Paul Miller: Welcome to Tech for with Bethnal Green Ventures. I'm Paul Miller and today I'm delighted to be joined by Tim O'Reilly. Tim's the founder of O'Reilly Media and one of the most influential thinkers in the technology industry from helping popularize the concept of Web 2.0 to shaping how many of us think about the relationship between technology, business and society. Now, back in 2008 and early 2009, Tim gave a series of talks and wrote an essay called Work on Stuff That Matters, which was really influential, I have to admit, particularly to the precursor to Bethnal Green Ventures, what we call social innovation camps, where we brought people together for weekend hack events to build prototypes of things that could solve stuff that matters.
Tim O'Reilly: Yeah.
Paul Miller: And it was a response to the financial crisis, but it quickly became something broader, almost like a manifesto for a different kind of industry. nearly 18 years on, ⁓ we it would be an interesting moment. ⁓ Now going through such massive technological change again ⁓ to the idea. So Tim, it's great to have you on the show.
Tim O'Reilly: Thanks so much for having me and I'm glad that I had an influence and that you took the ball and ran with it.
Paul Miller: So just to start off, let's rewind what was going on in 2008. What was it that prompted you to do those series of talks and to write the article?
Tim O'Reilly: Well. Well, was actually the specific on stuff that matters framing grew out of an earlier talk I gave called Why I Love Hackers, which I had given at ⁓ our E-Tech conference. ⁓ And was about how all these people who history forward because they're trying things, ⁓ just for the love it, for curiosity, and that that's a much more important starting point in the traditional narrative from Silica Valley where everything begins with entrepreneurs and venture capitalists. my experience of early ⁓ early open movement, these were people who were motivated by love and excitement and enthusiasm and not by, wow, we're going go out and make big buck. ⁓
Paul Miller: Yeah.
Tim O'Reilly: And so then, then ⁓ I gave it, I gave this talk and ⁓ Palka, who's now my wife and became a social entrepreneur, came up to me. She was starting this new conference in partnership with a rally called the Web2O Expo. And she said, I need a talk like that for my conference, but I need it for entrepreneurs, you know, not about hackers. And that's when I came up with the talk that became why, ⁓ so it became work on stuff that matters. And then I wrote the essay. specifically I was reacting to, you know, there was this sort of crazy period of social media. I where people were getting feted for ⁓ social media apps that were incredibly stupid. And it a kind of a period of hype. You know, did Dickens say? It was the best of times. It was the worst of times. In other words, a time much like our own. ⁓ You know,
Paul Miller: Yeah.
Tim O'Reilly: You know, that was that period. And then you had the whole metaverse thing coming out of meta and it was all clearly designed to drive a hype cycle. Now I think today's world with AI, there's a lot more substance to it, but I still think that Silicon Valley has lost its way in the sense that, you know, certainly the initial entrepreneurs who sort of waded into this AI. technology and started building towards it are, you know, we're profoundly motivated by curiosity and just the pure, the wonderful hacker research ethic. But it very quickly became so overloaded with capital. And I think in general, since 2000, the 2009, you know, 2009 financial crisis, when the response was to have near zero interest rates, capital became really cheap. You were looking to deploy massive amounts of it. Silicon Valley switched from a model in which entrepreneurs were really looking for product market fit. And instead they were doing what Reid Hoffman called blitz scaling. You that is you buy, you know, just move as fast as possible, seize market share. Don't worry about whether you actually have a workable business. You'll figure that out later. I just need the high ground. And, you know, and I thought of this in particularly this happened with, uh, you know, the ride hailing apps, uh, Uber and Lyft, you know, the actual technology was invented by a guy named Sunil Paul, who created a company called Sidecar. And he raised $35 million, which just coincidentally was about the same amount of money that Google had raised and all it took for Google to succeed. And Sunil's the guy who actually, ⁓ mean, was an SMS app for calling black cars for rich people. Rift a web app for putting together intercity carpooling. The VCs saw what Sunil was doing and they piled on, literally gave them billions of dollars. He was quickly out of business. He's the one really invented that he used GPS to locate the phone. the implications of what was really happening with maps and mobile. He was the investor. He got just wiped out. And basically, then they rushed to get market share by ⁓ capital-fueled capital-fueled marketing. And then of course, once they got to their IPO, was like, well, actually we have to raise prices, the service will get worse.
Paul Miller: Yep.
Tim O'Reilly: What was ⁓ casualty was the kind of experimentation that you typically see in an early period of technology. My guess is absent that, don't know that, Sunil probably wouldn't still have won. He wasn't the kind of cutthroat entrepreneur who typically does emerge from the pack. But you would have had lots of local companies adopting the technology. You would have had different business models that were ⁓ playing out. And this is as if somebody had come in at the early days of the web and anointed AOL with so much capital that nothing else could be funded. ⁓ So that was sort part of the framing of that time. And I worry that we're in that same kind of moment today where of users, the market of customers picking the winners,
Paul Miller: Yeah. Yeah.
Tim O'Reilly: capitalist picking the winners because you're building this enormous capital moat and it's not sure that it's going to play out that way because you know there's often a bet that goes wrong when you do that. You know the market is actually smarter than the venture capitalists which is why it's better when you actually the gradual ramp up to success rather than somebody right at the outset deciding that
Paul Miller: Mm-hmm.
Tim O'Reilly: you know, whether it was Netscape back in the day or, ⁓ you know, or OpenAI is the winner and you just give them all the money and let them take the high ground. ⁓ know, you have this period of experimentation and that works better. And we're seeing that with AI, for example, where people are actually, you know, inference costs starting to crush the whole economics of the thing. And it's becoming clear that this is not a viable business model as ⁓ constituted. even companies like Anthropic, which is, I think, really well at finding market and focusing on it and really serving it, is hitting real capacity And there's a big disjuncture between the ⁓ experience and narrative, which is good because it may mean that ultimately we'll get a more decentralized AI economy.
Paul Miller: Yes.
Tim O'Reilly: My experience, and I've talked about this from my early days of advocacy for the commercial internet in 1993, source in 1998, Web 2.0, was all about how we build a more decentralized technology economy opposed to one that's centralized by market capital, moats. And I think I still really believe that. So a lot of what I'm really talking about now is, what is the architecture of participation for AI? How do humans actually, what's the human economy look like? You know, I see someone like, you know, open and anthropic, all they're talking about is we're going to put all the, out of work. I go, well, it'd be your customer then. And, and I see them not even thinking about.
Paul Miller: Hmm.
Tim O'Reilly: the right problems. And so I'm trying to encourage them just like I was in that work on stuff that matters period. Although there I was really, sorry, back up a little bit there. When I said work on stuff that matters, what I meant was you're probably going to fail entrepreneur. So work on something you will have been proud to have done that needed doing. ⁓
Paul Miller: Yeah, yeah, because I mean, I think the first heading was work on something that matters more than money, wasn't it? Yeah. what did that feel radical at the time? Did people push back or?
Tim O'Reilly: That's right, that's right, that's right. No, I think there were a lot of people who loved it because I there's a real hunger for idealism and optimism out there. And yes, there are people who are just sort of totally obsessed with really honor wealth ⁓ in some You've just experienced in Silicon Valley where people come and they don't tell you about their customers, they don't tell you about even their revenue from customers, tell you about their fundraising. You know, it's just, it's this game, you know, where people are placing bets and it's, kind of a, a meme economy. And this is all the way up to the top, you know, Elon Musk is the world's richest man, not because he actually has built a company that makes a ton of money. He's just works the, the memes, you know, it's like all the other car companies are worth, you know, uh, don't know, 10 or 15 times earnings. Tesla's at its peak worth 1,500 years of earnings. And now you look at all the hype he was pushing about robots. And now that he got SpaceX as this new carrier, he sort of said, that really was harder than we thought moving on to the new hype machine. I wrote a piece that I wish more people would read, maybe. four or five years ago called Two Economies, Two Sets of Rules, is really focused on the idea that if playing with capital, it's not just that you have this advantage that compounds without labor and that it magnifies the returns, but also it's borrowing from the future. The fundamental fact of capital,
Paul Miller: Hmm.
Tim O'Reilly: is that people are betting on a future stream of earnings. That's why even in the boring industry, a stock is worth 10 or 12 or 15 or 20 times annual earnings. They go, well, we'll keep earning for some number of years into the future. You do some discounted cash flows and you say, this is what it's worth. And so you buy the stock. But once that becomes completely unmoored and you can have someone like Elon Musk, who's literally making hundreds of billions of dollars, promising that this is a company that pay itself back in thousands of years, ⁓ completely on board. You're very on board. so it's a hype story. You compare any of enterprises to the actual giants of technology. mean, obviously, ⁓ NVIDIA right now way out there, but Google and...
Paul Miller: Yeah, jam tomorrow as we call it.
Tim O'Reilly: and, ⁓ you know, Apple or whatever, there maybe 35 times earnings, you know, and Musk there hundreds of thousands sometimes. And, and, and it's, anyway, that's sort of a little bit of an aside, but the problem is that it has real social implications because there's a class of people who basically for whom a dollar that they earn through their labor is worth a dollar. And there's people for whom a dollar saved. in labor costs, for example, or a dollar of profit, which is valued at some multiple of that dollar. So effectively, the reason why companies are relentlessly optimizing for profit is the executives are paid in a different currency, which has a 10, 20, 30, 50, 100 times multiplier. So of course you want to save that dollar, you know? because that dollar may become $100, may become $1,000. And this is actually the mechanism by which ⁓ economy has ⁓ wrong. Because it's not just that, ⁓ AOC says, ⁓ there shouldn't be any billionaires. I don't believe that at all. There are people who totally their billions, right? But there are a lot of people who have...
Paul Miller: Mmm.
Tim O'Reilly: you know, are participating in this levered game, which is fundamentally unfair. And, you know, and so it's just this idea that when you realize that, okay, we have asked our companies, first of all, we, this is to me, I wrote a piece again, back in, I think it was 2020 called, we've already let the genie out of the bottle. And it was really, you know, corporate governance is our test case for AI governance. You know, and, uh, you know, what do we learn from it? And what we learned from it is, you know, very quickly, you know, we get, you know, basically starting in the early eighties with Milton Friedman's, the social purpose of a business is to make money for its shareholders. And then the shareholders will decide what to do with those profits. That quickly became Meckling and Jensen at Harvard kind of saying, Oh, well, the way to do that is to align the interests of the management with the interests of the shareholders by.
Paul Miller: Hmm.
Tim O'Reilly: stock-based compensation. And then all of a sudden you have everybody subject to what isn't effectively a rogue optimization function, you know, because they've said optimize for profit above all else, you know, and I'm kind of like, you know, we used to have something that was a little bit more balanced and now we really do have ⁓ an economy that we're literally the obligation supposedly of a company is simply to make as much money as possible for shareholders. we built a mechanism in the form of stock-based compensation by which all of the incentives ⁓ slave to that. So I go, well, okay, let's just imagine AI wants to rule the world. ⁓ already built the mechanism for that, ⁓ ⁓ basically, ⁓ not saying that AI will want to rule the world because I think there's a level ⁓ hype there and you already see. it is, I think it's Henry Farrell and Alison and Cosmo Shalizi and James Evans say AI is social and cultural technology. It's basically, it's taken in human values. ⁓ you see ⁓ not just the company, but Claude, the AI pushing when it's asked to do certain things. we're really in this interesting place where, I mean, it's always the case, know, things that seem terrible are also kind of wonderful because, you know, we're in this, you know, crucible where we're figuring out the future and, you know, and the future is of having these new, vastly powerful technologies that we don't quite know what to do with that are going to change our culture. And so I always think, you know, historically, and I go, yeah, you know, would we take back, you know, the printing press and widespread literacy, despite the fact that it led to hundreds of years of religious warfare? No, we wouldn't, but we certainly, if you were in those times, you might've wished to. so you understand some of the conservative impulses, like the future is coming.
Paul Miller: Hmm.
Tim O'Reilly: As Alvin Toffler said, the future always comes too fast and in the wrong order.
Paul Miller: Yeah. So, I mean, where do you see the kind of interesting people working on stuff that matters today? Are there sort of pockets that you've noticed in terms of people really thinking, maybe outside of that model that you've just been talking about?
Tim O'Reilly: Well, I mean, first off, I think there are people who are forced by necessity to work on stuff that matters and it's not necessarily do good stuff. You know, I mean, you look at, you know, Ukraine fighting for its survival they're pushing the future forward. ⁓ the future of warfare. think there are people in some of the big companies. do think that both DeepMind and anthropic really, you know, have very serious commitment to sort of using AI to try to, you know, make advances in healthcare. And so, you know, I don't think it's all, ⁓ all dark, even at the big labs. I, I, I'm actually more, jaundiced by, you know, some of the, ⁓ the, the smaller companies Silicon Valley, they're just jostling for their place in the trough. But even though there's just, there's always entrepreneurs who, kind of are, ⁓
Paul Miller: Hmm.
Tim O'Reilly: are driven by an impulse to serve a real audience. I don't think of AI, or sort of technology for good, just do-gooders. Although I do think there is a lot there. mean, Jen started Code for America, started the United States digital service, now she's sort of trying to work on government transformation. That's a classic do-gooder arc, but with a very different. you know, spin on ⁓ it is, but there are a lot of companies that do that. But I think of, for example, 11 labs, you know, when I talked to, to Maddie, the founder, and he's, he's just like, you know, he and his co-founder were like, they're in Poland and they're like, you know, you don't realize in Poland, you know, it's a small market and the movies are just dubbed by one actor, you know, all voices. And we go with AI, we could do better than that, you know? And that's like, that's a kind of AI for good. that where they're just going like, this is an underserved market and we could really make something better. And I do think that that thread does run all through Silicon Valley and it's best work. is somebody saying, wow, we could do better. But I think this also, what I'm seeing in a way that is AI is unlocking a kind of sort of bottom-up community hacker innovation again. I've seen, you was in the early web. was in early open source, but it really kind of got rung out of the system in the last decade AI kind of blew things open again, because there were more and more the big companies really had everything locked down. so, you know, there's that, you know, so a big part of AI for good is just turn people loose. And yes, will be a lot of people chasing the money. but there will also be people who are just like curious to solve problems, you know, and ⁓ learned with others and yeah. ⁓
Paul Miller: Yeah, I mean, one of the things I find really interesting. One of the things I think we find really interesting is that AI enables people who really understand problems to build solutions. So you've got this ability that actually almost, I think when we started working on this area, you'd have the people who understood the problems and then the engineers, like kind of as separate camps almost, but now those two can be much, much closer together AI, which is, ⁓ think is an interesting innovation moment where you've ⁓
Tim O'Reilly: That's right. Right. Yeah, it is. think there's an enormous opportunity for who are close to problems to solve them. And think that the capability, and I look at Jen's work, for example, ⁓ with government, think about the bad version AI. By the Henry Farrell has a great essay about this right now about ⁓ AI and Doge. the vision of as some kind of of top-down control, which embodied in Doge and the vision of ⁓ people Jen, is, no, AI is an enabler of people who are trying to do the work. And so actually bottom up. If you can actually enable, ⁓ in her she's looking at government, if you can enable people who are in public service, who really want to serve the public, to give them new tools to do that better, that's really powerful. And, and, ⁓ you know, she's, she wrote a great piece about what she calls horizon one, which is kind of the status quo horizon two, which is sort of short-term improvements. and then horizon three, which is really changing the status quo. she says, look, a lot of the problem is that a lot of technology is just keeping the old status quo alive and not utterly transforming it. And think that we, you know, ⁓ one the interesting challenges in the AI for good space is to think big enough.
Paul Miller: Yeah, and actually that's just one of the references you made in that original essay was to the Long Now Foundation and Where is long-termism in the ⁓ industry at the moment? ⁓ it disappeared or what's thinking about that these days?
Tim O'Reilly: Yeah. Mm-hmm. Yeah. ⁓ Well, I don't think it's the tech industry that's the problem. It is literally the design of the entire system. is baked in to our tax code. It's baked into all ways that things are financed. We are not optimizing for human flourishing. We are optimizing for... corporate profit, goes to the holders of capital. And so the system is seriously out of balance. so ⁓ people who are idealistic are locked into that system, which is why, you know, I think, you know, I've had a lot of talks with Eric Reese about this, has written a new book called Incorruptible. Now she's just out, this is really about his ideas for long, long-term idealistic companies. I'm admire what Eric's doing and I think he's identified a lot of things that help. But I don't think the solutions are at the company or the entrepreneur or the individual level. They really are systemic. You know, we literally have a system that demands that, you know, companies, you know, basically do what we as human beings would consider to be the wrong thing.
Paul Miller: How do we go about changing that? Where do we point it?
Tim O'Reilly: I wish I knew because I do think, there are steps that individual companies can take. But I think in some ways it's a social and cultural turning away from this crazy also a lot times I think in an odd way, the challenges of the century are what are going to save us from ourselves. you at climate change, you look at pandemics, you look at mass migrations, all of the four horsemen of the apocalypse that are coming at us sometime in this and we will have to stop playing games. We will have to get serious ⁓ in the way the people of Ukraine got serious with their crisis. again, it's not pretty. It's not what you normally think of as we're doing good, but like doing good is solving the problems that you have to solve in order to flourish. I do think that, that, you know, there are, first of all, there are a lot of people who already working on this. And I love this great quote from Richard Rogers. He was always trying to reassure kids. And I forget what, ⁓ some, had been some disaster and he had the show where he said, you know, whenever there are bad, bad things happening, look around for the people who are helping. There are always people who are helping. And just such a heartwarming message, you know? And I look at people who are, you know, working really hard on this energy transition, you know, and, you know, slammed down again and again, you know, my son-in-law, Saul Griffith has done a huge amount on, you know, electrification and energy and had a whole company. And then, you know, in comes the Trump administration, you know, all their government grants get pulled, everything, know, rewiring America, all the grants get pulled, you know, and it's just, it's really disheartening, but like, wow.
Paul Miller: Hmm.
Tim O'Reilly: You know, and that goes back to that original talk I gave at E-Tech in 2007, where I ended with this poem by Rilke, where he said, what we fight with is so small, and when we win, it makes us small. What we want is to be defeated decisively by successively greater beings. He was talking about Jacob and the wrestlers of the Old Testament who would go out and wrestle with angels who they knew they could not defeat. Yeah, and they would come away stronger from the fight. That's why he successively greater beings, not success or greater things. But, you know, and I look around and I see there are people who are, you know, fighting with, you know, wrestling with angels and, and, and, you know, and, you know, we are going to be building a very, very different world. And I think of the things that
Paul Miller: Yes.
Tim O'Reilly: I've learned maybe reading a lot of science fiction. Some people read science fiction and they got fantasies. I read lot of science fiction and I got fantasies. The ⁓ future world be very different from the past. I look at this a lot of... It's not a simple equation. of my favorite books is a book called The Passions and the Interests by Albert Owen. And the says it all, ⁓ arguments markets their ascendance. And he goes back Adam Smith and looks at all the early people. And ⁓ the title of book comes from the idea that The self-interest of merchants would be a gentler master than the passions of princes. ⁓ was the idea that they had at the time. And of course what we know what happened was that the merchants became princes and then you got both the passions and the Which course is what it was before, just a different. But you have this period of idealism. had a period of idealism with the internet and it took us a long way and it got some really good stuff before it got corrupted.
Paul Miller: Mm.
Tim O'Reilly: You know, idealism around AI and it's going to take us some in some good directions and then we'll get, you know, and it's just, you just have to keep, you know, fighting the good fight.
Paul Miller: Hmm. Yeah, I mean, we work with a lot of like really early stage founders, founders who are maybe close to the problem and trying to build a business because to some extent that's the tools they have these days is sort of, know, advice would you have for them these days? know, what's the, you know, how should they think about working on stuff that matters these days?
Tim O'Reilly: Well, I think the first part of it is you're already doing it if you're looking for problems to solve. When I think about this whole question of AI taking all the jobs, go, wait, look around. There's so much more to do. As as we're trying to do more and make things better for people. I think Nick Hanna, I recall, he said, technology is the solution to human problems. And that's really progress, right? So look for human problems and solve them. And if you're doing that, then as I famously said, it's my most famous quote ⁓ is, is in a business is like gas in a car. You have stop at gas stations, but you're not going on a tour of gas stations. And I feel so. ⁓ As long as you're focused on your mission, and you do have to learn how to run a real business. And I do think that there are certainly... you know, people who are focused on a mission, but who have not figured out how get, you know, product market fit, you know, where they are willing and ⁓ able pay for it, or, ⁓ you that match the cost structure of the business to deliver the thing that you, you know, that you are envisioning and you have to solve those kinds of problems. And I think you also have to realize that when the world changes,
Paul Miller: And that's important.
Tim O'Reilly: that maybe, you know, I mean, right now we're in the middle of huge, you know, cross currents that can wipe out somebody who's otherwise very successful and well-meaning and trying to do the best thing. And so I guess we need to hold it all lightly, you know, and I'm full of quotes. Another one that I love is from, a poem by W.S. Merlin where he says, we were not born to survive, merely to live. businesses too, you know? we engage with a problem, we try to make things better, we our best. at some we're gonna be, ⁓ you at point, you know, what I wanna say to all these people who are just so ⁓ obsessed with.
Paul Miller: Yeah, that's great.
Tim O'Reilly: seem increasing their wealth, they go, what's going to happen when you come to the end of your life? How are you going to feel about that?
Paul Miller: Yeah, Tim, I think we should end it there. Let's leave it with that question. ⁓ you so much for joining us. That's been super interesting. ⁓ yeah, thanks so much, Tim. ⁓ if listening and have suggestions for guests or topics, do get in touch. And if you're building a tech for good startup and looking for early stage investment, you can learn more about Bethnal Green Ventures at bethnalgreenventures.com. That's it for today. Thanks for listening.
Tim O'Reilly: Yeah.
Paul Miller: And it was a response to the financial crisis, but it quickly became something broader, almost like a manifesto for a different kind of industry. nearly 18 years on, ⁓ we it would be an interesting moment. ⁓ Now going through such massive technological change again ⁓ to the idea. So Tim, it's great to have you on the show.
Tim O'Reilly: Thanks so much for having me and I'm glad that I had an influence and that you took the ball and ran with it.
Paul Miller: So just to start off, let's rewind what was going on in 2008. What was it that prompted you to do those series of talks and to write the article?
Tim O'Reilly: Well. Well, was actually the specific on stuff that matters framing grew out of an earlier talk I gave called Why I Love Hackers, which I had given at ⁓ our E-Tech conference. ⁓ And was about how all these people who history forward because they're trying things, ⁓ just for the love it, for curiosity, and that that's a much more important starting point in the traditional narrative from Silica Valley where everything begins with entrepreneurs and venture capitalists. my experience of early ⁓ early open movement, these were people who were motivated by love and excitement and enthusiasm and not by, wow, we're going go out and make big buck. ⁓
Paul Miller: Yeah.
Tim O'Reilly: And so then, then ⁓ I gave it, I gave this talk and ⁓ Palka, who's now my wife and became a social entrepreneur, came up to me. She was starting this new conference in partnership with a rally called the Web2O Expo. And she said, I need a talk like that for my conference, but I need it for entrepreneurs, you know, not about hackers. And that's when I came up with the talk that became why, ⁓ so it became work on stuff that matters. And then I wrote the essay. specifically I was reacting to, you know, there was this sort of crazy period of social media. I where people were getting feted for ⁓ social media apps that were incredibly stupid. And it a kind of a period of hype. You know, did Dickens say? It was the best of times. It was the worst of times. In other words, a time much like our own. ⁓ You know,
Paul Miller: Yeah.
Tim O'Reilly: You know, that was that period. And then you had the whole metaverse thing coming out of meta and it was all clearly designed to drive a hype cycle. Now I think today's world with AI, there's a lot more substance to it, but I still think that Silicon Valley has lost its way in the sense that, you know, certainly the initial entrepreneurs who sort of waded into this AI. technology and started building towards it are, you know, we're profoundly motivated by curiosity and just the pure, the wonderful hacker research ethic. But it very quickly became so overloaded with capital. And I think in general, since 2000, the 2009, you know, 2009 financial crisis, when the response was to have near zero interest rates, capital became really cheap. You were looking to deploy massive amounts of it. Silicon Valley switched from a model in which entrepreneurs were really looking for product market fit. And instead they were doing what Reid Hoffman called blitz scaling. You that is you buy, you know, just move as fast as possible, seize market share. Don't worry about whether you actually have a workable business. You'll figure that out later. I just need the high ground. And, you know, and I thought of this in particularly this happened with, uh, you know, the ride hailing apps, uh, Uber and Lyft, you know, the actual technology was invented by a guy named Sunil Paul, who created a company called Sidecar. And he raised $35 million, which just coincidentally was about the same amount of money that Google had raised and all it took for Google to succeed. And Sunil's the guy who actually, ⁓ mean, was an SMS app for calling black cars for rich people. Rift a web app for putting together intercity carpooling. The VCs saw what Sunil was doing and they piled on, literally gave them billions of dollars. He was quickly out of business. He's the one really invented that he used GPS to locate the phone. the implications of what was really happening with maps and mobile. He was the investor. He got just wiped out. And basically, then they rushed to get market share by ⁓ capital-fueled capital-fueled marketing. And then of course, once they got to their IPO, was like, well, actually we have to raise prices, the service will get worse.
Paul Miller: Yep.
Tim O'Reilly: What was ⁓ casualty was the kind of experimentation that you typically see in an early period of technology. My guess is absent that, don't know that, Sunil probably wouldn't still have won. He wasn't the kind of cutthroat entrepreneur who typically does emerge from the pack. But you would have had lots of local companies adopting the technology. You would have had different business models that were ⁓ playing out. And this is as if somebody had come in at the early days of the web and anointed AOL with so much capital that nothing else could be funded. ⁓ So that was sort part of the framing of that time. And I worry that we're in that same kind of moment today where of users, the market of customers picking the winners,
Paul Miller: Yeah. Yeah.
Tim O'Reilly: capitalist picking the winners because you're building this enormous capital moat and it's not sure that it's going to play out that way because you know there's often a bet that goes wrong when you do that. You know the market is actually smarter than the venture capitalists which is why it's better when you actually the gradual ramp up to success rather than somebody right at the outset deciding that
Paul Miller: Mm-hmm.
Tim O'Reilly: you know, whether it was Netscape back in the day or, ⁓ you know, or OpenAI is the winner and you just give them all the money and let them take the high ground. ⁓ know, you have this period of experimentation and that works better. And we're seeing that with AI, for example, where people are actually, you know, inference costs starting to crush the whole economics of the thing. And it's becoming clear that this is not a viable business model as ⁓ constituted. even companies like Anthropic, which is, I think, really well at finding market and focusing on it and really serving it, is hitting real capacity And there's a big disjuncture between the ⁓ experience and narrative, which is good because it may mean that ultimately we'll get a more decentralized AI economy.
Paul Miller: Yes.
Tim O'Reilly: My experience, and I've talked about this from my early days of advocacy for the commercial internet in 1993, source in 1998, Web 2.0, was all about how we build a more decentralized technology economy opposed to one that's centralized by market capital, moats. And I think I still really believe that. So a lot of what I'm really talking about now is, what is the architecture of participation for AI? How do humans actually, what's the human economy look like? You know, I see someone like, you know, open and anthropic, all they're talking about is we're going to put all the, out of work. I go, well, it'd be your customer then. And, and I see them not even thinking about.
Paul Miller: Hmm.
Tim O'Reilly: the right problems. And so I'm trying to encourage them just like I was in that work on stuff that matters period. Although there I was really, sorry, back up a little bit there. When I said work on stuff that matters, what I meant was you're probably going to fail entrepreneur. So work on something you will have been proud to have done that needed doing. ⁓
Paul Miller: Yeah, yeah, because I mean, I think the first heading was work on something that matters more than money, wasn't it? Yeah. what did that feel radical at the time? Did people push back or?
Tim O'Reilly: That's right, that's right, that's right. No, I think there were a lot of people who loved it because I there's a real hunger for idealism and optimism out there. And yes, there are people who are just sort of totally obsessed with really honor wealth ⁓ in some You've just experienced in Silicon Valley where people come and they don't tell you about their customers, they don't tell you about even their revenue from customers, tell you about their fundraising. You know, it's just, it's this game, you know, where people are placing bets and it's, kind of a, a meme economy. And this is all the way up to the top, you know, Elon Musk is the world's richest man, not because he actually has built a company that makes a ton of money. He's just works the, the memes, you know, it's like all the other car companies are worth, you know, uh, don't know, 10 or 15 times earnings. Tesla's at its peak worth 1,500 years of earnings. And now you look at all the hype he was pushing about robots. And now that he got SpaceX as this new carrier, he sort of said, that really was harder than we thought moving on to the new hype machine. I wrote a piece that I wish more people would read, maybe. four or five years ago called Two Economies, Two Sets of Rules, is really focused on the idea that if playing with capital, it's not just that you have this advantage that compounds without labor and that it magnifies the returns, but also it's borrowing from the future. The fundamental fact of capital,
Paul Miller: Hmm.
Tim O'Reilly: is that people are betting on a future stream of earnings. That's why even in the boring industry, a stock is worth 10 or 12 or 15 or 20 times annual earnings. They go, well, we'll keep earning for some number of years into the future. You do some discounted cash flows and you say, this is what it's worth. And so you buy the stock. But once that becomes completely unmoored and you can have someone like Elon Musk, who's literally making hundreds of billions of dollars, promising that this is a company that pay itself back in thousands of years, ⁓ completely on board. You're very on board. so it's a hype story. You compare any of enterprises to the actual giants of technology. mean, obviously, ⁓ NVIDIA right now way out there, but Google and...
Paul Miller: Yeah, jam tomorrow as we call it.
Tim O'Reilly: and, ⁓ you know, Apple or whatever, there maybe 35 times earnings, you know, and Musk there hundreds of thousands sometimes. And, and, and it's, anyway, that's sort of a little bit of an aside, but the problem is that it has real social implications because there's a class of people who basically for whom a dollar that they earn through their labor is worth a dollar. And there's people for whom a dollar saved. in labor costs, for example, or a dollar of profit, which is valued at some multiple of that dollar. So effectively, the reason why companies are relentlessly optimizing for profit is the executives are paid in a different currency, which has a 10, 20, 30, 50, 100 times multiplier. So of course you want to save that dollar, you know? because that dollar may become $100, may become $1,000. And this is actually the mechanism by which ⁓ economy has ⁓ wrong. Because it's not just that, ⁓ AOC says, ⁓ there shouldn't be any billionaires. I don't believe that at all. There are people who totally their billions, right? But there are a lot of people who have...
Paul Miller: Mmm.
Tim O'Reilly: you know, are participating in this levered game, which is fundamentally unfair. And, you know, and so it's just this idea that when you realize that, okay, we have asked our companies, first of all, we, this is to me, I wrote a piece again, back in, I think it was 2020 called, we've already let the genie out of the bottle. And it was really, you know, corporate governance is our test case for AI governance. You know, and, uh, you know, what do we learn from it? And what we learned from it is, you know, very quickly, you know, we get, you know, basically starting in the early eighties with Milton Friedman's, the social purpose of a business is to make money for its shareholders. And then the shareholders will decide what to do with those profits. That quickly became Meckling and Jensen at Harvard kind of saying, Oh, well, the way to do that is to align the interests of the management with the interests of the shareholders by.
Paul Miller: Hmm.
Tim O'Reilly: stock-based compensation. And then all of a sudden you have everybody subject to what isn't effectively a rogue optimization function, you know, because they've said optimize for profit above all else, you know, and I'm kind of like, you know, we used to have something that was a little bit more balanced and now we really do have ⁓ an economy that we're literally the obligation supposedly of a company is simply to make as much money as possible for shareholders. we built a mechanism in the form of stock-based compensation by which all of the incentives ⁓ slave to that. So I go, well, okay, let's just imagine AI wants to rule the world. ⁓ already built the mechanism for that, ⁓ ⁓ basically, ⁓ not saying that AI will want to rule the world because I think there's a level ⁓ hype there and you already see. it is, I think it's Henry Farrell and Alison and Cosmo Shalizi and James Evans say AI is social and cultural technology. It's basically, it's taken in human values. ⁓ you see ⁓ not just the company, but Claude, the AI pushing when it's asked to do certain things. we're really in this interesting place where, I mean, it's always the case, know, things that seem terrible are also kind of wonderful because, you know, we're in this, you know, crucible where we're figuring out the future and, you know, and the future is of having these new, vastly powerful technologies that we don't quite know what to do with that are going to change our culture. And so I always think, you know, historically, and I go, yeah, you know, would we take back, you know, the printing press and widespread literacy, despite the fact that it led to hundreds of years of religious warfare? No, we wouldn't, but we certainly, if you were in those times, you might've wished to. so you understand some of the conservative impulses, like the future is coming.
Paul Miller: Hmm.
Tim O'Reilly: As Alvin Toffler said, the future always comes too fast and in the wrong order.
Paul Miller: Yeah. So, I mean, where do you see the kind of interesting people working on stuff that matters today? Are there sort of pockets that you've noticed in terms of people really thinking, maybe outside of that model that you've just been talking about?
Tim O'Reilly: Well, I mean, first off, I think there are people who are forced by necessity to work on stuff that matters and it's not necessarily do good stuff. You know, I mean, you look at, you know, Ukraine fighting for its survival they're pushing the future forward. ⁓ the future of warfare. think there are people in some of the big companies. do think that both DeepMind and anthropic really, you know, have very serious commitment to sort of using AI to try to, you know, make advances in healthcare. And so, you know, I don't think it's all, ⁓ all dark, even at the big labs. I, I, I'm actually more, jaundiced by, you know, some of the, ⁓ the, the smaller companies Silicon Valley, they're just jostling for their place in the trough. But even though there's just, there's always entrepreneurs who, kind of are, ⁓
Paul Miller: Hmm.
Tim O'Reilly: are driven by an impulse to serve a real audience. I don't think of AI, or sort of technology for good, just do-gooders. Although I do think there is a lot there. mean, Jen started Code for America, started the United States digital service, now she's sort of trying to work on government transformation. That's a classic do-gooder arc, but with a very different. you know, spin on ⁓ it is, but there are a lot of companies that do that. But I think of, for example, 11 labs, you know, when I talked to, to Maddie, the founder, and he's, he's just like, you know, he and his co-founder were like, they're in Poland and they're like, you know, you don't realize in Poland, you know, it's a small market and the movies are just dubbed by one actor, you know, all voices. And we go with AI, we could do better than that, you know? And that's like, that's a kind of AI for good. that where they're just going like, this is an underserved market and we could really make something better. And I do think that that thread does run all through Silicon Valley and it's best work. is somebody saying, wow, we could do better. But I think this also, what I'm seeing in a way that is AI is unlocking a kind of sort of bottom-up community hacker innovation again. I've seen, you was in the early web. was in early open source, but it really kind of got rung out of the system in the last decade AI kind of blew things open again, because there were more and more the big companies really had everything locked down. so, you know, there's that, you know, so a big part of AI for good is just turn people loose. And yes, will be a lot of people chasing the money. but there will also be people who are just like curious to solve problems, you know, and ⁓ learned with others and yeah. ⁓
Paul Miller: Yeah, I mean, one of the things I find really interesting. One of the things I think we find really interesting is that AI enables people who really understand problems to build solutions. So you've got this ability that actually almost, I think when we started working on this area, you'd have the people who understood the problems and then the engineers, like kind of as separate camps almost, but now those two can be much, much closer together AI, which is, ⁓ think is an interesting innovation moment where you've ⁓
Tim O'Reilly: That's right. Right. Yeah, it is. think there's an enormous opportunity for who are close to problems to solve them. And think that the capability, and I look at Jen's work, for example, ⁓ with government, think about the bad version AI. By the Henry Farrell has a great essay about this right now about ⁓ AI and Doge. the vision of as some kind of of top-down control, which embodied in Doge and the vision of ⁓ people Jen, is, no, AI is an enabler of people who are trying to do the work. And so actually bottom up. If you can actually enable, ⁓ in her she's looking at government, if you can enable people who are in public service, who really want to serve the public, to give them new tools to do that better, that's really powerful. And, and, ⁓ you know, she's, she wrote a great piece about what she calls horizon one, which is kind of the status quo horizon two, which is sort of short-term improvements. and then horizon three, which is really changing the status quo. she says, look, a lot of the problem is that a lot of technology is just keeping the old status quo alive and not utterly transforming it. And think that we, you know, ⁓ one the interesting challenges in the AI for good space is to think big enough.
Paul Miller: Yeah, and actually that's just one of the references you made in that original essay was to the Long Now Foundation and Where is long-termism in the ⁓ industry at the moment? ⁓ it disappeared or what's thinking about that these days?
Tim O'Reilly: Yeah. Mm-hmm. Yeah. ⁓ Well, I don't think it's the tech industry that's the problem. It is literally the design of the entire system. is baked in to our tax code. It's baked into all ways that things are financed. We are not optimizing for human flourishing. We are optimizing for... corporate profit, goes to the holders of capital. And so the system is seriously out of balance. so ⁓ people who are idealistic are locked into that system, which is why, you know, I think, you know, I've had a lot of talks with Eric Reese about this, has written a new book called Incorruptible. Now she's just out, this is really about his ideas for long, long-term idealistic companies. I'm admire what Eric's doing and I think he's identified a lot of things that help. But I don't think the solutions are at the company or the entrepreneur or the individual level. They really are systemic. You know, we literally have a system that demands that, you know, companies, you know, basically do what we as human beings would consider to be the wrong thing.
Paul Miller: How do we go about changing that? Where do we point it?
Tim O'Reilly: I wish I knew because I do think, there are steps that individual companies can take. But I think in some ways it's a social and cultural turning away from this crazy also a lot times I think in an odd way, the challenges of the century are what are going to save us from ourselves. you at climate change, you look at pandemics, you look at mass migrations, all of the four horsemen of the apocalypse that are coming at us sometime in this and we will have to stop playing games. We will have to get serious ⁓ in the way the people of Ukraine got serious with their crisis. again, it's not pretty. It's not what you normally think of as we're doing good, but like doing good is solving the problems that you have to solve in order to flourish. I do think that, that, you know, there are, first of all, there are a lot of people who already working on this. And I love this great quote from Richard Rogers. He was always trying to reassure kids. And I forget what, ⁓ some, had been some disaster and he had the show where he said, you know, whenever there are bad, bad things happening, look around for the people who are helping. There are always people who are helping. And just such a heartwarming message, you know? And I look at people who are, you know, working really hard on this energy transition, you know, and, you know, slammed down again and again, you know, my son-in-law, Saul Griffith has done a huge amount on, you know, electrification and energy and had a whole company. And then, you know, in comes the Trump administration, you know, all their government grants get pulled, everything, know, rewiring America, all the grants get pulled, you know, and it's just, it's really disheartening, but like, wow.
Paul Miller: Hmm.
Tim O'Reilly: You know, and that goes back to that original talk I gave at E-Tech in 2007, where I ended with this poem by Rilke, where he said, what we fight with is so small, and when we win, it makes us small. What we want is to be defeated decisively by successively greater beings. He was talking about Jacob and the wrestlers of the Old Testament who would go out and wrestle with angels who they knew they could not defeat. Yeah, and they would come away stronger from the fight. That's why he successively greater beings, not success or greater things. But, you know, and I look around and I see there are people who are, you know, fighting with, you know, wrestling with angels and, and, and, you know, and, you know, we are going to be building a very, very different world. And I think of the things that
Paul Miller: Yes.
Tim O'Reilly: I've learned maybe reading a lot of science fiction. Some people read science fiction and they got fantasies. I read lot of science fiction and I got fantasies. The ⁓ future world be very different from the past. I look at this a lot of... It's not a simple equation. of my favorite books is a book called The Passions and the Interests by Albert Owen. And the says it all, ⁓ arguments markets their ascendance. And he goes back Adam Smith and looks at all the early people. And ⁓ the title of book comes from the idea that The self-interest of merchants would be a gentler master than the passions of princes. ⁓ was the idea that they had at the time. And of course what we know what happened was that the merchants became princes and then you got both the passions and the Which course is what it was before, just a different. But you have this period of idealism. had a period of idealism with the internet and it took us a long way and it got some really good stuff before it got corrupted.
Paul Miller: Mm.
Tim O'Reilly: You know, idealism around AI and it's going to take us some in some good directions and then we'll get, you know, and it's just, you just have to keep, you know, fighting the good fight.
Paul Miller: Hmm. Yeah, I mean, we work with a lot of like really early stage founders, founders who are maybe close to the problem and trying to build a business because to some extent that's the tools they have these days is sort of, know, advice would you have for them these days? know, what's the, you know, how should they think about working on stuff that matters these days?
Tim O'Reilly: Well, I think the first part of it is you're already doing it if you're looking for problems to solve. When I think about this whole question of AI taking all the jobs, go, wait, look around. There's so much more to do. As as we're trying to do more and make things better for people. I think Nick Hanna, I recall, he said, technology is the solution to human problems. And that's really progress, right? So look for human problems and solve them. And if you're doing that, then as I famously said, it's my most famous quote ⁓ is, is in a business is like gas in a car. You have stop at gas stations, but you're not going on a tour of gas stations. And I feel so. ⁓ As long as you're focused on your mission, and you do have to learn how to run a real business. And I do think that there are certainly... you know, people who are focused on a mission, but who have not figured out how get, you know, product market fit, you know, where they are willing and ⁓ able pay for it, or, ⁓ you that match the cost structure of the business to deliver the thing that you, you know, that you are envisioning and you have to solve those kinds of problems. And I think you also have to realize that when the world changes,
Paul Miller: And that's important.
Tim O'Reilly: that maybe, you know, I mean, right now we're in the middle of huge, you know, cross currents that can wipe out somebody who's otherwise very successful and well-meaning and trying to do the best thing. And so I guess we need to hold it all lightly, you know, and I'm full of quotes. Another one that I love is from, a poem by W.S. Merlin where he says, we were not born to survive, merely to live. businesses too, you know? we engage with a problem, we try to make things better, we our best. at some we're gonna be, ⁓ you at point, you know, what I wanna say to all these people who are just so ⁓ obsessed with.
Paul Miller: Yeah, that's great.
Tim O'Reilly: seem increasing their wealth, they go, what's going to happen when you come to the end of your life? How are you going to feel about that?
Paul Miller: Yeah, Tim, I think we should end it there. Let's leave it with that question. ⁓ you so much for joining us. That's been super interesting. ⁓ yeah, thanks so much, Tim. ⁓ if listening and have suggestions for guests or topics, do get in touch. And if you're building a tech for good startup and looking for early stage investment, you can learn more about Bethnal Green Ventures at bethnalgreenventures.com. That's it for today. Thanks for listening.
