从代码洞穴到 Y Combinator：坚韧与创新之旅

我热爱科技、工程和科学，因为小时候每个周末我都会读一些远超我理解水平的东西。我一半都看不懂，但我觉得自己能看懂。

I love technology and engineering and science because I would just spend every weekend as a kid reading things that were way above my pay grade. I didn't understand half of it, but I felt like I could.

然后这就把我们带到了今天，机器能帮助你学习。

And then it sort of brings us to today where the machines can help you learn.

你在学校是个好学生吗？

Were you a good student in school?

是的。我的意思是，我当时没有意识到，我的童年太艰难了，所以我觉得取得好成绩、上大学就是你的出路。

Yeah. I mean, I didn't realize this at the time, like my childhood was so rough that I guess I viewed getting good grades and going to college as your way out.

你的出路。

Your way out.

那是我的……是的，那是隧道尽头的光。我十几岁的时候总是这么想。我当时有自杀倾向。我讨厌周围的一切，但当我坐在电脑前，我可以进入我的代码洞穴，我拥有控制权。除此之外，我生活中的一切都混乱不堪，我不知道什么时候会无缘无故挨打。但当我坐在电脑前，甚至在学校学代数或微积分时，感觉就像，哦，世界如此美丽、井然有序，我能理解它。

That was my... Yeah, that was the light at the end of the tunnel. That was always what I thought when I was a teenager. I was suicidal. I hated everything about it, but when I was in my computer, I could go into my code cave and I had control. Everything in my life outside of that was tumultuous, and I just didn't know when I would catch a beating for no reason or whatever. But when I was in front of my computer or even when I was in school studying algebra or calculus, it's like, oh, the world is so beautiful and well-arranged, and I can understand it.

是的。秩序。

Yeah. Order.

是的。

Yeah.

当你用你父亲的电脑时，你会玩游戏还是构建东西？

When you would be on your dad's computer, would you play games or would you build things?

我想游戏是我的初恋。

I guess games were my first love.

那当时是什么游戏？

What were the games then?

天哪，大概是 Atari 上的《间谍猎人》吧。但我真的很喜欢冒险游戏，比如《猴岛小英雄》或《夺宝奇兵》系列。老 Sierra 的冒险游戏真的很有趣，因为我不仅仅想要街机游戏。我想要故事。我喜欢那种你可以体验另一个人生活的感觉。如今，我喜欢人们玩的游戏，比如《荒野大镖客 2》，这是最好的游戏之一。

Oh my gosh, like Spy Hunter on the Atari, I guess. But I really loved adventure games like the Monkey Island games or the Indiana Jones games. The old Sierra adventure games were really fun because I didn't want just the arcade. I wanted a story. I loved this idea that you could inhabit another person's life. Today, what people play that I love is like Red Dead Redemption 2, for instance, one of the best games.

你还在玩游戏。

You still play games.

是的，我喜欢这类游戏，因为叙事故事，你可以进入别人的世界。那时的现实，我把它看作一种很好的解脱。

Yeah, I love those types of games because of the narrative story, and you can inhabit someone else's world. Reality back then, I viewed it as a nice reprieve.

什么是 Y Combinator？

What is Y Combinator?

Y Combinator 在这一点上是一个机构，你在网上申请，不需要认识任何人或知道任何事。你填写 12 个问题，录制一个一分钟的视频，我们有 16 位合伙人阅读这些内容并观看视频，试图找出谁将在未来 20 年构建所有技术。

Y Combinator is at this point an institution where you apply on the internet, and you don't have to know anyone or anything. You fill out 12 questions, record a one-minute video, and we have 16 partners who read those and watch the video and try to figure out who are the people who are going to build all of the technology for the next 20 years.

有多少人尝试报名？

How many people try to sign up?

是的，每年有 8 万份申请，大约 800 个名额。

Yeah, it's 80,000 applications a year for about 800 spots.

哇。

Wow.

每年。我们邀请你来旧金山。我们会见你 10 分钟。通常有两个合伙人在那里，10 分钟过得很快。我们只想知道你在做什么，为什么是你来做这件事？

Per year. We invite you to come to San Francisco. We'll meet you for 10 minutes. It's usually two partners in there, and 10 minutes goes really fast. We just want to know what are you doing and why is it going to be you who does it?

房间里的两位合伙人有什么背景？

What are the backgrounds of the two partners in the room?

通常，我的意思是，YC 的每位合伙人实际上都自己经历过这个项目并创建了一家初创公司，在很多方面他们做出了人们想要的东西。所以我们去做了，然后我们回来试图帮助下一代去做这件事。

Usually, I mean, every partner at YC actually went through the program themselves and created a startup, and in a lot of ways they made something people want. So we went and did it, and then we're back trying to help the next generation go do that.

告诉我你当时的经历。你记得你那 10 分钟里的什么？

Tell me about your experience of doing it. What do you remember from your 10 minutes?

哦，天哪。我记得和 YC 的创始人 Paul Graham 以及他的其他联合创始人坐在一起。他们基本上想看演示。我的初创公司，你只需在 iPhone 上拍一张照片，然后通过电子邮件发送，你就会得到一个博客。这很容易演示，然后我想他们立刻就明白了。所以一旦有人理解某物是什么以及为什么人们会使用它，那就回答了那个大问题。第一天，你会得到一件写着“做出人们想要的东西”的 T 恤。如果你看看那 10 分钟，那就是我们真正关心的。就像，这个人会做出东西吗？最疯狂的是，现在这比以往任何时候都容易。每个人都能做出东西。然后第二部分变得困难得多，重要得多，更难做对，但也更重要做对。那就是人们想要的东西。

Oh, man. I remember sitting down with the founder of YC, Paul Graham, and his other co-founders. They basically wanted to see the demo. My startup, you just took a photo on an iPhone and you emailed it, and then you got a blog. That is very easy to demo, and then I think they got it immediately. So once someone understands what something is and why people would use it, that sort of answers the big question. On the first day, you get a t-shirt that says 'Make something people want.' And if you look at the 10 minutes, that's all we really care about. It's like, is this person going to make something? And the craziest thing is that's easier than ever now. Everyone can make something. And then the second part has become much harder, much more important, much more difficult to get right, but much more important to get right. It's something people want.

Paul Graham 是 YC 的创始人吗？

Was Paul Graham the founder of YC?

是的，他是创始人。

Yeah, he was the founder.

跟我讲讲他。

Tell me about him.

哦，他是一位非常杰出的博学家。我想你会非常喜欢他。他写了一本书叫《黑客与画家》，这是一本完美的博学家之书。他是一名黑客，但同时，我最近见到他，他说他在全英国建立了最好的古典艺术收藏之一。他现在住在英国。所以，那有多酷？就像，他真的写了关于 Lisp 的书。Lisp 是 70 年代最受尊敬的硬核编程语言之一。我会说 Lisp 之于编程语言，就像当年的 CVGBs 一样。

Oh, he was sort of this really brilliant polymath. I think you would really dig him. He wrote a book called 'Hackers and Painters,' which is the perfect polymath book. He's a hacker, but at the same time, I saw him recently, he was saying that he built one of the best classical art collections in all of England. He lives in England now. So, how cool is that? Like, he literally wrote the book on Lisp. Lisp is one of the most respected sort of hardcore programming languages from the '70s. I would say that Lisp is sort of for programming languages what CVGBs was back in the day.

酷。

Cool.

我的意思是，它就像任何合法事物的起点。

I mean, it's just anyone who's legit ground zero.

它是起点。它是合法的。

It's ground zero. It's legit.

如果你知道如何编程并关心代码，你就知道 Lisp，他写了很多关于 Lisp 的书，字面意思叫《On Lisp》。所以我们从那里开始，他问，我如何解决金钱问题？他是哈佛大学计算机科学博士，之后他说，我需要在这个世界上谋生，我不能靠绘画达到目的，但我确实知道如何编程。那么我如何把编程——它本身是一种手艺和艺术——变成商业呢？

It's if you know how to program and care about the code, you know Lisp, and he wrote many books on Lisp, literally called 'On Lisp.' And so we went from that to a business where he said, how do I solve the money problem? He was a PhD at Harvard in computer science, and then afterwards he said, well, I need to make my way in this world, and I'm not going to get there by painting, but I do know how to program. So how do I turn programming, which is its own craft and art, into commerce?

那是什么时候？

And when was that?

我想这大概是 1990 年代初到中期。

I guess this was like early to mid 1990s.

所以他实际上是创建第一个 Web 应用程序的人之一。

So he actually was one of the people who created the first web application.

哇。

Wow.

所以网络是新的。但后来他意识到，我不必只做一个静态 HTML 页面。我可以把它连接到一个程序。今天你想到 Web 或互联网或计算机程序，它等同于我打开浏览器，我要去这个地方。但他是第一个真正把它连接起来的人。

So the web was new. But then he realized, well, I don't have to just make a static HTML page. I could just hook it up to a program. Today you think about the web or the internet or a computer program, and it's synonymous with I open my browser and I'm going to go to this thing. But he was the first person to actually hook that up.

是的，那太棒了。

Yeah, that's amazing.

而且它是为了制作最终成为 Yahoo Stores 的东西。所以它是帮助其他人建立企业和在线业务。

And it was to make what ended up becoming Yahoo Stores. So it was helping other people build businesses and online businesses.

直到我在他的文章里读到“我需要解决金钱问题”这句话，我才有了这个说法，我在网上找到了他的文章。现在我的朋友都在那个世界里。人们有点把 YC 说成一个邪教，我想。但我经常思考，因为它是一个邪教吗？它是一种意识形态吗？它可能是一种宗教。

I didn't have the words of 'I need to solve the money problem' until I read it in his essays, and I found his essays on the internet. Now all my friends are sort of in that world. People sort of talk about YC as a little bit of a cult, I guess. But I think about it a lot because is it a cult? Is it an ideology? It could be a religion.

它看起来很有生产力。

It seems productive.

是的。嗯，这就是我们试图让它成为的样子。

Yeah. Well, that's what we're trying to make it.

它与其他风险投资产品有什么不同？

How is it different than other VC offerings?

我的意思是，风险投资是一件非常有趣的事情，因为对我来说，它感觉停留在 1970 年代或 80 年代。风险投资在 50 多年前并不真正存在。然后这个领域的传奇人物，他们有点……我觉得他们像银行家，你知道，他们都穿着正装，扣子扣得整整齐齐。

I mean, VC is a very interesting thing because to me, it feels like it's stuck in the 1970s or the 80s. VC didn't exist really going back more than 50 years. And then the legends of the space, they're sort of... I felt like they're bankers, you know, they're all dressed up and buttoned up.

你听过仙童半导体（Fairchild Semiconductor）的故事吗？我记得他们当时给纽约认识的银行家们写了很多信，回复的人中有一位是阿瑟·洛克（Arthur Rock），后来他也成了资助苹果的人。

You ever hear the story of like Fairchild Semiconductor? I think they literally wrote a bunch of letters to a bunch of bankers they knew in New York and then one of the people who replied was Arthur Rock, who was also the guy who ended up funding Apple later.

哇。所以这有很长的传承，但我觉得它已经从那演变了很多。我认为如今世界上最好的投资者其实是曾经的创造者，而不是银行家。而且我觉得这种氛围很重要。世界上远不止金钱和数字。就像金钱和数字是在你做出人们想要的东西之后才来的。

Wow. And so there's a long lineage of that, but I think it's like really evolved from that. I think today the best investors in the world are people who are actually former builders instead of bankers. And I think that vibe matters a lot. There's a lot more to the world than just the dollars and cents. Like the dollars and cents come after you make something people want.

嗯。

Yeah.

所以我想如果我要描述，嗯，你知道，YC 是不是只是另一家风投？风投只关心钱。我想类比可能是媒体公司的上市公司 CEO。你知道，那些人的工作就是让股价上涨，不惜一切代价。而当我们处于早期阶段时，我们不必那样。事实上，如果你在人们还没有想法的时候就这么做，你会扼杀它。

So I think if I had to describe like, well, you know, is YC just another VC? VCs when it's just about the dollar. I guess the analogy would probably be like public company CEO of a media company. You know, those people kind of their job is to only make the dollars go up and that's it at at any cost. Whereas when we're early, we don't have to be like that. And in fact, if you try to do that early when people don't have an idea yet, you'll kill it.

那么，这是怎么运作的？人们回答 12 个问题，1% 的人被邀请参加会议。是这样吗？

So, how does it work? People answer 12 questions, 1% get invited to a meeting. Is that

哦，是的。大约 3% 到 4%。

Oh, yeah. About three or four%.

3% 到 4% 的人被邀请来见你 10 分钟。

Three or 4% get invited to meet you for 10 minutes.

是的。是的。实际上只有 1% 的人能获得我们的投资，我们给他们 50 万美元。这是一笔投资。所以，我们最终拥有每家公司 7% 到 10% 的股份。

Yeah. Yeah. Only 1% actually get an offer to be funded by us and we give them half a million dollars. It is an investment. So, we end up owning 7 to nish% of each company.

钱和 YC 做的其他事情相比，占多大比重？

How much is it about the money versus what else YC does?

我的意思是，现在有很多地方可以拿到钱。你在其他地方得不到的实际上是一个人群社区。当你处于早期阶段，还不知道要构建什么时，你能做的最好的事情就是让自己周围都是那些经历过这些的真正的创造者。而且这需要是一个安全的空间，你可以谈论你在做什么。我们实际上试图强制执行这一点。

I mean, there are lots of places to get money at this point. What you can't get anywhere else is actually a community of people. And when you're early and you don't know what you're even going to build yet, the best thing you can do is surround yourself with other people who are actually builders who have been there. And that it needs to be sort of this safe space where you can talk about what you're doing. And we try to enforce that actually.

除了现在钱比以前更容易获得之外，钱现在变得不那么重要了吗？

Beyond the fact that money is more readily available now than it used to be. Is money less important now?

我想是的。实际上，现在疯狂的事情是，能够从事这门手艺的行为突然变得重要了 100 到 1000 倍。笑话是 2025 年的感恩节是编码和代码生成方式发生完全疯狂转变的时刻。所以，Anthropic 发布了它的 Opus 4.5 模型，世界震动了，但没有人注意到，直到现在他们才开始注意到。我在一月份注意到，我从 2013 年中断后重新开始编码，现在我将要做的编码工作、编写人们可以使用的软件的工作量大约是 2013 年全年的 100 倍。

I think so. Actually, the wild thing now is like the act of being able to do the craft suddenly became 100 to a thousand times more important. The joke is Thanksgiving of 2025 was this totally insane shift in the way coding and codegen works. So, Anthropic released its Opus 4.5 model and the world shook, but no one noticed until they're only starting to notice now. I noticed in January where I started coding again after a hiatus from 2013 and right now I am about to do about 100x the amount of work coding writing software that people can use then the whole year of 2013.

哇。

Wow.

所以想象一下，如果这个房间里有 100 个人，那会挤得水泄不通。就像你可能无法在我们坐的这个房间里容纳 100 个人。但这就是我现在坐在电脑前的感受。在过去的 60 天里，我发布了两个不同的开源项目。其中一个叫 GStack，现在每天有 3 万人使用，而且是免费的。我直接把它送出去了。然后 11 天前，我发布了另一个叫 Gbrain 的东西，这是我的知识记忆系统。

So imagine like if there were 100 people in this room this room would be packed to the gills. Like you probably couldn't fit 100 people in the room that we're sitting in. But that's how it feels when I'm sitting in front of a computer right now. I've released two different open-source projects in the last 60 days. One of it which GStack is actually used by 30,000 people every single day right now and it's free. I just gave it away. And then 11 days ago I released another thing called Gbrain which is my knowledge memory system.

解释一下它是什么。

Explain what it is.

所以正在发生一件疯狂的事情，显然代码生成发生了。你可以使用像 Claude Code 这样的东西来编写比以前多一百倍的软件。

So there's this crazy thing happening where obviously codegen happens. You can use something like clogged code to write a hundred times more software than you did before.

你过去是专家级程序员，这有影响吗？

Does the fact that you were a expert coder in the past make a difference or no?

我认为实际上帮助很大。Claude Code 是 11 月以来的第一次革命，然后现在有一个新现象，我认为是从 11 月才开始发生的，就是有一个叫 OpenClaw 的东西，OpenAI 刚刚收购了它，但它是一个开源包。所以，我认为他们基本上只是把人员招进来了。

I think it helps a lot actually. Claude Code was sort of the first revolution from November and then now there's this new phenomenon that only happened since November I would say where there's this other thing called openclaw which openai just bought and it's an open source package though. So, I think they basically just kind of brought the people in.

好的。

Okay.

由一位叫彼得·斯坦伯格（Peter Steinberger）的人发起。我的意思是，这现在是一场社会运动。它实际上非常反企业。它就像自下而上的开源的定义。我现在完全相信它。可笑的是，我为我们的播客《光锥》（The Light Cone）穿上了龙虾装。互联网似乎不太喜欢那样，但我不在乎。但实际上，我穿着龙虾装的那一刻，对我来说更像是一个玩笑。我并没有真正理解这个叫 OpenClaw 的新现象。现在我明白了。我看到了这个信仰。

Started by this guy named Peter Steinberger. I mean, it's a social movement at this point. It's actually very anti-corporate. It's like the definition of open source from the bottom up. And I'm a total believer in it now. The ridiculous thing is I dressed up in a lobster suit for our podcast, The Light Cone. And the internet didn't seem to like that very much, but I didn't care. But I actually at that moment wearing the lobster suit, it was more kind of a joke to me. I didn't really get that there was this new phenomenon called openclaw. And now I get it. I see the religion.

那么，你知道电影《她》（Her）吗？

So, do you know the movie Her?

嗯。

Yeah.

你想拥有电影《她》中那个 AI 的一个版本，让你控制 AI，还是想使用由企业控制的那个？

Would you want to own a version of that AI from the movie Her where you have control of the AI or would you want to use it controlled by the corporations?

哦，显然自己控制总是更好的。

Oh, obviously it's always better to control things yourself.

所以，这就是为什么 OpenClaw 如此强大。我一直用这个例子。我的意思是，你用 iPhone 吗？

So, that's why Open Claw is so powerful. I use this example all the time. I mean, do you use an iPhone?

我用。

I do.

你用 Siri 吗？

Do you use Siri?

不用。

I don't.

没人用它。原因是事实证明苹果并不擅长做那些事。但如果我告诉你，我要发布一个真正能用的开源版本呢？这就是我过去 11 天基本上在做的事情。就像这个东西，那就是 GBrain。它实际上会连接到你的电子邮件、联系人、日历。它读取了我过去 18 年的 Apple 备忘录。它读取了我的电子邮件。它查看了我过去 18 年的日历。

No one uses it. And the reason why is it turns out Apple isn't great at doing that stuff. But what if I told you that I'm going to release like an open-source version of it that actually works? And that's what I've basically been working on the last 11 days. Like if this thing, that's what GBrain is. It'll actually connect into your email, contacts, calendar. It read my Apple notes for like the last 18 years. It read my emails. It went and looked at my calendar for the last 18 years.

然后它映射了一切。它查找了我遇到的所有人。它实际上查看了他们的情况，然后说：“嘿，你知道，你几年前见过这个人。这是他们现在的近况。”

And then it mapped everything. It looked up all the people I've met. It actually checked up on them and be like, "Here, this, you know, you met this person a few years ago. This is what they're up to now."

太棒了。

Amazing.

所以想象这个 AI，但与其让某个公司拥有并用它来监视你，不如说你真正拥有它。你拥有代码。你拥有数据。没有其他人能访问它。然后你可以实际告诉它做事，你可以自己编程。所以是的，我们正处于中间

So imagine this AI, but instead of the AI that some corporation owns and uses it to surveil you, it's like you literally own it. You own the code. You own the data. Nobody else has access to it. And then you can actually tell it to do stuff and you can program it yourself. So yeah, we're in the middle

而且你不需要知道如何编码就能做到。对吗？

and you don't need to know how to code to do it. Is that correct?

我的意思是，我正在试图找出如何让任何人都能做到这一点。所以这就是我目前正在进行的一些实验。所以我 11 天前发布了这个。它在 GitHub 上已经有 8000 颗星了。我们一直在用的类比是，当苹果推出 Apple 2 时，让苹果成为一家公司的是，它实际上是第一个非工程师也能使用的个人电脑的商业可用版本。那个时刻还没有发生在 AI 上。现在，它仍然是企业级的。它仍然非常非常昂贵。它需要大量的定制。你必须雇佣所有这些工程师来为你运行它。然后这映射到 70 年代计算机的样子。没错。就像史蒂夫·乔布斯（Steve Jobs）实际上就是在那个时代长大并生活的，当时他和史蒂夫·沃兹尼亚克（Steve Wozniak）坐在山景城的车库里创造苹果电脑。

I mean, I'm trying to figure out how to make it so that anyone can do it. So that's what that's some of the experiment that I'm in the middle of right now. So I released this 11 days ago. It's got 8,000 stars on GitHub. The analogy we've been using is when the Apple 2 came out by Apple, the thing that made Apple as a company was uh it was actually the first commercial usable version that a non-engineer could use of a personal computer. That moment hasn't happened yet for AI. Right now, it's still corporate. It's still very, very expensive. It's requires a lot of customization. You have to hire all these engineers to run it for you. And then this maps to how computers were back in the 70s. Exactly. Like Steve Jobs literally grew up and lived in that moment in age when him and Steve Wozniak were sitting in a garage in Mountain View creating Apple computer.

他们当时正处于必须焊接面包板来制造 Apple 1 的时刻。

They were at this moment where they had to solder a breadboard to create the Apple 1.

在那之前，是这些巨大的房间大小的计算机，没有人，没有普通人能拥有一台。只有机构和公司才能拥有。

And before that it was these giant room-sized computers, and nobody, no normal human being could have one for themselves. Only institutions and corporations.

是的。

Yeah.

所以，我们再次处于那样的时刻，而且是一种更强大的技术。我认为个人电脑确实革命化了世界。而这就像是下一个革命。它正在发生。AI 的自制计算机俱乐部正在发生，任何人都可以真正参与进来，成为这场革命的一部分。

So, we're exactly at that moment again and at a much more powerful sort of technology. I think personal computers did revolutionize the world. And then this is like the next revolution. It's happening right now. The Homebrew Computer Club for AI is happening right now, and anyone can actually come in and be a player and a part of this next revolution.

告诉我你经历过的每一次技术革命。

Tell me every tech revolution you've lived through.

天哪。我的意思是，我出生于 1981 年，所以我想我出生在个人电脑革命时期。我的第一台电脑是和爸爸一起的，因为我爸爸真的很喜欢……我的意思是，回想起来，我真的很感谢他对那件事如此热衷，因为，你知道，这非常幸运。

Oh my god. I mean, well, I was born in 1981, so I guess I was born during the personal computer revolution. My first computer was with my dad, because my dad really loved... I mean, in retrospect, I really thank him for being really into that, because, you know, it's super lucky.

是的。

Yeah.

我记得很早就听说过 Kanye，比如设法获得了一些早期的音频设备，这实际上是成为能够创作音乐的人的一部分。所以，我觉得我在 5 岁时得到一台 PC XT，我直到后来才学会用笔写字。我学会了打字。我写的第一篇故事，我想，是在 IBM PC XT 上的 WordStar 3.3 里写的，这非常有趣。但是，是的，个人电脑是一个非常大的想法。以前，只有 Digital 或 IBM 或者像你的大学或你工作的大型公司这样的巨型组织会允许你分时使用，而且非常非常有限，不能做很多事情。这有点好笑，因为我们正处于 AI 的那个时代。

I remember hearing about Kanye early on, like managed to get some early audio equipment, and that was actually a part of becoming someone who could make music. So, I feel like me getting a PC XT when I was 5 years old, I didn't learn how to write with a pen until later. I learned how to type. And the first story I ever wrote was, I think, in WordStar 3.3 on an IBM PC XT, which is very funny. But yeah, the personal computer was a really big idea. Before, only Digital or IBM or just these giant organizations like your university or a giant corporation you worked at would let you time share, and it was very, very limited and it couldn't do a lot of stuff. It's kind of funny because we're sort of exactly in that age with AI.

而你看到了一个将被民主化的时代。

And you see a time when that will be democratized.

是的，那就是“开爪”时刻。我认为那正在发生，那场革命正在发生。我认为很多人，十亿人正在使用 ChatGPT，而我看到的疯狂的事情，很多科技界人士正在看到，刚刚开始看到，我的意思是世界上可能有一千个人真正相信这一点，但每个人都将运行自己的 AI 智能体，完全由他们控制，由他们编程。

Yeah, that's the open-claw moment. Like I think that that's happening, that revolution is happening right now. I think a lot of people, a billion people are using ChatGPT, and the wild thing that I see, that a lot of people in tech are seeing, just beginning to see, I mean there's probably a thousand of us in the world that really believe this, but everyone is going to run their own AI agent that is totally under their control that they program.

它还会连接到 ChatGPT 或 Anthropic 吗，还是不一定？

Will it still be connected to either ChatGPT or Anthropic, or not necessarily?

不一定。目前，我认为会是，但希望是你能选择。我认为这实际上将是未来几年的大问题之一。有一种噩梦般的情景，一个公司或一个实体或一个政府控制着最大、最好、唯一的超级智能。我认为在那个单极世界里，真的会发生非常糟糕的事情。但好消息是，目前我们似乎还没有走上那条轨道。

Not necessarily. For now, I think it will be, but the hope is you'll be able to choose. I think that that's actually going to be one of the big enduring questions in the next few years. There's sort of this nightmare scenario where one company or one entity or just one government has control of the biggest and best and only superintelligence. And I think in that unipolar world, really, really bad things will happen. But the good news right now is it doesn't look like we're on that trajectory yet.

你对 AI 中的偏见有什么看法？

What are your thoughts on bias in AI?

这就是控制非常重要的地方。所以，实际上这是一个非常有趣的辩论，对吧？会有人说应该只有一个 AI，而我想控制它。

This is where control matters a lot. So, and actually this is a very interesting debate, right? There will be people out there who say there should only be one AI and I want to control it.

对。

Right.

总会有那样的人。

There will always be people like that.

是的。有点像那种世界政府的人。

Yeah. There's sort of like the one world government people.

是的。

Yeah.

然后还有那种叛逆的人，他们说：“不，不，不。我想当海盗。我不想加入海军。我想当海盗。”就我个人而言，我是海盗队。所以，你知道，我认为个人责任、个人自由、个人计算和个人价值。

And then there's the sort of renegade people who are like, "No, no, no. I want to be a pirate. I don't want to join the Navy. I want to be a pirate." And personally, I'm team pirate. So I, you know, I think that personal responsibility, personal liberty, and personal computing and personal value.

认为对一个人正确的事情对所有人都正确的想法是疯狂的想法。

The idea that the thing that's right for one person is right for everybody else is a crazy idea.

我绝对不想生活在那个世界里。

I definitely don't want to live in that world.

这是一个非人化的想法。

It's a dehumanizing idea.

完全同意。所以，我认为我们现在有选择。这就是为什么我认为个人 AI 运动非常非常重要。

Totally. So, and I think we have choices right now. And that's why I think the personal AI movement is really, really important.

现在，大约一年前，有一个来自中国的开源 AI。

Now, there was an open-source AI that came from China maybe a year ago.

哦，是的。是的。是的。那些是像 DeepSeek 或 Qwen 这样的模型公司。

Oh, yeah. Yeah. Yeah. Those are model companies like DeepSeek or Qwen.

当它们出现时，情况发生了什么变化？

How did that change things when they appeared?

它把所有人都吓坏了，英伟达股价下跌了 10%，我在那天买了，这很有趣。所以，嗯，我认为有两件事在发生。一是这对海盗有利。这对自由队有利。

It scared the crap out of everyone and Nvidia stock went down by 10% and I bought on that day, which was funny. So, um, I think there are two things happening. One is that it's good for the pirates. It's good for team Liberty.

因为它表明……

Because it showed that...

不可能只有一个。那里可以有很多很多模型公司，这将非常重要。但第二件可怕且可能糟糕的事情是，中国公司能够做到这一点的原因是他们实际上从美国大型模型公司那里窃取了智能。

There can't be just one. There can be many, many model companies out there, and that's going to be really important. But the second thing that is scary and maybe bad is that the reason why the Chinese companies are able to do that is they're actually stealing the intelligence right out of the giant American model companies.

所以两件事可以同时成立。可以是真的，而且讽刺的是，你知道，在北美和欧洲版权法霸权之外运作的中国公司可以这样做，然后他们可以发布它，然后同时他们做了一件不完全清楚是否合法或好的事情。

So two things can be true at once. It can be true that, and ironically, you know, Chinese companies that are operating outside of the American and Northern American and European hegemony of copyright law can do this, and then they can release it, and then at the same time they're sort of doing a thing that is not totally clear that should be legal or good.

不过，难道所有美国公司不也是建立在偷来的材料上的吗？

Aren't all the American companies built on stolen material as well though?

是的，这是一个有力的论点。是的，我的意思是你可以说那是合理使用，而且你知道，大型模型公司来说“实际上这只是我们的版权”有点奇怪。

Yeah, that's a strong argument. Yeah, I mean you could say that's fair use, and you know it's kind of odd for the big model companies to come and say actually this is just our copyright.

他们什么都没写，什么都没创造。他们只是把它堆在一起。

They didn't write anything. They didn't create anything. They just bunched it up.

没错。是啊。他们把它汇集起来，聚合了。

That's right. Yeah. They brought it together, aggregated it.

是啊。而且你知道，一方面，我们都真的真的……好吧，不是每个人都高兴他们这么做了，但你知道，我觉得很明显这里面有很多价值。它是个非常强大的工具。而这就是我们坐在这里讨论的争议。让我停下来思考的是，Sam Altman 在旧金山的家遭遇了不止一次，而是两次不同的暴力事件。

Yeah. So, and you know, on the one hand, we're all really really... Well, not everyone's glad that they did it, but you know, I think that it's clear that there is a lot of value to it. It's a really powerful tool. And then that's the debate as we're sitting here. It gives me pause that Sam Altman had not one but two different violent incidents at his home in San Francisco.

跟我说说这事。我对此一无所知。

Tell me about that. I don't know anything about that.

呃，有人朝他房子扔了燃烧瓶，然后就在所有新闻媒体公布了他房子和地址的照片的第二天，就发生了一起驾车枪击。

Uh, someone threw a Molotov cocktail at his house, and then literally the next day after all of the news media published photos of his house and the address, there was a drive-by shooting.

哇。

Wow.

是啊。我是说，没人受伤。呃，显然只是财产损失。为什么主流媒体对科技有这么大的仇恨？

Yeah. I mean, no one hurt. Um, obviously just property damage. Why does the mainstream media have such a hatred of technology?

我是说，一方面，这有点道理。主流媒体确实因广告的消亡而深受其害。你可以把新闻业的衰落追溯到……实际上我不知道为什么人们不把功劳多给 Craigslist 一点。我是说，你可以直接把报纸广告收入的下降归因于 Craigslist。另一方面，我觉得人们很高兴他们能上 Craigslist 卖掉自己的烤面包机。而且 Craigslist 是免费的。所以，我认为科技现在进入世界的方式有很多悖论。这就是一个很好的例子。

I mean, on the one hand, it sort of makes sense. The mainstream media has really suffered from the death of ads. You can trace it to journalism being actually broken down by... I don't know why people don't give Craigslist a little bit more credit for this actually. I mean, you can directly trace the decline of ad revenue to newspapers to Craigslist. And on the other hand, I think people are pretty happy that they can go to Craigslist and sell their toaster. And Craigslist is free. So, I think there's a lot of paradoxes in how technology comes out into the world right now. And that's a good one.

那……你该怎么办？比如，我们该保护报纸吗？你知道，是不是应该有法律阻止这种情况发生？另一方面，Craigslist 是免费的，它提供的服务比打电话给报纸花 20 美元卖车之类的要优越得多。所以，规则应该是什么？事情在变。这可以吗？然后我总想到，你熟悉那本有趣的管理书吗？就是《谁动了我的奶酪？》

That's... what are you supposed to do? Like, were we supposed to protect the newspapers? You know, were there supposed to be laws that prevented this from happening? On the other hand, Craigslist is free and it's giving a service that people are using that is way superior to having to call up the newspaper and pay $20 just to sell your car or something like that. So, what should the rules be? Things change. Is that okay? And then I always think about, are you familiar with this funny management book? It's um, Who Moved My Cheese?

不，我知道书名，但没读过。

No, I know the title, but I haven't read it.

是啊，我是说，你不需要读它，因为基本上就那点事。我觉得就是人们真的不想自己的奶酪被动，而科技是社会里主要的奶酪移动者，所以人们非常生科技人士的气，这说得通。

Yeah, I mean, you don't need to read it because that's basically the whole thing. I think it's just people literally don't want their cheese to be moved and technology is the primary cheese mover in society and so it makes sense that people are very upset with tech people.

未来有没有一个世界，所有内容片段都能上区块链，然后当它们被整合进 AI 时，你知道东西的来源。

Is there a world in the future where all pieces of content can be on the blockchain and then when they're integrated into AI, you know where things are coming from.

是的。基本上，很多是关于围绕它的法律是什么？我们怎么让人们得到报酬？我们怎么处理？有一段时间，有这样一个想法，艺术家应该通过像 Patreon 这样的东西获得报酬，我实际上投资了它。我是说，我见过 Jack Conte 和 Sam Yam。我觉得他们其实刚大学毕业。我是说，他们是大学室友。一个是出色的音乐家，一个是出色的计算机科学家。

Yeah. Basically, a lot of it is what are the laws around it? How do we get people paid? What do we do about it? For a while, there's this idea that artists should be paid via things like Patreon, which I actually invested in. I mean, I met Jack Conte and Sam Yam. I think they were actually right out of college. I mean, they were college roommates. And one is a great musician and one is a great computer scientist.

他们怎么想到 Patreon 这个点子的？你知道吗？

How'd they get the idea for Patreon? Do you know?

我觉得 Jack 有一个叫 Pomplamoose 的乐队，他在 YouTube 上，基本上……我想那大概是 Kickstarter 出现的时候，但 Kickstarter 有个问题，如果你运营一个乐队，你不是在做产品。你不是在做一次性东西。你不想为了赚钱每周都卖一件新 T 恤。所以他说，“我们为什么不直接那样做？直接采用赞助模式怎么样？”比如，我真的很喜欢这个音乐家。我真的很喜欢这个艺术家。让我每月付他们 10 美元。所以，我觉得 Patreon 会继续增长，那是个很好的模式，有点像是解决这个难题。

I think Jack had a band called Pomplamoose and he was on YouTube and he just basically... I think this was about when Kickstarter happened, but Kickstarter had a problem which is if you're running a band, you're not making products. You're not making one-off things. You don't want to sell a new t-shirt every week just to make money. So he says, "Why don't we just do that? How about the straight-up patronage model?" Like, I really like this musician. I really like this artist. Let me just pay them $10 a month. So, I think Patreon just continues to grow and that's a great model for sort of squaring the circle.

他们有竞争对手吗？还有别人做 Patreon 做的事吗？

Do they have competition? Does anyone else do what Patreon does?

我觉得他们现在对局势把握得挺好，但 Kickstarter 还存在，GoFundMe 也还存在，还有很多不同的东西，但我觉得 Patreon 把所有超级有创造力、只想要赞助的人聚集在了一起。而我对 UBI 的疯狂想法其实是，我们难道不应该扩展这个吗？关于当 AI 来了并取代工作时，所有人会做什么，有这种大规模、深度的辩论。我想我的答案第一是，我认为人们会做更多很酷的事情来互相服务。不只是做人们想要的东西，而是做一大堆人们想要的东西。这差不多是我的第一个答案。

I think they sort of have a good lock on what's going on right now, but Kickstarter still exists and I think GoFundMe still exists and there are lots of different things, but Patreon, I think, has just brought together all the people who are super creative who just want patronage. And my crazy idea for UBI is actually that shouldn't we just expand that? This sort of big, big deep debate about what are all the people going to do when AI comes and takes the jobs. And I guess my answer is one, I think people are going to do a lot more cool stuff to serve each other. Not just make something people want, but make a whole lot of stuff that people want. Like that's sort of my first answer.

然后我的第二个答案是，“人们想要的东西”这部分不必只是更多的企业 B2B 软件。不必是那样。它可以是更多的音乐。可以是更多的艺术。可以是更多有趣的派对。我听说，是 Ari Emanuel 吗？我听说他开始收购所有这些大型的现场活动之类的东西。

And then my second answer is the "something people want" part doesn't have to be just more enterprise B2B software. It doesn't have to be that. It could be more music. It could be more art. It could be more fun parties. I was hearing about, was it Ari Emanuel? I heard that he was starting to buy all of these big like live event things.

我就想，那太聪明了。实际上，那正是趋势所在，因为突然当每个人都有这么多空闲时间，他们会想要玩得开心。他们会想去派对。他们想要有趣的体验。他们会想体验作为人的感觉，而不是作为一个机械工人。不是 Aldous Huxley 说过吗，所有这些技术最终会导致每周 20 小时工作制，虽然还没实现，但也许现在是时候了。

I'm like, that's so brilliant. Actually, that's going to where the puck's going to be 'cause suddenly when everyone has all this free time, they're going to want to have fun. They're going to want to go to parties. They want to have interesting experiences. They're going to want to experience being a human more than being a drone worker. Wasn't it Aldous Huxley who talked about you know all of this technology will eventually result in a 20-hour work week and it hasn't happened yet but maybe now is the time.

那么在个人电脑革命之后，你经历的下一个是什么？

So after the personal computer revolution, what was the next one that you experienced?

我觉得下一个显然是互联网。我是说，我记得大概 1996 年左右，我就像个高中新生，在“网络日”到处跑，在我整个高中拉 Cat 5 网线。Al Gore 到处说信息高速公路会改变一切，它确实改变了。就像他说对了。实际上，我在 Fremont 长大，那是全国第一个有有线调制解调器的地方。所以，我可以在卧室里运行一台 Linux 机器，我确实这么做了。我学会了做网站和编程。老实说，那超级有趣。我是说，基本上 Paul Graham 是第一个这么做的，等到我做的时候，大概是 1995、96 年，大家都开始这么做了。

I think the next one was clearly the internet. I mean, I remember like 1996 or so, like I was like a freshman in high school running around and running Cat 5 networking cable all around my high school on net day. Al Gore was running around talking about like the information superhighway is going to change everything and it did. Like he was right. Actually, I grew up in Fremont, which was the first place in the entire country to get cable modems. So, I could run a Linux box in my bedroom, and I did. And I learned how to make websites and program them. And that was super fun, honestly. I mean, basically Paul Graham did it the first time, and then by the time I did it, it was like maybe 1995, 96 and everyone was starting to do it.

那时候你获取信息容易吗？

Was it easy for you to get information in those days?

我们还有黄页。你知道，我的有趣故事是我的第一份工作，我们当时住在公寓里，我环顾四周，看着当时三四岁的弟弟。我心想，伙计，我不想我们在公寓里长大。美国梦是我们应该住在一栋房子里。那我们怎么做到呢？然后我低头看脚下，有一堆黄页。我父母有点囤积癖，所以我不知道。就是有一大堆黄页。我拿起一本，翻到互联网部分。我学会了怎么做网页。互联网的有趣之处在于，在互联网上，没人知道你是一条狗或一个 12 岁的孩子。

We still had the yellow pages. You know, my funny story is my first job, we were living in apartments, and I looked around and I looked at my brother who was three or four at the time. And I was like, man, I don't want us to grow up in an apartment. The American dream is that we should live in a house. So, how do we do that? And then I look down at my feet and there's a stack of yellow pages. My parents are kind of hoarders, so I don't know. There's like a whole bunch of yellow pages. I picked up one, I flipped it to the internet section. I learned how to make web pages. The funny thing about the internet is on the internet, nobody knows you're a dog or a 12-year-old.

所以，我在 12、13 岁的时候赢得了所有这些网页设计奖项。然后那时我就想，如果我是个获奖网页设计师，那是个不错的头衔。我就可以直接在网上给陌生人打电话。于是，我找到了一家叫 Infolane 的公司。有趣的是，他们做城市网站。他们做了像弗里蒙特市这样的网站。我觉得那很酷。我就直接给他们打了电话，说：“嘿，我叫 Gary，我 14 岁，我是个获奖网页设计师，我在找工作，你们招网页设计师吗？”就这样我得到了第一份工作。我骑着自行车大约两英里到 Infolane，那是个很小的三间办公室。主要的技术人员是个叫 Andy McCrae 的家伙，他是个留着胡子的 Unix 程序员，饱经风霜，非常典型的新技术 CTO。然后 CEO 是个叫 David Hilstrom 的人。他梳着大背头。他穿着那种金融衬衫，白色领子，下面是蓝色条纹，还有百褶卡其裤和背带。说实话，他是个很迷人的人。他教会了我直视对方眼睛和有力握手的价值。那是我爸爸不会教我的，因为他自己从没学会。他搞不懂那部分，你知道吧。所以这真的很有趣。我家里一团糟。然后我可以骑着自行车去上班，听到那些古怪又超级聪明的人，他们不互相大喊大叫，只是把事情做好，做城市网站，你知道，服务公众利益。而我们只是通过在电脑前打字来做这些。还有一个设计师，我第一个合作的设计师叫 Jackie。我很遗憾不记得她的姓了。我觉得我从她那里学到了关于设计的一切。她教了我一种叫“眯眼测试”的方法。

So, I won all these web design awards as a 12 or 13-year-old. And then by then I said, well, if I'm an award-winning web designer, that's a pretty good tagline. I could just start cold calling people on the internet. And so, I found this company called Infolane. They, funny enough, made city websites. They made like the city of Fremont website. And I thought that was cool. And I just cold called them and I said, "Hey, my name's Gary. I'm 14 years old. I'm an award-winning web designer. I'm looking for a job. Are you guys hiring web designers?" And so I got that first job. Rode my bike about 2 miles over to Info Lane, this tiny little three-room office. The main tech guy was this guy named Andy McCrae, who's like a bearded Unix programmer, just grizzled, really new as tech classic CTO. And then the CEO was this guy named David Hilstrom. And he had the slick back hair. And he had like the finance shirt that had the white collar but then like the blue stripes underneath and then the pleated khakis and like the suspenders. And he was a fascinating guy honestly. Like he taught me the value of looking someone in the eye and the strong handshake. And that was something that wasn't something my dad was going to teach me because he never learned that. He couldn't figure that part out, you know. So this is really interesting. My crew at home. Everything was a mess. And then I could ride my bike and go to work and hear like quirky, super smart people who weren't yelling at each other just getting things done and making city websites, you know, like serving the public good. And we were doing it by just typing in front of a computer. And there was a designer I got to work with, my first designer I ever got to work with was named Jackie. I feel bad I don't remember her last name. I feel like I learned everything about design from her. She taught me this thing called the squint test.

她比你大多少？

How much older was she than you?

我想她肯定有 55 岁了。她大概……

I think she must have been 55. Like she was...

所以她是成年人，而你 14 岁。

So she was a grownup and you were 14.

是的，我 14 岁。我就坐在她旁边。

Yeah, I was 14. Like I was just sitting right next to her.

所以你把她当作导师？

So you consider her like a mentor?

哦，绝对是的。是的。我的意思是，我有很多导师。宇宙似乎就这样展开，就像这是你需要学习的人。你只需要回应召唤。我想你只需要在场。

Oh, absolutely. Yeah. I mean I had so many mentors. I mean the universe seems to just unfold and it's like here's someone you need to learn from. You just have to answer the call. I guess you just have to be present.

那互联网之后的下一个是什么？

And what was the next one after the internet?

天哪，互联网太大了。关于我的有趣之处是，我学会了做网站。我通过冷电话得到了第一份工程工作，在那里我学会了编程。那是旧金山一家叫 Adjacency 的设计公司，现在由我的朋友 Andrew Saither 经营。他的故事很疯狂。他就是冷……他没有冷邮件，因为财富 500 强公司没有电子邮件。他会冷寄这些漂亮的平面设计简报，说明互联网能为这些大品牌做什么，他署名是威斯康星大学出来的 18 或 20 岁年轻人。

Oh my god, the internet was so big. So the funny thing about me is I learned how to make websites. I cold called to get my first engineering job where I learned to code. It was this design firm called Adjacency in San Francisco, run by now my friend Andrew Saither. His story was crazy. He just cold... he didn't cold email because Fortune 500s didn't have email. He would cold send these beautiful graphic design briefs of what the internet could do for these giant brands, and he signed them as an 18 or 20-year-old out of University of Wisconsin.

好主意。

Great idea.

在 1998 年，就像……

In 1998, like...

只是对它能成为什么有远见。

Just had a vision for what it could be.

是的。所以那时网站不像现在这样。它们看起来像简陋可笑的“正在建设中”的标志和灰色背景。所以 Andrew Saither 想出了如何制作全出血的网站，看起来像杂志，那是他们想出来的东西，我很喜欢。我的意思是，我这么说只是因为技术在每一个时刻，都有一种对新事物的崇拜，以惊人的方式创造价值。

Yeah. So back then websites didn't look like the way they do now. They looked like dinky ridiculous like under construction signs and gray backgrounds. So Andrew Saither figured out how to make full-page bleed websites that look like magazines, and that was the thing they figured out, which I love. I mean I only say this because at every moment in technology, there is like a cult of the new that creates value in a way that is just astonishing.

你描述的每一件事都改变了一切，而且现在都还活着。

Each of these things that you're describing changed everything and are all still living now.

是的。我的意思是，我想……坐在这里的香格里拉，这是许多歌曲诞生的地方，对吧？所以无论是音乐、艺术、电影，还是坦率地说代码，代码实际上是其中之一。所以我得以和这些传奇人物一起工作，他们想出了这些。所以 Andrew Saither 为 Steve Jobs 创建了互联网上第一个 Apple 电子商务商店。所以我得以工作。

Yeah. I mean I think about... I mean sitting here in Shangri-La, this is the place that so many songs were born, right now, right? And so whether it's music or art or movies or frankly code, code is one of those things actually. And so I got to work with these legends who came up with it. So Andrew Saither created the first Apple e-commerce store on the internet for Steve Jobs. And so I got to work.

那是什么商店？

What was that store?

哦，他们卖……你实际上就是去 apple.com。那时能够做电子商务是个全新的想法。

Oh, they sold... You literally just went to apple.com. It was a brand new idea back then to be able to do e-commerce.

在线销售 Apple 设备。

Selling Apple devices online.

是的。所以你可以买一台 iMac，这很漂亮。我的意思是，我觉得那是我见过的最酷的东西。就像 Apple 是天上之物。

Yeah. So you could buy an iMac, and so this was beautiful. I mean, I thought that was the coolest thing I'd ever seen. Like Apple was this thing up in the sky.

他们制造美丽的产品，你知道。

They made beautiful products, you know.

在那之前，你必须去哪里买一个？

And before that, where would you have to go to get one?

哦，你必须去 Apple 商店。哦，我想 Apple 商店还不存在。哦，糟糕。

Oh, you'd have to go to an Apple store. Oh, I guess Apple stores didn't exist yet. Oh, shoot.

可能很难买到。我不知道……是的，你必须去零售商那里。

It was probably hard to get. I don't... Yeah, you'd have to go to a retailer.

而且我想在这个国家的许多地方不会有零售商。

And I imagine in many places in the country there wouldn't be a retailer.

是的，没错。而且你不能上网买它。所以，就像……是的，我们理所当然认为的所有这些东西都是由和你我没什么不同的人创造的，这真是深不可测。我觉得那简直是 Steve Jobs 的原话。他就是这么说的。

Yeah, that's right. And you couldn't go online to buy it. So, it's like... Yeah, it's unfathomable that all these things that we take for granted were created by people no different than you or me. I think that's like a straight-up Steve Jobs quote. That's what he said.

这是真的。

It's true.

他说：“一旦你意识到这一点，你就再也不会一样了。”

He said, "Once you realize that, you'll never be the same."

是的。不是特殊的人在做事。我的意思是他们很特别，但每个人都有一点特别。

Yeah. It's not special people doing these things. I mean they are special but everyone's a little bit special.

正是如此。

Exactly.

LMNT 元素电解质。你是否曾在剧烈运动或在阳光下度过漫长一天后感到脱水？你想最大化你的耐力并感觉最佳吗？将元素电解质加入你的日常。表现更好，睡眠更深。改善你的认知功能。体验稳定能量的增加，更少的头痛和更少的肌肉痉挛。元素电解质。在桑拿中饮用。清爽口味包括葡萄柚、柑橘、西瓜和巧克力盐。配方含有钠、钾和镁的完美平衡，让你全天保持水分和活力。这些矿物质有助于传导为你的神经系统供电的电流，让你发挥最佳状态。元素电解质无糖、生酮友好且味道极佳。矿物质是生命之本。所以访问 drinklnt.com/tetra，和元素电解质一起保持咸味。LMNT，下一场革命。

LMNT element electrolytes. Have you ever felt dehydrated after an intense workout or a long day in the sun? Do you want to maximize your endurance and feel your best? Add element electrolytes to your daily routine. Perform better and sleep deeper. Improve your cognitive function. Experience an increase in steady energy with fewer headaches and fewer muscle cramps. Element electrolytes. Drink it in the sauna. Refreshing flavors include grapefruit, citrus, watermelon, and chocolate salt. Formulated with the perfect balance of sodium, potassium, and magnesium to keep you hydrated and energized throughout the day. These minerals help conduct the electricity that powers your nervous system so you can perform at your very best. Element electrolytes are sugar-free, keto-friendly, and great tasting. Minerals are the stuff of life. So visit drinklnt.com/tetra and stay salty with element electrolytes. LMNT, next revolution.

天哪。所以在互联网之后，社交和它之间有点难分。我会说 Facebook 肯定是个大事。社交媒体，那是因为网络而发生的。对我来说有趣的故事是，我学会了制作所有这些网页，然后我以为网络死了。所以就像 2003 年我是斯坦福的计算机工程专业，毕业了，没有初创公司。就像我不能……所有初创公司在 2003 年都死了。Web 1.0 已经崩溃了。有趣的是，我其中一个夏天在 the man.com 度过。我的意思是，之前我们在谈论风险投资。回想起来，那是那种糟糕的风险投资。

Oh my god. So after the internet, it's sort of a tossup between social. I would say Facebook was a big deal for sure. Social media and that happened because of the web. The funny story for me is I learned how to make all these web pages and then I thought that the web was dead. So it's like I was 2003 a computer engineering major at Stanford, graduated, no startups. Like I couldn't... all the startups were dead in 2003. Web 1.0 had already crashed. It's funny. One of the summers I spent at the man.com. I mean earlier we were talking about venture capital. In retrospect, that's the kind of venture capital that was bad.

当你和一个穿西装的人、而不是一个建造者去炮制一个商业想法时，你得到的就是那种东西。我记得那是一家男性杂志，大概有 20 页的 PPT，全是各种流行词。基本上就是从一个脱节的新领域里拼凑出来的。这里有一堆抽象概念，如果它很热门，如果听起来很热门，那我就得赶紧跟上。但那就像蒙着眼睛导航。而建造者能做的是，他们知道这些东西是怎么建起来的。他们有眼睛，他们能看见。所以你能直接看到，嘿，那边有个东西，我要去把它建出来。然后他们就去建了，结果很棒。那么你会说今天最好的 VC 是那些建造者吗？

When you go in and sort of astroturf a business idea with someone who is a suit and not a builder, you get an idea like that. I think it was a men's magazine, some like 20-page PowerPoint about like all buzzwords. It's basically like mediated from a new sphere of people who are just disconnected. Like here's a bunch of abstractions and if it's hot, if it sounds hot, then I better get ahead of it. But that's like trying to navigate while you're blind. And what a builder can do is they know how this stuff is built. They got eyes. They can see. So you can just see like, hey, there's this thing over here. I'm going to go build that thing. And then they build it. And then it's awesome. So would you say the best VCs today are the ones who are the builders?

我会这么说。

I would say so.

嗯。

Yeah.

我的意思是，这几乎是必要的，尤其是在早期阶段。如果你不是先成为一个建造者，你不可能成为那么好的投资者，这说不通。

I mean that's just sort of necessary, especially early stage. It just doesn't make sense that you could be that good an investor without being a builder first.

从最初的提案到最终在世界上成功的东西，想法会变化多大？

How different do the ideas change from an initial pitch until the thing that ends up being successful in the world?

两种情况都有一些。像 DoorDash 的 Tony Shu，他们一开始就是那样，后来也一直那样。只是沿途做了一些小调整。他们说，“我们要成为最好的外卖。”然后微调是，当其他人都在做城市时，他们说，“哦，我们来做郊区。”就是那个小转变让他们发展得非常非常快，打败了其他所有人。有时像 Coinbase 的 Brian Armstrong。他进来时说，“我相信比特币会变得非常非常大。”

There's some of a little bit of both. Someone like Tony Shu from DoorDash, that's what they were and that's what they were always. And then there were like little adjustments along the way. They said, "We're going to be the best food delivery." And then the micro pivot there was like as everyone else was working on cities, they said, "Oh, we're going to do suburbs." And just that little shift allowed them to grow really, really fast and beat everyone else. Sometimes it's like Brian Armstrong at Coinbase. He came in and said, "I believe that Bitcoin is going to be really, really big."

嗯。

Mhm.

然后他就从那里倒推。所以他建的第一个版本，我记得叫 BitBank。他最初想做的其实只是为手机做一个比特币客户端。然后这就是创始人的行动。你知道，你在外面，你试图做出这个东西。然后你意识到，哦，这行不通。整个比特币客户端塞不进安卓手机。即使塞进去了，你也得坐在那里，插着电五小时才能让它可用。我的意思是，好吧，那不行。让我们转身，让它跑在服务器上。然后他做了，又不行。他说，“嗯，为什么？其实很难获得比特币。我建了一个能存比特币的东西，但你怎么把它放进去？”然后他去做了那个，就成了 Coinbase。所以

And then he just worked backwards from that. And so the first version of what he built, I think it was called BitBank. All he was trying to do actually initially was build a Bitcoin client for the phone. And then this is like the act of the founder. You know, you're out there, you're trying to make this thing. And then you realize, oh, it's not going to work. You the whole Bitcoin client can't fit on an Android phone. And even if it did, you'd have to sit there, leave it plugged in for five hours a day just to make it usable. I mean, okay, that's not going to work. Let's turn around and make it work on a server. And then he did that and it again didn't work. And he said, 'Well, why? It's actually really hard to get Bitcoin. I built a thing that holds Bitcoin, but how do you actually get it in there? And then he went and did that and it became Coinbase. So

太棒了。

Amazing.

所以这些东西总是在演变，总是在变化。真正伟大的创始人似乎能够直接问第一性原理的问题，然后去追求那个东西。

So these things always evolve. They always change. And really great founders seem to be able to just ask the first principles questions and then go after that thing.

带我了解一下这个 13 周的项目。

Walk me through the 13-week program.

哦，是的，我们现在正处于春季批次的第二周。说实话，前 10 周大部分时间就是构建。所以你进来，这在我脑子里还很新鲜。我们刚过完基础部分。有趣的是，你必须做多少基础工作，但我们有一群非常有经验、成功的创始校友，他们回来再次参加 YC。看着他们不得不回到初学者心态，很有趣。

Oh yeah, we're in week two right now of the spring batch. Honestly, most of the first 10 weeks is just building. So you come in and this is fresh in my head. We just went through the basics. And it's funny how much you have to do the basics, but we have a bunch of really experienced, successful former founders who are coming back and doing YC again, alums. And it's funny to watch them have to go back to beginner mind.

所以这些人是第二次参加这个项目。

So these are people who are going through the program a second time.

是的，完全正确。所以我们一些最好的创始人，比如 Rippling 的 Parker Conrad，第二次回来，越来越多的校友回来再次参加 YC。差不多。

Yeah, totally. So some of our best founders like Parker Conrad of Rippling came back through a second time and increasingly alums are coming back and doing YC again. Sort of.

你觉得他们这样做是出于怀旧，还是因为他们通过这样做得到了什么具体的东西？

Do you think they're doing it for a nostalgic reason or because they get something specifically by doing it?

我的意思是，我认为做得好的 YC 批次更像是一场复兴或一种体验，而不是一个教育产品之类的东西。我认为它是一种体验。然后我们的现实完全是由其他人构建的。所以前几周的首要价值实际上是脱离你的正常节奏，说，“好吧，我打乱了我的整个生活。我来到这里。我打算专心做这家公司，我要为其他人创造一个东西”

I mean, I think the YC batch done right is a little bit more like a revival or an experience than an educational product or something like that. I think it's an experience. And then our reality is constructed entirely with other people. And so the number one value the first few weeks is actually disconnecting from your normal rhythm and saying, "Okay, I displaced my entire life. I'm coming over here. I'm intentionally going to work on this company and I'm going to create a thing for other people

排除其他一切，我要只专注于那件事。仅这一点就非常有价值，因为否则，你知道，我们说的第一件事是，在其他任何情况下，你最好的朋友打电话给你，需要搬家，那个周末需要你的帮助。现在你有借口了。

To the exclusion of everything else and I'm going to focus on just that. And that alone is so valuable because otherwise, you know, the first thing we say is like in every other context, your best friend calls you and needs to move and needs your help that weekend. Now you've got an excuse.

嗯。

Yeah.

而答案是抱歉。

And the answer is sorry.

嗯。

Yeah.

总是明天。是的，现在我在做一件事。所以有趣的是，在你的生活中留出有意的空间，就是这样。

Is always tomorrow. Yeah, right now I'm doing a thing. And so it's kind of funny how much just making intentional space in your life is just that.

嗯。

Yeah.

然后以类似“相遇小组”的方式去做。就像你和所有这些其他人在一起，每周你都能看到他们的挣扎、他们的经历以及他们思维的演变。然后大约在第三或第四周，开始变成，哦，对他们有效。我落后了吗？你会说这是竞争性的，还是更像是每个人激励每个人？

And then doing it in sort of like an encounter session style way. It's like you're with all these other people and every single week you see their struggles and what they're going through and the evolution in their thinking. And then about week three or four it starts becoming, oh, it's working for them. Am I behind? Would you say it's competitive or more like everyone inspires everyone else?

理想情况下，是后者。有时是竞争性的。有趣的是，在我的 YC 批次里，我斯坦福的一个兄弟会兄弟实际上就在我那个批次。但我们的创业想法是发一封邮件得到一个博客。他们的是发一封邮件在相框上得到一张照片。所以我们和他们竞争。比如我知道他们要发布，我们得先发布。你知道，Parker Conrad 谈到他在做 Zen 的那个批次时，他在做企业健康保险，还有一家叫 Simply Insured 的公司，那真是给他屁股底下点了把火，让他把公司做得比竞争对手快 10 倍、好 10 倍。

Ideally, it's the latter. Sometimes it's competitive. What's funny is my YC batch, we had one of my fraternity brothers from Stanford actually was in my very YC batch. But our startup idea was post an email and get a blog. Theirs was post an email and get a photo on a photo frame. And so we were competitive with them. Like I knew that they were going to launch and we were going we needed to launch first. You know, Parker Conrad talks about that during his batch for Zen, he was doing health insurance for business and there was another company called Simply Insured and boy did that light a fire under his ass to make that company 10 times faster and better than his competitor.

嗯。

Yeah.

所以，主要是想激励，但另一方面，一点竞争从来不会伤害任何人。

So, mostly I wanted to inspire, but on the other hand, a little competition never hurt anyone.

听起来在这些情况下，那是现实世界中的直接竞争。

And it sounds like that's like real world direct competition in those cases.

哦，是的。

Oh, yeah.

这是你必须面对的。

Which you will have to encounter.

是的，那是非常真实的。这不同于，嗯，他建的东西和我很不一样，但我想和他一样好。

Yeah, that's super real. It's different than, well, he's building something really different than me, but I want to be as good as him.

是的，没错。这 13 周很大程度上是试图让人们专注于真正重要的事情，而不是我如何从 VC 那里筹更多钱？我如何打动金融兄弟？那些都不重要。唯一重要的是我们是否做出了人们想要的东西？

Yeah, that's right. A lot of this is the 13 weeks is a lot about trying to get people to focus on the thing that really matters instead of how do I raise more money from VCs? How do I impress the finance bro? None of that matters. The only thing matters is did we make a thing that people want?

知道 13 周后这一切就结束了，努力奋斗肯定更容易接受。

Going hard must be much more palatable knowing in 13 weeks this is over.

是的，这是一个过程。我的意思是，大多数公司，中位数公司筹集 200 万到 300 万美元，高端的人在 13 周后筹集 1000 万到 2000 万。所以竞争有真正的利害关系，然后让人们快速专注于某件事非常重要。我从未见过任何东西能以如此深刻的方式加速创始人。我花了几年的时间，有点像我的荒野岁月。

Yeah, it is a process. I mean most of the companies the median company raises $2 to $3 million on the high end people raise 10 to 20 million after 13 weeks. So there is there are real stakes to the competition and then just running people fast at a particular thing matters a lot. I've never seen anything speed founders up in such a profound way. I spent a couple years sort of in my wilderness years.

我在荒野里待了大约七年，说来好笑。说它是荒野有点夸张。我创办了自己的风投基金，非常成功。在那段离开 YC 的时间里，我尝试做种子前投资。我会遇到我真正喜欢的团队，给他们 50 万美元，然后说：‘去吧，这些文章都给你。’是的。我每周都去见他们，但他们待在 WeWork 的共享空间里，坐在咖啡厅里闲逛，就像以前在微软工作时那样。我就想，你还没改变任何东西。我想这和任何创意行业都一样。就是让自己跳出固有思维。换个地方，换一批人，换一种思考方式。这是我唯一的解释。我试着做种子前投资，让他们加速到创业速度，但从来都不管用。最后我实在受不了了，就说：‘好了，你们该去 YC 了。’如果他们足够优秀能进 YC，那他们就会加速。突然之间，神奇地，他们就会转变，一两年后，他们就能做到年收入 2000 万美元之类的。

I spent about, funny enough, seven years in the wilderness. It sounds ridiculous to call it the wilderness. I started my own VC fund that was very successful. And during that time away from YC, I would try to pre-seed. I'd meet teams that I really loved and I'd give them half a million dollars and I'd say, 'Go, here's all the essays.' Yeah. And I'd meet them every week, but they're in the co-working space in the WeWork and they're sitting there in the coffee room hanging out the way they used to when they worked at Microsoft or something. And it's like, you haven't changed anything yet. I imagine it's the same deal with any creative profession. It's like, just get yourself out of your head. Change your location, change the people you're with, change the way you think about things. And that's my only explanation. Like, I would try to pre-seed people and get them to speed up to startup speed and it would never work. And then finally I'd get frustrated and say, 'Okay, it's time for you to do YC.' And if they were good enough to get into YC, then they would speed up. And suddenly, magically, they would turn around and a year or two later, they'd be making $20 million a year in revenue or something.

你后来被邀请回去执掌 YC，这结果怎么样？

How did it work out for you to get invited back to run YC?

天哪，那是一次非常紧张的经历，因为我已经准备好了。我管理着数十亿美元，而且公司是我自己一手建立的，但 YC 的创始人保罗·格雷厄姆请我回去。

Oh man, that was an intense experience because I was good to go. I had billions of dollars under management and I had built the firm myself, but the founder of YC, Paul Graham, asked me to come back.

你知道他为什么选你吗？

Do you know why he picked you?

哦，这里有个有趣的故事。保罗和杰西卡的故事很吸引人，因为 PG，你会非常喜欢他。他就是连接着源头，老兄。他设计的 YC 的一切都有那种特质。深思熟虑、不张扬、简洁、克制。我想这就是让我回来的原因。而且我觉得 YC 本身远远超出了他的预期。我认为他因为创办 YC 而成了亿万富翁，而且到 2013 年左右，这一点已经很明显了。

Oh, here's a funny story. The story of Paul and Jessica is fascinating because PG, you would dig him so much. He's just connected to the source, man. Everything about YC that he designed has that quality about it. It's thoughtful and not ostentatious and concise and spare. And I guess that's what brought me back. And then I think YC itself way exceeded his expectation for it. I think he's a billionaire many times over from starting YC and it became sort of clear by, what, 2013 or so that that was true.

他做了多少年？

How many years had he done it?

那时已经七年了。所以他自己也经历了创办和运营 YC 的七年旅程。

Seven years at that point. So he himself went through a seven-year journey starting and running YC.

他有年幼的孩子，我当时没有意识到，但我正在 YC 工作。我算是门徒之一，后来 OpenAI 的萨姆·奥特曼也是通过 2005 年 YC 第一批走出来的。保罗写过一篇关于他的著名文章。他列了一份我们这一代五位顶级创始人的名单，他把萨姆列了进去，而那时萨姆还没有创造出后来那些成就。所以在当时，这在我们圈子里非常有争议，因为萨姆……

He had young children and I didn't realize it at the time but I was working at YC. I was sort of one of the proteges and then Sam Altman of OpenAI fame also came up through the first YC batch in 2005. Paul famously wrote an essay about him. He made a list of five top founders of our generation and he put Sam on that list when he had not created all the things that he's created since yet. So at the time it was very controversial in our community because Sam...

显然他能看到这一点。

Clearly he could see it.

PG 连接着源头，老兄。

PG's connected to the source, man.

他看准了。他看准了。

He called it. He called it.

2013 年，PG 带我和萨姆，还有我的合伙人哈吉·塔格去福基寿司吃晚饭。那时他正在寻找接班人。YC 已经变得如此庞大，如此制度化，对他来说工作量太大了，而且他从来没有打算把 YC 当作他余生唯一的事业。是的。另外，他有年幼的孩子，所以他依次问我们每个人：‘你觉得谁能运营 YC？’哈吉和我……我是说，哈吉和我都来自，怎么说呢，基本上是移民背景。对我来说，我一直是个大器晚成的人。我觉得我现在还在成长的过程中。就像，我还在发现自己到底是谁，以及如何连接……这永远不会结束，因为那其中有某种美好的东西。

In 2013, PG brought me and Sam and my fellow managing partner Harj Taggar out to dinner at Fuki Sushi. At the time, he was looking for a successor. YC had become so big and so like an institution that it was too much work for him and he never really intended on that being like the only thing he ever did for the rest of his life. Yeah. Plus, he had young kids and so he went to each of us in turn and said, 'Who do you think could run YC?' Harj and I... I mean, Harj and I come from, I don't know, basically immigrant backgrounds. And for me, I've just been... I'm a very late bloomer. I feel like I'm still in the process of blooming as it is right now. Like, I'm still discovering who the heck I am and how do I connect... that never ends because there's something beautiful about that.

这是一种很好的状态。确实如此。

That's a great state to be in. That's true.

我希望我能更早做到这一点。这就是问题所在。就像，我会更……如果事情总能更好，那就会更好。但重点是，你知道，哈吉和我看着 PG 说：‘没人能行，老兄。你得永远做下去。’而在那次晚宴上，萨姆·奥特曼说：‘我能做。’然后他真的做了。他接手了，这太棒了。我记得保罗……

I wish I did it earlier. That's the problem. Like, I would be more... things would be better if they could always be better. But the point is, you know, Harj and I looked at PG and said, 'Nobody, man. You got to do this forever.' And in that dinner, Sam Altman said, 'I could do it.' And he did it. He took it over, which was awesome. I remember Paul...

他做了多久？

How long did he do that for?

从 2013 年到 2019 年。然后基本上……那时 OpenAI 已经自成一体，萨姆不得不做出选择。

2013 through 2019. And then he basically... that was when OpenAI took a life of its own and Sam had to choose.

是的。

Yeah.

老实说，事情发生的时候，我必须承认萨姆。我必须说。我们是对手。说实话，有时他对我有点混蛋，但我意识到那只是事情的一部分。那是游戏宏大计划的一部分。你知道，他看准了。我记得 2014 年我决定离开之前为他工作，我说……每次开会他回来都会谈论 AGI。比如，他见了埃隆，他见了拉里和谢尔盖，世界以某种方式向他展开……就像他成为 YC 总裁时那样。我当时不理解。我就想：‘嘿，萨姆，我们只是想在这里投资一些初创公司。那很酷，但我们能谈谈这个 B2B SaaS 的事情吗？’然后我不想这么说，但看看 2014 年的自己，就像，天哪，老兄，你是萨列里，而萨姆是莫扎特。你知道，他在那里，他发现……他听到了，他看到了，他说：‘那是未来，我们要去实现它。’我必须承认。他做到了。这就像我生命中一个根本性的教训。我想我在努力拥抱……有什么东西阻碍了我，让我对显然会发生的事情视而不见，而我就在那里。科技和我的日常生活很大程度上都是即兴的。就像‘是的，而且’。你和创始人坐在一起，他们是很有创造力的人。就像，哇。而不是说‘不，这不会发生’。你知道，我应该说是的。然后就像，哦，是的，那怎么会发生？如果发生了，会发生什么，对吧？

And honestly, like, as it happened, I have to give it to Sam. I have to say it. We were rivals. Honestly, sometimes he was kind of a jerk to me, but I realized that was just a part of the thing. That's part of the grand plan in the game. You know, he called it. I remember working for him in 2014 before I had decided to leave, and I said... every meeting he'd come back and he'd be talking about AGI. Like, he met Elon, he met Larry and Sergey, and the world unfolded to him in a way... like when he became president of YC. I didn't understand it at the time. I'm like, 'Hey Sam, we're just trying to fund some startups over here. That's cool, but can we talk about this B2B SaaS thing?' And then I hate to say it, but I look at the 2014 version of myself and it's like, shoot, man, you were the salieri and Sam was the Mozart. You know, he was there and he found... he heard it and he saw it and he said, 'That's the future and then we're going to go get it.' I have to give it to him. He did it. And it was like one of the fundamental lessons for me in my life. I think I'm trying to embrace... there was something that blocked me, closed me off to what was clearly going to happen and I was there. So much of tech and so much of my life day-to-day is improv. It's like a 'yes and.' You're sitting with founders, very creative people. It's like, whoa. Instead of saying no, it's not going to happen. You know, I should be saying yes. And it's like, oh yeah, how does that happen? And if it happens, what happens, right?

而且想想 2014 年的那个画面就很有趣，我和萨姆以及当时在 YC 工作的其他合伙人坐在 YC 合伙人室里。对萨姆来说，氛围不是‘是的，而且’。氛围是，这家伙为什么要这么做？所以这实际上是让我夜不能寐的事情之一。我今天运营 YC，我想以非常不同的方式运营那个房间。是的。

And it's just funny to think about that 2014 snapshot for me sitting in the YC partner room with Sam and all the other partners who were working at YC at the time under Sam. The vibe was not 'yes and' for Sam. The vibe was like, why is this guy doing this? And so that's actually one of the things that keeps me up. I run YC today and I want to run that room in a very different way. Yeah.

我意识到有时候我就是那个带来‘这到底是怎么回事’能量的人，而人们也在这样想我。

And I realize sometimes I'm the one who's bringing that 'what the heck is going on' energy and people are thinking that about me.

我觉得 YC 非常有趣。

I think YC is very interesting.

听起来你可能不想以他为榜样。

It sounds like you might not want to use him as the example of how to do it.

也许不会。是的。我的意思是，我觉得他学到了很多。我觉得他变成了一个不同的领导者。我觉得他现在承受了很多压力，我真的很同情他和他的家人。另一方面，他不应该承受他现在所遭受的。

Maybe not. Yeah. I mean, I think he's learned a lot. I think he's become a different leader. I think that he's taking a lot of heat right now and I really feel for him and his family. On the other hand, he doesn't deserve what he's getting right now.

从创始人转变为运营 YC 这样的工作难吗？你想念你日常的工作吗？

Was it hard to go from a founder to doing something like running YC? Do you miss the day-to-day of your thing?

哦，是的。我现在可以鱼与熊掌兼得了。

Oh, yeah. I get to have my cake and eat it too at this point.

是的。

Yeah.

因为你们在开源这些东西，开源之后，说实话这变成了下一个吸引人的东西。比如我就是因为 Paul Graham 的文章才找到 Paul Graham 和 YC 的，而 Paul Graham 的文章仍然占了所有申请的大约三分之一。

Because you're open sourcing things, and open sourcing it and then honestly it's turning into the next thing that will bring people in. Like I found Paul Graham and YC because of Paul Graham's essays, and Paul Graham's essays are still about a third of all of the applications that come in.

这些文章是什么时候写的？

When were the essays written?

我是说贯穿他的一生，我想是从 80 年代末或 90 年代初开始的。

I mean throughout his life, starting I think in the late 80s or early 90s.

给我讲讲“做不能规模化的事情”吧。

Tell me about 'Do things that don't scale.'

哦，是的。我觉得很多你听到的建议，我们——我的意思是其中一些就像我们把所有建议都免费送出去。这在 YC 内部已经争论很久了。风投不会这么做。风投喜欢说这是一个，你知道，这是一个机构——

Oh yeah. I think a lot of the things that the advice that you hear that we—I mean some of it is like we give away all the advice for free. That's been a long debated thing inside of YC. VCs don't do this. VCs like to say that this is a you know this is an institution—

封闭的生态系统。

Closed ecosystem.

是的。你必须拿到我们的钱，而我们教你做任何事的唯一方式就是你得通过我们，他们是守门人。而我真的很希望我们不要那样。而且我觉得 YC 从来就不是那样的。你去任何其他孵化器，甚至任何其他做这种工作、帮助创始人的风投，看看他们的课程，都是我们的东西，因为我们在 YouTube 频道上免费提供——那里有 200 万订阅者。我觉得我们想继续这样做。说实话，我们想用知识武装人们，让他们去创造这些东西。现在这有多酷——

Yeah. You got to get our money and the only way we're going to teach you how to do anything is you got to come through us and they're gatekeepers. And then I really want us to not be like that. And I think YC has never really been like that. You go to any other incubator or even any other VC that does this job and helps founders and you look at their curriculum, it's all our stuff cuz we give it away on the YouTube channel—it's 2 million subscribers on there. And I think we want to just keep doing that. Honestly, we want to just like arm people with the knowledge to go and create this stuff. How cool is that right now that—

太棒了。

Amazing.

实际上，现在的工具比以往任何时候都强大得多。

The tools are way more powerful now than ever actually.

如今，生活的大部分都发生在网络上。Squarespace 是你在网络世界中打造梦想形象的基地。使用 Squarespace 一流的模板之一，设计网站变得轻而易举。借助内置的样式工具包，你可以更改字体、图片、边距和菜单，让你的设计完美贴合需求。通过 Fluid Engine（一个高度直观的拖放编辑器）发现坚不可摧的创造力。无需编码或技术经验。通过深入的网站分析工具了解你网站的表现。Squarespace 拥有你在网上取得成功所需的一切。创建博客、通过新闻通讯变现、制作营销作品集、开设在线商店。Squarespace 应用帮助你随时随地管理业务。在旅途中跟踪库存并与客户联系。无论你是刚刚起步还是已经管理着一个成功的品牌，Squarespace 都能让你轻松创建和定制一个漂亮的网站。访问 squarespace.com/tetra，今天就开始吧。

So much of today's life happens on the web. Squarespace is your home base for building your dream presence in an online world. Designing a website is easy using one of Squarespace's best-in-class templates. With the built-in style kit, you can change fonts, imagery, margins, and menus so your design will be perfectly tailored to your needs. Discover unbreakable creativity with Fluid Engine, a highly intuitive drag and drop editor. No coding or technical experience is required. Understand your site's performance with in-depth website analytics tools. Squarespace has everything you need to succeed online. Create a blog, monetize a newsletter, make a marketing portfolio, launch an online store. The Squarespace app helps you run your business from anywhere. Track inventory and connect with customers while you're on the go. Whether you're just starting out or already managing a successful brand, Squarespace makes it easy to create and customize a beautiful website. Visit squarespace.com/tetra and get started today.

你们每学期收到的申请有多相似？是不是有很多相同的想法？

How similar are the pictures that you get each semester? Are there a lot of the same idea?

哦，太多了。然后我们首先要做的就是试着让他们专注于一些对他们来说更独特的东西。

Oh, it's tons. And then the first thing we have to do is try to get them on something that's more unique to them.

那你们会给反馈吗？

And do you give feedback?

哦，是的。是的。是的。不，我的意思是，我回来后过去两三年我们开始做的最重要的事情之一就是我们必须一对一地见人。你知道，我现在团队的四分之一在印度，在新德里。这些人太厉害了，伙计。他们非常聪明。他们就像 22 岁的博士候选人。然后最有意思的事情是我合伙人 Aditya Gupta 刚在 Slack 上发的，我觉得那太迷人了。就像他们现在正在做 2008 年 Web 2.0 的想法。他们就像社交网络和辅导市场，这很有趣，因为在 YC 的历史上，我觉得那是最常见的想法之一。我记得 Dropbox 的创始人，著名的那次是哪一年？他在 MIT，当时在做辅导业务，他遇到了 Paul 和 Jessica，Paul 和 Jessica 说你是做创业公司的合适人选，但你不应该做辅导，伙计，然后他最终想出了 Dropbox。所以很有趣，一如既往，人们仍然在做社交网络、照片分享应用和辅导业务，但他们不应该，他们需要做真正独特的事情。

Oh, yeah. Yeah. Yeah. No, I mean one of the top things that we started doing the last two or three years once I came back is we have to meet people one-on-one. You know, a quarter of my team right now is in India and they're in New Delhi. These people are cracked, man. They are so smart. They're like 22-year-olds PhD candidates. And then the most interesting thing my partner Aditya Gupta just put in Slack that I just like that's fascinating. It's like they're just working on 2008 web 2.0 ideas right now. They're like social networks and tutoring marketplaces which is very funny because in the history of YC I feel like that's one of the most common ideas. I think the founder of Dropbox famously in what year was it? He was at MIT and he was working on a tutoring business and he met Paul and Jessica and Paul and Jessica said you're the right person to do a startup but like you should not work on tutoring man and then he ended up coming up with Dropbox. So it's very funny how same as it ever was like people are still doing social networks and photosharing apps and tutoring businesses but they shouldn't they need to work on things that are truly unique to them.

你们有没有收到一些非常边缘的、与众不同的想法？

Do you get any really fringe ideas that are unlike everything else?

哦，是的。一直都有。最疯狂的一个，真的——那个做得很好，可能真的能成功，有一家太空核聚变公司。

Oh, yeah. All the time. The craziest one I really—that's doing well and might actually pull it off, there's a space fusion company.

那是什么意思？

What does that mean?

字面上，国家实验室里有这些核科学家，他们三十多岁末、四十出头，但他们在国家实验室研究核聚变已经很多年了。他们带着研究中的模拟回来了。由于重力，在地球上进行核聚变非常非常昂贵。但他们有一个模拟和设计，表明如果你在太空，你可以以实际可行的规模进行真正的核聚变。他们基本上——他们必须去了解一下。这有点像坠入爱河。你会想，“哦，天哪，我应该改签航班吗？或者你——这不在我的计划里，伙计。这不是我想做的事，但我得去见一个女孩，对吧？”这就是其中一些创始人的感觉。

Literally, there are these nuclear scientists at the National Laboratories and they're in their late 30s, early 40s, but they'd been working on fusion at the national labs for years. And they came back with these simulations from their research. Fusion on the planet is very very expensive due to gravity. But they have a simulation and a design that shows if you're in space you can do real fusion at a scale that is actually viable. And they're basically—they have to see about it. It's kind of like falling in love. You're like, "Oh man, should I change my flight or you—it's like this is off my plan, man. Like this is not what I was trying to do, but like I got to see about a girl, right?" That's what it feels like for some of these founders.

是的，听起来很棒。

Yeah, that sounds great.

是的。所以，他们出发了。他们筹集了，你知道，我想 400 万或 500 万美元，这些人不是那种在 Google 工作过、上过斯坦福的寻常之辈。他们实际上是最有可能做到这一点的最聪明的科学家。在世界上大多数情况下，比如 100 个宇宙中有 99 个，那些人无法筹集到 500 万美元。但因为 YC 的存在，那些人开始行动了。如果他们成功了，我们将拥有——

Yeah. So, they're off. They raised, you know, I think $4 or 5 million and these are not normal beaten path worked at Google, went to Stanford kind of people. They are actually the smartest possible scientists to do it. And in most cases in the world, like 99 out of a 100 universes, those guys don't get to raise $5 million. But because YC exists, those guys are up and running. And if they succeed, we will have—

太空核聚变，

Fusion in space,

这太疯狂了。我不知道它会不会发生，对吧？他们甚至不知道模拟是否能在现实中实现，但我们必须去看看。

Which is wild. And I don't know if it'll happen, right? They don't even know if the simulations can actually happen in real life, but we have to see about it.

我们得弄清楚。被选中的人是不是更可能是那些提出你从未听过想法的人？

We got to find out. Are the people who get picked more likely to be the ones who come up with the idea that you've not heard before?

想法是那 10 分钟里人的函数。如果我们没有学到任何东西，那很可能就是不行。非常非常聪明的人。你真正想要的人，他们行动力很强，而且他们也很有洞察力。我想他们与源头相连。这可能是最难的事情。就像你看到就知道了，你遇到一个人，就像那个人与某种超越的东西相连——

The idea is a function of the people in those 10 minutes. If we don't learn anything, then it's probably a no. Really really smart people. The people you really want, they're really high agency and then they're also just so perceptive. I guess they're connected to the source. That's probably the hardest thing. It's like you know it when you see it like you meet someone it's like that person is connected to something beyond—

是的。他们能感觉到。这甚至超越了思想。是别的东西。

Yeah. They can feel it. It's even beyond thought. It's some other thing.

是的。所以当你遇到那样的人——我的意思是当我遇到 Brian Armstrong 时，我想，天哪，那家伙身上有点东西。

Yeah. So when you meet people like that—I mean when I met Brian Armstrong I was like oh my god there's something going on with that guy.

通常你会说，人比想法更重要吗？

Typically would you say is it more about the person than the idea?

我觉得人更重要。是的。他们关心什么？你必须有所追求。你内心必须有一种东西。我意识到我确实能在别人身上发现这一点。你只要和人相处，就会觉得，伙计，你内心有某种核反应堆。只是别失去遏制。

I think it's more about the person. Yeah. What are they about? You have to be about something. You have to have a thing in you. And I realized like actually I can spot that in other people. You just spend time with people and it's like, man, you got some sort of nuclear reactor in there. Just don't lose containment.

哦，是的。让我们想办法把它用在好的地方。

Oh yeah. Let's figure out how to use it for good.

没错。

That's right.

那么如果创始人被录取了，接下来会发生什么？

So if a founder gets accepted, what happens next?

他们会搬到旧金山。我觉得旧金山本身就是一段旅程。这里开放得多。从世界其他地方来到旧金山的人，描述的第一件事就是他们不再觉得自己怪异了。

They move to San Francisco. I feel like San Francisco itself is a trip. It's so much more open. The number one thing that people describe when they come from any other place in the world and they come to San Francisco is that they're not weird anymore.

他们不再觉得自己是局外人。

They don't feel like outsiders.

是的。他们是圈内人。就像突然你周围都是那些相信自己真的能戳一下宇宙，然后让它在别处冒出来的人。世界上很少有地方真的会发生这种事。其他地方都有“高罂粟综合征”：别尝试这个。你凭什么觉得自己能做到？

Yeah. They're insiders. Like suddenly you're surrounded by people who believe that they can actually poke the universe and have it pop out someplace else. And there's very few places in the world where that really happens. Everywhere else it's like tall poppy syndrome. Don't try to do this. Who are you to even believe that you could do this?

你觉得是什么让硅谷成为硅谷？

What do you think makes Silicon Valley Silicon Valley?

就是一群认真的极客，做出人们想要的东西。

It's earnest nerds just making a thing that people want.

那这些极客最初是怎么到那里的？

And then how did the nerds end up there in the first place?

我想你可以说基本上都是半导体行业的人。基本上是在二战后，大量研发发生了。休利特和帕卡德从斯坦福实验室起家。就是这样一种长期渗透的过程，世界上最聪明的人聚集在一个特定的地方做一件特定的事。

I guess you could argue there was like basically all the semiconductor people. Basically, it was post World War II. A lot of research and development happened. Hewlett and Packard started out of Stanford labs. There's just sort of this long percolating thing where you had the smartest people in the world collect in a particular place to do a particular thing.

这里面有政府方面的因素吗？

Is there a government aspect of it or no?

哦，当然。我的意思是，互联网全是国防研究。互联网之所以存在，是因为 DARPA 拨款用于实验比电话更具弹性的通信系统，以防核战争。所以如果你需要那个，你就需要像 TCP/IP 这样的东西，它是今天运行互联网的基本协议之一。所以这就是为什么每所大学都有 Unix 主机。然后他们会建立这些链接，不是普通的打电话或传真机之类的。那真的是你必须设计什么？你必须设计电子邮件。所以有着悠久的历史，国防和基础研究资金投入技术，然后商业化，成为每个人都在用的东西。

Oh, definitely. I mean, the internet was all defense research. The internet exists because DARPA gave a grant to experiment with communication systems that are more resilient than telephone if there's nuclear war. So if you needed that, then you needed something like TCP/IP, which is one of the basic protocols that runs the internet today. And so that's why you had these Unix mainframes at every university. And then they would make these links that were not, you know, normal pick up the phone and call them or fax machine and things like that. It was literally what would you have to design? You'd have to design email. And so there is a deep long history of defense and primary research dollars going into fund technology that then gets commercialized and then it becomes the thing that everyone uses.

如果世界上其他地方要出现第二个硅谷，你觉得会在哪里？

If there was going to be a second Silicon Valley somewhere else in the world, where do you think that would be?

我觉得现在是纽约市。我们告诉每个人待在旧金山的原因是，如果搬到旧金山湾区，创建价值十亿美元或以上公司的几率和可能性会高出 2.5 倍，而在纽约市则高出约 2 倍。

I think New York City right now. The reason why we tell everyone to be in San Francisco is the rate of and likelihood of someone creating a company worth a billion dollars or more is 2.5x higher if they move to the San Francisco Bay area and then it's about 2x higher if they're in New York City.

给我讲讲创始人的性格。你看到他们身上有什么共同点吗？

Tell me about the personality of the founders. Is there anything that you see that runs through many of them?

我最喜欢看到的两个特质，一个是认真，这是一个非常不寻常的特质。这尤其来自 Jessica Livingston。有一种观点认为最好的创始人是那种“ hustler”。但我要说，真正成功的 90% 的人并不是那样。他们实际上对自己想做的事非常认真。这与愤世嫉俗或卖蛇油的人形成对比。我们总是在打一场仗，总有一些创始人做得太过火。总有一些创始人做了不该做的事。我觉得最好笑的说法是，真正优秀的人知道能走多远，不会更远。我们资助了太多创始人，那些没成功的，很多时候是因为他们走得太远了。

I mean the two traits that I really love to see, one is earnestness, which is very unusual as a trait. This definitely comes from Jessica Livingston in particular. There's this idea that the best founders are sort of hustlers. And I would say 90% of the people that really really make it are not that. They're actually very earnest about what they're trying to do. And this is in contrast to like cynical or like a snake oil salesman. We're always sort of fighting this war where there are always founders who sort of take it too far. There are always founders who do things that they shouldn't do. I guess the funniest thing to say would be the really really good ones know how far to take it and not farther. And we just fund so many founders period that the ones that don't make it, a lot of times they don't make it because they went too far.

而且我还在想怎么处理这件事，因为如果你管得太严，你会失去所有怪才。但另一方面，显然你需要某种后果。所以我们正在想办法。YC 一直在增长。我们似乎每年增长 10% 或 20%。

And I'm still trying to figure out what to do about that actually because if you come down too hard, you're going to lose all the weirdos. But on the other hand, like clearly you need to have some sort of consequence. And so we're trying to figure it out. YC is always growing. Like we seem to be growing like 10 or 20%.

太过火，我的意思是直接做非法或不道德的事。难点在于锦标赛，每个市场都那么拥挤。我觉得它们都重新开放了。所以，在创业公司中有这样一个有趣的时期，回到浪潮，每当有新浪潮，我们都认为这是蓝海。你可以做任何事，真的是任何事，都会很棒。比如如果你在 2004 年或 2005 年做了一个社交网络，而且你做得很好，你可能已经干掉 Facebook 了。任何那些事都可能发生。然后随着时间推移，通常五年或十年后，公司筹集了数百亿到数十亿美元。收入上升。从蓝海变成红海。基本上，当没人知道哪头朝上，没有竞争时，就是蓝海。然后任何可能变得非常大的东西最终都会变成红海。在过去 15 年里，这变得越来越真实。然后 AI 基本上打开了它，一切又变蓝了。

Too far. I mean, just straight up doing things that are illegal or unethical. The hard part here is like the tournaments, every market is so crowded. I think they're all open again. So, there's this sort of interesting time in startups where going back to the waves, you have anytime there's a new wave, we have this idea that it's a blue ocean. You could just do anything, literally anything, and it'd be awesome. Like if you did a social network in 2004 or 2005 and you were good, you could have killed Facebook. Like any of those things could have happened. And then as time goes on, usually like five or 10 years in, the companies raise hundreds to billions of dollars. The revenue goes up. It goes from a blue ocean to red ocean. Basically, it's a blue ocean when nobody knows which end is up and there's no competition. And then anything that could be really really big ends up becoming a red ocean. And that became more and more true over the last 15 years. And then AI has basically opened it up and everything's blue again.

看起来可能会持续很长时间。

It seems like it might be for a long time.

不，我们等着瞧。我的意思是，我很难看到明年以后。我觉得这在很大程度上取决于模型会变得多聪明，以及以什么速度。我不想成为希望技术放慢速度的人，

No, we'll see. I mean, I hard to see even beyond next year. I think a lot of it depends on how much smarter the models get and at what rate. I don't like to be someone who wants technology to slow down,

但现在以目前的速度，我感觉很好。它进展得相当快，但还没有快到让人震惊，以至于创业公司无法存在或商业无法存在。事实上，它比大多数 AI 研究人员认为应该的速度要慢。我和 Bob McGrew 坐在一起，他是 OpenAI 的首席研究官，我的老朋友，当他们提出 01 和推理模型时，他说的第一件事就是他们把它发布给世界，但好像什么都没发生，他们想哦好吧，我们构建了它，他们应该现在就来，我说不不不，那发生在大约一年半前，我们现在才真正把它带到商业世界，现在才真正产生影响。

but right now at the current rate, I feel good. It's like it's progressing at a pretty fast clip, but it's not like mind-blowingly so fast that, you know, startups can't exist or business can't exist. In fact, it's progressing slower than most AI researchers would say it should. I was sitting with Bob McGrew who is chief research officer at OpenAI, old friend of mine, and one of the first things he said when they came up with 01 and the reasoning models was like they released it to the world and it's like nothing happened and they thought oh well we built it like they should all come right now and I was like no no no like it that that happened like maybe a year year and a half ago we're only now actually bringing it to the business world like only now is it actually having an impact

所以，我的意思是，这是给创始人的一个教训。即使你是对的，即使你创造了十亿人会用的东西，它会彻底改变他们的生活。就像，你真的发明了轮子。你可以发明轮子，但你仍然得把它塞进人们的喉咙，伙计。他们不会心甘情愿地来。

So, I mean, that's one of the lessons for founders. Even if you can be right, even if you created the thing that a billion people is, they're going to use it and it's going to change their lives completely. Like, you literally invented the wheel. You could invent the wheel and you still got to jam it down people's throats, man. It's like they're not going to come willingly.

人们没有轮子的时候不知道自己想要轮子。

People don't know they want the wheel if they don't have the wheel.

完全正确。

Exactly.

是的。

Yeah.

所以，这就是为什么说，哦，这其实是好消息。就像有人发明了轮子，你知道，有人发明了个人电脑，

So, that's where it's like, oh, this is good news, actually. Like someone invented the wheel, you know, someone invented the personal computer,

但然后

but then

现在怎么办？

now what?

是的，你可以去构建它。你可以把它带给人们，成为那场宏大戏剧的一部分。世界是一个舞台，在这一点上，你知道，整个世界都在被“vibecoded”。

Yeah, you get to build it. You get to bring it to people and you get to be a part of that grand play. All the world is a stage and at this point, you know, all the world is just being vibecoded.

你觉得大多数创始人是在解决个人问题，还是在解决普遍问题？

Would you say most of the founders are building something personal or building something for a general problem?

总是从个人问题开始。这就是所谓的“做不可规模化的事情”，对吧？

It always starts personal. That's sort of the do things that don't scale, right?

嗯。

Yeah.

基本上，很多建议都来自于我们看着创业者一次又一次地失败，然后我们把笔记写下来。就像，“哦，对，对，别那么做。” YC 之所以能这么有帮助，是因为创业公司大约有一千种死法，我们全都见过，而且可以把它们画成均值分布图。我们就能帮人们避开这些坑。还有保罗·格雷厄姆那句经典的话，现在越来越应验了：只要你不死，你就能成功。可能花五年，可能花十年，你知道，只要你一直在思考，做正确的动作，你总会在对的时间出现在对的地方。

Basically, a lot of the advice comes from we just see people fail over and over again and then we just sort of write down the notes. It's like, oh yeah, yeah, just don't do that. Like the reason why YC can be so helpful is there's about a thousand ways that startups fail and we've seen all of them and we can graph them on a mean distribution. We can just keep people from doing that. And then there's this classic line from Paul Graham which has become more and more true which is as long as you don't die you will succeed. Like it might take five years it might take 10 years you know like if you are just in there thinking about things if you do good reps you're going to be in the right place at the right time.

只是时间问题。

It's just a matter of time.

对。

Yeah.

13 周结束时会发生什么？

What happens at the end of the 13 weeks?

那时候你就要出去向世界展示你的成果，看看投资人是否愿意上钩。我们每年大约有 15 亿美元的投资人过来，他们平均投资额从 100 万美元到 400 万、500 万美元不等，最高端能到 2000 万美元投给某一家公司。我觉得自从我回到 YC 以来，中位数翻了一番。以前在演示日平均融资大约 100 万美元，现在每个批次接近 220 万美元。

So that's sort of when you go off and show the world what you got and you see whether the investors want to bite. And we have about a billion half dollars of investors per year who come and they're investing on average I mean anywhere from a million half dollars to four 5 million on the high end like $20 million into any given company. I think the median has doubled since I came back to YC. It used to be about a million dollars raised at demo day. Now it's closer to 2.2 million per batch.

演讲能力有多重要？

How important is the ability to pitch?

哦，超级重要。但最好笑的是，我们过去花了太多时间在这上面。我的意思是，这是其中一个微妙有趣的点。我觉得 AI 变得太强了。我们有这些脚本，真的能指导人。我是说，那是 SL 的办公时间，但对于其他一切，这其实是一个非常有趣的模式，我认为它会在整个社会发生。我们总是想着代码，现在世界上所有企业都在想“哦，我需要构建软件”，那是确定性的 0 和 1，但没人意识到，不仅仅是代码，还有技能。所以现在最新的事情是，那些曾经不可言传的东西，那些我必须坐下来教别人的东西。现在说出来有点吓人，但我认为它正在变成现实：我真的可以把我知道的、我们从成千上万的创始人那里学到的东西提炼出来。我们做到这一点的能力现在强了一千倍，因为我们真的可以用 CircleBack 或 Granola 来记录我们的办公时间，短期内我们只是战术性地使用。所以就像“哦，对，这是我们谈过的，这是我们应该做的。”

Oh, it's super important. And then the funniest thing now though is we used to spend so much time on that part. I mean, this is one of the subtle interesting things. I think that AI is becoming so good. We have these scripts that coaches people up literally. I mean that's SL office hours but for everything else this is actually a very interesting model that I think is going to happen across all of society where we always think about code all the businesses in the world right now think about like oh I need to build software and that's like deterministic ones and zeros and what no one has figured out yet is that like it's not just the code it's also the skills. So the newest thing that's happening now is that the thing that used to be ineffable, the thing that I had to sit with someone and teach them this stuff. Now that I say it, it sounds kind of scary and crazy, but I think it's becoming true that I can actually distill down what I know and what we've learned from like thousands and thousands of founders. Our ability to do it is a thousand times better now because we actually can literally use CircleBack or Granola and record our office hours and we're doing it in the short term just tactically. So it's like oh yeah this is what we talked about and this is what we should do.

解释一下 Granola 是什么。

Explain what granola is.

对。CircleBack 或 Granola 这类会议记录智能体。市面上有很多。你甚至可以在 ChatGPT 上点击麦克风按钮，它就会开始录音并听你的对话，然后总结你们聊了什么。但真正疯狂的是，你可以把这些记录加上你之前所有的记录，然后进行“分集”。意思是，让一个 LLM 逐行处理，不是输出全文，而是每 10 到 20 分钟总结一次，说“他们聊了这个，这个很有趣”。然后你可以再让另一个 LLM 来总结所有内容，提炼成原则，就像你会写进书里的那些东西。智能已经在那里了，它能够做到。

Yeah. Circle back or granola are these meeting recording agents. I mean there's a bunch out there. You can even go on chat GPT apparently and click the microphone button and it'll just start recording and listening to your conversation and it'll summarize what you talked about. But the really crazy thing is uh you can take that and you can take that plus all the other ones you did and you can diorize it. Meaning like you have an LLM go line by line and say instead of the full text it's like for every 10 or 20 minutes it's like this is what they talked about and this is what was interesting. And then you can take that and have another thing, another LLM come and summarize all of it and then distill it into principles and like the things that you would put in a book. The intelligence is already there and it's able to do that.

你觉得总结能力有没有问题？想想读一部伟大的文学作品和读摘要之间的区别。

Do you think there's an issue with the ability to summarize where think of the difference between reading a great piece of literature or the cliff notes?

它们是两回事。

They're two different things.

哦，肯定有区别。是的。在我们现在所处的“总结经济”中，我们会失去一些东西吗？

Oh, there's definitely a difference. Yes. Do we lose something in the summarization economy that we now live in?

我认为我们希望 YC 合伙人获得的体验，实际上应该是那种“我不再感到日复一日重复”的感觉。这开始在 YC 出现了。以前我必须参与每一次对话，教他们基础的东西，1、2、3、4。现在反而更好了，因为如果我们有我们的“Bookface”智能体，我们自己的 AI 智能体，它开始教人们这些更基础的东西。那么当我见到他们时，我实际上知道他们处于什么水平，然后我们讨论的是所谓的“分布之外”的东西。我们总是从 AI 的角度思考，什么是模型里已有的，什么是互联网上已有的，对，对比什么是前所未有的新东西。然后所有的阿尔法、所有有趣的事情、让你感到活着的事情、让你与源头连接的事情，都不在于“演奏热门曲目”。

I think the experience that we're hoping for with YC partners is actually the feeling that you should get is that like I don't get Groundhog Day anymore. And that's starting to be what we see at YC. So before I had to be in every single conversation per se and like teach them the basics 1 2 3 4. Now instead it's better because if I have our bookface agent, we have like our own AI agent that helps is starting to teach people these more basic things. Then when I see them, I actually know where someone is and then we're talking about stuff that's quote unquote out of distribution. We're always thinking AI in terms of like what's already in the model and what's already on the internet. Yes. versus what is new that has never been before. And then all the alpha and all the things that are interesting, the things that make you alive, the things that connect you to the source. It's not in like playing the hits.

嗯，

Yeah,

我可以整天演奏热门曲目。就像去年，我每天都是那样。每个合伙人每天都是，“哦，对，这是两句话的推介。这就是全部内容。我们可以把所有时间花在新新事物上，比如”

I can play the hits all day. And like last year, that's what my day-to-day was like. That's what every partner was like every single day was like, "Oh yeah, here's the two sentence pitch. This is what all covered. we can just spend all our time on like the new new and like

以及每个案例的具体细节。

and the specifics for each case.

对。所以这真的很重要。这就是人们应该感到兴奋的事情。当我们谈论如何让社会做更多有趣的事情时，我认为我们可以具体化。就像在你的身体里，你感到更兴奋去上班吗？为什么？对我来说，答案就是当我能和新的人谈论新的事情，而不是和同样的人谈论同样的事情。所以 AI 对于超级有创造力和高度智能体式的人来说是超级好消息，如果你有自主性，你可以把你的每一天设计成那样。第二件我认为非常重要的事情是我们在 YC 学到的：你必须写提示词。我们几乎对自己做了这件事，我们用 AI 来挑选人才，我们多年来一直这样做，但旧的 AI 并不聪明。我们称之为基于特征的。它实际上只是矩阵数学。那是在 Transformer 之前，在所有疯狂的事情发生之前。那东西太笨了，需要大量的数据标注。然后我们发现，15 位合伙人开始成为 API 之下的数据标注员。有一个非常疯狂的想法，是维尼特斯·拉奥提出的。他是互联网思想家，谈到整个世界正在被重构成像 Uber 或 DoorDash 那样的 API 线。有了 AI，危险在于组织中的每个人要么在 API 线之上，要么在它之下。你想在它之上。

Yeah. So that's really important. And so that's like sort of the thing that people should be really excited about. When we talk about how do we let society do more interesting things? I think we can embody it. It's like in your body do you feel more excited to come to work and why? And for me that answer is like when I get to talk to new people about new things instead of talking about the same people to the same things right and so that's where AI is super good news for people who are super creative and highly agentic like if you have agency you can design your day to be exactly like that. And the second thing that I think is really important that we learned at YC is you have to write the prompts. So we almost did this to ourselves where we use AI to like pick people and we always we have for like many many years but the old AI and then the old AI was not smart. It was we call it feature-based. It was literally just matrix math. It was before the transformer. It was before all of the crazy stuff that happened. And that stuff was so dumb that it required like an incredible amount of data labeling. And then what we found was the 15 partners started becoming data labelers under the API. There's this very crazy idea that kind of Venitates Rao talks about. He's this internet thinker that talked about all of the world is being recreated into these like Uber or Door Dash sort of API lines. With AI, the danger is like everyone in the organization is either like above the API line or below it. You want to be above it.

然后对我来说的理想是，组织里尽可能多的人都在 API 线之上。具体来说，这其实来自我的合伙人 Pete Kumman，他意识到我们需要让每个人都能拥有并修改自己的提示词。比如你用 Google 的时候，他们在你的 Gmail 里有 AI 产品，但它有点蠢，并不能真正按你的意愿行事。那是因为某个 Google 工程师和产品经理，他们没有经历过你的生活，不了解你和你关心的事情，而且他们不让你修改它，对吧？

And then the ideal to me is as many people in the org as possible are above the API line. And what that concretely means is actually from my partner Pete Kumman who figured out actually we need to allow every person to own their own and change their own prompt. So when you use Google for instance and they have their AI product in your Gmail, it's kind of dumb and doesn't really do what you want. Well, that's because some Google engineer and PM who didn't live your life who doesn't know anything about you and the things you care about and they don't let you change it, right?

对。

Right.

所以这就像一个基本的产品问题。当你使用它时，这就是闭源的问题。这是企业 AI 的问题，所有东西都被隐藏起来，而这些东西如此强大。这就是为什么 OpenClaw 让我震惊，它太强大了，显然就是未来。

And so that's like a basic product problem. Like when you use this, this is the problem with closed source. This is a problem with corporate AI, is that all of it is hidden away and this stuff is so powerful. This is why OpenClaw is blowing my mind and it's so powerful. It's so obviously the future.

如果你在开源领域做出了很棒的东西，有没有可能随着时间推移它发展成一项业务？

If you build something great in open source, isn't it possible over time that it grows into a business?

哦，是的。YC 已经资助了一些最大的开源成功案例。比如 Supabase，我一直在用，它最初是一个开源项目，然后通常你通过托管赚钱，我认为他们会成为最大的之一，个人觉得有一天他们可能比 AWS 还大。

Oh, yeah. YC has funded some of the biggest open-source wins. Supabase for instance, I use it all the time and that started as an open source project and then usually you make money on hosting and they're going to be one of the biggest I think someday they might be bigger than AWS personally.

哇。他们和 AWS 做的是同一行。

Wow. They were in the same business as AWS.

是的。他们帮助人们存储数据。

Yeah. They help people store their data.

酷。

Cool.

所以现在他们即将爆发式增长，因为突然每个人都会拥有自己的个人 Gbrain 或个人智能体，而那个智能体必须把数据存储在某个地方，很可能就是 Supabase。而它最初是一个开源项目，并不一定有商业化的意图，结果却变成了这么庞大的东西。但也不一定非得这样。开源领域的很多人就像 Linux 的 Linus Torvalds，那也很棒。那家伙简直能水上行走，我觉得地球上的每个人都应该为他的创造而深深感谢 Linus Torvalds。

And so now they're about to go hyperbolic because suddenly everyone and their mama is going to have their own personal Gbrain or they're going to have their own personal agent and that agent has to store its data someplace and it's probably going to be Supabase. And that started as an open-source project that did not necessarily have any intention of being commercial and then it just turned out to be this huge thing. And it doesn't have to be like that. A lot of people in open source are like Linus Torvalds from Linux and that's also awesome. Like that guy walks on water and I think every person on the planet owes Linus Torvalds an incredible thank you for what he created.

社交媒体是如何改变世界的？

How did social media change the world?

我认为社交媒体最悲观的观点是它某种程度上在摧毁民主，或者以不好的方式改变人们的思维方式。但我不相信这一点。我真的认为人们以前就相信你在社交媒体上看到的那些东西，只是你以前永远不会看到而已。但结果是，你必须在身边建立更多的认知盔甲，以及你允许进入的东西。我现在正试图把这教给我六岁和十岁的孩子。

I think the black pill on social media is that it's somehow destroying democracy or changing the way in a bad way, like how people are thinking. But I don't believe that. I really think that people believe all the things that you see on social media before. It's just that you would never see it before. The result though is that you have to have a lot more cognitive armor around yourself and what you allow in. And I'm trying to teach that to my six and 10 year old right now.

有效果吗？

Is it working?

我不知道。希望如此。他们真的很沉迷 YouTube。我为此感到很难过，但我相信很多父母都在为此挣扎。

I don't know. I hope so. They're really addicted to YouTube. I feel terrible that that's true, but I'm sure a lot of parents are struggling with that.

不过 YouTube 上有很多很棒的内容。我的意思是，你可以在 YouTube 上学到任何东西。

There's so much great stuff on YouTube, though. I mean, you can learn anything on YouTube.

所以，这就是我试图教他们关于社交媒体的东西。我觉得他们开始明白了。我十岁的孩子，我六岁的孩子还不明白。我十岁的孩子，我觉得开始明白了，就是，看，世界上每个人都需要自己解决金钱问题。有时候他们不得不去做 YouTube 内容，有时候他们做的方式是给你信息、帮助你、教你一些东西，而且有趣好玩。所以那些东西很酷。看那些节目。但还有另外一些人，他们试图解决自己的金钱问题，但方式却是伤害你、对你撒谎，或者只是说你想听的话，但实际上并不是能帮助你的东西。而社交媒体的关键就在于你必须能分辨这两者的区别。如果我能打个响指教会全世界孩子们一件事，那可能就是，你知道，如何分辨什么是对你真正有益的东西，什么不是。

And so, that's what I'm trying to teach them around social media. And I think they're starting to get it. My 10-year-old, my six-year-old does not get it yet. My 10-year-old, I think, is starting to get it that like, look, everyone in the world has to solve the money problem for themselves. And sometimes they have to go and make this YouTube content and sometimes they are doing it in a way that informs you and helps you and teaches you something and is fun and interesting. And so those things are cool. Watch those shows. But then there are all these other people who they're trying to solve their money problem but like in a way that hurts you or is lying to you or is just telling you what you want to hear but is not actually, you know, something that's going to help you. And the defining thing about social media is that you have to be able to know the difference between those two things. And if I could snap my fingers and teach the children of the entire world one thing, it'd probably be that, you know, like how do you tell the difference between something that is genuinely good for you and something that isn't?

我在想你是否能让你的 AI 智能体来组织这些，相应地组织 YouTube。

I wonder if you could have your AI agent organize that, organize YouTube accordingly.

我会把它加到我的清单里

I will add that to my list of

好主意，对吧？如果你真做了，请发给我，因为我也想看看。

a good one, right? If you do do that, please send it to me cuz I'd like to watch that as well.

这其实也是我们在 Y Combinator 一直试图资助的一个创业项目。所以，我们会资助它。

It's actually been a startup that we've been trying to fund at Y Combinator, too. So, we'll fund it.

是的，社交媒体太强大了，而且很可怕。再说一次，它就像终极的奶酪搬运工。

Yeah, social media is so powerful and it's scary. Again, it's like the ultimate cheese mover.

它对你的生活有什么影响？

How did it impact your life?

我的意思是，现在我在 X 上有大约 75 万粉丝，感觉就像一把巨大的摆动锤，偶尔会非常非常糟糕，但大多数时候相当棒。我非常感激创业圈让我基本上能推广所有我真正关心和相信的东西。我觉得这其实很不寻常，但我想这是一种契约。就像我推广的任何东西都不只是为了推广而推广。实际上我是在努力让自己达到我想教给孩子们的标准，那就是，好吧，我发的每一条内容，

I mean, now I have like 750,000 followers on X and it feels like this giant swinging hammer and then once in a while it's really, really terrible, but most of the time it's quite awesome. I feel really thankful that the startup world has let me basically shill all the things that I really care about and believe in. And I think that's unusual actually, but I guess it's sort of a covenant. Like anything that I shill is not just shill for shill stake. It's actually I'm trying to hold myself to that standard that I'm trying to teach my kids to, which is like, okay, everything I post,

你想让人们喜欢你喜欢的东西。

you want to turn people on stuff that you like.

是的。

Yeah.

改善你的生活。

Improves your life.

如果它改善了我的生活，我希望他们也能拥有。

If it improves my life, I want them to have it.

AI 已经如何改变了世界？

How has AI already changed the world?

我认为现在它才走了改变世界的 1%。我的意思是，有十亿人在使用它。所以，那相当疯狂。另一方面，也许那十亿人中有 1% 的人真正接触过我去年 11 月谈到的那种新型 AI，主要是因为太贵了。但好消息是它会变得更便宜。你知道，和 AI 研究人员待在一起，最让我印象深刻的是未来已经在多大程度上到来了。只是还没有均匀分布。你知道，威廉·吉布森在《神经漫游者》里说过这句话，我们现在就生活在这样的时代。那些现在每月花费 200 美元的模型比 99% 的人类都聪明。而我认为那十亿人中不到 1% 的人真正接触过或体验过它。我觉得这是我们可能需要努力的事情。我现在意识到，当我 16 岁或 14 岁的版本今天需要访问那个模型时，

I think it's 1% of the way into changing the world right now. I mean, you have a billion people using it. So, that's pretty wild. On the other hand, maybe 1% of those people have actually really touched the new type of AI that I was talking about in November of last year, mainly because it's so expensive. But the good news is it's going to get cheaper. You know, spending time with AI researchers, the thing that really jumps out at me is to what degree the future is already here. It's just not evenly distributed yet. You know, William Gibson said that from Neuromancer, that's where we're living in right now. The models that are, you know, cost $200 a month right now are smarter than 99% of human beings. And I think less than 1% of that billion people have really touched it or experienced it. I think that that's something we probably need to work on. I realize now when the 16 or 14-year-old version of me needs to have access to that model today,

你知道，那种 vibe coding 和我一直在做的所有开源，实际上确实花费很多。我记得 Mark Andre 和 Elon Musk 都发过关于这个的推文，像那些走在最前沿的人，比如我的朋友，你知道 OpenClaw 的创造者 Peter Steinberger，他之所以能做到，是因为他在职业生涯中非常成功，可以每年在 token 上花一百万美元而毫无感觉，而且没问题。但世界上有多少人真的能做到？真的不多。

You know, the sort of vibe coding and the all the open source that I've been doing, it actually does cost a lot. I think there was a tweet from both Mark Andre and Elon Musk about this like the people who are way out on the edge like my friends and you know Peter Steinberger creator of OpenClaw the reason why he was able to do that was he was so successful in his career that he could just spend a million dollars a year on tokens and not feel it and it'd be okay. But like how many people in the world could really do that? Like really not very many.

现在有没有更便宜的开源方式来做这件事？

Is there not an open-source way to do it now that would be cheaper?

已经接近了，但说实话，仍然很贵。你知道，还是很贵。

It's almost there, but honestly, it's still expensive. It's still expensive, you know.

那我们是说还要几年还是几个月？

So, are we talking about years away or months away?

我觉得还要几年。比如现在你每年花一百万美元在 token 上，就能做相当于一百个人的工作。所以，这通常花你一百万美元，却能带来数亿美元甚至数十亿美元的软件价值，对吧？现在地球上大概有一千个人在这么做。而 AI 研究人员带来的好消息是，这些更小、更便宜的模型其实可以更聪明。我认为这就是开源发挥作用的地方。事情会这样发展。

I think it's still years away. Like what you can do today spending a million dollars a year on tokens. At that point, you're talking about doing the work of like a hundred people. So, this would normally cost you like a million dollars is a multiplier to like hundreds of millions of dollars in software value, right? Or billions even. And that's happening for I don't know a thousand people on the planet right now. And then the good news from the AI researchers is that it turns out like these much smaller cheaper models they can be a lot smarter. And I think that's where open source comes into play. That's how that's going to happen.

是的。现在有一种棘手的相互作用，很多美国律师想说我们应该禁止中国的模型。而尴尬的是，这恰恰会助长另一种情况，即全世界只有一个玩家能接触到最好的模型，那将是相当灾难性的。

Yeah. There's this difficult interplay right now where a lot of lawyers in America want to say we should ban Chinese models. And then the awkward thing about that is that will actually play exactly into this other like there's only one player in the entire world that has access to the best models in the world and that would be quite disastrous.

有一个悠久的传统，就是缓慢地阅读神圣文本，让每个词沉淀、回响，并随时间揭示意义。与其急于下结论，这种实践邀请我们反思、倾听和专注。几个世纪以来，这种重复被用来贴近智慧。不是把文字当作信息来研究，而是把它们当作一次又一次被体验和经历的东西来接受。这个传统被称为 lectio deina。它起源于早期修道院生活，通过四个温和的步骤来接触经文：阅读、反思、重复和安息。读一小段经文，握住一个短语。沉默成为实践的一部分，为洞察自然浮现创造空间。Lectio 365 将这种古老的节奏带入当下。专为现代生活设计，它全天提供简短的引导式经文反思，以意图开始，在中午暂停，在夜晚放松。这些阅读与其说是收集信息，不如说是反复回归智慧。让熟悉的文字遇见新的时刻。这是一种让圣经智慧贴近日常生活、安静存在、稳定而鲜活的方式。Lectio 365 是免费资源。现在和永远在那里寻找灵感。了解更多请访问 lectio365.com

There's a long tradition of reading sacred text slowly, allowing each word to settle, echo, and reveal meaning over time. Rather than rushing to conclusions, this practice invites reflection, listening, and attention. For centuries, this repetition has been used to stay close to wisdom. Not by studying words as information, but by receiving them as something lived and experienced over and over again. This tradition is known as lectio deina. Emerging in early monastic life, it engages scripture through four gentle movements. Reading, reflection, repetition, and rest. A short passage is read, a single phrase is held. Silence becomes part of the practice, creating space for insight to surface naturally. Lectio 365 brings that ancient rhythm into the present moment. Designed for modern life, it offers brief guided scriptural reflections throughout the day to begin with intention, pause at midday, and wind down at night. The readings are less about information gathering and more for returning to wisdom again and again. Letting familiar words meet new moments. A modern way to keep biblical wisdom close, quietly present, steady, and alive within everyday life. Lectio 365 is a free resource. Find inspiration there now and always. Learn more at lectio365.com

对于一些 YC 的成功案例，有些想法在提出时听起来有多离谱？

For some of the YC success stories. How far out did some of the ideas sound when pitched?

我的意思是，比特币，我认为 Coinbase 显然是。DoorDash 看起来有点疯狂。Airbnb 可能是最著名的。Brian Chesky 现在在我的董事会里。当他来参加 YC 面试时，Paul Graham 说：“什么？人们会住在彼此的家里？谁会这么做？谁会这么做？”

I mean, Bitcoin, I think Coinbase clearly was. I mean, DoorDash seemed kind of crazy. Airbnb was probably the most famous one. Brian Chesky's on my board now. When he came to do his YC interview, Paul Graham said like, "What? People are going to stay in each other's homes? Who does this? Who would do this?"

是啊。

Yeah.

我们不再那么想了。我们只会说：“哦，那是 Airbnb。”关于 Brian Chesky，我最喜欢的故事是，当他们刚开始时，它不叫 Airbnb，而叫 Airbed and Breakfast。它确实是一个气垫床，而且他们必须提供早餐。在它开始起飞的时候，有人进来说：“嗯，我没有气垫床，但我可以用真正的床吗？”他说：“嗯，我们是气垫床和早餐。你买个气垫床放在床上怎么样？”

And we don't think about that so much anymore. We just say, "Oh, that's an Airbnb." My favorite story about Brian Chesky is when they started it, it wasn't called Airbnb. It was called Airbed and Breakfast. And it was literally an airbed and they had to provide breakfast. And at some point when it was starting to take off, someone came in and said, "Well, I don't have an air bed, but could I do a real bed?" And he said, "Well, we are airbed and breakfast. What if you buy an air bed and put it on top of the bed?"

到目前为止，

So far,

然后某个时候，他说：“哦，不不不。是的，就这样，去吧。就这么做。”你知道，结果那变成了大得多的市场。

and then at some point was like, "Oh, no, no, no. Yeah, it's just Yeah, go for it. Just do it." And, you know, that turned out to be the much bigger market.

你有没有对 YC 创始人能取得的成就感到惊讶？

Are you ever surprised by what the YC founders can accomplish?

哦，一直如此。太疯狂了。我的意思是，这种事一直在发生。

Oh, all the time. It's wild. I mean, it just happens all the time.

那一定是一种很棒的感觉。

It must be a great feeling.

我的意思是，没有哪种感觉能比得上和某人坐下来，他们说：“我有这个愿景。我相信世界应该这样运作。”然后你低头一看，它真的发生了。

I mean, there's no feeling like it to sit down with someone and they're like, I have this vision. I have a belief in how the world should work. And then you look down and it happens.

你接手后做了什么改变吗？

Did you change anything when you took over?

最大的改变是我把它变回了不是 VC，而是真正专注于最初的早期创业创始人项目。所以当我进来时，YC 有点像 Google 变成了 Alphabet。当你经历 Google Alphabet 时刻时，就像，哦，是的，我们在做八件不同的事情，我们有八个不同的 CEO 在各自负责不同的事情，而你有一个非常成功的东西，那是它的引擎。

The biggest change was that I turned it back into not a VC but actually focused on the initial early startup founder program. So when I came in YC had sort of gone from as if Google became Alphabet. When you do a Google Alphabet moment, it's like, oh yeah, we're doing like eight different things and we have eight different CEOs all working on different things and you had something that was very successful that was the engine of it.

我不得不回来说，我们实际上要专注于让 YC 变得伟大的那一件事

And I just had to come back and say we're actually going to focus on the one thing that made YC great

那将是

and that's going to be the

而且没有其他人真正在做。还有其他 VC。有人能做你的 A 轮或 B 轮。你只需要专注。专注就是说“不”。

and no one else is doing really. There are other VCs. There are people who can do your series A or B. And you just got to focus. Focus is saying no.

那很难。

and that was hard.

你觉得它会如何随时间演变？

How do you see it evolving over time?

现在，我只是想看清未来一两年。我想起互联网上的那个梗，全世界都在跳舞，然后角落里有一个孤独的人，独自一人，就像 AI 正在改变一切。然后每个人都在跳舞。每个人都在做自己的事。就像他们不知道。我觉得这个梗就叫“他们不知道”。现在我们只是在想如何改变自己。我们总是构建软件让 YC 变得很棒。当 Paul Graham 刚开始接受申请时，他著名地做了自己的软件。你会上网站，他做了自己的网络应用，让他们能审查数千份申请。现在我们正处于下一个时刻，回到这个 AI 导师，它会教你如何 vibe code。如果它还能帮助我们找出并识别那些应该获得百万美元 token 资助的人呢？

Right now, I'm just trying to see through the next year or two. I think of that meme on the internet where everyone in the world is dancing and then there's this one lonely person in the corner and they're alone and it's like AI is changing everything. And then everyone's dancing. Everyone's just doing their thing. Like they don't know. I think it's called literally the they don't know meme. Right now we're just trying to figure out how do we change ourselves. We always built software to make YC awesome. That when Paul Graham first started taking applications, he famously made his own software. You would go on the website and he made his own web app that would allow them to review thousands of applications. And now we're sort of in that next moment going back to this AI tutor that will teach you how to vibe code. What if it also helped us figure out and identify the people who should get million dollar token grants?

那太好了。

That'd be great.

我的意思是，这就是我们一直在问的。基本上，我们在这里是为了帮助那些像我们 18 到 30 岁或更早职业生涯时一样的人。我们如何真正找到这些人并帮助他们？那在今天会是什么样子？这就是目前的计划。就像，好吧，一切都变了，每个可能成为创始人的人，他们不再需要筹集十亿美元来实现梦想。比如，现在有 10 或 20 人的公司每年赚数亿美元

I mean, that's what we're asking all the time. Basically, we're here to try to help the people who were like us from when we were 18 to 30 or whatever, just earlier in our careers. How do we actually find those people and help them? and what would that look like today? And that's currently the plan. It's like, okay, well, everything just changed and everyone who could be a founder, well, they don't need to raise like a billion dollars anymore to build their dream. Like, there are companies right now that have 10 or 20 people making hundreds of millions of dollars a year

而且很快会达到数十亿。这看起来和我跟你谈到的那些成功案例截然不同，对吧？他们不像，如果你看看 YC 自 2005 年以来资助的一百家独角兽公司名单，你会发现，要达到规模并拥有客户，就会迫使你拥有成千上万的员工

and soon to be billions. And that looks radically different than some of the wins that I talked to you about, right? They don't to a tea like if you look at the list of a hundred unicorn companies that YC has funded since 2005. There's something about getting to scale and having customers that forces you to have thousands and thousands of employees

直到现在。

until now.

是的。而且问题是，其中一些真的非常低效。

Yeah. And then the thing is some of those are really really inefficient.

回到 Siri 甚至 Meta 眼镜的类比，你看那些组织，他们有成千上万人。为什么？那些人真的很聪明。他们真的把脑力和精力用在真实的事情上，还是只是午餐吃很好的生鱼片，有点像成人日托？

Going back to the Siri or even Meta Glasses analogy, you look at those organizations and they have tens of thousands of people. It's like why? And those people are really smart. Are they actually applying their brains and horsepower to things that are real, or are they going to have really nice sashimi at lunch and it's sort of adult daycare?

是啊，那也不错。说实话，这些都是很棒的工作。我有很多朋友在这些组织里，那些工作确实很棒。

Yeah, that's not bad. I mean honestly these are amazing jobs. I have a lot of friends who are in those orgs and those are incredible jobs.

但我也看到他们本可以做什么，以及社会多么迫切需要人类去做的事，那就是更加彼此服务。这怎么实现？我们怎么做到？

But I also look at what they could be doing and what society desperately needs human beings to do, which is to be even more in service to one another. How does that happen? How do we make that happen?

我担心的是，社会现状把人们锁进默认路径，比如去做管理咨询，去拿 MBA。这很安全，有养老金，有美好假期，开宝马。这没什么不好，那是很棒的生活，我希望更多人能拥有。但另一方面，

And my fear is that society, built as it is, locks people away into default paths, like just become a management consultant. Go get your MBA. This is safe. You'll get your pension. You'll have your nice vacations. Drive your BMW. There's nothing wrong with that and that's a great life and I want more people to have that. But on the other hand,

还有更多可能。

There's so much more.

上帝给了你大脑和感官，我们能做很多。我们怎么做得更多？怎么更好地服务彼此？技术当然不是坏事，它是为彼此解决更多问题的方式。

God gave you this brain and God gave you these senses and we can do so much. And how do we do more? Like how do we serve each other more? Surely technology is not a bad thing in that it is a way to solve more problems for one another.

所以我超级支持技术，超级支持市场，因为那是我衡量事物的方式。

And so I am super pro technology and super pro markets, cuz that's how I sort of score the circle.

跟我说说你跟机器的关系。

Tell me about your relationship to machines.

我爱机器。我的理念是，人类确实是从丛林里走出来的。我们能坐在这里，在美丽的地方谈论思想，而不是在某个地方勉强糊口，这其实是技术。是火，是轮子，是个人电脑，然后是互联网，这是不间断的链条。不只是机器。你会把书称为机器吗？

I love machines. I guess my ideology is that it does seem like humans came out of the bush. The reason why we get to sit here and sit in a beautiful place and talk about ideas instead of subsistence farming someplace is actually technology. It's fire. It's the wheel and it's the personal computer and then the internet, you know, like this is sort of the unbroken chain. And it's not just machines. Would you call a book a machine?

它是一种工具。

It's a tool.

它是一种技术。

It's a technology.

是的，是一种技术。

Yeah, it's a technology.

我在 PBS 上看过一部很有影响力的纪录片，叫《改变世界的机器》。那是 1995 年 WGBH 波士顿制作的关于个人电脑的纪录片。里面说电脑就是一本无限的书。但史前时代，在我们能写下东西之前，一切都丢失了，消失在时间的沙尘中。我们对人们怎么生活、做什么、关心什么毫无了解，直到我们发明了纸莎草纸、文字和书，然后随着活字印刷和古腾堡加速。于是识字率从一小撮祭司阶层突然变成人人都能识字。这就是每时每刻都在重演的事情。个人电脑就是这样，然后 iPhone 和移动革命也是这样。我们还没谈到那个。社交媒体就发生了。

There was this very influential documentary I saw on PBS called The Machine That Changed the World. It's like 1995 WGBH Boston documentary about the personal computer. And in it, it talks about that's what the computer is. It's an infinite book. But prehistory is before we can write stuff down. And it's lost. It's lost to the sands of time. Like we have no understanding of how people lived and what people did and what they cared about until we invented papyrus and writing and the book and then it accelerated with movable type accelerated with Gutenberg. And so literacy went from like a tiny priesthood to suddenly everyone could do it. And that's the thing that recurs every single moment. That's what the personal computer was. And then that's what the iPhone and the mobile revolution. We didn't get to talk about that yet. It's like social media happened.

嗯。

Yeah.

但有趣的是，Facebook 差点死掉，因为它几乎错过了移动端。Instagram 出现了，Facebook 不得不买下它，因为 Instagram 本来会杀死 Facebook，而不是扎克伯格。

But you know what's funny is Facebook almost died because it almost missed mobile. Instagram happened and Facebook had to buy Instagram because Instagram would have killed Facebook instead of Zuck.

Facebook 最初是桌面应用。

Facebook was a desktop application originally.

是啊，看起来像银行软件。

Yeah. It looked like bank software, you know.

嗯。

Yeah.

凯文·斯特罗姆本可以成为扎克伯格，但他卖得太早，这有点疯狂。这是历史重演。互联网电脑，是平装书电脑，突然变成联网电脑，突然变成 iPhone，然后现在是 AI，也就是智能电脑。它在加速，越来越快。我觉得这很棒。我的个人故事是，技术给了我一切，现在我希望每个人都能拥有。

And Kevin Systrom could have been Zuckerberg, but he sold too early, which is kind of crazy. That's sort of the history repeating itself. The internet computer, it's the paperback computer, suddenly the internet connected computer, suddenly the iPhone, and then now AI, like the intelligent computer. It's accelerating. It's happening faster and faster. And I think that that's awesome. Like I guess my personal story is that technology gave me everything that I have and now I want everyone to have that.

是的。对我来说，就像你必须加入一个祭司阶层，对吧？你必须学代数，必须拿全 A，必须进斯坦福。我是说，我反正做到了。那就像我隧道尽头的光。

Yeah. And for me, it's like you had to join a priesthood, right? You had to study algebra. You had to get straight A's. You had to get into Stanford. I mean, I did anyway. Like that was like my light at the end of the tunnel. Yeah.

但对于我的孩子，我希望他们那样做，但不是为了让他们进顶尖大学，而是希望他们热爱知识本身，对吧？我们未来走向哪里，我不知道我们是否还需要大学。那是后工业时代、20 世纪的老版本，就像给你一条车道，沿着这条路走，别问问题。

But like my kids, I want them to do that, but I want them not because I want them to go into the top college. It's because I want them to love the knowledge itself, right? And like where we're going, I don't know if we're going to need college, per se. That was the old post-industrial like sort of 20th century version of everything, like here's your lane and go down this path and don't ask questions.

感觉那些车道已经崩溃了。

Feels like those lanes have broken down.

是的，但我们必须找到新路。

Yeah, but we have to come up with the new path.

你总体上是个乐观主义者吗？

Would you say you're an optimist in general?

我觉得我必须如此。是的，我是个技术乐观主义者。我坚信技术可以也应该服务他人。

I think I have to be. Yeah, I'm a techno optimist. I definitely believe the technology can and should serve other people.

是的。如果技术没有做到，那其实不是技术问题，很可能是市场失灵的问题。说真的，为什么我不能拥有好的 Siri？因为苹果不允许那里存在市场。我们现在花很多时间在华盛顿和萨克拉门托，试图为此抗争。我们需要人们能够开创事业。任何破损的缝隙都很明显。Siri 就是个完美例子。最聪明的 AI 从去年 11 月就出现了，Siri 本可以那么聪明。但它有那么聪明吗？没有。为什么？因为没有市场。没有人创造开放访问。所以我们用着世界上最美的设备，但它们被严重削弱，以一种不该被破坏的方式破坏了。

Yeah. And then where it doesn't, it's really not a technology problem. It's probably a broken markets problem. Literally, why can't I have good Siri? It's because Apple won't allow a market to exist there. And we spend a lot of time in DC and in Sacramento right now trying to fight for that. Like, we need people to be able to start new things. And then any of the seams that are broken are obvious. It's like Siri is a perfect example. It's like the smartest AI has been around since November of last year and Siri could be that smart. But is it that smart? It is not. And why? There's no market. Like nobody created open access. And so we're using the most beautiful devices in the world and they're horribly crippled and they're broken in a way that it should not be broken.

跟我说说大科技与小科技。

Tell me about big tech versus small tech.

是的，小科技。我觉得 Siri 基本上就是这个问题。我们投资了一家公司叫 Beeper。我的前 YC 合伙人 Eric Migicovsky 创办的。它的愿景很简单，你应该能有一个单一的消息服务，是所有消息服务的超集。比如 iMessage、WhatsApp 和 Signal，为什么我们要有五六个不同的应用？我们应该有一个应用。但他直接撞上了没有市场。苹果说不，Meta 说不。为什么？为了公司的利益，而不是为了人民的利益。

Yeah, little tech. I think basically Siri in a nutshell is exactly that. Like we funded a company called Beeper. My former YC partner Eric Migicovsky started that. It had a very simple vision that you should be able to have a single messaging service that is a superset of all of them. So iMessage and WhatsApp and Signal, like why do we have like five or six different apps? Like we should have one app. But he ran headlong into there's no market for it. Apple says no and Meta says no. And why? For the good of the company, not for the good of the people.

听起来是个好的开源项目。

It sounds like a good open source thing to make.

是的，完全同意。问题是，你甚至不能做开源，因为即使开源，如果你用苹果，你必须越狱，对吧？那合法吗？我认为 DMCA 实际上有越狱规则，说如果是你自己的设备，你被允许这样做，但他们故意让它很难。所以结果是我们本可以拥有的最好技术，我们没有。为什么？为了大科技公司的利润。我认为在边缘地带，现在不重要。但当你谈到 AGI 和 ASI，超级智能就在这里。我认为这开始变得非常重要。

Yeah, totally. The thing is like you can't even make it in open source cuz even if it were open source, if you're using Apple, you have to jailbreak it, right? And is that legal? I think the DMCA actually has jailbreak rules that do say that you're allowed to do it if it's on your own device, but they make it intentionally hard. So as a result, the best technology that we could have, we do not have. For what reason? For profits by big tech. And I think like at the margins right now, it doesn't matter. But when you're talking about AGI and ASI, like super intelligence is here. I think it starts to matter a lot.

现在初创圈里有一个很大的争论，我觉得很有意思，也希望我们能找到答案。那就是 OpenAI、Anthropic、Google、DeepMind 这些公司拥有全球最前沿的 AI。争论的焦点是，随着模型越来越聪明，它们会不会一直保持开放？模型的使用权限是有可能关闭的。一旦发生，那意味着什么？是不是意味着我必须选边站？我是 Anthropic 的人。我们最终会不会陷入这种垂直的围墙花园？我是说，现在其实已经有点这样了，以后只会更糟。就像今天，你有绿泡泡、蓝泡泡。

There's a really big debate happening in startups right now that I think is very interesting and I hope we have an answer for it. That is, the OpenAIs, the Anthropics, the Googles, and DeepMinds of the world have the world's smartest frontier AI. And the big debate is, are they going to be open ongoing as these models get smarter and smarter? It is possible for the access to the models to close. And then once that happens, what does it mean? Does it mean that I have to choose? I'm an Anthropic person. Do we end up having these vertical walled gardens? I mean, that's sort of what we have now. It'll just be a lot worse. Like today, you have a green bubble, blue bubble.

但我们为什么会有这个？就是为了让苹果多赚钱。这又回到了建造者精神与那种尖脑袋经理人、金融兄弟精神的对比。这是两码事。我们支持建造者。如果建造者赢了，就意味着最好的技术能更快地到人们手里，也意味着小科技公司能蓬勃发展。

But why do we have that? It's just so that Apple has more profits. Again, it's sort of like the builder ethos versus the pointy-headed manager finance bro ethos. It's like, hey, these are two different things. And we're pro-builder. If the builder wins, it means the best technology gets in the hands of people faster, and it also means little tech can thrive.

现在的 AI 模型之间有多大差别？

How different are the AI models today?

它们有非常不同的个性。

They have very different personalities.

你会不会用一个模型做某项工作，用另一个模型做另一项工作？

Would you use one for a particular job and a different one for a different job?

当然。是的，Claude Opus 4.6 就像个多动症 CEO。他特别好玩，想法像每秒百万英里一样冒出来，真的能跟你产生共鸣。但他有时候也会自相矛盾，你知道，他会产生幻觉，有时记不住特别细节的东西，那是因为他脑子转得太快了。而 OpenAI 的 Codex 有点像那种几乎不说话、智商 200 的学者。你搞不懂它想说什么，它也不太理解你，但它是一种纯粹的机器智能，就是纯粹的聪明。所以我用的一个方法是，我主要跟 Claude 聊天，用 Claude Code，但在我的 GStack 技能里，我可以输入斜杠 Codex，然后它就把问题发给 Codex，我那位智商 200 的朋友。当我遇到问题，或者它卡住了，或者 Claude 就是搞不定的时候，它就会叫它那位智商 200 的自闭症哥们儿来帮忙。所以，简而言之，Claude 是多动症，Codex 是自闭症。而且这些模型每一个都有非常不同的个性。比如 DeepSeek，我发现它是……

Definitely. Yeah, Claude Opus 4.6 is ADHD CEO. So he's really fun. Like million miles per second ideas, really gets you, like really vibes with you. And then he actually like talks out both sides of his mouth sometimes, and you know he hallucinates and then can't keep track of like really detailed things sometimes, but that's 'cause his mind is going a mile a minute. And then Codex from OpenAI is sort of like the almost nonverbal 200 IQ savant. And so you're not gonna get what it's trying to say. It doesn't really understand you either, but it's a machine intelligence that is raw smarts. And so one of the things that I use is I mainly talk to Claude and I use Claude Code, but then in my GStack skill I can go slash Codex and then it sends it off to Codex, my 200 IQ friend. And when I have a problem or it gets stumped, or Claude just can't figure it out, it calls in its 200 IQ autistic buddy. And so yeah, in a nutshell, Claude is ADHD and Codex is autism. And every one of these models has like a very different personality. Like DeepSeek, I've discovered, is...

阴谋论者。

The conspiracy theorist.

你给它一个东西，它就会说：“嘿，你注意到这个了吗？”特别有意思。

It's like you give it something, it's like, "Hey, did you notice this thing?" It's fascinating.

听起来真有意思。

It sounds really fun.

是啊，真的很有意思。我觉得每个模型都有不同的个性，这是好事。我认为有多个模型，就像铁磨铁，互相砥砺。

Yeah, it's really fun. I mean, everyone has a very different personality and that's a good thing. Like I think having multiple models, it's sort of like iron sharpens iron.

嗯。

Yeah.

所以我希望那里有尽可能多的多样性。

So I want as much diversity there as possible.

而 Grok 比其他模型起步晚得多。它能赶上吗？

And Grok started much later than the others. Is it able to catch up?

我希望如此。我是说，我想要更多竞争，但很多当初构建它的人都离开了。所以，你知道，我希望它赢。我想要尽可能多的选择。而且 Elon 拥有所有的 GPU。所以，有趣的是，他能留住人才吗？问题是，外面总有更多聪明人，他们可以来把这件事做成。所以，我支持他们。令人惊讶的是，API 这块，他们好像不太推 API。我觉得他们就是想直接速通到 ASI。我想你在机器人上也会看到这一点，他们有点像垂直整合，可能跟任何人一样先进，但你知道，我们还没看到实际使用。所以，我想这就是那个大实验。永远不要赌 Elon 输。而且 SpaceX 的 IPO 对整个科技界来说都会非常棒，非常史诗。

I hope so. I mean, I want more competition, but a lot of the people who built it have left. So, you know, I want it to win. Like I want as many options as possible. And Elon has all the GPUs. And so, you know, the interesting thing is, can he retain the people? And the thing is, there's always more smart people out there who could come and make that work. So, I mean, I'm rooting for them. The surprising thing is, like, the APIs, like they don't push the APIs very much. I think they're just like trying to speedrun their way to ASI directly. I guess you'll see that also with the robots, like they're sort of vertically integrated and probably as advanced as anyone, but you know, we haven't seen the usage yet. So I mean, I guess that's the grand experiment. Like never bet against Elon. And the SpaceX IPO is going to be pretty fantastic and epic for all of tech.

是不是每家公司都应该上市？

Should every company go public?

我觉得这是创始人和股东的个人问题。不过我看到的趋势是，如果你建立了有价值的东西，会有很多二级市场，人们会直接来买。上市和私有的区别其实很微妙。如果要说的话，我认为普通人应该能买私有股份。那可能会大大打开局面。

I feel like that's the personal question for the founders and the shareholders. I guess what I'm seeing though is, if you have built something valuable, there's a lot of secondary, like people will just come and buy it. And the difference between public and private is pretty subtle. If anything, I think that normal people should be able to buy private shares. That would probably open things up quite a bit.

这可能吗？

Is that possible?

有点难。你得去 Forge Global 或 AngelList 这些地方，通过各种方式买私有股票。挺麻烦的。

It's kind of hard. You have to go to things like Forge Global or AngelList or all these different ways to buy private stock. It's kind of hairy.

不过听起来是个好主意。是啊，很多人都在这么做，而且可能应该有更多人能参与。我是说，我们家最近在二级市场买了一点 Anthropic 的股份。然后那些非常聪明的人之间有个笑话：如果你在二级市场买了点 Anthropic，那就意味着你能升级到二级面包线，跟其他 Anthropic 董事一起排队。你知道，在未来的 AGI 世界里，从三级升到二级可是件大事。

Sounds like a good idea though. Yeah, a lot of people are doing it and then more people probably should get access. I mean, our family recently bought a little bit of Anthropic on secondary. And then the joke among very smart people is, great, if you bought a little bit of secondary in Anthropic, it means you get to get upgraded to the level two bread line with the other Anthropic directors. You know, going from level three to level two is a big deal in the future AGI world.

大科技公司会不会竭尽全力阻止小科技公司存在？

Does big tech do everything they can to prevent little tech from existing?

我觉得这不是针对个人的。我认为资本主义意味着你应该出去拼命竞争。但与此同时，我也不是自由市场极端主义者，我不是自由意志主义者。我认为政府在市场不开放、准入不开放的时候，有角色和责任要扮演，就是为了给人们选择，对吧？比如，Anthropic 现在拥有最好的模型，这很棒，我很喜欢，我是它的超级粉丝，而且 YC 的几个校友其实是 Anthropic 的联合创始人。但与此同时，我有点希望他们能少一点反竞争。比如，OpenCode 是一个开源代码工具，我喜欢 Claude Code，但我也希望 OpenCode 能成功，能被使用。我觉得这些都是大问题，对吧？当你跟很多 Anthropic 的人聊，你会发现他们很多是 Google 出来的。你知道，Google famously 不再“不作恶”了。你知道，某个时候他们开始说：嘿，我们是个企业，我们需要赚钱，这没问题。我也不怪他们。但正是在这里，我认为政府有角色要扮演，当看门狗或监督者。我的意思是，如果你就让公司为所欲为，它们不是不道德，但也不是道德，它们是非道德的。然后人民的意志通过政府来体现。这就是为什么我说，人们应该投票。我们需要健康的媒体，我们需要健康的新闻业。

I don't think it's personal. I think capitalism means that you're supposed to go out and fight really hard. But at the same time, I'm not a free market maximalist in that I'm not a libertarian. Like I think that there are roles and responsibilities for government to play when there aren't open markets and open access, and just to give people choice, right? Like, you know, it's totally great that Anthropic, for instance, has the best model right now, and I love it, and I'm a big fan, and a couple of the YC alums are actually co-founders of Anthropic. But at the same time, it's like I kind of wish that they would be less anti-competitive. For instance, like if OpenCode, which is an open-source code harness, like I love Claude Code, but also I want OpenCode to be successful and to be able to be used. I think those are big questions, right? When you talk to a lot of the Anthropic folks, like a lot of them are ex-Google. You know, Google famously stopped being sort of "Don't be evil." You know, at some point they started saying like, hey, we're kind of a business and we kind of need to make money, which is fine. Like I don't begrudge that either. But that's where like I do think that there's a role for government to play as a guard dog or watchdog. I mean, if you just let companies do whatever they're going to do, like they're not immoral, but they're also not moral, they're amoral. And then the will of the people comes through the government. And this is why I'm like, people should vote. And we need a healthy media. We need healthy journalism.

只是不要垄断。

Just not monopolistic.

对，对。我是说，我们应该反垄断。我觉得垄断是坏事。

Yeah. Yeah. I mean, we should be anti-monopoly. Like I think monopolies are bad.

嗯。

Yeah.

在 YC 的语境下，什么是零基预算？

What is zero-based accounting in the context of YC?

YC。我想想，我回来当 CEO 的第一天，董事会开会，成员是 Paul Graham、Jessica Livingston、Brian Chesky 和 Carolyn Levy。那就是 YC 的董事会，加上我。

YC. I guess my first day coming back as CEO, the board met and was Paul Graham and Jessica Livingston and Brian Chesky and Carolyn Levy. That's sort of the board of YC and me.

我接到的头号指令就是：YC 3.0 应该是什么样？YC 1.0 是 PG 时代，YC 2.0 是 Sam Altman 时代。然后 YC 3.0，我们做零基核算——假设我们从零开始，我们现在做的每一件事还会不会重新选择去做？这就是我们做的。我觉得每个组织可能都需要这样，尤其是在这种领导层更替的时候，值得做一次。如果要改变，我们就得趁现在尽可能多地改变。

And the number one directive I got was like, what does YC 3.0 look like? YC 1.0 was under PG. YC 2.0 was in Sam Altman's era. And then YC 3.0, let's do zero-based accounting. Let's figure out if we were starting from zero, would we choose to do everything we do today again? So that's what we did. And I think every organization probably needs that, especially at a leadership change like that. It's worth doing. If we're going to change, we need to change as much as we can right now.

而不是躺在过去的功劳簿上。就像一次重启。

Not just rest on the laurels of what happened in the past. Like a restart.

没错。你怎么重新创立这家公司？

Oh yeah. How do you refound the company?

听起来很棒。很有意思。对你来说是个有趣的使命。

That sounds great. It was fun. Sounds like a fun mission for you.

看着这一切真是不可思议，因为这东西改变了我的人生。现在我们正大干一场，准备正面迎接 AI 这股浪潮。

It was insane to see, because I mean, this thing changed my life. And now we're rocking and rolling. Like we're ready to meet this AI thing head on.

当你创办一家新公司时，了解你所在领域的历史有多重要？

How important is it to know the history of the category that you're going into when you're starting a new company?

如果以 Airbnb 为例，也许没那么重要。但一旦你身处其中，你就得了解关于它的一切。事实上，有一篇文章叫《睡眠盲区》。如果 Stripe 的创始人知道这会有多辛苦、多糟糕，他们就不会开始了。这对很多创造性事业来说可能都是如此。几乎任何人，如果知道他们要经历那些……

If I go by the Airbnb example, maybe not so important. But as you're in it, you have to learn everything about it. In fact, there's an essay about it called 'Sleep Blindness.' If the founders of Stripe knew how much work and how terrible it would be, they wouldn't have started. And that's probably true for a lot of creative endeavors. Almost anyone, if they knew they had to go through the...

而且你花了太多时间思考……

And you spent too much time thinking about the...

你很可能就会选择去做别的事了。说实话，创业并不适合所有人。也许最需要破除的迷思就是：每个人都该创业吗？这就像问，每个人都该当音乐家吗？每个人都该成为某某某吗？答案是否定的。没有一种职业适合所有人。而且有些职业太难了，比如创始人，你最好知道这是你唯一能做的事，否则你会疯掉，因为每隔一天你就会说：“你知道吗，这太难了，我不干了。”

You probably just choose to do something else. And honestly, starting a company isn't for everyone. Maybe that's the biggest myth to dispel: should everyone start a company? Honestly, that's like asking, should everyone become a musician? Should everyone become X, Y, or Z? And the answer is no. There's no one-size-fits-all profession for someone. And some of them are so hard, like founder, that you better know that's the only thing you could do, otherwise you'll go crazy, because on every other day you're going to say, 'You know what, this is too hard and I'm out.'

Vibe coding 在多大程度上改变了局面？

How much has vibe coding changed things?

我觉得我们才刚刚开始看到它的影响。但现在最疯狂的是，人们的思想观念在多大程度上没有改变。比如我受到很多来自非常聪明的人的抨击，比如那些备受尊敬的资深系统工程师。有趣的是，他们中有些人点名攻击我，然后我去看他们的开源项目，却看不到任何进展。他们的进步速度和一年前、两年前、五年前一样。

I think we're just starting to see it. The wildest thing right now though is to what degree people have not changed their view. Like I'm taking a lot of attacks from really smart people, like really smart systems engineers who are really well-respected. And what's funny about it is some of them attack me by name, and then I go and look at their open source projects and I don't see any movement. They're progressing at the same pace that they were progressing a year ago, two years ago, five years ago.

嗯。

Yeah.

等我完成一些发布之后，我想我会进入那些开源项目，直接给他们提交一些 PR。用程序员的话说，就是免费送他们一份礼物，比如几万行修复代码。我觉得我应该真的送他们一份礼物，然后看看他们是否理解并接受。如果他们不接受，那也没关系。最讨厌 vibe coding 和 AI 的工程师，恰恰是那些最能从拥抱它中获益的人。

And after I get some launches done, I think I'm going to go into those open source projects and I'm just going to drop some PRs on them. In programmer terms, it's like I'm just going to give them a free gift of tens of thousands of lines of fixes to their software. And I think I should actually just give them a gift and then see if they understand and accept it. And if they don't, that's okay. The engineers who hate vibe coding and AI the most are the people who would benefit the most from embracing it.

YC 有没有出现过硬件公司，而不是软件公司？

Has anything come through YC that is hardware-based as opposed to software-based?

哦，太多了。是的。比如 Stoke Space，很可能很快就会和 SpaceX 一起成为可重复使用火箭技术的领先者。还有 Astronis，实际上是 SpaceX 的竞争对手，这很有趣。他们提供互联网接入。我想现在很多西南航空的航班都使用 Astronis 的互联网。

Oh, a ton. Yeah. I mean, Stoke Space, for instance, is probably going to be alongside SpaceX with one of the reusable rocket technologies soon. You know, Astronis is actually a competitor to SpaceX, which is fun. They're providing internet access. I think many Southwest flights now have internet from Astronis.

他们是 Starlink 的竞争对手。对。

They're a competitor to Starlink. Yeah.

所以，是的。有趣的是，我非常尊敬 Elon Musk。他去年在 Startup School 演讲，那太棒了。我也认为我们需要一百个那样的人，无论男女。我的意思是，我们需要……

So, yeah. What's funny about it is I both respect the heck out of Elon Musk. He came to speak at Startup School last year and that was awesome. I also think that we need a hundred of those guys or gals. I mean, I think that we need...

绝对。

Absolutely.

成百上千的人重新上马，去解决人类最棘手、最重大的问题。我们需要更多这样的人。所以如果回顾 YC 的这个时代，这大概就是我们的宏大而大胆的目标。我们如何不只是创建新的软件企业来解决人们的金钱问题？我们如何真正地创建新的全球性公司……

Hundreds of people who get back in the saddle and then try to solve the biggest hairiest problems for humanity. And we need a lot more of it. And so if I look back on this age of YC, that's probably the big hairy audacious goal for us. How do we not just create new software businesses that solve the money problem for people? How do we actually literally create new global companies...

制造酷东西……

That make cool stuff...

制造世界上最酷的东西，对吧？其中一部分实际上是给真正优秀的工程师一条发挥最佳水平的道路。我觉得很多人不想做银行软件，很多人不想帮人点广告，很多人不想当律师。很多人想做真正有意义的事情。

That make the coolest things in the world, right? Some of it is actually about giving really great engineers a lane to do their best work. I think a lot of people don't want to work on bank software. A lot of people don't want to help people click on ads. A lot of people don't want to be lawyers. A lot of people want to do something that's really, really meaningful.

然后我觉得我们的组织需要弄清楚如何帮助他们做到这一点？

And then I think our organization needs to figure out how do we help them do that?

是的。一个你可以围绕它、可以用手触摸的实体物体，确实有些不同。那和一款优秀的应用就是不一样。

Yeah. There is something different about a physical object that you can circle and that you can touch with your hands. That's just different than a great app.

哦，是的。我们正在进入那个领域。有趣的是，做这件事之所以难，很大程度上是因为 AI 和软件还不够。所以这有点像分形。我是说，来自 Android 的 Palmer Lucky 谈到了这个……

Oh yeah. And we're coming into that. The interesting thing is a lot of the reason why it's hard to do that is actually not enough AI and not enough software. So it's sort of fractal. I mean Palmer Lucky from Android talks about this...

而且我想 3D 打印也会有所帮助。

And I imagine 3D printing is going to help as well.

哦，是的，当然。所以我认为我们想要加速所有的事情。

Oh yeah, definitely. So I think we want to accelerate all of the things.

嗯。

Yeah.

我的意思是，下一步很明确。将是机器人技术，然后我们需要再次制造东西。你知道，我们有一家叫 Knox Metals 的公司，帮助美国公司获得金属，而不是必须拿起电话，比如。

I mean the next thing is pretty clear. It's going to be robotics and then we need to build stuff again. You know, we have a company called Knox Metals that is helping American companies get access to metal instead of having to pick up the phone, for instance.

“获得金属”是什么意思？我不知道那是什么意思。

What does 'get access to metal' mean? I don't know what that means.

哦，字面意思就是，如果你需要一大卷铝，你今天怎么得到它？你仍然必须拿起电话，然后被宰一刀之类的。我们需要一个完整的美国供应链体系。

Oh, literally like, if you needed a giant roll of aluminum, how are you going to get it today? You still have to pick up the phone and then you get gouged and things like that. We need a whole American supply chain system for supply chain.

哦，是的，那很酷。

Oh yeah, that's cool.

是的。有趣的是我们在这方面扮演着角色。这正是政府和监管可以发挥重要作用的地方。Knox Metals 是一个买卖大卷铝或任何你想买的金属的市场。但有一个小集团。

Yeah. It's interesting because we're playing a role in that. This is where government and regulation actually has a big role to play. Knox Metals is a marketplace for giant rolls of aluminum or whatever kind of metal you want to buy. But there's a cabal.

那是一个由少数经纪人组成的寡头垄断。他们开始排挤 Knox Metals。然后这就是社交媒体有趣的地方。这不是坏事，这是向权力说真话。那位创始人去了 Twitter，开始发帖，然后通过我们的人脉，Luther Low 是我们的政策主管。我们在华盛顿有全职人员，他认识 Gail Slater，她是华盛顿反垄断监管机构的一员。她只需要转发那条帖子并发表一个表情符号，然后你猜怎么着，Knox Metals 的所有反竞争垄断问题一夜之间就消失了。

There was like a small set, an oligopoly of brokers for that type of stuff. And they started freezing Knox Metals out. And then this is where social media is interesting. This isn't a bad thing. This is speaking truth to power. That founder went to Twitter, started posting about it, and then through our connections, Luther Low is our head of policy. We have a full-time person in DC, and he knows Gail Slater, who's sort of part of the antitrust watchdogs of DC. And all she had to do was take that post and post an emoji, and then you know what, all of the anti-competitive monopoly stuff went away for Knox Metals literally overnight.

有趣的是，科技界的人害怕政府。人们担心它的影响，但政府本质上是可以用来做好事的，如果真是那样的话，那将非常了不起。

So it's funny, people in tech are afraid of government. People are worried about its impact, but it's like government is something that can be used for good fundamentally as incredible if it was the case.

是的。关键在于接触，对吧？如果你能接触到他们，帮助他们，参与对话，这就是我们正在学习的。从小到大，我爸爸总是说：“孩子，你得远离政治。”但那就是来自亚洲、来自 20 世纪的那种氛围，你知道，我父母就是那样。比如我的祖父，他在中国南部的云南经营丝绸厂，后来工厂被没收，全家不得不逃亡。八个兄弟姐妹逃到了缅甸，祖父在缅甸重建了生意，但后来又被军政府没收了。

Yeah. It's about access, right? Like if you can access them and help them and be in the conversation, that's what we're learning. Growing up, my dad was always like, "Man, you got to stay away from politics." But that's what that was the vibe from like Asia, you know, from the 20th century. Like that was what my parents were. My grandfather for instance, he ran silk factories in Eunan in southern part of China and then they were seized by and the family had to flee. Eight brothers and sisters had to flee to Burma and my grandfather rebuilt that business in Burma and then it was seized again by the military hunter.

哇。

Wow.

我的祖父在缅甸去世，但八个兄弟姐妹都以移民身份成功到达了加拿大和美国。所以有很长一段历史，美国是那些想要创造和建设的人来的地方。移民是这段故事的一部分，而且是一个美丽的故事。我们必须保留这一点。我们还必须保留人们在美国创办新企业的能力。为此，我们还必须保留这样一种观念：创业不是坏事或邪恶的事，而是高尚的事。

And my grandfather died in Burma, but all eight brothers and sisters made it to Canada and the US as immigrants. And so there's a long history of like America is the place where people who want to create and build come here. Like immigration is a part of this story and it's a beautiful one. And we got to preserve that. And we also have to preserve the ability for people to create new businesses in America. And for that we actually also have to preserve the idea that starting a business is not bad or evil but noble.

如果你今天 16 岁，你还会去斯坦福读计算机科学吗？还是你会用 vibe coding 创办你的第一家公司？

If you were 16 years old today, would you still go to Stanford for computer science or would you vibe code your first company?

哦，天哪，我可能会试着去斯坦福至少读几年。说实话，我觉得我可能只想读斯坦福，兄弟。我觉得这其实是个很难的问题。

Oh man, I'd probably try to go to Stanford at least for a couple years. And honestly, I don't I think I might want to just do Stanford, man. I think that's actually the hard question.

那是一段很棒的经历。

It was a great experience.

那是一段很棒的经历。当我 18 岁时，我的童年如此艰难，以至于我在大学里感觉自己像 14 岁。我的整个 20 多岁都像是延迟的青春期。我不得不弄清楚自己是谁。

It was a great experience. And when I was 18, my childhood was so difficult that I felt like I was 14 in college. My whole 20s were sort of this delayed adolescence. And I had to figure out who I was.

是的。有趣的是，在 YC 有一种氛围，觉得 YC 只想资助像 18 岁或 20 岁的帕特里克·科里森那种类型的人。

Yeah. It's funny. At YC there's this vibe that YC only wants to fund the like 18-year-old or 20-year-old sort of Patrick Collison types.

我不认为这是真的。至少对我来说不是这样。比如我在 27 岁创办了我的公司。我只是还没准备好。我是个晚熟的人，还没有完全成熟。我的 YC 创业公司叫 Posterous，它实际上成为了前 200 名的网站。你给我们的邮箱发一封邮件，我们就会为你创建一个博客，这是从 iPhone 上获取照片最快、最简单的方式，现在听起来很疯狂，但基本上

I just don't think that's true. That's not true for me anyway. Like I started my company at 27. I just wasn't ready. I was a late bloomer and I just wasn't fully developed. My YC startup was called Posterous and it actually became a top 200 website. You send an email to our email address and we would make a blog for you and it was the fastest, easiest way to get a photo off of your iPhone, which sounds crazy now, but basically

那时候你做不到。

you couldn't do it back then.

是的，你可以那样做。是的。

Yeah, you can do it that. Yeah.

是的。

Yeah.

而那个领域后来变成了 Instagram。所以现在每个人都在用 Instagram。所以在某种程度上，科技领域一直在进行一场锦标赛，胜利者有点像《拜金一族》。第一名是你赢，第二名是一盒牛排刀，第三名是你被解雇。基本上科技竞争就是这样运作的。我们得到了那盒牛排刀，然后把公司以 2000 万美元卖给了 Twitter，这对我来说非常有力且有意义。

And what that space became was Instagram. So everyone uses Instagram now. So in a way like in technology there's a tournament happening all the time and the winner it's sort of like Glengary Glenn Ross. First place is you win, second place is a box of steak knives and third is you're fired. It's like that's basically how the competition of technology sort of works. We got the box of steak knives and sold that company to Twitter for $20 million which is very powerful and meaningful for me.

是的。但当我回顾那里发生的事情时，我觉得我需要成为现在的我才能正确地赢得那场比赛。当我真正审视时，我还没有完成 10 年的心理治疗。而那才是我真正需要的

Yeah. But when I look at what happened there, it's like I needed to be who I am now in order to properly win that tournament. And when I really look at it, I hadn't done 10 years of therapy yet. And that's what I actually needed

才能解决所有导致我创业失败的问题。我最终和联合创始人闹翻了。实际上，创业公司在 Instagram 上线那天就开始走下坡路，我们本应该注意到的。是的。

to be able to work through all of the things that caused my startup to fail. I ended up falling out with my co-founder. The startup started failing the day Instagram launched, actually, which we should have paid attention to. Yeah.

然后我还没有成长为一个完整的人，无法去和我的联合创始人说我的情况。

And then I just wasn't developed as a full human being to be able to go to my co-founder and say what was going on for me.

我只是忍气吞声。他对自己想做的事情有非常强烈的想法。我好像在互联网上读过一些建议，说如果你想让你的创业公司成功，你最好成为一个非常好的联合创始人。你是一个联合创始人。我读了那篇文章，然后我想，哦，那意味着我应该自我放弃。我应该照他说的做，对吧？

I just ate it. He had a really strong idea for what he wanted to do. I was like I had read some like advice on the internet about if you want your startup to succeed, you better be a really good co-founder. You're a co-founder. And I I read that and I thought like, oh, that means that I should self-abandon. I should just do whatever he says, right?

不，你是认真的吗？比如，

No, are you serious? Like,

我当时 30 岁，我希望我能回到过去，摇醒那时的自己，但我就是没准备好。比如，我童年有所有这些未经审视的心魔。你知道，我有时想知道我是否也在犯同样的错误。我现在可能只是不停地编码。那时候，我就像英雄工程师。比如，我写了所有的代码。我做了所有的设计。我甚至保持服务器运行，这非常不明智。然后现在经过很多心理治疗，我意识到哦，那是我的英雄情结，你知道，我必须支撑我的家庭，我的父母是我的父母，但我不得不像父母一样照顾他们，这导致了我现在的样子，但也导致了我的失败，导致我无法管理人员，无法转型，无法做出帮助我的创业公司的决定。我 2008 年的 YC 创业公司曾有机会。我们本可以成为竞争者。我经常想到这一点。你知道，如果你不了解自己，或者你没有解决内心真正发生的事情，或者你不知道自己如何做决定，如果你不是一个整合的人，这些东西都会显现出来。再说一次，

I was 30 at the time and I wish that I could go back and like shake the me of that time, but I just wasn't ready. Like, there are all these unexamined demons from my childhood. You know, I sometimes wonder if I'm making the same mistake, too. I might be just by coding all the time right now. Back then, I was like the hero engineer. Like, I just wrote all the code. I did all the design. I even like kept the servers up which was very inadvisable and then now after a lot of therapy I realize oh that's my hero complex you know I had to carry my family my parents were my parents but like I had to sort of parent them as well and that's what caused me to be the way I am but also it caused me to fail caused me to not be able to manage people not be able to transition not be able to make the decisions to help my startup my YC startup in 2008 had a shot. We could have been a contender. And I think a lot about that. You know, if you don't know who you are or you haven't worked through the things that are really going on inside of you or how you make decisions, if you're not an integrated person, this stuff all comes out. Again,

我只想从另一面来论证。有时候，那些完全整合的人无法达到突破所必需的非人、不切实际的极端。

I'll just argue the other side of that. Sometimes the people who are fully integrated can't go to the inhuman, unrealistic extremes necessary to break through.

是的。是的。

Yeah. Yeah.

比如有时候正是因为那些创伤，你才能不断锤打。

Like sometimes it's because of that damage that you can just keep hammering.

是的。

Yes.

我不是建议把它作为一种选择，但现实是，有时候就需要那样。

I'm not suggesting that as a choice, but it's just the reality that sometimes it takes that.

是的。我还没有调和这一点。我的意思是，这绝对是事实。比如当我们遇到那些完全投入的人，那能量来自某个地方。

Yeah. I haven't reconciled that yet. I mean, that's definitely true. Like when we meet people and they're all the way turned on, it's like that's coming from someplace.

它不一定来自积极的地方，但那可能是跨越差距的能量。一旦成功了，他们就可以在自己身上下功夫，并且有时间。在比赛中，你除了比赛什么都不能关注。

It's not always a positive place, but that might be the energy to get over the gap. And then once it works, then they can work on themselves and have time. During the race, you can't focus on anything but the race.

嗯。你说到了一些非常深刻的东西，我在这里要非常坦诚。这实际上是我现在正在挣扎的一个决定性的事情。

Mhm. You're hitting on something very deep and I'm just going to be super vulnerable here. Like that's actually one of the defining things I'm struggling with right now.

比如我内心有愤怒和火焰，我在心理治疗中看到了它。实际上，我会去拜访那个房间。

Like I have a rage and a fire inside me and I see it in my therapy sessions. Actually, I go visit that room.

是的。比如我可以去那个公寓拜访我七岁的自己，那里堆满了啤酒罐和技术书籍，很小，是个囤积者的空间，我睡在地板上，你知道，我们没有床，我们只有地板上的泡沫垫。

Yeah. Like I can visit my seven-year-old version of myself in that apartment with filled with beer cans and technical books and it's tiny and it's a hoarder space and like I slept on the floor, you know, like we didn't have beds like we had foam pads on the floor.

嗯。

Mhm.

我可以随时去那里，我能感受到我酗酒的父亲要回家的恐惧

And I can go there anytime and I can feel the dread of my alcoholic father is going to come home

而且我不知道谁会回家。

and I don't know who's going to come home.

他会不会已经喝醉了？他会不会打我？就像，会发生什么？但这也是当人们在网上看到我时，他们会想，这家伙怎么了？就像，这家伙到底怎么回事？你知道，或者他们看到我带着那种极端的强度去追求我想要的东西和我要实现的事情。那是因为我能走进那个房间，找到我七岁的自己，他一直都在那里。然后我拿一个杯子，就像是一个放射性杯子，他倒出像液态钚一样的东西，然后我带着它开始我的一天，我就有了力量。

Is he going to be drunk already? Is he going to beat me? Like, what's going to happen? But that's also when people look at me on the internet and they're like, what is wrong with that guy? It's like, what's going on for that guy? You know, or like they see me with like the very extreme intensity of what I want and what I'm trying to make happen. It's because I can go into that room and I go to my seven-year-old version of myself and he's in there all the time. And then I bring in a cup and it's like this radioactive cup and he just pours out like liquid plutonium and then I go into my day and I got it.

是的。你会说你和他的关系是，他并不总是主导一切吗？

Yeah. Would you say you have a relationship with him where he's not running the show all the time?

我的意思是，如果非要说的话，是我在主导，我在利用他。

I mean if anything like I'm running the show and I'm using him.

听起来像是一种健康的关系。

That sounds like a healthy relationship.

我希望如此。

I hope so.

那可能是你的超能力。

That could be your superpower.

你觉得呢？你认为是这样吗？我的意思是，我的治疗师说，“老兄，你得把那个孩子整合起来，去拥抱他。”然后我担心，如果我那样做了，我还能做我现在做的事情吗？

What do you think? Do you think so? I mean, my therapist is like, "Man, you gotta you need to integrate that kid and like hug him." And then I'm afraid, well, if I do that, will I be able to still do what I do?

我还不知道。我不知道答案是什么。

I don't know yet. I don't know what the answer is.

你觉得他现在有没有在某种程度上阻碍你？

Do you feel like he's in any way getting in your way now?

是的，一直如此。

Yeah, all the time.

你会说你有自毁倾向的一面吗？

Would you say there's a part of you that's self-destructive?

嗯，我已经戒酒了，所以这很好。

Well, I stopped drinking, so that's good.

听起来很棒。

That sounds great.

有趣的是，我确实有自毁倾向，我现在非常非常清楚这一点。但当我年轻的时候，我会想，“哦，不。我很好。我很正常。一切都好。我有这种极端的控制欲。要么是自我放弃，要么是支配，对吧？”而现在整合的问题是如何保持中间？我妻子总是说，“加里，你太极端了。比如，不一定要吃一磅西兰花或一磅牛排。有时候我们只想吃鸡肉，而我在努力。我在努力，你知道的。”

Funny enough, I do have a self-destructive streak that I am now very, very aware of. But when I was younger, I was like, "Oh, no. I'm fine. I'm normal. Everything's fine. I have this like very extreme control thing. It's either self-abandonment or it's domination, right?" And the integration thing right now is how do I, you know, stay in the middle? My wife always says like, "Gary, you're so extreme. Like, it doesn't have to be a pound of broccoli or a pound of steak. Sometimes we just want to eat chicken and I'm trying. I'm trying, you know.

我听到你这么说，我想，只要不是自毁的，就像你用的“正常”这个词。正常并不是值得追求的东西。

I wonder just hearing you say it, as long as it's not self-destructive, like you use the word normal. It's like normal is nothing to aspire to.

是的。那是中位数。

Yeah. That's the median.

是的。正常就是平庸。

Yeah. Normal is average.

是的。没错。我们想要非凡。

Yeah. Yep. We want to be extraordinary.

是的。有时候那部分是一种来自某处的火焰。只要它不烧到你

Yeah. And sometimes part of that is a fire that comes from somewhere. As long as it's not burning you

是的。

Yeah.

或者烧到你周围的人，并在世界上做好事。用这股火焰去做好事真的很好。

or burning the people around you and doing good in the world. Having the fire to use for good is really good.

是的，完全同意。

Yeah, totally.

我想现在对我来说，这个问题出现在哪里，我的意思是，我们要变得非常私人，但我认为这是我很多朋友实际上都在挣扎的事情，他们从不谈论它。所以，也许我们需要让谈论它变得正常化。

I guess where it's coming up for me right now, I mean, we're going to get super personal, but I think that it's something that a lot of people who I hang out with actually struggle with and they'd never talk about it. So, maybe we need to normalize talking about it.

是的，

Yeah,

这杯钚有一种有害的影响，那就是无法在家中出现或关闭。

this cup of plutonium has a dilitterious effect and that is not being able to show up or turn off at home.

嗯。

Mhm.

所以，你知道，我家里有儿子。我的妻子非常了不起，她一直支持我。我刚刚和我的个人 AI 一起做了一些关于沃伦·巴菲特的研究，他在 47 岁时著名地离婚了。我猜他的第一任妻子苏西搬到了旧金山，当时沃伦说苏西很棒，但她不理解复利的价值，然后多年后，我的意思是，我想最近，也许你知道，30 年后，他们没有离婚，只是分居。

And so, you know, I've got boys at home. I have my wife is incredible and she supported me all the way through this. And I was just working with my personal AI doing a bunch of research about Warren Buffett who at 47 got divorced famously. I guess his first wife Susie moved to San Francisco and at the time Warren talked about Susie's amazing but she doesn't understand the value of compounding and then years later I mean I think recently maybe like you know 30 years after they didn't get divorced like it just was a separation.

是的。

Yeah.

但你知道，在他七八十岁时，他开始说那是我一生中最大的错误。

But you know in his 70s or 80s he started talking about like that was the greatest mistake of my life.

是的。所以，我 45 岁。我刚满 45 岁。我想这就是我的想法。就像，我并不是那么独特。我是。

Yeah. So, I'm 45. I just turned 45. And I guess that's where my head goes. It's like I'm not that unique. I am.

但它是全天候的吗，还是

But is it 24/7 always or is it

它总是全天候的。所以，你你

It's 24/7 always. So, you you

我在努力停止。

I'm trying to stop.

好的。

Okay.

我在努力创造那个空间，但我发现这非常困难。

I'm trying to make that space, but I find it incredibly hard.

我理解。我是基督徒，所以我不太相信这个，但如果有轮回，我会活过 100 亿世来占据这个身体和这个思想，在这个地球上，在这个时刻，有一种使命感，我看到一切都排列好了，有些事情我需要完成，如果我不做，我不确定它们会不会被完成。

I understand. I'm Christian so I don't really believe this but like if there was reincarnation I would have lived 10 billion lifetimes to inhabit this body and this mind on this earth at this moment there's like a sense of mission I see like everything sort of arrayed out and there are things that I need to get done that like if I don't do them I'm not sure if they're going to get done.

是的。你怎么接受这一点？你怎么回家，基本上在门口脱下防火服，然后回家和妻子孩子吃晚饭？

Yeah. How do you sit with that? And how do you go home and basically take the fire suit off at the doorway and then go home and have dinner with your wife and kids?

是的，

Yeah,

我认为这正是人们现在比以往任何时候都更需要和想要的，那就是真正与他人建立联系。

I think that's what people actually need and want now more than ever, which is just actually connection with other people.

我认为这正是社会某种程度上的需要。比如在旧金山，我们一直在做一件叫做“法案 17”的事情，我们试图让更多科技人进入教堂，这听起来几乎与科技背道而驰。实际上，科技圈有一种非常强烈的无神论文化，我想我能理解。我成长过程中也是无神论者。我是无神论者。

And that's sort of what I think society sort of needs. Like in in San Francisco, we've been working on this thing called Act 17 where we're trying to bring more tech people into church, which sounds sort of almost antithetical to tech. There is actually a really strong atheist sort of culture around tech, which I think I get. And I was an atheist when I was growing up. I was an atheist.

是的。

Yeah.

因为我看着我小时候的经历，我就是不相信上帝应该是全知、全能和全善的。我看着我的成长方式，觉得其中一件事是错的。

cuz I looked at what I experienced as a child and I just didn't believe that basically God is supposed to be allseeing, all knowing and all powerful. And I looked at how I was growing up and was like one of these things is wrong.

告诉我一件你现在相信但你年轻时不相信的事情。

Tell me something you believe now that you didn't believe when you were young.

我认为最大的一个信念是，你知道，当我 22 岁左右时，我的朋友们正在创办一家公司，他们说，“加里，你在微软。你是我认识的最聪明的工程师。我们要创办这家公司。你得辞职来加入我们。”我当时想，“老兄，我不知道，老兄。不管你们在做什么。我有一份工作。好吧，得到这份工作太难了。你们在做什么？”他们最终说服我从西雅图飞到旧金山，我和他们的恩人共进晚餐。结果发现是彼得·蒂尔。那可能是在他给 Facebook 写 50 万美元支票后的几个月内。所以是我，我的两个哥们，乔·朗斯代尔和斯特凡·科恩，还有彼得·蒂尔。彼得说，“你在微软做什么，老兄？你得辞职。”我当时没意识到这是投资者或联合创始人必须做的一整套把戏，就是“你总是说服人们辞掉他们愚蠢的工作来为你工作，对吧？”所以这只是他进行的一千次对话中的一次。但我的哥们真的很想要我，我猜。所以他们让他说，“我知道你从来没有因为工作而有钱。”我说，“嘿，我有健康保险。”他说，“我知道这些都很重要。这样如何？让我们把这变成零风险的机会。这是我个人银行账户的 7 万美元支票。你兑现它，这是零风险。只要辞职。”我应该等到汤上完再说，因为我说，“不，非常感谢你，蒂尔先生。”

The biggest one I think is that, you know, when I was 22 or so, friends of mine were starting a company and they said, "Gary, you're at Microsoft. You're the smartest engineer we know. We're starting this company. You got to come and quit your job." And I was like, "Man, I don't know, man. Whatever you guys are doing. Like, I got a job. Okay, it was so hard to get this job. Like, what are you guys doing?" They finally convinced me to fly down to San Francisco from Seattle and I had dinner with um their benefactor. turned out to be Peter Teal. It was probably within a few months of him writing the $500,000 check to Facebook. And so it's me, my two buddies, Joe Londale and Stefan Cohen, and Peter Teal. And Peter is like, "What are you doing at Microsoft, man? You got to quit your job." I didn't realize this was like a whole shtick that you have to do as an investor or as a co-founder is like, "You're always convincing people to quit their stupid jobs to come work for you, right?" And so this was just like one of like a thousand conversations that he had. But my buddies really wanted me, I guess. So they put him up to saying, "I know you never had money having a job." And I was like, "Hey, I have health insurance." He's like, "I know all that's really important. How about this? Let's make this a zerorisk opportunity for you. Here's a check for $70,000 from my personal bank account. You cash it and this is zero risk. Just quit your job." And I should have waited until after the soup course because we were I said, "No, thank you very much, Mr. Teal."

我明年可能就能升到 60 级了。我以为我会升职，也确实升了，但那家公司后来成了 Palantir，而我拒绝了大概，我也不知道，25 亿到 40 亿美元的期权。

I might make it to level 60 next year. I thought I was going to get promoted and I did get promoted, but that company turned out to be Palantir and I turned down something like, I don't know, two and a half to four billion.

太疯狂了。

Crazy.

别为我感到遗憾。我是说，那没什么。而且我后来也以第 10 号员工的身份加入了。但我经常回想那个特别的时刻，我当时觉得，我不相信自己，不相信我的技能有价值。我觉得在微软有一份工作就像是一份礼物。

Don't feel bad for me. I mean, it was fine. And I also ended up joining as employee number 10. But I think a lot about that particular moment where I was like, I didn't believe in myself. I didn't believe that my skills were valuable. I thought that having a job at Microsoft was like a gift.

嗯。

Yeah.

那不仅仅是安稳。而是像，天哪，这是一份礼物。就像社会给了我一份工作。我有一个使命。

It was not just safe. It was like, oh my god, this is a gift. Like society has given me a job. I have a vocation.

你所有的力量和自主权都在这些庞大的组织里被消耗殆尽。那里很安稳，但你正在失去一些非常非常重要的东西。

All your power and agency is sapped away inside these giant organizations. It's safe, but you're losing something really, really big.

是的。然后另一件事，我觉得更重要的事，是意识到我也不知道如何珍视眼前的东西。这些人是我最好的朋友，我们一起打造产品。我了解他们。他们是我见过的最聪明的人。而且背后有一个人支持，你知道，他当时还不是亿万富翁。他卖掉了 PayPal。他算是那种超级有名、非常成功的创始人，一切都很完美。但我拒绝了。我问自己为什么？那是因为我的一生和我所相信的一切都是由别人为我构建的，现在我意识到，为什么我当时要关注《华尔街日报》的记者？他们离源头更近，但他们并没有连接到源头。像你、像 Paul Graham，或者我们认识的很多人，你们是在潮流之中的。而你在媒体上读到的东西，实际上是那些连接到源头的人六七个月前所做的事情的二三手转述。

Yeah. And then the other thing, the bigger thing I think was realizing I also didn't know how to value what was in front of me. These were my best friends who I built products together with. I knew them. They were some of the smartest people I'd ever met. And it was backed by someone who, you know, he wasn't a billionaire yet. He had sold PayPal. He was sort of this super famous successful founder, but everything was perfect. And I said no. And I asked like why? And it's because my whole life and everything I believed was constructed for me by people who now I realize why was I paying attention to the reporters at the Wall Street Journal at that moment. It's like they are closer to the source but they're not connected to the source. People who are like you or Paul Graham or like a lot of the people we know, they are in the flow. And the thing that you read about from media is actually the second and third hand retellings of stuff that the people connected to the source are doing from six or nine months ago.

所以我当时想，哦，他们在做企业软件，那并不热门。我是说，这种投资或花费一生去追逐热门东西的想法，我肯定这一定让你抓狂。

And so I thought, oh well, they're working on enterprise software that's not hot. I mean, this whole idea of investing or spending your life following what is hot is like I'm sure this must drive you crazy.

我不参与那套。

I don't participate in that.

是的。完全正确。对吧？就像如果你遇到一个人，他们说，“嗯，我有点想做点什么。我想做一首歌，有点像他们刚听到的那种。”我就会说，“谁在乎呢，老兄？你为什么那样？那是在讲别人的故事。”没错。你知道，你的故事是什么？你身上发生了什么？所以如果我能回到 22 岁，更珍视自己，然后也意识到整个世界都是虚构的，你不必从媒体或任何人，甚至你的导师那里接受权威，对吧？

Yeah. Exactly. Right? Like if you met someone and they're like, "Well, I kind of want to make something. I want to make a song that's kind of like whatever they had just heard." I was like, "Who cares, dude? Why are you like that's telling someone else's story?" Exactly. You know, what's your story? What's going on for you? And so if I could just go back to when I was 22 and value myself more and then also realize the whole world is made up and you don't have to take it on authority from media or really anyone even your mentors, right?

就像干掉你的导师，不是字面意义上的，而是在你心里。

Like slay your mentors, not literally, but just like in your mind.

不，他们是好意，但他们分享的是他们自己生活中的经验。

No, they mean well, but they're sharing their experience from their life.

是的。而你走在一条不同的路上。你不在他们那条路上。

Yeah. And you're on a different path. You're not on the same path as them.

实际上，我喜欢在指导别人的时候，或者说我不觉得我在指导。我只是在办公时间与人随意交流，我喜欢当我说，“嘿，这是我看到的。”而他们说，“不不不，你搞错了。”然后我说，“哦，你正在连接到源头。”

Actually, I love it when I'm mentoring someone or like I don't feel like I'm mentoring. I'm just literally just vibing with people in office hours, and I love it when I'm like, "Hey, this is like what I see." And they're like, "No, no, no. You have it wrong." And I'm like, "Oh, you're getting connected to the source."

嗯。

Yeah.

我想主动寻找那些建立自己的连接、构建自己对世界如何运作的最佳世界模型的人。这是全世界最独特、最重要的事情，值得去鼓励。而世界并不鼓励这一点。世界说，“闭嘴。”

I want to actively find people who are making their own connections and building their own world models of how the world works best. And that's the most unique, most important thing to encourage in the entire world. And the world does not encourage that. The world says, "Shut the up."

嗯。

Yeah.

世界就像，“你为什么做这个？”事实上，我经常听到这个。他们告诉我。但我已经学会了，实际上这是我不得不告诉很多连接到源头的创始人的事情，他们在做某事，他们在掀起波澜，做一些让人害怕、让人不安、给他们带来自尊心创伤的事情。对我来说，如果我能做一件像漂流瓶一样的事，我会说，“你超出目标了，兄弟。你超出目标了。继续前进。”

The world is like, "Why are you doing this?" In fact, I hear that all the time. They tell me that. But I have learned actually this is something I have had to tell a lot of founders who are connected to the source and they're doing something, they're making waves and doing things that scare people, upset people, they give them ego wounds. And for me, like if I could do like a message in a bottle, I'm like, "You're over the target, bro. You're over the target. Keep going."

嗯。

Yeah.

就像如果有人生气，你去看，发现那下面什么都没有。他们没有逻辑。没有第一性原理的思考，关于为什么你做的事情是坏的或错的。那只是情绪，他们说，“我讨厌它。”而你找不到更深层的东西。那么你就是超出目标了，兄弟。

Like if you've got someone who is mad and you look at it and there's nothing underneath that. Like they don't have a logic. There's no first principles thinking about why what you're doing is bad or wrong. It's just vibes and they're like, "I hate it." And you can't find something deeper. Then it's you're over the target, man.

重力。Tetragrammaton 是一个播客。Tetragrammaton 是一个网站。Tetragrammaton 是一个完整的知识世界。什么可能属于 Tetragrammaton 的领域？反文化，Tetragrammaton。神圣几何，Tetragrammaton。先锋派，Tetragrammaton。生成艺术，Tetragrammaton。塔罗牌，Tetragrammaton。绝版音乐，Tetragrammaton。生物动力法，Tetragrammaton。平面设计，Tetragrammaton。神话与魔法，Tetragrammaton。冷门电影，Tetragrammaton。海滩文化，Tetragrammaton。秘传讲座，Tetragrammaton。离网生活，Tetragrammaton。另类灵性，Tetragrammaton。精美物品的经典，Tetragrammaton。肌肉车，Tetragrammaton。新时代的古老智慧。进入后，体验当日的艺术作品。深呼吸，看看你被吸引到哪里。tetetrogram.com

Gravity. Tetragrammaton is a podcast. Tetragrammaton is a website. Tetragrammaton is a whole world of knowledge. What may fall within the sphere of Tetragrammaton? Counterculture, Tetragrammaton. Sacred geometry, Tetragrammaton. The avant-garde, Tetragrammaton. Generative art, Tetragrammaton. The tarot, Tetragrammaton. Out of print music, Tetragrammaton. Biodynamics, Tetragrammaton. Graphic design, Tetragrammaton. Mythology and magic, Tetragrammaton. Obscure film, Tetragrammaton. Beach culture, Tetragrammaton. Esoteric lectures, Tetragrammaton. Off the grid living, Tetragrammaton. Alt spirituality, Tetragrammaton. The canon of fine objects, Tetragrammaton. Muscle cars, Tetragrammaton. Ancient wisdom for a new age. Upon entering, experience the artwork of the day. Take a breath and see where you are drawn. tetetrogram.com