From Tech Investor to Author (with a16z's Andrew Chen) – Harvard Business Review

Azeem Azhar speaks to Andreessen Horowitz general partner Andrew Chen about their parallel career moves – from tech insiders to authors.
Andreessen Horowitz general partner Andrew Chen and host Azeem Azhar have something in common: they were both tech company insiders and early-stage investors before becoming authors. They explore why they decided to write their books, how writing intersects with their day jobs, and whether they’d do it all again.
They also discuss:
Further resources:
Mary Meeker’s Internet Trends 1995-2019
Status as a Service (Eugene Wei, 2019)
The Cold Start Problem (Andrew Chen, 2021)
The Exponential Age (Azeem Azhar, 2021)
AZEEM AZHAR: Welcome to the Exponential View Podcast where multidisciplinary conversations about the near future happen every week. I’m your host Azeem Azhar. Now, my guest today is Andrew Chen, a partner at Veg Capital Firm, Andreessen Horowitz, specializing in consumer internet startups. We do talk about that in this podcast, but we also talk about something else. Like me, Andrew released his first book this year and had to get to grips with transforming years’ worth of experience as an insider in the tech industry into a coherent thesis and a long form book. Both he and I have written in blogs, in tweets for some time, but a book was a different beast for the two of us. It was fascinating to hear about his experience of writing a book, how his experience of startups translated to the world of publishing and what he learned about his day job while moonlighting as an author. It’s a little different, but I do hope you enjoy it. Andrew Chen, welcome to Exponential View.
ANDREW CHEN: Hi, Azeem. It’s really nice to be here.
AZEEM AZHAR: You’ve got a new book out, which is something that I can empathize with because my first book was published back in September. And I’m so curious about, of all the hats that you have worn, why you felt you wanted to add book author to the other ones.
ANDREW CHEN: There has been something really magical and special for me to put a lot of content out online. I’ve been on Twitter for over 10 years now. I’ve been writing on my blog and nearly a thousand essays, and I just wanted something that was new and creatively challenging and thought that maybe it would cause lots of new, wonderful things to happen. I actually met originally Marc Andreessen and Ben Horowitz as a result of writing. So putting something out there in the world is amazing. Now, all of that said, if you ask me from a business ROI perspective, whatever, whether or not writing for three years on a book is better than continuing to put stuff out every week online. I can’t actually recommend it to other people. I can’t actually so far to say like, “Oh, it’s obviously a slam dunk.”
AZEEM AZHAR: It’s always hard to know what the return investment of a book is for every hour that you put into it. I don’t dare divide the direct commercial return by the hours spent because I might have to report myself to the payroll authorities, or the employment agency, or something for being below the minimum wage. That’s the nature of it, but there is something different, isn’t there? To the process of thinking when you’re writing a book compared to writing a blog post, let alone a tweet.
ANDREW CHEN: Yeah. Well, I think first that there are just some ideas that just aren’t just tweet ideas. That’s just true. Tweetable ideas are great when they fit the zeitgeist, they’re very opinionated, they’re very simple ideas. Those are the things that spread. We’ve obviously created this whole social media set of dynamics and loops and everything that are really about these kinds of opinions. And then allowing people to memetically reproduce these kinds ideas over and over and over again. And I think a lot of what I was trying to get across in the book is actually, it’s not as simple as, and we’ll get into the content, but it’s not as simple as like you have this idea, network effects. If you have it, it’s the silver bullet, it’s the best thing since. I wasn’t trying to make that argument. I was trying to make the argument actually, “Well, you have this thing, it’s really hard to get. Once you have it takes a lot of time to keep it once. And then once you are really big, you might lose it at any moments, a startup could emerge and eat your lunch. And have this much more nuanced discussion, which I think is much more of like a book length type examination as opposed to a whole bunch of a whole bunch of tweet.
AZEEM AZHAR: And we’ll definitely get into the nuance of the argument. I was curious though, of this idea of being, when you’re a product person, you go through a process of iteration, right? Your first version is a little bit shaky, it’s a skeleton. And through the process of learning, you make a more and more robust product. I think of a famous chart that Spotify had from maybe a decade ago, which was about how you build a working car and you start with a skateboard and then you slowly enrich it over further iterations. And I’m curious about whether for authors today, the Twitter sphere, then the blogosphere provide that. You float an idea through a single tweet that has to be very generic and high level. Because it’s only… Back in our day, it was 140 characters, right? And then you get feedback from experts and participants. And that allows you to strengthen the thesis, and that allows you to then come up to something that could stand the test of a book length chapter. Is that too much of a simplification?
ANDREW CHEN: I think that it is right that as I was writing the book, what I found was the very, very first thing that I did is an outline. And what is an outline? Well, it’s actually a rapid prototype of the thing that you’re trying to create to see whether or not it all holds together. And what is an outline ultimately made of? What’s see to do in that case is just to have a whole collection of topics. And you’re just writing a bunch of topics and like, who knows if it’s going to be interesting or not. I was joking with the friend describing it as a tweet fractal. I was like, “Okay, what I’m going to do is I’m going to write a whole series of tweets and opinions basically for each section. Each section will be made out of opinions and things that I would naturally want to tweet.” Like a case study with a really interesting hook to it, a conclusion that’s maybe counterintuitive and surprising. And what I did was, in the earliest days of the book, I took about 100 early interviews that I had, and I condensed it into a 30-page outline of all these things that felt like they could be tweets. And it made it much more interesting, I think, to then write, because I then had the foundational building blocks of many opinions to argue and build off of, as opposed to the books that are just… they just explain something. They don’t have an argument at the core of it. And those are just way more boring to read, and so I wanted something where at every moment at every corner there was something I was trying to say that went up above and beyond just unpacking a concept. I didn’t want people to feel like they were reading an encyclopedia.
AZEEM AZHAR: Well, it’s interesting, and in the book, of course you have a lot of case studies, but you are often on the inside of those case studies, right? So you’re not like a researcher looking from the outside and digging through the archives, you’re going through your own personal email in some cases, right?
AZEEM AZHAR: You are bringing those stories back out to life, but trying to construct them as an argument that stands up as a book, which I think is a different process. It’s a different process both because you’ve been a participant and because you’ve worked in product for a while, right?
AZEEM AZHAR: I found something similar as well that I’d be balancing up email exchanges that I would’ve had or personal discussions that I’ve had with the people or building the technologies with the theory and the academia that sits behind that. But it constructs this new format, are you really an author or are you an operator or are you an observer or are you somehow something that’s in between? The thing that was fascinating was just for the first time in many years, doing something completely afresh. I didn’t have the methodology, I didn’t have something I could turn to. And I was trying to bring what I knew about taking products out to this process of writing a book. And of course I’m working with experienced editors and the book world doesn’t look at the minimum viable product. As a concept it has, it’s all about draft, and it’s about the first draft, the second draft and the third draft. And for me, that would be six weeks work. That would be iteration one iteration, two iteration, three out of maybe 50 iterations in a two-year period. So there’s a different mental model that you have to grasp. So I actually, I think I found it quite challenging the first time round, but I could imagine for the next book and hopefully the publishing industry will-
ANDREW CHEN: The next book-
AZEEM AZHAR: … will grace me, [crosstalk 00:08:29] there’ll be a chance for a second book. I think I would be able to bring these two disciplines together in a way that will be less hard work for me. You talk about a kind of a fractal pattern, you talk about is distilling a thumbnail to 30 pages. Which feels to me like you are bringing those disciplines that you’ve learned as a product person from Silicon Valley to the book writing process. I’m not sure I’ve necessarily got that far with my book, but I can see that there is something that I’ve learned.
ANDREW CHEN: And one thing I actually would love to experiment with in whatever is the next big kind of creative project I do around these lines, and it probably won’t be a book. The one that I actually want to try is, well, how do business people actually want to consume ideas? They actually prefer slide decks. That has become the media that we’ve all standardized on for whatever reason. Good or bad that is it, right?
ANDREW CHEN: I look at pitch decks all day. And so I’ve also thought about the idea of taking, instead of writing 400 pages in a book, what happens if you actually write 1,000-slide deck of an idea? And that’s actually what you do when you publish the slides, you publish notes for the slides, you publish a narration. And maybe that’s actually what people want to listen to and how business people want to consume it. I haven’t seen it done just like a mega… 400 pages of written content probably translates to certainly several thousand slides very likely. And so that would be a way to do it.
AZEEM AZHAR: That’s interesting. Could that be done without you narrating the slides? I think about great totemic internet era slide decks like Mary Meeker’s internet reports.
AZEEM AZHAR: It’s been going for 25 years and it’s now up to several 100 pages, but I wonder the extent to which one can actually get the heart of the ideas without the analyst over the top to ask and answer the questions.
ANDREW CHEN: That is one of the things that is great about slides, is that if you want to, you can listen to it at 2X, and that’s an option. If you want to just flip through it and skip through all the whatever, and you just want to say, “Oh, well, okay, I get it, get it, get it.” You skip to the section that you want, you can do that. So I think there’s a lot of control and a lot of flexibility for the viewer that they have that they don’t get in the format of a book. And so I really think it could be a viable new way to communicate these ideas. And the Mary Meeker deck, which is obviously a really fantastic deck. The one I would have of it is, it is more a collection of just graphs, I think, than it is like trying to make some kind of original point like you should think about your product in XY. It’s not meant to make that type of argument, but I do think that you could.
AZEEM AZHAR: Yeah, let me give you the counterpoint to that, which is, I’m sure you remember the famous essay, The Cognitive Style of PowerPoint from-
AZEEM AZHAR: … a couple of decades ago, and some of the critiques about how PowerPoint or slide decks will tend to frame our thinking because of the limitations of the sort of single slide. And I think the famous anecdote or apocryphal story is the NASA slide deck around risks in the space shuttle where ideas couldn’t fit in. So I guess the thing that I felt, the freedom that I felt as someone who has had to spend too much of my life creating slide decks within a book was that I wasn’t constrained by, here are the 30 different ways that we represent ideas. Whether it’s cycles or value chains or interrelationships or pillars. And this is the limited space you have, because it all has to be a minimum of 12 points and fits on a slide that you could actually dispersively explore and unpack an idea over a few thousand words. It’s not without its limitations, but it certainly broke me free of the constraints of a slide deck.
ANDREW CHEN: Yes. I think that’s right. I think the other big part, just to add to all this is that, the modern book publishing industry I just find to be amazing is that the number of books that you have to sell to be a bestseller? No one reads anymore. These are very small numbers, and in the US, I think it’s supposed to be like 10,000 hard covers or something like that. Which is just amazing versus, Azeem, how many people are on your email newsletter? How many people are on your…
ANDREW CHEN: If you’re just looking at in terms of the reach of the ideas, the online format is superior in many ways. Also because it’s so shareable. And then in the book format, obviously you have the depth. But I also worry in many cases that we don’t have analytics on book. I guess actually Amazon does because Kindle.
AZEEM AZHAR: [inaudible 00:13:01].
ANDREW CHEN: But I have to imagine that it is just an insane decay curve where it’s like 90% of the readers will read chapter one. And then it’s like, by the time you’re on chapter three, it’s like 30% or something. And only like 2% of your readers make it all the way to the end. It’s probably something insane like that if I were to guess. And so as a result, it’s sort of like, well, really you put all this blood, sweat, and tears to make this thing that is long and readable the whole way through. But in reality, people just want to consume the first 50 pages and you have to get your idea out in a very short period of time.
AZEEM AZHAR: But it’s one of the few parts of, I guess, our lives where one produces something of that quality with the understanding that the back half, which you give as much love to as a front half is going to reach far fewer people, right?
AZEEM AZHAR: It breaks that idea that drives so many consumer internet companies, which is we focus on the things that have the most impact, the largest number of people that are the cheapest for us to do. That is the mental model of internet product manager and a book takes the reverse of that. I’m curious about something, which is a lot of your work. And just for listeners, Andrew is widely regarded, not just for his work on network effects, which is the topic of the book, The Cold Start Problem, but also his work on consumer internet companies, and most recently on new media formats. He’s the investor behind backing the teams at Clubhouse, which is an audio platform, and Substack, which is a newsletter platform. So you are somebody who’s also thinking about what do new forms of media and publishing look like? I’m curious about what your interaction with this oldest of publishing industries, a book publishing industry looked like. Did you have to check day job Andrew in at the cloakroom before you talk to your publisher?
ANDREW CHEN: Well, everyone I have interacted with in the publishing industry has been incredibly nice and really great at what they do. Most interesting about how the industry has evolved, I think is, in the same way as music and film and video games actually. We talk about that a little bit. As everyone knows, we used to be in an era where the publisher had so much power in the ecosystem. What they did was they had all the retailer relationships, all the little mom and pop bookstores, all the large retailers like Barnes & Noble. They had the ability to connect you with all the tastemakers, so this is the mainstream media, this is the economist, this is all those. They would write the check to the writer so that they, with an advance could support themselves. So they were kind of the financier in everything as well. And then they also just had literally the tastemaking of just having these editors that were really, really good at picking people and picking ideas. And that formula that I just described, we could apply that to music, we could apply that to video games, we could apply that to all those.
ANDREW CHEN: 20 years ago, you had kind of these monoliths, these oligopolies that lived in every of those media formats that did those. Now, I think what’s been interesting with the internet is like a lot of that has been inverted. It’s like, well, distribution is, you can self-publish, and you have Amazon and you have all that done. The power of mainstream media is weaker and weaker and weaker and my entire book campaign was very much just based on Twitter, in my newsletter, ultimately. And getting people to take pictures of the book-
ANDREW CHEN: … on social media, which hopefully some folks saw. Similarly, the tastemaking is still important, but simultaneously, you can find writers where they have already put out a lot to work on social media and you can just say, “Well, I love their work. I want them to just continue what they’ve been doing in their newsletter or their blog or whatever, but just a big book format.” And then the final thing that I think a lot of these folks of the ecosystem have is they have this financing arm, which still remains very, very important for people who are doing this as a real job. So I think this entire disruption in the digital world has really changed this whole system. And I think the dynamics are still very fresh. It’s still unclear how everything is going to evolve over time. What I spend my time doing in venture capital is I’m often betting on the new format over the old format. And part of the reason why I bet on the new format, there’s a great essay from Eugene Wei That’s called the status as a service, about social networks and the nature of delivering status to people. And one of the things he talks about is this concept of old money, which is, when you have an old social network, the structure of the network becomes ossified because a lot of the users that have been around for a long time, they have a lot of followers. Well, the problem with that is they will tweet things or publish videos or otherwise publish content and the whole system reinforces because they already have the followers. That the rich get richer, right?
ANDREW CHEN: And so what ends up happening is you get this old rich thing, where then the only way for a new creator to break in is not to go to the old platform. They want to go to the new platform because they need to find something new. And so I think zooming out, looking at now traditional and the first couple generations of social products and so on. I think that that’s one of the things that you have to figure out for yourself as a creator. And maybe you and I can be small sea creators in this context is that I think we often have to go figure out the next new thing. Because a book world has all of these dynamics that are very, very old, that we may or may not be able to break into.
AZEEM AZHAR: That idea of old money or old status money is such an interesting one, because that of course is part of the structure of publishing or indeed the traditional art world when you look at it from the outside.
ANDREW CHEN: That’s right.
AZEEM AZHAR: How is it that these people get all of the reviews or all the gallery space and so on. And it’s all to do with this sort of hidden structure of networks that reside a bit behind them. What I guess happened with the original social platforms is that you did see that rich get richer. That is part of the challenge of the network effect and the network scientists have modeled that. The people who first got to 100,000 followers on Twitter are most likely to be getting to a million and that. So it becomes hard to break into the project. I suppose, that begs a question, which is something that you must think about, which is, to what extent is that actually a kind of structural product problem? Or do we just wait for the next platform to come along and democratize the talent?
ANDREW CHEN: Well, so in the regional conception of the network effect, which has been described for many years, including, I have an excerpt from, from the AT&T or back then it was called the American Telephone and Telegraph Company. The idea that the telephone is useless on its own, its value really depends on its connections that it provides. And so the network effect, we keep referencing it, and so maybe just a level set is really about these products, where the more users that use the product, the more valuable they get. But simultaneously two things happen on the edges, so that’s a great idea on its own. But what happens is if that’s true, if you believe that, that also means that when your product has very few users, then it useful at all. And so there’s a phase that your product goes through before it becomes useful, before there’s enough users that you can call it useful. And then that’s the cold start problem, that’s kind of that phase. Then you get to a point where it is actually really useful. And that’s kind of a combination of phases that I referenced in the book, which is the tipping point and the escape velocity. And then there’s a very interesting thing, I think this is what we’re talking about now, which is every single one of these networks eventually faces some set of overcrowding and saturation and all these things. We all experience this when you use Slack and there’s just like too many messages, or you start follow like too many people on Twitter. I think I follow like 10,000 people on Twitter. And so at some point you need to throw in all the algorithms to help with discoverability, but to the point that I think you and I are making, you get to a point still where new entrants find it hard to come in. And so I think that is the challenge, which is that every single one of these networks does get to this S-curve where it has all these problems. And the problems are so difficult to entrants. It’s the one part of the book that I actually don’t have all these amazing solutions for. I’m basically like, “These are ongoing. And if you’re one of the major social networks, yeah, you’re going to continually fight trolls, fraudsters, regulatory entities, whatever. And you’re just going to fight them forever basically.” But what that allows new startups to do is to cherry-pick some little tiny part of these huge networks and to start running with it. I think as an optimist when it comes to technology, I think that’s what you have to believe and support in order to get the next generation going. And the thing that is hard, I think is when these platforms are so powerful like the Google, Apple platforms, networks are so powerful that it becomes very, very hard to see how a new entrant would enter. In which case that’s why you have to be so excited about Web3, because it’s truly the thing that looks like it is most likely to be the next big entrants into that.
AZEEM AZHAR: In a sense, competing with a media-based network, whether it’s a traditional centralized publisher with its taste making or whether it is something, whether tastemaking is more driven by algorithms. Or perhaps by lead users, which is the case on things like Instagram and Twitter. It’s easier to break in on the media side because we already have the consumption devices. Whereas if you’re trying to break in on the operating system side or the device side, you have to do so much more. I have a newsletter that started with my emailing to 21 people, and we now have like a couple of hundred thousand people five years later who receive it across different channels. But I could rely on the fact that they all had access to an email device and I didn’t have to make any assumptions. But actually with a deeper product like an operating system that has a hardware connection, actually to get people to switch also requires you to bring developers on board. Because I’m not going to move to a new phone platform that has no apps. And that raises a cost of entry, it raises a cost of competition.
ANDREW CHEN: That’s right. And I think one way to view it is, the cold-start problem, the cost of traversing this phase of your product where you don’t have enough users, you don’t have enough content, you don’t have enough any of these things. The cost of that is really the competitive moat of the incumbents. And so I worked at Uber during some of the hyper growth years. And you could just see that in order to launch a new city, you had to do a series of things in order to solve the cold-start problem. You had to go to a Craigslist. You had to post a bunch of listings and say, “Hey, I’ll pay you 30 bucks an hour if you want to drive. And as long as you sit on the platform, we’ll pay you 30 bucks an hour.” And you needed to get at least a couple dozen of these drivers onto the platform. Then you had to go to the of train stations, you’d go on Facebook, you’d do PR. You’d have to do all of these other things in order to get probably hundreds, thousands of riders excited about using the platform. And then once they started to transact with each other in a network, there was a moment where it would catch on its own and off you’d go. But that initial start-up cost was certainly hundreds of thousands of dollars, and maybe in some cases for the bigger cities, millions of dollars. The fact that you can measure that in dollars tells you something that’s really interesting. Now, here’s the question, which is once you get into a mode where for Uber, if Uber and Lyft or others are already in the city, and consumers are already comparing you against them, then the cost of start-up is even higher. Because why would you use this new service otherwise? And so what that tells you is, you have to actually buy your way, not into just a functional network, but one that is at parody with the incumbents. And so what that tells you is that that cost again is your moat getting more expensive and as a result, protecting the incumbents. And I think you’d say the same thing about the app stores, because you’d say, “Well, in order to build a viable third App store, you’d need to recreate some percentage of the top apps for Android and iOS.” And how much does that cost? Billions of dollars. It’s just an insurmountable moat at this point. And so you have to do something as Web3 has been doing or what the other viable things might be entering in through a side door like all the Metaverse stuff. And coming in through games or something like that where you’re building a subset of applications that people spend a lot of time on. And then maybe you can grow it from there. So we’ll see how it plays out.
AZEEM AZHAR: What you’ve encapsulated there is so much of the classic challenge of getting the network effects running, and we call it the empty disco effect. You have to just get enough people to dance to make it feel like there’s a party going on. But the thing about the disco or the nightclub is that it can still evaporate. The action can dissipate as quickly as it gets started. It’s almost like this chaotic process. With unclear what starts it and it’s unclear what causes it to fizzle out. I’m just curious about that question, which is, once you get this internal heat running, thinking back say, to Uber’s early days, to what extent does it really run itself? Or to what extent are you doing hard thinking and operational tweaks every single day to maintain it? And what do you measure? What is the measurement of the health of that network and that network effect?
ANDREW CHEN: Well, so first I completely groom you about the disco-bar effect that you were talking about. And I think that’s very easily encapsulated by the fact that if you’ve been using tech products for the last 20 years, how many different social apps and messaging apps did you start with, get excited about, used for a couple years, and then quickly abandon and go to the next one? Because I remember for me growing up that it was all about AOL Instant Messenger. I would talk to all my friends in AOL’s Messenger. And then a couple years later, everyone migrated to Myspace or Zango or LiveJournal. And then after that, it was, Facebook was so cool and you’d use Facebook for everything. And then after that, now we’re in Snapchat and Instagram and it’s going to continue and continue and continue. And so I think one of the maybe counterintuitive non-consensus opinions from my theory is that these companies and these products that we view as invincible empirically are just not. They seem to cycle every five years or so. Even the apps that we’ve been talking about are slowing down in terms of their innovation and the amount of consumer excitement. I think that’s absolutely a real thing. But to the other point, I think the way that you have to end up addressing these, once you have solved the cold-start problem is, every single one of these companies that has network effects. If it were so easy, they would just be able to have the products just run on their own, but that’s not what you actually see. What you see is that every single one of these tech companies we’ve talked about, whether it’s Dropbox and Zoom, I have a bunch of B2B examples in the book and Slack and so on. As well as marketplace companies, as well as social companies. So the network effects concepts covers those as the major category. Every single one of those companies ends up hiring thousands of very expensive employees immediately to spend all their time tweaking these networks and trying to get them to grow. And the reason why they do that is because ultimately, these networks are actually made up of smaller networks. So if you have a marketplace company, it’s actually a whole set of smaller networks in terms of verticals. You have collectibles, you have media, you have cars, you have this, you have that. If you are a B2B company like Slack, you’re a series of networks that are team-by-team or company-by-company. And you have to basically constantly innovate to go after new networks and to hold onto existing networks. Again, I think this is a good example where the network effect itself is like crazy silver bullet thing. It’s really something that you have to refine and amplify over time because it just gets harder and harder to grow these products as they get bigger.
AZEEM AZHAR: I think one of the things that I’m curious about though is, how do you tell whether to make a bet? And so, I’m just going to think about three companies that I have had various amount of excitement with that are in this field, and we’ll look at media ones.A few years ago there was a company called Meerkat, which did a sort of instant video. People got really excited about it, I was at least one of them. And Meerkat raised quite a bit of money and then died by about 2016. Then you have Clubhouse, which is one of the businesses you are invested in, which was absolutely the bell of the ball for very many months. And now’s got much, much quieter, but raised money according to the press at a $4 billion valuation. And then you have a third, which I was involved in as one of the earliest users was Substack, in the newsletter business, which again I think you’re involved in and couple of fantastic founders there. Again, a business that the market I think viewed as being worth 6, $700 million just a few months ago. And I guess the question is, how do you tell? What is the assessment in those early, early days around what the potential is and whether this is going to have a network effect that is not just going to grow, but establish that moat? There’s something going on, you’re making rational decisions based on uncertain information. What is that?
ANDREW CHEN: One of the big differences from when I was working at a normal company versus being an investor is that when you’re at a place like Uber, you want the hit rate of your projects to be very high. They should be 80 or 90%, because almost always, you’re taking something that is already working and you are making it incrementally better. When you’re looking at start-ups though, you have a very different question. So instead of asking, “Okay, is this an 80, 90% thing? What could go wrong?” Instead, you spent all of your time thinking, “What if this works? Can it be big enough? Can it transform an entire industry?” What is it about it that makes you [inaudible 00:31:21] So I think that’s one piece. There’s a lot of other evaluation criteria in terms of the founders and how amazing are they? We could spend a lot of time on that. But on the idea itself, which is I think what you’re trying to get to. The biggest things I tend to think about are, first, we invested in a company like Clubhouse and Substack when they are basically two or three people. They were very early. I was user 104 on Clubhouse. When we invested, I had 500, so 5-0-0 daily active user, very, very small. Substack, very tiny numbers as well. And I think one of the ways that you go about it is, if you believe in this concept that in order to solve the cold-start problem, you have to start by building a very small network. You have to start and build this tiny little network and prove that you can get it to work there. And by the way, you need to have a number in mind for like how big this network needs to be. So Zoom is like a two or three-person network, and then it’s valuable. Slack is like a five or 10-person team-based network. Something like Tinder or Uber, Airbnb needs to be hundreds of people. And I refer to that as an atomic network. And so basically, if you can show that you can build one atom of this larger network, and then you can build a second atomic network and the third atomic network. You could probably build 10, you could probably build 50, you can probably build 100. And so the question comes down to, can you prove in the early days that even if your network only has 500 people, that’s it’s working for those 500 people. And then there’s an exercise of extrapolation to say, “Okay, well, if it works for this 500 people, could it work for every everyone else?” And so, one of the things that I would end up looking at is basically for this group, I would go into the retention curves. How many users that signed on in the first day are around 30 days later. In the Substack case, if you’ve subscribed, what percentage of those readers are still paying subscribers a month in, six months in? And for both of those two cases where I got directly involved, I was able to get myself over the hump by saying, “If this works in both cases, they could be huge. They could transform the media industry.” Clubhouse, because there’s really no easy way to start up a podcast. As you know, starting a podcast, pain in the ass. You have to deal with a huge stack of tools, there’s all these issues with it. And what if you could make it as easy as a tweet or as Instagram? There’s something very powerful to the idea Substack, big, because you could basically self-serve and be a writer and make millions of dollars potentially. And you don’t need to deal with all these media corporations and conglomerates and all these things that are in the middle of that media, all of this. So that’s very exciting as well. And so once I look all the metrics and I convince myself even with a small user base that it’s there, then I tend to think about, “Well, how easy is it for these teams to add new networks over time? And can I prove to myself that they’re doing that?” And so then I’ll look at metrics like top level growth and so on and so forth. And then finally, I think the fun thing with these things is, there’s a lot of luck. There’s a lot of luck involved, there’s a lot of randomness.
AZEEM AZHAR: So you make these judgment decisions based on the vision of these companies, your belief in whether it can be a huge market. What the early data of retention shows you, how strong the founders are and how well they fit that market. These investments happen at the end of 2020 and early 2021, and by the middle of 2021, everyone’s talking about Web3 as this new computing platform. And you yourself have referred to Web3 as something that addresses the challenges of traditional ways of building companies. So what for you is the essence of Web3 in this landscape? How would you encapsulate Web3 to somebody who’s perhaps not in the industry?
ANDREW CHEN: I think what we are really seeing is introducing the concept of ownership and with its scarcity and with it, the technological underpinning of tokens to power all of that. And so purposely, I think the thing that’s caught on is Web3 versus Web 2.0. Web 2.0 was about users being able to read and also write to the internet. Users being able to talk to each other and connect with each other. But in that world, there’s still very much advertising-based businesses, there’s still very much ones as a content creator, you’re not able to gain upside or as early user, you’re not able to gain upside in the product, and Web3 is changing that. And it’s changing that across many, many categories, not just obviously everything we’re seeing in NFTs, but there’s a huge wave of games that are getting started, a huge wave of consumer applications that have Web3 dated. And so I think ownership is really the key differentiation.
AZEEM AZHAR: The thing that strikes me about Web3, and I am like you, fascinated about it. I’m midway through a series of essays explaining why I think it’s interesting. Part of my core hypothesis about why it’s interesting is really an evidence proof rather than a theoretical proof. It is that there are lots of developers and smart developers, the growing number of developers working on it and experimenting, And there’s a growing number of users. And there’s an economic flow that makes it sustainable. And that we’re starting to see not only applications that actually work in the real world. I love this one called Helium network, which is a distributed Wi-Fi style network that is built from the ground up. But there’s also weird and crazy experimentation. There is exploration all over the place, NFTs being one example of that. So all of that to me is evidence of what does a healthy emerging ecosystem require to grow. It is a network effect in itself of between the users and the developers and the flow of funds. So I’m also pretty excited by the potential. But if I reverse out and put a skeptics hat on it, have we really seen the thing? There was a moment for me and first time I fired up Clubhouse and I’m user 2090, so I’m a real Johnny-come-lately compared to you-
ANDREW CHEN: That’s so early. That’s so early.
AZEEM AZHAR: … Andrew. There was a magical moment, when I first fired it up. And I’m curious about whether Web3 is delivering that relative to the excitement and the noise around it. And what you see as the hard kernel of the reality, because you’re someone who has to look beyond the hype.
ANDREW CHEN: Right, right. Well, I think that if you use the metaphor as the original internet, the very earliest years of crypto, we were building protocols, we were building the equivalent of HTTP or FTP. All of these protocols that were out there, some of them ended up being really valuable, some of them ended up being less widely used like Gopher or the newsgroup protocols and so on. And so I think we had many years of just that, and many years of people working on basically non-consumer-facing functionality as a result. And these use cases that were more store value and frankly speculation were the primary drivers for all the energy. I think we’ve now entered a period where between DeFi, between NFTs. And then now you have these games like Axie Infinity, you actually have applications that consumers are engaging with and that actually are real in the same way that one of my favorite analogies here is like Magic: The Gathering or the Pokémon card game is like that as well. These are games where you want to collect the cards, and you can spend all your time just collecting the cards if you’d like. And that has value and is fun on its own. And then you can also play other people, and you can play very casually or you can play in very serious tournaments. And that’s what has a lot of the engagement. So I think that’s the era that we’re in. We’re in this like, a lot of people are trading baseball cards right now. You can’t really do anything with baseball cards, but it’s cool. It’s like art, you can put it as your profile. There’s a social status thing. Now we’re building these games on top so that you can, again like Magic: The Gathering, you can actually play and engage people through that. And then I think we’re just at the beginning. Your question is a very fair one, but I think it’s very clear we’re very much at the beginning. And I think for all the skeptics, from my vantage point, you’re often betting on the first derivative on this or the second derivative. You’re betting on the fact that the inflow of people and ideas is just increasing over time and then their products that they’re building, the diversity and the creativity of it is increasing over time. And all of that is going to reach consumers over the next five to 10 years in an amazing way.
AZEEM AZHAR: So if we look out in that 10-year period, say look to 2030, how can you imagine our interactions with Web3? Will we even think about them as Web3? Or will they just be the internet?
ANDREW CHEN: I’m always surprised by what are the things that technical jargon that makes it all the way to the consumer world. Normal consumers know what URLs are. Normal consumers have seen HTML. They ended up being able to see some of the building blocks. And in that way, I’ve been surprised that the term NFT, for example, actually caught on and is now a thing. But I do think that when you actually go through the flow today of what it’s like to use one of these Web3 applications, it’s just too much for a new user. You got to go onto an exchange and set up your wire transfer, and then you need better mask, and then you need to think about gas prices, and then you need to go to the… It’s just too much, it’s just too much. And so I think that the most streamlined way, if you were to think about it in the future, is that all of this functionality I think, has to be built pretty deeply into the stack. I don’t know where it goes. Is the question, does it go into the browser? Does it go into the operating system? Is it a set of connections like Facebook sign in that’s just so simple? It’s like connecting your Coinbase Wallet is already really nice, but doing it in a way where you can maybe users come in and instead of opening a bank accounts, they just do it on one of these new neobanks that supports crypto and then everything just works magically. But I think we have to go there in order for this to be a widespread product, but I’m super optimistic that we actually will get there.
AZEEM AZHAR: We have had such a wide ranging conversation. We started with this process of going from being a blogger and an operator and an investor, and now an author. Which is something that I recognize with explored your thesis around network effects. And of course, as it is a consumer internet in 2021, we’ve ended up in Web3 and NFTs where I guess all conversations have to end. But I think there’s an important point before we let you go on your day, Andrew, which is for us to understand how best to follow your ideas. And just remind us of the title of your book and where we can find you on the internet.
ANDREW CHEN: The Cold Start Problem is now available for general release. You can pick it up at your favorite retailer. If you want details on the book and the table of contents and other kinds of information, I’ve set up a website, for that. And I am returning to regular writing in 2022, and so I will be publishing new essays, new things at and my Twitter account is also @andrewchen. Thank you so much for having me, Azeem.
AZEEM AZHAR: No, it’s been our pleasure and have a happy 2022. Now, I hope you enjoyed my discussion with Andrew. I had a lot of fun. If you want to learn more about network effects, the secret source behind some of Silicon Valley’s biggest success stories. Listen to my podcast with James Currier, a general partner at NFX, a venture capital firm that specializes in leveraging those effects. Now, we’ve got two more episodes of Exponential View coming out over the next two weeks. I’ll be recapping the biggest trends of 2021 and giving my take on how the world will change in 2022 and beyond, so watch this space. To become a premium subscriber of my newsletter, go to, where you’ll get a 20% off discount. And to stay in touch, you can follow me on Twitter, it’s @azeem, A-Z-E-E-M or A-Z-E-E-M. This podcast was produced by Mischa Frankl-Duval, Fred Casella and Marija Gavrilov. Bojan Sabioncello is our sound editor.