Introduction

Patrick: [00:02:39] My guest today is Daniel Gross. Daniel is the founder of Pioneer, an extremely unique company, which he describes as a fully remote startup generator that helps talented people around the world figure out if their idea has legs. You can learn more about it at pioneer.app. Our wide-ranging conversation covers the art of asking great questions, the use of predictive modeling and psychometrics to identify talent, and why psychometrics are probably overrated and not that scientific. We then dive into exciting new frontiers for tech investing, ranging from GPT-3 to satellites. I really enjoyed this conversation and hope you will, too.

Movies as an Icebreaker

Patrick: [00:03:15] So Daniel, I don't think I've ever started the podcast by asking about movies, but there's two that you and I were talking about earlier that I thought would be a fun place to start. Those two are Crank, starring Jason Statham, and Whiplash, which is one of my all-time favorites. Why do you care about these two movies?

Daniel: [00:03:30] I love those two movies. I find them very useful as a screening tactic as well. Whiplash, to those maybe who haven't seen it, is a story of a musician, a drummer in particular, with a talented coach who gets pushed a little bit too much. I very much love that story. I very much empathize with it, the story of the insecure overachiever falling almost too much in love with the vehicle of his performance. You could cut it many ways, whether it's an inspirational or cautionary tale. But what I find interesting is there's a lot of questions you can't really ask people directly when you're talking to them in interviews, or social settings, or whatever. And I find their opinion on movies to be something that lets you have that conversation in a fairly coded way.

It's so interesting. I find it relaxes the conversation a little bit. Even as we're talking about movies now, it feels much less tense than a conversation about, I don't know, the top-performing stocks or something. So there's an element of play and novelty in using it in the interview.

Patrick: [00:04:22] What do you think that tests in people? So, the Whiplash is a great litmus test, and you could see how it would rub some people the exact wrong way, and others, like you said, see themselves in it or something. What is that line that's being played with there?

Daniel: [00:04:37] The Whiplash question is actually one I have used quite a bit in interviews. I find it separates people, I think, who have an adversarial relationship with their work, to maybe an inspirational relationship with their work. I understand if you have an adversarial relationship, in most jobs actually. I think we're both extremely lucky to have our jobs be our passions. But I think a lot of people, in Silicon Valley, in particular over time, developed very much give-and-take relationship with their work, where they want to do the minimum amount possible, such that they can return to life.

I think that's fine for many people. I tend to very much enjoy the presence and company of just a different style of person, one who very much would use the work as way to perform as something inspiring, as something to do more of. There's an element in Whiplash, almost a worshiping of achievement in that sense. Now, like everything, there's probably a dose-response curve, like a hormone for work, and anything can go too far. But I feel like it very quickly filters out someone who says, "You wouldn't want to take anything too seriously," to someone who says, "Gosh, I hate to say it but that was great."

I don't know what causes that split in people. That's of course, a very interesting thing, assuming they are working at a job that's a creative pursuit. We can construe two different styles of jobs, maybe jobs where the delta between the median and the tail is maybe 10% or 20%, so like a cab driver. Best taxi driver in New York City is going to be 15% better than the median. But in a creative pursuit, or in investing, or software engineering, the tail is literally 10 times...John Carmack is 10 times better an engineer than myself. In those situations, I think you really want to be looking out for people that are really interested in the job, and do work as something that they almost have to meter and dose out for themselves, as opposed to something that they're careful not to do too much of, so-to-speak.

Screening for Talent

Patrick: [00:06:15] You've laid an amazing transition to one of the topics that I want to cover at the end of your story, which is Pioneer. But it's a perfect excuse to talk about what you've learned from screening talent. Be helpful for you to first set some context, and describe what Pioneer is, what it does. And then I'm going to ask all sorts of questions about parsing through talented people.

Daniel: [00:06:35] Pioneer is a company that I run. It's kind of a cross between the two different careers that I've had over the course of my life. I initially started a search engine that got acquired by Apple, later on became an investor in a bunch of different startups and worked at a startup investment firm. Pioneer is kind of the culmination of those two things. It's part search engine, part venture investment firm. What we're trying to do is, we're trying to use the power of internet to find people around the world that are potentially interested in starting companies. We're almost try to, through software, both qualify them and figure out if they're good or not, and also, almost convince them to start companies, even create, so to speak, more founders, the same way you see Peloton creating more cyclists or Strava creating more runners. I've had the pleasure over the course of my career, both through Pioneer and through others, to invest in a bunch of different companies. I've had the chance to meet a bunch of different founders. And over the course of my time, having maybe been one.

I'm very interested in this topic. I think it's an interesting question for any asset allocator to what extent you believe, especially at the early stage, that you want to be looking for great markets or great founders. And like everything, the truth is somewhere in between. We can accept, maybe as an axiom, that being able to screen people, either using software or using your own mind, is an important thing, not just for investors, of course, but for anyone looking to hire, or really do anything significant in life. I find the question of talent to actually be the most important question in the world because it is, at the end of the day, required to do anything interesting.

Patrick: [00:07:56] As somebody that's done a fair amount of predictive modeling, I know from experience that often, the most important thing is defining the outcome very cleanly and very well, before testing certain variables or features against that outcome. I'd love to hear how you define talent. If the idea is to search for talented people, how do you know that you found one in some sort of modeling exercise?

Daniel: [00:08:17] Metrics for success is a fascinating problem, because at the end of your day, you're planting seeds. And it'll take a decade or so for it to grow into its tree. We have these moments at Pioneer where we're not quite sure, are we successful or not? Companies have, at the end of the day, a very short-term feedback loop to chase, which is revenue, which, obviously, can be abused, but I actually think is a general wonderful incentive structure for progress in society. For us, it's obviously less clear.

And there's metrics we can look at to figure out if a person is good. One thing Pioneer looks at is the fraction of folks we find that get funded by others, which is, right now, around 20%, 25% or so. So follow-on funding is something you can look at. But of course, WeWork, or the Vision Fund has taught us that, that alone isn't good enough. One thing I'm always on the lookout for with people. Are they moving quickly? Are they evolving quickly? In my view, at the end of the day, when you're betting on a founder, especially if the market's not super clear, what you're trying to do is, you're trying to predict that, that particular person will have as many shots on goal as possible, because occasionally, companies get formed and it's akin to someone sticking a toothpick in the sand in Saudi Arabia, and oil comes gushing out. Product-market fit on day three. Great, done. But more often than not, there's a tale of endurance there. What you want is the person who's going to plunge as many toothpicks into the ground as possible, who's going to take as many shots on goal as possible.

That's why I think derivative is pretty important because it's one thing to have energy. If you're searching through random space, you could actually spend a lot of time pivoting your company before you get somewhere. But if there's a little bit of intelligence there, the reinforcement-learning algorithm at the end of the day will end up making compounding better and better decisions.

What you want is the person who's going to plunge as many toothpicks into the ground as possible, who's going to take as many shots on goal as possible.

Investors often think about the magic of compounding and how hidden that is from most people. I think it's interesting to think of compounding, not just in CAGR, but in a person. That, to me, is one of the simplest things you can try to measure on a short term, to give you a sense of whether the person, company, or project are going somewhere long-term.

Patrick: [00:10:01] One of the things you mentioned, the blending of search engine and investment fund, that's so interesting. One of the mental models here would be something like page rank in search, right? Google's original competitive advantage where they base the quality of a site on, I guess, how often it was referred to or linked to by other sites. Is there some equivalent in talent for page rank? Is that a useful analogy for thinking about sorting through people?

Daniel: [00:10:25] This takes us down to the very interesting world of psychometrics and all sorts of ways we can try to use, to get a sense of whether a person is good or not. This would be remarkably important if it was possible, because we would find a way to highlight people that have been overlooked by every other possible system.

One thing I'm very tickled by is, Pioneer, as a business model, is economically incentivized to find people that have been overlooked by anyone else. That is to say the terms that we offer folks, once our system grades them as great, are not terms that a Stanford undergrad should take or accept. Sequoia will offer them better terms. But we're very much going after these underrated assets. So, if you could find a way to figure out if these folks are good or not, in software or mostly software, that would be great. But part of the issue is, though, that anytime you quantify anything the world, you are taking high-level dimensional space and you're reducing it to a lower-level dimension. And you're going to lose data.

Question is whether your lossy encoding technique is truly going to MP3, where at the end of the day, you don't notice the difference, or is it something more aggressive? Are you losing too many pixels or too many bytes along the process? In an attempt to reduce a person to a number, have you lost the magic? So to date, Pioneer's taken the approach of almost a, how would you say, a self-driving car algorithm, where we still use humans to occasionally guide the system and reinforce the data set. Nothing works better than really trying to evaluate the person based on the progress that they make.

So when you do sign up, you get a score, and that score is how good you and your project is. They're merged. So on the one hand, it's half page rank. On the other hand, it's a little bit half power score on a bike, in a sense that you can actively work to drive that score better, and you can see the feedback result. You're very eager to see that number go up week over week. Now again, like any maximization function, people will maximize it whether it's societally good or not. I really view founders as excellent hierarchy maximizers and terrible hierarchy selectors.

So people don't really know what they're doing, but they just want to drive the score that they have week over week. It's pretty much built like a video game in that sense where, the more progress you make on your project, the higher your score goes. And like page rank, there's a whole composite function of what goes into that score. It could be things like growing your revenue. It could be things like what other peers who are in Pioneer think about the progress that you've made this particular week, or a myriad of other tasks.

Our goal is to really shove more and more into this infinite portrait of what progress and accompanying an individual look like, such that, at the end of the day, we're fully able to quantify it. So, I actually view the resolution that we're currently using to understand someone to be fairly low. Maybe, it's a 16-bit bitmap. At some point, there's going to be a whole 4K addition of Pioneer, where there's thousands of different metrics you can pump into the system, Twitter followers that you have and, I don't know, YouTube views on your video, and revenue growth this particular month, revenue growth benchmarked against other companies, analysis of the way your company interacts over Slack.

As much as we can get to really not just quantify who's good and who's not, but also motivate people to get better. I think if we're able to build a reliable metric for day-to-day progress, it will not only be a great service for creating more startups, but I think it'll be a great thing for productivity, if you don't prove what you don't measure. I think part of the issue is, by the way, not just for Pioneer, but I think for anyone working on something early-stage or an investor is, it's not really clear at the end of the day what separates a good day from a bad day. There's no score.

We know that in markets and industries where there is a score, people do improve, and markets are efficient. For example, in sales, it's very clear what a good day is, it's very clear what a bad day is, and people are motivated to improve. You can measure who's great and who's not. If there's someone junior on the team who's really good, they're rewarded based on the merits of their work. In sports, this is extremely clear as well, a good day and a bad day. It's not really clear if you're an investor, what a good day or a bad day is. It's not really clear if you're an early-stage startup what a good day or bad day is. I mean, even for a software engineer, it's not really clear what a good day or a bad day is.

Patrick: [00:14:09] I'm curious to what degree still, psychometrics, and maybe you can define what that term means, what are some examples of psychometrics? What things you found that matter or are at least interesting to you, having explored all those avenues?

Daniel: [00:14:23] Yeah, psychometrics is a very interesting topic. At a high level, it is the attempt to use various mechanics, mostly self-assessment mechanisms, to get a sense of the individual, and to be able to bucket them into different categories. This, again, I think could be very easily used for evil, but for the most part, I think, stands to be a force for good, because it allows you to screen people without necessarily being subjected to the person's gender, race, language restriction, whatever.

Over the course of a couple of decades, centuries, a bunch of different psychometric techniques have been discussed. Myers-Briggs is a tremendously popular one that, in my personal very humble opinion, is horribly inaccurate and doesn't regress to anything really material. The Big Five Aspect Scale is currently viewed as the popular psychometric toolkit or framework. But I should mention, before we go any further, that psychometrics, I think, in general, there's two interesting things to observe about them, about the whole concept of them.

First is that, once someone is exposed to the concept of psychometrics, there's a period of infatuation where they tend to overuse it and assume that it is the decoder ring for life. It does try to take a very high-dimensional space, one of the highest dimensions we know, which is the individual, and reduce it. So you get really excited because you think, "Gosh, the Big Five Aspect Scale, I now understand everything." Over time, as you use it, it's funny. My sister's a psychologist. I remember I first came to her many years ago, wild eyed about the Big Five. She said ... I forgot ... I think she called it something like the Big Five Halo, and said, "Call me in a year. It'll fade a little bit." And she was very right. Someone listening here, I think you can do quite a bit of researching in Google, and maybe we'll send around some links after this. But you should realize, at the end of the day, that there's a modest level of efficacy here.

And then the second thing is, why it is not more efficient, and why the science here isn't better, I think, is a very interesting reason. Like epidemiology, I think, before the coronavirus, there were no free market stressors that were really applied to the validity of psychometrics. In my humble opinion, if there are no free market stressors, the science is false. So, I think we're realizing a lot of epidemiology and modeling viruses, those people don't know what they're talking about. Now, I don't blame them because capitalism, I find, is gravity that grounds things, that pulls them back to reality, as opposed to the ether of space and ideas.

A counterpoint here would be weather forecasting. We know its limitations very accurately. That is because it is grounded in the free market. So psychometrics, with the exception of the military, maybe air traffic control, which happens to use IQ tests, and two, three other organizations, it's not really grounded in reality. Which is one of the reasons why I think Pioneer has been able to make remarkable progress in the field. Not because we're that good, but because it’s very nascent days, because for the most part, the effect sizes in the world of psychometrics, in the Big Five, are extremely light. The biggest effect size that people are excited about, it is true that, I think that R-squared of IQ to earnings is .6 or something, but you're still not explaining a huge amount of variants at that point.

But that is the most celebrated effect size. So, I think we should be careful when trying to use this. The final thing I should say is, I think a lot of these things are useful because of the language that they give people. And it is useful, I think, after you would meet someone, or if you're in your team and you're working, and you're trying to discuss someone after the interview, it is useful to have common vernacular and taxonomy about how you describe someone.

To that extent, Myers-Briggs may be helpful if you all agree on that language. It doesn't matter that actual test of Myers-Briggs is valid or not. One reason, by the way, why a lot of these tests are just wrong is, they're based on self-assessment. Self-assessment is such a crazy thing. We're going to ask the person to evaluate themselves at one point in time. Call me at 8:00 in the morning after a cup of coffee and I'll give you a very different assessment from 2:00 in the morning about myself.

Patrick: [00:18:11] It seems like the quality of persistence becomes incredibly important in all this, especially when you're measuring literal progress in the variety of different ways, whether that's revenue or some other thing further upstream from revenue inside of a person's project or company. Do you think that persistence is, obviously you need some natural aptitude and probably some curiosity, but do you think ultimately persistence might be the most simple way of talking about all of this?

Daniel: [00:18:35] Persistence, power, or energy, are words we frequently use. I think that's important because you need to ensure the person's going to take many shots on goal. It's just your sense of how formidable they are to you. This may be a question you can ask yourself that reflects that on them. I think Paul Graham originally, Y Combinator, originally said that was one thing he looked for in founders is, he wants to walk away from the meeting feeling a little bit afraid. The other thing going on, I think, with persistence or energy is, the person is very interested in improving a particular dimension in their company or what have you, because they're performing at the end of the day for someone in particular.

And I think that can give someone a very elongated sense of endurance or persistence. Often, you'll meet founders who really look up to other founders or other leaders, if they get stuck in the right feedback loop, they will find themselves very much performing for that individual in a way that's constructive for society. There's also a destructive mechanism of this, where people spend a lot of time on Twitter, trying to get Elon to favorite one of their tweets, and dopamine hit from that becomes a thing that they chase. But that's, in my opinion, not that useful for the world, obviously. Like I mentioned, founders are very bad at selecting hierarchies to maximize and very good at maximizing whatever game that they're in. And maximizing status, I think, is a useless hierarchy because it's infinite. Capital is a great one to maximize, precisely because it's scarce. So you could meet a founder who's going to spend all day trying to get any of your previous guests to retweet their tweet, and that will be their entire game in life, that when they wake up in the morning, that's who they're performing for, and that's what they're really trying to do. But I find the best founders view the company, the revenue, as the vehicle through which they're performing. Obviously, I think that's much more healthy because for someone to click like on your tweet requires far less effort than parting with something of scarce value, like giving you money for your product.

Patrick: [00:20:22] You mentioned earlier, a term, and this is my last question on personality side. You mentioned the term that I think our friend, Graham, was the first one to introduce me to, which was insecure-overachiever. Everything you just said makes me think of insecurity and how that gets filled. Do you think, in your experience, with all these different pioneers, that insecurity is a good thing or bad thing?

Daniel: [00:20:40] I definitely think it removes the foundation from the building, and it creates a bit more of a shaky atmosphere. I think it can create a lot of energy in someone, precisely because, like a molecule searching for covalent bond, they're going to be looking for someone else to perform for, because they have that sense of insecurity and they don't know where they sit in the hierarchy. And they will need to find a way to validate their position. Be it in a dinner party, or online, or with their company. So that, I think, is often what creates, what propels a lot of that energy. Look, I say this as someone who's slightly insecure myself.

And I think I've got to a little bit better at this as I've aged. Actually was commenting to our friend, Graham, that I think one transition that one goes through in life, certainly that I've observed recently is, by developing, in some cases a family, in my particular case, a very small set of very close relationships, some of that insecurity does tend to go away because, what happens is, you find your place in your world.

I definitely think it removes the foundation from the building, and it creates a bit more of a shaky atmosphere. I think it can create a lot of energy in someone, precisely because, like a molecule searching for covalent bond, they're going to be looking for someone else to perform for, because they have that sense of insecurity and they don't know where they sit in the hierarchy. And they will need to find a way to validate their position.

And you find a hierarchy that you feel like you can sit in comfortably, regardless of the actions that you take. But often, I actually think this is one of the hidden reasons why immigration works, it is because you take a person out of a hierarchy from country, voluntarily, and you shove them into a new hierarchy, where they have no idea where they stand. I think that creates a lot of insecurities. Speak to anyone who came to San Francisco, who immigrated here. Could be from Wisconsin, could be from India, it doesn't matter. They have no friends, and so they don't know where they sit in the world, because they're in a brand-new place.

In many ways, I think the story of immigration is the story of someone who's fairly insecure trying to secure their place in the hierarchy. I think that can create greatness. That feedback loop can cut both ways. You can spend a lot of time, as I mentioned earlier, maximizing stupid things. I think the interesting question here, the interesting question is, whether you would wish that thing on your children, because I think there's something about the Whiplash story, the insecure-overachiever story, which is very good for society, very thrilling for the individual, but there definitely are ups and downs. I actually, although I feel somewhat blessed to have this dynamic myself, I feel like it gives me a level of immersion and flow that I couldn't access, I think, if I was just more confident in my place in the world. I do wonder if that's something that occasionally does cause discomfort.

Patrick: [00:22:47] You mentioned capital as an interesting thing to be optimizing around because of its inherent scarcity. Can you walk us through this concept you have of fast-twitch versus slow-twitch capitalism?

Daniel: [00:22:57] It's funny. I speak to a lot of people who are trying to figure out what career to take, especially people on the younger side in life, who are trying to figure out if they go into investing or operating a company or working at a company. I feel like there's so much capital sloshing around, there's a lot of seed funds being created.

I mean, I think the only thing that would me more common than starting a seed fund is downloading TikTok. There's a lot of people stepping into venture that, in my personal opinions, I don't think will actually be happy as venture managers because venture is kind of this form of slow-twitch capitalism, where the amount of time it takes for you to get the reward. We were talking earlier about early ways of figuring out if you're making progress or not. It’s very unclear in venture what you're doing for a very long time. And it takes a while to figure out if you're decent or not. There's a lot of people I'm noticing that are mistaking activity for progress. And the way you see this is writing a lot of investments into a lot of different companies, I think, in many ways, to fill a void inside themselves, just to get a sense that they're advancing in the world.

But of course, as we know, small checks into a lot of different companies doesn't really create for a great result. You want to be in an enduring position on the cap table. In a small set of companies is far better. Concentration is a goal. It is not a risk. I think that's because these are people that would be much better suited to fast-twitched capitalism, which is, I think, a company where there's things changing every day. There's revenue to be chasing, deals to be had, progress to be made, as opposed to slow-twitch, metabolism could be your other metaphor here, which just takes longer. And I think you have to be slightly more self-assured to do it.

Patrick: [00:24:30] You mentioned memes, a fascinating topic, and a huge driver of human behavior. What is your thinking here on their importance there, tend to be funny things online? But the funny things we see demonstrate how much a lot of what we do is just copying other people. What's your thinking on memes?

Daniel: [00:24:45] It's very important, I think, to understand in the world, for any kind of asset allocator or someone running a business, because I think it helps explain a lot of human behavior. I think dissecting them is very interesting, and I think, for every particular meme, you may want to think about, what is the hole in the world that this thing is filling? And what it possible? Even something as, why has Matcha become popular? Something I've been thinking about lately. What whole void in world is the Matcha latte filling? It's interesting to think about because this wasn't as much as a fad 10 years again, and now it's certainly an explosive meme.

I think a lot of it, interestingly, is a byproduct of Instagram. I think Instagram, once Instagram got popular, it created a shortage in the world of colorful experiences that were purchasable at a fairly low rate. What is something I can buy that's colorful, that will pop, in the Instagram filter? And I think, as a result, you started to see a lot of dynamic contrast in colors. Purple drinks, green drinks, orange drinks. The Matcha latte filled that void. When you walk around the street and you see people order their Matcha latte, you realize what they're actually doing is, they're somewhat following this meme, I think, which wouldn't've gotten popular without Instagram.

So being able to dissect these things, I understand, I think, the underlying systems behind why humans do really anything, I find super useful. I feel like sometimes I'm walking around the world with this little heads-up display when I'm able to think this way, and really understand the reason behind the reason.

Themes in Businesses Being Built

Patrick: [00:26:12] With all of this as background and backdrops, we got the discussion around Pioneer and urging people to become entrepreneurs, maybe even convincing ones that wouldn't have otherwise to do so. And also, this trend of lots of capital in the system. Lots of seed funds, or really, just lots of risk-seeking capital with rates so low. And I guess easier probably to get funded as an entrepreneur than ever before. With those two things in mind, what are you seeing as interesting as themes in the kinds of businesses being founded today? What is the frontier, if you will, of companies being formed?

Daniel: [00:26:47] Yeah. What I'll say before I get in is, what I find so interesting is how rapidly this scene evolves. Founders are these migrant species, kind of a wildebeest moving from savanna to savanna for whatever market is more interesting. One of the reasons why I think the study of memetics is so important is, that can help you predict where things are going to go to.

One thing that's quite popular now is, there's a lot of people working on various chat aggregation services. Matrix is an open-source chat ecosystem. As part of the ecosystem, there's all sorts of people that have built adapters, things that let you slurp in, WhatsApp, Signal, whatever messaging service you use to this federated service. And that's opened the door for, I think, half a dozen or so startups that are experimenting with bringing back AIM and ICQ. So that's one fairly active field. The other one that kind of adjacent, that's pretty active now has been interesting to watch enterprise RPA, Robotic Process Automation, which was certainly a celebrated category of 2018 and 2019, fade away, in replacement for search.

Another thing that's become fairly popular recently is, the whole ecosystem around Robinhood. Obviously, everyone kind of becomes a trigger now, but there's a lot of people building the picks and shovels for this, Bloomberg for Robinhood, or various trading apps, to help trade better. It kind of remains unclear whether what you really want to be building in this world is a Bloomberg or Reddit so to speak, what matters more, whether it's community or tooling, but that's a pretty active scene. The most rampant scene of activity, which really goes to speak to how startups get created, is founders solving their own anecdotal problems, is there are a ton, a ton, of people working on various forms of socializing online, really trying to solve the Zoom conundrum, so to speak, which is to say, why is it that meeting over Zoom is not as good as in the real world, and how can we make it different and better?

A lot of what I would say horsed those carriages in this world, in the sense that what they are, are people, very much trying to recreate the real world in the internet, instead of being internet first. I think it's very interesting if you look at pretty much any technology. The car is obviously the famous one. Horseless carriage metaphor. Pretty much any technology goes through the real-world variant of it. We can think of, I don't know, pen and paper, the first digital form of it, the word processor. And then the fully online variant of it, which would be Google Docs. You could think of an encyclopedia, to Britannica online, to Wikipedia, same effect. And it feels like a lot of the video conferencing 2.0 stuff is still stuck in the encyclopedia era. No one's really figured out the Wikipedia of that, which I think will look incredibly different from the way it is to interact in the real world. Those are some interesting categories going on. There's another one started fairly recently, maybe in the past two or three weeks, which is, there's a lot of people trying to use GPT-3 to build various companies.

Patrick: [00:29:23] Can we spend some time on that last one? We're recording this, I don't know when it was announced or released, but fairly recent over the last couple of weeks. Could you describe what GPT-3 is? It is equal parts thrilling, interesting, maybe scary. Describe where it came from, what it is, and then maybe we'll talk about building companies on top of it.

Daniel: [00:29:41] We can use the tools we discussed earlier to the lens of GPT-3, which is to say, I do think it is a meme. In all honesty, if you haven't heard of GPT-3, I'd be very curious to talk to you, because that must mean you're on a very different orbit than mine, because that's all my world sees. So at the end of the day, GPT-3 is a machine-learning model that's been developed by OpenAI, that's achieved a remarkable advance in the field of natural-language understanding, from writing essays, to summarizing text, to doing question answering, GPT-3 is not quite like a human, but it's surprisingly close.

One simple metaphor for someone would be, just assume that GPT-3 has ADD. And it's able to understand text of very shallow depth, so it can produce and understand things like tweets, or maybe a paragraph or two. If you ask GPT-3 to understand something of any deeper knowledge, it tends to really go off the rails in interesting ways, but it really goes off the rails. It can do things at shallow depth, much like images require shallow depth of attention. Beyond that, I think it still fails. I think it is extremely interesting to think about. We'll talk about business applications for GPT-3 in a moment.

But it is extremely interesting to ponder if one assumes that GPT-3 is over hyped, why exactly it is over hyped. What is causing this level of hype and insanity? One simple explanation is, I do think, much like the Matcha lattes, we're providing a product in a shortage of colorfulness in the world. GPT-3 is providing technological advance in a world that's short of it. We have yet to see something big and new in the world of technology, maybe since the iPhone, and this is kind of close to it. A computer that can write text, that resembles a human is kind of close to it. So as a result, people are quite excited about it.

The other thing going on with GPT-3 is, especially if you look online, it's very easy for people to experiment with. OpenAI didn't just release a research paper that no one could use. They actually released a nice little website where you could toy with the model and speak to it. The results it produces, I would say, maybe 6 times out of 10, are bad. But the 4 times out of 10, you cherry pick and you post on Twitter, and you get reactions from people. It's almost a tool of creativity in that sense. That really created a tremendous amount of hype and excitement around it.

I think it's a good thing for the world, by the way, regardless of whether this product is under hyped or over hyped, the hype will lead to capital investment, will lead to more people working on this. I can tell you as someone who ran search and machine learning Apple for quite some time, there were real problems in that space that, if we solved, could unlock tremendous amounts of societal and economic progress. So I really hope more people focus and spend their time on it. I think a very question is, to the extent to which you would want to build a real business on top of GPT-3, and exactly how. So, example business have been, there's some interesting demos you can see online of someone typing in a bunch of words like, "Give me an interface with a table, with two buttons at the bottom, I can click on." And GPT-3 can produce not just English language, but also programing language. So, it's able to produce that interface, and that's very exciting as a demo, actually. I think the interesting question is, will be this be a disruptive innovation, one that startups will be able to take advantage of, or is this such an obvious thing that it'll be a sustaining innovation for the large companies?

You could imagine someone creating, say, a Shopify clone where you just have to mutter, "Set up a website for me that looks a particular way," and it just automatically happens, versus Shopify just doing that. That is, I think, the first way in which I'd wonder whether these startups using GPT-3 are good or not. The second more interesting question is, will there just be tons of GPT-3 now. There are various numbers online. You could read about how much money this thing took to create. It's all in the ballpark of single-digit millions. And you can envision pretty much every large company under the sun maybe training their own similar model now. So, a very rapidly evolving landscape. Certainly, a lot of people working on it. Remains unclear whether this is going to be the type of thing that's going to be like, say, mobile banking, which is very much a sustaining innovation for the big guys, versus a huge disruptive innovation like Uber or Instacart coming out of mobile.

Patrick: [00:33:30] On sort of the opposite end of the complexity spectrum, there's a great post that you've put up called The Power of Ten Playbook, which I think is a nice little ode to not thinking too big too soon, and a great way to get started on a project, which, my experience in COVID so far, I'm curious if yours as well. It's almost everyone I know is trying something new as a side project, as maybe even a new career. They have extra time. They're not commuting. Whatever the reasons are. I love this little playbook. Could you describe why you wrote it and what you mean by it?

Daniel: [00:34:00] 2020's going to be a very interesting year in hindsight because we've run this great A/B test of what the world looks like if we turn everyone into dropouts. If you're in a university, literally, even if you're attending Zoom school, you're kind of a dropout. You don't really have to go anywhere. So, classes are not being taught significant uptick in startup activity. I think super interesting, hopefully good. The blog post that I wrote is just an attempt of mine to distill or maybe simplify the steps that one might take in starting a company.

So, we touched on this earlier, but broadly speaking, you can imagine the world getting split into two. Jobs that have very clear set of next steps. Working in the military, or working in Starbucks, or working as a manager in a large company, clear jobs. And then you've got opaque jobs. I think starting a company is an incredibly opaque job. It's very unclear what to do next. It's very unclear, if you're starting a business, how it's going to become Stripe, or how it's going to become Airbnb. That seems like an unfathomable goal.

In all honesty, even getting to, I don't know, $100,000 in revenue feels like an unfathomable goal. So, I think it's very important to break these things down into bite-sized steps. That post was really just an attempt to do that. It seems very hard and very unachievable to run a marathon. But it also seems very achievable to run 200 meters. A good running program will give you the step-by-step way to get there, so that at some point if you really want to, you find yourself running a marathon. You find your company doing tens of millions of dollars in revenue. It really starts with just trying to get a small set of people. Initially, just 10, which I think is a very achievable goal. A group of 10 people that are avidly using your product. This is important for two reasons. One is, it'll enable you to run the process of discovery around what's wrong in your product and what's right.

I have not yet to see a case where, without consulting anyone in the world, someone creates something and it's instantly successful overnight. Maybe Bitcoin is the only example of this. No consultation. Just put it out there and it works. The one thing you'll get out of it is, 10 people, you'll be able to refine what you're building, and you'll be able to search the space of product improvement in, hopefully, a good way. The second thing that's very important, that's an underrated reason is, you'll find yourself becoming more motivated to work on the product if it's 10 people, ideally, that you respect, because you'll want to be able to serve them in an interesting way.

We often find in Pioneer that, that alone is the catalyzing feedback loop to getting something done. It's not even revenue. It's just getting the attention of 10 people that like what you're building. Once you go to 10, you can try to go to 100. It's the next step. Once you go to 100, you can try to get to maybe $1,000 in recurring revenue. What I find so hopefully compelling about this concept is, it's extremely inexcusable to say, "I can't get 10 users." I mean, everyone's got to be able to find 10 people. The internet has billions of them.

The attempt was really to distill this unknown fairly opaque thing to a simple set of goals, that I think anyone can use to get started in starting an online business, that hopefully goes somewhere. It's worth mentioning that I feel like a lot of founders forget that everything big starts small. The reason people forget this is, there's media teams responsible for making sure you forget this. The first thing you do after the company's successful is, you hire folks in comms and marketing, and they'll do an excellent job of rewriting your origin story, so that it sounds like, from the moment you were six, you were working on whatever it is, Microsoft, or Amazon.com, of Facebook. And this was your life's vision to connect people. Whatever, whatever, whatever. I find it very strengthening to go look at early variants of witness of Amazon.com. You can watch some of the interviews with Mark Zuckerberg from 2005, where he says, "There's no reason Facebook should expand beyond the scope of Harvard."

It's worth mentioning that I feel like a lot of founders forget that everything big starts small.

Even SpaceX with all of its might, SpaceX with all of its might was originally conceived as, I believe, the Green Mars Oasis project, which I'll describe to you. It sounds like something you do when you're in LA. The idea was, "Hey, let's go buy some Russian rockets, launch them into Mars. We'll just get a photo of a green plant on Mars, and then we'll shutdown the company." That's it. It's a giant photo stunt, which, at the end of the day, is slightly ambitious, but it is not to the ambition of, "We're going to create interplanetary travel, the largest private space company in the world, and migrate the entire human populace to another planet." That was not the goal. The goal was, "Let's get a nice photo on Mars." So, you got to remember, everything big starts small. That post was really meant to distill that for people and give them a bit of a framework there.

Technology Companies Transforming into Utilities

Patrick: [00:38:05] One of the interesting transitions happening, you and I were chatting about TikTok a little bit is indicative of this trend that I think you've written about elsewhere too, which is the notion of technologies starting at a frontier, becoming what we call technology, but then eventually becoming a utility. Most people don't think of today's utilities as yesterday's technologies, but they were. Can you talk about that general transition, that general lifecycle of a piece of technology, a railroad, or something else, and how it relates to some of the giant what we still call technology companies today, maybe with TikTok as an example?

Daniel: [00:38:36] I enjoy reading a lot of old news. I find that, at the end of the day, the world moves in cycles. And the tricky thing is, the length of the cycle seems to be just ever so slightly longer than the length of the human lifespan. So, it seems really new and novel to us. But it's exactly what people went through in prior times. We don't read old news because, I don't know why, new news seems more important. I find old news is a great decoder ring to life. In particular, you can read quite a bit about just the terminology used to describe the Vanderbilt railroad empire, the Rockefeller oil empire, very much as technology.

Obviously, cars were considered technology. Roads were considered technology. If you go back far enough, maps are considered as technology. So technology, I think, is really just our way of saying the new thing. I feel like what we're watching now, if we look at the fights going on with Twitter adding various tags around Trump's tweets, or what's about to happen to TikTok is, what we call technology businesses today, a lot of them, especially the consumer-facing ones, are effectively media businesses. They're media businesses just like CNN is. When something moves from this era of being, say, a frontier technology, at some point it was cars, at some point it was social networking, to a utility, to something that's very popular. First of all, the wording around it, I think, needs to change, although this lags. So, you'd really, again, want to think about Facebook and Twitter as media businesses with all the upsides and downsides to it. We'll talk about how this folds into TikTok in a minute. But second, I think this means that they're going to have all the pros and cons of what a utility is. If you think about, say, PG&E or any other type of utility, the cons are, for the most part, aggressively regulated. It is nationalized, or at least federalized, or mandated statewide. The pros are, the regulatory lock-in creates a moat of sorts. You can't start another PG&E tomorrow. I think, over time, we'll probably evolve into a world where it'll be hard to start something that has the shape of Facebook, if they really manage to get the regulatory lock-in the regulators and weirdly Facebook want. So they're kind of pro and cons both ways. But we are in this interesting interstitial era where, I think the world thinks that these media businesses are technology businesses are in that middling phased you described, where they are in fact total utilities.

I think, over time, we'll probably evolve into a world where it'll be hard to start something that has the shape of Facebook, if they really manage to get the regulatory lock-in the regulators and weirdly Facebook want.

One distillation of this is, I think what's going to happen to TikTok, which is tremendously popular. There's actually a tremendous amount of US dollars invested in the entity that, as I understand it, is a Chinese entity. And it's kind of rapidly unfolding, but there are quite a few parties that are working on transitioning TikTok into an American entity because the United States has banned TikTok, at least in the military. India has banned it across its entire county. On the one hand, I feel like I speak to a lot of people that say, "That's crazy. What's happening? We're banning websites." On the other hand, if you think of TikTok as a utility, the United States is not going to let a Chinese entity run its power grid. It's not going to let a Chinese entity run 5G chips, certainly. So it's not going to let a Chinese entity run one of the largest media companies. At the end of the day, either the thing will get totally banned in the United States, which I highly doubt, or it'll have to split off and become somewhat of an American entity.

I think, weirdly, if everyone thinks this is bad for the capital stack in the company, but it's, of course, always a great thing. One of the greatest things that ever happened to any other previous technology monopoly is, they get some type of lock-in by the regulator. We had banks in the, say, 18th, 19th century, or oil in the early 20s. So that, to me, is a very interesting lens to view tech right now. I think the asset allocator, of course, wants to think about, "How do I avoid the utilities?" And "How do I find the things that’s either a frontier market or a commercial market right now?"

I think there's a lot of interesting activity in everything from satellites in space, I think remain mystifyingly unregulated. I think there will be a huge war in the years to come between Starlink and Geo, just about the airwaves. But I think, right now, it's total open season. To me, by the way, this is astonishing. It is astonishing that Elon is going to be able to, with Starlink, effectively be able to provide internet to the world in a way that is almost impossible to regulate. No one's really thinking about it that way. But when you think about it, he's going to put internet in the air. There's no towers you have to manage. There's no cabling you have to lay. It's literally going to be in the air. So that's a great example, I think, of an underrated frontier market.

Patrick: [00:42:41] In that specific example, why can't it be regulated? Also, what do you think it enables? Is that the equivalent of what everyone also talks about around platform shifts that create, the introduction of the iPhone or something, that creates an enormous opportunity to build stuff on top of? In your mind, does something like Starlink represent that sort of platform shift, assuming it's not regulated out of existence or something?

Daniel: [00:43:04] Two questions there. In terms of why it can't be regulated, I mean, I think it can be regulated by convenience. Great Firewall in China is an interesting example of this. Yes, technically you can bypass the Great Firewall. It's not hermetic. In many ways, you don't. So that gives a huge edge to, obviously, Chinese companies. So, I think you could do similar things with Starlink where, for example, you could ban the sale of, say, receivers in a particular county. Yeah, people will still smuggle them in. But the convenience factor matters much more.

It's hard to regulate iridium. It's hard to regulate something that is just in the air. In terms of it being disruptive, yeah, one of the most interesting questions I find right now, in disruptions, we can of course think of consumptive ways that'll disrupt the industry. But nonconsumption is always the most interesting form of disruption. This is the Clayton Christensen framework, obviously. Nonconsumption or new use cases that we couldn't've thought of before, that just really change everything. I don't exactly understand what the nonconsumption just around super cheap satellite launch will be. But it feels like there's a lot of interesting things there. If you could launch satellites, I believe an iridium satellite launch is maybe $50 million or so. With Starship, I think a SpaceX satellite launch is maybe $50,000. So, it's very interesting to think of what goes up in space and how that whole thing unfolds, if it's that cheap to launch things.

Internet satellites feel like the most obvious use case, initially, but certainly not long-term. I imagine weird stuff will come out of it. It would've been hard, I think, to predict Uber when the mobile phone came out. Very easy to think of, order a taxi on your phone, not increase the TAM of the market like Uber did. And I wonder what that will be for satellites. There's a different question, which is for someone allocating capital to it, what is the payback period, and to what extent is this something that should be infrastructure investment, versus venture capital investment. That's a whole other question. But I very much believe that the next decade, we can think of 2020 to 2030, that will be the decade of Starlink, of satellites, and of space.

Seeds vs Leeches

Patrick: [00:44:55] One final question on the ecosystem which you operate, and then my traditional closing question. So first, the final question on the venture ecosystem. You mentioned the amount of capital sloshing about down there. Say a little bit about this notion you have of a seed-versus-leech ratio in the venture world, and how you see that evolving.

Daniel: [00:45:13] First question we must ask ourselves is, is the venture game, a picker's game or is it an access game. That's the first question. Of course, the answer is, as you go across the cap table, you see it shift. This may be an inverse-U-shaped curve. As you can imagine, very early on, it's a picker's game, meaning, for example, with Pioneer, extremely early. We can pick whoever we want. Very late stage in the cap table, it's also a picker's game. I mean, obviously the company's public for the most part. Hedge funds are in the game of picking, not in the game of access. But at the A and the B, it is very much a game of access. The good companies are known, and the main question is, can you get into a company. I think if you're an investor, you have to realize where you're playing. And you have to develop different competencies based on where you're going to play. I think, if you are going to place in the access game, it is very interesting to think of, in my head, the way it's distilled is, your seed-leech ratio in the venture world. Anyone who grew up with BitTorrent would recognize.

There used to be all these BitTorrent communities, and BitTorrent is somewhat similar to venture, where access is constrained, in the sense that all of these communities, there's a limited amount of a certain resource, the latest TV episode that you want to watch. And all these communities would have a seed-leech ratio that they would assign to you, based on how much you're contributing to the network. That is to say, how much you're allowing other people to upload your copy of that TV show, versus how much you're leeching, how much you're taking away from the network.

I find this an interesting way to view relationships in general. I think it's very important, in venture, to think it about it this way, because you are going to want to have good relationships because you need to get invited to the party in order to win. You want to be that person that is voluntarily given access because there's a scarce resource. I've been thinking a lot lately about what a good seed-to-leech ratio is. A very simple way of quantifying this is, you can think of friends, people you interact with. And think of the number of interactions you have with them where, don't want anything from you, or they're giving you something, or they're just positive. There's nothing there that they're leeching from, versus the number of interactions where they're actually leeching something through you, where they want something from you.

It's very much my goal as an individual to not have to have a lot of interactions where I have a very high leech ratio or I always want something from someone, because I don't think those are fun. You think in the back of your mind the people who have a bad ratio there. I feel like I see this a lot in the venture world where every interaction I have with the person, they always want something from me behind the scenes. It leads to very un-genuine, un-fun conversation, in my opinion. You think of this conversation that we're having now. There's nothing we really want for each other. We're just trying to have a good time, and I think that's what creates flow. I think if you want to engineer success for yourself, in this market, I think the place I try to be is very much to not have interactions where I really want things from people for the most part. This doesn't work, I mean, to the extreme. At the end of the day, if you work at a company, obviously, occasionally, you need things from people, so that's a bit of a different dynamic. But you can think of friends as the tail end of this.

Friends are people where you almost never have intersections where you're leeching from them. It's always, you're seeding, there's no goal, you don't really want anything from them. So, I much strive to try to be that person in all interactions. The other way of distilling this in your mind is, scroll your WhatsApp, or your iMessage, or whatever. Who are the people that you almost always invite to the party, and what are the properties of those people? I wonder this a lot because, at least this insecure-overachiever's attempt to be a good actor in the world and have a sense that I stand in a decent spot in mankind's hierarchy. I would like to know that I would be that person. I know I'm certainly not always there, but I'm striving to be. So, I find this concept of just making sure you have a good seed-to-leech ratio with people to be an important one in that sense.

Good Interview Questions

Patrick: [00:48:55] It's a fascinating way to look at so many different things, maybe not public markets, as you said, so much, where you can get what you want if you're willing to pay the market price. But every other kind of investing and relationships, this seems like a great concept. It reminds me of, I think it's Adam Grant's book, Give and Take, a wonderful closing concept. My closing question for everybody is to ask for the kindest thing that anyone's ever done for you.

Daniel: [00:49:16] By the way, it's worth dissecting for a moment why this question is so good and why it works. For any interview question, I think it's worth wondering that. I think, in my view, as you ask it, there's two things going on. One is, it's novel. Most people don't ask this. Two is, you used the word kindness, and that relaxes the environment a little bit.

Patrick: [00:49:36] I've never thought about the question itself. Before you answer it, I'm just curious if there are other questions that have a similar quality, or other qualities that you find good or helpful. I guess I'm basically just asking you, what are some good questions?

Daniel: [00:49:49] I mean, this is the secret to life. I'll share one, which is, how do you think this conversation's going?

Patrick: [00:49:55] I love that question. I think it was you that introduced me to it, maybe a year ago or something like that. And I've used it constantly ever since. The reason I love it is because it is so hard to hide from. You have to say something that's interesting or has some weight to it. It's very hard to just say, "It's going okay." You just don't say that. You either say, "It's going great," and describe why. You say, "It's not going great," and feel the need to really defend that opinion by being very specific, which is why I love that question. I'm curious if you like it for the same reason or for a different reason.

Daniel: [00:50:29] I think there's a lot fun things going on with that question. I mean, it does make you smile and laugh, and that is, I think, the most important thing it does. It breaks the fourth wall on the interview, which, I think, is your goal in the interview, is to break the fourth wall. There's this ethereal quality to it, which is, what you're saying is, "Look at us in the game that we're in. Isn't this kind of silly?" I think that very much relaxes, especially, a candidate who's probably nervous. Second is, yeah, I mean, obviously, it causes the person to introspect in a new way they haven't done before. The third is, my way of thinking of your third point, it is impossible to not have an answer about it. It's got to be a cached answer. I mean, it's an experience you just went through. One mistake I think a lot of people making in interviews is, they ask questions for people that are novel and surprising, that, the person feels like they have to be performative in their answer.

The old Peter Thiel question about, "What's an area in which you disagree with the world?" When a person asks you that, I think the one flaw in the question ... I mean, now, it's become too popular ... but the one flaw in that question is, there's a performative aspect to the answer. So, you don't really get a genuine sense of the person. Whereas "How do you think this interview is going? How do you think this conversation is going?" It doesn't sound too intellectual, and I think that's very important if you really want to claw at what makes the human genuine. This is the other reason why ... Tying this back to our initial debate around films. That's a very easy thing to get a sense of the person. "What are some film that you like?" Because it's disarming.

Patrick: [00:51:54] Are there any other questions for you that are on your personal Mount Rushmore of question, that you think you'll find use for forever, like I do, the kindness question? And then we'll come back to that question.

Daniel: [00:52:04] There is one question that Pioneer asks, that's been, for us, the most revealing question, period. We collect a lot of these qualitative questions in addition to the quantitative data, to try and see which question is predictive. And probably, the most predictive one we have is, "What is something weird or unusual you built or did early on in life?" I find it so astonishing in how polar it is. Someone will tell you about, "I accidentally got my high school teacher fired." And then someone will say something incredibly bland. "I did a 10-mile hike." To me, it really explains people. Again, it's worth thinking about why that question works. I'm asking you for something factual. I'm not asking you to think or be highfalutin with your hoity-toity intellectual concepts. So something factual. Something simple. You could give me an earnest answer. It shouldn't take too long to think of. Really give me a sense of who you are.

Patrick: [00:52:51] To round out the idea, we're talking about interviews mostly, or I guess, maybe references would be another category. The reason that all these questions are interesting, is what? You're trying to get at some underlying substrate of the person, their attitude? Why are these so interesting to you, as questions?

Daniel: [00:53:08] I'm trying to understand, ideally, even better than you completely why you're doing what you're doing. I don't think I'm any brilliant person. I actually think other people could understand about me, why I do what I do, better than I can. One of the greatest lies is, when people tell you why they did certain things. I'll speak for myself. You were to ask me, "Why exactly did you leave Apple when you did," or "Why did you sell your company?" I don't know that I could give you the real answer. What I'm going to do is, I'm going to give you an intellectualized answer I thought, that I think you'd be interested in hearing.

It's so funny when you talk to people about, "Why did you leave job X and go to job Y?" "Oh, I felt I wasn't learning enough." Do you really have a daily learning number that you put next to every diary entry, and the learning number was low? Is that really why you left, or was it something deeper? Is what happened that, I don't know, you bumped into someone at the barbecue, and they mentioned their salary, and your salary was lower? And then for three months, this was in your head, that your salary wasn't what it was supposed to be? And at some point, the email came in from the recruiter, and inspiration struck? Or was it that, you had some conversation with your wife, and went down some rabbit hole and you realized in the conversation that you just really don't like your boss? The stated reason for things with humans is almost never the real reason, and it's often a third party, ideally, people much smarter than me, I hope, that can figure out what it is.

And in an interview, what I'm trying to figure out is, if it’s a founder we're meeting, or whether it's someone we're hiring to work at Pioneer, I'm trying to figure out, why are you doing this and what's driving you? I find ways to get there are, asking simple questions that the person can answer without needing to perform. The person is working against you when they start performing. By the way, another tactic I'll occasionally use is, I tell people, especially now over Zoom, where there's a bit more of a disconnect. I tell people up front, "Please stop performing and just be yourself with me. And I will try to do the same with you." Very simple other thing I highly recommend doing on this front, repeat the same question over and over again until you get the answer. I think a lot of people walk into interviews and they have this rubric of questions they need to ask.

The stated reason for things with humans is almost never the real reason, and it's often a third party, ideally, people much smarter than me, I hope, that can figure out what it is.

Maybe their boss gave it to them. They ask the question, they get the answer, and then they move on. There was literally no entropy exchanged. The question was, "What is your biggest weakness?" The answer was, "I work too hard," and then you moved on. You might as well have not had a conversation. That's what a draining conversation is in my opinion. It's that you're eating up oxygen without any novelty in it. I very much try to not have those, partially because I'm trying to understand the person. Partially, because nobody wants to be bored.

Patrick: [00:55:39] I have a friend who says, "Society and its trimmings are only three days deep," meaning, if you're by yourself in the wilderness or something for something, it all sorts of fall away. I like this version of it, which is, "The truth is only three answers away," or "three questions away," rather. It's a neat concept. I'm forced now to return to my actual question. I do love ... I've never explained this before, but since we're talking about questions. I do love asking it, just because it makes me happy at the end of interviews. It's not more complicated than that. And I think that orienting and priming people around some act of kindness is a great way to close. So, I'm curious for your answer to the kindest thing that anyone's ever done for you.

Daniel: [00:56:15] Okay. I have a bit of a recency bias, but something that's been on my mind almost every single day. A couple of weekends ago, I drove my car to the middle of nowhere in California. It was pretty warm. Parked my car on the side of the road. I don't really pay attention to where I parked it. I locked my door, put my keys in my pocket, and I got out for a run. I come back to the car, 12, 14 miles later, pretty dehydrated. Didn't bring any water with me. It's maybe 95 degrees. I'm pretty excited to get into the car, turn on the air conditioning, drink some water, and get back home. I then discovered that the car had sunk into the ground a little bit. It's very thin sand. I turn on the car, and start the engine, and hit the gas, and dig a great ditch for myself.

Situation becomes a little bit more alarming when I realize the car has almost no fuel left, literally, in middle of nowhere in California. I really treated myself here. I thought I was screwed, but it only took maybe, I don't know, 10, 12 minutes until someone's driving by on the road. They drive by and then they pass me, and then they turn around. Low and behold, over the course of the next 30, 40 minutes or so, three people stopped their car, turned around, and helped me tow my car out of the sand. To me, this was so simple, but so momentous for me. I truly had begun thinking, "Well, this is basically it. I'm stuck here and I'm going to make camp here for the next three days." But the fact that people went out of their way to help me, here's the interesting thing that I find in it. Now, every time I'm driving on any type of road, in particular, if it's desolate, and I see someone stuck, I have a real proclivity to try to help them because I was helped. I think that small effect, in many ways, describes the magic of Silicon Valley. Someone took a bet on me when I was 18 years old, and you see now in my career, I'm trying to take other bets on people. So there's a real pay-it-forward element, I think happens, to when you shower someone with an excessive amount of, with a surprising amount we should say, of effort when they feel like they don't deserve it. So I don't know if it's the most kind thing people have done to me, but that's a pretty big one.

Patrick: [00:58:13] I love it. So many of the answers come down to the same thing you said earlier, around seed and leech, that fundamentally, they're all seeding activities. That's a great place to close. Every time we talk, I feel like I take away 15 fun devices, new bits of learning, new ideas. You're always full of them. So, I've loved to have finally been able to do this on the record. I'm sure it will be the first of several. Thanks for your time.

Daniel: [00:58:37] Thank you so much. I enjoyed it as well.