Moth Fund
Moth Minds
Matt Clifford: Co-founder & CEO of Entrepreneur First

Matt Clifford: Co-founder & CEO of Entrepreneur First

MM.04 Culturally-encouraged entrepreneurship, the internet's variance amplification, and re-accelerating science
I talk to people and record it! Some call this sort of thing a podcast. I call mine Moth Minds and the premise is “interviews with high-agency humans."

Subscribe to hear the latest on a bi-weekly(ish) basis: Spotify, Apple, Google, Amazon, and Pocket Casts.


  • Cultures of entrepreneurship are created when the social currency is based on how ambitious your attempts at creating new businesses are as opposed to what credentials you’ve collected. Lowering the social friction to test an idea for a business and making it easier to stop working on it helps expedite the entrepreneurial process of narrowing in on a company-shaped idea that works.

  • The internet is the ultimate amplifier of extreme outcomes (variance) because it takes something that might have been observed or participated in by a few hundred/thousand people and makes it potentially accessible to billions of people.

  • ARIA is a bet that a new organization with a very different structure and a far greater risk appetite baked into the legislation can help re-accelerate science.

MM: Hi everybody. Today I’m joined by Matt Clifford, the co-founder and CEO of Entrepreneur First, the Chair of ARIA, the UK’s Advanced Research and Invention Agency, and the co-author of the book, How to Be a Founder

Our conversation covered the characteristics that define cultures of entrepreneurship, the goals of his work on ARIA, and how the internet functions as a variance amplifier. I hope you enjoy.

MM: Hello Matt and welcome to the podcast.

MC: Thanks so much for having me, Molly.

MM: By way of introduction, my first question for you is, you say that the internet is variance amplifying. Say more about what that means.

MC: Yeah. Well I think the way to think about what it means for the internet to be variance amplifying is probably to step back and say how do we get to the point where there wasn't that much variance to begin with? My sort of basic thesis about the world is something like the world was very, very weird for a long time. Then modernity came along and the main impact of modernity for a couple of hundred of years was to reduce variants. I'll explain what I mean by that for a minute.

And then the internet came along and it just threw a massive bomb in that and we're back to a world of extreme variance amplification. So what do I mean by variants? Well, I basically mean how extreme would you expect the outcome of any given social or economic process to be? And I think we've kind of got used to the idea that we have a set of institutions, certainly in the sort of democratic west as where you and I live, that damper variants, like the way we pick leaders.

Again, sort of if you ignore the last 10 years, it was sort of designed to pick the boring candidates. It was designed to pick the people that would do what the system expected. If you look at what our corporate leaders looked like 20 or 30 years ago, they were sort of satisfying the median desire of Wall Street analysts or the media or whatever, and that was normal because, in general, for most of human history, variance was a pretty dangerous thing. Variance is the thing you try and avoid. Variance is what kills you. And so, modernity did a great job of dampening variance. A lot of the institutions we love dampen variance.

What the internet does is it basically does the opposite. It basically amplifies the most extreme outcomes and a lot of your listeners will be familiar with this idea of it's called the Matthew Effect which sometime is called the rich get richer or where basically anything that attracts some sort of resource like attentional money then attracts more resource like attentional money. The internet just does that on steroids.

“The internet is the ultimate amplifier of extreme outcomes because it takes something that might have been seen or observed or participated in by a few hundred or a few thousand people and makes it potentially accessible to hundreds of millions of billions of people and that increases its variance a lot.”

The internet is the ultimate amplifier of extreme outcomes because it takes something that might have been seen or observed or participated in by a few hundred or a few thousand people and makes it potentially accessible to hundreds of millions of billions of people and that increases its variance a lot.

MM: That's really interesting. I love that description of it. It's also a good articulation of what a power law outcome looks like in the world when it's applied to everything. How does this change in the age of AI specifically — do you think it'll become more pronounced in the future or do you think it will even out over time?

MC: Well, I think one of the most exciting, some would say, terrifying things about where we sit right now, we're recording this in June 2023, is I think what the future of AI looks like is really up for grabs right now. And there are many different ways I think it could end up. But to paint sort of two, maybe not extremes, but two poles of what could happen. I think you can imagine a world where AI ends up being this very centralizing force where we end up with one big AI to paint very crudely. And it turns out that the returns to scale are so huge or the returns to getting to something like AGI first is so enormous that effectively you have one AI that is this the sort of dominant factor in the world.

Now I would see that probably ends up being variance dampening for reasons we can get into. I think there's another world where actually you see a lot of what's happening in open source right now, big very large language models being released in open source. You see a proliferation of innovation in that space. You see actually the compute needs for those models which are not fully competitive with say GTP 4, but pretty competitive with GTP 3.5. The compute cost of that is actually really low now. You don't need a huge compute and cluster to train one of those models.

You can imagine a world if that becomes the norm where there are millions or billions of different AIs, all of which are powerful, but there's not that centralizing force. If that world comes to fruition, I think that's a massive variance amplification because then when core insights are discovered, one, you'd expect that to happen in a very distributed way. It's not going to be limited to a small number of biddable people that can be hemmed in by regulation or other social forces. Then, I think, you get a massive explosion of variance amplification.

And to be clear, I should have said this in the answer to the first question, variance itself is neutral. I don't think by itself you can tell which of those is good or bad, except that I would say, you can imagine in the case of a world with a billion AIs, if bad things happen, they could get bad really, really quickly. Clearly, the opposite is also true. Maybe that's one of the best ways to material abundance.

MM: I like that a lot. That's a good answer. What is your take and how do you think about the question of is the rate of progress slowing down?

MC: I think that there are certainly domains that you can point to where pretty unambiguously to me the rate of progress is slowing down or at least when you adjust for the amount of input, the number of people working in them, the amount of capital going into them to pick two, one of which I know something about, one of which I know really nothing about. So I think it's pretty clear that the rate of progress in science adjusted for capital and people is slowing.

It takes more people and more money to get to breakthrough ideas today than it did a few decades ago. And we can talk a little bit about that. That's an area I know a bit about because I'm the chair of this thing called ARIA, which is a UK brand new R&D funding agency that's been given the mandate to try and solve that problem or at least contribute solving that problem.

I think the other one that comes up a lot, at least on Twitter, at least in the circles I hang out with, is it's sort of weird how hard it is to build things today. And if you measure progress, at least in part by the sort of physical assets that we have at our disposal and the ability to make ambitious things that are made of atoms not bits, I think, it's very clear that that has slowed our ability to do that quickly has slowed.

Our mutual friend Patrick Collison has that great page on his website, which I think is just, where he just points out a long list of things that physical assets that were built in the past in what today seems ludicrously short periods of time. And we just don't do that today. And that I see as the opposite of progress I suppose. Yeah.

MM: Speaking of ARIA, I'm curious, what was the genesis of your being involved in the mission behind it and what problems are you trying to solve?

MC: Yeah, so ARIA came out of a desire from various people in UK government over the last few years to really look at this question of, is science slowing down? Is the rate of scientific progress slowing down? And if so, what are some of the institutional arrangements that might be able to fix that? Now clearly, one hypothesis is, as you sort of already hinted at, maybe we just used up the low-hanging fruit, maybe it will slow down.

“ARIA is a bet that says that there is something about the institutional apparatus of funding science and the way this happens today that is responsible for this slowdown. By giving a new organization a very different structure and a lot of freedoms, can you create a beacon for what it might look like to re-accelerate science?”

ARIA is a bet that says that there is something about the institutional apparatus of funding science and the way this happens today that is responsible for this slowdown. By giving a new organization a very different structure and a lot of freedoms, which I can talk about because obviously ARIA is small, it's not going to reverse the whole thing, but can you create a beacon for what it might look like to re-accelerate science?

And so, ARIA was set up by Act of Parliament in 2022, just over a year ago, and I always tell people who are considering applying for jobs at ARIA that they should actually read the act because although I would never recommend people read grammar just incredibly short, and this is sort of the beauty of our, IT basically says ARIA exists to fund breakthroughs, scientific breakthroughs for the UK and the world and it can do more or less anything that isn't otherwise illegal in order to get there. So it can do funding, it can do investments, it can do grants, it can do prizes, it can do really anything. It has relative, I should say it's fully funded by governments entirely. A public sector organization has 800 million pounds, which is what, just over a billion dollars I guess. And unlike most public sector organizations, ARIA has the freedom to hire who it wants and pay them what it wants.

“One of the things that I think people who work outside government often miss is that because of the way accountability works in government, if you don't have things like a risk appetite baked into the legislation, then you risk your success being evaluated in the same way that people would evaluate how you would build a bridge. I’d argue that the way you want to approach building a bridge and the way you want to approach funding breakthrough R&D should have a very different risk appetite.”

It can procure in ways that would normally be outside the way that the public sector, at least in the UK procures, it has an explicit risk appetite baked into that act of parliament that says it has to do high-risk things. And these sound like very, maybe particularly from people from Silicon Valley, listen to this be like, these are pretty table stakes stuff. But one of the things that I think people who work outside government often miss is that because of the way accountability works in government, if you don't have things like a risk appetite baked into the legislation, then when you are asked to account in some years in the future for what you did, the risk is that you are evaluated in the same way that people would evaluate how you would build a bridge. And I would argue that the way you want to approach building a bridge and the way you want to approach funding breakthrough R&D should have a very different risk appetite.

So even just little things like saying in the legislation, the point is to make high-risk bets that may not work in pursuit of enormous outcomes, that provides a degree of protection for the executive team of ARIA to really go all in on what they think might work. So sorry, there's a risk of becoming a long answer, but broadly, this legislation was passed last year. I was appointed to be the chair last summer alongside an amazing guy who's our CEO called Ilan Gur, who's executive leadership of the organization. My job is to, I suppose, do the governance but also do the communication back into government to stop any concern from government that they should be getting more stuck into what ARIA does. And broadly what ARIA will do is two things. One, it will fund hopefully breakthrough science and technology through a mechanism not unlike DARPA in the US.

So I don't know if I should say have, we are in the process of hiring program directors who are themselves really talented scientists who otherwise would be leading research themselves and what their role is to take some of this budget and actually come up with a vision of a technology or a capability that they think is on the cusp of being possible. They imagine what it takes to drag that into the present by funding people across science all the way through to engineering to work together to bring it into reality so they're not building a startup. This is stuff that's typically going to be like a decade pre-commercial, but they're bringing a lot of startup mentality to that activity. And what they're trying to do is say, if this were possible, who would I fund and how to make it happen? And so, we're probably going to have sort of 6 to 10 of these programs going, all of which have to be ambitious enough that if they're the only one that succeeds off that portfolio, it's still worth having done the whole thing.

MM: Truly very similar to startups. Yes.

MC: Right, exactly. And I think I have to say credit or maybe premature, but credit to the UK government that they really were willing to recruit people into these two roles, the chair and the CEO, that had a very different ... I have a very different background from the people that typically fund R&D funding bodies in the uk. And I think Ilan has quite a difficult different background as well. He's an entrepreneur, he is a very accomplished scientist, but he is also a deep tech entrepreneur and a founder of deep tech entrepreneurs. And it was a bet on taking quite a different mentality about how you might do this thing and applying it to a core problem in science. And it's very early days, I don't think we can, it will be years before we know if it's really paid off. But I would say the thing I have been paying closest attention to so far is what is the caliber of people we are attracting to our team who want to be part of this? And so far on that, I feel like in terms of talent density, we're in a really exciting place.

MM: That's amazing. Going back to the core problem that ARIA is trying to solve, if you could change anything about the existing institutions that are funding science, aka academia, what would you change?

MC: Yeah. So I think when I step back and I think, what have I learned about talent and outside success from spending the last 12 years at Entrepreneur First helping people build companies. The same themes that I see there are the themes that I see in academia as being the potential, the bottlenecks. And of course there's a little bit of a risk that in all you've got as a hammer, everything looks like a nail. But I think what I would say is the thing I've learned about entrepreneurship and startups is if you're going to just do one thing, probably that thing should be find the most talented people that you can and then remove all the barriers and pursue it from them pursuing their most ambitious idea. That's sort of broadly, if you, obviously there's loads, loads of more nuance to it. And I guess, you and I are pretty aligned mission-wise on that, just find extraordinarily talented people, remove barriers, and encourage them to be the most ambitious version of themselves.

I think if you look at academia, I think the same rule applies. Find extraordinary talented people, remove the barriers and encourage them to be the most ambitious version of themselves. So then the question is, why doesn't that happen already? And for me, the number one answer in academia, and one of the reasons I'm so excited about ARIA is that the thing that stops people being the most ambitious version of themselves is time and time again I hear if I try and do that, I just won't get funded. And that's because of the way, again, painting, I don't know if it's exactly the same in the US although I do hear enough things from friends in the US that have looked at this problem to think it's broadly the same. If you go walk around the really great research labs and the really great universities in the US and the UK and you find ... You go around and just say, who are the smartest postdocs here?

And then you go write a list of those and go talk to those people and you say, "What are you doing so cool? What's your biggest idea?" And then they tell you and then you say, "Well, are you working on that?" And they're like, "Of course not." And you're like, "Why not?" And they're like, "Well, because they would just never get funded to do that. If I wanted to get funded to that first I need to get N number of publications with Y number of citations. And so I'm doing this thing because that will definitely get me the publication. And then by the time I'm like 45, maybe I'll be able to do something close to what I think I should do."

And I find that really worrying and depressing because if you look at the history of science, that's not actually where the breakthroughs have come from in general. They typically come from people making anomalous, controversial, weird bets relatively early in their careers and going all in. And my worry is the thing I would change is making that easier to fund today. And I think we have got to the point where rightly, again, we can go back to variance dampening, variance amplifying. What our funding system does today is it reduces variance. Effectively, no one has consciously set out to say this, but what it effectively does is said, how do we minimize the risk that we fund something really stupid? How do we minimize the risk that there's a headline in one of the newspapers that we spent a hundred million dollars on something that was never going to work? And that's variance damping.

“If you look at the history of science, the breakthroughs have typically come from people making anomalous, controversial, weird bets relatively early in their careers and going all in.”

And in some ways that's a good thing. Accountability does tend to be variance damping, as I said before, it's neutral whether this is a good thing or not. Equally, if you think about the real breakthroughs, the things that have moved civilization on from which we get all the things that we care about, a lot of them looked pretty stupid at the start. A lot of them carried a high risk of a bad headline at best and other risks. And actually, I think what I would do is I would at least create pockets like ARIA that are variance amplifying where you say, "Go all in on your most ambitious idea. We've got your back. If it turns out to be the wrong idea, you're not going to be impoverished. It doesn't end your career. And crucially, we're going to fund many such things such that the thing that does work pays in inverted commas for all the rest."

And so, I totally understand why we probably shouldn't take all global science funding and put it through a variance-amplifying machine. There's a lot of stuff that just has to get done where actually you do want incremental progress. But I think taking a portion of that and saying, "Let's throw it through a variance-amplifying machine," is super valuable in terms of human progress.

MM: It's creating an environment for outliers or something before it ...

MC: Absolutely.

MM: Yeah.

MC: Absolutely. And there are very few, again, going back to the theme you started with, if you think about where outliers really welcome in society there's not that many places. Companies are very dampening machines. Yes, they want big ideas, but they sort of mainly want to deliver predictable earnings per share growth quarter after quarter. And you mainly do that by making incremental improvements. And so, corporations select for competence and skill, but they don't select for variance. They select against variance.

Startups, at least in the first few years, the opposite. And that's why probably you and I find them so exciting is that you're selecting for variance, but it's a dangerous thing to do. That's why so many startups, it's not why so many startups fail, but it's an artifact of the same fundamental uncertainty. And so, again, I just think you are exactly right, just as startups have become, certainly in the US I think increasingly in the UK, the obvious center of gravity for people who consider themselves outliers, it would be great if science felt a bit more like that to more people.

MM: Absolutely. Yeah. It's spreading the extremely ambitious, huge jump changes culture of technology into other areas. Shifting more to speak about your work with EF and entrepreneurship more broadly, what do you think are the highest leverage ways to increase entrepreneurship? And potentially, it sounds like there are a lot of parallels here.

MC: Yeah. Well I think there's a few things. I mean, I think obviously, given what I do and what you do, we're probably somewhat sympathetic to the idea that it is broadly going to be about talent. And I really do believe that. So I think specifically, if you buy what I just said that most mainstream institutions are hostile territory for outlier individuals, then that probably suggests that there are a lot of people who are in those institutions who are uncomfortable and feel dissatisfied and disillusioned and constrained and shackled, all these things who actually would be great founders.

And I think one of my slightly more controversial theses about what Silicon Valley gets wrong is that Silicon Valley is that one place in the world where nearly everyone that you meet has already realized what I just said and already brought out, right? And so, there's a huge selection effect in who Silicon Valley people talk to. And so, I think there's this myth that not everyone but many people in Silicon Valley have, which is that anyone who should be doing this has already figured it out. And the problem with that belief doesn't actually affect, I think, very much Silicon Valley because there is a critical mass of people that have figured it out.

But the problem with that belief is that it basically, and it is the opposite of Silicon Valley nearly every other respect, it basically imposes a scarcity narrative over entrepreneurship rather than abundance narrative. Because it basically says, the supply of great founders is more or less fixed. And we know this because anyone who's not doing it already is not suited to do it. Whereas my mentality is the opposite, I would take an abundance mentality. I would say there are many people who have these successful careers who today probably look very entrepreneurial because they figured out the game and they figured out how to get the promotion and the promotion and the promotion until they're very successful that had you place that same talent, that same ambition in an entrepreneurial environment.

They'd have figured out a different utility function and could have been wild as successful entrepreneurs. And so, one of my beliefs is that, as I said, the supply is not fixed and the number one thing we can do is try and increase the supply. And the number one way we can do that is by effectively providing environments that make it more legible, more acceptable, more celebrated to be a founder. And the thing, again, sorry a long answer, but I've believed this for a long time, that's why I created EF with Alice a long time ago now.

“One of my beliefs is that the supply [of founders] is not fixed and the number one thing we can do is try and increase it by effectively providing environments that make it more legible, more acceptable, and more celebrated to be a founder.”

But it's been pretty cool over the last few years to see some academic evidence for this. So, there's a pretty cool paper that came out couple years ago that I wrote about at the time, which basically looks at what happens in years when banks hire fewer people. And when I say banks, this is a UK context, it's talking about the banks that are the sort of employer of choice for a lot of the most talented people, whether or not they ever thought they wanted to be bankers.

So in the UK, the financial services industry is this massive talent suction machine. And so, what it looks at is when banks just reduce their hiring, what happens? Do more people become founders? And the paper basically has two conclusions. One that I think is really intuitive and one that is not intuitive. The intuitive one is when banks have to hire fewer talent and ambitious people, more people become founders. And those founders are more successful on average than people who became entrepreneurs by choice in the period.

In other words, the exact opposite of what we'd expect if you believe that everyone who is going to figure it out has figured it out already. And I think the reason, I think, that's true is that in reality, we are so constrained and shaped by culture, by the culture we're embedded in and the sort of water we're swimming in that it's hard for us to imagine that because we think of an Elon Musk and we're like, "How could Elon ever ended up being a, I don't know, partner at McKinsey or a managing director at Goldman Sachs?" And I buy it, he specifically probably wouldn't.

But I don't think it's crazy to think that there are lots of people who actually just got so absorbed in the game that they found themselves in that it actually wasn't obvious to them. And so, I believe if we can get a lot more of these lost founders who end up in the wrong system, winning the wrong game, as we often say at EF, then we can increase the supply of really great founders. And that's what we do at EF. We want to be the natural home of people who are like, "Wait a minute, the game I'm playing is not the game I want to win."

MM: I really love that. What strikes me about your approach is it's raising the baseline from bringing more people in who wouldn't otherwise be thinking that way. And then also, encouraging and creating the optimal environments for the Elon Musks and the people who will be outliers. And that seems like a really, really poignant strategy. Instead of focusing on the middle and being like, "Let's just make it incrementally better in small ways," when those people would probably be filtering that way anyways, like you said to Silicon Valley.

MC: Yeah. I think that's right. What I'm absolutely not saying is we should just be cheerleaders for entrepreneurship. Everyone should be an entrepreneur. You're in a bank, you should be an entrepreneur. I'm not saying that at all. Actually, I think, in general, advising people to be founders is terrible. You shouldn't do that. However, at the very upper end of the ambition and talent distribution, there are people who would be great founders. And a lot of the skill we think in what we do at EF is trying to tell the difference between the very smart, the average banker in the UK is very smart. It's a very selected full profession.

It doesn't mean they should be founders. We're trying to find those outliers to use the word you introduced earlier. And our view is that entrepreneurship is a high skill profession. It's for only really exceptional people thrive in as founders. And so, it is really about saying, if you live in a society and a culture that every time you do something well and you excel, it nudges you further towards taking a particular career path. That career path is going to be, I'm not just talking about banks by the way, it's true of all anything that attracts talent. You're going to find these tracked careers are full of people who were actually brilliant but sort of just didn't know that this was an option.

“Tracked careers that attract talent are full of people who are actually brilliant but just didn't know that this was an option.”

And when I look at the very best founders that we've worked with and you're like, "Why didn't you do this sooner?" Very often it's just that, "I didn't know it was an option."

MM: Yeah, 100%. What do you think once they realize it's an option, what are the most common bottlenecks that appear after that they have to overcome to become a great founder?

MC: Yeah. This is not designed to be a pop, but Alice, my co-founder and I wrote a book more or less exactly about this question, which came out last year. It's called How to Be a Founder. And the premise of the book is there's tons of books about startups out there, but they all sort of start from the perspective that you start the company and you want to make it better. Whereas our book starts from the premise of maybe you're not even sure you should make the leap into it. And the reason I bring it up is that whenever I get reader feedback and Alice gets reader feedback on it, it's almost always the bit that people pick out is the bit at the start where we're like, here are five myths that hold people back once it's on their radar.

And I won't go into them all, but I would say in general, a lot of the way that entrepreneurship is depicted in popular culture, in my view, is deeply unhelpful for helping people decide whether this is something they should try. So just to use one example, well, I'll use two because I think they're both very dominant in what we hear. One is this idea of readiness. Like, oh, I'm not ready to be a founder. I now understand I can do it and I think I probably will do it one day, but I'm not ready. Now of course, in the limit that's probably true, right? There's a point at which you shouldn't be.

And experience helps and experience always helps. But this is the problem is that I think in cultures where entrepreneurship is not normal, people get very quickly into this idea that there's like, if only I knew X or if only I had experience doing Y, then I would feel ready and then I would start. And of course, the problem is once you've got X and Y, you discover A and B and there's always something that, yeah, if you had more experience you'd be ready.

And the big thing that we argue at EF in our jobs and in the book is that you shouldn't think of your ability as a founder as some sort of analytical fact about yourself. Something that can be ascertained if you do right, ask the right questions and fill out the right survey. It's nothing like that. Your skill as a founder is developed by the act of getting started. In video game language, it's procedurally generated by the work that you do. And so you're never going to feel ready. If the bar you hold is like, oh, I don't know enough about X. No. Guess what? You and you never will.

In fact, the whole point of startups, the whole reason for startups have an opportunity to exist is because they're dealing in the unknown. If they're dealing in the fully known, a big company would be doing it already. I think this idea of readiness is, I think, is a very corrosive myth for people that is now on their radar. The second one, and I will stop after this when I went to all five, is about just feeling that you need to have the perfect idea. And I think the number of people that would rather weekend after weekend polish a Google doc with a beautiful business plan and idea, rather than just get out there and talk to customers is extraordinary.

And I think there is this, if you look at films and books and TV show sort of depictions of entrepreneurship, there is this idea of the light bulb moment, the fantastic idea that it hits you and then what you want to do. In my experience, approximately zero companies, especially the great ones come from that. They usually start from, you either observe a problem or you build a thing and there's some organic pull on it and the organic pull pulls you into a set of conversations and brings you a set of resources that allows you to imagine the next phase, which in turn gives you a set of conversations and resources which allows you to imagine the next phase.

And it's much more the ambition and vision is a product of getting started rather than what gets you started.

MM: I love that. Are there any stories in particular that you think do a good job of depicting entrepreneurship in an accurate way?

MC: Well, I think one of the reasons, I mean I think there's lots of problems with it, but I think one of the reasons that the film The Social Network had such a big impact is that it does sort of push on two things that are really important and potentially seem contradictory if you've not thought hard about them. One is that Zack is clearly an accidental entrepreneur. He builds something and there is that pull and it brings, again, as I just said, a set of resources that allows him to take to the next level. And he's clearly exceptional. He's clearly an exceptional person.

And I think however many years on it's clear, well 20 years on that, he's clearly exceptional. I mean he's been able to ride that journey. And I think often people think of those two conclusions as opposite. Well, if you're exceptional then you must have got it all figured out on day one. And clearly actually what The Social Network shows us is he wasn't lucky. There's lots of people that could have had the pull and not been equal to the challenge. Equally, he wasn't fully intentional.

I think often when we meet people who are like, I think, I have this feeling that this is what I have to do, I should be a founder. I just don't want to be playing this game of climbing the corporate ladder. They're like, but what I need to do is really either think harder about an idea or I need to go and work in place X in order to learn about what thing Y that I want to work on. It's like, yeah, maybe. But I think, again, we're so influenced by other water we're swimming in.

“We meet people who say: ‘What I need to do is really either think harder about an idea or I need to go and work in place X in order to learn about what thing Y that I want to work on.’ It's like, yeah, maybe, but we're so influenced by other water we're swimming in.”

One of the things I'm proudest of EF is that it creates an environment in which people can just explore. And obviously, a lot of the time when they explore, they don't find anything. And that's obvious natural. Startups are weird and find something that's actually worth spending 10 years of your life on is weird. But if you are still in your job, how on earth are you going to find out whether or not you have the perfect idea? You just can't do it. You almost have to make the choice first that you're going to start exploring and start pulling on threads that interest you in order to be in the water that allows you to make that decision.

MM: It's really well played. It reminds me too, I think that a lot of times when people make excuses about not being ready or needing more experience before they're ready to make the leap, it's really because they're just afraid of failure or getting rejected or messing up. And so anything that you can do to make environments where actually taking shots on goal is more celebrated than just staying stagnant. And it's totally fine to fail as long as you keep trying or make a calculated decision and really learn things from that. But it's hard to do.

And so I’m curious, how do you endeavor to create that culture? Is it just having critical mass of a certain mentality and really building it from the get-go or how have you approached it?

MC: So I think certainly, I think critical mass is a huge, huge benefit. You're absolutely right that people are afraid of failure. And when you really push people on it, it's almost never commercial failure. Maybe 10, 20 years ago people worried about, oh, what if I go bankrupt? I think I literally never hear anyone say that today. What they really mean is social failure. What they really mean is, I'm going to quit this job that I competed for really hard, that impressed my parents and my peers when I got it. And I'm going to go and instead of doing this thing, which even though I hate it, it creates legible achievement points for my immediate circle.

I'm going to do this thing that has no legible achievement points where there is no bath, there's no pray, there's no mark scheme, all these things. And so, I suppose what we have always felt was if you could change the peer group, that would be a massive win. If the people around you are suddenly actually if the social currency rather of the people around you is how ambitious are your attempts rather than did you get the badge? That's a big win.

“The social currency of the people around you is how ambitious are your attempts rather than did you get the badge.”

But also, I think, there was something really special that happens when everyone in the room knows that everyone else in the room gave up something really valuable to be there. And I think what it does is it means that you've already done the hard bit. You've already got the kudos of being the ones that tried. Now I don't think you should valorize trying too much. I think you still got to valorize succeeding. That's important.

But I think there is a thing where you're like, how do you create, it's really short, I think we give people about 12 weeks to decide whether they've got something and for us to make that decision too. How do you maximize the benefit of that 12 weeks? Partly by finding ways to forge bonds really, really quickly. And I think the sense that everyone in that room could be making more money, getting more badges, impressing their parents more, doing something else. It's this shared identity on day one that I think is really, really powerful.

MM: Yeah. Reminds me a lot of small group theory too, filtering for people that have some common thread between them all that ties them together. And there's like some shared values as well.

MC: Yeah. And I think one of the hardest things that we do at EF is help people form co-founding teams from scratch. So start companies with strangers effectively or very nearly strangers. And you, again, talking about conventional wisdom, this was probably, and maybe still is, the bit that experienced people from startups when we got started were like, that's the bit that's never going to work, so you should just drop that bit right now. Yeah, and I totally get why, again, I think there's a lot of selection bias like, of course, if there's never been a systematic attempt to help people start companies with strangers, then if you select on who are the people that I see successfully starting companies, it's going to be people starting companies with people they know. I mean that's just like a very obvious statistical artifact.

But that said, I do think it's true that there is a really hard problem to solve if you are starting a company with a stranger, which is have you any idea whether there's a compatibility there? The controversial bit I believe is that the same question actually applies to start a company with anyone. When someone's like, "Oh, I'm going to start this thing with my best friend from work." I'm like, "Cool, have you ever started a company together before? Because if you haven't, I kind of don't care about any previous experience you have social or work."

Because starting a company is really different from anything else and it's going to put pressures on your relationship and on your ways of working and the demands on you that are totally different. And so, what we realized at EF was the stupidest thing we could do, which is what we did for two years because we largely started doing stupid things, is to try and think of this as an analytical problem where you make people take quizzes or let people sort of spend loads of time getting to know each other and go for drinks together and figure out who they're friends with.

And they'll pick the people that are the closest proxies for the people they would've picked had they been to school or work with these people before. That was turned out to be totally wrong. The thing that turned out to work was to abandon any pretense that this is an analytical exercise, instead just really lower the social friction to just getting started, but also really lower the social friction to just stop it. So, a failed team at EF might last 48 hours in the sense that you're literally like this is a poor people who've all given up something valuable to be here. They've all come to join this community because they're very ambitious and very talented and they want to start companies.

“The thing that turned out to work was to abandon any pretense that this is an analytical exercise and instead just lower the social friction to getting started and stop — a failed team at EF might last 48 hours.”

The opportunity cost for being in a bad team is really, really high. You've only got 12 weeks. And so, all we do now is we say, "Get in your first team in the first few days and start working." And really the only way you should evaluate whether or not this is a good team is are you productive? Are you getting more done than you would expect given what you know about yourself? If you are, keep staying in the team until that's not true. If you're not, break it up, get out, try someone else.

And actually, what we find is that when you lower that friction to getting out of the teams where it's like just norm, everyone knows it's fine if you broke up, everything gets really easy for getting into teams. And so people actually try a lot of things. And so, the one thing that everyone thought, including ours I have to say, will break EF. The idea that the teams will just be really non-resilient after they graduated from us has turned out not to be true. At least, the data I have suggests that EF teams that raise a seed round are less likely to lose a co-founder before series A than at least London based teams that form organically in the wild as it were.

MM: That's amazing. It reminds me of a micro version of the stats on entrepreneurship being mainly just about how many shots on goal you had. It's the same with trying a bunch of different formations of team. Because people, yeah, you don't really know actually how the combination will work and how symbiotic it'll be. And then, at a larger scale when they're actual companies, whether there's a real market in demand there.

MC: Totally. Totally. So much learning by doing, right?

MM: 100%, yeah.

MC: Anyone who tells you otherwise say, "I don't get it." The whole point of working in the realm of deep radical uncertainty is you usually just have to try stuff to learn anything.

MM: And it takes a lot of courage to have that resiliency. Yeah.

MC: Yeah.

MM: At a higher level, I'm curious, if you could change anything about how the government operates or institute any policies that would help encourage entrepreneurship, what would you change? Or what would the policies be?

MC: Yeah, I mean, again, a risk of just having a single theme. I think the single biggest thing that let's take a UK or a US government could do, because they're the places I know the best, is just make it really, really easy for the world's best talent that want to come to those countries, to build companies, to do so. I mean, the US in particular having Silicon Valley is the world's great asset of having, sorry, ecosystem asset, right? I mean, it's just extraordinary.

So if there are tens of thousands, even hundreds of thousands of people who are incredibly talented, adding to that ecosystem will increase the value of the ecosystem. It's the world's best free launch to let them come and do it. And I have to say one thing that has been pretty good in the UK over the last decade is making it easier and easier for entrepreneurs from outside the UK to get visas to come here.

We fed into that a little bit at EF and I think the exceptional talent visa in the UK, I believe is the best visa in the world for founders right now. And it's really good. But I think any government that is serious about entrepreneurship basically just has to have an open door policy when it comes to extreme talent, particularly extreme technical talent and founder talent. But that's sort of harder to define. So that's one. I think, in general, the other things are things about make it really easy to invest in startups, certainly not tax punitive to do so. Make it really easy to be an employee in a startup, particularly when it comes to options.

Basically just make sure the incentives are really, really well aligned for talented and ambitious people to make their bets in this sector versus something else. But ultimately I see these things as talent challenges. Again, hammer, nail, et cetera. But it's striking to me when I look outside the UK, how many countries make it really hard for great people to either move there or stay there. I mean, the craziest ones are where a country will take grad students in technical subjects, give them degrees and then be like, "You can't stay." These are the people that are going to build the things.

To me, that is such low-hanging fruit that almost everything else feels like an afterthought. But I guess if I had to point to other things, and again, bias because it's partly what I do. But I do think investing in fundamental R&D, everything we know about the economics of innovation is that investing in fundamental R&D is really, really good for long-run innovation. And so making sure that you have a really great academic sector, particularly in science and technology, is really worthwhile. Creating the incentives that make it really easy to spin breakthroughs out of universities into companies that don't mean that the founders on day one already own less than half the company and all this stuff. The somewhat low-hanging fruit there as well.

“Investing in fundamental R&D is really, really good for long-run innovation. And so making sure that you have a really great academic sector, particularly in science and technology, is very worthwhile. Creating the incentives that make it easy to spin breakthroughs out of universities into companies.”

But in general, I would say, if you've nailed basic talent access and basic capital incentives, you've probably done 90% of the work.

MM: Yeah. Do you think that there's any areas of improvement in terms of the venture model? Because venture very much is incentivizing the 10X outcomes, these huge innovation outlier outcomes, and that's wonderful. But I think that there are a lot of things that don't fit into that model that I think venture is becoming more self-aware that perhaps they're missing things because of that. But I'm curious to hear your thoughts on that.

MC: Yeah, no, I mean it's a great question. I think it's one of those things where there's a joke about hedge funds that hedge funds aren't really an asset class. They're just a compensation scheme that's been generalized. And there's an element of truth in that. But the serious insight behind that joke is that it is an incredibly valuable thing for an asset class to be legible to capital allocates. So the really great thing about structuring yourself a venture fund, if you want to fund innovation of any kind, is that LPs understand it. And I don't mean that at the level of, oh, do they intellectually get it? I mean, LPs are smart people.

I mean, does it fit into a sell in their grid when it comes to asset allocation? And for large pools of capital, of course they have to view the world somewhat like that grid. They generally are not going to do things that they can't cover systematically. And so, one of the reasons that venture ends up taking on a lot of things that don't really fit very well with venture is because venture's super legible to LPs. And so you can raise money to do it.

“One of the reasons that venture ends up taking on a lot of things that don't fit very well is because venture's super legible to LPs. And so you can raise money to do it.”

I mean, I have learned that both the hard and the, I don't want to say easy way, but in both positive and negative sense that EF is not a very natural fit for a traditional fund structure because so much of what we do to earn the equity we get is not the cash we invest but the work we do. And we spend 10 years trying to retrofit that into a fund structure. And it sort of worked. And we've been lucky to have great LPs and also deliver really great returns to those LPs.

But last year we were able to restructure as a permanent capital vehicle where all the capital we raise from investors now going forward comes into the company. And that allows us to be much more flexible in how we think about our operations as a source of value. It's very hard for a fund to think of operations as a source of value unless the assets under management is so huge that the management fee is a corporate-sized revenue stream. But EF isn't that scale. Very few funds are.

And so, we found that we were very lucky to find investors that resonated with who were at least many of the investors in the funding round that made that possible were not traditional LPs. Some were as well very forward-thinking ones that were very lucky to have. But finding first principles, thesis-driven, pools of capital that are willing to try new structures is hard because, in general, you don't want something to fail because of the structure. Because in the same way that no one got fired for buying IBM, no one actually got fired for buying 2 and 20 with a 10-year closed-ended fund cycle like that. That's right. That's the normal thing.

Sorry, I've sort of slightly got away from where you started, but my broad point would be, in general, if capital is the bottleneck, the hard thing is finding initial investors that can help you validate structures that are not the classic 2 and 20 10-year closed ended fund.

And so I think it's great to see that there's a lot of people that made a lot of money in tech in the last 10, 15 years who are actually really key to do first principles thinking about the right structures for these things. And that's why if you read the list of people that kind of backed EF's series C last year, that allowed us to do this transition to this permanent capital model, it's largely a list of entrepreneurs. It's people like John and Patrick Carlson, Reid Hoffman, Matt Mullenweg, et cetera, et cetera, they're not constrained in the same way that an institutional LP would be. And so I think that's really great.

I would love our billionaires to do more non-traditional structure in the way that they invest because it's what eventually pulls the rest of the capital allocation world towards things that today look quite niche.

MM: It's all about the incentives you design for yourself. It is interesting too because I feel like venture, in general, it's just like a funding model, but then you can do what you will with it, but you do create incentives by the capital you take from and the expectations you set more than anything else.

MC: Yeah. And it will be interesting whether things change a bit. As VC goes through this huge transition of getting used to a non-zero interest rate world because part of the problem has been frankly that it's been too easy to make money in VC. And I mean that in a literal sense, as in people thought they were making money that they were not making. And therefore, things like funding hardware, funding deep tech, funding anything that's going to take a long time, became very, very hard because it felt like the opportunity cost was free money investing in SaaS. It turns out that money was not free.

So you can tell a negative story that maybe things will get worse because there's going to be a general sort of cyclical outflow of capital from VC. But you can tell a story that it will get better or it will get better for stuff that is not mainstream today in that the opportunity, the perceived opportunity, cost of funding weird stuff has actually reduced.

MM: That's a great answer. How do you see the next 10 years of EF playing out and how does it fit in?

MC: Yeah. So we sort of feel like EF is just embarking on the start of its second chapter right now. And it really was that flip to a different structure that enabled it in the ... I think where we've got to is we really believe that the category that we've created is going to be extremely valuable. And the next 10 years are really figuring out what is the scalable model for ensuring that we capture the full benefit of that. And I think what we've come to the conclusion of is that we probably thought about it too much in a top-down way before.

We created this product. By which I mean the system of EF. And then we expanded to other countries by hiring great operators in those countries, giving them the product saying go to the races. And actually, I think, we've just made a big switch in that we now would say the really scarce resources are the people that can build the bottom-up relationships with great talent really early in their career.

It's not a plug-and-play machine. It's not a machine where you crank the handle and kind of good people in, good companies out. If you really want to build needle-moving companies that start by forging incredible relationships very early in people's careers and being genuine career coaches for people as they decide even whether they should be founders, never mind how to be a founder, and then taking them on that journey.

And so, we've done a big restructure at EF over the last year where we've moved away from this top-down model and really gone all in on this idea that we want to create a new career path for people that we will call talent investors. And it's sort of unlike, I think, it's a little bit like being a VC, it's a little bit like being a talent agent. So a little bit like being an operator, but it's really none of these things. It's a new thing, which is finding people that will be the thought partners, counselors, and coaches to the next generation of Patrick Collisons’ and Elon Musks’ and Matt Mullens’ and all these people.

And I think a book I would really recommend to all your listeners is this book that came out, I think, last year called The Power Law by Sebastian Mallaby, which is a history of venture capital. And it's great. There are loads of great things about the book. But the thing I loved about the book is you realize for the first 20 years of what we now call VC, everyone was sort of feeling around in the dark. All the things that are now, like completely rigid structures of VC were off for grabs, and people were feeling around it.

And one thing I often say to people who apply to work at EF as talent investors is we've been doing this for 10 years, but if you look at the history of venture capital 10 years in, no one figured anything out. That's where we are right now. We've done enough to earn the right to be here. But in terms of where we'll be when someone writes the book about talent investing, we're basically equivalent to day one for the first 10 years of the rounding era. And the exciting thing for me is you read that book and you read about the impact that people like Mike Moritz have had on that whole asset class.

And you say, "Who's the Mike Moritz of talent investing?" Well, it's not me. I mean, I hope that my investments today turn out to be great. But my guess is that Mike Morris of talent investing hasn't yet applied to work, but they will. And I think they'll have a comparable impact on talent investing that Mike had on venture capital and that's what I'm excited about the next 10 years, is actually finding the next generation of people that make my track record, which I have to say is actually pretty good, but just by out of water. And that's what I'm really excited about right now.

MM: What a wonderful answer. Well, thank you so much for coming on the podcast, Matt, it was a pleasure to have you.

MC: Molly, thanks so much for having me. I really enjoyed it.

Subscribe for more episodes to come: Spotify, Apple, Google, Amazon, and Overcast.

Moth Fund
Moth Minds
Interviews with high-agency humans