主题综述

伟大创始人的画像 · What Makes Great Founders

主题综述

更新日志

主流共识

不同视角之间,有三条几乎所有人都点头的特质。

第一点:近乎偏执的专注与原始驱动力——"焦点是对好主意说不"

David Senra 研究了 400+ 创始人,提炼出的第一条就是专注的定义:

"Focus is saying no to a good idea that you really want to do because it distracts you from a great idea."
「焦点,是对一个你真的很想做的好主意说不——因为它会让你从一个伟大的主意上分心。」

Marc Andreessen 把同一种东西描述成原始驱动力,而非"改变世界"的愿景——而且他强调,真到了至暗时刻,撑住你的只有这种最原始的理由:

"A lot of people express that as like ambition, you know, ambition to change the world, improve the world, humanity. … I think those founders are great. … But I do think there's a more fundamental ambition, which is I want to build something of my own. … You need some reason to get out of bed in the morning. … You have to have like a primal reason when things are really, really bad, when the shit really hits the fan."
「很多人把它表达成'改变世界、改善人类'的野心。……我觉得这些创始人很棒。但我认为有一种更根本的野心——我想造一个属于我自己的东西。……你需要一个早上爬起来的理由。……当事情糟糕透顶、当狗屎砸到风扇上时,你必须有一个最原始的理由。」

这一条在新语料里从一个完全不同的行当得到印证——设计师出身的 Paul Bakaus 谈"为什么尊敬 Jobs",给出的理由不是审美,是编辑式的说不:

"I actually really greatly respect people like Steve Jobs because Steve was one of the greatest editors in the world, right? He said no to so many things as opposed to saying yes to so many things."
「我真的非常尊敬 Steve Jobs 这样的人,因为 Steve 是世界上最伟大的编辑之一,对吧?他对太多事情说了'不',而不是对太多事情说'是'。」
Paul Bakaus (Impeccable) · What Happens to Design After AI?

第二点:独立思考——不被外部影响、敢于面对真相

Ben Horowitz 把它放在"问尖锐问题"上:

"I think really good companies, the very, very, very, very best companies tend to have founders and CEOs who ask pretty aggressive questions."
「我认为真正好的公司——最最最好的那些——往往有那种问非常尖锐问题的创始人和 CEO。」

Josh Kushner 把它放在"不被赞美扭曲"上:

"External validation can distort even the most disciplined minds. And just as we have ignored critics in the past, we guard ourselves against the lure of praise."
「外部认可会扭曲哪怕最自律的头脑。正如我们过去忽略批评,我们也要警惕赞美的诱惑。」

新加入的 Founders Fund 一篇把"面对真相"翻译成了组织机制——在 FF,不肯拿自己声誉去跟 Peter Thiel 对赌的人,做不成任何一笔交易:

"If you're not personally super high conviction, I really want to do this deal. I'm going to fight for it. I'm going to be willing to argue. I'm going to be willing to put my reputation on the line, disagreeing with Peter. If you're not willing to do that, you're just never going to get a deal done."
「如果你没有个人层面的极高信念——'我真的想做这笔交易,我会为它争,我愿意辩论,我愿意押上自己的声誉、跟 Peter 唱反调'——如果你不愿意这么做,你就永远做不成一笔交易。」
Trae Stephens (Founders Fund) · Uncapped #53 | Trae and Delian from Founders Fund

配套的还有一个可量化的机制:FF 的"1% 条款"——任何一笔投资,按美元额的 1% 留给经手人自掏腰包跟投。用真金白银测量信念,而不是听会议室里的雄辩。

第三点:没有公式——不能照搬别人的模板

Senra 这一条说得最重,且直接点名 Jobs——他正和 Daniel Ek 合写一本"创始人原型"的书,起因就是这个陷阱:

"When he started Spotify 20 years ago, everybody tried to imitate Steve Jobs. The founders today, who are they gonna imitate?"
「他 20 年前创办 Spotify 时,所有人都在模仿 Steve Jobs。那今天的创始人,又该去模仿谁?」
David Senra(转述 Daniel Ek)· What David Senra Learned Studying 400+ Founders

Horowitz 没有"单一模具",但他给了一个所有模具共享的底色——对真相的态度:

"If you're running away from the truth to preserve feelings, that's a very dangerous thing in a tech company. And the corollary to that is it's really important that like bad news travels fast. … You need that bluntness. Andy Grove used to call it constructive confrontation."
「如果你为了照顾情绪而逃避真相,这在科技公司里非常危险。它的推论是——坏消息必须传得快。……你需要那种直率。Andy Grove 当年管它叫'建设性对抗'。」

"无公式"这条连管理研究员的人也认。OpenAI 首席研究官 Mark Chen 被问"什么样的研究者能产生大影响",答案同样是没有单一模具:

"Not every researcher is the same. I think there's a lot of different types of impact. There are the people who just take an idea, it's very clear, and they'll just implement it before anyone else. There are also the people who just come up with the crazy Almost too crazy but somehow not that crazy and they really convince you in a different way of seeing the world or another completely different type of project."
「不是每个研究者都一样。产生影响的类型有很多种。有的人拿到一个非常明确的想法,能比任何人都先把它实现;也有的人提出那种近乎太疯狂、但又没那么疯狂的想法,用一种完全不同的方式说服你看世界,或者做一类完全不同的项目。」

分歧在哪

共识止于"专注 + 独立 + 无公式"。再往细看,每个观察者的镜头都不一样——他们盯着不同的特质,而这些镜头之间有真实的张力。下面是九种视角(外加一条来自研究组织的旁注),以及它们分道扬镳的地方。

视角一 · David Senra(研究者):"founder-problem fit"——不是天赋,是匹配

Senra 最反直觉的判断:决定成败的不是某个特质,而是人和问题的契合度——而且逐字稿里这其实是 Daniel Ek 的框架,比 PMF、甚至比 founder-industry fit 更根本:

"Daniel's point is more important than product-market fit or even founder-industry fit. He believes the most important is founder-problem fit. Think about Demis Hassabis from DeepMind. There's one great company he had in him. It was DeepMind. He was put on this planet to do what he is doing."
「Daniel 的观点是:比产品-市场契合、甚至创始人-行业契合更重要的,是 founder-problem fit。想想 DeepMind 的 Demis Hassabis——他体内只有一家伟大公司,就是 DeepMind。他生到这世上,就是为了做他正在做的这件事。」
David Senra(转述 Daniel Ek)· What David Senra Learned Studying 400+ Founders

这句"他体内只有一家伟大公司"也顺带回应了"团队 vs 单一力量"之争——在 Senra 眼里,最持久的公司往往由一个与问题死死咬合的单一力量驱动。

但 Senra 自己又给团队留了一扇门——伟大团队能救平庸主意,平庸团队会毁伟大主意:

"If you give a great idea to a mediocre team, they will ruin it. If you give a mediocre idea to a great team, they will either fix it or throw it out and create something new."
「把伟大的主意交给平庸团队,他们会毁了它。把平庸的主意交给伟大团队,他们要么修好它、要么扔掉重造。」

这一轮刷新,fit 框架第一次拿到了准自然实验的支持。 Uncapped #53 里,Delian Asparouhov 复盘他的导师 Keith Rabois 离开 Founders Fund 回到 Khosla Ventures——同一个人,两种机构文化,两种结果:

"I do think his work style, personality, how he likes to do things, Actually just does fit into how KV's culture, et cetera, like operates a lot better than I think it did like at FF."
「我确实认为他的工作风格、个性、做事方式,就是更契合 KV 的文化和运作方式——比在 FF 时契合得多。」
Delian Asparouhov (Founders Fund) · Uncapped #53 | Trae and Delian from Founders Fund

主持人 Jack Altman 顺势把它推广成一般命题——几乎是 Ek 框架的逐字重述:

"I think like we probably all are great in one environment and terrible in another. It's probably, you know, there's people who are being an amazing founder for one thing, horrific for the next. I think so much of like life is like getting yourself into the place that you're supposed to be in."
「我们大概都是在一种环境里伟大、在另一种环境里糟糕。有人做某件事时是惊人的创始人,换一件事就一塌糊涂。人生的很大一部分,就是把自己放进你本该在的位置。」

同一个 Rabois,在 FF 是摩擦源、回到 KV 如鱼得水——这正是 fit 框架此前最缺的东西:同一个人、两个语境、相反结果的对照。

视角二 · Marc Andreessen(投资人):"撞墙留洞"——relentlessness 高于一切

Andreessen 的镜头盯的是不可阻挡性,他的比喻是这批里最生动的:

"I want the founder who leaves a founder-shaped hole in any brick wall that he runs into, like a cartoon character."
「我想要那种创始人——像卡通人物一样,撞上任何砖墙都会留下一个'创始人形状的洞'。」

他的核心特质清单是 IQ + 勇气 + 野心,而他对"特殊的人"的态度近乎绝对——无条件下注:

"When you have special people, you should back them almost without consideration of other factors. And when you don't, you shouldn't."
「当你遇到特殊的人,你应该几乎不考虑其他因素地支持他们。遇不到,就不该支持。」

这跟 Senra 的"fit"是直接冲突的——Andreessen 信的是可识别的特质(特殊的人,无视语境去押),Senra 信的是语境匹配(同一个人换个问题就不行)。两人对"伟大创始人是天生可识别的、还是和问题匹配出来的"给了相反答案。

而这一轮里,Andreessen 清单里的"高 IQ"还挨了一记来自同行的正面撞击——Thiel 认为 IQ 过了 130 反而"权衡太多"(见视角八)。这是七个月来特质清单内部第一次出现"上限"论:特质未必是越多越好的标量。

Andreessen 还有一条关于心态的:

"Life just gets a lot simpler if you just assume everything is your own fault."
「如果你假设一切都是你自己的错,生活会简单很多。」

视角三 · Tarek Mansour(创始人本人):"outlier imbalance"——异常本身是能力

Mansour 从创始人内部给了一个独特的镜头——你需要某种异常的不平衡才能突破市场天然的排斥:

"I think if you want to achieve outlier results, you need some sort of outlier, I'm going to call it imbalance. You need some sort of outlier, unusual thing that is part of your history, something."
「我认为要做出异于常人的结果,你需要某种异常——我称之为'不平衡'。你需要某种异常的、不寻常的、属于你个人历史的东西。」
"A startup is intrinsically something that the world does not want to exist."
「初创公司本质上是这个世界不希望存在的东西。」

他还给"天真"翻了案——聪明的天真是超能力:

"If you're very smart and you're naïve, the naïveté is not coming from a lack of understanding or a lack of deep thinking; it's coming from a lack of experience."
「如果你很聪明又很天真,这种天真不是因为缺乏理解或缺乏深度思考——而是因为缺乏经验。」

这条跟 Andreessen 的"high IQ"形成一个有意思的角度差——Andreessen 要高智商,Mansour 要"高智商 + 天真",多了一个"未被经验污染"的维度。

视角四 · Josh Kushner(投资人):"脆弱即自信"——和 Andreessen 的撞墙相反

Kushner 的镜头跟 Andreessen 几乎对立——他押的不是不可阻挡的强势,而是承认无知的能力:

"Many people view vulnerability as a sign of weakness, but if you are willing to say, I don't know the answer, that is actually the most extraordinary sign of confidence."
「很多人把脆弱当作软弱,但如果你愿意说'我不知道答案',那其实是最非凡的自信标志。」

但 Kushner 也给了一个跟 Andreessen "渴望"一致的判断——选最渴望的那个:

"If you have to pick between the most educated person, the most experienced person, or the person who wants it the most, you always pick the person who wants it the most."
「如果必须在受教育最高、经验最丰富、最渴望的人之间选,永远选最渴望的那个。」

Andreessen 的"撞墙留洞"和 Kushner 的"敢说我不知道"——两个顶级投资人盯着几乎相反的特质,却都自信地说"这就是伟大创始人的标志"。这是这个题目里最锋利的一处分歧。

视角五 · Garry Tan(YC,看过最多创始人):"多听十倍" + spiky advice

Tan 的镜头是反 founder-mode 邪教的——伟大创始人不是少听,是多听后自己判断:

"The cargo cult around great founders is, oh, let me listen less. Actually, you have to listen 10 times more than the other people who don't listen."
「围绕'伟大创始人'的邪教式模仿是'让我少听点'。恰恰相反——你必须比那些不听的人多听十倍。」

但他对"伟大创始人怎么用建议"的判断又是独立的——好创始人能接住尖锐建议、也能果断拒绝坏建议:

"What's really valuable is to have people around who have seen enough and are intuitive enough about ideas and technology and teams, that they can give you even spiky advice, sometimes like the worst advice. But good founders are like, oh, that's bad advice. I'm just not going to do that."
「真正有价值的是,身边有人见识足够多、对想法、技术和团队有足够直觉,能给你尖锐的建议、有时是最糟的建议。但好的创始人会说'哦,那是个坏建议,我不会那么做'。」

视角六 · Alfred Lin(Sequoia):"maniacally aligned on mission"——对齐力

Lin 的镜头不在创始人个人特质,在他让整个公司对齐的能力:

They just are maniacally good at making sure that their companies are completely aligned on the mission and where they are and how they get there.
他们极度擅长一件事——确保整个公司在使命、现状、路径上完全对齐。
— Alfred Lin (Sequoia) · Alfred Lin - The Future According to Sequoia

这是这些视角里唯一一个把"伟大"定义在组织层面而非个人特质的——其他人都在描述创始人是什么样的人,Lin 在描述创始人能让组织变成什么样。

视角七 · Keith Rabois(PayPal Mafia / 投资人):"罕见组合"——不是某个特质,是不该同时出现的特质同时出现

Rabois 的镜头跟所有人都不一样——他不找单一特质,找通常不会在同一个人身上共存的能力组合。"top 15 个基点的人造就了对世界和风投回报都重要的公司"这个说法,其实是主持人 Jack 开场时对 Rabois 此前公开言论的转述;Rabois 本人在本集里给出的门槛更极端——超能力要进到世界前 1–10 个基点(0.01%–0.1%):

"Every founder that really succeeds has a superpower, and that superpower is something like they're in the top one to ten basis points in the world on some trait."
「每个真正成功的创始人,都有一种超能力,这种超能力是他们在某个行业中位居世界前 0.01% 到 0.1%。」
Keith Rabois · Comparative Advantages

而识别他们的方式,是看罕见的能力叠加——他用 Reid Hoffman 看准 Max Levchin、和 Jack Dorsey 举例:

"Reid exactly nailed Max's combination of superpowers. It's those two traits don't go together. Jack Dorsey, another very successful founder. Jack actually can do three. He's actually a pretty good technologist. He's a world-class design and has world-class design taste."
「Reid 精准地看中了 Max(Levchin)那种'超能力的组合'——那两种特质本来不会同时出现。Jack Dorsey 是另一个非常成功的创始人——Jack 能做到三样:他是个不错的技术者,又是世界级的设计、有世界级的设计品味。」
Keith Rabois · Comparative Advantages

而且他对自己这门手艺的诚实,是给整组"特质清单"打了个折扣——连他这种顶级猎手,命中率也就 Ted Williams 级别:

"You have to be contrarian and right … an early stage investor call 40%. 40% is pretty much Hall of Fame, you know, Ted Williams kind of stuff, if you can be that accurate."
「你得既反共识、又正确。早期投资人——算 40% 命中率吧。能到 40% 基本就是名人堂、Ted Williams 那个级别了。」
Keith Rabois · Comparative Advantages

Rabois 这条把整个题目往上拔了一层:如果连最会看人的人,事前命中率也只有 ~40%,那"伟大创始人画像"作为一个事前可用的清单,本身就是可疑的。

这一轮,"罕见组合"从两个不相干的行当各拿到一票。 设计界的 John Maeda 被问"为什么品味在会议室里总输给速度",答案落在领导者的罕见组合上:

"I mean, it's all about the leader, right? You mentioned Steve Jobs, a unique combination of design, business, technical sense. That's very rare."
「一切都取决于领导者,对吧?你提到 Steve Jobs——设计感、商业感、技术感的独特组合。那非常罕见。」
John Maeda (Microsoft) · What Happens to Design After AI?

硬科技那头,Trae Stephens 给出同一结论的否定式表述——只有一半组合的创始人,撑不起公司:

"it's not sufficient to have a really strong technical founder. You have to have somebody that's equally good on the business side. Otherwise, It's just not going to work."
「只有一个非常强的技术创始人是不够的。你必须有一个在商业侧同样出色的人。否则就是行不通。」
Trae Stephens (Founders Fund) · Uncapped #53 | Trae and Delian from Founders Fund

Maeda 还给这种组合起了名字——conviction,并把它和赌注的目标绑在一起:

"It's that combination in a leader when they have conviction. It is a bet, but it's a bet aimed at the global maximum. Not the local maximum."
「就是领导者身上的那种组合——当他们有信念的时候。它是一注赌注,但赌的是全局最大值,不是局部最大值。」
John Maeda (Microsoft) · What Happens to Design After AI?

视角八 · Trae Stephens & Delian Asparouhov(Founders Fund):Thiel 的镜头——IQ 有最优值、方向性正确高于精确正确

新增视角。Founders Fund 是"押独行创始人"名声最响的机构,这期访谈难得地把 Thiel 的识人逻辑、以及 FF 对自己判断力的诚实复盘同时摆上了桌面。

第一条:IQ 不是越高越好。 Trae 转述 Thiel(语境是聊"希望孩子有什么",但 Trae 明说这就是 FF 招人的逻辑——招的是能在辩论里站住的高智商,同时接受它的代价):

"I think like 130 is like roughly the right IQ because anything over 130, there are too many tradeoffs."
「我觉得 130 大概就是'恰好的智商'——因为超过 130,权衡就太多了。」
Peter Thiel(Trae Stephens 转述)· Uncapped #53 | Trae and Delian from Founders Fund

这条直接顶撞视角二:Andreessen 的清单里 IQ 是单调加分项;Thiel 说它有最优值、过量要付别的代价。它同时给 Rabois 的"罕见组合"补了一个机制解释——如果每种极端特质都自带 tradeoff,那"不该共存的特质共存"之所以罕见,正是因为大多数人付不起多重极端的叠加代价。

第二条:方向性正确 > 精确正确。 Trae 观察 Thiel 二十年,提炼出的不是"独立思考",而是一种更激进的操作:

"It is less important to be exactly true than it is to be directionally true because it's really hard to shift the overton window."
「'精确正确'不如'方向性正确'重要——因为把奥弗顿之窗挪动一点点,实在太难了。」
"No matter how annoyed you are with him, he's always right. On some timeline, he's always right."
「不管你被他惹得多恼火,他总是对的。在某个时间尺度上,他总是对的。」
Trae Stephens(谈 Peter Thiel)· Uncapped #53 | Trae and Delian from Founders Fund

这给"独立思考"共识加了一个此前没人说过的细节:伟大创始人不只是不从众,而是故意超调——把话说到不合理的程度,只为把整个组织或市场的可讨论范围拉过来一点。Trae 说自己在 Anduril 高管会上也刻意扮演这个"完全不讲理"的角色。

第三条:事前识别的诚实账本。 FF 以"论题驱动"著称,但 Trae 这次把账本翻开了:

"If you think about where most of those gains came from, it was actually Pavement Mafia. It wasn't like a thesis that turned into great returns. It was a network that turned into great returns."
「如果你想想那些收益大多来自哪里——其实是 PayPal Mafia(转写讹作 Pavement Mafia)。不是论题变成了回报,是网络变成了回报。」

甚至网络中心的案子,事前也不显然——若不是 Luke Nosek 押上职业生涯,"I don't know that Founders Fund would have done the early SpaceX investments"(我不确定 FF 会做早期 SpaceX 投资);而 Anduril:

"When I pitched the Founders Fund team on Anduril, they were basically like, who's running this company? It's not you and it's not Palmer. So you need to come up with a better solution."
「我把 Anduril 讲给 Founders Fund 团队时,他们的反应基本是:这公司谁来管?不是你、也不是 Palmer——你得想个更好的方案。」

连身处网络中心、事后回报最大的两家公司,事前都近乎被自家否掉——Rabois 那条"~40% 命中率就是名人堂",又多了一份来自最强机构内部的证词。

第四条:画像不止于特质,还有任务选择。 Trae 把镜头从"你是什么样的人"推到"你选了什么任务"。他的例子是美国没有前沿芯片制造:他判断"One of those outcomes is wildly probable, 90 plus percent in the next 10 years"(未来十年台湾现状改变的概率 90% 以上——一条挂在时间线上的可证伪预测),重建产能需要"Elon plus tens of billions of dollars"(一个 Elon 加上数百亿美元),而硅谷最有能力的人却在去开风投基金:

"We really need them to be founders. We need them to be heroes. It's a big ask, but like, I'm shocked that no one has taken it more seriously because it feels like the most important thing that people could work on right now."
「我们真的需要他们去当创始人。我们需要他们当英雄。这是个很高的要求,但我很震惊没人更认真地对待它——因为这感觉是此刻人们能做的最重要的事。」

这和 Senra 的 founder-problem fit 是同一枚硬币的两面:fit 说"人要配上问题",Trae 说"最重要的问题正等不到配得上的人"。

视角九 · Ev Randle(Benchmark,前 Founders Fund):"great founders are always in style"——范式作废时,创始人是唯一不变量

新增视角。Ev Randle 值得单列,是因为他刚好在两家哲学相反的机构都干过(FF 的真相寻求 + 论题网络,Benchmark 的平等合伙 + 只投人),而他最终落点在"人"这一侧。先看他对 AI 时代的判断——软件投资的全部黄金法则已经作废:

"We used to have golden rules or like North Star metrics in investing. When you think about software, there was like sort of five or six things that mattered. And it was very legible and it was very spreadsheetable. And now all of the golden rules of the past that defined like the spreadsheet investing era, We're all gone in the new AI paradigm."
「我们过去有投资的黄金法则、北极星指标。说到软件,要紧的就那五六件事,非常易读、非常能装进电子表格。而现在,定义了'表格投资时代'的那些过去的黄金法则,在新的 AI 范式里全没了。」
Ev Randle (Benchmark) · Benchmark's AI Bets

正因为商业模式层面全是流沙,他把不变量押在创始人身上:

"the benefit that we have is that like great founders are always in style. Whereas like these business models can go in and out of style and things become popular and less popular."
「我们的优势在于:伟大创始人永远不过时。而这些商业模式会一阵流行、一阵过时,东西红了又凉。」
Ev Randle (Benchmark) · Benchmark's AI Bets
"we don't invest in categories, we invest in people."
「我们不投类别,我们投人。」
Ev Randle (Benchmark) · Benchmark's AI Bets

他给的证据有点反讽:入职 Benchmark 前,他以为这家的 AI 组合是精妙的主题布局(垂直 AI、水平 AI、开发者、太空数据中心各占一格);入职后发现对类别"literally no thought"——全部是从创始人出发(比如投 StarCloud 在 Elon 公开看多轨道数据中心之前几周敲定)。而这些创始人会自己走向大类别:

"great entrepreneurs are always in style. And like they always pull rabbits out of hats and they end up being magnetized and, you know, drawn towards where are these really massive categories and they often end up pioneering them."
「伟大的企业家永远不过时。他们总能从帽子里变出兔子,总会像被磁吸一样走向那些真正巨大的类别——而且常常就是他们开创了那些类别。」
Ev Randle (Benchmark) · Benchmark's AI Bets

表面看,这是给视角二(Andreessen"押特殊的人")投的一票,但理由不同:Andreessen 断言特殊的人可识别;Ev 的理由更像贝叶斯式的——当所有商业信号都失效时,人是唯一还能下注的稳定变量。"押人"可以在画像不可事前识别的情况下依然是理性的。

他还留下一个对画像本身的扰动项——AI 流动性冲击将改写"谁去创业":Anthropic 3800 亿美元估值的一轮若在 1.5 万亿市值上兑现流动性,"it would return 35 times that of the Snowflake pre-IPO round"(等于一轮之内装下 35 个 Snowflake pre-IPO),随之而来的问题是这批实验室员工"Do they start new companies? Do they stay at these labs?"(他们会去创业,还是留在实验室?):

"I just don't know if people really understand and are ready for the impacts that this much liquidity could have on the ecosystem."
「我不确定人们真的理解、并且准备好了这么大规模的流动性会对整个生态造成的冲击。」
Ev Randle (Benchmark) · Benchmark's AI Bets

旁注 · Mark Chen(OpenAI 研究负责人):换个物种,同样的识别难题

Mark Chen 管理的是研究员不是创始人,与本题只弱相关,但他被问到"面试时能否识别出未来的顶级研究者",答案与 Rabois 的 40% 同构——事前识别难,事后 6–12 个月才显形:

"It's a hard problem before someone comes to OpenAI. I think that's genuinely true."
「在一个人来 OpenAI 之前,这是个难题。我认为这是真的。」
"But it is really hard to tell out of the gate. Usually in, you know, let's say six months to a year, it's pretty clear who has the strongest trajectory and who's going to make a lot of impact."
「一开始真的很难判断。通常要过六个月到一年,谁的轨迹最强、谁会产生大影响,才会变得非常清楚。」

他也给"高风险信念"的价值投了一票——"we take a lot of high-risk bets. I think it's what's allowed us to stay at the frontier so consistently over time"(我们下很多高风险赌注,这正是我们能持续留在前沿的原因);推理范式(o1)当年在公司内部也要顶着"why do something when you have a machine that works?"(既然手上的机器好用,为什么还要另起炉灶?)的合理质疑,靠 Jakob、Ilya 等少数人的信念硬推起来。这与 Maeda 的 conviction、FF 的 1% 条款是同一主题的三种形态:伟大组织都在发明测量信念的机制。

都没说透的

"very often somebody would walk into the room with Steve and Steve would just test them and say, like, you know, I think that's a stupid idea. Even though he might think that's a really great idea, but just to test them, right?"
「经常有人走进房间见 Steve,Steve 会故意测试他们:'我觉得那是个蠢主意'——哪怕他其实觉得那是个好主意,就为了测试。」
"And once he got that signal, once he got that signal of conviction, he often would say, yes, let's do it."
「而一旦他收到那个信号——那个信念的信号——他往往就会说:好,我们干。」
Paul Bakaus (Impeccable) · What Happens to Design After AI?

顶回去的勇气(撞墙)和接受测试而不玻璃心(谦逊),在这个轶事里都是信念的观测值——测的是同一个底层量。但这仍是轶事,不是整合框架。

"I think both at OpenAI and at other labs, you're starting to see a lot of the work become mostly orchestration focused, right? Like the researchers coming up with ideas. And the model's great enough to do the implementation execution by itself."
「无论在 OpenAI 还是其他实验室,你都开始看到大量工作变成以编排为主——研究员出想法,而模型已经好到能自己完成实现与执行。」
"It's going to be hard to teach the models good taste. We noticed that."
「教会模型好品味会很难。我们注意到了这一点。」

Mark Chen 还给了一条可证伪的背景判断:缩放定律已在近 10 个数量级上成立、他认为没有理由不继续成立——如果他对,执行侧的贬值只会加速。但"哪些传统创始人特质被 AI 归零了"的清单,仍然没人给出。

我的看法

判断(不是事实):把九个视角并起来看,我仍然认为——"伟大创始人有哪些特质"这个问题本身被问错了。这些清单(IQ、勇气、专注、脆弱、不可阻挡、天真、对齐、组合、方向感)之所以彼此冲突却各自自信,是因为它们都是事后从成功者身上反推的,而成功是多条不同路径的并集。这一轮刷新让我在两处加码、一处减码:

1. 加码 fit。 Rabois 在 FF/KV 的对照是 fit 框架第一份准实验证据——同一个人、两个语境、相反结果;Jack Altman 的总结("我们都是在一种环境里伟大、换一种环境就糟糕")几乎是 Ek 框架的逐字重述。fit 从 Senra/Ek 单一来源,变成了三个独立来源。 2. 加码"组合 + 权衡"。 Thiel 的 IQ-130 上限给 Rabois 的罕见组合补上了机制:每种极端特质都自带代价,所以多重极端天然罕见。Maeda 的 Jobs(设计×商业×技术)与 Trae 的"技术强还不够"从设计与硬科技两个行当分别复核了组合论。特质不是可加总的标量清单,是有代价约束的组合优化——这能同时解释为什么单项清单彼此打架、以及为什么伟大如此稀少。 3. 减码"画像可疑所以别用"。 Ev Randle 给了一个我此前没认真对待的反面理由:当范式剧变把所有商业信号作废时,"押人"未必因为人可识别,而是因为人是唯一剩下的稳定变量。也就是说,"画像"作为决策工具,可以在"不可事前识别"成立的同时依然是理性的下注对象。这削弱了我从 Rabois 40% 推出的"清单无用"结论——清单可以既不准、又是可得信息里最好的。

把握程度:中等偏上(上一轮为中等)。上调的理由是 fit 与组合论各自获得了独立来源的支撑,且 FF 的自我复盘(回报来自网络而非论题、Anduril 险被否)压低了"特质可事前识别"的对立面。仍然最弱的一环没变:fit 无法事前打分——如果不能在创始人和问题相遇的早期评估契合度,它对决策者就仍然只是事后叙事。这个实操缺陷这一轮没有得到任何新解。

还想知道什么

取材

语料里还有多个高相关、本轮未纳入的投资人视角(Pat Grady、Shardul Shah、Shaun Maguire 谈 Elon、Sarah Guo、Mamoon Hamid 等)——后续刷新可继续加视角,把"多视角"这条做得更厚。