从联席总裁到联席 CEO：Spotify 非同寻常的接班内幕

所以今天早上我给 Daniel E 发了短信，我说：“嘿，我在和 Gustav 录节目。你认识他差不多二十年了。告诉我该问他什么。”他回复说：“让他讲讲他是怎么为 CEO 这个角色做准备的。我们做了一件很不寻常的事。你能描述一下发生了什么吗？”

So, I texted Daniel E this morning. I was like, "Hey, I'm recording with Gustav. You've known him for, you know, two decades. Tell me what I should ask him." And he goes, "Ask him to tell you how he prepared for the CEO role. We did something quite unusual. Can you describe what happened?"

Daniel 非常重视准备。你知道，Daniel 做决定不是特别冲动的那种。所以这也是那种我认为在他心里酝酿了很久的决定之一。大概三年前，他就让我和 Alex Nordstrom 升任联席总裁，开始替他打理 Spotify 的日常运营。他当时还是 CEO，所以他逐渐给我们越来越多的自主权，让我们按周、按月地管理业务。所以当真正到了接任 CEO 的时候，那部分我们已经烂熟于心，而且我们已经运营了三年，包括完整的盈亏和资产负债表等等。但还是有很多新东西要学，你知道，那些原本只有 Daniel 负责的事情，比如公关、政府关系、作为公司公众形象等等。但我们对接手业务已经准备得非常充分了。我觉得这是 Daniel 的独特之处。我为 Daniel 工作了将近 18 年，我觉得除了 Daniel，我不会再为别人工作了。我遇到过好老板，也遇到过不太好的老板，Daniel 是那个最好的。而且他从一开始就非常善于放权。我记得我加入 Spotify 的时候是 2008、2009 年，当时我们要去美国，Daniel 认识 Meta 的人，当时叫 Facebook，我们想和他们做深度整合。我当时在 Spotify 还是新人，Daniel 就说：“你能不能去和 Zuck 谈谈，通过 Facebook 联系他们，把这事搞定。”我当时想，这对一个新人来说真是巨大的信任。所以他一直超级放权，我想这也是为什么我能待将近 18 年，因为你得到了很多责任，几乎每年都像换了一份新工作。所以这个他在三年前开始的联席 CEO 或联席总裁的合作，就是其中一段旅程，我和 Alex 得以站出来，尝试接管整个业务。我们实际上彻底改变了公司运营的方式，和 Daniel 的运营方式完全不同。

Daniel's very big on preparing. You know, Daniel, he's not very rationous decisions. And so, this was one of those decisions as well that I think grew with him for a long time. So already about three years ago, he asked me and Alex Nordstrom to step up and become co-presidents and sort of start to run the day-to-day of Spotify for him. He was still the CEO and so he's been gradually like giving us more and more rope to run the business day to week to week and then month to month. So when it actually came to sort of taking over CEOs, we already knew that part like by heart and we had ran it for 3 years, the full P&L and balance sheet and so forth. And then there was still a lot of new stuff to learn. You know, all the stuff that was Daniel's responsibility alone, like, you know, PR, government, being the public face of the company. But we were very well prepared for taking over the business. I think that's unique with Daniel. I worked for Daniel for almost 18 years. And um I don't think I'll work for anyone except Daniel again. I've had both great bosses and not so great bosses. Daniel is the great one. And uh he's incredibly uh delegating you know already from the start. I remember when I joined Spotify was back in 2008 2009 and then we were going to go to the US and um Daniel uh knew u suck that meta then Facebook and we wanted to do this pretty deep integration with him and yeah I was new on the job at Spotify and Daniel just said like can you just go there and like talk to Zuck and met them like over Facebook at the time and like set this up. I'm like that's a lot of trust for someone who's new. So he's always been like super delegating and that is also I think why I've stuck around for almost 18 years because you get so much responsibility. It's like you get a new job almost every year. And so this like co-CEO or co-president partnership that he started 3 years ago was one of those journeys where you know Alex and I got to step up and try taking on the full business. We actually changed completely how we run the company from the way Daniel runs it.

再多讲讲。

Say more about that.

你知道，说到组织模式本身，这不是我想出来的。我觉得 Steven Sowski 说过很多次，你不可能赢。没有一种组织模式，你知道有很多种，而且都有效，比如亚马逊以某种方式组织，是万亿美元公司；苹果用另一种方式，也是万亿美元。Elon 做他自己的事，也产生了万亿美元公司。所以没有唯一正确的路，你只需要选一个适合你性格的。所以我实际上认为组织是相当基于个性的。Daniel 有非常独特的个性。我觉得他不是那种典型的，我是说，你了解他，但他不是那种典型的阿尔法男性创始人，我认为这也是为什么他在这种行业如此成功的原因之一，因为如果你想想 Spotify 是什么，是的，它是一家做硬核技术创新的科技公司，但它也是一家媒体公司，在一个有很多音乐厂牌、图书出版商、播客出版商的行业里，你可以想象，这些媒体行业里有很多非常强势的人物。所以他需要既能凝聚团队，打造世界级的技术，又能让这些有个性的人合作，去做他们个人不想做、集体也不想做的事情。我仍然不完全理解他是怎么让这些人做他想做的事的，但肯定不是通过阿尔法男性的强力。而是通过不具威胁性、非常讲逻辑。他非常谦逊，但我觉得也许大多数人，特别是美国人，会把谦逊误认为软弱。其实恰恰相反。他极其坚韧，极其坚韧，从不放弃。所以他有非常独特的领导风格，这塑造了公司。而且正如我所说，他的领导风格有很多不可思议的优势。有一件事我觉得我确实选择做得不同，就是 Daniel 喜欢星型模式，和很多人一对一地谈。他会和我谈一些事情，单独开会，然后和 Alex Nordstrom 谈同样的事情，但从商业角度，然后可能和公关、财务谈，他就是用星型模式，他不喜欢所有人都在场的会议，这有时会让人沮丧，你知道，就像他对我说一些事，对 Alex 说一些事，然后我和 Alex 见面，我说“我们要做这个”，Alex 说“不完全对，他不是这么跟我说的”，然后我们回去找他，当面指出。但他喜欢那种星型模式。Alex 和我选择做非常不同的事情。我们选择让整个公司同步。所以你可以把 Spotify 看作一个双职能的组织。我大体上负责产品和技术，Alex 大体上负责商业和内容，但我们不是做分而治之，那是个好主意，分而治之因为更有效。但它有代价，就是你们不同步。所以我们不做分泳道的分而治之，我们做这个我称之为“同步游泳”的事情，我们一起游，这比竞技游泳难得多。当你搞砸了，看起来很难看，但当你做得好，看起来很美。所以我们决定不搞那种通常的，你知道，我和我的直接下属开会谈产品和技术，他和他的开会谈商业和内容，然后我们的人在组织里某个地方见面，然后争论出结果，这是大多数公司的运作方式。我们选择不搞我们自己的领导会议。我们只有一个会议，所有高级副总裁都参加。所以我们每周二有一个叫 E 团队的会议，三个小时，所有内部职能的高级副总裁都参加，比如市场、广告、订阅，也包括所有产品和技术职能，都在同一个会议上。

You know when it comes to organizational models in themselves I this is not something I came up with. I think you know Steven Sowski said this a bunch of you just can't win. There's no organizational model you know there are many and they all work like Amazon is organized in a certain way trillion dollar company. Apple in another way also trillion. Elon does his thing you know also produces trillion dollar companies. So there is no right way. You just have to pick one that fits your personality. So I actually think organizations are quite based on personalities. And Daniel has a very specific personality. I think he is not the typical sort of um I mean you know him well but he's not the typical sort of alpha male founder which I think uh is one of the reasons why it's so successful in this type of business because if you think about what Spotify is yes it's a technology company doing sort of hardcore technology innovation but it's also a media company working in an industry with a bunch of uh you know first music labels and then book publishers and and you know podcast publishers, some of these very very big personalities in the media business as you can imagine. So he needed to be able to both rally a team and build world-class technology, but also get all of these people with the big personalities to work together and get to do something that they didn't want to do individually and not as a group. And I still don't understand exactly how he managed to get all of these people to do sort of what he wanted, but it certainly wasn't through like alpha male force. It was through not being threatening, being very logical. Uh he's very humble, but I think uh maybe most, you know, Americans specifically can mistake humbleness for weakness. It's the opposite. He's insanely tenacious. Insanely tenacious. Just never gives up. So, so he has a very specific leadership style and and that shaped the company. And as I said, there are a bunch of incredible advantages to his leadership style. One thing that I think uh you know that I do choose to do differently is Daniel loves to to talk in a star pattern one-on-one with many people. So he would talk to me about something separate meeting then with Alex Nordstrom about something the same thing but from you know the business angle and then maybe with PR and finance he just talked in star patent he doesn't like to have meetings everyone in the room and that can be frustrating at time you know it's like he says something to me something to Alex and then Alex and I meet I'm like this is what we're going to do and Alex is like not quite that's not what he said to me and so then you know we go back and we call him out on it you know uh but he likes that star pattern Alex and I chose to do something very different. We chose to synchronize the whole company. And so you can think about Spotify as sort of a two function or. I run product and technology by and large. Alex runs sort of business and content by and large, but instead of sort of doing the divide and conquer, which is a good idea, you divide and conquer because it's more effective. Uh but it has the cost that you're you're unsynchronized. So instead of doing the divide and conquer in separate swim lanes, uh we do this thing which I've I've called synchronous synchronized swimming where we kind of swim together which is much harder to do than than competitive swimming. And when you mess it up, it is ugly to look at, but when you do it well, it's a beautiful thing to look at. So we decided to instead of having our the usual sort of you know I meet with my director reports and we talk about product and technology he meets with his and talks about business and content and then our people meet in the org somewhere and they fight it out which is how most companies work. We chose to say we're not going to have our own leadership meetings. We have a single one with all of our uh SVPs. So we have a single meeting every Tuesday called the E team for three hours with all the SVPs of of all the internal functions like you know marketing ads subs but also you know all the product and technology functions in the same meeting.

这个会议有多少人参加？

How many people are in this meeting?

大约 14 人。

It's about 14 people.

好的。

Okay.

我们每周开会，讨论整个公司的所有事情。

What happens is we meet every week to talk through the entire company.

所以我们讨论的是阻碍用户体验的机器学习问题，比如我们想推出某个音乐视频功能，但那个功能往往卡在许可谈判上，而许可谈判属于 Alex 的部门。所以 ET 会议的根本原因是，在 Daniel 任期内，我们开了太多会，总有人说‘我被那个人卡住了’或者‘那个人没来开会’。于是大家就说‘我们私下再谈吧’。我们听腻了‘私下再谈’这句话，所以我们就开了个会，规定再也不许说‘私下再谈’。所有职能部门的代表都要参加这一个会，这样当你说‘我被许可卡住了’时，许可部门的人就坐在对面，我们当场把问题谈清楚。所以我们改变了公司的运作方式，建立了这个同步运营模式，但它在时间上代价很高。每周 14 到 15 个人开三个小时的会，这是大量的领导层时间，而且这些对话并不是对每个人、每个时刻都相关。有时我们花很多时间讨论内容，可能对某些人不太相关，或者讨论广告策略，可能对个性化团队的人不太相关。但酷的是我们什么都谈。我们谈损益表和资产负债表，也谈机器学习问题。所以现在有 14 个人拥有完整的 CO 视角。

So we talk through machine learning problems that are blocking someone's experience, problems where we want to ship this music video feature and so forth. But often that music video feature is blocked maybe on a licensing discussion that sits in Alex's org. So the whole reason for the ET meeting was that previously, under Daniel's tenure, we were in so many meetings where someone said, 'Well, I'm blocked on that person, or that person is not in the meeting.' So people said, 'Let's take it offline.' And we got so tired of hearing this phrase, 'Let's take it offline,' that we did a meeting and we said, 'You're never allowed to say let's take it offline again. Everyone that represents all functions should be in this one meeting so that when you say, I'm blocked on licensing, well, licensing is right there across the table. Now let's talk it out in real time.' So we changed how the company operates and we have this synchronized operating model that is expensive in time. You have 14 or 15 people times three hours per week. That's a lot of leadership time, and the conversations are not relevant for everyone at every time. Sometimes we talk a lot about content that may not be that relevant, or ad strategy that may not be that relevant to someone in personalization. But the cool thing is we talk about everything. We talk about P&Ls and balance sheets, and we talk about machine learning problems in there. So you have now 14 people that have the full CO perspective.

对，我正想说你说‘不相关’这个词。我刚读完凯利·约翰逊的自传，他在洛克希德内部搞了臭鼬工厂。他的理念就是部门之间没有隔阂——工程、设计、制造，都是一回事。我们一起工作，我们都想要同样的上下文，同样的共享知识基础。

Yeah, I was going to say you use the word 'not relevant.' I just finished reading Kelly Johnson's autobiography. He's the one that did Skunk Works inside Lockheed, and his whole thing was like no separation between any department—engineering, designing, manufacturing—it's all the same thing. We're all working together. We all want the same amount of context, the same shared base of knowledge.

对。

Yeah.

所以其实它是相关的。

So it's like it is kind of relevant.

这只是时间尺度的问题。所以，体验部门的人对广告技术栈了解很多，这在当时看起来不那么相关，但这个人理解我们如何赚钱，这会影响他们未来构建产品时的产品决策，因为猜怎么着？产品里有广告。所以他们不会做出违背我们变现策略的决策。所以这非常相关，只是时间尺度更长。所以我把它看作投资，但这是反常规的。我想很多人会说你不应该坐在一个会议上。我听到有人说‘一旦你觉得会议不完全相关，就离开’。

It's just a question of time scale. So this thing about the person in experience understanding a lot about the ad stack—it won't seem that relevant right then, but then the fact that this person understands how we make money is going to affect their product decisions in the future when they build a product, because guess what? There are ads in the product. So now they're not going to make decisions that go against our monetization strategy. So it is very relevant. It's just relevant on a longer time scale. So I think of it as investment, but it is contrarian. I think many people would say you should not sit in a meeting. I hear people saying, 'The second you're in a meeting and you feel it's not fully relevant, leave.'

埃隆也这么说。

Elon says this.

埃隆也这么说，对吧？但我说了，他没错。他创立了地球上最成功的一些公司。只是方法不同。我们选择了这种方法。所以我们有一个非常资深的团队，拥有公司的完整视角，我认为这对我们公司是对的，因为许可方、我们拥有的关系、我们签订的合同，以及我们如何构建产品之间，有太多依赖关系。像 Spotify 这样的产品很难分而治之。而且我认为有一个真理：最终你的组织架构图会体现在你的产品里。

Elon says this, right? And like I said, he's not wrong. He's built some of the most successful companies on earth. It's just different approaches. We've chosen this approach. So we have a very senior group that has a complete perspective of the company, and I think that's right for our company because there are so many dependencies between licensers, relationships we have, contracts we have, and how we can build a product. It's very hard to divide and conquer a product like Spotify. And I think there's this truth out there that eventually your org chart shows up in your product.

对。而且因为我们有单一体验战略，我们的战略实际上就是一个超级应用。我们试图先构建音乐，然后是播客、书籍，更多的东西集成到一个产品里。我认为我们最大的风险是把组织架构图发布出去。所以如果我们拆分它，让每个人并行运行，我认为大约 6 个月内会感觉很好。然后体验就会开始崩溃，因为你把所有的复杂性都外部化给了用户，而不是在内部解决冲突，最终提供单一体验。所以在组织模式方面，有职能型组织，它有清晰的领导、清晰的所有权，但当你希望两个职能协作时，据说很难合作。有矩阵型组织，它的好处是优化用例，但甚至不清楚你向谁汇报。然后有分部型组织，比如 Riot Games。你按游戏组织，把技术平台拆分成独立的组织。亚马逊实际上有点像游戏组织，分部制，下面有强大的平台。在这个尺度上，我选择非常接近苹果，因为我认为他们解决了构建一个表面下非常非常复杂——极其复杂——但感觉像是一个人为了一个人而构建的问题。

Yeah. And because we have a single experience strategy, our strategy is literally a super app. We try to build first music, then podcast and books and more things into a single product. I think our biggest risk is to ship the org chart. So if we divide it and set everyone just run in parallel, I think it will feel very good for about 6 months. Then the experience will just start crumbling because you're externalizing all that complexity to the user instead of taking the fight internally and coming up with a single experience. So in terms of organizational models, you have functional orgs which have the benefit of clear leadership, clear ownership, but supposedly hard to cooperate when you want two functions to work together. You have the matrix org which has the benefit that it optimizes for the use case, but it's unclear even who you're reporting to. Then you have the division-based org, like Riot Games. You organize by game and you break out the technology platform into its own org. Amazon is actually kind of a gaming org, like divisional with a strong platform underneath. And on that scale, I've chosen to be actually very close to Apple because I think they cracked the problem of building something that is very, very complex underneath the surface—insanely complex—and it still feels like it was built by a single person for a single person.

告诉我你认为苹果是怎么运作的。

Tell me how you think Apple's run.

我觉得苹果有趣的地方在于，从外部看，它是一个职能型组织，对吧？职能型组织——我曾在雅虎工作过一段时间。

What I think is interesting about Apple is when I look at it from the outside, it's a functional org, right? And functional orgs—I used to work at Yahoo for a while.

在哪里？

At where?

在雅虎。我把我的一个公司卖给了他们。

At Yahoo. I sold one of my companies there.

你学到了什么？

And what did you learn?

我在那里没学到什么，但我学到了很多不该做什么。那家公司当时也完全一团糟。我想我在两年半左右经历了四任 CEO。但所有 EVP 和 SVP 都互相争斗至死，试图互相隐瞒信息。所以按理说，我认为苹果不应该有很好的条件来构建那种产品。我一直惊叹为什么那些职能负责人不互相争斗？通常都会这样。我通过与那里的人交谈得出的唯一答案是，这实际上与领导层的任期有关，这也是我试图在 Spotify 内化的原则之一。所以他们的高级领导层任期很长，这意味着这些职能负责人找到了合作同步的方法。所以有时是硬件职能创新并带头，然后软件和服务跟进；有时是软件职能带头，他们跟进。这只有在领导层有超高信任的情况下才有效，我认为这可以追溯到史蒂夫·乔布斯，他们设法保持了这一点。但理论上，我认为职能型组织往往会陷入政治斗争，如果没有任期，没有关键因素。所以在 Spotify，我的大多数直接下属，很多人已经为我工作了 14、15 年。我想平均可能是七、八年。然后职能型组织就能运作。如果你对领导层没有信任，我认为职能型组织只会陷入政治斗争。

I didn't learn anything there, but I learned a lot about what not to do. It was a completely messed up company back then as well. I think I had four CEOs in like two and a half years or something. But all the EVPs and SVPs, they fought each other to the death and tried to hide information from each other. So supposedly on paper, I think Apple should not be very well positioned to be able to build that kind of product. I always marvel that why aren't those functional leads just fighting each other? That's what usually happens. The only answer I've come to as I've spoken to people there is that it actually has to do with tenure of leadership, which is one of the principles I've also tried to internalize at Spotify. So they have very long tenure at the senior leadership, which means that these functional leads have found a way to work together to synchronize. So sometimes it's the hardware function that innovates and takes the lead, and then software and services follow, and sometimes it's a software function that takes the lead and they follow. That only works if you have super high trust in the leadership, which I think stems all the way back from Steve Jobs, and they managed to keep that. But in theory, I think functional orgs tend to dissolve into politics without the tenure, without the key. So at Spotify, most of the people that work for me, many have worked for me for 14, 15 years. I think the average is maybe seven, eight years. And then it works to have a functional org. If you don't have trust in the leadership, I think functional orgs just dissolve into politics.

我想告诉你本期播客的赞助商 RAMP。我最近读了很多关于 SpaceX 的文章。SpaceX 是全球最有价值的私营企业之一，其历史中一个核心主题就是不断挑战和质疑成本。Ramp 帮助全球许多最具创新力的企业做到这一点。使用 Ramp 的企业中位数能削减 5% 的开支。SpaceX 证明了一点：对控制成本的宗教般执着实际上有助于增加收入，因为你可以追求原本无法实现的机会。我们在 Ramp 的数据中也看到了这一点。使用 Ramp 的企业中位数还能增长 16% 的收入。所以当你的企业使用 Ramp 而竞争对手没有时，你就拥有了一个随时间复利的巨大竞争优势。Ramp 是唯一一个旨在让你的财务团队更高效、更快乐的平台。我认识的许多顶级创始人和 CEO 都在使用 Ramp。我自己的企业也在用 RAMP，你也应该用。访问 ramp.com 了解它如何帮助你的企业节省时间、节省金钱并增长收入。那就是 ramp.com。

I want to tell you about the presenting sponsor of this podcast, RAMP. I have been reading a lot about SpaceX lately. SpaceX is one of the most valuable private businesses in the world and one of the main themes in the history of SpaceX is constantly attacking and questioning your cost. Ramp helps many of the most innovative businesses in the world do exactly that. The median company running on ramp cuts their expenses by 5%. And one thing SpaceX has demonstrated is that a religious dedication to controlling costs can help actually increase revenue because you can pursue opportunities you couldn't otherwise. And we see that in the ramp data too. The median company running on ramp also grows their revenue by 16%. So when you're running your business on ramp and your competitors are not, you have a massive competitive advantage that compounds over time. Ramp is the only platform designed to make your finance team faster and happier. Many of the top founders and CEOs I know run their business on ramp. I run my business on RAMP and you should too. Go to ramp.com to learn how it can help your business save time, save money, and grow revenue. That is ramp.com.

这正好呼应了你之前说的，没有唯一正确的做法，而是有多种不同的方式。我刚刚和 Mr. Beast 聊过这个话题，因为你知道，我特别好奇，尤其是在媒体行业，顶尖创意人才非常难管理、难整合、难留住。我问他，你的人员流动率是不是很高？我不是说制作团队的人，而是你真正最顶尖的五位人才，那里是不是有很多流动？他非常崇拜创始人。所以我们每次聊天都会聊企业家历史。我觉得有趣的是，你拿 Larry Ellison 来说，他认为 Oracle 成功的秘诀之一是他们把产品团队的核心成员保持了大约二十年，或者十五年左右。而 Elon Musk，Larry 也指导过他，Elon 说他少数会去请教的人之一就是 Larry Ellison，但 Larry 说的却相反。他说我想要新鲜血液，我想尽可能不断地换人。这就很有意思了，Oracle 超级成功，Elon 也超级成功。

This speaks to your previous statement that there's not one right way to do it. There's multiple different ways. I just had this conversation with Mr. Beast and we talked about this and cuz you know I was like very curious especially in media like your top creative talent is so hard to to manage and to integrate to keep. I was like do you have a lot of turnover on I'm not talking about like the people on the production. I mean your actual your top five most talented people. There's a lot of turnover there. And um he's a huge fan of founders. And so we talk about entrepreneur history every time we we we speak. And I was like what's interesting to me is you take Larry Ellison who he believed that one of the secrets of success Oracle is the fact that they kept I think he said the core kernel of the product team together for like two decades or something like that or maybe decade and a half. And you take Elon Musk who again Larry mentored. Elon says that you know one of the few people he goes to for advice is actually Larry Ellison who says the opposite. He's like I want fresh blood. I want to keep turning through people as possible. It's like, oh, it's like Oracle's super successful. Elon is super successful.

没错。

Exactly.

我觉得这就是关键。你无法赢，你只能少输一点。

I think this is the whole thing. You can't you can't win. You can only not lose as much.

所以我的思考方式是，你有一堆组织原则。就像我加入 Spotify 时，我在 Yahoo 的时候说过，我再也不想在大公司工作了。

So the way I think about it, you have a bunch of organizational principles and as I said as I joined Spotify and had to like when I was at Yahoo, I said I never want to work in a big or

嗯。

Yeah.

我三十多岁加入 Spotify，从没想过它会发展到几千人。最终我不得不面对大公司的问题。然后我读了很多关于组织原则和模式的书，当时我读了很多 Steven Senowski 的作品。

I joined Spotify when I was like 30 something. I never expected it to get to thousands. Eventually I had to do the big or thing. And then I try to read a lot about organizational principles and models and and I think at the time and I read a lot of Steven Senowski.

嗯。

Yeah.

我认为他的论点是你无法赢，没有正确的组织。公司采用一种组织类型，就会在某方面擅长，在另一方面糟糕。人们看到糟糕的方面就说，如果我们改成矩阵式，我们就能在这方面变得很棒。他们改了，现在他们在这方面很棒，但在原来那方面又很糟糕。你只能选择维度。所以我们在 Spotify 做的是，我们说我们必须选择，最好的结果是你为对公司重要的事情进行优化，而在不那么重要的事情上表现糟糕。

And and his I think the thesis is you can't win. There is no right org. And companies they sit with one type of org and then you're going to be good at something and terrible at something. People look at the terrible thing and say like well if we flipped to like a matrix or we' be great at this. They flip now they're great at this and terrible at the old thing. You just you just have to choose dimensions. So what we did at Spotify was we said you have we have to choose like the best outcome is that you're optimized for the thing that is important for the company and you suck at the thing that is not as important.

最糟糕的结果是你在不重要的事情上很擅长，而在重要的事情上很糟糕。

The worst outcome is you're really good at the non-important thing and you suck at the important thing.

所以弄清楚什么重要，然后为此优化，并接受你在其他方面会平庸。我们选择优化的一个方面是单一体验。我们非常擅长构建一个实际上能做十亿种不同事情的单一体验，而大多数人甚至没有意识到后端更加复杂，因为你有音乐业务，那是一个版税池，就像从固定池中按百分比分配，音乐业务的结构与有广告的播客业务非常不同。所以当你在 Spotify 中点击某个东西时，它会触发各种不同的商业模式，后端极其复杂。

So like figure out what's important then optimize for that and just accept that you're going to be average at the others. And so one thing we chose to optimize for was the single experience. We're really good at building a single experience that actually does a billion different things and most people don't even realize that on the back end it's even more complicated because you have a music business that is a royalty pool so like it's percentages out of a fixed pool music business that has a very different structure podcast business that has ads. So when you click something in Spotify it triggers all kinds of different business models like on the back end it is insanely complicated.

我显然是个超级爱读书的人。我以读书为职业，但我在 Audible 上也有几百本有声书。我记得问过 D，到底为什么没人想到，哦我喜欢这本书，直接按播放就好了，而不是像 Audible 那样，我只能听五分钟，还得处理积分，很尴尬。我当时想，你们后端到底是怎么做到的？他也向我解释了后端的复杂性。

I'm I'm a huge obviously love to read. I read books for professionally for a living, but I also have hundreds of audio books on Audible. And I remember asking D like how the hell like why did no one think like oh I like this book just press play it to go instead of like a five minute like audible like I can listen for 5 minutes I have to do the credits like awkward and I was like how the hell did you do that on the back end? He was explaining to me like the complexity on the back end too.

是的，后端是一个非常复杂的模型。

Yeah it's a very complicated model on the back end

尤其是因为你们在设计这个超级应用。

especially because you guys are designing this super app though.

没错。

Exactly.

如果是三个不同的应用，你本可以让不同的团队并行运行。所以，你们在优化单一体验，但你们有没有一个组织原则，比如北极星？这就是为什么我花了很多时间，我非常钦佩，我过去 14 个月里去了斯德哥尔摩大概四次，因为我觉得最好的创始人通常能用几句话解释他们的人生使命。比如我们提到了苹果，让我们谈谈 Steve Jobs。Steve Jobs 用三个字做到了：极致出色的产品。他 25 岁时你会听到他说，临终时他也在说极致出色的产品。你想做出极致出色的产品，那就跟我来。但你知道什么对他重要。

Let me if it was three different apps you could have again you could have had like different teams running in parallel. So, you're optimizing for a a a single um experience, but do you have I feel you have an organizing principle like a northstar and this is why I spend so much time and I really admire like I've been to Stockholm like I think four times in the last like 14 months because I I think of like um the best founders usually can explain just a handful of words like what what their life's mission is. So like we we mentioned you mentioned Apple. Let's talk about Steve Jobs. Steve Jobs did it in three words. He said insanely great products. You could hear him when he was 25. He talked about insanely great products when he was dying. Insanely the great products. You want to make insanely great products, then you come with me. But you knew what was important to him.

而且我觉得，我不知道你们内部是否这样做，但

And I feel I don't know if you guys do this internally, but

我和 Alex、Daniel 以及你聊过很多。我觉得你们的组织原则是“时间花得值”。

I've talked to, you know, Alex and Daniel and you a bunch. I feel your organizing principle is time well spent.

如果你看我们关注什么、优化什么，你有一个战略视角，然后有一个商业视角，再有一个情感视角。战略视角是我们很久以前就意识到，世界上最大的挑战将是分发。你知道当 iPhone 问世时，感觉分发是免费的。人们每天去应用商店，看看免费和付费应用排行榜，但那个时代结束了。通过 Facebook 的安装广告有很多分发，但那些也消失了。

If you look at what we focus on and what we optimize for, you have a strategic lens, then you have sort of a a business lens, and then an emotional lens. So the strategic lens is really that we saw a long time ago that the biggest challenge in the world is going to be distribution. You know when when the iPhone came out, it felt like distribution was free. People went to the app store every day and looked for like you know what's the top charting free and paid app that stopped working. There was a lot of distribution through Facebook with all the install ads. That all went away.

所以从战略角度看，当我们进入播客领域时——那是我们做的第二个产品——我们审视了一下，然后说，好吧，这又是一次权衡。我们可以做一个独立应用，对我们组织来说会更容易，因为我们可以拆分组织，人们可以并行运作等等。但最大的问题是设计体验，还是真正的问题在于获取分发？然后我们看了应用商店，我想大概有 10 个，但至少有 7 个非常好的播客应用，比如 Overcast 之类的。当我们看它们的使用量时，我觉得 Apple Podcast 仍然占了 98.5%。

So from a strategic point of view, when we were going into podcast, which was like the first second product we did, we looked at that and said, okay, there again it's all a trade-off. We could do a separate app. It would be easier for us as an org because we can divide the org and people can run in parallel and so forth. But is that the biggest problem to design the experience, or is the biggest problem actually getting distribution? And then we looked at the app store, and I think there was maybe, probably 10, but at least seven really good podcast apps in there, like Overcast and all these apps. And when we looked at their usage, I think Apple Podcast was still like 98 and a half percent.

所以问题不在于没有足够好的播客产品，而在于它们没有获得分发。

So like the problem wasn't that there wasn't a good enough podcast product. The problem was that they didn't get distribution.

所以我们选择承受做一个单一应用带来的所有复杂性，以换取触达当时已经有的 3 亿多用户。但这难得多。我们选择在某处承受痛苦和成本，以在别处获得好处，我们说，你知道，这只是软件，它应该能适应使用场景，对吧？这在当时是反潮流的。然后对于有声书我们也做同样的事。所以这背后有战略视角。我们深切关心的第二件事是，当我们需要排序时，我们确实把用户放在我们自己之上，放在出版商和唱片公司之上。我们确实优先考虑消费者体验。

So we chose to take that pain of doing a single app with all the complexity that comes with that, for the benefit of reaching what was then already 300 and something million users maybe. But it was much harder. We chose like pain and cost somewhere to get benefit somewhere else, and we said, you know, it's just software, it should be able to adapt to the use case, right? Which was contrarian at the time. And then with audiobooks we do the same thing. So there was a strategic lens to that. The second thing we care deeply about is when we have to prioritize, we do prioritize the user over ourselves, over publishers and labels. We do prioritize the consumer experience.

你说我们把用户放在自己之上，这非常重要。所以，几个月前，我们在瑞典一起赛车。然后我们休息了一下，吃了午饭。有我、你、丹尼尔，还有在 Prima Material 工作的皮娅。第一，我知道你超级好胜。顺便问一下，输给一个女孩你感觉如何？

You saying we prioritize the user over ourselves is super important. So, a few months ago, we were racing cars together in Sweden. We then took a break from racing cars and we had lunch. It was me, you, Daniel, and Pia, who works at Prima Material. One, I know you're super competitive. How did you feel about losing racing to a girl, by the way?

输给皮娅我感觉很好。她是个很棒的车手。她完全配得上胜利。她是我最喜欢的人之一。我愿意为 Prima Material 做任何事。我爱那里的整个团队。但这就是我有点理解你哲学的地方，因为我们之前也有过多次对话。我们有一次很长的午餐。但这就是我认为你们在科技行业很特别的原因，也是我喜欢和你们共度时光的原因。而且 Spotify，你知道，在你最喜欢的 AI 聊天应用之外，就像你手机上最好的应用是什么？对任何使用它的人来说，Spotify 都是前两名。但就是那个——

I felt great about losing to Pia. She's an awesome driver. It was very well deserved. She's one of my favorite people. I do anything for Prima Material. I love the entire team there. But the com— this is where I kind of understood your philosophy, because we also previously had multiple conversations. We had this really long lunch one time. But this is what I think you guys are special in the technology industry and why I like spending time with you. And Spotify's, you know, out of whatever your favorite AI chatbot like chat app, it's like what's the best apps on your phone? Like Spotify is in the top two, you know, for anybody that uses it. But it's that—

你在午餐时谈到过。你知道现在可以往 Spotify 里加入一些黑暗的参与模式，那会让人们更多使用应用，提高留存率，也许带来更多用户，但对他们不好。

You talked at lunch. You knew there's dark engagement patterns you can put into Spotify right now that would have people use the app more, would have more retention, maybe bring more users, but it's not good for them.

是的。所以这是我们的指导原则。关心用户意味着什么？你可以关心用户体验，但如果你以更深层的方式真正关心用户，那就是他们对你所做的事情的感受。Spotify 的情况是，你可以说我们是侥幸，因为我们从音乐开始。音乐被大多数人认为是几乎毫无疑问的好事。也许某处有坏音乐，但你知道，很少有人觉得音乐对世界是坏事。而且它恰好有非常大的受众。几乎没有人不听音乐。这比社交网络还大，对吧？不是每个人都用社交网络，但每个人都听音乐。

Yeah. So this is our guiding principle. What does it mean to care about the user? You can care about the user experience, but if you really care about the user in some deeper way, it would be how they feel about what you do. And what happened with Spotify is you can say that we lucked into it because we started in music. Music is by most people considered almost an unequivocal good. Maybe there's bad music somewhere, but you know, very few people feel that music is a bad thing for the world. And it also happens to have a very large town. There is almost no one that doesn't listen to music. It's like bigger than social networking, right? Not everyone does social networking, but everyone does music.

所以我们开始于这个对世界非常积极的领域。

So we started in this space which was very positive for the world.

而且我认为这是丹尼尔的深思熟虑的决定。他对音乐充满热情。他想做有意义的事情。我不认为这是运气，因为如果你看他现在做的事情，它处于同样的对世界和欧洲有益的领域，而不仅仅是更多的钱。

And I think that was a deliberate decision from Daniel. He was passionate about music. He wanted to do something meaningful. I don't think it was luck, because if you look at the things he's doing now, it's in that same space of being good for the world and for Europe etc., rather than just more money.

是的。所以我不认为这是偶然，但你仍然可以说我们是侥幸。但后来我们意识到这很重要。我们带来了很多快乐。所以当我们考虑下一步做什么时，我们看了播客，我们看到了这个东西，你知道，当时互联网上的模式已经是越来越短的内容形式，越来越短的注意力持续时间。看起来世界迷失了，就像这就要下地狱了。但然后你看播客，你说不，同时有这些非常长的深度讨论，人们用完整的句子说话，能解释完整的思想。他们不一定受到攻击。他们能反驳等等。我们说我们认为这实际上对世界有益。我们不知道它会变得多大。所以我们不得不冒险，但我们认为这是一个商业机会，也对世界有益。那是我们的视角。所以我们进入那个领域，把播客建进了 Spotify。有声书也是一样。我们认为书籍是世界上最重要的事情之一，我对书籍非常热情，但它也是一个好的商业机会。我们很幸运在瑞典，那里由于某种原因在媒体习惯上相当先进。所以在瑞典，有声书已经是非常大的事情。我们对市场有相当高的信心。所以我们持续进入这些我们认为和用户认为时间花得值得的领域，但我们从未在外部完全阐明这一战略。我们只是最近才开始谈论它。

Yeah. So I don't think it was an accident, but still you could say we lucked into it. But then we realized that this is important. We're bringing a lot of joy. So when we looked at what to do next, we looked at podcasts and we saw this thing where there was this, you know, the pattern on the internet already back then was shorter and shorter form content, shorter and shorter attention spans. And it looked like the world was lost, like this is just going to go to hell. But then you looked at podcast and you said no, at the same time there are these very long deep discussions where people speak in full sentences, get to explain full ideas. They don't necessarily get attacked. They get to retort and so forth. We said that we think this is actually good for the world. We didn't know how big it was going to be. So we had to take a chance on that, but we thought it was a business opportunity and good for the world. That's our lens. So we went into that and we built podcast into Spotify. And it was the same thing with audiobooks. We think that books is one of the most important things that happen in the world, and I'm very passionate about books, but it's also a good business opportunity. And we were fortunate to be in Sweden, that for some reason tends to be pretty advanced in media habits. So in Sweden, audiobooks is already very big thing. We had quite high conviction over the market. So we've kept going into these things that are considered by ourselves and users time well spent, but we never fully articulated that as a strategy externally. We've only recently started talking about it.

但你们在内部是否将其阐明为战略？

Did you articulate it as a strategy internally though?

很长一段时间，我们有一个叫“无悔”的战略，但我们基本上没有对外说过。我不知道“无悔”是不是最好的。

For a long time, we've had a strategy called no regrets, but we largely haven't said that externally. And I don't know if no regrets is the best.

我觉得时间，你知道，那是个低标准。

I think time, you know, it's like that's a low bar.

但我们一次又一次地说，这真的是无悔吗？这来自哪里很有趣。我们通过第三方匿名调查了我们的用户。我们一直这样做。有人来问你，你觉得这些媒体服务怎么样，而不知道他们代表谁。所以你得到真相。这就是我们如何理解我们与竞争对手等的表现。所以我们让他们调查所有大型媒体平台。这就是“无悔”的来源。我们问了这样的问题：你觉得你花在 Spotify 上的时间怎么样？所有平台，你花在 Spotify、YouTube、Apple Music、Amazon、TikTok 上的时间，都只是了解你的感受。

But we've said it again and again, like is this really no regrets? What that comes from is interesting. We surveyed our users through a third party anonymously. We do that all the time. Someone comes and asks you what do you think of these media services without seeing who they represent. So you get the truth. That's how we understand how we're doing versus competitors and so forth. So we had them survey across all the big media platforms. And this is where the no regret comes from. We asked the question like how do you feel about the time you spent using Spotify? All of them, the time you spend on Spotify, YouTube, Apple Music, Amazon, TikTok, all of them, just understand how you felt about it.

但然后我们也问了相反的问题。之后你后悔了多少时间？有趣的是，当我们拿回那份调查时，有两件事让我惊讶。一件不太让我惊讶。

But then we also asked the opposite question. How much of your time did you regret afterwards? And what was interesting was when we got that survey back, there were two things that surprised me. One didn't surprise me so much.

我们实际上是互联网上遗憾度最低的内容，就我们调查的平台而言。或者反过来说，是互联网上时间利用最充分的，Z 世代表示他们几乎 90% 的时间都觉得很值得，事后感觉非常好。所以那很好。但让我惊讶的是相反的情况，不点名是哪些平台，在许多大型平台上，人们几乎 60% 或更多的时间都感到后悔。而且这些都是年轻人。让我震惊的是，在我看到那份调查之前，我知道这些平台参与度极高，但我错误地以为他们是因为想待在那里才待在那里的。我以为他们可能很享受。结果当你问他们时，他们说：‘不，我被困住了。我对花的时间感到糟糕。’

We were actually the lowest regret content on the internet in terms of the platforms we asked about. Or if you flip it around, some of the time most well spent on the internet, with Gen Z saying that they value almost 90% of the time they spend, they feel very good about it afterwards. So that was good. But what surprised me was the opposite, without naming which ones, on many of the big platforms people regretted almost 60% of the time they spent or more. And these were young people. What shocked me was before I saw that survey, I knew that these were insanely high engagement platforms, but I kind of mistakenly thought that they were there because they wanted to be there. I thought they were probably enjoying it. Turns out when you ask them, they're like, 'No, I'm trapped. I feel horrible about the time I spend.'

所以，你错误地以为他们待在那里是因为喜欢。它只是非常吸引人。

So, you were mistaken in thinking that they were there and they loved the time. It's just very captivating.

所以我们决定，好吧，我们内部有这个说法已经有一段时间了。让我们把它变成一个真正的战略，成为互联网上时间利用最充分的内容之一。所以当我们看某件事时，比如我们现在谈健身，为什么？因为我们觉得很少有人后悔做锻炼。

So, then we decided, okay, we've had this as a saying internally for some time. Let's just make it an actual strategy to be some of the most time well spent on the internet. And so when we look at something, you know, we're talking about fitness now, for example. Why? Because we think very few people regret doing a workout.

你说谈健身是什么意思？

What do you mean you're talking about fitness?

我们刚刚宣布了一些与多家公司的合作，我们还在加大投入。你知道，Spotify 上近 70% 的人会边用 Spotify 边锻炼。所以我们在健身方面投入更多，想成为更好的健身应用。所以这是一个领域，就像音乐、播客、书籍一样。我们只是投入更多。

We just announced some partnerships with a bunch of companies and we're investing in more. You know, almost 70% of people on Spotify work out with Spotify. So we're investing more into fitness, into becoming an even better fitness app. So this is one area, kind of like music, podcast, books. We're just investing more in it.

那会是什么样子？我的搭档 Rob 今天早上在中央公园跑步时听 Spotify。所以他是在用你的应用锻炼。

What would that look like? My partner Rob was running through Central Park this morning listening to Spotify. So he was exercising while using your app.

好的。这将在几周后推出。所以我会告诉你我们将在投资者日谈论什么。我是个跑步者。我对跑步充满热情，我们看到很多人用 Spotify 跑步，他们自己做播放列表。我希望能够实现的一件事，而且很快就能实现，就是你进入 Spotify，你告诉 Spotify，这是 AI 的一个好用例。我想你对 Spotify 说：‘我想要一个 8 分钟一英里的跑步播放列表，要符合我的品味，但要在重拍上，要么在我的双脚上，要么只在一只脚上，对吧？而且我希望你每周更新。’这是一个相当复杂的任务。所以我们想要做的是首先计算 8 分钟一英里意味着什么。那是什么意思？嗯，大致如果你的平均身高，可能每分钟 165 步左右。好的。现在我们要去找音乐，要么是 160，要么是它的一半 80。所以要么每一步，要么隔一步。好的。你会在那里找到一些音乐，但不多。所以现在你想看得更广一些。然后你实际上想把它们加速或减速到正好 160 或正好 80。然后你实际上想把它们混合在一起，完美地匹配节拍过渡，对吧？然后你实际上想在它上面叠加评论，说：‘现在你要进入间歇部分。现在你要放慢速度’等等。所以这些就是我们正在构建的体验。这就是我说的深入健身的意思。那是无遗憾的时间领域之一。

Okay. So this is going to come out in a few weeks. So I'll tell you kind of what we're going to talk about at Investor Day. I'm a runner. I'm passionate about running and we see a lot of people running with Spotify and they make their own playlists. One of the things I want to be possible, that's going to be possible very soon, is that you go into Spotify, you just tell Spotify, this is one of the good use cases of AI. I think you say to Spotify, 'I want a running playlist for like an 8 minute mile and I want it to be in my taste, but I want it to be on the downbeat, either on both my feet or just one of my feet, right? And I want you to update it weekly.' It's a pretty complicated task. So what we want to be able to do is to first calculate like an 8 minute mile. What does that mean? Well, roughly if your average height, maybe 165 steps per minute or something like that. Okay. Now we got to go and find music in either 160 or half of that, 80. So it's either in every downstep or every other downstep. Okay. You're going to find some music there but not a lot. So now you want to look a bit broader. And then you actually want to speed them up or slow them down to exactly 160 or exactly 80. And then you actually want to mix them together with perfect beat match transitions, right? And then you actually want to overlay commentary on top of it saying like, 'Now you're going to go into the interval part. And now you're going to slow down' and so forth. So these are the kinds of experiences that we're building. This is what I mean with going deeper into fitness. That's one of those no-regrets areas of time.

没有人会在健身房待一小时或跑一小时步，然后出来感觉：‘天哪，我后悔 70% 的时间。’

No one spends an hour in the gym or an hour running and comes out feeling like, 'Oh my god, I regret 70% of that.'

不。你不会。所以这一直是我们的指导原则。我觉得有趣的是，你知道，我 50 岁了。我一生做过很多有趣的事。你必须弄清楚你真正想把自己的一生花在什么上，对吧？我想把自己的一生花在让我个人感觉良好、让我感到自豪的事情上。你感觉好，但你知道，一个人感觉好的方式是如果他们做的事情实际上是在为他人服务。这是，我知道你懂哲学，这是几千年前人类就有的东西。这不是新想法。一千年前是真的。今天也是真的。现在也会是真的。我和 Milan，他坐在那边，在你来之前我正和他说话，他在 Spotify 工作。我和他谈这个，因为他说：‘你现在在商业排行榜上有两个不同的播客。’我说：‘听着，这个新节目，我上瘾了。我睡不着。这些对话让我兴奋。我是自私地做这件事。这是另一种教育形式。但我真的觉得 Founders Podcast 是我毕生的事业。当我推出它时，我感觉我知道它对世界有益。我知道这些对话对世界有益。我不会做播客垃圾。我想让它有娱乐性，但核心是，我想让你被这些成就伟大事业的人启发和教育。’然后重点就是，知道你也可以做到。

No. You don't. So it's been our guiding principle. And what I think is interesting about it is, you know, I'm 50. I've done a lot of interesting things in my life. You got to figure out what you actually want to spend your life on, right? And I want to spend my life on doing something that I feel good about personally and that I kind of feel proud of. You feel good, but you know, like the way a human is going to feel good is if what they're doing is actually in service of other people. This is, I know you're into philosophy, this is a human thing for yourself thousands of years ago. This is not a new idea. It's something that was true a thousand years ago. It's true today. It will be true now. Me and Milan, who's sitting over there, I was talking to him, he works at Spotify, right before you came. I was talking to him about this because he's like, 'You have two different podcasts on the business charts right now.' And I was like, 'Listen, this new show, I'm addicted to doing it. I can't sleep. I get wired from these conversations. I'm doing it selfishly. It's another form of education. But I really feel that Founders Podcast is like my life's work. And the way I feel when I put that out, I know it's good for the world. I know these conversations are good for the world. I'm not going to make podcast slop. I want it to be entertaining, but at the core, it's like I want you to be inspired and educated by these people that have accomplished great things.' And then the point was just like that knowing that like you can do it, too.

是的。那对世界有益。我这样做感觉很好，它不依附于其他任何东西。不是像‘除非我能赚 X 美元，否则我不会做这个’。我不在乎那个。我在乎的是钱会自己来。我在乎的是向世界输出好东西。

Yeah. That is good for the world. I feel good doing that and it's not tied to anything else. It's not like, 'Oh, I wouldn't do this unless I can make X dollars.' I don't care about that. I care that the money will take care of itself. I care about putting out good things to the world.

我认为你说得对，每个人都会到达那个点，历史上每个哲学家都到达了。问题只是你生命中多晚才到达？我认为你比一些人早得多。

And I think to your point, everyone gets there and every philosopher in history got there. The question is just how late in life do you get there? I think you got there a lot earlier than some people.

我花了 32 年才找到我的道路，又挣扎了五年半才真正能维持生计。现在每天醒来，我对自己所做的事毫无怀疑。我要再次建设。我在挖掘那些古老书籍，有人做了了不起的事，从中寻找想法和见解，我们可以传递给下一代创业者。我在进行这样的对话，我们之前谈过，每个人都说：‘哦，你为你的听众问了这个完美的问题。’我自私地进行这些对话，因为我知道有 1000 万个 David 在听这个，他们有同样的兴趣，想知道你内心是什么。你说你 50 岁了。

It took me 32 years to find my path and another five and a half of struggling until it actually could sustain me. And then now I wake up every day like I have no doubts about what I'm doing. I'm going to build again. I'm mining these old books of somebody did something amazing for ideas and insights that we can push down to the next generation of entrepreneurs. And I'm having conversations like this where we talked before, I was like everybody's like, 'Oh, you asked this perfect question for your audience.' I'm having these conversations selfishly because I just know there's 10 million Davids listening to this that have the same interest that want to know what is in you. You said you're 50.

你一直在……你卖掉了第一家创业公司，然后你工作……你的生活经历太独特了。我和丹尼尔之前就在聊：谁会卖掉自己的创业公司，在大公司工作 18 年，然后当上 CEO？

You've been working at... you sold your first startup, then you work... you have such a singular life experience. It's what me and Daniel were talking about earlier: like, who the hell sells their startup, works at a big company for 18 years, then becomes CEO?

这是一条不同的路，一条不寻常的路。

It's a different path, it's an unusual path.

我想聊聊 AI，因为……

I want to talk about AI, because...

你能先说说之前那件事吗？

Can you say something on the previous thing first?

我觉得“时间花得值”这件事有意思的地方在于……这些话说起来容易。我认为每个人都会说服自己，他们是在为世界做好事，即使那些不是的人也一样，因为大多数人……你知道，对自己完全诚实很难。完全诚实地说“我其实没在为社会做好事”，很少有人这么做。他们说服自己。

What I think is interesting with this time well spent is... these things are sort of easy to say. I think everyone convinces themselves that they're doing good for the world, even the ones who don't, because most... you know, it's very hard to be honest with yourself. Totally honest and saying, like, 'I'm not actually doing anything good for the world.' Very few people do that. They convince themselves.

我有个例子。嗯。

I have an example of this. Yeah.

我想聊聊这个。

That I want to talk about.

嗯。

Yeah.

我们显然会收到各种推销，别忘了……希望你别忘了你的想法。但听着，没有不敬的意思。我最喜欢的电影之一是《教父》。我最喜欢的场景之一是索洛佐去见马龙·白兰度，他说……马龙说：“嘿，别人怎么谋生不关我的事，但我不能掺和进去，因为你在向孩子贩毒，我对这个没兴趣。”我也有同感。就像，我在意……我教给孩子们的原则之一是，第一条，第二条是“管好你自己的事”。但有人向我推销，他们说：“我们想合作，作为预测市场之一进行深度合作。”我在电话里跟他们聊，我一点也不了解他们，但他们描述自己的方式，他们说：“我们在建造一台最大真相追寻机器。”我拿起手机，我打开……我不会说公司名字。我输入网址，然后我说：“老兄，我在你网站上看到的第一件事是谁会赢得新英格兰爱国者队的掷硬币。”如果你只是说我们想赌博、赚大钱，我尊重。但别给我来那套……你在建造最大真相追寻机器。

So, we get obviously all these pitches, and don't forget... hopefully don't forget the way you're thinking. But like, listen, no disrespect. One of my favorite movies is The Godfather. And one of my favorite scenes is where Sollozzo goes to meet Marlon Brando, and he's like... and Marlon's like, 'Hey, it's none of my business how another man makes his living, but I can't get involved in that because, you know, you're dealing drugs to kids, and it's not interesting to me.' And so I feel the same way. It's like, I mind... one of my principles I teach my kids is like, number one, rule number two is like, mind your own business. But I had somebody pitch me. They're like, 'We want to partner, do a deep partnership as one of the prediction markets.' And I'm on the phone with them, and I didn't know anything about them, but the way they describe themselves, they're like, 'We're building a maximum truth-seeking machine.' And I pull up on my phone. I go to... I'm not going to say the company. I type it in, and I go, 'Dude, the first thing I see on your website is who's going to win the coin flip for the New England Patriots game.' If you're just saying we want to gamble and make a ton of money, I respect that. But don't give me this other... you're building maximum truth-seeking machines.

嗯。但这就是我的观点。我认为几乎每个人都得说服自己所有这些。那么，既然要呼吁我们，那这样做到底有没有意义？

Yeah. But that's my point. I think almost everyone has to convince themselves of all that. So, to kind of call on us, then does it actually mean anything to do this?

嗯，这意味着你得避开一些东西。所以我们最近做了一个决定……你知道老播客正在变成视频播客，我们也跟进了。

Well, it means you have to stay away from some things. So one decision that we did recently was... you know how old podcasts are turning into video podcasts, and so we follow that.

我做视频的唯一原因是你们骚扰我。

The only reason I do videos is 'cause you guys harassed me.

没错。所以视频很受欢迎，但有个问题，那就是……你知道，很多人，尤其是家长，他们觉得自己的“时间花得值”现在开始看到这些他们不想看到的视频，对吧？所以他们觉得“时间花得值”在下降。然后我们面临一个决定：视频对参与度有好处，但这些人觉得这对他们的“时间花得值”不好。所以我们做了一个决定，有点反参与度的决定，允许任何人，任何人，如果他们想关掉视频，就可以关掉，只听音频，不仅是为了孩子，也是为了自己。所以这可能意味着我们在那个人身上损失一些参与度。但如果你要有这些原则，你就得接受它们，它们会带来后果。现在，我们仍然认为这是非常好的生意，因为我们大部分收入来自订阅者为他们体验付费。你知道，几乎 90% 的收入。我认为当你每月用钱包投票时，你不会为参与度或花费的时间付费。你会为你感受到的价值付费，对吧？如果你觉得这是“时间花得值”，如果你觉得这是高价值，你会说：“我想为此付费。”所以我认为当你用自己的钱付费时，这与优化对用户有利的东西非常非常一致，而在纯粹的参与度、纯粹的广告模式中，你的激励显然是不惜一切代价最大化花费的时间。所以我认为对我们来说，这是一种奢侈，因为我们大部分收入来自订阅，没有那种压力。也许我们锁定了那个商业模式，也许是有点故意的。

Exactly. So they are very popular, but there was a problem, which was that... you know, a lot of people, specifically parents, who felt that they had this time well spent, now they started seeing these videos in there that they didn't want to see there, right? So they felt that their time well spent was going down. And then we were faced with the decision: video is good for engagement, but these people feel it is not good for their time well spent. So we made the decision, sort of the anti-engagement decision, to allow people, anyone, to turn off video if they want to, and just listen, not just for their kids but for themselves. So that may mean that we take some engagement hit on that person. But if you're going to have these principles, you kind of have to live with them, and they're going to have consequences. Now, we still think this is very good business because most of our revenue is from subscribers paying for their experience. You know, like almost 90% of the revenue. And I think that what you pay for when you vote with your wallet every month, you're not going to pay for your engagement or time spent. You're going to pay for the value you feel, right? If you feel that this was time well spent, if you feel that this was high value, you're going to say, 'I want to pay for this.' So I think when you pay with your own money, that is very, very well aligned with optimizing what is good for the user, whereas in a pure engagement, in a pure advertising-based model, your incentives are clearly to maximize time spent at any cost. So I think it's a luxury for us, because most of our revenue comes from subscription, to not have that pressure. Maybe we locked into that business model, maybe with some deliberate.

这正是我想问你的。你选择订阅模式时不可能预见到这一点，但基于你的第一个产品——音乐，它必须是订阅模式。

That's what I was going to ask you. There's no way you could have predicted that when you chose a subscription model, but it had to be subscription model based on your very first product, which is music.

是的。所以现在我们看到了，我们正在加倍投入，我认为这对 Spotify 来说是正确的赌注。我认为人们越来越意识到他们的时间是如何被捕获的，以及他们对生活的控制越来越少。顺便说一句，这让我们进入 AI 的话题，每个人都在谈论，现在对 AI 有很多负面情绪，我理解，而且有充分的理由。生成式 AI 有可能成为你听过的最令人上瘾的算法，因为现在我们可以如此深入地了解你。那里有黑暗的潜力，对吧？但我想说的是，我们想说的是，有……但它是一种双重用途的技术。你可以选择用生成式 AI 做别的事情。所以我们选择做的，例如，是把对算法的控制权还给用户。所以你可以做的另一件事，我们正在做的，是你可以说：“嘿，这是我们用 LLM 认为的你，基于你所有的收听记录，这是我们在音乐上认为的你。这些是我们认为你感兴趣的播客。这些是我们认为你感兴趣的有声书。”然后我们可以放一个框，你可以说：“不，你错了。也许那是我做的，但那不是我想成为的。我想要更多……我想要更多的传记。我想回到古典音乐，这是我试过的。”但因为我主要听 EDM，主要听 EDM。所以现在我可以进去说：“现在我想听古典音乐，”然后我就得到古典音乐。我认为描述生成式 AI 的最好方式是，最终计算机理解了英语。就像，过去只有大约 100 万 GitHub 开发者能跟计算机对话。现在我们都能了。所以我认为消费公司可以而且应该让每个人都能用简单的英语跟他们对话。所以这意味着在 Spotify，我们过去有这些用户研究团队，他们和大约 10 个用户坐在一起，试图深入了解他们的感受和行为。然后我们把这 10 个用户的应用推广到一百万个用户。

Yes. So now that we see it, we're doubling down on it, and I think it's going to be the right bet for Spotify. I think people are getting more and more aware of how their time is getting captured and how they have less and less control over their life. This, by the way, leads us into AI, which everyone is talking about, and there's a lot of negativity around AI now, which I understand and for good reasons. There is a promise for generative AI to be the most addictive algorithm you have ever heard of, because now we can understand you so deeply. There is potential for darkness over there, right? But what I'm trying to say, what we're trying to say, is like there is... but it's a dual-use technology. You could choose to do something else with generative AI. And so what we're choosing to do, for example, is to give back users control over the algorithm. So the other thing you can do, which we're doing, is you can say like, 'Hey, here's who we think you are using LLMs, and based on all your listening, this is who we think you are musically. These are the podcasts we think you're interested in. These are the audiobooks that we think you're interested in.' Then we can put a box where you can say like, 'No, you're wrong. Maybe that's what I do, but it's not what I want to be. I want much more... I want even more biographies. I want to get back into classical music, which is something I've tried.' But because I listen to mostly EDM, mostly EDM. So now I can go in and say like, 'Now I want to have classical music,' and now I get classical music. The best way I think to describe generative AI is that finally computers understand English. Like, it used to be a small population of about 1 million developers on GitHub who could talk to computers. Now we all can. And so I think consumer companies could and should give everyone access to talk to them in plain English. So it means that at Spotify, we used to have these user research teams that sat with like 10 users, trying to understand deeply what they felt and what they did. Then we took those 10 users and tried to apply that to a million users.

如果你能对每一位用户，全部 7.61 亿人，随时进行那种深度用户研究，因为他们和你对话，你也用高保真英语回应，体验就会根据他们的需求调整，那会怎样？

What if you could do that deep user research with every single user, all 761 million of them, all the time, because you talk to them and they talk back in English, high fidelity, and the experience just adapts to what they want?

所以我现在的使命，某种程度上是把这些算法的控制权交还给消费者。这有点反主流，但我认为这是正确的赌注。

So my mission now is to sort of give back control to consumers over these algorithms. And it's contrarian, but I think it's the right bet.

像你和我这样的人可能想要更多控制权，但你有 7 亿用户。你觉得他们中真正有多少比例……我觉得人们就是希望你替他们做。

People like you and I will probably want more control, but like you have 700 million users. What percentage of them do you actually think... I feel like people just want you to do it for them.

不，你说得没错。我觉得世界上所有事情都遵循幂律分布，对吧？比如听音乐——有些人听得少，有些人听得极多。一切都是幂律。但我觉得有趣的是，Spotify 从中获得的东西——比如你看歌单功能——有些人创建很多歌单。大多数人只有一个歌单，可能是他们的喜欢歌曲，也许两个歌单，你知道，一些独立音乐收藏。有些人有几百个歌单。所以 Spotify 的故事，以及我们最初为什么擅长个性化和推荐，是因为我们有大约 100 亿个用户歌单，那是人们仔细标注的：这首歌和那首歌搭配得好，和那首也搭配得好。这是一个极其宝贵的数据集。但你说得对，它严重偏斜。是少数人——不是几百万，但也不是几亿——做了大量工作，让其他人受益。那些说这些歌搭配得好的人，意味着那些不太热衷音乐的人，他们跟不上新发行或旧目录，却从这些人的工作中获得了极好的推荐。所以我认为这里也会如此。事实上，我知道确实如此。我们看到有些人参与度很高，一直和 Spotify 对话，说不对，那错了，那不好，你应该这样。他们做了很多工作，让自己的 Spotify 变得更好，但我们也知道，另一个看起来很像他们的人，却没有时间或知识去做这些，其实可能想要同样的东西。

No, you're not wrong. I think everything in the world is sort of a power law, right? Whether it's music listening—some people listen a little, and then some people listen extremely much. Everything is a power law. But I think what is interesting, what Spotify has gained from—if you look at playlisting, for example—some people playlist a lot. Most people have one playlist, maybe their liked songs, maybe two playlists, you know, some indie favorites. Some people have like hundreds of playlists. So sort of the story of Spotify and why we got good at personalization and recommendation in the first place was that we had, like, I don't know, 10 billion user playlists, which is people carefully saying this track goes well with this track, goes well with this track. It's an incredibly valuable data set. But to your point, it was heavily skewed. It was a few—not a few million, but not hundreds of millions—that did a lot of work that benefited the others. Those people saying these tracks go well together meant that the others who weren't that into music, they didn't keep up with the new releases or the back catalog, they got amazing recommendations from the work of these people. And so I think that's going to be true here as well. And in fact, I know it is true. We see some people engaging a lot, talking to Spotify all the time, saying no, that's wrong, that's bad, you should be like this. They're doing a lot of work. They're making their Spotify better for themselves, but then we know that someone else who looks very similar to that, but doesn't have the time or knowledge to do all of that, actually probably want the same thing.

而他们可以从其他人那里受益。

And they can benefit from the other.

所以他们从其他人那里受益。

So they benefit from the other.

我有个类比——因为大家都说，‘哦，你知道，每个人都会构建他们想要的一切。’我当时想，‘我不觉得他们会去构建。’告诉我这是不是一个愚蠢的类比。我很好奇你个人的看法。但比如，你看在社交媒体上发帖有多容易？你放一张照片，或者点两个按钮，就完成了。而且差不多是，你知道，90% 的内容来自 1% 的人，然后接下来 10% 的人会做，然后大约 90%、85% 或 90% 的人只是潜水，什么都不做。但我觉得这很好地代表了人性，大约 90% 的人不会构建任何东西。

The analogy I had—because everybody's like, 'Oh, you know, everybody's going to be building everything they want.' I was like, 'I don't think they're going to build.' Tell me if this is a stupid analogy. I'm curious your personal opinion on this. But like, if you even look at how easy is it to post to social media? You put a picture or like two buttons and you're done. And it's something like, you know, 90% comes from 1%, and then you have the next 10% do, and then like 90, 85 or 90% just lurk, they do nothing. But I feel like that's a good representation of humanity, where like 90% is not going to build anything.

有个著名的说法：90——对，1% 创造，9% 策展，90% 消费。

There's this famous one: 90—right, 1% creates, 9% curates, and 90% consumes.

对，差不多是这样。我觉得这在很大程度上是对的。

Right, something like that. And I think that's largely true.

这就是我说的幂律。但正如我所说，那 1% 创造的人和那 9% 策展的人，对另外 90% 的消费者来说极其宝贵。然后让你能替他们做，我认为这是大多数人类想要的。

That's what I mean with the power law. But as I said, that 1% that creates and that 9% that curates, it's incredibly valuable for the other 90% that consumes. And then allows you to do it for them, which I think is what most humans want.

Deal 是最好的创始人把世界变成他们人才库的方式。我研究历史上最伟大的创始人如何运作已有十年，他们都有一个共同点：他们明白招聘和雇佣最优秀的人才是最重要的优先事项。A 级人才会识别其他 A 级人才，这就是为什么像 Ramp、Shopify、11 Labs、Uber 和 DoorDash 这样的顶级公司都使用 Deal。我认识的许多顶级创始人都在使用他们的产品后亲自投资了 Deal。他们发现，Deal 是世界上构建全球招聘基础设施最好的公司。Deal 将帮助你的企业在全球任何地方雇佣、支付和管理任何员工。这样你就可以在任何地方留住最优秀的人才，并把剩余时间专注于你最擅长的事情——为客户提供价值。11 Labs 的创始人对 Deal 能给你的公司带来的价值有一个很好的描述。他说：‘我们建立 11 Labs 是为了打破语言和沟通障碍。借助 Deal，我们能够在任何地方雇佣和支持杰出人才，从而加速创新，把更多声音、故事和想法带到世界的每个角落。’超过 40,000 家企业信任 Deal。今天就访问 deal.com/enra，了解他们如何帮助你的企业。那就是 deal.com/enra。

Deal is how the best founders turn the world into their talent pool. I've been studying how history's greatest founders operate for a decade, and one thing they all have in common is they understand that recruiting and hiring the very best talent is your most important priority. A players recognize other A players, which is why top companies like Ramp, Shopify, 11 Labs, Uber, and DoorDash all use Deal. Many of the top founders I know have personally invested in Deal after using their product. And what they discovered is that Deal is the best company in the world at building infrastructure for global hiring. Deal will help your business hire, pay, and manage any worker anywhere in the world. So you can retain the best talent anywhere and spend the rest of your time focusing on what you do best, delivering value to your customers. The founder of 11 Labs has a great description of the value Deal can give your company. He said, 'We built 11 Labs to break down language and communication barriers. With Deal enabling us to hire and support exceptional talent anywhere, we can accelerate our innovation and bring more voices, stories, and ideas to every corner of the world.' Deal is trusted by over 40,000 businesses. Learn how they can help your business today by going to deal.com/enra. That is deal.com/enra.

我觉得这里需要一些背景。我 Spotify 上收听量第四的播客是《Spotify：产品故事》，由你讲述，那是 2021 年、23 年出的，记得吗？

There's some context I think people need here. So the fourth most listened to on my Spotify rap for podcast was 'Spotify: A Product Story,' which you narrate, that came out in 2021, 23, remember?

差不多吧，跨了几年，是很久以前的事了。

Something like that, over a couple of years, came out a long time ago.

它讲述了历史，我觉得更多公司应该这样做。我经常跟创始人说这个，因为本质上，我们为什么不直接讲述自己公司的历史呢？然后你采访当时在场的人，比如肖恩·帕克在里面，丹尼尔·埃克也在，那些扮演过角色的人，即使他们只扮演了几年的角色。

And it tells the history, which I think more companies should do. I tell founders about this all the time, because essentially, why don't we just tell our own company history? And then you interview the people that were there at the time, like Sean Parker's in there, Daniel Ek's there, people that played a role even if they only played a role for a few years.

是的。

Yeah.

但你有整整几集，我想是 2021 年或 2022 年的，专门讲你如何使用 Spotify。你知道，人们谈论 AI 已经 60 年了，但即使在当时，谈论的人也比现在少得多。告诉我你当时在 2021 年、2022 年对 AI 的看法，与现在相比有什么不同。

But you have entire episodes, I think from 2021 or 2022, dedicated to how you're using Spotify. You know, people have been talking about AI for 60 years, but even back then, way less people were talking about that now. Tell me the difference of how you view AI back then in 2021, 2022, to Spotify, to today.

嗯。

Yeah.

所以我对 AI 的兴趣，我想你们很多人都知道，我从小到大一直很感兴趣，后来成为了一名工程师。在大学里，那是 90 年代末，我就对它感兴趣了。那时候有神经网络，但你什么都做不了，计算机不够强大。然后我进入了所谓的 IT 行业，创办了自己的公司，后来加入了 Spotify。当我来到 Spotify 时，我们已经有了一套大型的音乐播放列表数据集。那里有一些非常有才华的人，特别是一个叫 Eric Bernardson 的家伙，当时是 2009 年，他在做机器学习，用的是所谓的协同过滤，跑在世界上最大的 Hadoop 集群上。我觉得这太酷了——这种规模的数据和这些算法。我很久没想过这些了，所以深受启发，又回到了这个领域。后来 DeepMind 开始做基于强化学习的东西，他们有个叫深度 Q 网络的东西，开始打败 Atari 游戏，如果你还记得的话。那时候它看起来像个玩具，就像所有东西一样，但我已经深深着迷于一个问题：从哲学角度看，我们人类到底是如何运作的。我没有好的答案，没有人有，但我有个想法，也许它是计算性的，因为人工神经元就是模型。但在 90 年代，很难相信这些来自 Demis、David Silver 和整个 DeepMind 团队的深度新网络。你可以看到它的 Scaling（规模扩张）在应对越来越复杂的问题。所以我深入进去了。我回到家，又开始写代码，自己做了循环神经网络、LSTM 和 GRU，纯粹为了好玩，在假期和晚上。我们继续做推荐问题，那是老派 AI，叫机器学习。但当 Transformer 论文在 2017 年 6 月左右发表时，那篇论文叫《Attention Is All You Need》，我在最初几天就读了，完全被震撼了。我开始跟任何愿意听的人讲。所以我早就准备好了。然后我们开始非常积极地投资这个方向。我想如果你外推一下，如果你真的相信那条指数曲线，我测试过几次，那么最终计算机将能够开始跟你说话、推理，你也能跟计算机说话。我们很早就开始为此准备了。我们收购了一家叫 Sonantic 的公司来生产语音。我的意思是，今天你在互联网上到处都能买到语音，但按分钟算很贵。所以如果你想为 7 亿用户每天提供几分钟的语音，你会破产的。所以我们收购了这家公司，用不同的技术非常非常便宜地生产语音，甚至在 LLM 足够聪明、能写出语音脚本之前，因为我们就是赌在拦截那条曲线上，而不是等待它，然后再去构建。所以我认为这源于我对 AI 的个人兴趣，再加上一点运气，能在 Spotify 这样的地方，能接触到大量数据。当你看到它时，你只能下注。我认为这种兴趣是个人兴趣。Spotify 是我尽可能接近 AI 哲学层面的借口，也就是作为人类意味着什么。

So my interest in AI, I think as many of you know, I was always interested growing up and then became an engineer. In university, which was in the late '90s, I was interested in it. There were neural networks back then, but you couldn't do anything; computers were not powerful enough. Then I went into what was called IT, and I did my own companies, and I started at Spotify. When I came to Spotify, we already had one of these large datasets of music playlists. There were some very talented people there, specifically a guy named Eric Bernardson, who at the time, this was 2009, was doing machine learning, something called collaborative filtering, on what was the world's largest Hadoop clusters. I thought, this is super cool—this scale of data and these algorithms. I hadn't been thinking about it for a long time, so I got very inspired by that and kind of got back into it. Then what happened was DeepMind started doing these reinforcement learning based things; they had this thing called the deep Q network that started beating Atari games, if you remember that. It looked like a toy back then, like everything does, but I was already so hooked on the idea of how do we actually work as humans from a philosophical point of view. I didn't have a good answer, no one did, but I had this idea that maybe it is computational, because it was the model of the artificial neuron. But it was hard to believe in the '90s with these deep new networks from Demis and David Silver and the whole DeepMind team. You could see the scaling of it handling more and more complex problems. So I got very deep into that. I went home and I started coding again, and I did my own recurrent neural networks, LSTMs, and GRUs, just for fun, during my vacations and evenings and nights. We kept doing our recommendations problem, which was sort of old school AI, called machine learning. But then when the transformer paper came out, I think in June of 2017, a paper called "Attention Is All You Need," I read it in the first few days and was completely blown away. I started talking to anyone who wanted to listen. So I was primed for that. Then we started investing quite aggressively towards this. I think if you extrapolate it out, if you really believe in that exponential curve, which I've tested a few times, then eventually computers would be able to start talking to you and reason, and you would be able to talk to computers. We started preparing for that quite early. We bought this company called Sonantic to be able to produce voice. I mean, you can buy voice anywhere on the internet today, but it's quite expensive per minute. So if you want to serve 700 million users with a few minutes of voice per day, you're going to go bankrupt. So we bought this company producing voice very, very cheaply with different technology, even before the LLMs were smart enough to produce the script for that voice, because we just bet on intercepting that curve instead of waiting for it and then building for it. So I think it comes from a personal interest in AI, and then a bit of luck being at a place like Spotify that had access to lots of data. When you see it, you have to bet on it. I think the interest is a personal interest. Spotify is my excuse to get as close as possible to the philosophical side of AI, sort of what it means to be a human.

再多说一点。

Say more about that.

嗯，我想你经历人生，我小时候想了很多事情，有几个问题让我学到后非常困惑。一个是这个观念——我不记得具体什么时候，但有人告诉我，你身体里的所有原子，或者说 99% 的原子，每七年就会全部换掉。我说：“等等，什么？每七年？什么？你什么意思？”他说：“是啊，每七年它们就全换了。”但那意味着我不是同一个人。所以如果我认为我是我的原子，我以为我是我的身体。显然这就是我，而你告诉我，如果七年后我见到你，我就是另一个人了。那我到底是什么？如果我不是我的原子，我是什么？你知道，如果你看一块石头，它就是它的原子。你看你自己，七年后回来，同样的原子。但对你我来说，你的原子可能曾经是——首先，大多数原子永远不会被毁灭或创造，对吧？它们基本上是一样的。所以你的身体里可能有一堆拿破仑的原子什么的。当你死了，它们会去别人那里，对吧？所以只是那些原子的构造，在很短的时间里是你，对吧？所以我意识到，你不是你的原子，你是你原子的结构。而结构的另一个词是信息。所以我就想，好吧，那是 DNA。就像 DNA 是蓝图，你以某种方式排列原子。那就是你。然后有人说：“不，那也不是真正的你。”因为如果你长时间思考某件事，如果你长时间弹钢琴，你思考某些模式，你的大脑会生长。你会有一个过度发达的音乐中枢。如果你擅长语言，你的语言中枢会发展。最著名的例子是英国的出租车司机。他们必须背下所有街道，当你给他们做脑部扫描时，他们有一个非常发达的 3D 中心，用于三维思维。所以显然它不是静态的；你思考的东西会以新的模式移动那些原子，而且甚至不是同样的原子；原子会随着时间被换掉。所以现在你不是你的原子，你甚至不是原子的结构；你真的是你的思想。字面上，不是比喻。唯一恒定的是处理信息的模式，对吧？所有做处理的原子都会被换掉。所以现在我想，好吧，我是信息处理。这有点让人费解。然后我接受了它，因为我想，好吧，现在我可以某种程度上永生，因为如果我是信息处理，我以某种方式思考。那才是我真正的样子。你可以用其他原子来实例化我，等等。然后如果我把其中一些带入我的孩子，如果他们以类似的方式思考，比如我 32% 还活在他们身上，因为我就是我的思想，然后我传递我的思想。如果你递归地说，好吧，现在把它们转移到计算机上，你就能永生，因为你就是你的思想，如果那个信息处理在那里，那才是真正的你，不是你的原子。

Well, I think you go through life, and I was thinking a lot as a kid about things, and there were a few problems that really dumbfounded me when I learned about them. One was this notion—I don't remember exactly when, but someone told me that all the atoms in your body, or 99% of them, get switched out every seven years. I'm like, "Yeah, wait, what? Every seven years? What? What do you mean?" Like, "Yeah, every seven years they're all switched out." But that means I'm not the same person. So if I thought I was my atoms, I thought I was my body. Clearly this is me, and you're telling me that if I see you in seven years, I'm someone else. So then what am I? If I'm not my atoms, what am I? You know, if you look at a rock, it is its atoms. You look at you, come back seven years later, same atoms. But in you and I, your atoms may have been—first of all, most atoms never get destroyed or created, right? They're largely the same. So you may have a bunch of Napoleon atoms in your body or something. And when you die, they're going to go to someone else, right? So it's just the construction of those atoms that is you for a very brief time, right? So I got to that realization that you're not your atoms; you're the structure of your atoms. And another word for structure is information. So then I'm like, okay, so it's DNA. Like DNA is the blueprint, and you arrange the atoms in a certain thing. That's you. And then someone said, "No, that's not really it either," because if you think about something for a long time, if you play the piano for a long time, you think certain patterns, your brain will grow. You're going to get an overdeveloped musical center. If you're good at languages, your language center develops. The most famous example are cabbies in the UK. They have to learn all the streets by heart, and when you brain scan them, they have a very overdeveloped 3D center for three-dimensional thinking. So clearly it's not static; what you think about moves those atoms around in new patterns, and it's not even the same atoms; the atoms get switched out over time. So now you're not your atoms, you're not even the structure of the atoms; you literally are your thoughts. Literally, not as an analogy. It's the only thing that is constant is the processing information pattern, right? Everything like the atoms that do the processing get switched out. So now I'm like, okay, I'm information processing. And that was kind of a mind-bender. And then I got comfortable with it, because I'm like, well, now I can sort of live forever, because if I'm information processing, I'm thinking in certain ways. That's really what I am. You could have instantiated me with some other atoms, etc. Then if I bring some of that into my kids, if they're thinking a little bit in the same way, like 32% of me is still living in them, because I am my thoughts, and then I transfer my thoughts. And if you're recursive, you say, well, now transfer them to a computer, and you get to live forever, because you are your thoughts, and if that information processing lives there, that was really you, not your atoms.

所以，我们可能是计算生物、信息处理体，这种想法，我认为是当初让我进入 AI 领域的原因。但在大学时，二十多年前，甚至更久，在 90 年代末的大学里，你什么都做不了。电脑不够强大。所以当你试图实现这些想法时，你构建一个神经网络，比如几个节点，它就慢如蜗牛，毫无亮点，你会想，不，这不可能是它。

So this whole thing of like we probably are computational beings, things information processing, I think it was what got me into AI in the first place. But then at university, two decades ago, more than that, in university in the late '90s, you couldn't do anything with it. The computers were not powerful enough. So when you try to instantiate those ideas, you built a neural network like a few nodes and it slowed to a crawl and it was unimpressive, and you're like, no, that can't be it.

好的，这很有意思。所以你已经思考这个问题几十年了。你必须有这些有利的事件发生，现在让你拥有，我想你们是世界上第二大订阅用户群，仅次于 Netflix。我说得对吗？

Okay, so this is interesting. So you've been thinking about this for multiple decades. You had to have this favorable occurrence of events that now lead you to have, I think you guys are the second largest subscriber base in the world behind Netflix. Am I correct?

是的。

Yes.

而且你们有庞大的数据集。现在技术可以赶上你从大学就开始谈论的东西了。

And you have this huge data set. Now the technology can catch up with the thing you've been talking about since you're in college.

是的。这是一个很好的借口，让我可以整天思考、学习这些东西，和业内最优秀的人一起工作。你知道，我有幸遇到所有这些改变世界的人。我见到他们，和他们交谈，问一些我那些有趣又愚蠢的问题。所以，能从事自己热爱的事情，这是一个幸运的位置。但我感到很幸运，因为我可以这样做，而且我可以在一个我认为仍然对世界有益的领域里做，并且我能产生影响，我可以把 AI 引向我认为好的结果，比如把控制权还给用户。

Yeah. It's a good excuse to get to spend all my day thinking about, learning about these things, working with the best people in the business. You know, I get the fortune of meeting all of these people that are changing the world. I get to meet them and talk to them and ask my interesting stupid questions about, you know, things. So, it's just a fortunate position to be in to get to do something that you're passionate about. But I feel very lucky because I can do that, but I can do it in an area where I think it's still good for the world and I can have impact on, I can steer this towards what I think is a good outcome of AI, like giving back user control, for example.

我不想跳过这一点，因为我真心相信我们有过足够的私下对话，我知道你和所有其他大公司设计产品的人都谈过。我们过去私下讨论过，就像，他们中的一些人，你知道，做出了决定，他们知道，他们优化了。他们说，"嘿，我们跑了这个测试。这对参与度很好。对心理健康很糟糕。"然后他们说，"推出去。"这对我来说太疯狂了。这太疯狂了。这就是我们在开始录音之前谈到的，就像我告诉你的，听起来很傻，但我深深渴望一个史蒂夫·乔布斯还活着的世界，因为我显然热爱技术，但我希望那些构建我们所有人都在大规模使用的技术的人，去构建工具——技术就是工具——来增强人性，而不是试图取代它，或者让你变得更笨、更胖，或者只是被吸进手机里，变得抑郁，我觉得现在很多情况就是这样。

I don't want to skip over that because I truly believe we've had enough private conversations and I know you talked to all the other like people designing products at all the big companies. We talked about this privately in the past and it's just like yet some of them, you know, have made decisions where they know, they have optimized. They're like, "Hey, we ran this test. This is great for engagement. It's terrible for the mental health." And they're like, "Push it." That's insane to me. That is insane. And this is what we were talking about before we started recording was like I told you it's like it sounds silly but like I deeply desire a world in which Steve Jobs is still alive because I obviously love technology but I want the people building the technology that we all use on the scale that we all use it at to like build tools and that's all technology is that enhance humanity instead of trying to replace it or to make you dumber or fatter or just you know sucked into your phone and depressed which I feel is a lot of what's going on now.

是的。

Yeah.

这就是我认为你之前说的，比如播客，我们有这个快速增长的东西，对世界有益，假设你在听对的播客，当然你应该听 David Sun 和创始人的播客，同时这些人却在用 15 秒、20 秒，你知道，5 秒、1 秒的滚动摧毁他们的大脑。这让我想起你之前说的，我想，哦，你回去读杰夫·贝索斯的股东信，他谈到为什么发明 Kindle。他说 Kindle 是对短注意力的解药。他担心人们不再阅读了。

Which is I think what you said earlier was like well podcast we have this thing growing massively that's good for the world assuming you were listening to the right podcast and you should be listening to David Sun and founders of course at the same time when these people are like destroying their brains in 15 20 second you know 5-second 1-second scrolling and it reminded me when you said that earlier I was like oh that you go back and read Jeff Bezos's shareholder letters he talks about why he invented the Kindle. He said the Kindle was an anecdote to short attention spans. He was worried that people weren't going to read anymore.

顺便说一句，作为产品，Kindle 给我留下了深刻印象。不仅仅是技术，比如被动屏幕等等。人们忽略的是商业模式创新。他们做了 WhisperSync，它在后台静默更新，因为他们购买了大量的移动数据。这是一个很棒的产品。

I'm deeply impressed by the Kindle, by the way, as a product. And it's not just the technology like the passive screens and so forth. What people miss is the business model innovation. They did the whisper sync where it silently just updates in the background because they bought a certain amount of, you know, mobile data. It's a great product.

解释一下 Kindle 的商业模式创新。

Explain the business model innovation on the Kindle.

嗯，通常你必须为数据付费，如果你懒惰的话，但他们没有。你可能会说，这是 Kindle，现在去买一个每月订阅的移动套餐来更新书籍。但他们做的是与运营商签订协议。他们计算了平均书籍大小，并做了一个固定成本协议，所有数据几乎永远包含在硬件价格中。他们一次性购买了这东西，然后它就神奇地更新了。当时那很神奇。在所有这些产品创新之上，还有很好的商业模式创新，这是我非常热衷的。我相信技术是变革的必要因素，但不是充分因素。我认为你可以用技术造成混乱，比如盗版，但真正改变的是当你把新技术与新的、往往逆向的商业模式结合起来。这就是 Spotify 关于访问与拥有的故事。顺便说一句，这也是为什么我也希望史蒂夫·乔布斯还在，但也不希望，因为你知道，他多次试图杀死我们。

Well, normally you would have to pay for data if you know, if you would have been lazy, which they weren't. You would have said like here's a Kindle, now go and buy a mobile subscription monthly for that for the books to update. Then what they did was they signed deals with these carriers. They calculated the average book size and they made a fixed cost deal where all the data sort of forever was included in the price of the hardware. They bought this thing for one-time cost and it just magically updates. At the time that was magic. Those good business model innovation on top of all the product innovation which is something I'm very passionate about. I believe that technology is necessary ingredient for change but not sufficient. I think you can cause havoc with technology like piracy but when things really change is when you take a new technology and marry it with a new often contrarian business model. This is the Spotify story of access versus ownership, for example. Which is by the way why I would also love Steve Jobs to be around, but also not because, you know, he tried to kill us many times.

有趣的是，我们有过这样的对话。我们之前请了 Jimmy Iovine 上节目，你知道，Jimmy 是我一直想见的前三个人之一，因为我喜欢《Defiant Ones》，那部关于他和 Dr. Dre 的纪录片。是的。我看了不知道多少遍，现在大概 10 遍了。奇怪的是，我和 Jimmy 是朋友，我们住得很近。嗯，他信誓旦旦地说，他说，我本来要杀了他的。就像 Apple 拖了我的后腿。我说，Jimmy，绝对不可能。他说，你是一家大公司里的小角色，那家公司制造了有史以来最成功的消费产品。而他们是小人物，在一个小地方，瑞典大约有 500 万人。如果他们 Spotify 失败了，他们就会死。如果 Apple Music 没成功，他们也没事。我想他说当他开始和你们打架时，你们只有，我不知道，300 万或 1000 万或 1500 万订阅者。

It's funny, we had this conversation. So, we had Jimmy Iovine on the show earlier and, you know, Jimmy was in like top three people I wanted to meet of all time because I love Defiant Ones, the documentary of him and Dr. Dre. Yeah. And I've watched it, I don't know, now 10 times or something. And it's weird because me and Jimmy are friends and we're like, we live close to each other. Um, and he swears up and down. He's like, I was going to kill him. It's just like Apple held me back. I was like, Jimmy, there's just no way. He's like, you're like a small part in a huge company that built the most successful consumer product of all time. These were small guys in a tiny there's like 5 million people in Sweden and something like that. And they had if Spotify like there was they only had one thing they had I always love this idea was like if you had to pick between this is Josh Kushner's quote where it's like if you had to pick the person that's like the most experienced or the smartest or wants the most, you always bet on the person that wants it the most. They had to if they didn't succeed, they would have died. If Apple Music didn't work out, they would have been fine. And I think he said when he started the fight with you guys, you guys only had like, I don't know, three million or 10 million or 15 million subscribers.

然后你们就把他们打得落花流水。

And then you just smoked them.

当时很多人非常非常害怕。

A lot of people were very, very scared at the time.

因为他是个传奇。现在仍然是。他是个杀手。我是说，老兄。

Cuz he was a legend. And he still is. He's a killer. He's a, I mean, dude.

直到今天，即使他退休了，我的意思是，我和 Daniel 经常谈论这个，因为这太搞笑了，因为他们显然也是亲密的朋友。嗯，是的，他是一个令人敬畏的人，但再次，我认为面对一家更大的公司，这非常迷人。

To this day, even though he's retired, I mean, me and Daniel talk about this all the time because it's just hilarious because they're obviously close friends too. And uh yeah, he's a formidable individual but again I think being onto a bigger company and again like it was just very fascinating.

但想象一下，你是一家相当小的瑞典初创公司，你觉得自己擅长产品，然后世界上最受尊敬的产品公司，以及你最敬仰的人——史蒂夫·乔布斯——说他们要来对付你。不仅如此，他们还收购了 Beats Music，并请来了 Dr. Dre 和 Jimmy Iovine。就像，我们要干掉你。他们在内部会议上就是这么说的。他们给了我们六个月。那是一段艰难的时期。

But imagine that you're a fairly small Swedish startup and you think you're good at product and then the biggest most respected product company in the world and the person you look up to the most, Steve Jobs, says they're coming after you. But not only that, goes and buys Beats Music and brings in Dr. Dre and Jimmy Iovine. Like, we're going to kill you. That's what they literally said in their internal meetings. They gave us six months. Those were hard times.

是的，我们非常努力地工作，但最终成功了。这样的时刻有过几次。

Yeah, we worked very hard, but it worked out. There have been a few of those moments.

不过，关于这一点，有一句很棒的话，因为如果你想想 Jimmy 在音乐行业的经历，那真是令人着迷。他们在纪录片里讲述了，你知道，从在唱片厂扫地开始，然后成为工程师，再成为制作人，然后成为一家新唱片公司最成功的创始人之一，再然后打造产品。我推荐人们听他的故事，是因为能在多个领域、跨越几十年都取得成功的人，那不是运气。他们一定知道些什么。

There's a great line though about this, because if you think about Jimmy's experience in the music industry, it's just fascinating. They tell the story, obviously in the documentary, you know, starting out sweeping the floor in the record plant, and then becoming an engineer, then a producer, and then one of the most successful creators of a new record company, and then building products. The reason I recommend people hear his stories is because people that can succeed in so many different domains over multiple decades, that's not luck. There's something going on that they know.

是啊，你不会倒霉那么多次的。

Yeah, you don't get bad luck that many times.

是的，当然。所以从你的角度来看，那一定很可怕。但我想到的是《权力的游戏》里的一句名言，我对此非常着迷，它说：边缘之人往往最终掌控中心。

Yeah, for sure. So it had to be terrifying from your end. But the one thing I think of is this great maxim from Game of Thrones, which I'm obsessed with, and it says those on the margins often come to control the center.

是的。

Yeah.

而且你想想，没有比这群瑞典人更边缘的了，他们在一个小国里试图接管并建立你们所建立的公司。

And if you think about it, couldn't have been more marginal than these group of Swedish guys in this tiny country trying to take over and build the company you guys built.

我们没有被他们压垮，是有一些坚实的原因的。

There were some good hard reasons for why we didn't get crushed by them.

你觉得是什么？

What do you think they were?

嗯，我认为内容方面非常令人担忧，因为如果只关乎内容和音乐人，你有嘻哈音乐，它在美国有最大的吸引力，还有那些有权势的人。但有一些结构性原因。一是当我们制定战略时，我们押注了三件事，它们基本上都是刻意与苹果对着干的。第一是高端。我们相信苹果，即使他们做了免费产品，他们也会在广告上挣扎。他们试过 iAd，但他们想要太多控制，所以没成功。所以我们赌他们不会做好免费层级。因此我们非常专注于高端和真正好的免费层级。第二是个性化，因为苹果当时几乎可以说是反对数据的。我们赌他们永远不会擅长个性化和推荐，而我们相信那是未来。第三是普遍性。我们赌他们会偏爱自己的产品，永远不会在三星电视或安卓手机上做得好。所以这就是我们的三个赌注：高端、个性化、普遍性。我必须说它们结果非常好。他们仍然在挣扎。他们有一段时间有免费层级，一个电台，后来撤回了。普遍性，我认为他们仍然在非苹果硬件上做得不好。而个性化，我想说我们更好。所以有真正的战略原因，现在仍然有，为什么我们坚持下来了。但对你来说，对我们来说这不是可选的。我们要么死，要么烧，要么背水一战。

Well, I think the content side of the business was incredibly worried because if it was only about content and only about musicians, you had hip-hop, which has the largest appeal in the US, and all these powerful people. But there were some structural reasons. One is when we created our strategy, we bet on three things and they were all basically counter positions to Apple deliberately. One was premium. We believe that Apple, if they even did a product, a free product, they would struggle to do advertising. They tried the iAd, but they wanted so much control it didn't work. So we bet that they wouldn't do a good free tier. And so we focused a lot on premium and a really good free tier. The second was personalization, because Apple was, I would almost say, against data at the time. We bet that they would never get good at personalization and recommendation, and we believed that was the future. The third was ubiquity. We bet that they were going to prefer their own products and never get good on a Samsung TV or an Android phone. So these were our three bets: premium, personalization, and ubiquity. And I have to say they panned out really well. They're still struggling. They had a free tier for a while, a radio, they pulled it back. Ubiquity, I think they still struggle to be good on non-Apple hardware. And personalization, I want to say, we're better. So there were real strategic reasons, and still are, for why I think we held out. But it was also, to your point, for us this was not optional. We either died or burned, or we did this back against the wall.

我认为，就你的观点而言，我不知道当时 Apple Music 之前是什么，但它肯定没有 iPhone 重要。

Where I think, to your point, I don't know what prior Apple Music was at the time, but it was not more important than the iPhone for sure.

我在读《从 0 到 1》时发现了我一直最喜欢的一句话。这句话是：“我注意到的唯一最强大的模式是，成功的人会在意想不到的地方发现价值，他们通过从第一性原理而非公式来思考商业。”这正是 AppLovin 在广告平台 Axon 上所做的。Axon 让你在移动游戏中接触到超过 10 亿的潜在新客户。Axon 让你捕捉到全神贯注的注意力。Axon 广告是全屏视频广告，平均观看时长为 35 秒。这种留存率让其他广告平台望尘莫及。你可以在几分钟内启动 Axon。你设定目标，Axon 实现它。没有复杂的设置，不需要专业知识，Axon 可以快速扩展。他们可以把你的广告展示给超过 10 亿的潜在客户。其他企业已经看到了立竿见影的效果，每天花费达到数十万美元，收入增加了数百万。所以你想在竞争对手都上 Axon 之前快速开始。你可以通过访问 axon.ai/enra 来实现。那就是 axon.ai/enra。

I found one of my all-time favorite quotes when I was reading the book Zero to One. The quote says, "The single most powerful pattern I have noticed is that successful people find value in unexpected places and they do this by thinking about business from first principles instead of formulas." That is exactly what AppLovin has done with their advertising platform, Axon. Axon connects you with over a billion potential new customers inside mobile games. Axon allows you to capture undivided attention. Axon ads are full screen video ads that are watched for an average of 35 seconds. That is retention that blows other ad platforms out of the water. You can launch on Axon in minutes. You set the goal and Axon achieves it. There's no complex setup, no expertise needed, and Axon scales quickly. They can put your ads in front of over a billion potential customers. Other businesses have seen immediate results, have scaled to hundreds of thousands of dollars of spend per day and increased their revenue by millions. So you want to get started quickly before all your competitors are on Axon. And you can do that by going to axon.ai/enra. That is axon.ai/enra.

每个人都会知道 Spotify 是什么。但我觉得很多人不知道你是谁。你只做过几个播客。你和 Patrick 做的那期我觉得很棒。但我想深入探讨一些我对你的了解，我认为这些非常不寻常。我们谈过一些社交应用的危险，而你构建了自己的 AI 智能体。你能解释一下你是如何消费 X 的吗？你构建了什么？

Everybody's going to know what Spotify is. I don't think many people know who you are. You've only done a few podcasts. The one you did with Patrick I thought was excellent. But I want to go into some of the things that I know about you that I think is very unusual. We talked about the dangers of some of these social apps, and you have built your own AI agent. Can you explain how you consume X, for example? What you've built?

这是我们正在 Spotify 做的事情，我们几周前发布了这个功能，叫做“保存到 Spotify”。当你使用 Claude Code 或 Codex 之类的工具时，你可以让这些智能体去为你查找信息，写你的播客脚本，然后生成播客，我们允许你将其上传到 Spotify，但以私密方式，对吧？因为你可能在里面放私人信息。也许你想为面试做准备，就像我今天做的那样。我不想让那在 Spotify 上到处都是。所以我们现在支持私密播客或个人播客。所以我做的是，我使用这些智能体，结合我的个人信息，比如我有的文档和我想准备的东西，然后根据我在不同服务中的兴趣。我的所有兴趣，Spotify 从我的所有收听中非常了解我的兴趣，它会出去看看世界上发生了什么，然后给我一个非常个人化的更新，关于世界上可能让我感兴趣的事情。所以这有点像个人算法，过滤掉很多我不想听到的噪音和疯狂，只给我我想听的东西。这是我希望给每个用户的用户控制，但如果你愿意投入工作，你现在就可以拥有这种用户控制。有趣的是，我没想到这一点，但我通过这个发现了这么多音乐和播客。

So this is something we're working on at Spotify, and we released this just a few weeks ago, something called Save to Spotify. When you sit in your Claude Code or Codex or something, you can just ask these agents to go out and look at information for you, write your podcast script, and then generate a podcast, and then we allow you to upload that to Spotify, but privately, right? Because you may put private information in there. Maybe you want to prep for your interview as I did today. I don't want that to be all over Spotify. So we now support private podcasts or personal podcasts. And so what I do is I use these agents and I take a combination of my personal information, so documents I have and things I want to prep for, and then what I'm interested in based across my different services. All my interests, Spotify knows my interests pretty well from all my listening, goes out and looks at what's happening in the world, and then it gives me this update that is deeply personal about what's happening in the world that I might be interested in. So it's kind of like a personal algorithm that filters out a lot of the noise and the crazy that I don't want to hear and just gives me the stuff that I want to hear. This is the kind of user control that I would like to give every user, but it's the kind of user control that you can have now if you're prepared to put in the work. And what's interesting is I didn't expect this, but I found so much music and podcasts through this.

然后它会过滤到我感兴趣的人，我在不同社交媒体上关注的人等等。所以这就是我在乎的人之间讨论的话题。比如，你和 Evan Spiegel 的那期节目一次又一次地出现，在我的个人播客里有很多评论，说这内容对你来说真的很有趣，你应该听听，因为里面有产品智慧等等。我听了整期节目，顺便说一句，我真的很喜欢。我在考虑借用他的框架，比如我们过去有各种想法但资源有限，现在我们既有想法又有资源。我真的很喜欢这个。

Then it's filtered to the people that I'm interested in that I follow on different social media and so forth. So it's the thing that is being talked about among the people I care about. So for example, your episode with Evan Spiegel came up again and again with lots of commentary in my personal podcast saying like this stuff is really interesting for you. You should listen to this because there are product wisdoms in there and so forth. I listened to the whole episode which I really like by the way. I'm thinking about stealing his framing of like you know we used to have all the ideas but limited resources. Now we have all the ideas and all the resources. I really like that.

关于 Evan 的事，你知道人们会跟我说，是啊，但看看他的市值。我就说，我不在乎他的市值。我在乎的是你认为他真正与众不同。我和他相处了很久。最后，我就想让他赢。我不在乎他能不能搞定。然后每个人都像，哦，但他的股票跌了 80% 之类的。所以，我就去查了 CEBT，然后我问，Steve Jobs 经历过多少次 75% 的下跌？你知道，大概四次。这在每个人的职业生涯中都会发生。这孩子还年轻。他不是孩子了。这家伙还年轻。让我们看看未来十年会发生什么，你知道，但事实是他有很多别人没有的非常非常有趣的想法。而且他在社交媒体上发明了一种每天有十亿人使用的模式。

The thing about Evan, you know people give me they're like yeah but look at his market cap. I was like, I don't care about his market cap. I care about that you think he is truly differentiated. And I spent a bunch with him. At the end, I was like, I just want him to win. I don't care about his like figure it out. And then everybody's like, oh, but his stock's down 80% or something like that. So, I just went to CEBT and I was like, how many times did Steve Jobs suffer a 75% decline? You know, it was like four times. Like, this happens in everybody's career. The kid is young. He's not a kid. The guy is young. Like, let's see what happens over the next decade, you know, but the fact is like he's got a lot of very very interesting ideas that no one else has. And he invented a pattern on social media that a billion people use every day.

他发明了好几个，对吧？比如阅后即焚的图片故事，所有这些。

He invented several of them, right? The disappearing image stories, like all of these.

你觉得听那个人讲一个小时不有趣吗？你疯了。

You think that guy is not interesting to hear talk for an hour? You're crazy.

当然。不，我喜欢他。也许因为我是产品出身，他是设计出身，很接近，但我确实喜欢他。

Of course. No, I like him. Maybe it's because I'm a product guy. He's a design guy, so close, but I like him for sure.

这是你身上最迷人的一点，我认为这真的非常重要，值得深入探讨，而且人们如果不这样做会很危险。你一直在努力消除噪音。这个想法甚至在你把它变成音频之前，你本质上是在把文本变成音频，现在有了你刚才提到的功能。但当我们谈到这个，或者也许我听你说过，也许我在播客里听到你讲过，但你之前可以读它，对吧？它就像剥离掉 X 上发生的所有疯狂，或者人们为政治争吵，就是这样。我在乎的人是什么？他们在谈论什么不会让我焦虑？不会提高我的皮质醇水平或让我难过。所以我真的让我的智能体过滤掉愤怒诱饵、点击诱饵、政治内容。我只是要求，这就是你现在可以拥有的用户控制，但很少有人理解你可以拥有，而且太复杂了，但你可以拥有所有这些控制。所以我认为大多数人其实知道自己想要什么，但你知道你会在当下被捕获。就像你想要某样东西，然后有人把糖果放在你面前，你就会拿走那块糖果。所以有趣的是，你能不能让人们提前决定自己的未来。这就是品味档案和这些智能体背后的想法，我并不比别人好。我经常被捕获。所以我特别讨厌路怒。我对它非常着迷。我会陷入路怒的黑暗深渊，因为这是最奇怪的事情。比如那些 supposedly 完全正常的聪明人，他们浪费了整个生命，还进了监狱。

This is one of the most fascinating things about you, and I think it's really really important to drill down on, and it's really dangerous for people not to do. You are constantly trying to eliminate noise. The idea even before you turn this into an audio, you're essentially taking text and turning it into audio now with the feature you just talked about. But when we talked about this or maybe I heard you I think you told me about it, maybe I heard you talk about a podcast, but you were taking it, you could read it though before, right? It was like stripping out all like the crazy that happens on X or the people fighting over politics and it's just like that. What are the people I care about? What are they talking about that's not going to induce me? Like it's not going to raise my cortisol levels or make me sad. So I literally ask my agent to filter for rage bait, filter for clickbait, filter for politics. I just ask this is the kind of user control that you can have now that very few people understand that you can have yet and it's too complicated but you can have all that control. So what I think is most people kind of know what they want but you know you get captured in the moment. You know it's like you want something and then someone puts candy in front of you, you're going to take that candy. So what is interesting is can you give people the chance to decide their own future ahead of time. This is the idea behind like the taste profile and behind these agents like I'm not better than anyone else. I get captured all the time. So my specific pet peeve is road rage. I'm so fascinated by it. I just go down dark holes of road rage cuz it's the weirdest thing. Like these supposedly fully normal smart people, they waste their entire lives and go to jail.

他们向车里开枪，杀死孩子，比如

They shoot into cars and kill kids like

然后事后你意识到，不，他们是很正常的人。只是在车里不知怎么就失控了。所以我陷入那个兔子洞，看几个小时的路怒视频，对吧？但我现在能做的是，我可以说不要给我看路怒，因为我知道会发生什么。所以这叫什么？提前预谋。

and then afterwards you realize that no, they were pretty normal people. It's just somehow inside your car you lose it. So I go down that rabbit hole and I watch hours of road rage, right? But what I can do now is I can say like don't show me road rage cuz I know what's going to happen. So that like what do you call it? Ahead of time premeditated.

对，预谋。

Yeah. Premeditated

预谋媒体。

premeditated media.

对。

Yeah.

这是我认为每个人都应该至少能接触到的东西，如果他们想要的话。它对我有效，我认为对很多人也会有效，而且我认为这种需求不会减少。所以，我认为它会变得更大。所以，我碰巧认为这也是好生意。

Is something that I think everyone should at least have access to if they want to. It works for me and I think it will work for a lot of people and I think that need is not going to get less. So, I think it's going to get bigger. So, I happen to think it's also good business.

你刚才说话时我突然想到一件事。我很好奇谁影响了……所以，我们基本上在谈论你的信息摄入，对吧？以及你让什么进入你的大脑，你谈到我们本质上就是我们的思想。让陌生人进入你的大脑并影响你，那太疯狂了。我昨天和 Mr. Beast 谈过这个，他就说我认为人们不应该用 Instagram，因为它让所有年轻孩子都在比较，比较是快乐的小偷，他们整天做的就是把别人的精彩片段和自己比较，这对让这些东西进入大脑非常非常有害。我最喜欢的一件事，我想 Daniel 去年在这个播客里说过，你知道，我见到人时总是很好奇，尤其是成功人士，我总是问谁告诉你真相，因为很少有人。他们很多人周围几乎都有一群马屁精，过滤着进入他们大脑的信息。而他们会做的是寻找他们可以信任的人，他们会告诉他们真相。我想他讲了你成为他（Daniel）的这样的人的故事，因为你把他拉到一边，你说，嘿，你在负责产品。但你不太擅长。你有点这个。他的反应是他回家，我想他告诉他妻子，他说，天哪，我得解雇这个人。就是你。然后他想了一下，他说，哦，他是对的。就像他帮了我一个忙。

Something that just popped to mind when you were speaking just now. I'm very curious like who influenced... So, we're talking like essentially your information diet, right? And what you're letting into your mind, you talked about we literally are our thoughts. Like you're crazy to let people get into your brain and affect you. And I was talking to Mr. Beast about this yesterday where he's just like I don't think people should have Instagram cuz it's like it's just making all these young kids like comparison is the thief of joy and all they're doing all day long is comparing you know somebody else's highlight reel to them and it's just like very very bad for to let that into your brain. One of my favorite things, I think Daniel said it on this podcast last year where, you know, I'm always curious when I get to meet people, especially successful people, is like who tells you the truth is a question I always ask cuz it's like very few. A lot of them kind of have like this almost like cloud of sycophants around them that are like filtering their knowledge that's getting into them. And what they'll do is like they'll seek out people they can trust and they'll tell them the truth. And I think he told the story of you being one of those people for him, for Daniel, because you pulled him aside, you're like, "Hey, you're running product." And you're like not good at it. You're kind of this up. And his response was he went home. I think he told his wife, he's like, "Oh my god, I have to fire this guy." It was you. And then he goes then he thought about it. He goes, "Oh, he's right." It's like he just did me a service.

很少有人那么坚强能承受那种，但这就是他成功的原因。

Very few people are that strong can take that kind of but that's why he's successful.

是的。所以我认为这真的很重要。那么谁为你做这件事？

Yes. And so I think this is really important. So who does that for you?

所以我做了类似的事情。这又回到了长期任期的概念。所以我有些长期共事的人，但和他们有巨大分歧。他们已经意识到，当他们反驳我，对我生气时，不会有任何后果，对吧？但那种舒适感需要很长时间才能建立起来。

So I do a version of the same thing. This is back to the thing of long tenure. So I have people who are long tenured but like huge disagreements with them. And they've realized that when they talk back to me, they get upset with me, there are no consequences, right? But that takes that comfort takes a long while to build.

你不能指望一个从美国企业界来的人，尤其是他们，能当面顶撞你，因为通常会有后果，你只会被解雇。

You can't expect someone to come in, especially if they come from corporate America, and expect them to stand up to you because usually there are consequences. You just get fired.

哦，是的。但人们不知道你是棕带。

Oh, yeah. But people don't know you're a brown belt.

不不不不，我是蓝带。

No, no, no, no. I'm a blue belt.

哦，蓝带，那你也能勒他们。

Oh, a blue belt, so you could also choke them.

有很多人也能把我勒晕。所以，那是一条冒险的路。

There are many people who could choke me out. So, that's a risky path to take.

这是任期带来的。我身边有一群人，我觉得也希望他们能自在地说：‘不，你错了。我觉得你就是没理解。’我这么认为是因为他们确实这么做了，而且他们知道我不能总是控制自己。我对某些事非常热情。我会说：‘不，你错了。那很蠢。’等等等等。然后他们知道不会有后果，所以他们还是说了。我尽量在单独谈话或小团体里这么做。有时在大会上，一个和我共事很久的人，非常自在，就直接说：‘不，我觉得那是个蠢主意。’我会说：‘不，你的主意才蠢。’然后周围的人都想：‘哦，现在大家都要被解雇了。’所以你必须随着时间建立那种信任。我也有类似的经历。

It's a tenure. I have a bunch of people around me who I think and hope feel comfortable saying, 'No, you're wrong. I think you're just not understanding this.' And I think so because they do and they've learned that I can't always control myself. I'm very passionate about something. I'm like, 'No, you're wrong. That's stupid.' Blah, blah, blah. And then they know it's not going to be consequences, so they say it anyway. I try to do that more in one-on-ones or small groups. Sometimes there's a big meeting, someone who has worked with me for a long time, feels very comfortable, just says, 'No, I think that's a stupid idea.' And I'm like, 'No, your idea is stupid.' and everyone around is like, 'Oh, now everyone is going to get fired.' So you have to build that trust over time. So I have a version of the same.

好的。所以我们在对话一开始就谈到了任期，当时我问 Mr. Beast 关于他人才流失的问题，就是那些最有才华的人，他说这里有一种价值，因为负责运营的那个人已经在那里很久了。他们曾经住在一起。他说我和这个人聊了不止 1 万小时，而是 3 万小时，他们走到片场，因为他负责运营，他们同时运营着大约 10 个项目。Mr. Beast 可以走到片场，在他抬头之前，那个人，他叫 Tyler，就会说，我已经知道这个了，这就是为什么我做了这个决定，我们要这么做，哦，你可能不喜欢这个镜头，但我们要拍这个镜头，因为这样我们可以从这个角度拍，然后实际上什么都没发生。他们就像《星际迷航》里的瓦肯心灵融合。所以就是这样。他说了最棒的话，他说你 100 小时得不到那个。

Okay. So we talked about tenure at the very beginning of this conversation where I never finished a thought when I was asking Mr. Beast about turnover on his talent, like the most talented people in there, and he's just like there's a value here because the guy that runs his operations has been there. He used to live together. He's like I've spent not 10,000 hours talking to this guy, 30,000 hours where they walk on set and since he's running his operations, they're running like 10 at a time. Mr. Beast can walk on set and before he'll look up and the guy, his name's Tyler, he's like, I already know about this and this is why I made this decision and we're doing this and oh, you may not like this shot, but we're doing this shot because that allows us to shoot from this angle and then literally does nothing has even come out. It's like they're what's the thing from Star Trek with the Vulcan mind meld. So that's what happened. He said the greatest thing, he goes you don't get that in 100 hours.

所以任期的一个好处是信任。人们让你保持诚实，除非你变得太自满，这总是成功带来的风险。但另一个好处是效率，比如人们知道你不需要给太多背景。任期也有一些缺点，和 Elon 的相反，你得不到那么多新鲜血液，所以你需要小心。也许最终你只是一群老人，对吧？所以你必须控制这一点。所以我努力确保我们也招聘一些新人。但这两个好处就是效率和信任。

So one benefit of tenure is the trust. People keep you honest unless you become too full of yourself, which is always a risk with success. But the other benefit is this efficiency of like people know you don't have to give as much context. There are some downsides to tenure which is opposite of Elon's you don't get fresh blood as much and so you need to be careful. Maybe you're just a group of old people eventually, right? So you have to control for that. So I try to make sure that we also hire some new. But these are the two benefits just efficiency and trust.

那群老人你必须非常小心。想想 Walt Disney 手下最有才华的人。就像那九个人，他们是动画师。我想 Walt 死后他们被称为老人。我刚刚和皮克斯创始人 Ed Catmull 谈过，他们作为年轻的技术专家进去，试图向他们推销一种新的动画方式。而 Walt 已经死了，这些人老了。没有新鲜血液。他们说计算机永远做不了你说的那些东西。他们拒绝了。拒绝这些新想法是非常不人道的。

The group of old people you have to be very careful with. Think about the most talented people that Walt Disney had. It was like these nine guys, there are animators. I think they used to call them the old men after Walt died. And I just talked to Ed Catmull, the founder of Pixar, and they're going in there as a young technologist trying to pitch them here's a new way to animate. And Walt's dead, these guys are old. There's no fresh blood. They're like the computers will never be able to do the stuff that you're talking about. It's like they reject. It's very inhuman to reject these new ideas.

所以，这又回到了你优化某些东西，但你必须减轻其他东西的风险。所以，如果你优化了任期，你必须减轻没有新鲜血液、职业阶梯缓慢的风险。所以，也许有才华的人觉得，好吧，那要花 10 年才能爬上职业阶梯。所以，你必须为此创建项目，给他们曝光机会。

So, it's back to like you optimize for some things, but then you have to mitigate for the other things. So, if you're optimized for tenure, you have to mitigate for the risk that you don't get fresh blood, that your career ladder is slow. So, maybe people who are talented feel like, well, that's going to take 10 years to get up the career ladder. So, then you have to create programs for that to give them exposure.

Gustav 花了 18 年才当上 CEO。

It took Gustav 18 years to get to CEO.

是的，花了 18 年才当上 CEO。哇，真长。终于做到了。但这是同样的原则。

Yeah, took 18 years to get CEO. Wow. It's a long time. Finally made it. But it is the same principle.

那么，你们怎么获得新人才呢？比如你们在 Spotify 具体在做什么？因为我知道你们有同样的想法。你们和 Shopify 的 Toby 以及我朋友 Ramp 的 Kareem 跟我谈过的想法一样。

Well, how do you get new talent then? Like what are you guys doing literally at Spotify? Because I know you have the same idea. You guys have the same idea that Toby from Shopify talked to me about and my friend Kareem from Ramp did.

顺便说一句，Toby 那集也很棒。

Another great episode by the way, the Toby episode.

是的。他是最棒的。但你们对招聘尖峰人才没问题。

Yeah. He's the best. But that you guys are fine hiring for spikes.

是的。

Yeah.

那么你今天怎么做的？你怎么进入你的大型平台业务，我想刚庆祝了 20 周年，对吧？

So how do you today? How are you getting in your huge platform business around, I think just celebrated 20 year anniversary, right?

嗯。

Um.

那你们怎么获得新的年轻人才呢？你怎么考虑这个？

How are you getting new young talent then? How do you think about that?

有几件事。一是你获得初级人才，然后我们有这个新星计划。我们确实会识别，让经理们识别这些是冉冉升起的新星，所以每年我们都有一个项目。大概 20 到 30 人，他们参加一个项目，去不同的市场，比如印度、美国、德国等，了解不同市场的不同问题。然后他们去和 Alex 坐下来谈。然后他们和我坐下来。这实际上是一个内部 MBA 项目，你学习商业、技术、产品，并在整个组织内建立关系。所以现在这些人 12 个月后出来，他们认识 Spotify 在世界各地的每个人。所以他们可以完成任何事情。所以现在他们会开始更快地上升。所以我们有这些具体的项目来识别人才，帮助他们加速，而不是有机地做事情、被注意到、慢慢往上爬。所以这是一件事。但偶尔你也必须招聘更高级别的人，对吧？所以如果你看看 AI 开始出现或当时被称为机器学习的时候，我觉得我们在深度学习方面特别不够强。我们知道老派的矩阵分解之类的东西。所以我们收购了波士顿一家叫 The Echo Nest 的公司，就是为了靠近 MIT，然后我们在那个领域有了一批人才，并确立了自己的地位。所以有时你必须进行收购，当这是一个全新的技能时，比如如果你需要快速在加密货币领域做大。是的，你可以等几年让组织理解加密货币，学习，你可能必须收购或在高级别招聘。你必须减轻任期的问题。

It's a couple of things. One is you get junior talent and then we have this rising stars program. We literally identify, ask managers to identify like these are rising stars, so then we have a program for them every year. It's like maybe 20-30 people, and so they get to go on a program where they go out into the different markets like India, US, Germany, etc., and learn about the different problems in different markets. Then they go and sit with Alex and talk to him. Then they sit with me. It's literally this internal MBA program where you learn business, technology, product, and you build relationships across the org. So now these people come out 12 months later, they know everyone in Spotify across the world. So they can get anything done. So now they're going to start rising much faster. So we have these specific programs to identify the talent and help them accelerate versus the organic path of them doing something, getting noticed, and slowly working the way up. So that's one thing. But then every now and then you also just have to hire at a more senior level, right? So if you look at when AI started happening or was called machine learning back then, I felt that we just weren't strong enough in deep learning specifically. We knew the old school matrix factorization stuff. So we did an acquisition of a company called The Echo Nest in Boston just to get very close to MIT and then we had a bunch of talent in that field and establish ourselves. So sometimes you have to do acquisitions and just get when it's a completely new skill, something like if you need to get big in crypto quickly. Yeah, you can wait for years for the organization to understand crypto, learn, you may have to acquire or hire at some senior level. You have to mitigate for the problems of tenure.

如果我要做一些牵强的类比，也许这是苹果在机器学习和 AI 方面遇到的挑战之一。

And if I'm going to make some far-fetched analogies, maybe this is one of the challenges that Apple had with machine learning and AI.

平均来说，他们稍微太老了，看不到那个。

On average, they were slightly too old to see that.

什么事情让你夜不能寐？你觉得 Spotify 目前最薄弱的环节是什么？

What's the thing that keeps you up at night? Like what do you think the weakest spot for Spotify is right now?

让我夜不能寐的事情并不出人意料：AI 最终意味着什么。我非常确信，媒体习惯会因为 AI 而改变。所以我认为坐以待毙不是一个选项。但我的思考方式是，世界是以微波形式前进的。你得到一项新技术，它常常会引发一片混乱。然后出现某种新的商业模式，接着事情就变了，对吧？最初的 Spotify 就是这样。我们算是搭上了廉价宽带和个人电脑的微波。然后突然间，它们变得无处不在，宽带速度也足够快，你不再需要拥有文件了。你可以直接流式播放，对吧？我们意识到，我们围绕它建立了一种商业模式，对下载行业等造成了巨大的颠覆。然后你等上几年，一切看起来都很好。接着智能手机出现了，现在一切又变了。有些人没有电脑。我们当时只在电脑上有免费层。我们没有免费层。事实证明，虽然个人电脑和互联网、桌面端的延迟很低，你可以流式播放。但如果你在 2008 年尝试通过蜂窝网络流式播放一首歌，它甚至无法开始。要 20 秒才能开始，还会卡顿。所以一切又变了。然后我们不得不适应这一点。我们做到了，我们想出了离线同步的商业模式和移动模式等等。但接着你会进入这些稳定期。所以如果我是一名分析师，试图预测 2015 年到 2025 年这十年，你知道，世界会是什么样子？我只要简单地外推更多的移动订阅、更多的广告，更多的一切，就会做得非常好，你会非常接近实际发生的情况。但如果你把时间提前 10 年，2005 年到 2015 年，你会外推个人电脑和互联网。你会错过智能手机。你会错过一切。我们会错过 Uber。我们会错过整个世界。所以问题是，我们处于哪个时代？我们处于外推的舒适区，还是处于宏观变革之中？我认为我们正处于一个一切都将改变的时代。所以我无法告诉你什么会改变，但我非常确定事情会改变。但我们作为一家公司的思考方式是，最可怕的时期是变革时期。猜猜 Spotify 增长最快是什么时候？

I think the thing that keeps me up at night is not surprising: what AI truly means in the limit. I'm pretty sure that media habits are going to change because of AI. And so I don't think it's an option to just sit still. But the way I think about it is the world moves in microwaves. You get some new technology. It often causes a bunch of havoc. Then there's some new business model that emerges and then things change, right? This was the original Spotify. We kind of rode on the microwave of cheap broadband and PCs, right? And then all of a sudden they were so ubiquitous and the broadband was so fast, you didn't actually need to own files anymore. You could just stream them, right? We realized that we built a business model around it that became very disruptive to the download industry and so forth. And then you wait a few years, everything looked great. Then the smartphone came along and now everything changes again. You have people that don't have a computer. We only had a free tier on the computers. We had no free tier. Turns out that while the PC and the internet, the desktop had such low latency you could stream it. If you try to stream a song over cellular back in 2008, it wouldn't even start. It was 20 seconds to start. It would chop. So like everything changed again. And then we had to adapt to that. And we managed to we came up with the offline sync business model and the mobile model and so forth. But then you get these periods of stability. So if I was an analyst and I try to predict the world between 2015 to 2025 those 10 years just you know what will the world be like? I would have done really well to just extrapolate more mobile subscription more ads just more of everything and you would have gotten really close to what happened. But if you would have shifted that 10 years earlier, 2005 to 2015, you would have extrapolated PCs and internet. You would have missed the smartphone. You would have missed everything. We missed Uber. We would have missed the entire world. So the question is which era are we in? Are we in the comaters of extrapolation or in the macro change? I think we're in one of these where everything changes. So I can't tell you what's going to change, but I'm pretty sure that things are going to change. But the way we think of it as a company is the scariest periods are the periods of change. Guess when Spotify grew the most?

变革时期。

Periods of change.

变革时期。当事情稳定时，市场份额保持稳定。你不会蚕食市场份额。所以当有变革时，就有风险。你可能会失去市场份额。但这也是你拥有最大机会蚕食市场份额的时候，如果你采纳这种变革的话。所以我的原则就是永远做第一。做第一，并且第一个采纳它。就像世界将会改变。接受这一点，并走在曲线前面。所以我认为这是我们做新事情的大好时机。但这也是我一直在思考的事情。它到底意味着什么，什么才是正确的赌注？正确的赌注是押注用户控制等等这些，你知道，而其他人则会押注与我们完全相反的方向。所以让我夜不能寐的是，我们仍处于这场变革之中。它还没有结束。没有人确切知道它会走向何方。这个时代将会出现 Uber 和 Airbnb。也许它们就是 OpenAI、Anthropic。也许不是。也许它还没有发生。

Periods of change. When things are stable, market share stays stable. You don't eat market share. So when there's change, there's risk. You can lose market share. But that is also when you have the most opportunity to eat market share if you adopt the change. So my principle is just always be first. Be first and adopt it first. Like the world is going to change. Just accept that and get ahead of the curve. So I think this is a time of great opportunity for us to do new things. But it is also the thing I'm thinking about constantly. What does it actually mean and what are the right bets? Are the right bets to bet on user control and these things are, you know, whereas others are going to bet the complete opposite of us. So that's what keeps me up at night: we're still in this change. It hasn't finished yet. No one knows exactly where it's going to end up. There are going to be the Ubers and Airbnbs of this era. And maybe they are OpenAI, Anthropic. Maybe not. Maybe it hasn't even happened yet.

这是一个完美的结束点。Gustaf，谢谢你接受采访，伙计。和你聊天总是很愉快。

That's a perfect place to end. Gustaf, thanks for doing this, man. Always good to talk to you.

谢谢你，伙计。

Thank you, man.

感激不尽，伙计。

Appreciate it, man.

希望你喜欢这一集。请记得在你收听的地方订阅并留下评论。另外，请务必收听我的另一个播客《Founders》。近十年来，我痴迷地阅读了 400 多本历史上最伟大企业家的传记，寻找你可以在工作中使用的想法。你在这个节目中听到的大多数嘉宾最初都是通过《Founders》找到我的。

I hope you enjoyed this episode. Please remember to subscribe wherever you're listening and leave a review. And make sure you listen to my other podcast, Founders. For almost a decade, I've obsessively read over 400 biographies of history's greatest entrepreneurs, searching for ideas that you can use in your work. Most of the guests you hear on this show first found me through Founders.