Early to Every Wave - Former Y Combinator President | Geoff Ralston

What makes Silicon Valley special? Why is it that so much innovation and venture-funded companies come out of there?

Part of the reason is the social construct of creativity and innovation being the height of experience and career potential. It's different from other places where starting a company might be seen as a failure to get a job. In Silicon Valley, if you say you're doing a startup, the reaction is, "Awesome, what are you doing?"

Do you know the root cause of this culture?

Root causes are harder to discern, but it probably comes from a combination of success stories and role models. Maybe it's something in the water or the sunshine in California, but it's a mix of things.

How did venture capital start in Silicon Valley?

It was kicked off by an East Coast company called Advanced Research Development Corporation, started by Georges Doriot. He funded Digital Equipment Corporation, turning $70,000 into $24 million. That success story helped create Sand Hill Road and the idea of limitless potential.

How do you understand the right timing for what to work on?

Timing is crucial for a startup's success. You can have a brilliant idea and execute well, but if your timing is wrong, you might fail. It's often guesswork, intuition, and luck.

What was your inspiration for RocketMail?

I believed in the importance of connecting computers. I was inspired by the idea of a connected world from science fiction. When I saw the World Wide Web, I knew it was the future and left my job to pursue it.

How did you decide to sell your company to Yahoo?

It's about maximizing value and personal readiness. With RocketMail, there was a belief that Yahoo would provide a bigger platform for growth, even though I initially thought we had a great future on our own.

What was your first month at Yahoo like?

It was a shock moving from a 30-person startup to a 300-person company. It took time to get used to the structure and having a boss.

How was the growth at Yahoo?

The growth was explosive. We were getting tens of thousands of new users daily. It was amazing until you compare it to today's growth rates, like ChatGPT reaching a billion users in two years.

How did you manage such fast growth?

Managing growth involves navigating organizational structures and learning from the people around you. It's about hiring people who are better than you in certain areas and learning from them.

How do you manage large teams without ending up in bureaucracy?

Bureaucracy is necessary to manage large organizations. It's about creating social structures to organize people. You learn by trial and error and from the expertise within the organization.

How do you hire people better than you?

You can tell in conversations if someone has insights you hadn't considered. A good interview reveals what it's like to work with someone and whether they'll add value beyond what you or your team can do.

What was the story behind Lala's acquisition by Apple?

We believed our future was limited and sought an acquisition. We had a deal with Facebook that fell apart, which was a huge blow. We reconstituted the company and eventually maneuvered to find a good acquisition deal with Apple.

Were you worried about the fallout from the Facebook deal?

I don't think it was well-known, and we managed to do smaller deals with Facebook and Google that helped us regain some momentum.

How did Imagine K12 merge with Y Combinator?

After Lala, I wanted to do something in education. Tim Brady and I started Imagine K12, inspired by YC's model. PG supported us and suggested I become a YC partner, which helped us learn how to run an accelerator.

What do you teach founders as a YC partner?

I help them understand their startup's purpose and instill belief in their vision. It's about reinforcing their belief in their startup, which is crucial for success.

What was your focus as president of YC?

I aimed to simplify operations and make YC more professional to ensure its longevity. When COVID hit, we had to adapt to remote operations, which was a significant challenge.

I remember you pointing towards having a thousand startups per batch. I'm just wondering, like-

And we're not that far now. I mean, we got to close to 800 startups a year and then pulled back somewhat. Cause I do think we grew probably a little bit faster than-

How did you know that? Like, how did you know that you grew a little bit faster?

Faster than we ought to have. I think there was a discomfort amongst the partnership about how many companies we had and our ability to focus effectively on helping each and every one of those startups be successful. So we pulled back, but not that much. Now, today YC funds something like 500 startups a year on that order of magnitude. And that feels like, I think it's a pretty good level. And I couldn't speak for Gary, but I would suspect that if Gary could grow that, he would. And maybe he will over time. He's expanded the number of batches from two to four. So it's spread out a little more evenly during the year. And I think that's been effective.

What is the ceiling of that? Is there some sort of-

What's the ceiling of great founders, of the number of great founders in the world? I don't know. Especially with the emergence of AI as a transformative technology that it will be, that more and more people will find entrepreneurship as a career path that's successful. So I don't know what the ceiling of that is. I hope it's higher. I hope it's a lot higher. I think there's amazing entrepreneurs in every ecosystem, in every geography around the world. And I don't see why YC couldn't fund a thousand startups a year and more.

And choosing the next person, like got to join after you and so on. Like, how can you make this judgment?

How can you make the judgment about who is going to join later? How do you know what-

How do you do succession?

Exactly. It's hard. A lot of people blow it in succession. I think you try to set your criteria as carefully as you can. Think through your criteria as carefully as you can. At an organization like YC, you try to get as much consensus on those criteria as you can. And then, you know, do your best to find the right people. Sometimes you get lucky and sometimes you don't, to find the people that you need. Succession is a super hard problem. I wouldn't say it's solved. Sometimes it works great and sometimes it's really painful. And we've seen that in lots of different companies. Look at Apple, right? Steve tried to arrange succession for himself and that worked out terribly, so he had to come back. And you could argue with Tim Cook, it's worked out pretty well. But he learned a lot by then, by failing. That's probably the best way to learn anything is by failing. One of the reasons Silicon Valley is what it is, is that we have this culture of failing up, right? Of learning from your failure and then succeeding later on.

Is there something you recall from Steve Jobs that you learned from him, you mentioned earlier?

He impressed upon me, you know, it's super funny because I just saw a quote from, it was from somebody that Gary reiterated, but it's one of the things I remember most vividly from Steve. When the acquisition happened, we were invited over to Steve's house to sort of get his stamp of approval on the acquisition that Eddie Q, who ran Apple Music at the time, he wanted to do this acquisition, so he brought us in front of Steve. And a lot of the meeting was just Steve telling us stories about Apple and how Apple ran. And he said this super interesting thing. He said, you know, when I came back from Next and I was, I took over Apple again, I looked at Apple and I had this realization. We had $5 billion in sales, which seemed like an awesome number to me. The problem was we had $6 billion of expenses. And that seemed like a problem, but I said to myself, all I have to do is find a billion dollars of expenses that we didn't need to have, maintain that $5 billion and maybe grow it. And all of a sudden then we'd have degrees of freedom to do what we wanted to do. So all I had to do was to say no to a whole bunch of things. And there was just this dialogue that I, this is just like the last day or so I saw there was dialogue with people saying like, the key thing to do is to know when to say no. That was like Steve Jobs' superpower. And it's true. And Gary said, oh, well, yeah, cause I'm like a yes and person, but it is true that one of the key components of success is to know when to say no.

I used to do office hours with, while I was president of YC, with founders that I would call my priority office hours. But what they were really about was trying to predict the future. I'd say, look, roll forward five years when you're a successful billion dollar corporation, you will have made a bunch of decisions amongst the priorities you have today. You all have ordered those priorities and made those decisions that will have been substantive until you becoming this billion dollar company. And your trick today is to figure out which of those decisions they are, which priorities to make, what to say yes to, but more importantly, what to say no to. Cause you know, as a startup founder, you have every day an infinite number of things you can do. I don't even know why you're taking time to do this podcast because you should be working on your startup. And you know that in the back of your head, you have a million things to do and you have to decide where you spend your time, what to do, but equally, well, you have to decide what not to do. And so learning how to say no, I think maybe that's like the thing that I recall the most from everything that I spent time with Steve on.

During the era of Sam Altman, you probably work with him closely. You saw everything about AI from the first seat and you decided to do safe. Can you walk me through like, what are you foreseeing and safety in AI is so important?

I'm not the first to say this and I won't be the last, but AI is part of the world. Perhaps the most transformative technology mankind has ever created, maybe will ever create. It's a fundamental change in the way we are all going to interact with the world. This incredibly powerful technology is being rolled out as we speak in almost every domain imaginable because if you think about it, what AI represents is cognition, is thought, is intelligence and in almost every human domain, adding intelligence adds value, at least potentially. But like any super powerful technology, there's dual use. You can use it for great good, amazing good. The promise of AI is incredible or you can use it in ways that all of us, I think, would consider bad, evil. The peril of AI is equally real and it's imperative to me, for all of us, for my children, for everyone's children, that we build a world where they can be flourishing members of the human project, the human project. And to do that, I think we need guardrails. We need to build an AI future that's a safe and secure, some would say resilient future where humanity can be resilient to that future. I don't think there's anything more important that I or I would argue anyone else can work on today because I think this is a precarious time because we don't really understand exactly how this technology is going to impact the human project. We just know for sure it's gonna have a massive impact. So I hope that we can do it as safely as possible and so I'm doing my own little part, I hope, to make this infiltration into society of intelligence beyond what we could have imagined even a few years ago as beneficial as possible.

Are there any practical areas you can point on of how we can basically control our intelligence and not become nuclear that we don't even can comprehend where this can go?

The control problem's a real one. What many people worry about is what happens if we create an intelligence that's so much more intelligent than us that we lose control of it. And there are arguments up and down the spectrum of how bad that could get or how realistic that scenario is. I have tended to focus on more practical concerns. Some of them I think are rather mundane in the sense that they don't necessarily enter the realm of catastrophic. Finding ways for agents to be successfully deployed within corporations so they don't delete all the corporation's data. That's really important to the corporation but it's probably not a catastrophic threat to the human project.

Just a small parenthesis there. We've been facing this problem in Terra and actually building bureaucracy was one of the solutions. As in layer of agents.

Agent bureaucracy, yeah, 100% yes. And building bureaucracy, building a structural compliance that these agents have to live with. That's a smart thing and it's great and we should do that, we should build those guardrails. But there's a set of concerns that are not mundane but only connect to the loss of control problem we were discussing a second ago where the super powerful technology enables bad actors to do really bad things. The area I'm most focused on right now is biosecurity. We mentioned COVID. We just went through a pandemic that was really quite mild by some measures. The mortality rate was pretty low. So if you got COVID, your chance of dying was pretty low. And that seemed like it was a sort of a good dry run. However, this mild pandemic killed somewhere between seven and 30 million people depending on who you believe. Millions of people lost their lives due to this quote unquote mild pandemic. And it cost the world trillions of dollars, maybe as many as $14 trillion in lost productivity. We know what happens when you have a global pandemic. AI will enable the creation of dangerous pathogens that could create a much worse pandemic and it could do it soon. We know what we need to do to build a worldwide, think of it as a worldwide immune system to help protect us against future pandemics, whether they be natural or AI enabled. And I think it's urgent that we get on that now. And I'm doing what I can to help create that worldwide immune system or work with the people who can create that immune system to make us all safer all the time from dangerous pathogens.

Have you seen any interesting startups in this area?

Yes, I have.

Give me some examples. Like how are people approaching this?

There's a few examples. I'll only name a couple. Some of them aren't completely public yet, but there's a startup I know well that is trying to build what some people call bio radar, which one of the important things that you wanna have happen when a dangerous pathogen is let loose is to detect it as early as possible. Because once you detect it, you can isolate it and hopefully prevent it from spreading to the rest of the world. This is super important. And there's startups working on detection tools that go way beyond what we have today. There's startups who are working on filtering tools to make it less likely that with the aid of AI, you could synthesize DNA to create a very dangerous pathogen. There are startups working on cures. If you are infected with a dangerous pathogen that will help you fight off that pathogen and survive and maybe stop it from spreading. Across all of those different domains, there are different startups working today.

I think we have five minutes. I don't want to keep you too much, too long. I don't know if you, fast questions here to ask. What's your opinion on Peter Thiel's competition is for losers?

He was stating a point, making a point, which is mostly irrelevant to most startup founders most of the time. It was really a point to venture capitalists, which is to say, if you can find in his terminology, a startup that goes from zero to one, where you can like, where startup, where competition becomes irrelevant, that's how you make the real money. Like investing as he did early on in Facebook, where Facebook actually made it so there was no real competition for Facebook. That's the best thing. As a startup founder, there's a different set of attitudes you have towards competition. One, the first attitude you should have as a startup founder is to ignore your competition. There's an old YC saying that goes that startups die from suicide, not murder. If a startup is going to fail, it's usually because of internal stuff that happens where they kill themselves. So don't worry about it at first. Later, as the company matures, of course you worry about competition. You don't say competition is for losers. It's like, oh my God, these people are competing with me. And in almost every market, they will compete with you. And you're going to try to out-compete with them. Now, if you can find yourself in a position where the competition is irrelevant, where you've surpassed that, good for you. But in most domains, most successful companies have competition. Even Amazon has some competition, right? You could argue maybe not that much with AWS, but selling goods, they have lots of competition. All the car companies, Tesla has lots of competition, maybe more than they like. Every humanoid robot company, every AI company, there's competition, there's always competition. He was just making a point that the best companies to invest in are the ones where there's, in the end, no competition. So yeah, sure, competition is for losers. Or more of the point, he should really have said it the other way around, which is for the winners, the ultimate winners, the biggest winners, there's no competition.

Marc Andreessen got this lately, saying that retrospection is useless. What is your take?

I think that was embarrassing. I don't know why he said that. I suspect that Marc is a very retrospective, introspective person. Again, I think sometimes these people are posturing and making points that aren't real. He's probably read Marcus Aurelius and he knows that it didn't come about in the 20th century. Sometimes I think he's putting out a narrative that he doesn't even believe himself because he knows that's not true and he knows how important introspection is. And I think the best founders, the best builders are thinking about why they're building things. Brian Chesky used to give a talk, the founder of Airbnb, to every YC class. You probably saw it where he says, do good for the world. How do you do good for the world unless you're introspective about what you're doing and why you're doing it? Or looking at retrospectively about, oh, am I like a tobacco company killing a lot of people even though I'm making a lot of money because I'm addicting them to my product? Is that what I wanna do? Is that what I wanna create? Is that what I wanna leave during my short time on this planet? Is that the legacy I want to create? And most of us care about what the legacy we create. Again, we're here for a short time and it's not all about the maximum amount of pleasure that we can get for ourselves, it's we're thinking about our legacy, whether it's our children or the works we do here. And the best people, the people I look up to the most care very deeply about precisely what legacy they're leaving and what sort of positive impact they're having on the world.

What did you try to teach to your children?

How many seconds left to answer that question? Let me not answer so much what I tried to teach them. My hope for my children was that they would end up being people I admired. Right? My children are all adults now and I admire every single one of them. And it's because they're all good people, they're kind and they're thoughtful and they care about what's important in the world and not what's unimportant. And however you teach that, if you have accomplished that with your, in some ways it's your greatest legacy, well, then you've done a good thing for the world because those are the kind of people that add value to the world in whatever they do.

I don't think we can end with a better answer, so I'm not going to ask you any other questions. That's been incredible. Thank you so much.

Fantastic. Nice talking to you, Kyriakos. Super.