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 210 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
Crazy but repeatedly successful decisions imply a superior model of reality
“If I see someone who is pretty successful do a series of crazy things over and over again, and yet they work out every single time, either they have root access, To the simulation. Or their model of reality is better than mine in some key way that I don't unde…”
Sundheim: The biggest investing mistakes come from selling great compounders too early
“The biggest mistakes are selling the Costco's too early, because you know, the IRR is totally dependent upon what you assume the exit multiple is, like, and that is difficult to be precise about, you know, what the right exit multiple is for a business.”
Sundheim: Private market illiquidity is a huge benefit for holding great companies
“That's the nice thing about the private markets is that Once you invest, you can't sell, and usually for the best companies, that's a huge benefit ultimately.”
Sundheim: Managing risk after a position moves destroys capital
“Risk management cannot happen after the fact. Like, anything you do after the fact is not risk management. It's actually just destroying capital. Risk management has to happen before the fact. Meaning, like, you size these positions so that, like, When it goes…”
Sundheim: US markets price turnaround stories far faster than Europe
“Like, the U.S. Is pretty quick at, like, seeing change happening, and then pricing in that change. In Europe, I find you know, once a company, like a company like Rolls had underperformed for so long, I guess that European mutual funds, they just kind of got i…”
Sundheim: Investors always make the most money from multiple expansion
“You're always gonna make the most money from multiple expansion. Just like it's hard to have that differentiated view on a company's growth rate.”
Scaringe: Vertical integration becomes a huge structural cost advantage at scale
“When you're producing less than 50,000 units a year, the engineering fixed cost and the operational fixed cost to run that is very high, so it's a structural cost challenge. But you reach an inflection. Well, just to run an engineering team to engineer all tha…”
Scaringe: Legacy EV launches fail due to uncompelling products, not customer hesitation
“There's been a number of EV launches from Existing manufacturers that haven't gone that well, and I think we get the causality backwards. We say, oh, it's because customers aren't ready to buy EVs. And I say, well, no, it's because the EV is not that compellin…”
Lütke: Using third-party software forces adoption of vendor worldviews
“If you use software by others, you have to buy into their vision.”
Lütke: Internal software shapes companies faster and more effectively than policy
“It's the sort of missing ingredient people talk about, you know, people talk about incentives shaping companies, and then policies shaping companies. So, those tend to be the two tools. And I think then, sort of generally, like, if people are really insightful…”
Lütke: Outperforming products come from places that reject consensus decision-making
“It's when something outperforms, it's usually the product of a place that doesn't make decisions by consensus, right, as we know.”
Collison: People cannot be sufficiently optimistic about businesses that truly work
“People, people are not capable of being sufficiently optimistic about a business that really works. They, it just, it blows through all their projections.”
Lütke: Advantages of any org structure occur immediately; disadvantages emerge over time
“Well, and the advantages of any organization structure occur immediately. The disadvantages only appear over time.”
Andreessen: MBA hiring choices are the ultimate VC contrarian indicator
“This in turn is why the employment decisions of graduating Harvard and Stanford Business School students are such a great indicator. Yes. Possibly the best indicator of all of what's happening in the market, because if they go into tech market, it's, the marke…”
Songhurst: A Series A lead investor's status predicts a startup's success
“The single strongest correlation of how a company will perform is which, how high status a VC does a series A is within the stack ranking of VCs. It is far more predictive, sadly, than, you know, my own selection or any other variable I can find. It's almost d…”
Andreessen: Westward expansion selected the most risk-tolerant people for Silicon Valley
“It's not an accident that both Silicon Valley and Hollywood are the place, places that they are because the people involved went west as far as they could before they were literally stopped by the Pacific Ocean. Right, like, it was the ultimate selector in the…”
Songhurst: Mega-cap tech companies are built by second-generation institution builders
“There is a trend where, if you go to the absolute cutting edge of tech, they're so sort of wilderness people that they don't have the conscientiousness. They, you know, go flying instead of turning up to the meetings. And it's almost like you get a second gene…”
Andreessen: Transparency and peer-to-peer communication dissolve centralized institutional authority
“True transparency, true free speech is just, is a fundamental solvent basically dissolving all centralized institutional authority.
and the reason for that is centralized institutional authority is never perfect, and it often has problems, and in fact, it of…”
Songhurst: The centralized nation-state is downstream from centralized media
“The extreme version of this for good or ill is that the centralized state is an outcome of centralized media. The nation state is downstream from the newspaper.”
Wu: General AI intelligence alone will not solve messy software engineering
“Software engineering in the real world is so messy, you know, and there's all sorts of these things that come up. And I think in practice, you know, most disciplines look like this. And I would say the same thing about law or medicine or, and so on. And so whi…”
Wu: Solved AI benchmarks shift bottleneck to defining real-world benchmarks
“The thing is when that happens, I don't think what we end up with is just pure ASI end of humanity, human knowledge work or whatever. I think the thing that we end up in is basically a point where the hard question is, all right, now what is the benchmark, rig…”
Wu: Seat-based SaaS pricing makes no sense for AI agents
“Seats don't really make sense when it is like, the AI themselves are arguably seats as well, you know, like they're doing a lot of the labor too. And then on the other side, I think you know, usage obviously just goes so naturally with the cogs themselves beca…”
Levels: Personal wealth generally reflects total value added to society
“Getting rich is simply a representation of your added value to the world economy. In most cases, of course, as edge cases generally, The money on your bank account is how much value added to society.”
Vogt: Prior Robotics Standards and Tools Are Now Worthless or Outdated
“I think it would be that robotics today is a completely different field in a different industry than it was five years ago. Like all of the things we thought we knew, all the businesses that were tried and failed, all of the tools that have become best practic…”
Vogt: Modern Robotics Requires Adaptability via Neural Networks, Not Repeatability
“Today you don't need a robot that's repeatable. You need a robot that's adaptable, like powered by neural networks, and if it makes a mistake or doesn't approach this object at exactly the right angle, it doesn't matter. It can correct that.”
Vogt: Mimicry and LLM-Style Tools Make Domestic Robotics Tractable
“If you're willing to completely let go of that and embrace today's tools, it's not the programming thing. You have to show a robot how to do a task and it can mimic that and let, you know essentially chat GPT like technologies instruct these things at a high l…”
Vogt: Scaling from 80 to 400 people reduces per-person productivity by 90%
“When we go from. 80 people to 400 people, our productivity per person is going to drop by 90%. And are we going to sign up for that and understand that it won't get better until we're past 400 people? I mean, that's the reality of the situation.”
Vogt: Selling a startup to a legacy corporation cannot scale its vision
“I think my heart was in the right place, but I was naive about the ability to get like a large corporation, which is like, it's like an aircraft carrier. You can't steer it. You can't get it to change its focus. Yeah. It's going to do what it wants to do or wh…”
Platform value comes from close contacts, not total network size
“What Snapchat proved was actually that it's not about the size of your network or the number of people you can contact on your network, it's about who you actually talk to. And that tends to be a much smaller group than the total size of the Potential network,…”
Protecting hardware innovations is much easier than software in the AI era
“That is one thing I love about hardware. It's way easier to protect our innovations and our ideas compared to software, especially today in the age of AI.”
Staniszewski: Solving conversational voice AI inherently solves text agents too
“If you fix the voice agent, you'll have fixed text piece or like inherently as well.”
Collison: Real-world industry experience is required to discover massive B2B markets
“You talk to university students and often their ideas for companies are pretty half-baked. Yeah, it's find my friends. It's like a college textbook exchange app, you know, but you know, there's like the five apps or whatever. Whereas so frequently what happens…”
Cacioppo: Tying brand identity to a category standard backfires when competitors emerge
“In the early days, before we had competitors, we tried to basically make this call response of, like, someone says, talk to you, someone says Vanta. And this really close association, which in the early, again, when we were just competing against consultants. …”
Collison: Investor updates with lots of words and no metrics signal failure
“I feel like investor updates with a lot of words and no metrics.
Oh yeah, those are bad.
Those are bad.
And actually like, no investor updates is fine.
Like you didn't have to send me.
Yeah, yeah, exactly.
Yeah, no with either very good or very bad.
Yeah.
met…”
Dolgov: Autonomous AI relies more on metrics and data than architecture
“And of course, you know, people like to talk about architectures, but architecture, you know, is important, but really a lot of it comes down primarily to your metrics, to your evaluation mechanisms, to, you know, all of the training recipes, and of course, ne…”
Dolgov: Each additional nine of AV reliability requires 10x more effort
“It's deceptively easy to get started, but it is super hard to go, you know, the full distance and, you know, the number of knives, right? That you have to, like, there's, The standard, you know, engineering rule of thumb that, you know, every next nine takes, …”
Lopes Lara: Fintech and healthcare startups cannot operate by asking forgiveness
“I think that the approach we took from the start was that financial services or healthcare, I think you can't ask for forgiveness. I think there's a big difference between losing people's money, see what goes wrong, like an FTX example that can go very wrong w…”
Mansour: Banning prediction markets heightens risks by driving trading offshore
“Because if you ban it,
You're actually heightening the risks that you're trying to prevent, because now that activity is going offshore, right?
Like, and where you cannot monitor it or police it or do anything to protect it.”
Taylor: Agent-to-agent phone calls are temporary until better harnesses emerge
“I think it's great that our agents have spoken over the telephone already in English. ... But I don't think it's like the long-term future because there's so much value that you can provide. And put another way, the agents using a sophisticated... Application …”
Taylor: Grounding AI agents is harder for famous brands than obscure ones
“It's actually easier when the internet has never heard of you and you want to make a well-grounded agent, it's actually pretty easy because there's no temptation from the LLMs to go off script. So actually, I would say. Ironically, the harder challenge is when…”
Taylor: Applied AI startups must build ahead of models and discard code
“So the interesting part about building an applied AI company is you can't have the luxury Of waiting for all the models to catch up with your aspirations. But you know they will. So you have to have the best technology and have to be comfortable with throwing …”
Taylor: AI token usage does not strongly correlate with delivered business value
“I would argue there's not a strong correlation between token usage or utilization and value. There may be, but there's not always”
Taylor: Overly lean startups risk losing to slightly larger, well-staffed competitors
“If all of a sudden for some you know, clever reason you want to prove you can, the idea that, like, a competitor might have 10 people and beat you is probably more likely than even having a ten billion dollar company.”
Langley: Tech sophistication for policing should scale with crime severity
“The framework that we use is, I don't know how you land this, is that the severity of a crime should be commiserate with the sophistication of technology.”
Pope: The best AI inference chip is also a great training chip
“I think the best inference chip today will be a train, a really good training chip as well.”
Pope: Mixture of experts maps well to systolic arrays, attention does not
“The mixture of expert layer maps really well, but the attention does not.”
Collison: B2B payments need a DNS-style central clearinghouse for bank accounts
“If you looked at the bill pay system from the outside, you would say, we should have DNS for companies.
So rather than a company
Like, sending you their bank account details.
You should, like, look them up in some central clearinghouse, and that way you confir…”
Thompson: In algorithmic media, absolute volume of good content beats hit rate
“What actually matters, and this is a key thing, people get hung up on relative numbers, and what matters is absolute numbers. So it is better to have .one percent of your content is good if your content is like in the billions or trillions, as opposed to, oh, …”