Patrick Collison discusses the value of learning from first principles, why he still writes himself, and his experience dropping out of college twice to start companies.
要点 · TL;DR
即使有先进的 AI，认知 L1 缓存知识在解决问题时仍然更快、更有价值。 Even with advanced AI, cognitive L1 cache knowledge remains faster and more valuable for problem-solving.
写作和人际沟通是模型仍缺乏的基本技能，因此亲自去做很重要。 Writing and interpersonal communication are fundamental skills that models still lack, so doing them yourself matters.
基于具体、真实的客户问题创办公司对成功至关重要。 Starting a company grounded in a concrete, real customer problem is crucial for success.
核心观点 · Key points
即便有了先进的AI，认知一级缓存中的知识在解决问题时仍然更快、更有价值。 Even with advanced AI, cognitive L1 cache knowledge remains faster and more valuable for problem-solving.
写作和人际沟通是模型仍欠缺的基本技能，因此亲自去做很重要。 Writing and interpersonal communication are fundamental skills that models still lack, so doing them yourself matters.
创办一家基于具体、真实客户问题的公司对成功至关重要。 Starting a company grounded in a concrete, real customer problem is crucial for success.
从一开始就拥有生产用户，即使在公开发布之前，也能提供必要的依据和反馈。 Having production users from the very beginning provides essential grounding and feedback, even before a public launch.
在AI时代，初创公司应考虑更宏大、更发散的起点，而不是狭窄的利基市场。 In the AI era, startups should consider more ambitious and divergent starting points rather than narrow niches.
Stripe的数据显示，现在是创办企业的最佳时机，更多初创公司正更快地取得成功。 Stripe data shows it's a better time than ever to start a business, with more startups succeeding faster.
反共识 · Contrarian takes
辍学创业并不冒险；成本微乎其微，之后也没人在意。 Dropping out of college to start a company is not risky; the costs are minimal and nobody cares later.
如果你现在不创业就会被困在永久下层阶级的担忧被夸大了；机会是充足的。 The fear of being trapped in a permanent underclass if you don't start a company now is overblown; opportunities are abundant.
等待两年才发布Stripe是正确的选择，与“尽早发布”的信条相反，因为他们很早就有了生产用户。 Waiting two years to launch Stripe was the right call, contrary to the 'launch early' mantra, because they had production users early.
精益创业方法正变得不那么有效；在AI时代，反精益策略更为成功。 The lean startup approach is becoming less effective; anti-lean strategies are more successful in the AI era.
AI不会导致霸权集中化；相反，它将创造一个更加去中心化、多赢家的世界。 AI will not lead to hegemonic centralization; instead, it will create a more decentralized world with many winners.
即使有强大的AI，自己写文章仍然更胜一筹；模型生成的文章并不引人入胜。 Even with powerful AI, writing essays yourself is still superior; model-generated writing is not compelling.
本期章节 · Chapters（共 13）
开场与初次会面Introduction and Early Meeting
从基本原理学习的价值Value of Learning from First Principles
个人选择与写作Personal Choices and Writing
辍学创办StripeDropping Out and Starting Stripe
关于辍学与紧迫感On Dropping Out and Urgency
Stripe为何两年才上线Why Stripe took two years to launch
Schlepp盲点与智力回报Schlepp blindness and intellectual rewards at Stripe
考虑成功与长期承诺Considering Success and Long-term Commitment