are actually being told they have to do something, is to say no. And we can't do that. That's a terrible idea, and that will break everything, and that's a big catastrophe, and, and, and, and awful, and you don't…”

“Like, we can't build gasoline that doesn't burn. And that's very different to saying, well, I don't like that, or I don't want to do that, or that's bad for my margins. And I think, you know, the sort of, the thread that I just see over and over again is not Understanding the difference in those kinds of questions.”

“it used to be that you basically, the thing was to understand what was happening in web two, and then the thing is to understand what was happening in smartphones. And now there's a fintech thing. So how do we think about the intersection of Square, which is now called Block, and Stripe, and, um, Web Three, and Google,…”

“Another big piece of it, of course, is that instead of spending the money on advertising, they spend the money, instead of spending the money on rent, they spend the money on advertising. So, you know, there's no, there's, you know, it's like a law of thermodynamics, a conservation of energy. You know, if you don't…”

“but it's also the feedback loop, and I think that's the change from ASOS, and certainly from Zara and H&M, and in some ways the change from Netflix, but it's what you have on TikTok, and what you have on, um, on Shein is, is you have that very quick feedback cycle.”

“And what you have in TikTok and what you have in Shein is you've got enough people looking at it that you've actually got that feedback loop. So basically the Shein, the Shein user base or the TikTok user base is almost like a giant mechanical turk. Um, telling the system whether you, we're telling the system what's…”

“it's always kind of much more useful to try and find the middle ground between, like, of course you can't predict what we'll be doing on the internet in 15 years' time, but to try and get specific about, well, what is it that you think that would do or not do, and how would that work? Um, because at least then you can…”

“a very large part of buying a book is you don't know what book you want. And so the job, the job of the retailer is not to be, can some retailers at the end point to a logistics chain? Other retailers, they are providing something else, which might be service or experience or marketing or advertising.”

“the point is, traditional TV, there was syndication. Whereas we're streaming, there isn't. Netflix buys the show, and it's the show. And so, as a Performer or producer or writer, you get the money that you get, and that's it. You don't get a show at the back end. And indeed, for Netflix, there isn't really a back end,…”

“The challenge, of course, is YouTube and Instagram have forms of media that are much easier to monetize, and much easier to think, how would you create something rich and engaging that you could monetize, whether it's e-commerce, or advertising, or something else.”

“so they build Prime as a way of reinforcing your loyalty to the platform, and that means they buy TV shows. That's not because they want to be in the TV business, it's because they want to drive Prime, which drives the core Amazon business.”

“what's happening in tech now, what, so tech today is spending a lot of time thinking about what happens in 2030, and meanwhile, like the retail industry is messed up by stuff that tech was obsessed by in 1995 or 2000, like broadband networks.”

“They want to turn it, particularly for Google, Facebook, Amazon, they want to turn it into automated self-service software. And if it turns out that you kind of can't do that, and you need a bunch of salespeople, or you need a bunch of offices, or you need a bunch of guys in pickup trucks, they've got the money to hire…”

“there's a sort of a mental model that I use, which is sort of the conclusion to the presentation that I published in December again, which is that, um, my sort of first 50 or 60 years of the car industry is, um, What's a car? And the second 50 or 60 years of the car industry is what happens when everyone has a car? And…”

“when you're a company the size of Google or Apple or something, um, there are things that are probably a perfectly good business, but they're just way too small as a proportion of that company for it to be worth, worth you donate it devoting the kind of effort that you would need to put into it to make it work. Um,…”

“And of course you might be a surfer and they might come and bite your leg off and then wander off and get bored. And there you are missing a leg, which is of course something that has happened to a bunch of tech, a bunch of industries outside of technology.”

“There's, there's the, how, what, there's this tool that you're using, and then the form that that tool Drives you to produce. And then there's a network around, um, the discovery that people find you and see your stuff. And the network of course is much the most important part because otherwise you're just kind of…”

“Clubhouse, while you were driving people to Clubhouse from your Twitter account. To begin with, because it didn't really have a network, and it was, you know, that was one of the things they had to try and build, was to build a network. And the same thing, and the same thing with Substack. People said that Substack is…”

“obviously if I moved from Instagram to some new tool, none of the people who followed me in Instagram will see me in this new place. And that's part of that is that the, the Instagram app or, you know, the Twitter app or whatever is based on creation and consumption. Well, it's a creation and a consumption platform…”

“in email or the calendaring, you can't actually change the primitives. All you can do is change the client. Whereas, but the client can't do change very much because the primitives are still the same.”

“Instagram, Snap is basically your friends, whereas Instagram is sort of a mix of your friends and your interest graph, so celebrities, designers, architects, whatever it is you're interested in.”

“TikTok is a different kind of format, but it's also a different network. And it's actually much more of a YouTube sort of network where it really isn't people that you know. Um, obviously it's all about the recommendation because it's partly because it's all about the recommendation engine. So it's not about, you know,…”

“I mean, clearly the, the issue there is that you can do whatever you want on the blockchain, but you can't get it, when you get it off, it's the exit points that are subject, that are subject to the law”

“It's a curated, but it's a, it's a magazine. It's like saying, I bought World of Interiors, and there was nothing about Ukraine in it. Well, no, there wasn't. It's World of Interiors. I mean, there's probably. There's probably something in the editorial, but that's what Instagram is, you know, and you've turned yours…”

“There's a huge difference between blocking and not amplifying.”

“Layered onto that, you have a huge amount of sort of scamming because basically you've added software to money and connected to the internet, and it's relatively early, immature software.”

“With open source, you could download the code and look at it, but you couldn't see it running inside Google. Whereas with a blockchain computer, you can see the code running and all the rules and all the transactions as they happen.”

“I could export my data, but most of my data is who's liked my pictures and who else they liked and who else follows them and who follows them. And A, I don't own this, but more to the point, if I am Like, go to a new network. Like, there's nothing in that new network.”

“don't clone office, make Google drive. Like that's making something that where it matters that it's the internet, make something where it matters that it's crypto.”

“And you read this and you think, okay, that's not a mass market consumer experience any more than the command line was, any more than open source coding was, and you can believe everyone will learn the command line and everyone will write the code, but they promise you they won't, and so you're going to need to…”

“And in tech, when people say five to 10 years, that's sort of the edge of science fiction. It's like, yeah, this is really interesting. But we don't know yet.”

“the stores are sort of a self-funded marketing operation. ... they broke even in their own right, but they're not just that, that's not actually what they're there for. They're not the end point to a logistic system.”

“You know, you can have 10 different department, you can have 50 department stores, and you can have 10 different supermarkets, but it makes no sense to have 10 different Amazons. However, it does make sense to have a hundred different sake brands, or a hundred different DTC brands selling this in different ways,…”

“obviously every brand wants to have a direct Consumer relationship. It doesn't follow that every consumer wants to have a direct relationship with the brand.”

“Web three is an attempt to Shift the discussion from saying this is digital gold and that's all it is, and it's, and it's a cryptocurrency. It's to get away from talking about currency and to say, no, it's a platform, which I think I'm, if I haven't said explicitly, like, yeah, I get that. I believe that. Definitely.”

“there's tech being obsessed with what will happen in 10 years. There's what's actually happening now, which is what tech was obsessed by 10 years ago. And then there's everything that's happening in retail and TV and advertising, which is basically like streaming and e-commerce, which we were obsessed by in, like,…”

“if all you can do is tell me that one barbarian king defeated another somewhere in the Central Asia, um, what use are you to the public? What does this tell me? Who cares? So if all you can tell me is that Amazon has ads now, Or that there's a new fast fashion app from China. Who cares? This is not interesting to me.”

“sort of the other half of them, or probably three quarters of them, are enterprise SaaS, workflow, two-sided networks, marketplaces, digital transformation, They're basically continuing in the curve of deployment of ideas that people in tech were obsessed by 10 years ago.”

“If you're in that 20 trillion dollar retailing industry, um, having a good website is not a tech conversation anymore. It's a retailing conversation.”

“the entire global recorded music industry is what, 16, seventeen billion dollars. Global online advertising is three hundred billion dollars. You know, Apple has, what, a hundred and fifty billion dollars of cash? Like, music is just a rounding error for the tech industry.”

“if your brand is bad, or uncool, or toxic, or whatever, and you change the brand, if you haven't actually fixed the reasons why the brand was toxic, then those things will just leach into the new brand, and you won't fool anyone.”

“And like, all this time you're spending showing how I can type on an AR screen, like, do you really think the future of this is going to be typing? That just feels like such sort of, let's take what we do on our Macs and put them in VR, which is so totally not what happened on smartphones.”

“Digital is now as much your identity as your music or your clothes. Um, or will become part of your music and your clothes and your identity and your self-expression and how you regard yourself.”

“quite a lot of the regulation or the more sensible regulation that one sees isn't supposed to change everybody's life or produce some dramatic structural change in the industry. It's supposed to solve some particular class of problem or some bad behavior or some corner case”

“car companies didn't do safety because they thought it didn't sell, and they were right, it didn't sell. That wasn't a selling point. You couldn't sell a car by saying it was more safe, except unless you were Volvo, and so what had to happen instead was you needed that to be applied at an industry-wide level by…”

“one of the phenomena you have now is to sort of push towards the sort of strongest common denominator, that whatever economy is large enough that you have to operate there and has some rules, you have to kind of have to follow those rules everywhere, which is why you get all these American companies having to work out…”

“Apple leapfrogs the rest of the industry, and then in the next nine months, the rest of the industry will leapfrog Apple, and then in 12 months, Apple will leapfrog, shake another jump forward, and they've got to lock on the hand of the industry, and that's it.”