The Ledger
Every statement that passed quotation and attribution checks. Mix any filter with any other: certainty 1/5, debate potential 5/5, or both at once.
why aren't all 537 resolved? a statement only gets an assessment when the public record can support or contradict it. opinions and what-ifs never can, and 0 checkable ones are still open, waiting for their date. predictions held up or didn't; assertions are supported or contradicted. on every card: ▮▮▮▮▮ certainty · ▮▮▮▮▮ debate potential. speakers are clickable
Evans: Bottom-up enterprise adoption only ever succeeded for Slack and Notion
“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.”
Evans: Open-source clones of commercial products never succeed
“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 sour…”
Evans: Near-zero AI error rates render semantic understanding questions practically irrelevant
“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…”
Evans: Decentralized social networks with multiple clients are an illusion
“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 t…”
Evans: WhatsApp proves harmful content is not caused by ads or algorithms
“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 d…”
Evans: Bottom-up sales only reach 5% of the enterprise market
“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 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 wai…”
Evans: A software company's moat is go-to-market, not writing code
“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 pr…”
Evans: There are no network effects in building AI models yet
“There are no network effects in building the models yet. There may be in the future, but at the moment we don't know.”
Evans: AI product heads cannot set roadmaps because research breakthroughs dictate capabilities
“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 d…”
Evans: Undifferentiated tech with divergent market shares creates market instability
“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.”
Evans: The delusion that AI will turn non-coders into programmers is wrong
“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 that everyone who doesn't write code will …”
Evans: AI foundation models lack network effects and sustainable moats
“There's no network effect that we know of yet. And so unlike previous 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 …”
Evans: Lack of network effects makes ChatGPT's early lead fragile
“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 …”
Evans: Cheaper AI and Jevons paradox will not guarantee pricing power
“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. But that doesn't mean you'll be able to charge muc…”
Evans: Big enterprises are running AI trials, but few use it broadly
“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…”
Evans: Testing LLMs on complex logic is like testing Apple II for five-nines uptime
“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 problem…”
General web search is a natural monopoly comparable to utility companies
“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.”
Evans: Probabilistic AI is a feature, not a standalone product
“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…”
Evans: Enterprises require dedicated product UI over raw ChatGPT prompts
“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…”
Evans: TikTok's Real Threat Is Algorithmic Control, Not Data Collection
“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.”
Evans: Engineering Diversity Alone Cannot Fix AI Dataset Bias
“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.”
Evans: Massive scale transforms generative AI tools qualitatively compared to Photoshop
“There's a difference when you do, something can be the same theoretically, just at greatest in principle, but a much greater scale and practical and accessible and everything else. And in fact, making something much greater scale and much more accessible, but …”
Evans: Newspapers' core crisis is losing their former distribution monopolies
“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.”
Devins: Vision Pro is a destruction test for spatial computing headsets
“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 minu…”
Evans: Enterprise AI deployment takes three years, not three months
“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 actually turning it into software that's not just ChatGPT.”
Evans: Making work cheaper and accessible often increases total employment
“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 can make a poster with fonts, that doesn't mean you have fewer graphic designers. …”
Brody: Selling solely to a celebrity fanbase cannot create an exitable business
“And then there's a third factor, which is the product has to be good. Because if you're building a product that the world doesn't need, doesn't matter if it's authentic, doesn't matter if the, you know, inevitably your audience will dry it. You know, you'll, y…”
Evans: Science lacks a theoretical model for LLMs and AGI
“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 d…”
Benedict Evans: Augmented reality requires a strict visual realism threshold to work
“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.”
Benedict Evans: Spatial computing turns the metaverse into a mere app icon
“With a feature phone, internet was one icon on your telephone. Whereas on a smartphone, the telephone is one icon 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 i…”
Benedict Evans: If a $500 Vision Pro fails, the VR headset dies
“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 not…”
Benedict Evans: Fixing LLM hallucinations does not require achieving AGI
“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 might not know and give you make something up, or …”
Benedict Evans: Banning facial recognition is absurd because ML is commoditized
“But in general, machine learning has become a commodity very quickly which is sort of the absurdity of people saying, well, we have to ban face recognition, or you can ban it, but that won't work. Cause it's just a commodity technology.”
Evans: AI summarizing content without referral traffic makes publisher expropriation claims valid
“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.”
Evans: Generative AI does not destroy search advertising economics
“I mean, 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 ch…”
Evans: Technologists confuse engineering architecture with actual end-user problems
“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 …”
Evans: Synthesizing styles from millions of images isn't copyright infringement; direct copying is
“If I've now looked at millions of pictures of wizards and I draw something, do I have to pay them? Like, no. If I copy your thing, do I have to pay you? Well, yes.”
Evans: Existing copyright frameworks fail to map onto generative AI
“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 map very well to what's happening here. It's kind of like taking live performance rights from the early 19th c…”
Evans: Privacy concerns span all foreign apps, but algorithmic risks isolate TikTok
“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,…”
Evans: Amazon ads function as price discrimination to squeeze supplier margins
“This is effectively a way for Amazon to squeeze more margin. Out of its suppliers as much as it is advertising”
Evans: Dismissing generative AI over early errors misses the conceptual breakthrough
“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 idio…”
Evans: Moving to 3D Computing Is a Bigger Shift Than Mobile
“But the other side of it is 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. And so the shift in character to move to three D is actually much bigger than the shift. …”
Evans: Software productivity advances via automation and ML, not bigger screens
“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…”
Evans: Founders wanting autonomy shouldn't take $20 billion acquisitions
“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 pr…”
Evans: Streaming platform scale is driven by content budget, not network effects
“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.”
Evans: Content matters more than software in media disruption
“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 counter argument to the sort of software reach the world thesis.”
Evans: As software commoditizes, value shifts to the domain product itself
“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 the clothes, or the TV, or the light, that's the thing. The product is no longer technology, the technology is just some part buried …”
Evans: Tech cannot unify consumer healthcare like it can consumer banking
“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 differen…”