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 1,057 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
Nadella: AI fundamentally alters software economics by introducing true marginal costs
“This one is interestingly enough, both a tech shift, a business model shift, because this is the first time you have marginal cost software that's not like COGS of the SaaS world, but true marginal cost. And three, the way you produce your artifact, your softw…”
Nadella: Gaming's real competition is short-form video, not other games
“After all, the gaming's competition is not other gaming. Gaming's competition is short form video.”
Chesky: Social networking was effectively 'uninvented' in 2012 by performative media
“Social networking may be the most popular product of all time that was invented and then uninvented. It was literally uninvented because around 20 12, it became social media. And the moment it became social media, your friends became your followers. You stoppe…”
Starting a startup at 18 or 19 carries minimal downside risk
“People think that starting a company or taking a break from school is this very high risk move when you're young. They say, oh, how do you find the bravery to do that? What I need to, what I try to push on people is it will never get better. Like when you are …”
Ryan Cohen: Corporate Leaders Without Personal Financial Stakes Disregard Shareholders
“Ultimately, what you see is that when people don't have their own money on the line, they don't give a shit, and they're focused on all kinds of other stuff that ultimately doesn't matter to shareholders”
AI cannot replace human sales reps because buyers crave human connection
“AI doesn't have a soul. It's not that human connectivity. It's not, you know, at our depth, AI is not born. It's, you know, AI is made. So, you know, we, there's a humanity that still needs to be nurtured and governed and developed and grown. People want to bu…”
Altman: Consumers will eventually value originality over human or AI origin
“You know, stuff that is technically written or drawn or filmed by a human, but is completely derivative and much less original than what an AI has generated, and I think that will be what people really care about long term.”
Gilbert: standard interview content cannot support a great business
“If you are a person who is being interviewed, your incentive is to go and do as many interviews as possible. Therefore, that is not a scarce commodity. Therefore, you can't build a great business on it.”
Zuckerberg: Brands will become more important as AI expands information access
“I think the brands are going to become more important, right? I mean, I basically think that, like, all kind of economic theory assumes that people have access to perfect information. And I think the internet took us a step closer to that, and AI is going to t…”
Karp: Software should be priced downstream of actual value creation
“The logic of software should be We charge you something downstream of value creation. That sum is a percentage of the value we create. It's better for both sides because it's significantly less than the value creates. Good for us because there's a multiple on …”
Karp: LLMs only create enterprise value within an ontology wrapper
“The value of LLM is when it's essentially in an ontology wrapper, because to actually create value, you have to be able to take the output, serialize it, and deserialize it in the context of the business. So the logic, actions, and security of the business, an…”
Karp: The woke left and right both undermine meritocracy
“The thing that unifies the woke left and the woke right is they don't like the consequences of meritocracy. They want to work back to the inputs. So in that, that like just will screw society. It's like, you've got to be able to allow people to succeed whereve…”
Cuban: Retail private markets require platforms to act as market makers
“Is there a way, whether it's tokenization or other approaches to that?
Probably, but it's going to have to go back to the future where somebody is going to have to take some risk and be a market maker.
Yeah.
If Vlad or whoever is a market maker so that there's…”
Coogan: In AI, the best product cannot win without distribution
“Distribution is king. It's very clear that, that even if you built a better product, the best product is not going to win, and we see this with the LLMs.”
Andreessen: Tech products become obsolete the moment they achieve perfection
“I think technology products become obsolete at the precise moment they become perfect. And what I mean by perfect basically is like, yeah, it's like the perfect idealized complete product. Like it does everything you could possibly ever imagine, everything a c…”
Andreessen: Copying Apple's Last-Mover Strategy Has Massive Survivorship Bias
“And I guess, yeah, what I would say is like, look, that clearly works if you're Apple, right? Yeah. And so it clearly works if you're Apple. But I would say there's a fine line between that strategy and just, and simply becoming obsolete, right? And so the pro…”
Andreessen: Tech Products Become Obsolete the Moment They Become Perfected
“One of the things I've observed over the years is I think technology products become obsolete at the precise moment they become perfect. And what I mean by perfect basically is like, yeah, it's like the perfect idealized complete product. Like it does everythi…”
Chaudry: Prompting AI is entertainment that directly competes for consumer attention
“I think my first POV on AI right now is that I think we don't talk about
AI and prompting as entertainment enough.
And what I mean by that is not like its ability to produce entertainment, but the fact that it is entertaining to chat with an LLM.”
Hays: AI-only development is too hard to debug without canvas tools
“Granted, that is a theoretical threat to the company, but the problem with AI-only development processes is that they are extremely difficult to debug or decompose. Figma gives a natural canvas to add in AI.”
Shkreli: AI Labs Should Partner with PE to Acquire Enterprise Data
“I sometimes joke that the big AI companies should become LBO shops. And what they can do is they can partner with KKR or Blackstone and all they get is the data. KKR and Blackstone pride the capital, get all the returns. Fine. But all the data comes back to, y…”
Patel: AI's primary bottleneck is the inability to learn on the job
“The key problem I see
Is this, the models can't do on-the-job training. So if you think about a human employee you might have some. And these human employees, the good thing about them is that, you know, you train them for six months or a year, and over time …”
Patel: Prompt correction is an order of magnitude worse than human learning
“I think the, this is at least an order of magnitude less efficient and less less capable than the way humans learn.”
Patel: Tacit experiential knowledge cannot be captured in a system prompt
“This, like, tacit knowledge that you build up through practice is not this, like, written instruction manual that you can just write out as a system prompt.”
Patel: AI automation is bottlenecked by task depth, not task breadth
“I think the bigger problem is not just the width or the width of the pool, how many different tasks you have to RL on, but it's a depth in the sense that a job doesn't involve doing a thousand different five minute tasks individually. It's the fact that you're…”
Patel: Aggregated on-the-job AI learning would functionally produce superintelligence
“Even if there's no software progress out of that point, that algorithms aren't improving, just that ability to learn on the job from everything in the economy would functionally produce what looks like a super intelligence, right? No human will be, will have m…”
Prince: Prioritizing raw performance over power efficiency is a losing chip strategy
“We've seen this game before with Intel saying, oh, that doesn't matter. Like just get the most performance possible. You can water cool things, all kinds of things. That, that tends to not be the winning strategy over the long term, and so we do think that the…”
Depue: RL task specification scales with compute far better than pre-training data
“Because like tokens, you kind of hit diminishing returns pretty quick in terms of just like more internet texts or more like human written math solution examples. And it doesn't scale super well, but the thing, the nice thing about the task specification is yo…”
Gallagher: Weakened Congress forces members to focus on building social media platforms
“The modern Congress has become so structurally weak that people channel their ambition into building a media platform, social media in particular, and that almost becomes their full-time job. And that's really a problem because we need Congress to focus on the…”
Gallagher: Reforming Congress paradoxically requires an over-empowered president
“The things we need to do to enhance the power and the functionality of the modern Congress actually require an over empowered president to make them happen. Like otherwise Congress won't do it on its own.”
Huberman: Funded research grants often fail the critical laboratory deletion test
“I sat on study section for a long time, and I can tell you there were clusters of grants that didn't solve what we call, what good labs call, the deletion test. If this laboratory didn't exist, would it change the direction of science? You have to pass the del…”
Cannon: Users share more candidly with AI interviewers than human moderators
“In fact, they like, have even expressed that like, they'll share more because there isn't a person. It's this idea of like social desirability bias. You're like, I don't want to come off as a person that is X, Y, and Z to another human.”
Moore: China has a video AI advantage from cheaper PhD labelers
“And China has really benefited there because there are so many more PhDs than there are here, and it's much cheaper for these companies to hire them to do these dense labels for the video data.”
Hotz: Robotic arm hardware is good enough, robotics is strictly software
“The arms are already good enough. It's the off the shelf arms are fine. It's all software. Again, it's always all software. Autonomous vehicles are all software. Robotics is all software”
Chen: AI reasoning only emerges at large scale requiring massive compute
“When you look at reasoning you just don't see that happen at small scale, right? There's like a certain scale at which it starts becoming signal bearing and that requires you to have resources, right?”
Dixon: Blockchains solve political trust problems preventing bank coordination
“There's a lot of interest from the banks to think about ways to use blockchains to create essentially marketplaces there, you know, maybe permissioned marketplaces where only kind of certain participants can participate. For them, the benefit is, you know, why…”
Dixon: Blockchains solve coordination problems whereas AI solves intelligence problems
“If AI, you know, is solving all the problems in the world that, that they need more intelligence, right? Blockchains try to solve the problems in the world that need more coordination or more, you know, kind of collective collective action, getting a bunch of,…”
Balaji: Startup societies must scale demographics like enterprise SaaS
“In the same way, like you first kind of have to build a product that works for like the individual remote worker, then maybe the couple where it's a two body problem, then like, you know, the early family, then the scaled up family, then like the whole communi…”
Balaji: Zero-Knowledge Tech Is to Crypto What Transformers Are to AI
“Zero knowledge arguably is, and Vitalik's also made this analogy, ZK arguably is to crypto what the transformer is to AI in the sense of, It takes a bunch of things that were previously special case stuff, and it puts it into kind of a common framework where y…”
Balaji: Stablecoins Are an Intermediate Step to Pure Crypto
“Sablecoins are an intermediate form that facilitates this transition from the, you know, fiat world to the pure crypto world. Kind of like putting newspapers online was an intermediate form.”
Contextual Awareness Resolves Noisy Non-Invasive Neural Speech Signals
“There's a really good story to be told about kind of doing this for speech and getting really terrible speech. I have no illusions as a neurotechnologist that you will get really terrible speech out of this. But the key thing is that this does not matter becau…”
Sankar: AI value will accrue to toolmakers, not model providers
“If AI is electricity, like in electricity, all the value didn't accrete to the people who made the turbine generators. It went to the people who made the tools that ran on electricity.”
Sankar: Defense consolidation's real cost was eliminating founder-led heresy
“The real issue is that consolidation bred conformity, and the conformity pushed out all the heretics. It pushed out all the founder personalities that you need to make this stuff really work.”
Van Lancker: Founders cannot outsource conviction to investors with misaligned portfolio incentives
“That investor or that angel investor is going to place who knows how many bets, and you're gonna place one during this period of time. You are an investor of your time, and that is one thing that you can do during this period, and you have to take that really …”
Russ d'Sa: AI hallucination is a feature in voice interfaces
“People talk about hallucinations as this bad thing, but in a lot of ways, for like, kind of this speech-to-speech interaction model, where you're talking to an AI, Hallucination is a feature.”
AI is a sustaining innovation for software incumbents, but disrupts Google
“In the sustaining versus disruptive innovation, I think AI is like pretty clearly a sustaining innovation in almost all cases. I think it like might be disruptive to Google. And so assuming like a scale up is still Competent and can still build new product. I …”
AI startups must either advance frontier research or obsess over customer needs
“If you're going to build an interesting company, I think you either have to attack it from, ah, you have the handful of researchers that can literally push the state of the art forward, which is like maybe eight organizations in the world, or You know and unde…”
OpenAI's limited bandwidth creates startup opportunities outside their top ten priorities
“If you have some strict stack ranking of the teams within open AI, like only 10 problems can have like the top 10 teams on it. And so if you're any, if you're doing anything after like the top 10 problems that they have, then you're like by definition don't ha…”
Will Brown: AI scaling faces diminishing returns on capital investment
“The quality bump over things that are much smaller is just like the, we're hitting diminishing returns on capital investment is a lot of it. Like they're taking out of the API because like the, they can sell other things with the same GPUs and make more money …”