Tobi Lütke 的领导手册:第一性原理(First principles)、无限游戏(infinite games)与最大化人类潜能
在多数商业领袖追求温和与周全时,Shopify创始人Tobi Lütke选择了一条迥异的道路。本文深入剖析了其独特的领导哲学,揭示了驱动他打破现状的底层逻辑:对现状的不满与坚定的乐观主义。Lütke主张以百年愿景倒推当下战略,并通过被称为“Tobi龙卷风”的极致敏捷迭代,果断斩断无效努力,拒绝在错误方向上浪费团队的生命。文章最深刻的洞见在于其人才观——他将“最大化人类潜能”视为公司的核心产品。通过直接甚至严苛的反馈,他不断挑战团队边界,试图唤醒人们未曾察觉的巨大潜力。这不仅是一份管理手册,更是一次关于超越平庸、重塑自我认知的思想对谈。
Tobi Lütke 的领导手册:第一性原理(First principles)、无限游戏(infinite games)与最大化人类潜能
对现状的不满与进步的动力
**Tobi Lütke:**你的播客是由建造者为其他建造者开设的播客。我认为人们能问建造者的最有趣的问题就是,你的能量来源是什么?我的能量来源是对现状的不满。现在有太多书探讨技术将如何导致反乌托邦。就像真正思考过这件事的人,不会愿意出生在距今20年前的世界里。我认为今天就是未来的反乌托邦。我们有义务尝试构建那些能引领……迈向进步的产品。
**Lenny Rachitsky:**我见过几句沿着这些思路的名言,描述了你思考这些问题的方式。“如果大多数人都以某种方式做事,我默认就不想那样做。”
**Tobi Lütke:**世界上存在一种美学,那就是商界人士穿着西装打着领带,他们说话比我高深得多,通常还没有口音。他们通常拿着一根教鞭,戏剧性地指着身后的图表。这种美学与卓越表现有多大重叠?悲观主义听起来极其深奥。乐观主义听起来总是很蠢,或者至少很天真。在商业领域中,最强大的不可量化的事物是乐趣和愉悦。
百年愿景与逆向思考
**Lenny Rachitsky:**我不知道还有哪家公司是这样运作的,创始人对产品需要走向何方有着百年的愿景,并以此倒推。
**Tobi Lütke:**我经常谈到展望未来然后逆向思考,对吧?就像在这个问题上,我们会希望20年前已经做了什么?我们有非常长期的计划。在100年的时间跨度上,你无法谈论某个软件项目,但你可以谈论使命本身,无论哪些事物能在这一特定时间框架内留存80年。创业精神非常珍贵。Shopify 的存在,基本上就是为了让创业变得更普遍。
**Lenny Rachitsky:**你有什么想留给听众的话吗?
**Tobi Lütke:**我真心认为,这个星球上没有一个人哪怕接近了他们的最大潜能。不断提醒人们他们自身的潜力,实际上是一件很棒的事。
节目开场与主题引入
**Lenny Rachitsky:**今天的嘉宾是 Tobi Lütke。Tobi 是一位不需要介绍的人,所以我会说得非常简短,直接进入这场干货满满的对话。如果你喜欢这个播客,别忘了在你最喜欢的播客应用或 YouTube 上订阅和关注,这是避免错过未来剧集的最好方式,也对播客帮助巨大。言归正传,为您请出 Tobi Lütke。Tobi,非常感谢你的到来。欢迎来到播客。
**Tobi Lütke:**很高兴来到这里。我很期待我们的对话。
**Lenny Rachitsky:**我听了你很多其他的访谈,也和你公司里的很多人聊过。我想尝试做一些不一样的事情。在听你分享建议、接受采访以及和你的员工交谈时,基本上有两个主题反反复复地出现。一个是第一性原理思考,另一个是最大化人类潜能。我现在只是先埋下这些种子。我不会直接问这些问题,而是会用我准备的许多问题从侧面切入。我听到 Shopify 的人描述过一种叫“Tobi 龙卷风”的东西。
“Tobi 龙卷风”与快速迭代
**Tobi Lütke:**哦,哇。好的,这是个开局。我喜欢。
**Lenny Rachitsky:**什么是“Tobi 龙卷风”?
**Tobi Lütke:**我想说,“Tobi 龙卷风”是将大量的变革管理、对话、冲突或真实的讨论压缩在极短的时间内。我看到某个东西,它不……它不太好,我就会进行一次对话。我非常快地学到了一些东西,比如,我需要更新我的先验认知,或者,酷,我们换个做法。此时,一个项目可能会被叫停,我们重新聚在一个房间里,然后开始一个产品的新版本。而目前在这个特定项目团队中的每个人,都不再是该项目的成员,但他们成为了以不同方式构建的下一个版本的创始人。这可能会让人感到有些猝不及防。我的意思是,我当然希望如此,人们也是这么告诉我的。这也是他们欣赏这家公司的地方。就像什么是最好的,最终变得非常重要。
**Lenny Rachitsky:**基本上,就是你走进一个聊天室说“嘿,这个”……“我们要结束这个项目”……“我们试试别的”,因为你发现或意识到这是一个糟糕的想法。有些人会抱怨这种做法,比如“哦,Tobi 终结了我们一直在做的东西”。我的视角是,你意识到我们只是在一个行不通的东西上浪费时间,我们不应该这么做。在这方面还有什么要补充的吗?
**Tobi Lütke:**没有。这就是全部。再说一次,一旦我设想某个东西可能不是正确的工作方向,我要么是错的,如果是这样,我理解为什么就超级重要;我要么是对的,如果是这样,让人们在一个成不了的事情上工作就极度不公平。还有第三种情况,我可以忽略它,但那是对我作为 CEO 和创始人职责的违背,我绝对不愿意这么做,所以这不是我认为有效的一条前进道路。我知道那是很多人选择的做法。所以是的,压缩时间很重要。我们在职业生涯中的时间相当有限,对吧?我们的职业生涯没那么长。如果幸运的话,你在行业里有40年时间。大多数人在学校度过的时间更长,然后如果足够幸运的话可能会晚点离开。所以实际可用的时间甚至比这还少。
我想,在职业生涯的尽头,你会想要去做尽可能多让你感到自豪的事情。当你回首时,你会想说,“天哪,我们做出了这个东西,这绝对是对我所关心的使命的一项不可思议的贡献,而且这家公司里充满了和我一样关心它的人。”并且他们也非常自豪能在这里工作。甚至可能很乐意和我一起工作,并且真的很高兴我们能一起在项目上花时间。是的,如果大家都在优化一个可能本不该存在的东西,这一切就不会发生,对吧?因此,我认为这是更好的做法。
**Lenny Rachitsky:**顺便说一下,我很喜欢这种描述方式……把时间压缩以做出决定,而不是专注于用最友善、最温和、最讨喜的方式呈现。当你给人们反馈,说某样东西不是你认为它应该有的样子,或者没有达到它能达到的最好程度时,你的表达通常非常直接,通常也很难听进去。对我来说,这表现出的似乎是你正在试图最大化他们的潜力,你在试图推动他们去做更好的东西。在这方面,你认为这是一种处理反馈的方式吗?
**Tobi Lütke:**我真的、真的、真的认为,这个星球上没有任何一个人接近了他们的最大潜力。我只是觉得每个人都比他们自己认为的要好得多得多得多得多得多。而我们没有在这个水平上表现出来的原因是一系列的想法,也许是某些培养我们技能和手艺的方法尚未被发现,因此我们无法利用它们。这是一种仅仅把注意力集中在相当缺乏野心的事情上的环境,在这一点上,你会陷入与简直世界上所有其他人竞争的境地,因为每个人都没什么野心。
我发现不断提醒人们他们自己的潜力,实际上是一件很棒的事情。而且我有这样的历史:在判断人们的潜力方面,我比他们自己更正确。现在,在某种程度上,这常常注定让我感到失望,对吧,对自己感到失望。顺便说一下,我在这里也是在说我自己。我认为我的潜力远远超过了我所发挥出来的,我讨厌这样,所以我正试图培养我明天需要的技能,并不断挑战自己。我对自己的要求比对任何人都严格。按照某种折现率,我对周围的人也采取同等标准,特别是那些显而易见非常耀眼的人。
潜力即产品
在职业生涯中花很长时间和人们在一起,然后对他们保持高标准,意味着他们经常能完成他们自己都想象不到的事情。对我来说,这是最美妙的事情。坦白说,这是我整个职业生涯的一条贯穿始终的主线,因为这个……我的产品就是如此。我希望我的产品能让人们比他们自己认为的更成功。事实上,让他们对自己在在线商店和业务上所构建的东西变得比最初设定的目标更有野心。
因为类似的事情也发生在我身上,对吧?我曾在某个时候开了一家单板滑雪店。我最初并没有打算打造 Shopify。如果你致力于在下一步跟随你的好奇心,并在选择这些步骤时将学习最大化,它会带你从一个地方走向另一个地方,而你实际上会发现,世界上关于人类潜力和进步充满了谎言。也许人们在这点上并没有恶意,但他们绝对是困惑的。学校教导你,你必须在12个月的期限内学会某一块特定的数学,而不管你到底理解了多少。这就好像结果是可变的,却要在一段固定的时间内对一个可变的结果进行评分,这与我见过、学过或目睹过的关于如何实际学习东西的任何经验都毫无关系。顺着这条线索找下去,你只会发现个人成长是没有速度限制的。
在某种程度上,Shopify 对于这种信念来说一直是一个绝佳的实验实验室。我只是看到它被证实为真。当然,听到你尊敬的人对你说:“嘿,我认为你完全有能力把这件事做得好得多,因为我认为你可能很早就看到了项目中存在路径 A 和路径 B。你出于方便可能选择了路径 B,即使你知道那不是正确的做法。而实际上我对你的期望更高,我期望……” “我认为下次在你的职业生涯中遇到这种情况时,你应该走路径 A,因为……” “基于你的信念。”因此,这很难听进去,对吧,因为它是正确的。但这同时也是极其有价值的,对吧?我喜欢的是这样一种环境,人们彼此对真正的潜力负责,而不是对他们当前的水平加上或者……加上一些,我不知道,一点点额外的东西负责。
**Lenny Rachitsky:**在这里有太多我想继续追问的线索了。学校的例子就是非常切题的一个。现在我们正在为儿子挑选幼儿园,而他们……他们在描述他们的教育理念时,我就会想,我不知道我为什么要相信这是正确的方法。这让我只想花所有的时间去研究哪种教育方法是有效的。我知道这还只是幼儿园,也许还没那么关键。
教育与过拟合
**Tobi Lütke:**你押注了吗?我也有三个孩子,这在某种程度上是每个父母都要面对的决定,对吧?你的许多听众可能是机器学习产品的产品经理,所以也许这能引起共鸣。机器学习有一个有趣的现象,你就像……你在大量数据上进行训练,并且希望你能从中得到一个能正确预测你想让它预测的东西的模型。这其中最大的问题是过拟合(overfitting),对吧?什么是好的表现,一个损失函数(loss function)?它是一种启发式方法,因为它不是该模型未来实际要执行的任务,而是某种替代你希望它未来执行任务的代理指标。预测欺诈、预测下一个词,随便什么。所以过拟合基本上就是模型学会了如何在基准测试或适应度函数(fitness function)上作弊。
所以这有一个商业上的类比,那就是……它被称为古德哈特定律(Goodhart’s law)。它字面上和过拟合是完全一样的东西,只不过是用在商业上。古德哈特定律只是说,“任何成为目标的指标都不再是一个好的指标。”完全相同的事情。普遍真理是,几乎任何竞争领域都会用不同的术语为自己发明这些东西。顺便说一下,我认为这也是为什么专注于个人成长并广泛学习如此有趣的原因,因为你最终会发现事物背后这些隐藏的和谐之处,那些明显是持久正确的见解。所以过拟合、古德哈特定律是一回事。学校优化的是什么?据说是分数,对吧?事实上,学校里的过拟合字面上就是孩子们为了获得分数而作弊,对吧?你得到了另一个类比。然而,什么是适合孩子们的正确损失函数呢?你做出决定了吗?
**Lenny Rachitsky:**没有,我还没有,我们才刚刚开始走这条路。你告诉我——
**Tobi Lütke:**这是那种[听不清]的事情。你确实需要深入哲学层面才能弄清楚这个问题。然后之后你才能去构建,去找到你喜欢的学校。对我们来说,仅仅是保持好奇心,这与取得好成绩是完全不同的目标。但我只是认为每个人生来都极度好奇,而学校习惯于把这种好奇心从孩子身上剥离。从字面上看,孩子有一个基础模型,你在学校里对它进行微调(fine-tune),然后它就失去了好奇心的神经元(neurons),因为事实上,仅仅因为有趣就漫步到其他话题中去探索是不被鼓励的。我不知道,这有点不符合播客的节奏,但我确实经常思考这些问题。有趣的是,这些事情是如何不断重复出现的。
核心团队的无KPI运作
**Lenny Rachitsky:**所以当 Archie 上播客的时候,他是 Shopify 的增长负责人……我不知道这是不是他的官方头衔。基本上主导了很多增长工作。他谈到核心产品团队,除了增长团队之外,在没有KPI、没有具体目标的情况下运作。而决策主要由品味和直觉驱动,也就是你、Glen 和其他一些领导者。很多人听到这个,他们的反应是“我”……“首先,我不相信。其次,当没有数据告诉我们什么是对的和好的时候,一个人怎么能那样运作呢?”所以我的问题是,一个人如何才能成功地以那种方式运作?一家公司要那样运作需要什么条件,因为很多人会尝试然后失败?
**Tobi Lütke:**这非常接近我之前和[听不清]谈到的内容。古德哈特定律是真实存在的。当一个指标成为目标的那一刻,它就不再是一个有用的指标,对吧?我认为这或多或少是一个精确的表述。为什么?因为没有哪个指标本身能作为复杂商业的完整启发式方法,因为商业是复杂的。一家公司里有数以百万计的不同张力,你不可能通过优化一个固定值来让它们都保持和谐。确实,我们没有KPI,至少没有硅谷意义上的 OKR,但我们极其依赖数据。我们在系统上投入了大量的金钱和时间,基本上让我们触手可及一切。我把这个演示发给了其他创始人,他们完全被我们能够深入挖掘的方式震惊了,基本上可以挖掘出15分钟前刚刚形成的群组(cohort)中每个组成的原子级细小部分,就在一个季度、一个月或一周的末尾。
在很多不同的地方,这是其中之一,在产品中也是如此,只是没有为了可量化的事物而过拟合。每个人都在为所有事情竞争,但它是高度可量化的,因为……这很有趣,就像一场游戏。你调整一个数字,0.1比少0.1要好,这是一种即时满足感。但我只是认为,你用产品能做的最有价值的事情,与碰巧完全可量化的事情之间的重叠部分,大概只有20%,这就留下了80%的价值空间,是那些只看可量化事物的人无法触及的。Shopify 对不可量化的事物感到自在,比如品味、质量、激情、爱、恨,这与人们拥有的强烈情感同在。
工匠在把工作做好时感受到的那种深层次的满足感,如果你允许它发挥作用,实际上比通过单元测试(unit test)是一个更好的代理指标(proxy)。单元测试可能无法通过,而单元测试会在15分钟后通过,因为我们已经修复了它们,或者调整了一两件事物让它们支持我们。我们有系统可以准确地告诉我们事情是否走错了方向,Shopify 中有一个极其复杂的推出系统,永远保留对照组(holdouts)并在每一次实验等等中将所有事物与所有事物关联起来。但如果你把它想象成飞行员的驾驶舱,决策仍然是由飞行员做出的,我们认为这会导致更好的结果。这和我们的产品是一样的,商业有很多 A/B 测试工具和所有这些东西,弄清楚你的转化率(conversion rates)是什么当然非常重要。但是你代表了你的品牌吗?这是一个不可量化的问题,你为你建造的东西感到骄傲吗?你觉得它是你自己的吗,对吧?
我认为商业中需要对不可量化的事物有更多的接受度。商业世界中最强大的、并非不可量化的事物是乐趣和愉悦。如果人们在做事时感到有趣,那只是处于那么多其他事物的上游……抱歉,是下游。我认为,如果所有指标都在指向下方,但每个人都说,“我的天哪,我获得了多得多的乐趣,”我认为接下来会发生的事情,伴随一定的时间延迟,就是所有指标都会开始上升。如果那没有发生,我们就会调整方向。
我们之所以明确不使用 OKR 和这些东西,是因为……如果你想把不可量化的事物作为稳定存在的东西……让人们实际上真正地服从它们,并真正习惯于有人只是说,“嘿,这真的非常棒,他们正在发布这个,”那么你需要对企业进行某些调整,不要让每个人过多地想起他们可能来自的那些公司,在那些公司里获得晋升的唯一方法就是把指标拉高。我想这有点像[听不清]的对话。签语饼上写“Shopify 不做 OKR 或者不做指标”之类的话倒是很好。但这实际上只是因为指标会把我们带到一个[听不清]函数中,在这个函数里我们经常只服从于更多……有时带点情绪但通常更不可量化的事物。
**Lenny Rachitsky:**我想如果有人听到你描述这种专注于喜悦、乐趣、爱和愉悦,也许……这很容易被忽视。
对抗商业审美的偏见
**Tobi Lütke:**这听起来完全很愚蠢,对吧?再说一次,世界上存在一种审美,那就是商人穿着西装打着领带,他们说话比我通常复杂得多,没有口音,有一头浓密的头发。他们谈论指标,站在 PowerPoint 演示文稿前,通常拿着一根教鞭,戏剧性地指着他们身后的饼图,并且极具魅力,极其……所以那就是我们的审美。这种审美与我们的表现有多少重叠?我不知道,但他们中的一些人……一些像这样的人做到了。我认为世界在某种程度上被设定为会让我们误入歧途,因为我们关于最优状态是什么样的故事在太多方面实在太不正确了。乐观总是听起来很蠢,或者至少很天真,悲观听起来极其老练。指标驱动听起来极其老练,谈论乐趣听起来很天真。
好吧,首先,我总是忽略人们普遍的想法。不知怎么地,这对我来说是很自然的事情,我对此感到非常幸运。但我现在实际上已经认识到,世界上几乎所有的 alpha 都在于……恰恰是那些不明显但真实的事物,以及那些被人们当作天真之类而忽视的事物。地球上最成功的商人是埃隆·马斯克(Elon),无论你能想象到哪个方面,他都不符合关于最老练的商人应该是什么样的任何观念。我认为我们生活在一个反事实事物正在获胜的世界里,因为我们的……因为审美只是在引导我们误入歧途。
**Lenny Rachitsky:**这是一个绝佳的过渡,引出了我想花点时间讨论的另一个主题,那就是第一性原理思考,埃隆·马斯克是这方面的经典案例。老实说,我认为你如今是另一个最经典的案例。我们将继续讨论你与其他公司运作方式截然不同的所有方面,这些都是例证。但我想读一段 Glen Coates 分享给我的关于他如何看待你的话,这为了解你的第一性原理思考提供了一个有趣的视角。以下是他对你的评价:“Tobi 骨子里是一个真正的未来主义者,他痴迷于事物在未来应有的样子。数据驱动本质上是锚定在用户和技术当前的行为方式上。他从未真正向我明确表达过这一点,但了解他的人我想,任何主要基于事物现状或过去的设计,在他看来,都劣于那种滑向事物可能或应该成为的样子的冰球的设计。”这能引起共鸣吗?
**Tobi Lütke:**是的。我的意思是,我认为那是正确的。那实际上非常有趣。你的播客是一个建造者为其他建造者做的播客。我认为人们能问建造者的最有趣的问题就像,“你的能量来源是什么?你从哪里获取能量?”我认为从根本上说,世界存在于室温下。几乎所有的公司都在那个温度下运行,嗡嗡作响,无所作为。有某些个体能够向企业注入热量。创始人非常擅长做这件事。
任何人听说过的所有初创公司都有在注入热量的人,因为如果没有人向企业注入热量,在室温下,你无法超越其他任何人。如果没有人向这个企业注入热量,你就不可能比所有人都热。所以从根本上说,公司有一种能量的注入,它来自创始人和最优秀的领导者,就像你邀请过的所有来自 Shopify 的播客嘉宾一样,是一组完美的人物阵容,他们都是放热的(exothermic)。他们就像能量的源泉,导致了我们得以享受的所有惊人成果。
所以问题是能量从何而来?这也是我们的另一个讨论,它很快就会涉及到情绪。实际上,有一个非常……所以我前段时间看了《最后之舞》,迈克尔·乔丹的 Netflix 纪录片,当然,但有一个场景,他只是,我确定这是一个超级著名的故事,他只是捏造了一个别人告诉他的洞见,以便他随后可以去彻底摧毁他们,而他后来当然这么做了,因为很难想象有谁比他更放热。所以我们知道他的能量来源是什么。是竞争。潜在地,是洞见,或者是愤怒,诸如此类的东西。
能量来源与反现状
我不是……我的能量来源是对现状的不满。我的基本信念是所有这些关于技术的谈论,所有的……这么多书都在讲技术只会导致反乌托邦。你知道反乌托邦是什么吗?是今天,相比于它在20年前或任何时候将会变成的样子。你可以把这个应用到人类历史的任何部分。我不是在做一个关于未来的声明。我是在做一个关于地球体验的几乎是全局性的声明。
真正思考过这个问题的人,没有人会想出生在今天的20年前而不是今天。所以我认为今天是未来的反乌托邦,我认为我们有义务尝试去构建那些以大大小小的方式引领进步的产品。但是是的,我认为如果有人来对我说,“嘿,我们去做这件事吧。我们四处看了一下,这是人们解决这个问题的方式,让我们做一个更好的版本吧,”我会说,“那不是我们要做的工作。”因为你遇到的每一件事,每一个解决方案,每一个产品,每一个存在的事物都是路径依赖的(path-dependent),高度、高度、高度路径依赖的,而且通常是路径依赖于必须做出妥协,基于在做决定时是真实的但现在已经不再真实的事情。
整个领域……那是什么?我忘了这个领域的名字。乔姆斯基的领域,语言学研究领域。很酷。我们现在有自回归模型(autoregressive models),就像,我们实际上不需要建立一个完整的……我们不需要研究语法的结构来让机器也参与到口语中去。结果发现,我们实际上可以直接在互联网上训练。所以这在当时是不可能的,因为你没有合适的架构,但现在有了。
所以我认为你必须要做的是,实际上,当你提出一个新产品或讨论一个新产品时,你必须从第一性原理推导它。你必须说,“鉴于我们现在拥有的每一个基本构建块,我们将如何解决这个问题?”为此,要做到这一点,你实际上必须理解现在存在的所有构建块的力量和可组合性(composability),这是一个很高的要求,没有人能完美地做到这一点。
但这样一来,你继续说,“好的,很酷,所以这就是我们实现这个东西的方式。这就是它今天将被实现的方式。”然后我们现在可以说服自己走捷径。“也许我们实际上应该开始像其他人那样做。也许我们推导出的正是每个人都在做的正确的事情。”有时现状中编码的智慧比你预期的要多得多,我认为这超级令人愉悦。然后你发现了这一点,所以当你据此采取行动时。但不可以接受的是跳过这个练习,去做和其他人一样的事情,因为这再次是对产品领导力的放弃。
所以是的,我会说我变得极其怀疑,如果我收到一个提案,去做一个和其他人做的相同事情的好版本,因为我只是发现在我们这个领域,这极少是最好的解决方案。
公理与商业逻辑
**Lenny Rachitsky:**我有看到过几个符合这路线的引语,描述了你思考这些东西的方式。“如果大多数人都以某种方式做,我默认不想那样做。如果你想做成世界级的,你不能像其他人那样做。”
**Tobi Lütke:**是的。我甚至不认为这是我持有的一种观点。我认为这实际上是……我们基本上处于公理(axiom)领域。如果你想做得比现有事物更好,你必须做得不同。这并没有说明在你做完之后它最终是否会更好。它也可能更糟。但如果你做同样的事情,你不可能做出更好的东西。这在公理上是不可能做到的。
它不符合阿基米德逻辑(Archimedean logic),然而它是某种……你会惊讶于有多少商业计划实际上是以这种方式不符合阿基米德逻辑的,就像,“让我们做一个我们已经一直在做的东西的好版本,我们将占据1%的市场,”诸如此类的东西。就像,“相信我。”我觉得这有点可爱。
从第一性原理推导的实践
**Lenny Rachitsky:**你能分享一个你用这种方式处理问题的例子吗?我想这应该是在不断发生的。你也提到你天生如此。所以我认为对于一个人来说,仅仅坐下来学习,像 Tobi 那样思考是很困难的,但我很乐意帮助人们开始以这种方式处理问题,所以也许一个例子会有所帮助。
**Tobi Lütke:**我认为这完全是可学的,我想。所以我鼓励人们基本上保持逐步思考的练习,去做就好。它会很快成为一种习惯,因为它只是,它也只会表现得更好。
例子。最开始的例子就是 Shopify 本身。这很酷。所以当时有很多电子商务软件,它们之所以是那个样子,是因为路径依赖,因为在 2004、2005 年想要它的人,都是现有的零售商,因此他们有复杂的业务需要移植到线上,包括他们所有有些拜占庭式的业务逻辑。“我想制作能在未来的互联网上表现优异的电子商务软件,并且我相信我们可以让在线上开启新业务比在物理世界中更容易,因为物理世界受到许多法规的拖累,还有租赁等前期成本。所以让我们针对这种情况进行优化,构建一些非常直观易用的东西,以至于在死胡同职业中感到沮丧的人可以在午休时间里为建立自己的业务取得进展,这最终让他们能够以自己的方式去做这件事。”
所以这种运气的偏向,最终证明它其实也是解决所有企业级案例的更好准备,因为没有人必须……企业级软件过拟合于销售过程,也就是它赢得了 RFP,因为它拥有有史以来的每一个功能,或者至少有一种方法可以在每一个 RFP 条目旁边打上勾。但它们并不好,对吧?一个 RFP 是采购世界中过拟合的绝佳例子,因为它根本无法告诉你其背后软件的质量。但老实说,这种情况一直都在发生。
搜索的局部最大值
就像在这里,我们正在为 Shopify 上以及整个 Shopify 的产品进行更高质量的检索搭建舞台。我认为我们在搜索方面一直处于局部最大值(local maxima),并且我们认为,特别是随着新模型的进步,现在有些事情变得可能做了,而这些是以前无法做到的,因为又一次因为这一层又一层的路径依赖,没有人过来说:“那是最好的方法吗,也许我们应该定期重建这个组件?”我们现在可以在搜索方面做得更好,从而带来更令人愉悦的体验。所以这是一个现在正在进行的有趣项目。
**Lenny Rachitsky:**我觉得这非常有趣,因为我所寻找的就像是第一性原理思考的 Tobi 算法。埃隆·马斯克分享了他那些著名的思考方式。其中之一是从金属的成本开始,以帮助你了解火箭应该花费多少。然后他有这个五步法。首先,决定我们需要这个东西吗?然后弄清楚如何优化它,接着是自动化它。根据我目前听到的,我很好奇你是否考虑过这个?如果没有,这感觉像是你未来一篇非常好的博客文章,即 Tobi 第一性原理算法,但我会分享几个到目前为止在你描述中我听到的东西。
Tobi 第一性原理算法
**Lenny Rachitsky:**一个是分析现有解决方案所依赖的路径,几乎就像是当初它被构建出来时成立的那些假设。另一个是看这个过拟合,它是为了什么而过拟合?它在过度解决哪些也许并不必要的问题?在关于你如何处理问题这方面,有没有类似这样的思路?
**Tobi Lütke:**这可能太书呆子气、太技术化了。你是对的,但我应该稍微理清一下我自己。我的大脑运行在一种度量语言上,但我想它并不能直接转化为文字。我更多地是从编程构造、纯函数、过度陈述这些方面来思考这类事情的。并且我认为在任何时刻,最好的决定是,我认为完美的产品负责人几乎就像是一个针对高质量产品的恒温器。就像你在设定,说:“我想构建一些非常非常棒的东西,并且我将经历一个比恒温器所做的要复杂得多的系列过程,”恒温器基本上就是检查温度,然后做出开空调或暖气的决定。你会让你重新推导字面上的每一个有价值的决定,每一个基础假设,每一个基础的 ABC 方向。并且你希望看到自你上次推导出下一步以来,在此期间你所做的观察。在现在已更新的状态、更高保真度的信息上重新运行整个函数,你会得出完全相同的东西吗?
从就地避难到永久远程
**Tobi Lütke:**有时候在构造中,在基础假设的树中相当早的地方,就发生了改变。我们都有一个例子是 COVID 初期,当时我们突然有了就地避难的指令。所以 Shopify 拥有那些令人难以置信的优质办公空间,而且我们是一家非常依赖线下办公的公司,我们……我们的楼层平面图因为我认为我们真的为如何打造极好的……我从我的联合创始人 Daniel 那里获得了很多创始人能量,来为创造性工作构建伟大的协作空间,那里有很多快乐的意外,人们互相偶遇等等。无论如何,我们非常非常非常非常坚定地要做那件事。但在这个嵌套函数的构造中的某个地方,你必须重新运行以及基础假设。在栈中是一个相当基础的布尔值(Boolean),即人们被允许离开房子吗?答案是肯定的。当那个值翻转时,并不是就像,“好吧,在这里让我们尽我们所能做到最好。”实际上是整棵树现在移动到了一个不同的地方。并且你可能降落在一个非常遥远、不同的地方,因为你会在每一步做出同样质量的决策,但你需要重新运行一些将你带到一个完全不同着陆区的东西。所以然后这也很容易让我们说,“酷,我们将永远只做远程办公?走吧。”因为我们意识到这种暂时的就地避难会导致一系列事件,这将使得那成为对公司来说最优的最佳权衡集合。所以我认为这是……我能把它变成一个精辟的五步法吗?我不知道,但也许它能在某些人那里产生共鸣,他们能帮我弄清楚这个编码的英语语言,因为我解释它的方式一直很书呆子气,即使对我自己也是如此。
**Lenny Rachitsky:**这个远程工作的例子绝对是我想要谈及的,所以我很高兴你提到了那里。那么在这个决定中,关于你是如何得出像“哦这个布尔值改变了所以我们应该重新思考这个?”这样的结论的,你还能分享些什么吗?因为我知道你大概,你不是……我不知道。你是在洗澡的时候突然就像,“哦哇,因为这点我们真的应该远程办公”吗?这实际上是如何发生的?然后我还想问另一部分——
办公地点的优化函数
**Tobi Lütke:**是的,确实差不多是这样,因为又一次,我重新运行了所有的输入并弄清楚会发生什么。这里还有其他几件事情也已经开始让坚持线下办公的决定出现松动。我们起步于加拿大的渥太华,这是一个拥有一百万人口的城市,它有一些科技底蕴和优秀的大学。但一百万人口,它的人口密度就是不够,也没有足够深的人才库来支持一家将要发展到一万人的公司。所以任何运行我在这里谈论的关于新创业公司地点策略的优化函数的人,如果他们做对了,都会得出这样的结论:每个人都应该坐在同一张桌子周围。只有当你做不到这一点时,你才会说,“好吧,让我们都在相同的几个房间里。”只有当那也行不通时,你才会说,“好吧,让我们分散在同一层楼。”如果那行不通,在同一栋楼里。如果那行不通,在同一个城市。如果那行不通,至少留在同一个时区,或者确保它们之间有很好的枢纽连接。
这就是它的运作方式。就像在这个过程中某个地方的某个假设被打破了,你最终会落到决策树的不同一侧。但当时我们身上发生的是,我们已经分布在四五个城市了,那时我们已经坚持在同一个时区了。但我的体验是,有时在渥太华办公室,我会经历整整一天10个小时的会议,这很正常,但我每一个……有时我觉得我身边没有其他任何人,因为每一个会议都是和另一个办公室的人开的,有些人远程拨入等等。所以存在一种尴尬的混合状态,而我们在这一路上最终只做出了好的决定。
这实际上是最危险的事情。大多数时候,你之所以会落入决策树糟糕的部分,陷入路径依赖的环境中的局部最大值,恰恰是因为你只做出了好的选择。人们认为做出一个好的选择就能让你免于犯错,或者认为一个想法存在缺点就注定会使这个想法失去资格,这两点都是不正确的。所以你需要做的是……所以那是决策的一个叠加层。
COVID 开始的那一刻,我们还遇到了 shoppers 爆炸式增长的情况,因为在 COVID 期间我们实际上是人们的一项资产,是本地企业的电子商务。我们对此非常非常认真,试图让更多的企业在 COVID 中生存下来,让小企业在这场特殊的灾难中生存下来,否则它们就不会存活,这很重要,因为小企业往往在任何时代变得脆弱时最先被消灭。所以我们需要扩充人员,所以关于在哪里扩充人员出现了一个非常奇怪的问题。很显然,如果你能在任何地方招聘人员,那将是更好的输入。所以一旦那个决定翻转,你就能看出现在说“我们要远程办公”实际上是多么容易,因为没有回头路了。即使有了这个决定,对未来来说更好的一组权衡是接受建立一家远程公司的巨大额外困难,这要困难得多。
这不是你应该向任何人推荐做的事情,因为这就像试图在科罗拉多州的阿斯彭跑出马拉松世界纪录一样,就像那里没有足够的氧气来做到这一点。但如果你最终成功了,你就是个真正的 Chad,那非常酷。所以我觉得困难本身很有趣。而且同样,我在互联网上度过了足够多的青少年时光,所以我知道可以纯粹通过远程聚集起惊人的文化,比如,我不知道,维基百科,魔兽世界,副本公会至少可以有优秀的文化。所以我们就像,“好的,酷。让我们来到一个新的稳定部分。”那个稳定部分是不要把办公室搬到线上,让我们把互联网引入一家公司,然后我们就像,“走吧。”
深入底层与第一性原理
**Lenny Rachitsky:**我想谈谈你提过好几次的那个我认为非常棒的人才栈(talent stack)概念。但在我们谈到那个之前,我想换个方向,我认为这是你能够在第一性原理思考上取得成功的一个有趣的基础要素,也就是你通过实际上花费大量时间写代码,深入到金属层(getting to the metal),触及你正在解决的问题的核心。我其实听过一个故事,说你们有一次大型的全体峰会,你们把所有人都飞过来,大家都在一块儿,有一个为期三天的黑客马拉松。描述的方式是你戴着耳机坐在桌前写代码,就像任何 IC(独立贡献者)工程师一样,没人会知道你是这家公司的 CEO,这家公司的创始人。对你来说,为什么写代码和在代码中工作仍然如此重要?
**Tobi Lütke:**这个故事是这样传开的,这很有趣。它是对的,但我不知道为什么这会令人惊讶。那是我的快乐所在,能够清空我日历上的空闲日子,和所有这些加入我们旅程的非凡之人一起待到午夜,只是去构建东西。就像是,我做我的工作,以便那样的工作能为其他人存在。那是我真正想要的工作,那是我找不到的工作,因为没有人为我创办过公司。但至少在过去的大多数日子里,我觉得实际上还是没有。
所以我就是热爱写代码。它是最伟大的,我不知道,爱好和追求之一,我做这个已经很久了。我非常非常早地接触到了它,它完全贴合我的大脑,我非常欣赏编码背后的手艺。我受过训练成为一名学徒……抱歉,我在德国做过程序员的学徒,那里有双元制教育系统,你可以做这样的事情。所以我一直在公司里专业地度过一整天,而且我是认真的,从我刚满16岁起就在编程,我想第一天是在我刚满16岁之前,当我开始我的学徒期时。
所以我热爱它。我的屏幕上在你左边现在有一个 cursor,它打开了一个 Juniper notebook,我在里面处理我正在玩的一些预测数据。我试图尽可能对我得到的东西进行合理性检查,我试图发现……在数据中发现新的洞察对我来说是一场游戏。我只是觉得,我不知道,这很酷,也许这就是你说的那张照片。有一张别人拍的照片,真的很有趣,我在那里用我的笔记本电脑和一群工程师在一起,然后 Harley,也就是我们的总裁,在台上打碟。这是在凌晨1点左右拍的,这张照片实际上对我们俩来说真的很珍贵,因为它以一种方式完美总结了我们的关系,这些快乐的意外是美妙的。在这些漫长的旅程中,我做 Shopify 已经20年了,所以我很欣赏这些 artifacts。但是是的,所以……
**Lenny Rachitsky:**需要明确的是,这不寻常的原因是,也是为什么有人告诉我这个故事的原因,是大多数 CEO 不会这样做,不会就坐在那里和团队一起写代码。我认为这个故事很重要的原因,以及我认为你的第一性原理方法之所以有效的原因,是你实际上身处工程细节之中,如果你想一想,这类似于埃隆·马斯克,就是深入其中做这件事,理解这件事是如何运作的,而不仅仅是凭空想出一些主意。所以我想问的是,如果你想像第一性原理那样思考,真正贴近底层,贴近最底层的硬件,这到底有多重要?
**Tobi Lütke:**嗯,是的。第一性原理思考始于,我想埃隆·马斯克把它说成物理学,我认为这有点太偏向于原子了。我认为仅仅从原子构建块开始才真的是一个正确的起点。原子构建块就是我们正在使用的计算机,计算机是我们的乐器。我们用它们来创造人们喜欢以软件形式接收的音乐,所以你必须了解它们是如何工作的,至少要像我这样工作的话。
我的工作方式是最优的吗?当然不是。这就是我关于如何行事、如何工作的美学观点。我可能不符合大多数上市公司应该是什么样、甚至应该把时间花在哪里的传统看法,但我认为我之所以成功,是因为我不试图迎合任何东西,除了我学到行之有效的方法。而我学到的行之有效的方法就是,尽可能深入细节。真的是、真的是、真的要去理解我们正在做决定的东西,并且愿意不要在沉没成本谬误(sunk cost fallacy)上投入太多筹码。尽可能地让业务或你所在的公司部门对沉没成本谬误产生免疫力,因为这能让你看到更好的解决方案,等等。
我想我们当时在做的,如果我没记错的话,是在黑客日期间,我们正在梳理将整个 Shopify 合并为一个巨大的单体仓库(monorepo)的利弊,我们正在勾勒我们需要的目录结构和工具,而且……我们需要的工具,而且……再说回单体仓库,现在对于公司来说,它非常像那种选 A 门还是 B 门的事情。这是一个后果重大的选择,在某个规模下同意它是不正确的,然后在我的观念里它又会变得非常正确,但在那个节点上这需要巨大的努力。所以,这是一种实际上由我处于独特位置来参与的事情,因为它实际上也是一项商业战略。这是一种投资,一项非常真实的投资,去说:“嘿,让我们改变构建系统的方式。让我们找出最好的方法是什么。”
会有变革管理。会有一些人对说是或否有非常强烈的意见。坦白说,“Tobi 这么说了”,是有帮助的。这也压缩了大量时间,因为每个人都知道我的行事方式。任何人都可以带着关于任何事情的更好主意来找我,如果你是对的,我会改变主意。但我会非常强烈地坚持我的观点,直到被说服它们不是正确的想法。也许这就是你早先谈到的 Tobi 龙卷风现象的另一个方面,人们确实在战斗。再说一次,我们时代的审美把很大筹码放在了一致性上。我记得有各种各样的政客因为在某些时候被称为反复无常者而输掉竞选,这我……虽然,我更属于梅纳德·凯恩斯学派的……“当事实改变时,我改变我的看法。先生,你会怎么做?”听起来这是一个更好的操作系统。
如何进行第一性原理思考
**Lenny Rachitsky:**这太棒了,因为我正在提取更多你思考事物的算法。你刚分享的几件事是,(1.),你需要身处细节之中。这就是思考如何作为一名第一性原理思考者取得成功,(B.),在你的情况里是代码。在埃隆·马斯克的情况里,就像,造出那个东西并在工厂里,睡在地板上。正如你所说,他非常基于 Adams。你则更数字原生。然后还有,不要过于依赖某些成本。就像你在这个 Tobi 龙卷风的情况里说的,“我知道你一直在这个项目上工作,我们要终止它,因为如果我们现在做这个相比于做那个,它不会效果更好。就继续吧,因为我们一直都在做。”然后,回到你之前分享的东西,分析你走过的路径,之前的产品和解决方案假定了它们的路径依赖性,然后不要看它们可能过拟合(overfit)的东西,那是不正确的。我很喜欢。好的,这实际上是一个很好的过渡,引向另一些与你共事的人建议我问你的事情,Farhan,你的工程主管。我问他,“一窥 Tobi 内心世界的最佳方式是什么?”而这实际上与你刚刚描述的方向一致,也就是,问题仅仅是,与你意见相左的最佳方式是什么?
异议的价值
**Tobi Lütke:**我想就是直接与我意见相左。说实话,我立刻就会爱上它。我真的很渴望它。这非常有趣。我知道这真的很让人惊讶。而我实际上更欣赏它,因为它需要勇气,坦白说,我确实发现……但它也让我立刻更信任这个人,部分是因为,首先,我认为他们会做他们认为正确的事情,而不是方便的事情。同意一个群体的意见往往要方便得多。但实际上不仅如此,他们有勇气在那个时候就这么做。我认为勇气真的是、真的是很罕见的。我在工业界发现的高智商比勇气多得多。我发现甚至可能天才都比勇气多。我喜欢那样。我希望它不需要那样。这就是为什么我努力对它表现得非常欢迎。
当有人与我意见相左时,我倾向于立刻停下来并说,“酷,让我们弄清楚为什么会有分歧。”我发现这几乎从来不是处于一种,“我就是觉得我们应该用不同的方式做这件事”的状态。我寻找的是提供未言明的基础假设。我们的分歧点在哪里?因为你可能是对的。事实上我发现,当到了这一步时,人们往往是对的。有时是我持有的一个未言明的基础假设是不正确的。人们告诉我,然后我非常高兴我们谈论了这件事,因为我将永远不再唠叨这件事,因为现在我知道因为萨班斯-奥克斯利法案(Sarbanes-Oxley)之类的东西我们做不到它。就像,“嗯,酷。”我脑海中关于萨班斯-奥克斯利法案(这是对上市公司的监管)的心智模型并不完美。我没有深入研究过它的所有细节,所以当我说,嘿,我们应该用某种方式解决手头的任务时,我不会在脑海中参考它。所以,这非常棒。
如何表达异议?我真的很喜欢辩论。如果每个人都同意某件事,我会主动扮演魔鬼代言人(devil’s advocate)。再说一次,特别是一个提案感觉像是我在它进入会议之前就能预测到会是一个提案,并且其中没有任何令人惊讶的东西时,我会让自己成为一个关底的终极 Boss,只是说,“我打算说这并不是那么好。我认为我们能做得更好。”我想让人们然后为了决定的正确性进行更深入的论证,而这会导致一种形式的分歧。我认为所有这些东西最终都在建立信任。
**Lenny Rachitsky:**我喜欢这也触及了再次最大化潜能的想法,团队的潜能,员工的潜能。
决策的本质与产品的灵魂
**Tobi Lütke:**因为当然,真正重要的决定,我们是不讨论的。这是最重要的事情。Shopify 可能每天做出……什么?……数百万个决定,比如,用这种方式写这段代码,是的,我要加上这个单元测试(unit test),也许我要跳过一个单元测试。这可能是未来生产审计之间的差异。数以百万计的这些微小的决定。所以,你主要不是在雇佣工程师或者会计,你是在雇佣在他们的专业领域和他们监督的领域内做出优秀决定的人。鉴于此,我发现决策作为一个概念实际上真的被研究得太少了。
我认为这些是我们能学到一起做决定的例子。因为我认为虽然很多决定是独立做出的,我们是一家产品公司,我们肩负着使命,我们希望我们的产品感觉像是某一个人做出的东西,就像任何作者都试图写一本明显读起来是出于单一头脑的书一样。因为人们能看到并发现这一点,如果这没有发生的话。你最终会得到一个看起来像电视遥控器一样的东西,这边有一个 Netflix 按钮……就像你可以从遥控器上反推 [听不清 00:59:41] 一样。这很重要。
利用这些时刻,同时将决策过程内化,去说,嘿,让我们为尽可能多的人创造非常高效的聚在一起共同做决定的机会,不是以民主的方式,而是清楚地知道谁需要说服谁。因为否则这最终只会花费太长的时间。给每个人一个改变我的想法而不是 Glen 的想法的机会,等等,我发现这是一个极好的做法。我认为这是一个更优的处理方式。
**Lenny Rachitsky:**说到意见不合,这让我想起我听过的一个故事,实际上是关于“Tobi 龙卷风”的,也就是你的运作方式。我很喜欢。它最终成为了你第一性原理思维方式的一个有趣的缩影。这个故事是,我想你曾经在某个时候说过,让人们给你见解的最好方法,就是在互联网上说一些他们不同意的话。
**Tobi Lütke:**是的。
**Lenny Rachitsky:**你通常采用的方法就是,你在 Slack 群组里发帖说,“嘿,我认为这个产品行不通,原因如下。”这最终会为你创造出最多的信息。在这方面有没有什么值得分享的经验?
**Tobi Lütke:**我不认为那是我的一个好方法。我必须稍微小心一点。对于和我一起工作很多的人,我会这样做,只是因为这很好玩。但连我都能看出,那会让实习生压力山大。
**Lenny Rachitsky:**你们有很多实习生。我想 Farhan 提过,明年你们会有 1000 名实习生。
**Tobi Lütke:**是的,我们刚刚开始。我昨天在办公室,里面绝对全是实习生。太棒了。
**Lenny Rachitsky:**天哪。喜欢实习生,精力太充沛了。
百年愿景与逆向工作法
**Lenny Rachitsky:**你分享的这个关于打造一个产品、朝着同一个愿景建设的想法,这实际上让我想起了每次我请到 Shopify 的人来做客时基本都会谈到的事情,也就是你一直保持的这个百年愿景。我不知道还有任何其他公司是以这种方式运作的,创始人对产品需要发展到哪里有一个百年的愿景,并以此向后推演。你能谈谈这种运作方式吗,为什么你觉得它有帮助,它实际上是如何运作的?
**Tobi Lütke:**我经常谈论展望未来然后向后思考。在这件事上,我们会希望自己在20年前、10年前或5年前做了什么?我们未来的自己会希望我们做出什么决定,这是很有用的。我发现展望未来通常极具价值。我绝对不是为了转手卖掉而建立 Shopify 的。我有很多机会把 Shopify 卖给各种各样的人。我没有考虑过,因为这只是一段太有趣的旅程,而且我认为那太……如果所有公司的终点[听不清]都是收敛于四五个能买得起它们的人,我认为这在思想上只会造成太多的单一文化。
我喜欢 Shopify 是与众不同的。我认为这很好。我认为对很多、很多人来说,这是一个非常糟糕的工作场所。但对某些人来说,它是世界上最好的工作场所。这太好了。我认为这就是我们想要的。我们想要更多这样的东西。我们希望人们能够像逛橱窗一样挑选一个他们能取得巨大成功的地方,因为那个地方的信仰体系就像手套一样完全贴合你。
100年。我们有非常长远的计划。100年的时间,你无法谈论这个软件产品,但你可以谈论使命本身,在这个特定的时间范围内剩下的任何能存活80年的东西。创业精神是无比珍贵的。Shopify 的存在,基本上就是让创业变得更普遍。这就是我们想在世界中促成的事情。我认为我们在这方面已经取得了一些成功,但同样,这件事没有速度限制,也没有停止点。正如我不断说的,世界是令人难以置信地路径依赖的。因此,如果我们是一条路径的一部分,我们可以促使它更符合我们看重并希望看到的事物。这是建立公司的美妙之处。它们可以产生持久的影响。
我们正在基于我们能做的所有事情做出决定。哪些事情对我们所从事的长期追求具有最长远的价值,以及我们如何让创业常态化?这对我来说是一件非常重要的事情,只是因为人们在这上面花的时间不够。这全都是经济学。我们的[听不清]生活完全依赖于商业。我们都是一个极其具有创业精神的环境的一部分,并且实际上将创业颂扬为一种勇敢的甚至光荣的行为。
长期主义与重复博弈
但在世界大部分地区并非如此。事实上,大多数人从未遇到过任何从事这项活动的人。人们并没有把这看作他们的选项之一。我希望我们能做出这些选择。我认为长期的关注是重要的。它对决策也非常有效。我知道这对大多数人来说可能是直觉,但很少被实践。你请过很多 Stripe 的人来做客。Stripe 和 Shopify 有着非常长期的合作伙伴关系。Stripe 和 Shopify 可能拥有科技史上最有价值的合作伙伴关系,或者至少是历史上排名前几的之一,因为我们当时都是非常小的公司,而且我们决定,嘿,让我们在一个我们都将赢得各自市场的假设下工作,并携手合作。所以,Stripe 的[听不清]用于支付,我们被允许使用 Stripe 的一部分等等。
当你处于这样的合作伙伴关系中时,你基本上是在玩重复的囚徒困境(iterated prisoner’s dilemma)。每一个实例,每一轮你都可以做出选择:合作还是背叛,在另一个人合作时背叛,合作者效应。如果双方都合作,每个人都能得到一分。如果一方背叛而另一方合作,你会得到很多分,全都在一瞬间。在商业中做一个好的合作伙伴就像这个企业的棉花糖测试(corporate marshmallow test),公司往往会以一种非常滑稽的方式失败。
如果你看过孩子们做真正的棉花糖测试的视频,聪明的孩子实际上会把椅子转过去,不看棉花糖,把双手压在身下,然后就开始发抖。因为他们知道等待未来得到两个棉花糖是正确的做法,但是天哪,如果他们只是看着一个棉花糖,他们就会把它吃掉。大多数 CEO,大多数公司甚至都无法成功做到这一点,只是热衷于打折提取未来的利润,甚至背叛那些在长期来看会更有价值的合作伙伴关系。
如果你谈论的是像100年这样的长时间框架,那是毫无疑问的。但显然,玩重复的囚徒困境的正确方式是双方都合作。在很长一段时间内这样做,比任何瞬间的背叛在那一刻可能带来的便利要有价值得多[听不清]。这涉及大量的产品决策。当我们在决定路线图时,最终会受到这种想法的很大影响,因为我会收到各种关于我们应该做什么的提案,为什么我们不对系统做这种改变,这里面有这么多钱,诸如此类的事情。我的反应是,很酷,但这些几乎总是以打折提取未来利润的形式出现。
但我们有着很长的时间跨度。我们不要打折提取,让我们……而且,过一段时间你往往会损害整个业务,因为你最终只是……如果你开始榨取价值,你的客户是会注意到的。我相信,至少从领导者的视角来看,最好的公司就像某种驾驶舱,或者是一个装满用于变现的刻度盘或拉杆的房间。我的工作就是在这个房间里尽可能多地安装这些拉杆,然后一个也不拉。因为我认为,如果我们和客户坐在桌子的同一边,让他们成为创业者,这是一项使命,而我们希望支持他们取得更大的成功,这就是我们所处的商业模式。在我们的变现系统中,我们只从销售额中获取非常非常小的一部分。因此,我们有动力让每个人都尽可能成功。再说一次,当我收到投资银行家之类的发来的 PowerPoint,告诉我说我拥有巨大的定价权,可以大幅度改变价格时,我的反应是:“是啊,但如果这种事发生在我身上,我会离开 Shopify 的,这怎么算是好事呢?”他就像:“为什么?”事情就这样了。
正和游戏与终极增长黑客
**Lenny Rachitsky:**你在某处写过一句话,对我来说极其精辟地概括了这一点:“在足够长的时间线上,与你的客户玩正和游戏是终极的增长黑客手段。”是啊,太美妙了。
**Tobi Lütke:**我觉得这话相当漂亮,而且它也就是……你很难反驳它。
**Lenny Rachitsky:**确实。
**Tobi Lütke:**正和游戏有着令人难以置信的回报,尤其是在软件领域,在那里你可以很容易地体验到指数级增长。我把建立公司看作……我希望有一个比国际象棋更好的比喻,因为国际象棋是一个完全信息博弈的游戏,这完全不符合实际。但作为商业的比喻,我喜欢国际象棋的一点是,它基本上是两个游戏,而你必须同时擅长这两者。一个是你要学习的局面战:出子,让你的棋子在棋盘上获得影响力。这非常非常重要。然后还有战术,你通过战术训练来学习它。你去弄个解谜训练器,不断练习战术,你培养出对战术的直觉,去感知它们。这两者实际上是超级相互独立的。
商业的局面战与战术
商业界只谈论战术。它只谈论我们做了这件事的转化率。我们做了一个 A/B 测试,改变了蓝色的颜色,然后红色的转化率上升了。我们神化那些简单的技巧。老实说,我认为这并不重要。你在战术上需要足够好,以免倒闭。你肯定不能被追缴保证金。但如果你真的在做局面战,你可以运用的所有潜在战术的价值总和,都会留存在你这里。局面战就像,你在地图上占据的是什么领地?你扮演什么角色?你从商户那里获得了多少信任?商户是希望从你这里得到更多还是更少?你是那种他们试图从软件支出中优化掉的东西,还是那种他们要求用来包揽所有其他软件支出的东西?他们依赖你吗?你是团队的一部分,还是被当作工具箱里的一个工具,通常被遗忘,只有在执行某些任务时才会被拿出来?你的产品涵盖什么?你针对的是哪些行业?等等,等等,等等。这就是局面战。棋子之间配合得多好?人们是否喜欢更加深入地依赖你?如果你把这点做好了,战术就是你的了,你可以雇佣很多极其擅长发现和运用战术的人。但如果你在这方面做得太过,你最终会通过战术榨取掉你所创造的全部价值,而这些价值本该是你通过局面战来获取的。如果你那样做了,你的油箱里就什么也不剩了。这就是我们都能看到的那些公司,它们只是到了一个点,然后就慢慢淡出了。那些就是被战术耗光了自身位置的公司。
寻找短期与长期的平衡
**Lenny Rachitsky:**显然,这里的技巧在于找到这两者之间的平衡:短期内,如你所描述的,不被追缴保证金,同时还要向前看。对于正在听这期播客的创始人来说,有没有什么启发式原则,比如他们心里会想:“我该怎么做?我应该在多大程度上思考未来而不是现在?”我猜是这样吧?你是如何尝试画出这样一张饼图,既能立即推进目标并向投资者展示我们做得非常出色,同时又能高瞻远瞩的?
**Tobi Lütke:**有没有好的启发式原则?首要目标是:不要死。再说一次,最终,那些成为历史性公司的企业就是那些没有死掉的公司。这听起来非常基础,但实际上它比表面看起来更具有可操作性。过了那个阶段,我的观点是:只关注局面战。我认为,这就是创始人有时与公司其他成员之间存在一点分歧的地方。我认为创始人为人们创造了有限的、能赢的游戏,而这些游戏极大地服务于使命所暗示的无限游戏的发展,服务于在棋盘上的位置质量。顺便说一下,这是一个无限的棋盘。只需把国际象棋棋盘想象成初始地形,但棋盘上有,我不知道,迷雾,而且这是一个随着时间的推移你将去探索的、大得多的地形。我认为这是在脑海中构想建立公司体验的最美方式。它是一场探索,是对使命所暗示的一个问题的协同探究。在此过程中,你将得以探索你的使命有多正确,你的决策质量有多高。你会学到很多,这就是为什么我认为这是一件有价值的事情。但他们往往,创始人和公司的使命是保持一致的,但它们,再说一次,是不可量化的东西。它们是追求。也许是作为超越了,超越了我们所有职业生涯时间跨度的时间跨度,因此这让人们很难现在就深深地关心它们。但他们的确关心那些实例,关心沿途的那些游戏,关心为了探索地图的一部分而必须完成的[听不清],关心必须执行的战术。
《有限与无限的游戏》的启示
**Lenny Rachitsky:**我不打算在这里带我们深入探讨,但我知道你是《有限与无限的游戏》这本书的超级粉丝。我实际上曾经在读书会上读过一次,它让我的大脑豁然开朗。我会放个链接。大家应该去看看。你在另一个播客上谈过这个。我不打算[听不清]。
**Tobi Lütke:**是的,这是一本可爱的书。James Carse 做了一件令人难以置信的、未被充分赏识的……我认为他们本质上写了最好的商业书籍之一。他们试图写一本硬核的哲学书。我不知道它是否实现了其预期目的,但我认为他实际完成的东西更有价值。也许,可悲的是,他在死前可能都不知道这一点,那将是非常可悲的。
**Lenny Rachitsky:**这本书对人们来说有点难读……
**Tobi Lütke:**是的。
Lenny Rachitsky:……所以我的建议是,坚持读下去,试着去理解他想要表达的意思。
**Tobi Lütke:**我认为阅读……我不知道具体是几章,但是……前几章真的能帮助你掌握整个故事的脉络,他的洞见。书的其余部分是他的思想应用的例子。我读他这些例子的心得是,他并没有完全欣赏他自己想法的洞见。它最终变得非常局限于局部,而且视野非常狭窄。这是一个比我认为他真正……更宏大的想法。这就是为什么我说我不认为他完全欣赏他自己想法的质量,这……
**Lenny Rachitsky:**我觉得我们需要一个 Tobi 版本的这本书,加上一篇前言,还有……
**Tobi Lütke:**我听说 Simon Sinek……
**Lenny Rachitsky:**嗯哼。
Tobi Lütke:……写了一本类似标题的书。我其实一直没抽出时间去读,但我想如果他写的就是这个,他会非常擅长解读它。
**Lenny Rachitsky:**听起来就是这样。我觉得他的书就是那本书,只是换了一种写法。我想回到你谈到的一点,也就是专注于创业以及你们合作的那些商家。对我来说,这是最大化人类潜能的又一个例子。我的理解是,人们总是在谈论 Y Combinator 和斯坦福,以及所有这些创造了众多公司和创始人的地方。仔细想想,Shopify 做到的规模比它们高出好几个数量级,而且我认为你们没有得到足够的认可,无论是你们创造的企业数量,还是你们改变的生活数量。
创业生态的塑造者
**Tobi Lütke:**那是因为我们不需要这个认可。这才是重点。
**Lenny Rachitsky:**有哪个平台是吗?不是。
**Tobi Lütke:**我们不需要这个认可。我的意思是,也许就算你想要,你也可能得不到,但我们不需要探讨这个,因为我们确确实实、从字面意义上就不想要这个认可。Shopify 是一家在背后推动的公司。我们不想被写进故事里,我们只想让你做你自己的事。想要在台前包揽一切,这种想法太自大了。这也是促使我们去帮助人们建立自己的东西,而不是从建立一个市场起步的原因,因为在通常的市场里,每个人只能以神奇地恰好等于你利润率的价格租用一个小板块。所以是的,我觉得我们不想要这个认可。我对此感到自豪,但那是内在的。我想我不需要太多的外在赞赏。有数百万、数以千万计的人每天都在使用 Shopify,这代表了商业,并且实际上在全球范围内创造了大量的就业机会,尤其是在北美和欧洲。所以我认为这真的令人欣慰。我确信可以说,如果没有 Shopify,我们许多商家可能依然会存在。
隐于幕后的哲学
我也认为你可以说有相当一部分商家是因为 Shopify 才存在的,因为我们观察到的情况其实是,这是整个 Shopify 旅程中我最喜欢的一点,也是我们严格确定下来的,那就是因为我真的、真的、真的把优秀的用户体验(UX)、清晰易懂的界面放在首位。你可以用出色的用户体验驯服巨大的复杂性,让它以人们能理解的方式呈现。我认为这在某种程度上对软件来说几乎是一种道德义务,因为当软件表现糟糕时,它会让人觉得自己很笨,而在我看来,机器不应该对人们产生负面影响。这颠倒了优先级的层次结构,原本机器是作为工具被人们挥舞,让人们在自己原本就擅长的领域变得比他们想象的更强大。这就是所有这些需要契合在一起的时候,但除了这个十年来的道德观点之外,你当然可以对我们的确定结果持相反意见,但我们的结论是,每一次我们让复杂的事物变得更简单,实际上就会有更多的商家在这个平台上存在。
降低复杂性的价值
所以我认为这很直观,或者至少指明了方向,但有时候人们会说,“嗯,难道你不需要拥有所有这些功能吗?”这就是软件世界里的 RFP 视角,对吧?“难道你不需要拥有所有功能,然后人们就可以执行他们的计划了吗?”其实不是的。想想创业者的心理状态,我的意思是,尤其是如果你是初次创业者,他们是毫无经验的。他们不知道该做什么。他们有点害怕。顺便说一句,再次投身于构建某种东西是一种纯粹的勇气,而且通常是一种很难对他人隐藏的行为。你周围的人会知道你在做这件事。去问问任何一位创业者,告诉你不要做这件事的人数之多实际上是令人震惊的。人们不希望别人仅仅是走出一个已经存在的、他们用来向自己解释周围世界的盒子,并且有可能达到更高的高度,因为那会在某种重要意义上否定他们自己的人生故事。
所以实际上已经有很多唱反调的声音在打击人们了,这并没有太大帮助。所以现在你面临的情况是,人们可能在劝你放弃,你大概也没多少钱,你会在心里问自己,我做的到底对不对?希望会有一些鼓励的因素。希望有一种热情。希望你是在努力创造某种东西。然后就会出现一种让你在某种程度上感到错愕的情况,比如软件突然开始谈论 API,这是你这辈子从未遇到过的事情,或者你以为在文本配置屏幕上应该有的选项,考虑到我们所知道的本地税收等情况,却根本说不通。接下来发生的事情是这样的。而 Shopify 的美妙之处在于,基本上没有税务配置屏幕,它就是正确的。软件的神奇之处在于,我们实际上可以直接接管这部分复杂性。我们知道各地的税收是多少,我们就是替你搞定了。你不需要去想税收的事。在这种情况下,如果有人遇到了让他们错愕的事情,让他们停滞不前。如果在倒霉的一天,这意味着他们可能会关闭浏览器然后说,“你知道吗?去他妈的,我不是这块料。”或者,“我用的这个软件太烂了,它太笨了,根本理解不了我脑子里的想法,”或者别的什么。每个人对同一件事都有不同的表现,但结果就是进步停止了。并不只是那些本来就不该存在的生意会以这种方式停止。事实上,这也是数据再次证明的,很多很多次,许多生意都非常接近这个临界点。我可以告诉你 Shopify 就遇到过很多次,而且我在创业旅程中拥有的工具比大多数人都多。我是个计算机程序员,我喜欢那些东西。你告诉我接下来几年我可以每天花14个小时编程,我会说,“我靠,走起。”
用户体验与创业者的困境
大多数人不是这样的。所以我拥有不公平的优势,这让我能够克服学习曲线上非常技术性的攀登,但我知道大多数人无法拥有这些优势,因此会在测试中流失。换言之,降低复杂性,打造优秀的用户体验(UX),创造能够自动处理税收或支付或任何这类事情、甚至欺诈的软件,实际上催生了更多的创业行为。这是我能遇到的、对我生命中最重要问题的最好回答,因为在此之前,做 Shopify 这件事在绝对意义上、在一百年的意义上会是有价值的,这只是一种直觉。在那之后我就知道了,对吧?所以这也意味着,现在看到那些过度复杂或根本不该存在的东西,会感到生理上的痛苦,因为我知道,就像所有这些试图取得进展、本可以存在、本可以现在雇佣员工并让客户满意的生意,就因为我们做错或做得很糟糕的事情而在半路死掉了。所以这成了我继续前进的巨大能量来源。
Lenny Rachitsky: 我对此深有同感,我能感受到那种痛苦,想到只要一个按钮坏了,或者按钮的颜色不对,导致转化率本可以更高,就意味着你没有创造出更多的企业、更多的创业者。你描述的那种不敢发布东西的恐惧感,我在开始写简报时完全体会过。我当时的解决方案是,“我就把这当成一个实验来发布。看看它的发展,别太担心。我可能偶尔写写博客。”这确实有助于降低心理预期,我能看到像这样的事情,以及能帮助人们克服那个障碍的建议,有多么大的作用。
Tobi Lütke: 是的,我认为这是在用户体验(UX)领域探索不足的一点。随便举个例子,我绝对是被惊艳到了,也许 TikTok 也有同样的功能,我不知道,但当时我在 Instagram 上发布一个视频,它允许我对这条 Reel 进行一次试运行。这似乎是一个新功能,就是说先把它展示给圈外的几百个人看,如果他们喜欢,它才会发布到我的个人主页上。我当时的反应是,“我不需要这个。”但我认为这是最……说实话,我直接关掉了 Instagram,心想,“我靠,这可能是遇到过的最深刻、最有洞察力的软件设计了。作为一个好点子的鉴赏家,我从未觉得有哪样东西如此有价值。”感觉这作为传统 A-B 测试这个概念的阶跃式升级,因为它太容易理解了,我就想,“世界上还有哪些东西应该有在圈外进行的试运行?”基本上所有东西都需要,不是吗?
这真是一件了不起的事情。就像我说的,世界上最罕见的东西,甚至不是创造力或天才,而是勇气。所以让我们降低所需的绝对勇气值吧。说实话,这绝对是那种你可以据此开创整个职业生涯或重塑一个行业的东西。如果你真的从第一性原理出发去追求它们、研究它们,而不是仅仅因为有了点洞察力,就想着怎么把我本来就想做的事稍微做得好一点,这些事情就会释放出难以置信的价值。后者当然也不错,人们也应该一直这么做,但这通常不是大部分价值最终落脚的地方,大部分价值是在新想法出现时才显现出来的。所以我这么激烈地谈论一个 Instagram 的功能,这有点奇怪,但我确实只是真的很高兴,因为也就是几天前刚发生的事。
Lenny Rachitsky: 我很喜欢关于勇气的这番讨论。你举的例子也让我想到,我觉得第一次在另一个城市使用交友软件也是这方面的一个例子,那就像是一场试运行,没人认识你。
Tobi Lütke: 但即便如此,难道不应该有个功能,比如,“嘿,把我的资料发到另一个城市,告诉我效果怎么样。”
Lenny Rachitsky: 这样人们就会喜欢。
Tobi Lütke: 我的意思是我基本上没怎么约过会,我的意思是至少在过去 25 年里没有。
Lenny Rachitsky: 对,是指做播客以来的 25 年。
Tobi Lütke: 所以我处于左右滑动约会之前的时代,所以我对这个的概念非常古老。也许我该多看看交友软件。我很确定大多数事情中都有一些惊人的用户体验(UX)。
Lenny Rachitsky: 我能想象。我觉得约会越来越转向线下了。我感觉这是人们现在想做的事。他们对滑动感到厌倦了,但不管怎样,扯远了。趁你还在这里,我可能还有几个问题。一个是关于人才栈的想法。你以前提到过这个概念,这也是我见你用来描述它的词,关于专注你独特的才华和好奇心,从而引导你走向最大的机会,尤其是在职业生涯早期。你能谈谈你在这方面的心得吗。
人才栈与好奇心驱动
Tobi Lütke: 是的,再说回来,我只活了一次,所以我无法对我做出的所有决定进行蒙特卡洛模拟,然后找出哪些最终成了承重墙,对吧?在这一点上,有一件不断发生的事情让我觉得相当有趣。但我会对一些绝对随机的事情产生好奇心,这些事情确实相当冷门,然后像变魔术一样,它成了一年后让我能够做出非常重要选择的方式。它最终成了一个我学到的更好的类比,或者一种看待事物的不同方式,或者我在这个专业领域发现的另一个想法,实际上也就是另一个基础思想的重新包装,这让我能够去寻找更多的例子。
就是这么有趣。所以甚至在职业生涯早期,我就只是追随我的好奇心。我喜欢编程,喜欢电脑,互联网出现时我也喜欢互联网,我就想,“酷,我要找一个我觉得有价值的任务,”我一直从事零售业,对于应该如何通过互联网把零售带给人们,我有很多话要说。但这正是我所有感兴趣的事物的一个美丽的交集。在此基础上,我发现了 Ruby,我很喜欢这项技术,然后现在我有极大的动力去修补、探索一个显然正在兴起,但在未来明显非常有价值的领域。
但我这样做不是为了追随金钱。我这样做是因为我喜欢通过做事来学习,我喜欢修补东西,所以这是一种通过卖滑雪板来为我的修补提供资金的方式,然后它引向了其他事情。我基本上一直都在做这种事。这非常有趣,对我来说非常有趣。我不知道这算不算什么好建议,但也许是一个可以证明它非常有效的例证。不过我有时会担心,这样的事情最终会有点像,你所需要做的就是买一张彩票,然后设定为彩票中奖者,对吧?所以,谁知道呢?
Lenny Rachitsky: 这确实引起了共鸣。这让我想起 Brian Armstrong 曾经分享过的一句话,也就是他之所以——
Tobi Lütke: 顺便说一句,他也是,如果你开始培养一群第一性原理思考者,你必须把他加进去。我认为有很多创业成功、成功的创始人、CEO,尤其是上市公司的,在这一点上惊人地相似,只是投射出一点不同,也许来自不同的背景,但 Brian 在这方面非常强[听不清 01:32:06],顺便说一句。
Lenny Rachitsky: 好的,这就是下一个被提名上播客的人,Brian。我们会安排的。但有趣的是,对于是什么让他能够创建 Coinbase,他基本上给出了完全相同的建议,他的背景是经济学、编程和密码学之类的东西。这就是那个韦恩图,就像是,“这就是我独特擅长的东西,并且有机会赢下它。”
Tobi Lütke: 是的。所以这正是我们非常积极地告诉客户的话,对吧?Kevin Kelly 在 2005 年写过一篇文章说,“在未来的互联网上,你只需要 1000 个铁杆粉丝。”这就是互联网,我认为也是 Shopify 所庆祝的,与其试图去制造牙膏,这当然是一个巨大的市场但很难差异化,不如想办法弄出三个不同事物的三重交集,然后把它做到极致。我和我的孩子们、家人去加勒比海度了个寒假,我喜欢和他们一起玩扑克。我喜欢玩扑克,这是一个很棒的不完全信息游戏。我认为这对孩子们很有价值,所以我找到了这个,我可能在什么地方提到过这个,但我找到了这个,我想它叫海盗黄金吧。
这是一个极其惊艳的扑克套装,筹码看起来就像是直接从宝箱或者《加勒比海盗》里拿出来的。想想看其中的交集,在线德州扑克,高品质的扑克套装,而不是……因为你本来可以用任何东西玩,对吧?市场上有人愿意花钱买这样的东西,而且是海盗主题的。你会觉得,“但我是个超级粉丝,我把它告诉你,人们可能真的会买,因为我对它充满热情。”热情实际上是最好的营销,而且他们至少有一千个铁杆粉丝,为一些人做出惊艳的东西,比做出每个人都想要或者能容忍的东西要好得多。我认为这非常非常好,所以我认为这也适用于你的职业生涯,顺便说一句,我们都是创业者,所以我们准备好思考任何职业了。
职业建议
如果你真的要我给职业建议,那就是我们都是产品,我们在从事创业,如果你愿意的话,我们基本上是软件即服务,或者是人才即服务(talent as a service)。我们以所谓的雇佣形式在出售订阅,但坦白说,这实际上只是重新包装的订阅,因为它带有一些额外的保护,这非常好,但它仍然是同一回事,它是一个双向选择的公司。就像你会说,我能做这些事情,我对公司的价值将远远大于公司花在我身上的钱。理想情况下对公司来说有正向的投资回报率,然后公司要么说“我同意”,要么说“我不同意”。这实际上就是雇佣关系。所以我认为人们应该把自己看作一个产品。你的职业生涯不是基于你找到了什么导师,或者是否得到了晋升。
关键在于你在哪方面好到让人无法忽视?你能有多好?你能做到绝对出色吗?我的意思是你不必是世界级的什么的,但你必须……理想情况下你要找到这样一个东西,你处于大约五个事物的微小交集空间里,或者就是我们比其他任何人都懂得多,然后大家都会来找你。就像我说的,我们使用 Ruby。Ruby 运行得不是那么快。Shopify 本来希望它非常快。我们找到了那些在动态语言、动态语言垃圾回收、即时编译方面获得博士学位的人,来让 Ruby 变快。他们在 Shopify 过得很开心,Shopify 现在变得非常快,这非常好。因为我们需要这个,每个人都从中获益了,因为我们把它合并回了 Ruby Core,现在每个人都可以拥有一个 JIT 编译器(JIT compiler)。所以这太棒了。
Lenny Rachitsky: 我觉得每一个故事都像是可以延伸出其他几小时播客对话的分形。我知道你得去运营这家九千、一万、一万人的公司,去做实际的工作。所以我就到此为止。我们几乎聊完了我希望能聊完的所有内容,但我还有无数的问题。也许为了结束这次访谈,你有什么想留给听众的吗?有什么最后的小干货是你可能还没提到的吗?
产品管理的核心
Tobi Lütke: 如果我能再次和那些关注产品和产品管理的人谈话,这就是我想说的,我认为这可能在我们刚才所有的谈话中只是隐晦地体现出来了。世界上每一个产品,归根结底,其质量仅仅是创造它的人对产品有多在乎的反映。如果做产品的人不在乎这个产品,是不可能做出伟大的产品的。我实际上认为这是产品领导者一个非常重要的角色,要确保团队在乎。我认为这可以通过为使用它的人建立同理心来做到,但也可以由产品领导者具有传染性地做到。产品领导者必须在乎。不要参与你不在乎的产品工作,因为你无法生产出给你派发任务的人所寻找的东西。
我认为再说一次,有太多东西进入了这个无法量化的对话中,我们已经讨论过了,但这太重要了,因为它不仅仅关乎简报文档,不仅仅关乎对齐利益相关者。有时在某些地方,其中一些事情是它的一部分,我认为它们有时与人相关。但我认为更高阶的位是,如果你在做产品,你基本上有免费的道路。你必须让团队坐下来,看到他们看不到的角落。你只是必须比其他任何人都更好地理解正在做的这件事,因为这是一个角色。现在,你不必自己做这件事。团队里的每个人都是你的资源,你可以向他们提问,从而确定角落后面是什么等等。无论是在工程还是用户体验(UX)等方面,但这非常重要。其次,你必须对真正在乎这件事表现出放热的传染性,因为仅仅这一件事就能让产品好十倍。这种改变有多大简直疯狂。
创始人模式
Lenny Rachitsky: 老实说,我觉得你在这里描述的就像是创始人模式(founder mode)的产出。在乎,表现得放热且兴奋,并且看到角落后面。我觉得这本身就可以单独做一期对话了。
Tobi Lütke: 是的,我同意。我不认为创始人模式,我的意思是,我认为,是的,创始人模式是一个非常有价值的术语,但我确实认为我在关于向系统中注入热量的讨论中,回答了关于我对创始人模式看法的问题。那就是我做的。只是在创始人运营的公司里,创始人模式更容易存在,因为人们不喜欢那些向事物中注入热量的人。人们会与他们对抗,人们可能会把他们赶出去,如果老板是这样的,老板就可以保护其他像这样的人。因此我认为创始人运营的公司可以更长时间地保持创新,因为有一个基底可以让这种情况长期存在。所以无论如何,这就是我对创始人模式讨论的拆解,我认为这是一个非常迷人的近期讨论。我觉得它非常有价值。我确信它在各种方面被误解了,但我认为它比人们认为的更具可操作性。
Lenny Rachitsky: Tobi,我很感激你继续抛出干货,即使我知道你得走了,顺便说一句,你的视频现在看起来棒极了。这剪辑太棒了,我们就像在做一道彩虹从你身上流过的特效。
Tobi Lütke: 是的,我们真的在这里击中了彩虹,“走吧,去发明。”我希望你得到了你希望的黄金,然后……
Lenny Rachitsky: 太多黄金了。
Tobi Lütke: ……我这个充当小矮妖的在这里把它送达了。
Lenny Rachitsky: 多棒的比喻。好的,Tobi,非常感谢你做这期节目。这太不可思议了。我觉得这将帮助很多人建立伟大的公司,从第一性原理感谢你。
Tobi Lütke: 好的,我得跑了。拜拜。
Lenny Rachitsky: 好的,拜。
术语表
| 原文 | 中文 |
|---|---|
| alpha | alpha |
| API | API |
| Archie | Archie |
| Archimedean logic | 阿基米德逻辑 |
| autoregressive models | 自回归模型 |
| axiom | 公理 |
| Boolean | 布尔值 |
| Brian Armstrong | Brian Armstrong |
| Chad | Chad |
| cohort | 群组 |
| composability | 可组合性 |
| conversion rates | 转化率 |
| corporate marshmallow test | 企业的棉花糖测试 |
| Daniel | Daniel |
| devil’s advocate | 魔鬼代言人 |
| Elon | 埃隆·马斯克 |
| exothermic | 放热的 |
| fine-tune | 微调 |
| First principles | 第一性原理 |
| fitness function | 适应度函数 |
| founder mode | 创始人模式(founder mode) |
| Glen | Glen |
| Glen Coates | Glen Coates |
| Goodhart’s law | 古德哈特定律 |
| holdouts | 对照组 |
| IC | 独立贡献者(IC) |
| infinite games | 无限游戏 |
| iterated prisoner’s dilemma | 重复的囚徒困境 |
| James Carse | James Carse |
| JIT compiler | JIT 编译器(JIT compiler) |
| Kevin Kelly | Kevin Kelly |
| Lenny Rachitsky | Lenny Rachitsky |
| local maxima | 局部最大值 |
| loss function | 损失函数 |
| monorepo | 单体仓库 |
| neurons | 神经元 |
| overfitting | 过拟合 |
| path-dependent | 路径依赖的 |
| proxy | 代理指标 |
| RFP | RFP |
| Sarbanes-Oxley | 萨班斯-奥克斯利法案 |
| Simon Sinek | Simon Sinek |
| sunk cost fallacy | 沉没成本谬误 |
| talent as a service | 人才即服务(talent as a service) |
| talent stack | 人才栈 |
| Tobi Lütke | Tobi Lütke |
| unit test | 单元测试 |
| UX | 用户体验(UX) |
此文档由 AI 分片翻译(translate_long_document)
Tobi Lütkes leadership playbook: First principles, infinite games, and maximizing human potential
Tobi Lütke (00:00:00): Your podcast is a podcast by a builder for other builders. Here’s the most interesting question I think people can ask builders, what is your energy source? My energy source is dissatisfaction with status quo. There are so many books are about this … Technology leading to dystopia. Like no one who really thinks about this would want to be born into a world 20 years before today. I think today is the dystopia of the future. It behooves us to try to build the kinds of products that leads in … Towards progress.
Dissatisfaction and Drive for Progress
Lenny Rachitsky: There’s a couple quotes along these lines I’ve seen that describe the way you think about this stuff. “If most people are doing it a certain way I by default don’t want to do it that way.”
Tobi Lütke (00:00:34): There’s an aesthetic in the world that exists which is that business people dress in suit and tie, they are speaking much more sophisticated than I do, usually without an accent. They usually have a stick and show dramatically at the chart that is behind them. How much is that aesthetic overlapped with outperformance? Pessimism sounds extremely sophisticated. Optimism always sounds dumb or at least naive. The most powerful unquantifiable things in the word of business are fun and delight.
I don’t know of any other company that operates where the founder has this 100-year vision of where the product needs to go and working backwards from that.
Tobi Lütke (00:01:11): I talk about look in the future and then think backwards a lot, right? It’s like what would we want to have done 20 years ago on this? We have very long-term plans. At 100 years you can’t talk about this software project but you can talk about the mission itself, whatever things that will survive for 80 years that are left on this particular timeframe. Entrepreneurship is just precious. Shopify exists, basically, to make entrepreneurship more common.
Is there anything you want to leave listeners with?
Tobi Lütke (00:01:37): I really, really, really think that there is not a single person on this planet who is even close to being at their maximum potential. Reminding people of their own potential constantly is actually a wonderful thing to do.
Tobi Lütke (00:04:24): I’m glad to be here. I’m excited for our conversation.
I’ve listened to so many of your other interviews, I’ve talked to a bunch of people that work for you. I want to try to do something a little different. There’s basically two themes that emerged over and over and over as I’ve listened to you share advice, and interview, and in talking to people that work for you. One is thinking from first principles, the other is maximizing human potential. I’m just going to plant these seeds for now. I’m going to not ask about these directly, I’m going to come at these from the side with the many questions that I have for you. Something that I’ve heard people describe at Shopify called the Tobi tornado.
Tobi Lütke (00:05:01): Oh, wow. Okay, that’s a start. I like it.
What is the Tobi tornado?
Tobi Lütke (00:05:08): A Tobi tornado, I would say, is a whole lot of change management or a conversation or conflict or real talk compressed into a very short timeframe. I see something it doesn’t … It’s not good, I have a conversation. I have learned something very quickly about hey, I need to update my priors, or cool, let’s do it differently. At which point a project might be stopped and we get back together in a room and then we start a new version of a product. And everyone who’s currently on the team, of that particular project, is no longer on the project but they are the founders of the next version which is built differently. And that might be a bit whiplashy for people. I mean, I certainly hope that’s true, it’s certainly what people tell me. It’s also what they appreciate about the company. It’s like what is best ends up mattering a lot.
Basically, it’s you going into a chat room being like “Hey, this” … “We’re going to end this project we’re” … “Let’s try something else” because you’ve discovered or realized this is a bad idea. Some people complain about this practice of like “Oh, Tobi kills the we’ve been working on.” My lens is, you realize we’re just wasting time on this thing that is not going to work and we shouldn’t do it. Is there anything there along those lines?
Tobi Lütke (00:06:32): No. This is everything. Again, once I imagine something might be not the right thing to work on I’m either incorrect at which point this is super important that I understand why or I’m correct at which point it’s super unfair for letting people work on something that isn’t going to make it. There’s a third wave which is I could also ignore it but that’s an abjection of my CEO and founder responsibility that I’m absolutely not willing to make so that’s just not a path forward that I see valid. I understand that’s what a lot of people choose to do. So yeah, compressing time is important. We have a fairly limited time in our careers, right? Our careers are not that long. If you’re lucky you have 40 years in the industry. Most people spent more time in school and then maybe leave later if they’re so lucky that they can. So it’s not even that.
I think you want to do the maximum amount of things you can be proud of at the end of your career. When you look back you want to be saying, “Hey, holy shit, we shipped this thing which was absolutely an incredible contribution to a mission I cared about at a company that was full of other people who cared as much as I did.” And also are very proud of working. And maybe even thinking of working with me and being really, really glad that we spent time on projects together. Yeah, none of this happens if everyone’s sort of optimizing a thing that probably shouldn’t be there, right? And, therefore, I think it’s the better thing to do.
100-Year Vision and Reverse Thinking
Lenny Rachitsky: And, by the way, I love this way of describing … Of compressing it in time to just make a decision, not focus on making it come across the most kind, nicest, sweetest way. When you give feedback to people when something is not the way that you think it could be or as good as you could be it’s often very direct and often hard to hear. To me this comes across as you’re trying to maximize their potential, you’re trying to push them to do something better. Is there anything there that you think is a way of approaching feedback?
Tobi Lütke (00:08:49): I really, really, really think that there is not a single person on this planet who is even close to being at their maximum potential. I just think everyone is way, way, way, way, way better than they think. And the reason why we’re not performing at this level is a series of ideas, maybe certain approaches for cultivating our skills and our crafts that have not yet been discovered and, therefore, we could not take advantage of them. It’s an environment that just narrows the focus on fairly unambitious things at which point you get stuck competing with literally everyone else in the world because everyone’s unambitious.
I have found that reminding people of their own potential constantly is actually a wonderful thing to do. And I have a history of being right about people’s potential more than they are themselves. Now, in a way, this dooms me fairly often to be disappointed, right, in myself. By the way, I’m talking about myself here too. I think I have way more potential than what I bring to bear and I hate that so I’m trying to cultivate the skills that I need for tomorrow and constantly challenge myself. I’m harder on myself than on anyone else. By some discount rate act equally to the people around me, especially the ones who just so, obviously, are brilliant.
Spending time and longer time in careers with people and then holding them to a high standard means that they accomplish very often things that just they didn’t imagine they could. To me this is the most wonderful thing to see. And frankly this is a throughline for all of my career because this … My product is that. I want my product to cause people to be more successful than they thought they could. And, in fact, become more ambitious about what they are building with their online stores and their businesses than they are actually initially set out to do.
Because something like this happened to me, right? I started a snowboard store at some point. I didn’t set out to build Shopify. If you are committed to following your curiosity as to the next step, and optimize for maximum amount of learning when you choose these steps, it takes you from one place to another and you actually realize the world’s full of lies about human potential and progress. Maybe people are not malicious about it but they’re definitely confused about it. School teaches you that you have to learn this particular piece of math in this 12 month period, and it doesn’t matter how much you understand it. It’s like the outcome can be variable and will grade you on a variable outcome for a fixed amount of time which has nothing to do with anything I’ve ever seen or learned or see … Or witnessed about how to actually learn things. You follow that thread and you just find that there is no speed limit for personal growth. In a way Shopify has been a wonderful experimental lab for this sort of conviction. I’ve just seen this to come to be true. And, of course, hearing from someone that you respect that “Hey, I think you had it in you to do this thing significantly better because I think you probably saw fairly early in the project this sort of path A path B. You chose path B potentially out of convenience even though you knew that wasn’t the right thing. And I actually expected better of you and I expected” … “I think the next time this happens in your career you should go path A because” … “Based on your conviction.” And, therefore, that’s hard to hear, right, because it’s right. But it’s also extremely valuable, right? What I love is an environment of people who are holding each other accountable to the actual potential rather than sort of their current level plus or … Plus some, I don’t know, a little bit extra.
There’s so many threads I want to follow here. The example of the school is such a pertinent one. To me right now we’re looking at preschools for our son and I’m … They’re describing their education philosophy I’m like, I don’t know why I should believe this is the right approach. And it makes me just want to spend all this time researching what education approach works. I know it’s just preschool and maybe not as critical yet.
Tobi Lütke (00:13:38): Did you have a bet yet? I have three kids too and this is sort of a decision that every parent faces, right? So many of your listeners are probably product managers of machine learning products that maybe this resonates. So there’s a funny thing about machine learning which you’re just like … You train on a lot of data, and, hopefully, you get something that predicts the thing you want it to predict correctly out of it. The biggest problem of this is overfitting, right? What does good look like, a loss function? Which is a heuristic because it’s not the actual task that the thing will do in the future it’s something that proxies to the task that you want the thing to do in the future. Predict fraud, predict the next word, whatever. So overfitting is basically model learning how to cheat on the benchmark or on the fitness function.
So there’s a business analogy of this which is that … It’s called Goodhart’s law. It’s literally the same thing as overfitting just for businesses. Goodhart’s law just says, “Any metric that becomes a goal ceases to be a good metric.” Same exact thing. The universal truths are things that almost any competitive field will invent for itself by different terminology often. And I think this is also, by the way, why it’s so interesting to focus on personal growth and learning a lot about a lot because you end up finding these sort of hidden harmonies behind things, the things that are clearly enduring correct insights. So overfitting, Goodhart’s law are the same thing. School optimizes for what? Marks supposedly, right? In fact, overfitting in school is literally the kids cheating to get marks, right? You get another analogy. What is however the right loss function for children? Have you made a decision yet?
No, I have not we just started down this path. You told me-
Tobi Lütke (00:15:55): It’s the kind of thing [inaudible 00:15:57]. You have to actually go fairly deep in philosophy to figure this out. And then again afterwards you can build, you can find the schools that you like. For us it was just maintaining curiosity. This is a completely different goal from being good at marks. But I just think everyone’s born extremely curious and school has a habit of getting it out of kids. Literally, there’s a foundation model of a child, and you fine-tune it at school, and it just loses the neurons of curiosity because it’s actually discouraged to meander into other topics and explore them just because they’re interesting. I don’t know. This is sort of not the beat of a podcast but I just think about this a lot. It’s funny how these things just recur constantly.
Podcast Opening and Topic Introduction
Lenny Rachitsky: So when Archie was on the podcast, he’s the head of growth at Shopify … I don’t know if that’s his official role. Basically drives a lot of the growth. He talked about how the core product team, outside of the growth team, operates without KPIs, without specific goals. And decisions are driven by taste and intuition primarily you, and Glen, and some other leaders. And a lot of people heard that and they’re like “I” … “First, I don’t believe that. Second of all, how does one operate that way when there’s no data to tell us exactly what is right and good?” So the question I have is just how does one operate in that way successfully? What does it take for a company to work that way because a lot of people will try it and fail?
Tobi Lütke (00:17:25): This is very close to what I just talked about before with [inaudible 00:17:28]. Goodhart’s law is real. The moment a metric becomes a goal it’s no longer a useful metric, right? I think that’s more or less a precise wording. Why? Because no metric by itself is a complete heuristic for a complex business because business are complex. There’s a million of different tensions in a company and you can’t all keep them in harmony by optimizing for one fix. It’s true that we don’t have KPIs and we don’t have at least OKRs in the Silicon Valley sense but we are extremely data-informed. We have invested enormous amounts of money and time into systems that give us basically everything at our fingertips. I sent this demo to other founders and they’re completely bowled over by the way we can dig into basically every constituent atomic bit part that makes up the cohort that just got formed 15 minutes ago by the end of a … At the end of a quarter or month or week.
In a lot of different places, this is one of them but also in its products, it’s just not overfitting for the quantifiable. Everyone competes for everything but it’s highly quantifiable because it’s … It’s fun, it’s like a game. You tweak a number and 0.1 more is better than 0.1 less. That’s an immediate gratification thing. But I just think the overlap of most valuable things you can do with a product, and for things that happen to be fully quantifiable, it’s like maybe 20% which leaves 80% of a value space unaddressable by the people who will only look at quantifiable things. Shopify is comfortable with the unquantifiable things such as tastes, quality, passion, love, hate. It’s with the strong emotions that people have.
The sort of deep satisfaction that a craftsperson feels when they’ve done a job well is actually better proxy if you allow it to be then the … Do a unit test pass. A unit tests might not pass. And the unit tests will pass 15 minutes later because we already fix them or adjust the one or two things so they support us. We have systems that tell us exactly if something goes the wrong way. There’s an extremely sophisticated rollout system in Shopify that forever holdouts and correlates everything with everything for … In every experiment and so on and so on and so on. But if you think about it as a cockpit for a pilot. The decisions are still made by pilots and we think this leads to better results. It’s just the same with our product. There’s plenty of A/B testing tools and all these things for commerce, and it’s, of course, really important to figure out what your conversion rates are. But are you representing your brand is an unquantifiable question? Are you proud of the thing that you have built? Do you feel it’s your own, right?
And so I think there needs to be more acceptance in businesses or for unquantifiable things. The most powerful, not unquantifiable things in the world of business are fun and delight. If people have fun when they’re doing something that is just upstream from so … Sorry, downstream from so many other things. I think that if all the metrics are pointing down but everyone says, “My God, I’m having so much more fun,” I think that the very next thing that will happen with some time delay is all metrics will start going up. And if that doesn’t happen then we adjust course.
The reason why we specifically don’t have OKRs and these things is because … If you want to hold the unquantifiable as things that are stable and exist … That people actually do really defer to them and really actually learn to be okay with someone just saying, “Hey, this is actually just really great and they’re shipping this” then you need to make certain edits to a business that don’t remind everyone too much of the companies they might have come from which the only way to get promoted is by driving the metric up. It’s a bit of a [inaudible 00:22:24] conversation I suppose. Good on the fortune cookies saying, “Shopify doesn’t do OKRs or doesn’t do metrics” and so on. But it’s actually just because the metrics take us [inaudible 00:22:37] function where we often defer to just more … Sometimes a little bit emotional but generally less quantifiable things.
I imagine if someone were to hear you describe this of focus on joy, and fun, and love, and delight, maybe … It’s easy to dismiss that.
Tobi Lütke (00:22:58): It sounds completely idiotic, right? Again, there’s an aesthetic in the world that exists which is that business people dress in suit and tie, they are speaking much more sophisticated than I do usually without an accent, have a full head of hair. They talk about metrics, they are in front of PowerPoint presentations, they usually have a stick and show dramatically at the pie chart that is behind them. And highly charismatic, highly … So that’s our aesthetic. How much is that aesthetic overlapped with our performance? I don’t know but some of them … Some people pull it off who are like this. I think the world is sort of stacked to lead us astray based on our stories about what optimal looks like are just so incorrect in so many ways. Optimism always sounds dumb, or at least naive. Pessimism sounds extremely sophisticated. Metrics driven sounds extremely sophisticated. Talking about fun sounds like naive.
Well, first of all, I’ve always ignored what people think generally. That came pretty natively to me somehow which I’m very lucky about. But I’ve now actually learned that almost all of the alpha in the world is now in the … Exactly the things that are unobvious but true. And the things that people dismiss as naive or so. The most successful business person on planet Earth is Elon and he conforms to no idea of what the most sophisticated business person ought to be like in any which way you can imagine. I think we live in a world where the counterfactuals are winning because our … Because aesthetics are just leading us astray.
”Tobi Tornado” and Rapid Iteration
Lenny Rachitsky: This is an awesome segue to the other theme that I wanted to spend some time on which is thinking from first principles, Elon is the classic example of that. Honestly, I think you’re the other most classic example of that these days. And we’ll keep talking about all the ways you operate very differently from other companies which are examples of this. But I want to read a quote from Glen Coates, he shared with me, of how he sees you that gives an interesting lens into your first principles thinking. So here’s what he said about you. “Tobi is at his heart a true futurist, he’s obsessed with the way things should be in the future. Being data-driven is innately being anchored in the way users and technology are behaving today. He’s never really said this to me explicitly, but knowing him I think any design that is drawn primarily from the way things are or were is one that he sees as inferior to one that is skating to the puck of the way things could or should be.” Does that resonate?
Tobi Lütke (00:25:55): Yeah. I mean, I think that’s correct. That’s actually really interesting. Your podcast is a podcast by a-
Tobi Lütke (00:26:01): Your podcast is a podcast by a builder for other builders. Here’s the most interesting question I think people can ask builders is like, “What is your energy source? Where are you getting energy from?” I think fundamentally the world exists at room temperature. Almost all companies are running at that, humming along, doesn’t do anything. There are certain individuals who can inject heat into businesses. Founders do this very well. All the startups anyone’s ever heard of have people who are injecting heat because if no one would inject heat into the business, at room temperature, you cannot outperform anyone else. You can’t be hotter than everyone else if no one’s injecting heat into the concern. So fundamentally, there is a injection of energy into companies that comes from founders and the best leaders, like all the people you’ve had on the podcast from Shopify have a perfect set of cast of characters of people who are just exothermic. They are just like wellsprings of energy that leads to all the amazing results that we get to enjoy.
So the question is where does energy comes from? And that’s another one of our discussions which very quickly goes into the emotions. Actually, there’s a really… So I watched The Last Dance, the Netflix special of Michael Jordan a while ago, of course, but there was one scene where he just, I’m sure this is a super famous story and he just made up an insight that someone told him so that he would then go and just want to destroy them afterwards, which he then of course proceeded to do because it’s hard to imagine anyone more exothermic than him. So we know what his energy source is. It’s rivalry. It’s potentially it’s insight or it’s anger, something like this.
I am not… My energy source is dissatisfaction with status quo. My fundamental belief is all this talk about technology where all… So many books are about just technology leading to dystopia. You know what dystopia is? Today, compared to what it will be in 20 years ago or any. And you can play this for any part of human history. I’m not making a future statement. I’m making a almost totalical statement about the experience on planet Earth. No one who really thinks about this would want to be born into a world 20 years before today rather than today. And so I think today is the dystopia of future, and I think it behooves us to try to build the kinds of products that lead in towards progress in a small way or a big way. But yes, I think if someone comes to me and says, “Hey, let’s go do this thing. And we’ve looked around and here’s how people solve this problem, let’s make a good version of that,” I’m like, “That was not the job.” Because everything that you encounter, that every solution, every product, everything that exists is path-dependent, highly, highly, highly path-dependent, and often path-dependent based on having to make compromises, based on things that were true at the time a decision was made but are no longer true.
The entire field of… What was it? I forgot the name of a field. Chomsky’s field, the linguistic research field. It’s cool. We now have autoregressive models that are just like, we don’t actually need to set up a complete… We don’t need to research the structure of grammar to be able to make machines also engage in the spoken word. We actually can just train on the internet, it turns out. So that was not possible back then because you didn’t have the right architecture for this, but now it is. So I think what you have to do is to actually have, when you come up with a new product or you discuss a new product, you have to derive it from first principles. You have to say, “How would we solve this problem given every fundamental building block that we have available right now?”
For that, to do that, you actually have to understand the power and the composability of all the building blocks that exist right now, which is a tall order and no one is perfect at this. But so this way, you go ahead and say, “Okay, cool, so this is how we are implementing this thing. This is how it will be implemented today.” And now we can talk ourselves in taking shortcuts. “Maybe we should actually start up doing it the way everyone else does. Maybe we derived exactly what everyone else does as the correct thing to do.”
Sometimes there was a lot more wisdom encoded in the status quo than you expect, which is I think is super delightful. Then you figure that out and so when you act on it. But what isn’t okay is skipping the exercise and doing the same thing everyone else does because that is again a abdication of product leadership. And so yeah, I would say I become extremely suspicious if I get a pitch to do a good version of the same thing everyone else does because I just find that in our space specifically very rarely to be the best solution.
There’s a couple quotes along these lines I’ve seen that describe the way you think about this stuff. “If most people are doing it a certain way, I, by default, don’t want to do it that way. And if you want to do something world-class, you can’t do it like everyone else.”
Tobi Lütke (00:31:46): Yeah. I don’t even think that’s an opinion that I hold. I think that’s actually… We are basically in axiom territory here. If you want to do something better than what exists, you have to do it differently. That does not make a statement about if it will be better in the end after you do it. It could also be worse. But you can’t get something better done if you do the same thing. It’s like axiomatically not possible to do. It’s fails Archimedean logic, yet it’s something… You would be amazed how many business plans are actually failing Archimedean logic in this way like, “Let’s do a good version of this thing that we’ve already been doing and we will capture 1% of the market,” and just like this stuff. It’s like, “Trust me.” I find it kind of cute.
Is there an example that you can share of you approaching the problem this way? I imagine it’s constantly happening. You also mentioned you’re born this way. So I think it’s hard for someone to just sit down, learn, think the way Tobi thinks, but I’d love to help people start to approach problems this way, so maybe an example might help.
Tobi Lütke (00:32:54): I think this is entirely learnable, I think. And so I encourage people to just have a practice of think step by step essentially and just do it. It’ll become a habit pretty quickly because it just, it also just outperforms.
Examples. The very first example is Shopify itself. It’s cool. So there was lots of e-commerce software and it was all the way it was because of path dependence because everyone who want it in 2004, 2005, e-commerce was an existing retailer and therefore they had complex businesses that needed to be ported online, including all of their somewhat Byzantine business logic. “I wanted to make e-commerce software that would do very well on the internet of the future and I believe that we can make it easier to start new businesses online than it is in a physical world because the physical world is encumbered by a lot of regulations and also upfront costs for leases and so on. So let’s optimize for that case and build something that is so intuitive to use that frustrated people in dead-end careers can spend their lunch breaks making progress towards building their own business, which then eventually allows them to do it on their own way.”
And so being fortune was biased, eventually it turns out to be a much better prep for also solving all the enterprise cases because no one had to… Enterprise software is overfit to the sales process, which is that it wins the RFPs because it has every feature ever or at least a way of putting a checkbox next to every RFP line ever. But they’re not good, right? An RFP is a great example of overfitting in the world of procurement because it tells you nothing about the quality of software behind it. But honestly, this happens all the time. It is just like here, we are setting the stage for much more, much higher quality retrieval for the products on Shopify and across Shopify. I think we’ve been in a local maxima on search and we think that, especially with the advances of the new models, certain things are now possible to do that could not have been done yet because again of these layers and layers of path dependence and no one coming and saying, “Is that the best way to do and maybe we should rebuild this component periodically?” We can now do a better job if search that will leads to much more delightful experiences. And so this is a fun project that this happening right now in.
Potential as the Product
Lenny Rachitsky: I think this is really interesting because what I’m looking for is like the Tobi algorithm of first principles thinking. Elon’s got these famous ways of thinking he shared. One is start with the cost of metal to help you understand how much a rocket should cost. And then he’s got this five-step. First, decide, do we need this thing? Then figure out how to optimize it, then automate it. What I’m hearing so far, and I’m curious if you’ve thought about this? And if not, this feels like a really good blog post in your futures, the Tobi first principles algorithm, but I’ll share a couple of things I’ve heard so far as you’ve described it.
One is analyze the path that existing solution has relied on, almost like the assumptions that were true for it to be built back and it was built. And the other is this overfit, what is it overfit for? What is it over solving that maybe isn’t necessary? Is there anything along those lines of just how you approach problems?
Tobi Lütke (00:36:43): This is probably too nerdy technical. You are right, but I should figure myself out a little bit. My brain runs on a meter language but isn’t directly something I can translate into words, I suppose. I think more about things like this in terms of programming constructs, pure functions, overstate. And I think in any moment, the best decision to do is I think the perfect product lead is almost like a thermostat for high quality product. It’s like you’re setting, saying, “I would like to build something really, really great and I’m going to go through a series which is much, much more complex than what a thermostat does,” which basically checks the temperature and then makes a decision of air con or heating. You make you re-derive literally every decision that is valuable, every foundational assumption, every foundational ABC direction. And you want to see the observation you’ve made in the meantime since you last derived the next step. Re-running the entire function over the state that is now updated, the higher fidelity information, would you come to the very same thing?
Sometimes fairly early in the construct in the tree of foundational assumptions, change is made. An example we all had was beginning of COVID when we suddenly had shelter in place. So Shopify has that incredibly good office spaces and we were very in-person company and we’ve… Our floor plans because I think we really added something to the understanding of how to put great… I got a lot of founder energy from my co-founder, Daniel, to build great collaborative spaces for creative work with lots of happy accidents, people running into each other and so on.
Anyway, we were very, very, very, very determined on doing that. But somewhere in this construct of nesting functions, you have to rerun and foundational assumptions. It’s in the stack is the fairly basic Boolean of are people allowed to leave a house? Which was yes. The moment that flips, it’s not that just like, “Okay, over here let’s do the best what we can do.” It’s actually that the entire tree now moves into a different place. And it can be a very far place, different, that you land because you will make the same quality of decision on every step, but you need to rerun something that takes you to a completely different landing zone. And so then this is also easy for us to say, “Cool, we are going to be remote only forever? Let’s go.” And because we realized that the temporal shelter in place would cause a series of events that would make that for optimal best set of trade-offs for the company. And so I think this is… Can I put this into a piffy five-step? I wouldn’t know, but maybe it resonates with someone who can help me figure out the coded English language for this because it has been nerdy way the I explained it, even to me.
This remote work example is something I definitely wanted to touch on, which is I’m glad you got there. So in this decision, is there anything more you could share about how you got to that place of like, “Oh this Boolean changed so we should rethink this?” Because I know you probably, you weren’t… I don’t know. Were you in the shower and just like, “Oh wow, we should really go remote because of this”? How did that actually come about? And then I wanted to ask on another part-
Tobi Lütke (00:40:27): Yeah, pretty much actually yes because again, I rerun over all the inputs and figure out what happens. And here’s a couple of other things that were starting to fray on the decision to be in person as well. We started in Ottawa, Canada, which is a city of a million people, and it has some tech heritage and good universities. But at a million people, it’s just not population dense enough and has a depth of talent pool that it can support a company that is going to 10,000 people. So anyone who runs this optimization function I’m talking about here about location strategy for a new startup will come to the conclusion if they do it right that everyone should be sitting around the same table. Only if you can’t do that, do you say, “Okay, let’s be all in the same couple of rooms.” Only if that doesn’t work anymore, you say, “Okay, let’s be spread over one floor.” If that doesn’t work, in the same building. If that doesn’t work, in the same city. If that doesn’t work, at least stay in the same time zones or make sure that there’s good hub connections between.
This is how it works. It’s just like some assumption somewhere along the line is invalidated. You end up in a different side of a decision tree. But what was happening to us was we already were in four or five cities. We were adhering to the same time zone at this point. But my experience was I went through, sometimes at the Ottawa office, through an entire day where I had 10 hours of meetings, which is fairly normal, but I do every single… I don’t think I had a single other person sometimes with me because every one of them was with people in another office. Some people are dialing in remote and so on. So there was an awkward hybrid-ness which we ended up in making only good decisions along the way.
This is actually the most dangerous thing. Most of the time you end up in a bad part of a tree, in a local maxima of a path-dependent environment, by only making good choices. People think that making a good choice inoculates you from making mistakes or that the presence of a downside of an idea ends up disqualifying the idea. Both of those things are incorrect. So what you need to do then is… So that was an overlay to decision.
The moment COVID started, then we also had this thing of shoppers exploding because we were actually an asset to people during COVID, e-commerce for the local businesses. And we took that very, very seriously trying to make more businesses survive COVID, small businesses survive this particular calamity than otherwise won’t which is important because small businesses tend to be wiped out first anytime the times turn fragile. And so we needed to staff up. And so there’s a very weird question about where to staff up. So clearly the better input there would be if you could hire people everywhere. And so once that decision flipped, you can see how this is actually now super easy to say, “We are going to remote,” because there is no turning back. There is no… Even with this decision, it’s a better set of trade-offs for the future is to accept the vast additional difficulty of building a remote company. It’s way harder.
It’s not something you should recommend to anyone to do this because it’s the same as trying to run a world record marathon run in Aspen, Colorado. It’s like there’s not enough oxygen up there to do that. But if you end up pulling it off, you’re a real Chad. That’s very cool. So I find difficulty itself interesting. And again, I spent enough of my teenage years on the internet to know that there’s amazing cultures that can be put together purely remote like, I don’t know, Wikipedia, World of Warcraft, Raiding Guilds can at least have excellent cultures. And so we are like, “Okay, cool. Let’s come to a new stable part.” That stable part was don’t port the office online. Let’s port the internet into a company and then we are like, “Let’s go.”
Education and Overfitting
Lenny Rachitsky:
Tobi Lütke (00:46:46): It’s so funny that that’s the way the story goes. It’s right, but I don’t know why that is surprising. That’s my happy place, being able to clear out free days of my calendar and being there till midnight with all these remarkable people who join us on journey, just building stuff. It’s like, I do my job so that that is the jobs that exist for other people. That’s a job I actually want. That’s the one I couldn’t find because no one did a company for me. But at least back in most days, I think still not really.
And so I just love coding. It’s one of the greatest, I don’t know, hobbies and pursuits. I’ve done it for a very long time. I came across it very, very early. It fits my brain like a glove. I appreciate so much of the craft behind coding. I am a trained apprentice in… Sorry, I’ve apprenticed as a programmer in Germany, which has a dual education system that you can do such things. So I’ve been professionally in companies spending all day. And I really mean it, programming ever since I just turned 16. I think first day was just before I turned 16 when I started my apprenticeship.
And so I love it. I have a view on my screen left of you is a cursor right now, which is opened to a Juniper notebook where I’m working on some projection stuff that I’m playing with. I try to sanity-check as much as I can of what I get. I try to find… It’s a game for me to find new insights in data. And I just think, I don’t know, it was cool. Maybe this is the picture you talk about. There is a picture that someone captured, which is really fun where I’m on my laptop with a bunch of engineers and then Harley, so president, is on stage DJing. This was taken at 1:00 AM or something like this. It’s a picture which actually really precious to two of us because it perfectly summarizes our relationship in a way which these happy accidents are wonderful and these long journeys, I’ve been doing Shopify for 20 years now, so I appreciate these artifacts. But yeah, so…
To be clear, the reason this is unusual is, and why someone told me this story, is most CEOs do not do this, don’t just sit there and code along with the team. And the reason I thought this story was important is and the reason I think your first principle approach works is you’re actually in the engineering details, similar to Elon if you think about it, of just in the weeds doing the thing, understanding how the thing works, not just coming up with ideas out of pontifications. And so I guess is there anything there of just how important it is, if you want to approach thinking from first principles, it is to be really close to the metal, to the bare metal?
Tobi Lütke (00:50:11): Well, yeah. First principles thinking starts from, I think Elon puts it as physics, which I think is a little bit atoms-coded. I think just from the atomic building blocks really is a right starting point. Atomic building blocks are the computers we are using. Computers are our instruments. We use them to create the music that when people appreciate to receive in the form of software. So you’ve got to understand how they work, at least to work the way I do.
Now, is the way I work optimal? Of course not. This is my point about the aesthetics of how to behave, how to work. I probably don’t conform to the traditional view of what the most of public companies should be like or even should spend their time, but I think I’m successful because I don’t try to conform to anything other than what I’ve learned works. And so what I learned works is be in as many details as you can. Really, really, really just understand the stuff that we are making decisions about and be willing to don’t put too much stake into the sunk cost fallacy. Try to inoculate business or your parts of a company from a sunk cost fallacy as much as possible because that allows you to just see better solutions and so on.
And I think what we were doing, if I remember right, is at hack dates is we were working through pros or cons of just merging all of Shopify into one huge monorepo and we are sketching out directory structures and tooling that we would need, and…
Tobi Lütke (00:52:00): … tooling that we would need and… Again, monorepo, now for companies, it’s a very much one of those door A, door B kind of things. It’s a very consequential choice that is incorrect to go say yes to at a certain size, and then it becomes very correct in my mind to say yes to, but at that point it’s an enormous amount of effort. So, it’s a kind of thing that actually is something I’m uniquely positioned to be involved with because it’s actually a business strategy thing as well. That’s an investment, a very real investment to say, “Hey, let’s change the way we are building system. Let’s figure out what the best way is.”
There’s going to be change management. There’s going to be some people that have very strong opinions on yes or no. And frankly, “Tobi said so.”, helps. It also compresses a lot of time, because everyone knows the way I book. Everyone can come to me with better ideas about anything, and if you’re right, I will change my mind. But I will hold my opinions very strongly until the point of being convinced that they’re not the correct ideas. Maybe that’s another aspect of this Tobi Tornado thing you talked about earlier, people do fight.
Again, the aesthetics of our times put a lot of stake into consistency. I remember various politicians losing campaigns because they were called flip-floppers at times, which I… Although, I’m more of a Maynard Keynes school of… “When the facts change, I change my opinion. What do you do, sir?” It sounds like a better OS to go by.
Running the Core Team Without KPIs
Lenny Rachitsky: It’s awesome, because I’m extracting more of your algorithm of how you think about stuff. A few things you just shared are, ( 1.), you need to be in the details. This is thinking about to be successful as a first principles thinker is (B.), you need to be in the details in your case code. In Elon’s case, it’s like, build the thing and be at the factory, sleeping on the floor. As you said, he’s very Adams based. You’re more digitally native. And then also, don’t be so reliant on some costs. Like you said in this Tobi Tornado case, “I know you’ve been working on this project, we’re going to kill it, because it’s not going to work better if we do that now versus this. Just keep going, because we’ve been going.” And then, coming back to the stuff you’ve shared previously is, analyze the path that you’ve been on, that the previous products and solutions have assumed the path dependence of them, and then don’t look at what they’ve overfit potentially, that isn’t correct. Love it. Okay. This actually is a good segue to something else that someone that you work with suggested I ask you, Farhan, your head of engineering. I asked him, “What’s the best way to get a glimpse into Tobi’s mind?” And it’s actually along the lines of what you just described, which is, the question is just, what’s the best way to disagree with you?
Tobi Lütke (00:55:02): I think just disagree with me. I immediately love it, honestly. I really crave it. It’s very funny. I know how this really surprises people. And I actually appreciate it even more because it requires courage, and frankly, I actually do find that… But it also makes me immediately trust the person more, partly because I, first of all, think they will do what they think is right rather than what is convenient. Agreeing with a group tends to be much more convenient. But more than that actually, that they’re courageous enough to do it right then. I think courage is really, really rare. I found a lot more high IQ in industry than courage. I found a lot more maybe even genius than courage. I like that. I wish it wouldn’t require that. This is why I try to be very inviting of it.
When someone disagrees with me, I tend to immediately stop and say, “Cool, let’s figure out why there’s disagreement.” And it’s almost never, I find, in the, “I just feel like we should do this differently.” What I’m looking for is offer unstated foundational assumptions. What is our divergence point? Because you might be right. In fact, people often are, when it gets to this point, I found. Sometimes it’s an unstated foundational assumption that I hold that is incorrect. People tell me and then I’m so glad we talked about it, because I will stop forever nagging on this thing because now I know we can’t do it because of Sarbanes-Oxley or something like this. It’s just like, “Well, cool.” My mental model of Sarbanes-Oxley, which is regulation for public companies, is not perfect. I have not in-depth studied all the details for it, so I will not consult this in my mind when I am saying, hey, we should solve the task at hand in a certain way. So, this is very good.
How to disagree? I really like debate. I will play devil’s advocate actively if everyone agrees on something. Again, especially if a proposal is something that feels like I could have predicted would be a proposal before it got into the meeting and there was nothing surprising in it, I will make myself an end boss of a level and just say, “I’m going to say this is just not that good. I think we could do way better.” I want people to then argue more in-depth for the veracity of the decisions, and that leads to a form of disagreement. I think all these things end up building trust.
I like that that touches also on this idea of, again, maximizing potential, the potential of the teams, the potential of the employees
Tobi Lütke (00:58:14): Because of course, the really important decisions, we don’t talk about. This is the most important thing. Shopify probably makes… What?… millions of decisions every day, like, write this code this way, yes, I’m going to add this unit test, maybe I’m skipping a unit test. Might be the difference between a future production audit. Millions and millions of these tiny decisions. So, you’re not hiring engineers primarily or accountants, you’re hiring people who make excellent decisions, given their specialization and areas they’re overseeing. And given that, decision making as a concept is actually really understudied, I find.
I think these are instances where we can just learn to make decision making together. Because I think while a lot of decisions are made independently, we are a product company, we are on a mission, and we want our product to feel like something that a single person made, in the same way how any author tries to write a book that clearly reads that it came from one mind. Because people can see and spot this, if this doesn’t happen. You end up with something that looks like a television remote, where there’s a Netflix button over here… It’s like you can reverse [inaudible 00:59:41] from the remote control. That’s important.
Using these moments and also bringing decision making inwards to go and say, hey, let’s have very efficient opportunities for as many people that can possibly get together, make decisions together, not as a democracy but with clear who needs to convince who. That because otherwise this ends up just taking way too long. Give a opportunity for everyone to change my mind over Glenn’s mind, and so on, is an excellent practice, I find. I think this is a more optimal way of going about it.
Fighting Biases in Business Aesthetics
Lenny Rachitsky: Speaking of disagreeing, it reminded me of a story I heard, about the Tobi Tornado actually, the way you operate that. I love it. It ends up being this interesting microcosm of your first principle’s way of thinking. The story is, I think you’ve said at one point that the best way to get people to give you insights is to say something they disagree with on the internet.
Tobi Lütke (01:00:42): Yeah.
That’s often the way you approach this is, you post in a Slack group, “Hey, I don’t think this product is going to work, and here’s why.” And that ends up creating the most information for you. Is there anything along those lines that might be helpful to share?
Tobi Lütke (01:00:56): I don’t think that’s one of my better ways of doing that. I have to be somewhat careful. The people I worked with a lot, I will do this, just because it’s funny. But even I can see that that would seriously stress out the interns.
You guys have a lot of interns. I think you have 1,000 interns this coming year, is what Farhan shared.
Tobi Lütke (01:01:20): Yeah, we just started. I was in the office yesterday and it’s absolutely full of interns. It’s great.
Oh, man. Love interns, so much energy. This idea that you shared about building one product, building towards one vision, this actually reminds me of something that came up basically every time I had someone from Shopify on, which is this idea of a 100-year vision that you keep. I don’t know of any other company that operates in this way, where the founder has this 100-year vision of where the product needs to go and working backwards from that.
Can you just speak to that, of that way of operating, why you find that helpful, how that actually works?
Tobi Lütke (01:01:54): I talk about look in the future and then think backwards a lot. What would we want to have done 20 years ago on this, or 10 or 5 years ago? What’s the decision our future selves would want us to make, is useful. I find future casting to be generally extremely valuable. I definitely did not build Shopify to flip. I had lots of opportunities to sell Shopify to various people. I didn’t consider it, because it’s just too interesting of a journey and I think that’s too… If a endpoint [inaudible 01:02:34] all companies is a convergence on a set of four or five people who can afford them, that just creates too much of a monoculture, I think, in thinking.
I like that Shopify is different. I think this is good. I think it’s a terrible place to work for many, many, many people. It’s the best place in the world to work for some people. That’s so good. That’s what we want, I think. We want more of this. We want people to be able to window-shop for a place where they can be enormously successful, because the place’s set of beliefs just fits you like a glove.
100 years. We have very long-term plans. 100 years, you can’t talk about this software product, but you can talk about the mission itself, whatever things that will survive for 80 years that are left on this particular timeframe. Entrepreneurship is just precious. Shopify exists, basically you make entrepreneurship more common. That is the thing we wanted to cause in the world. We have had, I think, success doing this already, but again, there’s no speed limit and no stopping point for this. As I keep saying, the word’s unbelievably path dependent. And therefore, if we are part of a path, we can cause it to be more in adherence to the things that we value and would like to see. This is the wonderful thing about company building. They can have lasting impact.
We are making decisions based on all the things that we can do. What are the things that will be most long-term valuable for the long-term pursuits we are on, and how can we normalize entrepreneurship? It’s a thing that’s really important to me, just because people don’t spend enough time on it. It’s all economics. Our [inaudible 01:04:44] living entirely depends on businesses. We are all part and tag of an environment that is extremely entrepreneurial and actually celebrates entrepreneurship as a courageous act and a glorious act even.
But that’s not true in most of the world. In fact, most people never encounter anyone who engages in doing that. It’s not something people see as one of their options. I want us to make these choices. And I think the long-term focus matters. It’s also really powerful for decision making. And I know this is potentially intuitive to most, but also rarely practiced. You’ve had a lot of people from Stripe on here. Stripe and Shopify have had a very long-term partnership. Stripe and Shopify had potentially the most valuable partnership, or at least one of the top ones in history, of technology, because we were both very small companies and we decided, hey, let’s work on an assumption that both of us are going to win our markets, and work together. So, Stripe’s [inaudible 01:06:02] for payments and we’re allowed a part of Stripe and so on.
When you’re in a partnership like this, you play basically iterated prisoner’s dilemma. Every instance, every turn you can make a choice: coordinate or defect, defecting if the other person collaborates, collaborator effect. If both collaborate, everyone gets a point. If one defects and the other one collaborates, you get a lot of points, all at one moment. Being a good partner in business is like this corporate marshmallow test that companies tend to fail in a very funny way.
If you see the videos of kids doing actual actual marshmallow tests, the smart ones actually turn the chair around, look away from the marshmallow, and sit on their hands, and just go vibrating. Because they know it’s the right thing to wait for getting two marshmallows in the future, but man, if they just look at a marshmallow, they’ll just eat it. Most CEOs, most companies can’t even successfully do that and just engage in pulling future profits forward at a discount or even defecting on partnerships that would be much more long-term valuable.
If you’re talking about long timeframes, like 100 years, there is no question. But clearly, the correct way to play iterated prisoner’s dilemma is coordinate for both sides. It’s way more valuable [inaudible 01:07:27] doing that over long periods of time than any momentary defection could possibly be convenient at a moment. It’s a huge amount of product decisions. When we are deciding roadmap, end up being very influenced by this, because I get pitches for things we should do and why don’t we do this kind of change to the system, there’s so much money in it, and all these kind of things. I’m like, cool, but almost always come in a form of putting future profits forward at a discount.
But we have a long time horizon. Let’s not take the discount, let’s… Also, you often compromise the entire business after a while because you end up just… Your customers notice if you’re going into value extraction. The best companies, I believe, at least from a perspective of a leader, resemble some kind of cockpit, or maybe a room full of dials or levers for monetization. My job is adding as many of these levers as possible to the room, and then not pulling any of them. Because I think we do best if we are on the same side of a table of our customers and we had them become entrepreneurs, which is a mission, and we want to support them to be more successful, which is the business model we are in. We get a very, very small stake of the sales in our monetization system. And therefore, we are incentivized to make everyone as successful as possible.
Again, when I get a PowerPoint from investment bankers or so, where they tell me I have enormous pricing power and I could massively change the prices, I’m like, “Yeah, but I’d leave Shopify if that happens to me, so how is that good?” He’s just like, “Why?” It just goes like that.
Energy Sources and the Anti-Status Quo
Lenny Rachitsky: There’s a quote that you wrote somewhere that, to me, identifies this point so succinctly, “On a long enough timeline, playing positive-sum games with your customers is the ultimate growth hack.” Yeah, beautiful.
Tobi Lütke (01:09:46): I think that’s pretty piffy and it’s also just… Try to argue with it.
That’s true.
Tobi Lütke (01:09:54): Positive-sum games have incredible returns, especially in the worlds of software, where you can experience exponentials very easily. I see company building as… I wish there would be better analogy than chess, because chess is a game of perfect information, which is totally incorrect. But one thing which I like about chess as an analogy for business is that it basically is two games, that you have to be good at at both. There’s a positional game that you learn: Develop your pieces, gain influence by your pieces over the board. That’s really, really important. And then there’s tactics, which you learn tactic training. You go get a puzzle trainer and you drill tactics and you learn the intuition to sports tactics, sense them. Both of them are actually super independent of each other.
The business world only talks about tactics. It only talks about the conversion of we did this thing. We did an A/B test and we changed the color of blue, and conversion red went up. We lionize the easy hack. I don’t think that’s important, honestly. You need to be good enough at tactics to not go out of business. You can’t get margin code, sure. But the sum total of all the value of a potential tactics that you could employ stays with you if you are actually are doing the positional game.
The positional game is like, what is the territory on the map that you are taking? What role do you play? How much trust do you have of merchants? Do merchants want more from you or less? Are you the kind of thing they’re trying to optimize out of their software spend, or the one that they ask to subsume all other software spend? Do they rely on you? Are you part of a team, or are you used as a tool in the toolbox, usually forgotten, sometimes coming out when certain task is being done? What does your product cover? What industries are you addressing? And so on, so on, so on. This is the positional game. How well do pieces fit together? Do people like relying even deeper on you?
If you do that well, the tactics are yours and you can hire a lot of people who are extraordinarily good at spotting tactics and using them. But if you do it too much, you end up extracting through tactics the entirety of the value that you have created and that is yours to take through a positional game. And if you do that, you have nothing left in the tank. That’s for companies that we all see, that just got to a point and then just fade. Those are the companies that got tacticked out of their position.
The skill, obviously, is finding the balance between these two things of short-term, not get margin calls, as you described, but also think ahead.
Is there any kind of heuristic for founders listening to this who are like, “How do I do this? How much should I be thinking about the future versus now?”, I guess? How do you try to create this pie chart of driving goals immediately and show investors we’re killing it, while also thinking ahead?
Tobi Lütke (01:13:22): Is there a good heuristic? Objective number one is: Don’t die. Again, in the end, for companies that become historic companies are the ones which did not die. It also sounds very basic, but actually it’s more actionable than it might seem. Past that point, I would argue: Just focus on the positional game. I think this is where a bit of a discrepancy of the founder and the rest of the company sometimes lies. I think the founders create finite, winnable games for people, that are very much serving the infinite game of developing, that the mission implies, the quality of position on the board.
By the way, it’s an infinite board. Just picture the chess board as the initial terrain, but there’s, I don’t know, fog on the board and it’s a terrain that’s much larger that you will explore over time. I think that’s the most prettiest way of mentally thinking about the experience of building a company. It’s an exploration and a collaborative inquiry into a question that is implied by the mission. And you will get to explore how correct your mission is and how good your decision making is along the way. And you get to learn a lot, and this is why I think it’s a valuable thing to do.
But they often, the founders, and the mission of a company are in alignment, but they are, again, non-quantifiable things. They are pursuits. Maybe as time horizons that go past, all of our time horizons of our careers, therefore that makes it hard for people to care deeply yet. But they do care about the instances, the games along the way, the [inaudible 01:15:24] that have to be done to explore a part of a map, the tactics that have to be executed.
I’m not going to take us here, but I know you’re a big fan of this book, Finite and Infinite Games. I actually had it once at a book club and it blew my mind open. I’m going to link to it. People should check it out. You’ve talked about this on another podcast. I’m not going to [inaudible 01:15:40].
Tobi Lütke (01:15:40): Yeah, it’s a lovely book. James Carse did an incredible, underappreciated… I think they essentially wrote one of the best business books. They is trying to write a hard philosophy book. I don’t know if it works for its intended purpose, but I think there’s more value in what he actually accomplished. Maybe, sadly, it may be that he didn’t know before he died, which would be very sad.
It’s a little hard to read for people…
Tobi Lütke (01:16:08): Yes.
… so just stick with it and just try to wrap your head around what he’s trying to say, is my advice.
Tobi Lütke (01:16:14): I think reading… I don’t know how many it is, but… the first couple of chapters really helps you get your arms around the story, his insight. The rest of the book are examples of his ideas applied. My takeaway from reading his examples is that he did not fully appreciate the insight of his own idea. It just ends up being very locally limited and very narrow-focused. It’s a much grander idea than I think he really… This is why I say I don’t think he fully appreciated quality of his own idea, which…
I think we need a Tobi version of this book, with a foreword, and…
Tobi Lütke (01:17:02): I heard Simon Sinek…
Uh-huh.
Tobi Lütke (01:17:04): … wrote a book of a similar title. I’ve actually never gotten around to read it, but I think he would be very good at interpreting this if that’s what he did.
Axioms and Business Logic
Lenny Rachitsky: That’s what it sounds like. I feel like his book is that book, written in a different way.
I want to come back to something you talked about, which is focusing on entrepreneurship and the merchants that you all work with. To me, this is another example of maximizing human potential. The way I think about this, people always talk about Y Combinator and Stanford and all these places that create all these companies and founders. If you think about it, Shopify does this orders of magnitude beyond, and I don’t think you guys get enough credit for that, the amount of businesses you create, the amount of lives you change.
Tobi Lütke (01:17:43): That’s because we don’t want the credit. That’s the point.
Is there a platform?
Tobi Lütke (01:17:47): We don’t want the credit.
No.
Tobi Lütke (01:17:48): I mean, maybe if you wanted it, you could also not get it, but we don’t need to explore that because we actually, literally don’t want the credit. Shopify is a company that pushes from behind. We don’t want to be written into the story. We want to just…
Tobi Lütke (01:18:00): … want to be written into the story, we want to just kind of have you do your thing. There’s just too much grandeur in wanting to front it all. This is also what caused us to go and help people build their own things rather than start with a marketplace that everyone gets to lease a little component of at prices that corresponds to magically exactly your margin, as this usually goes in marketplaces. And so yeah, I think we don’t want it. I’m proud of it, but that’s intrinsic. I don’t need very extrinsic appreciation for it, I suppose. There’s millions, millions and millions of people that use Shopify daily and that represents the business and it causes major amount of employment around the world actually, definitely in North America specifically and Europe. And so I think that’s really gratifying. I’m sure there’s a way of saying that many of us businesses might actually have existed without Shopify.
I also think you can make a case that a good number exists because of Shopify, because what we’ve observed is that this was actually, it’s my favorite thing in the entire Shopify journey that we sort of rigorously determined, which is that because I really, really, really prioritize good UX, legible interfaces. You can tame enormous complexity with great UX in a way that makes sense to people. I think this is actually almost a moral obligation to do for software because when software goes bad, it makes people feel dumb and machines do not get to negatively influence people in my mind. That is the inversion of the priority of the hierarchy of how machines are tools wielded by people to be more powerful at the things that they’re already great at than they might imagine. That’s when this all needs to fit together, but beyond the moral point, which is 10 years, and you could definitely take the other side of what we have determined is that every single time we make a complex thing simpler. It is actually that more businesses will exist on a platform.
So I think this is intuitive or at least directional, but sometimes people are not like, “Well, don’t you just need to have all these features?” This is the RFP view of a world of software, right? “Don’t you need to just have all the features then people can implement their plan?” Well, no. Think about the mental state and entrepreneurship entrepreneurs, they are, I mean especially if you’re first entrepreneurs, they’re unsophisticated. They don’t know what to do. They’re kind of scared. By the way, engaging in building something again is an act of pure courage and usually one which is very hard to hide from others. The people around you will know that you’re doing this thing. Ask any entrepreneur. The amount of people who tell you to not do this is actually stunning. People do not want people to just step out of a box that exists that they sort of explain to themselves around them and have a chance of reaching higher because that would sort of invalidate their own life story in some meaningful way.
So there’s actually a whole lot of nay saying that discourages people already, which is not super helpful. So now you have a situation where people are telling you to quit potentially, you don’t have a lot of money presumably, you’re asking yourself, are you even doing the right thing? Hopefully there’s some elements of encouragement. Hopefully there’s a passion. Hopefully you’re trying to create a thing. Then comes to a situation that stuns you in some way, somewhere where suddenly the software starts talking about APIs where this is something you’ve never encountered in your life before or that just the option that you think it should have on the text configuration screen doesn’t make sense given what we know about local taxes and so on. Here’s the thing that happens. So the beautiful thing with Shopify is there’s basically no tax configuration screen. It’s just correct. The amazing thing about software is we can actually just overtake this piece of complexity.
We know what the taxes are everywhere and we are just doing it for you. You don’t have to think about taxes. In this case we have someone encountered something that stunned them, that stopped them in their tracks. On the wrong day, what that means, it’s like they are going to close a browser and they say, “You know what? Fuck it. I’m not cut out for this.” Or, “This piece of software I use sucks and it’s too dumb for understanding the thing I have in my head,” or something else. Everyone has a different rendering of the same thing, but just like progress stops. It is not just the businesses that shouldn’t exist anyway that stop in this way. In fact, it’s actually, and this is again what the data says, so many businesses get very close to this point many, many times. I can tell you Shopify did many times and I have more tools on entrepreneurial journey than most. I’m a computer programmer and I love those things. You tell me I can spend 14 hours programming for the next couple of years. I’m like, “Holy shit, let’s go.”
Most people aren’t like that. So I’ve had unfair advantages that allowed me to overcome very technical climbs on this learning curve, but I know that’s like most people can’t have these advantages and therefore will churn out of a test. And so in other words, lowering complexity, making good UX, creating software that just autopilots taxes or payments or any of these kinds of things, fraud, actually causes more entrepreneurship. That is the best thing, the best answer to the most important question of my life that I’ve encountered because up to this point it was an intuition that doing the Shopify thing would be valuable in absolute sense, in a hundred-year sense. Afterwards I knew, right? And so that also just means now seeing things that are overly complex or shouldn’t even be there, it’s just physically painful because I know, it’s like all these business that are trying to get somewhere and could exist and could now employ people and delight customers just died along the way because of something we did wrong or poorly and so on. And so this is a great source of energy for me to keep going.
Practices Derived from First Principles
Lenny Rachitsky: I feel that, I feel that pain of the idea of anytime a button’s broken or it’s not as… the button’s wrong color and more people would be converting means that you’re not creating as many businesses, many entrepreneurs. The feeling you described of being scared to launch something, I exactly felt that when I started the newsletter. And my solution to that as I launched it is, “I’m just launching this as an experiment. We’ll see where it goes, just don’t worry about it. I might blog once in a while.” And that really helped just lowering the stakes and I could see how just things like that and advice that can help someone get over that hump.
Tobi Lütke (01:26:01): Yes, and I think this is an under explored thing in the word of UX. Random example, I was absolutely delighted by, maybe TikTok has the same thing, I don’t know, but I was posting a video to Instagram and it allowed me to run a try run of the reel. It just seems like a new feature and just saying it shows it to a couple hundred people out of network and if they like it then it’s going to post to my profile. I’m like, “I don’t need that.” But I thought that was one of the more… I honestly just closed Instagram and was like, “Holy shit, this is probably one of the most profound insightful pieces of software I’ve encountered. And as a connoisseur of good ideas, I’ve never had this as a valuable thing.” It feels like a step function upgrade to the traditional A-B test as a concept because it’s so understandable and I’m like, “What else in the world should have try runs that run out of network?” Basically everything, isn’t it?
This is an amazing thing. Like I said, the rarest thing in the world, it’s not even creativity or genius, it’s courage. So let’s lower the net amount of courage needed. Honestly, that’s one of those things you can probably run an entire career or reinvent an industry on. These things unfurl into incredible amounts of value if you really pursue them and come at them from first principles rather than how can I do the things I want to do anyway slightly better because of insight. That’s good and people should always do it, but not usually where most of the value ends up becoming, like manifests when a new idea comes around. So I talk radically about an Instagram feature. That’s a weird thing to do, but I was actually just really to delighted because again, it just came up a few days ago.
I love this thread of courage. The example you gave reminds me also of I feel like going on a dating app at a different city for the first time is another example of this where it’s a dry run, nobody knows you.
Tobi Lütke (01:28:06): But even then, shouldn’t there be like, “Hey, post my profile to a different city for telling me if it’s any good.”
SO people like it.
Tobi Lütke (01:28:14): I mean I’ve never dated basically so I mean at least not in the last 25 years.
Yeah, of the podcast.
Tobi Lütke (01:28:20): So I’m pre swipe left and right dating, so my conception is very low of this. Maybe I should look more at dating apps. I’m pretty sure there’s some amazing UX in most things.
I imagine. I feel like dating is going offline more and more. I’m feeling like that’s what people are trying to do. They’re tired of the swipe, but anyway, going to go in that direction. Maybe a couple more questions while I have you here. One is this idea of the talent stack. You’ve mentioned this concept before and this is the term that I’ve seen you describe it as of this idea of the power of focusing on your unique talents and curiosity and that leading you to the biggest opportunities, especially early in your career. Can you just talk about what your insight there.
Tobi Lütke (01:29:04): Yeah, again, I only lived one life, so I can’t Monte Carlo all the decisions I make and just figure out which ones ended up being load baring, right? I’m at this point pretty amused by the following thing which keeps happening. But I am getting curious about some absolutely random thing and it really is fairly far-flung stuff and like magic it becomes the way that ends up allowing me to make a very important choice a year later. It’s ends up being a better analogy that I’ve learned or a different way to see something or another idea that I found being represented in this area of expertise that is actually just again, another repackaging of another foundational idea, which allows me to go and look for more examples.
And it’s just funny this way. So even early in career just I followed my curiosity. I love programming and I love computers and I loved the internet when it came along and I just like, “Cool, I’m going to find a task that I find valuable,” which I was always engaged in retail and I have a lot to say about how retail should be brought by the internet to people. But this is a beautiful intersection of all the things that I find interesting. And then on top of it, I found Ruby, which I loved as a technology and then just now I was highly motivated to tinker, explore a space that was clearly emerging but just felt very obviously of value in the future.
But I didn’t do it because I was following money. I did it because I like learning by doing stuff and I like tinkering the things and so this was a way of financing my tinkering by selling snowboards and then it led to other things. And I’ve kind of been doing that thing all along. It’s great fun, great fun for me. I don’t know how much of a advice that is, but maybe a proof point that can work really well. I sometimes worry though that things like this end up being a little bit like all you need to do is buy a lottery ticket, set for lottery ticket winner, right? So it’s, who knows?
It does resonate. It reminds me of something Brian Armstrong once shared, which is the reason that he-
Tobi Lütke (01:31:37): He’s also, by the way, you have to add him to the set of first principle thinkers if you’re starting to cultivate one. And I think there’s plenty of startup success, successful founders, CEOs, especially for public companies are surprisingly alike in this projecting slightly different, maybe coming from different backgrounds, but Brian is extraordinarily strong at this [inaudible 01:32:06] by the way.
Okay, here’s to the next nominated for the podcast, Brian. We’ll get on it. But interestingly, he had basically exact same advice for what allowed him to create Coinbase is his background was economics, coding and cryptography or something along those lines. And it’s the Venn diagram, is like, “This is the thing I’m uniquely strong at and have an opportunity to win it.”
Tobi Lütke (01:32:29): Yeah. So this is what we very actively tell our customers, right? There’s a 2005 essay by Kevin Kelly saying, “In the future on the internet you just need a thousand true fans.” It’s what the internet and I think what Shopify celebrates, that instead of trying to create toothpaste, which is of course a huge term but hard to differentiate. It’s much better to figure out a triplicate intersection of three different things and nail it. We took a winter vacation with my boys and my family in the Caribbean and I like playing poker with them. I like playing poker and it’s great limited information game. I think this were valuable for kids and so I found this, I might have mentioned this in some place, but I found this pirate gold I think it’s called.
It’s just amazing poker set, poker chips that look like they come right out of the treasure chest or in Pirates of the Caribbean. Think about just the intersection, so this is online Texas Hold’em, quality poker set instead of… because you can play with just anything, right? People were in the market for spending money on something like this and pirate themed. You’re like, ” But I’m such a big fan, I’m telling you about it and people might actually purchase it because I’m enthusiastic about it.” And because enthusiasm is actually the best marketing and they have at least a thousand true fans, and it’s just better to make amazing things for some people than make something that everyone wants maybe or would tolerate. And I think that’s really, really good. So I think that works for your career too because by the way, we are all entrepreneurs so we’re ready to think about any career.
If you actually ask me for career advice is we are products, we are engaged in entrepreneurship, we are basically software as a service if you want, or talent as a service. We are selling a subscription in the form of our employment we call it, but it’s actually just rebranded the subscription frankly because it comes some extra protection, which is very good, but it’s still the same thing, it’s a double opt-in company. Like you would say, I can do these things and I will be of significantly more value to the company than you will have to spend on me. There’s like a positive ROI ideally for a company and a company either says, “I agree or I don’t agree with you.” And that’s actually what employment is. So I think people should think about themselves as a product. Your career is not based on mentors you find or getting a promotion or not.
It’s about what are you too good to ignore in? How good can you be? Can you be absolutely work? Can you be excellent? I mean you don’t have to be work class and stuff, but you have to… Ideally you find a thing where you are just in this sort of tiny space of five intersections or so versus we just know more than everyone else and then everyone calls you. Like I said, we use Ruby. Ruby was not that fast. Shopify would rather have it very fast. We found people who got PhDs in dynamic language, garbage collection dynamic language, just in time compilation to make Ruby fast. And they’re having delightful time at Shopify and Shopify is now very fast and it’s very good. And everyone gets to profit from us needing this because we merged it back into Ruby Core and now everyone can have a JIT compiler. So it’s great.
Local Maxima in Search
Lenny Rachitsky: I feel like every story is a fractal into other hours of other podcast conversation we can have. I know you have to go run this nine, 10, 10,000 person company and do real work. So I’m going to end it there. We got through almost everything I was hoping to get through, but there’s infinitely more questions I still have. Just to maybe end it, is there anything you want to leave listeners with? Any last nugget that maybe you didn’t mention?
Tobi Lütke (01:37:03): If I get to talk to again people who look at product and product management, here’s my thing that I think maybe only implicitly came through in all this. Every product in the world, the quality at the end of the day is simply a reflection of how much the people who created it gave a shit about the product. And it is not possible to make great products if the people work on it do not give a shit about the product. And I actually think this is a very important role for product leaders to make sure that the team gives a shit. And I think this is something that can be done with building empathy for people using it, but also it can be done infectiously by the product leader. The product leader has to give a shit. Do not engage in product work on product that you don’t care about because you cannot produce the thing that the person will give you a task is looking for.
And I think again, so much goes into this unquantifiable conversation which we already had, but boy is that important because it just isn’t about the brief document, it isn’t about the aligning stakeholders. Sometimes in some places some of those things are aspects of it and I think they correlate sometimes with people. But I think the higher order bit is if you are in product, you have basically free roads. You have to have the team sits, look around corners that they don’t see. You just have to understand this thing that’s being done better than everyone else because that’s a role. Now, you don’t have to do this by yourself. Everyone on the team is of a resource to you and you can query them and therefore determine what is around the corners and so on. But in engineering and UX and so on, but that’s very important. Second is you’ve got to be exothermically infectious with actually caring about this thing because just that one thing alone will make a 10 times better product. It’s crazy how much of a change this makes.
I feel like honestly what you’re describing here is like founder mode, what the outputs are. Giving a shit, being exothermic excited and looking around corners. I feel like this is its own conversation here.
Tobi Lütke (01:39:37): Yeah, I agree. I don’t think founder mode, so I mean I think, yeah, founder mode is a super valuable term, but I actually do think that I answered the question about what I think about founder mode in the discussion about pumping heat into systems. That’s what I do. It’s just that founder mode in a founder run company, I think this has an easier time existing because people dislike the people who are pumping heat into a thing. People fight them, people probably take them out, and if the boss is like this, boss can protect the other people like this. And therefore I think founder run companies can be innovative longer because there’s a substrate by which this can stay a thing for a long time. So anyway, that’s my tear down on the founder mode discussion, which I think is a really fascinating recent discussion. I find it’s very valuable. I’m sure it’s misunderstood in all sorts of ways, but I think it’s more actionable than people think.
Tobi, I appreciate you continuing to drop nuggets, even though I know you have to go, by the way, your video looks incredible right now. This is amazing editing, we’re doing like a rainbow streaming across your body.
Tobi Lütke (01:40:53): Yeah, we really hit the rainbows here like, “Let’s go and invent.” I hope you’ve got to gold that you were hoping and-
So much gold.
Tobi Lütke (01:41:00): … I was sufficient leprechaun here to deliver.
What a metaphor. Okay, Tobi, thank you so much for doing this. This was incredible. I feel like this is going to help so many people build great companies and thank you from first principles.
Tobi Lütke (01:41:13): Okay, I got to run. Bye-bye.
Okay, bye.
Thank you so much for listening. If you found this valuable, you can subscribe to the show on Apple Podcasts, Spotify, or your favorite podcast app. Also, please consider giving us a rating or leaving a review as that really helps other listeners find the podcast. You can find all past episodes or learn more about the show at lennyspodcast.com. See you in the next episode.
Glossary
| English | 中文 |
|---|---|
| alpha | alpha |
| API | API |
| Archie | Archie |
| Archimedean logic | 阿基米德逻辑 |
| autoregressive models | 自回归模型 |
| axiom | 公理 |
| Boolean | 布尔值 |
| Brian Armstrong | Brian Armstrong |
| Chad | Chad |
| cohort | 群组 |
| composability | 可组合性 |
| conversion rates | 转化率 |
| corporate marshmallow test | 企业的棉花糖测试 |
| Daniel | Daniel |
| devil’s advocate | 魔鬼代言人 |
| Elon | 埃隆·马斯克 |
| exothermic | 放热的 |
| fine-tune | 微调 |
| First principles | 第一性原理 |
| fitness function | 适应度函数 |
| founder mode | 创始人模式(founder mode) |
| Glen | Glen |
| Glen Coates | Glen Coates |
| Goodhart’s law | 古德哈特定律 |
| holdouts | 对照组 |
| IC | 独立贡献者(IC) |
| infinite games | 无限游戏 |
| iterated prisoner’s dilemma | 重复的囚徒困境 |
| James Carse | James Carse |
| JIT compiler | JIT 编译器(JIT compiler) |
| Kevin Kelly | Kevin Kelly |
| Lenny Rachitsky | Lenny Rachitsky |
| local maxima | 局部最大值 |
| loss function | 损失函数 |
| monorepo | 单体仓库 |
| neurons | 神经元 |
| overfitting | 过拟合 |
| path-dependent | 路径依赖的 |
| proxy | 代理指标 |
| RFP | RFP |
| Sarbanes-Oxley | 萨班斯-奥克斯利法案 |
| Simon Sinek | Simon Sinek |
| sunk cost fallacy | 沉没成本谬误 |
| talent as a service | 人才即服务(talent as a service) |
| talent stack | 人才栈 |
| Tobi Lütke | Tobi Lütke |
| unit test | 单元测试 |
| UX | 用户体验(UX) |
Reformatted by reformat_english.py