Kickstarting Network Effects

Doss: The content here is for informational purposes only, should not be taken as legal, business, tax, or investment advice, or be used to evaluate any investment or security, and is not directed at any investors or potential investors in any A16Z fund. For more details, please see A16Z.com/disclosures.

Doss: Hi, and welcome to the A16Z podcast. I'm Doss, and today's new episode is all about how to kickstart network effects. It features two entrepreneurs who have built or are building companies with network effects. Alexis Ohanian, co-founder of Reddit and now an investor, and Paul Davison, co-founder of Clubhouse, as well as A16Z general partner Andrew Chen, whose new book, *The Cold Start Problem, How to Start and Scale Network Effects*, comes out just this week. You can find more on that at coldstart.com.

Doss: The conversation that follows is one I recently moderated live on Clubhouse. We initially focus on how social media platforms can first win over creators before moving to discuss how Web3 is changing the incentives and dynamics around network effects.

Doss: Now, at A16Z, we've covered network effects a lot in our content over the past decade. Much of that you can find at A16Z.com/network-effects, including an exclusive excerpt from Andrew's book, which is also on future.com. But we begin this conversation with Andrew answering the most basic question. What exactly are network effects?

Andrew Chen: So if you go back to the original definition, it is the idea that a product becomes more useful as more users are on it. And so that could be anything from a messaging app where if your friends are on it, the more of your friends that are on it, the more useful it is. But also, it's important in a workplace and B2B context where if your colleagues are all using Dropbox, then you're going to use Dropbox too because all the files are in there and because it's more useful.

Andrew Chen: And so the funny thing with this is it's the thing that makes many of tech's largest companies, from Airbnb and Pinterest and Uber and Zoom, so powerful. And simultaneously, it is a huge pain in the ass for startups because it means that when you first get started, if your favorite people and your friends and your colleagues aren't using your products, then you face the cold start problem, which is nobody wants to use these products until everyone else is on them.

Doss: So I guess I want to get Alexis and Paul into this. How did you, the two of you with Reddit and Clubhouse, approach that? And specifically, kind of how did you get your first hundred users? What did it take?

Alexis Ohanian: Oh man, this is dating me because this was 2005. And so Twitter didn't exist. I mean, really the only other social media site that I knew of in 2005 in June, when we launched Reddit, was like Facebook. And so the way that I jumped it was I ran a PHP BB forum in college for about two years. I went to this community and I said, look, y'all, you know, for the last two years, we've been doing something really fun on this forum, maybe 750 members at that point. And I said, I'm working on something new. It's called Reddit. Like, come take a look. And just maybe 20 of them showed up. And it was because I had spent all this time in college building this forum that there was at least a sort of foundation of community.

Alexis Ohanian: And then the co-founder and I just faked a bunch of other users. Steve just built a very simple way for us to create a new account in like two seconds and start posting. And so I, as the non-technical sort of product manager designer, just spent a lot of time creating content under pseudonyms. That worked for a few months, and eventually enough other normal people started showing up, like real people, that we had solved it. The sort of best practices in 2005 are laughable compared to 2021, where you have no excuse not to at least get 100 people to show up and kick on a test flight.

Doss: That's amazing. And I want to come back to this idea of faking users because I know it came up in Andrew's book and he called it Flintstoning. But Paul, can we maybe hear from you first on how did you approach getting your first 100 users?

Paul Davison: Yeah. So in the earliest days of Clubhouse, the whole app was a single room. There was no audience. Everyone was a speaker. There was actually no follow graph. Everyone followed everyone. And we worked really hard to get critical mass and density within that one room. So we tried to find a bunch of friends who we thought were interesting who sort of were in a similar peer group and we made sure that every time they came on someone was in the room to greet them.

Paul Davison: So Rohan and I had a Slack message set up so every time a new person would join, Rohan and I would be on the phone and we'd hear and I would say I gotta go, I gotta go, someone's joining, and I'd jump in and say hey, how's it going, welcome to Clubhouse, I'm Paul, how'd you hear about us? And then half the time they would just leave because they didn't know who I was, but the other half of the time they would stay and someone else would join and someone else would join.

Paul Davison: I remember taking one friend through user onboarding and he got to the end and he saw all the people who he was following and he said, wait, I don't remember you asking for contact permission. How did you know who all my friends are? And I said, no, no, those are just all the people on the app. We tried to get a good initial community going. And then I think this is a pretty common pattern with a lot of online communities where oftentimes with communities, you start with a single community. And as it grows, there's this natural mitosis where smaller communities form, sub-communities form. It's kind of interesting.

Paul Davison: Like if you think about platform models and how a lot of networks grow, like with Facebook, they built Facebook.com and they got millions of people to come in and use the platform so they could talk with each other. And then they launched their developer platform so people could build apps to reach that audience or Apple built the iPhone and then they built the app store so people could reach that community. It often naturally scales out like that. So at Clubhouse, our approach was to build this small rooms product and to help people come together in a single small room and talk and grow a really great community that way. And that is what has led to the growth of the creator ecosystem within Clubhouse.

Doss: So the sequencing matters a lot. Andrew, does that kind of match up to the model that you have for how network effects unfold in the early days?

Andrew Chen: Yeah, that's right. And I think there's kind of two ways to think about it. So one is that there is a recurring pattern that has happened in the tech industry where many of the biggest tech products that we see actually start out in really small places. They start out in college campuses. They start out in individual teams inside of larger companies. And when you look at that recurring pattern, what you see is that it doesn't matter how many broad kind of users that you have. What really matters is, okay, do you have a little tiny network where there's enough people that it's stable and everyone wants to communicate with each other or share files with each other or buy and sell stuff with each other?

Andrew Chen: And what happens is almost every product, you can start to think about it like, okay, what is the atomic network of this product? And so it turns out that, for example, when I talked to the Zoom and Slack founders and CEOs, they only needed two or three people before they could see that there was a recurring pattern that people would use the product versus something like Airbnb, where when you talk to that team, they would say oh, we actually need something like 300 listings in a city before that city becomes viable. And same thing as what we saw at Uber: you needed to have at least a couple dozen drivers at any given time such that you could get a pickup time under 15 minutes on a consistent basis before a city would become viable.

Andrew Chen: And so that becomes kind of the core articulation of the Cold Start problem, which is like, okay, well, instead of trying to get lots and lots and lots of users all at once, and it might be completely untargeted, it just works so much better to do it one at a time. And in Alexis's case, and for Reddit, you really see that in the concept of subreddits. They have this very natural way of expanding from the geeky early adopters that Reddit were all the way to sports and politics and all the hundreds of thousands of subreddits that exist today.

Doss: That's such a great segue. I actually want to read something here. It was a quote from *Cold Start Problem*, Andrew, that I thought was really fascinating. And it said, for networked products, the curation of the network, who's on it, why they're there, and how they interact with each other is as important as its product design. And so I think this is kind of an interesting point. How do you get the right network? How do you think about different types of users and creators? And then how do you roll out the right features as the network starts to grow. It's kind of like being a party host. You spend a lot of your time, you know, the kind of stuff Paul's talking about, just making people feel welcome, sort of role modeling the behavior you want to see.

Andrew Chen: When I coach founders now, it's about helping them understand their job is to get to a minimum viable community. And you role model behavior in the same way that if you walk into my house for a party and you see me, as the host, doing a keg stand, you immediately have a sense of how you should behave. Whereas if you walk into my house for a party and I'm wearing nice clothes and I'm sipping on some wine and having some polite conversation, you instantly have some guidance for how you should behave. It's the exact same thing, even when you step into an online forum or an online community.

Andrew Chen: And this is more relevant than ever in 2021. Minimum viable community is going to be more important than minimum viable product because with everything happening around no code and just all of the kind of aptitude around shipping software now, building that V1, a product that doesn't suck, is actually like not that hard. Building a minimum viable community of people who really care, and you're seeing this now all over with Web3, where so many of these, you know, PFP projects, the really well-executed ones, they're a case study in minimum viable community because there's, in some cases, literally nothing else other than maybe some JPEGs.

Andrew Chen: And I look at that and I get really excited because I think this is such a human act to build community. And it's something that's going to be so important now, not just the act of doing it, but even the metrics that we measure, the analytics and the rigor that hasn't yet been applied to community management is going to get there. But I still think we're in this very early, early days understanding of it. There's a lot to learn from the tactics of Web2, but the growth that we're going to see on Web3 can level it up in such a big way because now the community has ownership. Now the users actually have ownership, which they never did before.

Doss: Totally. Yeah. No, I mean, and you bring up Web3, which is so it's, it's such an interesting, massive trend, partly because it's irrelevant because so much of it is about networks, but, but Alexis, I was actually going to ask you a question, which is on, on like sort of community curation front, I'm sure you have startups that you work with for your investment fund, where they come to you and they either say, hey, I want to be on the cutting edge of culture, or hey, I want this to appeal to really young people. I'm curious if you have an example of a company talking to you about that and kind of what you did in order to make that curation happen.

Alexis Ohanian: So using your, I know it was just a random specific example, but the like, I want to appeal to younger people type thing. Like if a founder CEO is asking that question, they're probably not going to be able to do it. There is a authenticity. That is really a big part of it. It is, if we just go back to the 101 of community building, like the idea that everyone is showing up at your house for a party or you're coming to a meeting space, whatever. If you don't genuinely believe the things you believe, if you don't genuinely embody the sort of environment you want to create, you're going to have a hard time.

Alexis Ohanian: And so what's exciting is when it works, it's coming from a really genuine place. So like in particular, this new generation that default thinks of themselves as content creators, as well as consumers just gets it because they, they seek out relationships differently. They've been molded in the shadow of Zuck and, and want to create something that looks different, wants to create something that I vibe with just because even though I'm not of the generation, like it feels real.

Alexis Ohanian: You know, I look at it with all of these cash grabs and NFTs and the projects to me that I think that are going to be the ones that endure and stand out are the ones that are coming from that place that actually understand the community side of it. And a lot of it is not, it's not tactical so much as it is vibes. It's kind of like, you know, when you see it and you feel it when you feel it. I kind of love that you had this example of coming into a party and it kind of brings up two questions for me. I told you, Alexis, I'd come back to this idea of like Flintstoning or faking users. And with where you are now, like if you were to start Reddit today, would you go for that same approach? No chance. Is that a good idea or a bad idea today? And can you fake users and still have the authenticity of a platform?

Alexis Ohanian: So I would definitely not do it. The reason it worked was also because we didn't have comments back then. And so for the first couple of months, like, yes, we were faking users, but they were just submitting links. So I think it gets into a different level of skullduggery if you're actually faking conversations, which we never did because we didn't have the comments. And also you don't have the excuse now, right? I went to, because remember 2005, there's no Hacker News, there's no Product Hunt, there's no Twitter, there's no place to talk about the thing that you've built for people who might even be remotely curious. There was no Reddit. And Facebook was still limited to colleges. And so today, no, you shouldn't have to fake users and you shouldn't fake users because you have no excuse to not get at least a hundred or a thousand to come on and at least learn from those early few.

Doss: This whole concept of Flintstoning, especially in the context of Reddit and a lot of the other services, let me just take a moment to define it actually. If you remember the old school cartoon, the Flintstones, they had like a car basically right before you, you know, you drive it, you'd yell yabba-dabba-doo. And then you'd actually use your feet and actually like just kick. And then, you know, the car was actually just the Flintstones family just walking. And so the whole idea there is that there's a lot of cases where early, early on in a product, you just don't have the users, you don't have the capability, you don't have the network. And so what you do is you just have the employees of the company, or you can kind of use some cleverness in order to try to fill that gap in the short term.

Doss: You know, the Reddit kind of users is one example. A more recent set that's been really interesting is the Instacarts and the DoorDashes of the world early on, they would put grocery stores and restaurants that they didn't have formal relationships with on the platform. And it would look like you could order from them and you could. And so if you saw, you know, your favorite Chinese restaurant, they would just send somebody to the place and order for you, collect the actual order, and bring it to you. And the restaurant would not know that they were in an app.

Doss: In the early days, Paul, I remember I used to think of Clubhouse as the talk to Paul app on Clubhouse because Paul would appear in every room and talk. And Paul was using his own engagement to try to like solve the cold start problem for everybody you know because he just be talking to all his friends that were showing up. And like, that's a real strategy. That's like a real thing that works.

Andrew Chen: Yeah, yeah. I mean, like, I guess at a more general level, the way I think about it is sometimes the approach that you take to solve it is to construct a network structure that allows for it to work at a small scale. And then the tricky part, I think, is structurally setting it up to go beyond that one community. So like one of the principles we've always had is creator first. So when we say creator, we always say like creator with a lowercase C. It just means the person who created the room. And we kind of internalized in the early days that there was going to be this tension, right? There will be times where the wishes of the creator are at odds with the wishes of a listener or a would-be listener.

Andrew Chen: Like say, Alexis, you start a room and I want to be in your room, but you don't want me there. Or you have a great stage going on and I want to come up and speak my mind, but you don't want me to. We decided in the early days, like every single time you have that situation, we will pick you. We will always pick the creator. And that allows people to form their own communities and their own clubs and their own norms. And for that mitosis to happen.

Doss: Yeah, that's fascinating. I want to come back to this idea of kind of keeping control of the platform while giving control to your creators. You know, we've talked about all the different ways you can get them up and running, you know, kind of this idea of Flintstoning. It's not just faking users. It's kind of in some ways doing the things that don't scale in those early days. But I kind of want to shift now and talk about like, as you start to get through those atomic networks and as your network starts to take hold and you have that community, how does the relationship to creators change as a startup grows and kind of what are the new problems that start to emerge?

Alexis Ohanian: Yeah. OK, well, so we set some interesting precedents by accident. The history here is started in 05, sold in 06. I stayed until early 2010 and then I came back in late 2014. And during that time that four five years span a lot of stuff got ossified or solidified that probably shouldn't or that definitely shouldn't have that instilled a really unhealthy amount of power and control among a small group of moderators who just happened to be in the right place at the right time without going too much into the details of it like for that period of time for about a decade of Reddit's life the default communities or subreddits that you would see and be subscribed to as soon as you created an account were like a list of, I don't know, 15, 20. And so all the mods of those communities, every time Reddit would add a new user, would automatically get subscribers, but only those communities.

Alexis Ohanian: And so over a decade, what that built up was a power cabal of like 20 communities and their mods. And so a lot of the work of the last six years, you know, has been to unwind that unhealthy power dynamic where a really small group of moderators were able to have total control of the front page just because we made a bad product decision that didn't get fixed. But I think culturally we just missed making Reddit a welcoming place for creators. You know, creator shows up on Reddit and says, hey, there's a there's a subreddit of my fans. And it's like, yes. And then they're like, can I have it? And the answer is no. And then it's like, OK, well, can I be a mod? No. Maybe if the mods let you. Okay. All right. Can I post content there? Like not really. It's not designed for you to post content. It's designed for your community to post content.

Alexis Ohanian: Now the pro of that is you create an environment where people feel like everything is really from the bottom up and an authentic representation of what the community wants it to be. The con of it is you lose out on the sort of goodwill and engagement of lots of people who drive lots of audience and do lots of community building. What a lot of people missed with the creator economy was not, oh, these people are making funny YouTube videos and lots of people are watching it and now they can pay their bills. What people were missing from the creator economy is that these people are building tribes, they're building cults, they're building communities, And they care about what it shows to the world, that they show a representation of allegiance to this community. That is something that I think is a tension every platform is going to have, especially as crypto starts to give more and more power to the creators.

Paul Davison: Because at the end of the day, if we're honest with ourselves, the people who drive values, the disproportionate amount of value on these platforms are the ones who make the content. For us, there are a couple of things I'd say. One is as you grow, the way that you communicate with the community evolves. One of the wonderful things about live group audio is you get to talk with people while they're using your product. And in the earliest days, when it was all one room and everyone was a speaker, it was literally Rohan and me in a room for like 14 hours a day talking with people why we built it. They would verbally report a bug. We'd fix it and ship an update 10 minutes later and go in and verbally thank them.

Paul Davison: And that's evolved, right? We started doing this weekly town hall where every Sunday at 9 a.m. we come and we talk with the community. In the early days, we would just invite everyone up on stage and everyone would just talk at the same time. And it went from an hour to two hours, to three hours, to four and a half hours. And at some point we had grown so big that we said, this format doesn't scale. We need to evolve that. You have to have systems that scale.

Paul Davison: And the other thing I would say is that you have to recognize that as the community grows, the needs evolve, right? Growth can stress the system in different ways at different times. Like I mentioned, we started out as a small group product, exclusively. And as you grow and new users come on, it's harder and harder for everyone to find their people and find their communities and to be immediately oriented. And then as you get more of a professional creator class coming on the platform, they have entirely different needs. There are entirely different engagement loops that you need to build for them.

Paul Davison: When we think about things like launching on Android and getting to general release and launching search and clips and replays and pinned links. A lot of these are designed to help those professional creators consistently grow their audience and monetize. So to me, it's really those two things. One is that you have to come up with new ways of having that relationship with the community that scale. And then you also need to recognize that with each wave of growth, you have a new set of people that might have a different orientation, might have different needs, different incentives, and you need to build tools and features that work for them.

Andrew Chen: I just want to build on what Paul just said about this professional creator class, because I think that this is sort of a recurring theme that you end up seeing across many of these different products. And I'll talk about marketplaces. Before Andrew Norwoods, I worked at Uber. And one of the amazing things that you would experience at Uber is you would go to the office and on your way, walking into the lobby, you'd have lines of drivers actually holding signs and protesting. And on one hand, it's like, wow, that's crazy. On the other hand, actually, in the book, I talk about this idea that every network has kind of an easy side of the network, these folks that are easier to get, they're less involved consumers or the viewers or the buyers, and they kind of will check things out and then they'll bounce. And then there's usually a set that get really, really into it. And they do a lot of the hard work for social networks that ends up being the creators for marketplaces. It tends to be the sellers for workplace tools. It tends to be the people who actually start and organize the projects.

Andrew Chen: You know this is sort of the hard side of the network. And so what tends to happen over time as both Alexis and Paul have alluded to is that they start out as kind of normal users of the platform but then eventually they come to professionalize over time. They tend to professionalize over time and take it more seriously and want more tools and want more analytics. And especially they want new ways to monetize. And if you have ways of doing that, then amazing, they'll stick around and build more and more and more. And you've seen that on everything from the YouTubes and TikToks and all those products, as well as marketplace companies and so on. But if you don't, the issue becomes because they're professional and they provide this higher level of service, and they often serve content or sell the best products or do all the things, they're also the most damaging if they decide to switch from your platform to a competitor's platform.

Alexis Ohanian: I mean, I think that's well said. And when we think about business model. This is another thing that's evolved a lot over time. Back in the earlier days of social networks, it was just assumed that everything was going to be an engagement platform and you'd have to get to massive, massive scale and then start to monetize. But now it's increasingly easy to build a business that is aligned with the needs of the creators, where you're only growing as a platform if they are growing individually as creators. And that's something that's changed a lot over time, which I think is a really healthy development.

Alexis Ohanian: And to Alexis's point, because of Web3, a lot of this business model around creator is going to be baked in from day one, right? Which is very different than how it used to be. It used to be, you know, you build this huge platform and then at the end, you'd kind of be like, okay, should we do ads now and do rev share? Yeah, exactly. I mean, yeah, right. Or in the early days, like a lot of platforms still don't do rev share, right? I mean, it's really about the platform growth. And I think the dynamic has changed a lot.

Alexis Ohanian: It has. And I still remember advice I was given in 05, which was just get big enough to run ads and you'll make money. That was what Facebook aspired to do and was sort of the North Star for all of us. The first revenue Reddit actually got was from merch. And this was the biggest fight we had had at this point because I really wanted to sell some merch. And at the time, it was hard to do. Like Stripe didn't exist. So I had to hack together a storefront, asked a friend to take some photos with the shirts, It's bought like maybe 300 that sat in the apartment. And, you know, this is only a few months into Reddit, but I could already sense like there's a community here. They want to support us. Like, let's sell T-shirts. We sold out within a few hours. And the next day I took garbage bags, packed all of them myself, hand signed the notes, you know, thanked people for supporting us, and then took them all down to the post office in a couple of garbage bags to ship out. And that was such a strong indication.

Alexis Ohanian: And so I look at what happens with PFPs and it just seems so obvious now. You know, I regret taking as long as we did with Reddit gold and actually not doubling down on those community elements even sooner. Because it was actually Drew Curtis at Fark, who I was complaining to over drinks in like 2008, 2009, who said, dude, ads suck. You should just run a system like we have. They had a total, it was called Total Fark. And I think it was $10 a month you would pay, 15 bucks. And you would just have a little logo next to your name when you posted on Fark that made you feel special because you were part of this club. And so it was a status symbol, just a little fav icon size, like 16 by 16 pixel image next to your username, whenever you posted and you get access to a special forum. And I'm like, dude, Drew, I mean, really people pay money for this? He's like, no, you don't understand. They pay money for this. And then once you allow them to gift it to one another, they will spend even more. And it blew my mind. And so I come back to the team. like hey we got to do some kind of total Fark thing eventually this ships as Reddit gold although it was shipped on April Fools as Reddit mold but regardless I think I keep going back to Web3 but the flywheel I always wanted to create that is happening right now and this is probably a controversial thing to say but I really believe that we're going to look at this blip of Web2 as this weird mistake where we just did the best with what we had because you know yeah 2005 running online billboards on websites seemed like a pretty good way to make a sustainable business. But mixing it with community, mixing it with social dynamics, mixing it with all the wrong incentives that it creates when at the end of the day, you just want engagement. At the end of the day, you just want people coming back and clicking and hopefully seeing some more ads. It's actually really unhealthy. And I relish the fact that this system is going to lose quickly because a better user experience is emerging or has emerged now that the revenue models actually are aligned between the platforms and the actual creators.

Doss: You guys have really hit home that Web2 to Web3 is a big shift. What should somebody starting a company today keep in mind at this particular moment in time? It's such a huge shift and the entire playbook is getting reinvented. So much of what's happening in Web3 and getting communities excited about your NFT drop and everything is all happening on Discord. It's happening on social media. And it's less so happening as, for example, like no one's tried, I haven't seen anyone try to do a college by college expansion program around a Web3 product because it's Web3 being inherently decentralized and global. A lot of use cases that people are the most excited about, because there's all the financial use cases, but a lot of the things that people are most excited about are things like the intersection of games and NFTs and virtual goods, for example. And in order to build something like that, you have to think about it not just as like, hey, let's make this game fun. If you go and look at Axie Infinity, you look at Zed Run, you look at a lot of the successful crypto games out there, it ends up being about how do you create the economic incentives and the economic game around all the design so that people play it in a very, very different way than they would play like a Fortnite and my colleague Chris Dixon talks about this a lot which is whenever there's a invention of a new uh platform you look at the the sort of um transition from web to mobile as one and then now from from you know web 2 to web 3 as kind of the other you know broad set people start by basically just trying to figure out like oh well what's something that worked on the web that i should now put into a mobile app and i think that's what is is very tempting for folks to think about, which is like, oh, yeah, we're going to have the social network, and we're going to build the Web3 social network. In reality, the things that people end up being the most excited about, whether that's Instagram or Clubhouse or any of these, are using capabilities that only exist on the phone. And then simultaneously, if you go back and think about it, the web didn't die, it didn't disappear. What actually happened was mobile and the web now coexist with each other. And you end up using desktop apps, you end up using the web, websites have continued to grow bigger and bigger and bigger as more people have used it. And also mobile has layered on new apps, new interactions, etc. on top of all of it. So I think the other big thing that founders are going to have to think about is where do they want to play? Do they want to do something that's truly crypto native, crypto first? Or do they want to do something that is potentially an extension and it's more like a hybrid? And I think there's going to be a lot of those opportunities.

Doss: Well, I just want to give a huge thank you for everybody joining us this morning, especially to Alexis, Paul, and Andrew for coming on and talking about network effects. For more on this topic of how to kickstart and scale network effects and beat the cold start problem, check out Andrew's book, *The Cold Start Problem*, which is out December 7th. And you can check that out at coldstart.com. So thank you so much, guys. This has been a fantastic conversation.

Alexis Ohanian: Thank you so much.

Paul Davison: Awesome, appreciate you guys.

Andrew Chen: See you later.