by Y Combinator2/23/2018
Reid Hoffman interviews Sam Altman in this uncut interview from the Masters of Scale podcast.
Craig Cannon [00:00] – Hey, how’s it going? This is Craig Cannon and you’re listening to Y Combinator’s podcast. Today we have an uncut interview from the Masters of Scale podcast and in it Reid Hoffman, the co-founder of LinkedIn, interviews Sam Altman. Alright, here we go.
Reid Hoffman [00:16] – I’m here with Sam Altman, President of Y Combinator, who is a good friend and has been involved in many scaling things. Let’s start with your entrepreneurial. What got you in entrepreneurship, how you started Loopt, why you started Loopt?
Sam Altman [00:34] – First of all, thank you for having me here. I fell into it accidentally. I went to college to be a computer programmer. I knew that was what I wanted to do. And I started college after the dotcom bubble had bust so startups weren’t on anyone’s find. I remember one thing I was surprised by in college, as a freshman, was that I thought people would still be excited about startups and if you said you were working on a startup people just sort of laughed at you in not a nice way. And I actually didn’t want to work on a startup. I worked my summer of my freshman year in the Stanford CS department as a researcher and I loved that. Out of that grew a project, which eventually developed into Loopt, but it started as just a project that we worked on after class. It would not have been a startup if it were not for Y Combinator. It got to the point where we had worked on it during spring quarter and it was really fun. I’m very ashamed to say that I had been planning to go be an intern at Goldman Sachs that summer. I accepted a job offer and I realized I was having much more fun with all three of us working on this project. And we all kind of knew who Paul Graham was. We had followed him online and he posted this thing saying, “Hey, not excited about your summer job? Come hack on your project, make a startup.” And that seemed like it would be more fun than being an investment banker so we applied to YC and flew out and interviewed and got funded. We were actually the first company ever funded by YC and then it just kept going.
Reid Hoffman [02:26] – Is there anything that even this the very beginning, is there anything from now, having done Loopt and a bunch of things we’re going to get into that if you could call that younger self of yours going into YC, that you would tell yourself to do differently? Like key things.
Sam Altman [02:47] – One general thing that I didn’t understand then and learned pretty quickly, but would have saved me quite a bit of heartache is about how to calibrate risk. Most people worry way too much about risk. When you’re young and you have nothing to lose is absolutely the time to take risks and it’s the time, unfortunately, that most people are the most risk averse in their lives. They want to work for a few years, build up savings then they’re just going to start out. They’re going to do what their parents want, whatever. I ended up in the right place, but it could have gone either way and I was very, totally stupidly nervous about the risk so this idea that most things are not nearly as risky as they seem is a powerful one and one that I always try to tell people in that position. You’re a poor college student with no money and a no reputation, if you do a startup and fail you’re like two years older with no money and reputation it’s fine. It’s actually much harder to wait and let your life ramp up and then do it. That’s one thing. Another thing is why do you think I’m super easy to work with today, by I was like infamously difficult to work with when I was 18 or 19 and I would have put more effort into trying to be better about that.
Reid Hoffman [04:09] – What specifically would you have done to be easier to work with?
Sam Altman [04:16] – A lot of it is how you set and communicate expectations with others and also realizing that if you’re the founder of the company and you wanted to work 100 hours a week and be super focused and productive that’s cool, but most other people you hire, especially as you get bigger, have other lives and you need to understand that. Again, everyone learns this lesson quickly, but it would have saved some pain along the way if I had learned it earlier. The other thing that I think I got wrong and 19 year olds starting companies often get wrong is because they evolve fairly organically from projects, you never take the time to realize all of a sudden I’m running this company with 10 people and we’re doing this and we’ve raised all this money and do I really believe that this is going to be a market that will support a giant company. There are several checkpoints along the way where people don’t give that enough thought.
Reid Hoffman [05:26] – How do you think that they should make those checkpoints, because from other conversations I know that both you and I think this whole total addressable market, TAM, thing is frequently very illusory.
Sam Altman [05:40] – The most interesting companies start with a TAM of nearly zero and it’s like the very bad investors are the one that’s focused on the TAM of today. The good investors are focused on the TAM of 10 years from now. The thing that I have seen be most predictive for … TAM is total addressable market. The thing that I have seen be most predictable for a large TAM down the road is how much the people that are using it today use it and love. One of the things that was obvious when people got iPhones. Even though only a few million of the iPhones sold, the people that had them used them everyday and loved them. It became like their most precious item. I remember shortly after the iPhone came out I was in a developing world country. It was really quite poor and people had nothing, except they all had a smartphone and once they had one … You read these statistics and people need to do some lightweight journalism about would you rather give up your smartphone or x, it doesn’t really matter what x is, they’re going to keep the smartphone. You could have predicted with a lot of certainty, many people did, that this was going to be a large market. It was small in 2008, 2007; but it was guaranteed that it was going to grow very quickly, because of how much people loved it. The internet, in the early days, was the same thing. And I think a lot of other trends people jump on too early, because a lot of people dabble, but put it on the shelf. A lot of people have bought VR headsets and put them on the shelf.
Sam Altman [07:24] – It doesn’t mean that VR will not be big someday, but I think you can make a high conviction that on starting a VR company today at least the TAM will be huge. It still feels like we’re at the PalmPilot, not the iPhone era.
Reid Hoffman [07:39] – For those who remember PalmPilots.
Sam Altman [07:42] – I have one still, that I got when it was a ridiculous thing to have. PalmPilot 7 was the internet connected one and I loved it, but other people didn’t. It was very easy to tell the difference between the reaction of that and the iPhone.
Reid Hoffman [08:00] – I think you probably know this, but the first use case for PayPal was splitting the dinner check on PalmPilots.
Sam Altman [08:07] – I do know that and apparently Levchin had set it up so that it could only happen once, because it was actually cryptographically secure and he stayed up for nights in a row or something, it’s demo and the camera didn’t work and it wouldn’t do it again. And they only had two that it work it on.
Reid Hoffman [08:22] – And we did the closing of the first financing on it in BUX. Classic Silicon Valley. Loopt, then how did you get to YC?
Sam Altman [08:37] – Loopt ran for seven years. We got acquired and I was trying to figure out what I wanted to do next and I decided I was going to partly take a midcareer sabbatical, race cars, fighter planes, travel the world, all that kind of stuff, but I didn’t want to totally disengage from working and I would try to invest for a while. I took the money I made from Loopt and I also raised some outside capital and I became a very traditional seed investor. I’d invested a little bit during Loopt, a very small check. Yhat was possible back in the day, because the valuations were so much lower and the companies raised so much less money. I remember like in 2010 I invested in Stripe and had $17,000 in that bank account at the time. Maybe it was 2009. And I invested like $15,000 of that into Stripe and it was for a meaningful percentage of the company. It was just a different world. And that’s kind of like as a founder, not making much money. That was the size of check you could do, but it was still possible to like once a year make an angel investment as a founder. And now I just think you can’t, because the amounts of money in the seed rounds have gone up so much. But because it was a different world I had invested a little bit during Loopt. I think I made three investments. And I thought I liked it so I was like well I’m going to do this as my sabbatical job. And I did that from 2012 to 2013 and I turned out to be fairly good at it, but I didn’t like it at all. You just have to try things to know,
Sam Altman [10:17] – but it turned out that I don’t want to be a seed investor or a VC. I liked running a company. I did not like being on the sidelines. I didn’t get the adrenaline rush I get out of being in the trenches of running a company, which is something that I think a lot of founders miss when they start investing. You figure out an approach to this I like for investing, which is continue to be very involved with operating one company and I think that’s a model that is very under explored and we’ll see more of. I did this and I was like, you know what, that was fun and it did pretty well. I think it’s a ridiculously highly compensated job, but I didn’t want to keep doing it. I was thinking about things I wanted to do and Paul Graham sort of jokingly said, a number of times over the years, “I’m going to retire and you should take over YC.” I had kind of bucketed that in my head, this is investing and I don’t like investing. But I started talking to him about that more seriously. I was looking at two other things.
Reid Hoffman [11:37] – Goldman Sachs.
Sam Altman [11:38] – No, I was thinking about going to run a big public company or start a company that I had been excited about for a while. Finally, I was like I’m really going to think hard about what I want to do and YC seemed like such a promising and underutilized thing and so important to what I cared about, or at least I realized that there was this set of four or five things that I really deeply cared about and YC had the best platform of anything I was looking at or maybe anything in Silicon Valley to go address all of those and that even though there were things about it I wasn’t excited, I could get other people to do those or I could change YC in big ways and that worked out basically as I expected, which was great.
Reid Hoffman [12:25] – Go into a little bit more detail about what the things were, like this is one of the things you see the potential scale, see this asset, it could be so much more in Silicon Valley. Which ones were those?
Sam Altman [12:40] – One was the kinds of companies that we were funding. At the time YC was mostly funding software companies, but I had a lot of conviction that we could apply the same thing that made YC work so well for software companies to companies in a lot of the areas that I cared about: AI, aesthetic biology, energy. And that the same model would work. Now people are like oh yeah, hard tech, everyone wants to invest in hard tech, but at the time it was like this is a really dumb thing. One of the things that’s funny, as a side note and just as a note to anyone that tries to do anything where you take a company in a different direction or scale it, is that it is always funny to read the articles from the same journalists that when you say you’re going to do this thing: Sam in crazy, completely unqualified, this is not going to work, YC is going to die, going after hard tech companies is so stupid to like a year, 14 months later, this is great, Sam is a genius, this is like predestined he was going to take over YC, it’s ridiculous that YC’s doing any software companies and all this other stuff. I think you just have to ignore all of that and just say I have a high level of conviction and we’re going to try this thing. And most people will tell you it’s not going to work, if it’s something new. Most people are afraid of things that are new and you just do it and as long as it’s … Again, risk is miscalibrated. It’s probably not that risky, probably won’t kill the company and probably undervalued if everyone else says it’s stupid.
Sam Altman [14:18] – We were able to do that and the first thing was expand YC into all these different directions. The greatest companies are created on the fleeting edge of what people are working on and by 2014 there were already a lot of people making mobile apps. There will still be great mobile app companies, I’m sure, but it’s like fish in a barrel in some of these areas where we’re able to just go pick out in the world the best quantum computing company, the best self driving company, which recently got acquired by GM, the best nuclear fusion company, the best synthetic biology company, just all the way down the list and no one was competing to the rest of those companies. Meanwhile like another photosharing company, we have to go to a lot of work to convince them to do YC, because they’re like, “Well, every VC in Sand Hill Road wants to give me three million dollars.” That was one area where we expanded. Another was just more companies. We significantly upped the number of companies we fund every year.
Reid Hoffman [15:14] – To what, for example?
Sam Altman [15:15] – Maybe it was like from 100 to 280, something like that now. We expanded geography so we plan on now funding companies from all over the world, which is a logistical nightmare, but really good for the kind of companies we can serve. We raised a later stage fund, because one of the other things we realized, especially with the hard tech companies, you can fund a lot of these companies. At some point they need to raise 50 or 100 million dollars. Not a lot of investors are doing that so to support those companies we need this large pool of capital. We wanted to, on the other side, really increase the top of our funnel so we started teaching MOOCs, basically. Like hey we’re going to try to distill how to start a startup into a class and make that available around the world. We started a research lab. There are some things that are important to us and our mission and our vision for the world that don’t fit as a for profit company that we still want to do. Fundamentally YC is sort of this new university. We are a collection of smart people that have some sort of shared vision of the world we’d like to build and the tools we think work to get there and that’s a very flexible structure so we’ve been able to do all these things.
Reid Hoffman [16:32] – In terms of the efforts, the scale YC has this global geography, this early and late stage, there’s research that you’re also taking non profits through YC as well as part of that. When you think about it and you say, “Okay, I saw this potential to have a massive scale impact and I went in,” which were the things you wish you had doubled down on earlier and which are the things you would have changed in that scaling process?
Sam Altman [17:11] – Two things are really important to get right when your going to try to scale an organization a lot. One is a very clear vision and culture and the other is a reasonably clear work structure. We were good on the vision and culture, this idea that YC wants to produce the most innovation in the world and then do that in such a way that we make the future great for everyone, not for seven people. We’ve kind of talked and took over about that the mission, the vision, the culture support that and we’ve done a very good job of staying true to that. One thing I’m particularly proud of the organization on is, without me having to legislate it, if there’s a company we look at where we think we’d make a lot of money, but we think it might be bad for the world we won’t touch it, which most other investment firms unfortunately have a hard time with. We do things that other investments firms would not, like fund basic income research or support open AI or fund nonprofits so I think we got that right early and getting that right is really critical, because if everyone believes the same thing there’s a lot less conflict and also if people are going to kind of organically march in the right direction you need a lot less organization. Usually it’s some and we could have put a better structure in place earlier. We’ve gone through a few iterations. We now have one that’s good, but initially tried to sort of just have no structure at all and that would have worked if we had stayed at like 10 people working at YC,
Sam Altman [18:54] – which is what it was when I took over. Fell apart pretty quickly at 30, 40 people, something like that.
Reid Hoffman [19:00] – Was it a deliberate attempt at holacracy or was it just kind of like oh let’s try not to bother with the work dynamic.
Sam Altman [19:05] – Definitely not a deliberate attempt at holacracy, which I don’t think I’m a fan of, from what I know about it. The growth in the number of people in an organization sneaks up on you and you can completely get away without any structure until you can’t and it’s a pretty quick flip.
Reid Hoffman [19:30] – Was there anything you would have applied capital to more fiercely or anything that you would have said, “I should have recruited these people into the organization earlier?”
Sam Altman [19:46] – We run super light on capital and that is to our credit. There are all these things that would have been smoother and better if we’d applied more capital to, but the trade off is it would have hurt our culture and that reflects on companies. I like that when entrepreneurs come to YC they drive to a shitty industrial part of Mountain View and they walk into this building that looks like it’s … We did a nice job on the inside, but it’s like nice and very … You’ve been.
Reid Hoffman [20:23] – I have.
Sam Altman [20:24] – It’s nice in the very cheap sense. There’s no gleaming marble or you know…
Reid Hoffman [20:29] – What you see on Sand Hill Road a lot.
Sam Altman [20:30] – We still have a CFO that yells at people if she finds there was a $50 cheaper flight to buy. In all these specific instances we could have applied more money to problems and it would have helped, but there is something culturally important to us about frugality, because we want that reflected in our start ups and our start ups have the kind of bond with us where they reflect what we do. On the whole it is good we did not try to go solve a bunch of problems by mega amounts of capital.
Reid Hoffman [21:11] – Let’s go to the YC selection process, because we’re going to go into what are the lessons in YC companies. And one of the things I learned from the New Yorker profile, which I had no idea, because I’ve never seen it, is you occasionally bring a sword in with you to interview an entrepreneur.
Sam Altman [21:27] – No, that’s not true. Either they wrote that wrong… I remember what happened. I love engineering history of all sorts. I collect these things like Concord engines and just things I think are important engineering milestones. I had bought this Bronze Age sword and I was in office hours and the reporter was sitting with me and office hours are kind of this drudgery that sometimes… It’s like the 16th meeting of the day. I needed some energy so one was a phone call and the sword had just came, beautiful Bronze Age sword and it had just arrived. I had been waiting for this thing. I’d flown it over from Europe in this big crate and I got it out and it was stunning, perfect. The first thing you do is pick it up and swing it and see how it’s weighted, how it feels. This particular one had the nicks where it’d hit peoples’ helmets a couple thousand years ago. It’s a little dark, but …
Reid Hoffman [22:36] – Or maybe bones.
Sam Altman [22:38] – Too much of a nick for a bone, too deep of a nick in the metal. I was so excited and I was on the phone and it was kind of like a not particularly exciting conversation so I picked it up when were in my house and I just started, while I was on the speaker phone, swinging it around and fighting this pretend enemy, because I was so excited. I had just got this. I had been waiting for it for so long. And I didn’t realize until I put it down at the end that was probably really dumb, that’s probably going to make it into the profile, because the reporter was sitting there watching. You just kind of forget after someone’s with you for weeks. But I have never swung a sword at someone during and interview process.
Reid Hoffman [23:17] – It was less the swinging sword, more did you bring props with you to the interviews.
Sam Altman [23:22] – No. This was just in my house and it had just arrived. We don’t have props.
Reid Hoffman [23:26] – Tell me a little bit about the YC interview process. This is one of the things you’ve really refined over the years, in order to get really good.
Sam Altman [23:34] – There are a few big ideas, actually a very small number of big ideas that make YC work and one of them is that there is a giant amount of arbitrage left, because most investors in Silicon Valley will only fund you with an intro or they’ll only talk to you with an intro and it is, honestly I think, by and large a fairly insular network. One of the things with YC is if we can build a process that’s not too painful where we can look at everyone smart that wants funding from us, that we can unlock a huge amount of value, because no one else does that. Or at the time no one else did that. Now other people do it, but I don’t think they do it as well. We have a brand or something. I think we still have an edge. But we are willing to look at tens of thousands of people a year with no intro, who don’t know us. And there’s a lot of really smart people that, just because of the circumstances they were born into or the country or whatever, aren’t plugged into the Silicon Valley network and we are bridge to that and we get compensated for that, but that idea that you should have an open application and you shouldn’t require an intro is great. Doing it well in practice is hard, because it is very tiring and very draining to talk to 40,000 people a year so we have a lot of people, a lot of software. This is like a secret I don’t even mind telling. Other people will copy it well. We spent so much time and money building really great internal software that only 50 people in the world use and yet it lets us run this process
Sam Altman [25:14] – that no one else can run.
Reid Hoffman [25:16] – And video tape them and you cross check the results–
Sam Altman [25:20] – Yeah, we cross check the results. We watch companies that we say yes or not to. When we say yes to a company we have to say how strong of a yes it is, how well we think they’re going to do and we really sweat every mistake.
Reid Hoffman [25:30] – Tell me also, before we get to some of the interesting companies, about your going global efforts. What were the the things to scale to more than just the Silicon Valley network?
Sam Altman [25:45] – The number one best thing we did for that was the MOOC, the class that we teach. That got incredible distribution worldwide and then that brought people from these other countries and what we find is that when we fund someone from another country or city for the first time, they go back and everybody else there is like, “I can get into YC now, it’s not so hard,” so we there’s like this incredible chain reaction effect once we get the first good company in a new area. We get on planes a lot. Partners just go fly around the world and talk and meet companies. That is phenomenally effective. It won’t be too much longer before we have more companies applying from outside the U.S. than inside the U.S.
Reid Hoffman [26:33] – And then you were also doing a fellows program at some point?
Sam Altman [26:35] – Yes. It was sort of like an on ramp version to YC. One of the things we realized that there was more demand for that program than we were ever going to fill. We were shocked. We thought we were going to get like 200 applications for the first fellowship. We got like 7,000. This is not going to work as designed. We have now evolved that and the class that we’ve taught together into this new thing that we’re going to try next year for the first time, which is halfway between a lecture series and a Y Combinator class where people have individual advisors and office hours and they have to report their metrics every week, but they can do it remotely and anyone can do it.
Reid Hoffman [27:17] – Now let’s go to a couple of the great, just amazing YC companies. There’s a number. We talked about Stripe a little earlier, is one of your earlier. I actually hadn’t realized the background economics. You were more edgy in your angel investing than I was, because started pretty early with the crazy, “How much are your savings?” I think you had a higher percentage of your savings that you were deploying.
Sam Altman [27:44] – Risk management has never been my strongest suite.
Reid Hoffman [27:47] – I think I went up to 50%. I think that was somewhere roughly around 90.
Sam Altman [27:51] – It’s different when you have no money at all though. There’s just some level beyond which the percentage doesn’t matter.
Reid Hoffman [27:58] – Let’s talk about a few of them and what are the key lessons for how the scaling, what we learned and what common here in Silicon Valley, what kind of scale. Let’s start with Stripe, where obviously the Collisons are amazing and the company is doing spectacularly well. What did it look like in the early days, why did you think it’d be great, what did they learn about how to scale?
Sam Altman [28:21] – The number one lesson that I have learned or that YC has learned about how to scale well is that the first thing you have to do is build a product that is so good people spontaneously want to use it and tell their friends about it and if you can do that you still have to blitz scale, but it’s the easy kind. You have too much demand. The hard kind of blitzscaling is where you try to start scaling up before the product is really great and then most of your effort at scaling is to generate demand. The number one most important insight about how to blitz scale is that the good kind of blitzscaling is when you are not having to generate demand as you go, but that you first got the product right. In many of these cases: Stripe, Dropbox, Airbnb… it took a long time to get the product right, but they were obsessed with that and then when they did all their effort is, okay we have so much demand that without much more effort we know this is going to keep growing 20-30% a month for years. That’s a real problem. It’s a high class problem, but it’s still a real problem. How do we build that? That is the kind of scaling that works and that has generated Facebook, Google, a lot of… It’s the same playbook. The kind of blitzscaling that we have seen go badly is we have a mediocre product, we have raised hundreds of millions of dollars and our VC is beating down our throats to hire more salespeople to grow faster.
Reid Hoffman [30:01] – Any particular examples?
Sam Altman [30:03] – I don’t want to name names. There’s so many to pick from. Thankfully most of them are not YC. One thing that is pretty good, and again a few exceptions to this, we try to beat that idea out of people during YC and thus most of the mistakes in Silicon Valley of that sort in the last decade have not been ours.
Reid Hoffman [30:28] – One interesting thing that I think is a slight variant on the theory you just gave that applies to Airbnb is actually in fact one of the things I think I’ve learned about some scaling things is initially you have to do things that don’t scale in order to get into scaling. Actually they spent a bunch of time kind of out in the desert, not getting transactions and I think it was actually Paul Graham who gave them this advice, said, “Look, go to New York. Go door to door. Explain to people, just get them into it.”
Sam Altman [31:06] – Paul Graham wrote an essay out of that experience, the Stripe experience, that somewhat we can talk about. The essay is called “Do Things That Don’t Scale.” I think it’s in the top four most important essays for a new founder to read. This is almost universal among, not perfectly, but almost universal among our best companies, where initially you have to go get users manually or do things that you could never do with 10 million users. And it’s actually a sign of bad entrepreneurs, in my experience, when people that have a company with no users, no product and no revenue say they won’t do something because, “Well that’s not going to scale.” What that means is I am lazy and don’t want to go get my hands dirty or I think knocking on doors and taking photos is baloney, in the Airbnb guys’ case.
Reid Hoffman [32:01] – Which is not what they did. They actually went and did all those things.
Sam Altman [32:03] – Exactly. The way that you build a really great product is to be very close to your customers and the way that you do that is to do things that don’t scale. It’s super important. It’s a critical piece, even in 2016 in Silicon Valley, still not fully internalized advice.
Reid Hoffman [32:31] – Are there any cases that come to mind for you that were where someone built a really good product that had that potential for love, that actually failed to scale?
Sam Altman [32:47] – Twitter is the example everyone uses. They say, “Well if the company had scaled better, it would have been a 200 million dollar company not a 20 million dollar company,” or 10 or whatever it’s worth now. If Twitter is the best example you can find that is not working, that still says something. There are others. It’s more often where I think the founders were really good at building a product, but they’re not good at all at building a company, so it never even got far into the public conscious, but it’s rare. It happens surprisingly rarely.
Reid Hoffman [33:27] – What advice do you give YC founders on hiring to scale?
Sam Altman [33:33] – Vinod Khosla has this soundbite that I have always loved, which is that the team you build is the company you build. And a lot of founders, particularly young founders, but a lot of founders are afraid to hire people that are a lot more competent or experienced than they are. And there’s some truth to this, because a lot times hiring really experienced people backfires and they actually turn out not to be good at all, but it can be magic when it works right and the thing that I think you need most to scale well and quickly is two or three senior team members that you trust and are really good at scaling an organization, especially if you’re a CEO. The recipe that has worked pretty well in Silicon Valley is a relatively inexperienced CEO who identifies a small number of direct reports who are really great at scaling things up pretty quickly and know how to do that. And if you don’t hire those people, generally you will just suffer.
Reid Hoffman [34:48] – Do you, in YC, teach the companies anything about culture building?
Sam Altman [34:52] – Not enough. We do do some of that in the early days, which I think we’re fairly good at, but culture building is not this thing that you do in the first three months of your life and then stop. Shit gets a lot more important as the company gets well out of YC. One of the things that we’re thinking about doing at YC is programs that continue to teach our founders more, well after they’re out of YC. The model used to be that you would go through YC and then you’d get a board member who would only take eight boards and would spend a couple days a month with you and do things like help you build the culture as the company scales, but just the dynamic of Silicon Valley venture has changed so much and there’s so many more companies and deals per investor that a lot of these things that used to get taught one on one by a board member just don’t get taught at all so I’m interested, in terms of what we can do to teach that to our alumni.
Reid Hoffman [35:53] – One thing you said during the Stanford seminar is what a founder needs is something like idea times product times execution times team times luck. Is there any particular where luck is a random number between 0 and 10,000? Is there anything in particular you think that people should keep in mind about how to play luck, how to make luck, how to factor luck into their strategy?
Sam Altman [36:23] – There a lot of ways that you can manipulate the outcome and there’s a lot of things you can do unrelated to luck and certainly you can be really lucky, but if you don’t do the right things well the company will still be worth nothing or very little. But I think also things like try to come up with a good idea and try to pick a market that’s going to be really big and trying to hire a great team and trying to execute really well. Those are all the ways you minimize the effect of luck. That said, it’s a chaotic universe. So sure, you can do a whole lot to minimize that and you should. I also generally believe that if you do really well on those four categories, even if the first or the second company doesn’t work out you can tell people that are eventually going to be successful at something. It takes a lot of tries.
Reid Hoffman [37:19] – If they keep at it, especially. One of the things I think is the way that I think about managing luck is, one of the things YC provides founders, which is really good, is a network and it’s the use of the network, both for seeking opportunity and also seeking risk assessment. One of the pieces of advice I give founders is to constantly go to smart people and say, “What do you think … What land mines am I going to run into? What risks might I have?” Not just, “Do you like my thing or not,” because I’m sitting with you, I’m probably going to say, “Oh sure, your thing’s great,” unless I’m deliberately trying to help you by being edged. What do you think about those as key things for luck management?
Sam Altman [38:02] – I agree. The way I have always tried to think about it for myself is that luck is a big factor, but I’m going to keep working and eventually, because it’s a random variable it’s going to swing my way and I think that’s roughly the right mindset to have. If you don’t acknowledge the role of luck at all, I think you’re wrong in a dangerous way were you sort of just are not a great human. If you can’t look and say I got really at some point it’s probably bad, but if you’re also like well it’s about luck and I have no chance, the world is against me and I’m just going to sit here and complain, that’s not going to work either. The roughly correct mindset is luck is important, but I’m eventually going to get lucky and I’m just going to work really hard until I do.
Reid Hoffman [38:56] – That may be one of the better definitions of optimism that I’ve heard. One of the challenges of being the president of scale thing is that all of a sudden the demands on your time go ferocious. For example, you get a deluge in email looking for meeting requests. How do you manage that?
Sam Altman [39:23] – I don’t think it’s rude to send someone an email asking them to meet. I also don’t think it’s rude not to respond or to write a quick thanks, but I’m super busy. I still don’tmanage this well, because I still feel more guilty than I should about not doing things, but I try to be pretty rigorous about not doing what I don’t want to do. Everyone finds their own productivity systems that work well for them. I’m a lists person so I make lists everyday of what I need to get done and I make annual lists of the big picture things I want to do. Because YC is a services organization, I’ve just accepted that I’m going to be very interrupt driven and that most of my time is going to be dynamically responsive to our companies that need help and I just accept that that’s our fundamental job, I’m going to devote 50-60% of my time to that and I just leave it empty, because I don’t know when it’s going to come up, but other than that I just don’t do stuff I don’t want to do and this is an underutilized strategy, because it feels rude. And again, because I feel it’s rude, don’t do nearly as good of a job of it as I would like. The other answer that no one likes to give is just work a lot, just do a lot of hours. People talk about working smart all they want, there’s nothing that makes up for working smarter and a lot of hours.
Reid Hoffman [40:56] – What got you into coding?
Sam Altman [40:59] – I don’t remember not being into coding, honestly. I got a computer for my eighth birthday and I already knew how to program it, very rudimentary programming. I learned at some point before that at school, but I don’t remember if it was just …
Reid Hoffman [41:15] – The New Yorker thought it was analyzing area codes.
Sam Altman [41:24] – New Yorker writers come up with a lot of theories, but I don’t think they’re particularly right on that one.
Reid Hoffman [41:28] – Got it. Because I was curious. That was in the category of things I didn’t know in addition to the amateur swordsman.
Sam Altman [41:38] – I have a brain that is naturally inclined to enjoy things like puzzles and naff and coding is a fun way to do that.
Reid Hoffman [41:51] – And then for the last personal question of this sort. Many people have noted your affinity for cargo shorts.
Sam Altman [42:00] – I’m not wearing them today.
Reid Hoffman [42:01] – I know.
Sam Altman [42:03] – It’s cold outside.
Reid Hoffman [42:03] – When it’s warm, cargo shorts. Is there any particular reason cargo shorts?
Sam Altman [42:08] – Honestly, I don’t think they’re that ugly and I find them incredibly convenient. You can put a lot of stuff. I still read paperback books. I like paperback books. I like to carry one around with me. I have an iPhone 7 Plus, which works really well in cargo pockets. I carry computer chargers, cables. They’re just efficient. Why people care about that so much, I can’t tell you.
Reid Hoffman [42:35] – I don’t know either, but it’s somewhat your Batman utility belt.
Sam Altman [42:38] – You just can carry a lot of stuff.
Reid Hoffman [42:40] – Moving back to blitzscaling and to scaling. You and I have talked a bunch about what are the theories of what makes Silicon Valley unique is the networks of talent and networks of practice in order to scale things quickly and some of that is to actually build a product that actually goes fairly global pretty quickly. Is there anything in the theories that I’ve been doing that you think I currently have a hole in? Because you’re familiar with the whole range. Is there anything you would critique the current blitzscaling theory?
Sam Altman [43:29] – I certainly agree incredibly strongly that the magic of Silicon Valley is the dense network of people that have this knowledge, these connections, the willingness to help for free. The network is so dense that all the different pieces you need, of which there are a lot, to blitzscale are all together. And if I disagree with anything it would be how quickly that will spread outside of Silicon Valley. It may happen relatively quickly. There’s still a question of how long does Silicon Valley remain the absolute dominant force in startups and at this point a lot of that knowledge and talent an people and capital and cultural mind share and whatever is seeping out in a way is good. I don’t know if it’s against the theory, but at least just a relative weakening of Silicon Valley.
Reid Hoffman [44:37] – Let’s do one more version of this, because I think it’s useful. The central theory is part of what Silicon Valley itself has is network effects, where it’s network effects of the talent, network effects of the knowledge sharing, network effects of company and business formation, therefore creating another Silicon Valley or other Silicon Valleys is actually going to be super difficult, because you need to have that network effect density in order to make it happen and your counter theory is actually the networks are actually being built in these places, because the culture of entrepreneurship is spreading?
Sam Altman [45:13] – Yes. I think it’s just spreading.
Reid Hoffman [45:19] – And do you think the question of scaled talent and the ability to play the scale games is sufficiently spread?
Sam Altman [45:30] – Well I think if you look at Beijing, it’s spreading there.
Reid Hoffman [45:34] – Actually China is its own case. There’s blitzscaling in Silicon Valley, there’s blitzscaling in China.
Sam Altman [45:38] – Maybe L.A. is a more interesting case. L.A. is not Silicon Valley. One of the things I’ve always said that makes Silicon Valley work is that startups are the number one thing. In New York it’s finance, D.C. it’s politics and in L.A. it’s been media and it’s still definitely not startups, but it’s interesting that startups, there’s a number of startups in L.A., Snap mosy prominently, but others that are doing really well. In fact I bet there’s more billion plus dollar companies in L.A. the last few years than in New York, or at least more total market cap. L.A. has not been traditionally thought of as a hot spot for startups at all, at all and yet somehow while everyone’s talking about New York is the second Silicon Valley. There’s more evidence of that in other cities: L.A., Seattle. And I think that’s an interesting example of where it was close enough that people fly back and forth all the time, they move, you can get people to move from Silicon Valley down to L.A. And at this point I’d say you have that whatever that is, that density of network and talent and capital and knowledge and everything else in L.A. And that happened relatively quickly.
Reid Hoffman [47:00] – Actually that part of the network extension, because when I think about how to build more Silicon Valleys it’s the connectivity of Silicon Valley, which is one of the things I think YC does so well and that’s actually a part of how you get it to spread.
Sam Altman [47:11] – For sure.
Reid Hoffman [47:13] – I’m going to go to lighting round, unless there’s anything else. Something that’s in you pocket, beside your phone?
Sam Altman [47:22] – Very often nothing, honestly. I have been trying to evolve down to just a phone. I still carry one credit card and a driver’s license sometimes, but very often nothing but a phone.
Reid Hoffman [47:35] – Artificial intelligence fills you with hope or dread? Pick one.
Sam Altman [47:39] – Hope.
Reid Hoffman [47:41] – Your favorite place to thing big?
Sam Altman [47:45] – I’m pretty happy on any hike. I don’t think big in an office very well, although sometimes I guess I look at the biggest ideas in the last few years, my office at home works… I think getting the feeling of a good office down is really important for good thought and I think I finally managed to do it at my house. And I think people get this wrong all the time. What you actually want is a small… This is a famous… I think it’s da Vinci, where what you actually want is a small office with really good natural light and comfortable chairs. I finally got that. But I’m a big hiker and I like being out by myself or with friends, but way away from any structures.
Reid Hoffman [48:39] – I’m just beginning, because we moved headquarters of LinkedIn, to condition the office I want and it’s going to have a lot of custom wood when I get there. It’s not there yet. When I do I’ll have you come over, but it’s to give it the tactile feel.
Sam Altman [48:53] – I saw what I thought recently was the perfect office, which was a Japanese tea house basically, with beautiful custom wood by itself in a forest: glass, wood, a couch, a table, that was it.
Reid Hoffman [49:13] – Perhaps post ranchette. What job would you take if you were out of work tomorrow.
Sam Altman [49:22] – I would just sleep for a while if I were out of work tomorrow, not think about work, but a few months later… Actually I would go study physics and then I would get some job related to that probably, I don’t know.
Reid Hoffman [49:40] – One object from your childhood that you could never throw away?
Sam Altman [49:47] – There’s a lot of those. Probably photographs, honestly. But I still have my original computer and the stuffed animal I slept with when I was a kid. I don’t think I could throw away those either.
Reid Hoffman [50:08] – What’s one question no one has ever asked you?
Sam Altman [50:19] – I don’t know if there’s an interesting one.
Reid Hoffman [50:23] – What’s the one outstanding talented job candidate that got away?
Sam Altman [50:29] – Someone that I was going to hire? In our line of work, it’s more like most of the people, the job candidates, are founders that we didn’t fund and that’s actually what keeps me up at night. It’s not even the famous ones. Not many times, but there have been a few times where we funded someone and they went on to do really well, but that’s not the interesting failure mode, to me. The interesting failure mode to me is people that we said no to that would have been the next Zuckerberg or Chesky or whatever and there are three or four people over the years that I have turned down at interviews that didn’t go on to do anything successful, but it still eats at me and I’m pretty sure I was wrong in a big way.
Reid Hoffman [51:19] – The one thing you wish your phone could do?
Sam Altman [51:27] – I wish there was sufficiently good AI on my phone that it would only interrupt me for things I needed to see and also that it would decide if reading something would make me more or less happy, which probably means it would never show me anything on Twitter. That would be an easy one, but there’s a lot other stuff where I think I would love it if it would curate what I saw better.
Reid Hoffman [51:51] – Favorite app?
Sam Altman [51:57] – Gmail probably. Actually the new Gmail app sucks. The old Gmail app, I would say.
Reid Hoffman [52:03] – All time favorite book?
Sam Altman [52:05] – Oh, that’s a hard one. I can’t pick one all time favorite book. My favorite book of the year is a book called Pandemonium. The subtitle’s something like the coming of the machine age from 1650 to 1850 and it was just what it was like when people who had never seen a machine before had to think about the future of automation. An incredible read.
Reid Hoffman [52:35] – Best movie ever?
Sam Altman [52:40] – I’m not a huge movie person. I love all Star Wars movies. I love Dark Knight. I love American Beauty, to pick something that is not a sci-fi movie. I’m not a huge movie watcher.
Reid Hoffman [52:56] – Messy desk or clean desk?
Sam Altman [52:59] – I don’t really do a desk. I work on a couch most of the time.
Reid Hoffman [53:02] – No desk. Is there a single poem or passage from literature that you’ve memorized?
Sam Altman [53:09] – I’ve memorized a lot of poems and a lot of literature. One image that I love is relevant to this conversation. There’s a poem called Ozymandias, by Shelley, and the image is one that I think I often share with founders of a certain type and it has been very effective in correcting bad behaviors so I will use that one. In the desert is the wreckage of a statue, this giant statue of a king, that is now fallen and it’s ruins and it’s all by itself, just the sand. And the inscription that’s still visible on the statue is, “My name is Ozymandias, king of kings: Look on my works, ye mighty, and despair!” And there are a lot of startup founders who, in the moment, fell like they’re the most important thing ever in the world, their companies going to be incredible. These are usually founders that have accomplished nothing that have this feeling of forever, I’m going to be the most important person, my company’s going to be the most important company and we’re going to be this huge success and I’m king of kings, look on my works in despair. What eventually happens to everyone is that you end up in ruins, but startups often do that pretty quickly and it’s a pretty quick change of fortunes. And I think keeping that image of any statement you make like that is going to someday be in a collapsed statue in a desert has been an effective mental image for startup founders that got way too arrogant.
Reid Hoffman [55:02] – That’s great. Actually I may use it too. The single greatest embarrassment of your career?
Sam Altman [55:08] – Oh, I can’t pick one. There’s so many. One of the things that is good and bad about investing in startups is it’s deeply humbling. People forget that every time you make an investment you do it with a belief that company is going to be successful or at least that you’re paying for positive expected value. And so many investments that I’ve been so confident about have utterly failed, so completely wrong. That doesn’t actually bother me that much. The painful one is the other kind of error where you’re like, “This company’s going to suck,” and then it goes on to be really successful. The most painful version of that of all for me is when I’m already an investor in the company, they ask me to put more money into a future round and I say the price is too high and I’m not going to do it and that happened many, many times. Those are the most embarrassing errors to me.
Reid Hoffman [56:07] – What are the techniques that you teach YC founders on triage, because one of the key things is to let certain fires burn while you’re solving others, because startups are inherently dead at the beginning? How do you go okay it’s totally fine to let those fires burn while you’re working on the other ones?
Sam Altman [56:33] – I like to draw the matrix of urgent, not urgent, important, not important. And everyone gets urgent, important first and not urgent, not important last, but people screw up the other two. I like to talk about that you have to let the urgent and not important things not happen or get someone else to do them. A lot of founders find someone on their team relatively early on, that they rely on to take care of that stuff, and never think about it. That’s one important frame of mind. I think another one is you only have to do really well with a few things to do well and you can do badly at a lot of things, but if you don’t do well at those few things, it doesn’t matter how you do everything else. What you don’t want to be is the founder that gets everything but building a great product perfect. You just have to ruthlessly prioritize is this going to do it or not and you want to be aware of the areas where people like get tricked; like going to conferences, going to Necker Island or something. These things end up being huge time wasters, but they’re fun and they superficially feel like important networking or something. We try to talk to people about here are the normal tricks founders fall for that are bad uses of time.
Reid Hoffman [58:12] – Great. Well thanks Sam. As always, a huge pleasure and thanks for talking with us.
Sam Altman [58:17] – Thank you.
Craig Cannon [58:19] – Alright, thanks for listening. As always, you can see the transcript at blog.ycombinator.com and if you’d like to hear more rom Masters of Scale you can find them at mastersofscale.com. See you next time.
Y Combinator created a new model for funding early stage startups. Twice a year we invest a small amount of money ($150k) in a large number of startups (recently 200). The startups move to Silicon