Ep. #6, The Infinite Nature of Software with Adam Jacob

John McBride: Welcome back, everybody to The Open Source Ready Podcast. I'm here again with Brian. How are you doing?

Brian Douglas: I am doing fantastic. Thanks for having me on this show, John.

John: Well, I think it's really your show, but I've taken over for a few weeks, which has been a lot of fun.

Brian: Yeah, it's our show now.

John: It's our show now, exactly.

Brian: It's open sourced actually.

John: It's all open source, we're all maintainers of this show. We're here this week with Adam Jacob, who is the legend co-creator of Chef and now System Initiative. Adam, how are you doing?

Adam Jacob: I'm doing great. Yeah, things are good.

John: Thanks for coming on. We're very excited to have you on and chat about open source, and I know on Twitter, now called X, you've always had some really, really solid takes on how open source works and you've been around the ecosystem.

Why don't you give our listeners just a little introduction to who you are and some of your work?

Adam: Yeah, so the thing that made it so people knew my name was that I created a configuration management tool called Chef, and was sort of a big deal.

Now I'm the CEO of a company called System Initiative that is sort of the future of how you're going to do DevOps automation. I won't spend too much time pitching you on System Initiative, but suffice to say, it's dope and you should check it out.

Through that journey, most of my career is really as a systems administrator and sort of free-software person, open-source person, which then turned into a business person and then tried to figure out sort of how to square the circle between what I call my sort of like free-software, hippie truth, where there are some things there that I just believe really deeply and that really transformed my life.

And then also wanting that to live in the right context in a harmonious way with how I think about building businesses and how I think about that part of what I also do, which over time I've become even more sort of a business person.

Like I'm CEO of a company now. I did the math the other day and I'm almost 20 years into running an open-source company in one form or another. And so like that's another piece of me that's really strong.

And so that's led me on quite a journey of just sort of analyzing how I feel both philosophically about open source, how I feel about its impact on my life and the lives of other people, and then primarily how to think about it in the context of business.

I think there's a lot of other ways to think about open source. There's thinking about it as, "Oh, how do we pay maintainers?" There's a lot of aspects of sort of open source sustainability.

My particular line of thought is around how do you think about it in the terms of like startup ecosystems and that kind of business?

John: Yeah, well, let's get into it. I was at KubeCon in Salt Lake City a couple weeks ago, and you gave this absolutely incredible talk at Startupfest that Kelsey Hightower was putting together, and it was sort of at the intersection, your talk about free software, open source and fair source, and the commercialization of all that.

So I just would love if you could give for our listeners what that is, what those definitions are, and maybe where you land kind of among those.

Adam: Yeah, okay, so let's talk about just sort of definitions of free software, open source, fair source.

So I think the best way I think about free software is, primarily, it's kind of a political philosophy. And it sort of stems from this particular truth, which is that software by its nature is infinite. So if you think about what it means to have created some software, like if you have power and compute that can run it, and I give you a copy of some software, I lose nothing. And that is unique kind of in the history of the universe, right?

Most everything else, when you think about it as a thing that creates value or does something for you, it's finite in some way. And software is infinite. If I pass it to you, nothing happens. You can take it and do what you want.

That realization then leads to this idea that if that's true, then we can fundamentally alter our relationship to software as people. So the impact that software has on us as people can be different because it no longer has to be extractive, right?

It can no longer be that capital, or the lack of software, the lack of resources, in this particular way, like drives how humans work together or how we behave. Instead, we can sort of be in control of our own destiny.

And that sort of leads you to the four freedoms and this idea that software should be free because, by nature, it is, right? Software's nature is infinite, therefore human beings should have software be free because it's a mechanism of creating greater human freedom in the broader universe.

John: And this is really kind of a Stallman sort of approach where with the GNU Project and all that.

Adam: Yeah, yeah, that's certainly the way most people think of when they think of free software. And I don't think you can separate those ideas from his ideas, right?

I think open source was a reaction that says, along with free software came this new kind of development model, like this nature of software as being infinite and people being able to take it and do what they want to do with it creates a new kind of open development that says we can work together now to build software in interesting new ways.

Maybe software we couldn't have built alone, maybe software where the industry could decide that it needs a particular thing and it could all work together to do it.

And so open source builds on top of the political philosophy of free software, but really downplays it, sort of pulls it out of the lexicon and just says, "You know what this is about is open development models." What's great about this is how we collaborate together.

It's how, it's now everybody's podcast. And that's the great part. Forget about all of this "Star Trek" political philosophy mojo and let's just talk about how good it is for the economy.

Like let's talk about how as an industry consortium, we can build Kubernetes and we could crush Amazon, and then Amazon's like, "Not so fast, like I'll just join Kubernetes."

John: And now they have EKS, right?

Adam: And now they have EKS and everybody's like, "Oh, our plans were foiled or whatever." So open source's focus is on those open development models and on making it palatable for business, because free software as a political philosophy, terrifying to business, right?

Like just conceptually, it's like "Star Trek," like, "What if there was no economy?" And you're like, "I don't like that, that's how I eat."

So then fair source says, "Hey, all this open development model stuff, I like. I like that people can like collaborate with me. I like that they can see the source code on GitHub. I like that, in theory, you could like patch the software or whatever, but I don't like the part where people can take my software."

I don't know if people can see me or not, but I made air quotes. So if you couldn't hear them, there's air quotes there. "My software, the stuff I produce primarily, and then make money with it because what I want to do is make money with it."

And so they basically have said, "Hey, I want a little of this open development model that I get from open source, but I don't like the open part where you're allowed to like do whatever you want with it if it means I don't get mine."

And that's both a competitive question, right? Just like, "How can I own this market or the monetization of my software and sort of reserve some of those rights for myself?" while also sometimes maybe delaying the open source of it, right?

Saying, "Hey, I understand that it's a good idea that this thing become free eventually," free as in speech, not free as in beer.

And so that's kind of how you think about it. Another way to think about fair source is basically like a software patent that happens to be causing you to have this moment, patents exist so that the inventor of something can have a monopoly for a short period of time, right?

And essentially we're doing the same thing, we're just doing it now with software licensing. We're saying, "Hey, I want to have a monopoly over this software for a couple of years in a rolling way. So like every time I change the software, the clock resets. And I'm going to just sort of do that into perpetuity, because I don't like all this messy open-source things. Like I don't like all the parts where people can take it and do stuff I don't want them to do, like compete with me."

So that's the philosophy side. And then the business side is basically, what I've learned over the years after many false starts and attempts is that you have to really have a hard demarcation between when you're talking about what I was just talking about and business.

And if you're talking about business, just business, forget about open-source business, forget about the rest of it, businesses are relatively straightforward to think about.

You have a total addressable market, that's the total number of people in the world who could ever buy your product multiplied by the price they pay you.

You have a serviceable addressable market that says, "Yeah, those are all the people who could ever buy it," but you're probably constrained by something, features, geography, whatever, that's your serviceable addressable market.

Same calculation, smaller number of people on the left times the price they pay. Then there's the obtainable one of the people who could buy your product, how many will, right? And you can then analyze your strategy based on its impact on those three numbers.

And so when you're thinking about, "How do I build a business that I want to be really good?" what open source tends to do, what we've done mostly over the years, is this thing called open core, which basically says, "Hey, I'm going to open source some huge piece of what I do."

Chef did this, right? We built the Chef client, you could use it to run Facebook and not pay me a dollar." Crazy value, right? But the cost was nothing. And so then I went to Facebook and I said, "Facebook, you should pay me to use Chef."

And they were like, "Eh, like I don't know." And so like eventually I convinced them to pay me a little and let me use their logo. It was great, Facebook was an incredible customer. What an incredible partnership.

But some other large companies that remain unstated would say just like, "No," and Facebook didn't pay me a lot of money, paid me a little money.

And so then I would go build some proprietary software that wasn't open source, and I would say, "Here's the special sauce that's going to convince some other market segment to pay me," right?

The problem with that is that it starts from zero. So if you're doing that analysis of like, "What's the price that you pay?" the right-hand side of the equation, right?

The number of people who could buy your product is fixed pretty much. You're not minting new companies that often, we're not minting more people that often. So number of people on the left-hand side of the equation is fixed.

The problem is the right-hand side of the equation is the only one you can move. And when you decide to give your software away for $0, you sync the right-hand side of that equation pretty dramatically.

It goes from like maybe a big number, hundreds, thousands of dollars on average, to tens of dollars on average, and it is a brutal climb.

And so it took me a long time to sort of be able to express it and show it to people that like, "Hey, there's this strategy that is, how do you think about using open source to create a bigger community and open development model to move into the world? That strategy can be incredibly effective. It's so effective actually that we have public companies who have done this really inefficient monetization strategy, but they were so good at building this huge base of people who could adopt that they basically overcame this incredible boat anchor that is the incredibly low price that we ask people to pay for the software."

John: Do you mean companies that, like big public companies, like an AWS, that has an open-source offering?

Adam: Not so much AWS. AWS is minting Scrooge McDuck money. AWS is not an open-source company, right? Like they play with open source, but they don't open source their product. I can't go look at the source code for EC2 and be like, "What if I wanted to change it?"

Like it's not going down like that, right? And I object, there are people in the world who would say they are because they like use open source. Like there's people who say GitHub is an open-source company, and like, this is dumb.

Like Git is open source. GitHub? Get out of here, miss me with that, right? They are not an open-source company. I'm not even sure they ever liked open source, then I think I would've like, if you had asked me when GitHub started, right, I'd have taken that bet like 100 times.

If I was like, "Are they open-source true believers?" I'd be like, "Of course they are." No, no, no, not at all.

John: I'm sure Brian has a hot take there on that. (laughing)

Brian: Yeah, I mean, I spent a long time at GitHub as an employee, and you're not too far off, like open source is a means to it ends for GitHub, and I think they did a great job-

Adam: I agree.

Brian: in building a really good platform that was accessible to open-source users.

Adam: Definitely.

Brian: And they do have open-source code that's not the main platform.

Adam: Yeah, exactly, yeah, there's like stuff they do, but it's not like GitHub.

Brian: Yeah. And I was going to ask a question too as well, 'cause you're talking about like public companies, like HashiCorp comes to mind, like Elastic comes to mind.

Like I was going to ask a question because with what you've built, like do you think there's a timing thing, where 20 years ago, you could start thinking about open source, build a huge adoption, maybe build that market, but today can you start a movement and open source and build a company?

Adam: Absolutely, you can. Yeah, yes, yes, of course you can. Because the question there isn't about open source or not, it's about product market fit or not.

Build something incredible and open source it, and build a real collaborative community around it, it is like pouring gasoline on a fire.

Now, if you open source something that doesn't have product market fit, the product's not good enough, no one cares. Like open sourcing, it doesn't make someone care, you know what I mean?

If the product's not good. Right? So like yeah, but absolutely, you can, right? And I think when I talk about inefficiencies, it's interesting because when we talk about companies being successful or not successful, right?

Like HashiCorp, incredibly successful. So I'm going to talk smack for a second, but like let's just be clear. Like they're not mad, right? They did a fantastic job.

And the amount of penetration they have into the largest companies in the world for the amount of revenue they extract is insane, right? Like they should be way bigger than they are, right?

And you can kind of look at this sort of around the horn, if you look at how, when I talk about efficient monetization, I'm talking about like, given your market penetration, how much capital, how much money do you make?

And so like Red Hat's the best example here. They have like a 1% share for Red Hat Enterprise Linux. I think CentOS or the clones all combined is probably, I don't know, I don't have the numbers in front of me, so I don't know for sure, but let's call it between 25% and 30% for the clones.

But on a less than 1% share of the market, they made $3 billion the last time that they reported their revenue as a public company.

John: Wow.

Adam: $3 billion with a B, right? They made a billion dollars in ARR on Kubernetes. You know how many people probably in the container market buy OpenShift? Like not that many.

Like you all work for the Linux Foundation, I bet you talk to Kubernetes people all the time, I bet the vast majority of them do not buy Red Hat's version of Kubernetes.

And they made a billion dollars just every year for living, just for breathing. And like there is something different they're doing, you know what I mean?

And we've been making excuses, people come up with all kinds of convoluted things. But the truth is, like what they do different is they sell their product for money exclusively. The software is open source, right?

There's nothing about, you could take OpenShift right now, I forget what they call it. They have an open source, there's a fork, just like a clone you can run that is that software, but you don't, right? Instead what you do, if you want it, they build a good product.

It has product market fit in the enterprise, and they ask the enterprise to pay them money to have it, and they don't apologize for it. And what we have built is an organization of people and of entrepreneurs who don't know or don't think about how to segment their own market, right?

And so they're afraid of forks, they think they have to give the software away for nothing, because there's always somebody who wants it for nothing, right? Like there are people who will get mad at you if they can't have all the things that you give to Morgan Stanley for $0.

And like you just need to say to those people, "No, like you can have the software, right, 'cause I believe in software freedom. So like you want to take the software and do some stuff with it, please. Can you take my product and do stuff with it for nothing? No."

John: This is great, 'cause I think I'm almost having my own like come-to-Jesus moments talking about this and like looking into my past, 'cause when I was at VMware working on Tanzu, like we were, and this is all in hindsight, given Broadcom, et cetera, et cetera, we were almost like afraid of not having the goodwill of people in the open-source community and ecosystem.

So then we made Tanzu Community Edition, which was sort of like a Rancher open-sourcey competitor thing. But I remember in your talk you said like, "These companies open core themselves," and that's exactly what we did with--

Adam: That's exactly what you did. And it was a terrible strategy.

John: Yeah.

Adam: It is a terrible strategy. And we just don't talk about it as bad strategy. Like we talk about it as good community building, we talk about the goodwill of it. We're like, "Look at all the goodwill VMware did with Tanzu."

John: Exactly.

Adam: But let's be honest, screw a goodwill, right?

Get the money. Like at some point, get the bag. And like you weren't there for goodwill. You know what buys a lot of goodwill? Money.

Like Red Hat does a ton, a crazy amount of open-source software. They open source everything they touch. They're in the Linux kernel, they're in Kubernetes, they're in everything that matters to you.

They're doing good work all the time. You know why? They minted money. Like how many Tanzu developers are left?

John: Zero.

Adam: You know why? They didn't mint money. But good job getting all that goodwill, you know what I mean? Like what are we doing? What are we doing?

And then we talk about how open source has a sustainability crisis. It does not, right? Like if you think about it from a free software point of view, like think of all the software that Tanzu built.

If it had all been free software and you had actually built a thriving community around it and you had said, "What we're going to do is sell it for money to the largest enterprises in the world," everybody else, "We can't wait for you to take that software and sell it to someone else."

"You want to build Tanzu Edge Edition? Go ahead, you just can't call it Tanzu, you got to call it whatever, Coffee Cup Edge Edition. And like I can't wait for you to have that piece of the market."

And you just segmented it that way and you let it happen. Like that money, everybody else would've found money too, right? Because there's money to be found. Like there's money all over the place in these segments.

But we just don't normalize making it, right? And instead we put those things at odds. And then for the community, we don't do anything to explain to our own communities how we expect our communities to function, right?

We don't say, "Hey, part of the reason that it's a community isn't because I give you everything. That's not a community, that's a free lunch, that's a lunch line. A community is one where you want something and I want something, and together, we help each other get it."

John: Hmm.

Adam: And it means you have to want something, other than me to give it to you for free.

Like there has to be something in there for you. And when you come to me and you're like, "Oh, I want to work with you because I have something I want," now we get to be together, and now we're a community, now we're doing the work, now we're actually like building bonds that matter and we're building things that can last.

And people talk about it as like, "Oh, the only true open source is in foundations and all that stuff," that's a smokescreen. Like if the question we're asking is, "How do we build communities that survive?" the answer is, they help each other to thrive.

And in our world, the one we have today, you thrive by making money. And if you're not making money, it sucks. Like the free software hippie in me would like the "Star Trek" version of like, "I'm in on all of that. And also, what you need is to get paid."

John: Yeah, I think that's incredible context for maybe the next part of this, which I think is fair source.

Adam: Yeah, I mean, fair source, if you think about it as just competitive pressure, right? So let's use Sentry as an example.

So Sentry, in a crowded market, they do basically like application performance monitoring. They started from like reading credible error traces, and then grew into broader application performance monitoring.

Had an open source, Sentry was open source for a long time. Now it's not. The reason for that was because they had a much bigger player who wanted to come into their market where that larger player's strategy was to basically take open-source software and smush it together into like a rolled-up platform that then they would just vend back to people.

And so they were going to take Sentry's product and include it in their own and like sell it for more money, or less money, and compete with Sentry directly.

And Sentry was like, "This is not good for us." And honestly, I don't blame them, right? If that's the situation you're in, right? Sentry wasn't like, their open source hadn't driven them to be the defacto standard for application performance monitoring in the world, do you know what I mean?

They didn't have a dominant market position. They had a less dominant market position that was going to get further eroded by a larger player who decided to come and take their software and make a bunch of money on it by competing directly with them, and they didn't want that to happen.

I don't think I blame them. Like that seems like a reasonable setup. It's also been sort of bandied about as like a way to protect yourself from like predatory big cloud or whatever. And that's just dumb.

So like if Amazon sees what you build and thinks to themselves, "There's a market I want," Amazon will compete with you in that market. That's it, 'cause they're a shark, they eat to live.

Microsoft will compete with you in that market too. They'll just say nice things to you while they do it. They'll like come to you and talk about how friendly they're going to be or whatever, but like they're all coming for you.

And like your choice here is not whether they're coming for you, your choice is how. So like you get to make a competitive choice about like, if it's open source, we can talk about white labeling, we can talk about branding, we can talk about all kinds of stuff.

If it's proprietary, they're just going to come at you by building a different product. And you might win. Like that could be a competitive situation you want to be in where you force them to build a better database than you or whatever, fine.

But that's sort of the context you should think about with fair source. I think the argument that says fair source is really open source, and that people who call open source the OSI definition of open source are somehow pedants or whatever, like you can just miss me with that whole line of conversation.

Like there was a lot of effort that went into deciding what open source was and is, the people who are like, "But I don't like it, and it's just I get to use words the way I want to." Like I think they're being petulant babies.

And like I'm pleased that fair source exists because they stopped doing that and they started saying, "Here's what we value. This is what we believe. This is why you should do this."

And I have deep respect for that, because while I may not align 100% with their values, I align 100% with the part where you articulate your values, where you own them, where you tell other people what their choices are.

Do you know what I mean? And you know they're going to find people who agree with them.

John: Yeah, something that you've said in the past about these words mattering, open source, fair source, et cetera, et cetera, something that I really worry about is how that's getting kind of twisted in the AI space today. And I think you've even had some hot takes on like Llama is open source.

Adam: Yeah, it's not.

John: It's really not. And now there's an OSI definition around open-source AI.

Adam: And it's not very good. And I was reflecting on why I felt like the open-source AI definition from OSI wasn't very good, and what I came to realize was that it's because it doesn't have free software.

The problem is that like when open source exists to make open development models accessible to business, clear to business, right, we don't have an AI business problem, you know what I mean?

Like Facebook figured it out, they were like, "I want to give Llama away for you, and then that'll be good 'cause you'll use it." And that's true, they do.

It has a free software values problem, which is like, in a world where software is infinite and wants to be free because that is its nature, what does it mean to decide to have artificial intelligence that can't be utilized by normal people to solve their problems as they see fit?

Like we have a fundamental mismatch in the nature of that software's ability to alter the trajectory of human freedom. And like that's a real bummer.

And because the OSI's definition of AI sort of removed that soul from the equation, you read it and I'm just like, "Well, I don't need any of this shit. Like I'm not sure I agree with you. I don't know if I believe you, because you didn't actually tell me anything about why it's good for me. You just told me why it was good for business, and motherfucker, I knew it was good for business. The whole world was telling me it was good for business. That's not a hard sell. I don't need you to convince me, like everybody's convinced."

Like tell me how it's good for human freedom. Tell me how it's going to change the trajectory of my life. Tell me how it's going to change how a kid growing up with nothing, but who winds up with a computer from his uncle, or his aunt, or whatever, like suddenly figures out that they can become an AI software developer 'cause they can look at how the software was built and written, and they can use it to transform their lives, and there's no limit on their ambition because it hurts no one. Tell me that story.

John: Yeah, the studying part of that, like.

Adam: Not just studying, turn it into their life. Like the limitation, like the Llama limitation, that's like, "Hey, it's all good in the hood until you get too big. Everything's fine, until you get too big," like okay.

John: And then pay us a lot of money, right?

Adam: But then I pay you a lot of money, and it's unclear what those terms are, it's unclear where the mechanisms are. Like that's putting a cap on what's possible for someone to do. Do you know what I mean? And I think it's just kind of, it's not wrong, right?

It's just, it could be so much better. Like the nature of software is infinite and it does call you toward using that nature to transform what it's like to be a human being. And it's a real missed opportunity to have that move backwards in a technology that's as transformative as this could be.

John: Yeah, is there a hardware component of that? 'Cause something that I feel deeply around the like freedom part of it is, like some of these models, even the Llama like 70 billion parameter one, which is not that big, but like it'd be so much better if I had a cluster of Nvidia GPUs or an H100, but on my MacBook it kind of sucks.

Adam: Yeah, and this brings us back to the little caveat on software's infinite nature, which is, if you have compute and power, then it is infinite in nature.

And so you don't have the compute and you don't have the power, and so as much as we're talking about how great it would be if we transform the lives of someone by their ability to use those things, like, "Well, except, call me when you have," but at the same time, you're like, "Well, Google wasn't Google before it was Google," you know what I mean?

Like they were just dudes in a garage, until they weren't. And part of what made that possible was that like a lot of that early search technology, like search like AI, I'm old enough to remember the internet pre-Google and pre those things. And like Google was a lot better than WebCrawler.

John: Hmm.

Adam: And I mean, not like a little bit better, it was like crazy better, right? And the industry spent a decade talking about how search would be ubiquitous, we sold enterprise search appliances that we're going to index all the data and the enterprise so that you could do full-text search across all the information in the enterprise, and it would be this transformative summarization machine.

And it should sound a whole lot like what we're selling now into the enterprise with AI, where we're like, "Here's this magic robot, going to slurp up all these exabytes of data, and then what we're going to do is feed it back."

I mean, we kind of know, we sort of did it. We just don't remember, most of us, 'cause whatever. You got to be of a certain age.

John: Yeah, yeah, man, that's fascinating. It really kind of breaks my brain, 'cause I think for a long time just this whole idea of like open source was kind of just ubiquitous with how you, at least like when I was coming into like my software development career in computer science, it was like, "Oh yeah, you just like put it on GitHub, and then you'll find success, period."

Adam: Yeah, the success part's always been a problem. But the on GitHub part, I mean, open source as an idea and as an open development model is so successful that it's basically like fluoride.

Like all you have to do is drink it, and everybody's kind of an expert on what fluoride does for people. Like if I asked you both, "What's fluoride do?" you'd be like, "Makes my teeth good."

True. And if you ask most people, "What's open source mean?" they'd be like, "Means I can get the source code and it's on GitHub," and you'd be like, "True."

John: Yeah, any thoughts or hot takes, Brian, or questions on your side?

Brian: No, I feel like we got a quite a few hot takes from Adam already, so I think we're good on that front.

But no, I think we're in kind of a wild time to be alive, as a developer rather. And yeah, I just enjoy every minute of it and getting to soak this up with Adam and go down memory lane with you, it's been great.

Adam: Yeah, I mean, it is an interesting time. And like the part that makes me most excited about this moment is actually that there is like an interesting call to build.

And as a software developer who builds products and who like thinks about how to transform industries or how to transform people's like day-to-day work, like, aw, it's such a good minute, you know what I mean?

There's like all these new technologies. And people have forgotten how the like low-level technologies work.

So like if you know how computers work, like in really deep detail, there's actually opportunity now for you to build businesses you couldn't have built 15 years ago because everybody knew how they worked and so there was no surprises.

Now, suddenly, you can be great at some really low-level nichey detail and turn that into a superpower that nobody else can do. And like there's a lot of really interesting fun opportunity.

John: That's an interesting topic, just on like how, I don't know, the knowledge gaps have changed. and obviously there's like so many technologies, so many different areas of expertise where maybe 20, 30 years ago, it was like, "I know C, and I can like work in a Linux system."

We were all trying to figure it out. Have you seen those gaps change, or grow wider, or?

Adam: What's happened is we've just gotten better and better at letting people use the technology to do more and more powerful interesting things.

And so like your ability to be a consumer of technology and get incredible things done has only increased and is only increasing now even more. Your ability to understand is different.

So like the analogy I use for people is like, if we had a Model T that was just in buckets, like in parts, and the three of us were locked in a room with a bucket of Model T parts and all the tools we needed to get a Model T to work and a can of gasoline, we would get out of that room, like we would build that car, we'd figure it out.

If you took apart my Audi and put it in boxes, and then you locked us in a room and said we can't leave until we can drive that Audi out of the room, we die there. You know what I mean?

And maybe Brian or John, maybe either one of you know a lot about modern cars, but like I don't.

John: The microcontrollers alone. Like all the microchips and everything in there, ugh.

Adam: How do you start it once we took it apart? I don't even know.

Brian: Doesn't Audi run on Kubernetes now or is that BMW?

Adam: Please tell me it doesn't.

John: Oh boy.

Adam: Oh, that hurts my heart. But yeah, okay, and like, right? Even worse. And so like that's the way you can think about it is that like it's not like the old version was better.

Most people wouldn't trade their Audi for a Model T for their daily driver because the Model T can break your arm when you start it, and that is an untenable situation. So now we've just gotten to the point where complexity is undoubtedly higher.

Also, it's mostly focused at making it possible for end users to do things that they couldn't do before, which is fantastic. Like if that's where complexity leads you, take complexity, like all day, every day, twice on Sunday, eat the complexity.

And so like I think now we're in this moment where that ability to understand, but here is how the system works, here are how these pieces are put together.

And if I can use those pieces together in new ways, reflecting new advances in technology, like that is a new opportunity to build because most people, 20 years ago, everybody knew how the fundamentals worked.

I had this terrible moment as a product person in the last six months where I was explaining a feature about basically like an audit log. And I just assumed that the people I were talking to understood how to then do full-text search across all the data and like think about faceting, and of course they didn't.

And that's because I'm 47, and I lived through the part where we all had to learn everything there was to know about how search indexing was built because that was the number-one technology in the universe that was changing how people access the internet, and so every single person roughly in my age bracket and older knows a ton.

Everyone knows enough to be dangerous about how those things are put together. And the people I was talking to did not, they were 25, and they were like, "Oh, I've never," like they never had to think about how Amazon's homepage searches data and facets sweaters, do you know what I mean?

They had no idea what that was. And I'm like, "Oh, yeah, okay, let me explain that to you."

John: Wow.

Adam: But I didn't think to explain it to them, I just assumed they knew. And they said they knew, because they didn't know what I was even talking about, 'cause I never even said the words, 'cause that's how fundamentally I just believed everybody in the world knew how that worked. Obviously dumb, anyway.

John: The layers of complexity. I could chew your ear off about all these things. But, Adam, I got to ask you, are you ready to read?

Adam: Yeah, I'm ready to read.

John: Great, let's start maybe with Brian has one. Take it away, Brian.

Brian: Yeah, yeah, mine's pretty quick. I just saw, was it last month? StackBlitz had announced that they made revenue in four weeks, $4 million in four weeks based on this new AI feature, and it's called bolt.new. I don't know if any of you saw this come through.

Adam: Mm-hmm.

Adam: But I was actually really surprised 'cause I remember actually talking to Eric at StackBlitz and a few other folks on the prac side to get some feedback on this product, and essentially it's just, if you've ever seen Vercel's v0 where you sort of type in what you want to build and it'll give you like a full-stack JavaScript application out of it.

And that was enough to make $4 million in revenue, which I'm just like dumbstruck.

Adam: Yeah, I first heard about that because I was at a venture capital party at re:Invent and all the venture capitalists were like, "Have you heard about this company that made all this money?"

John: In four weeks. That's crazy.

Adam: Yeah, I mean, God bless 'em.

John: Yeah, I'm a big fan of Eric. Eric's, he's a fighter, he's been working on StackBlitz for like maybe seven years, and like he's now just started to get some like hardy traction.

But if you don't know Eric, you should google, "Teenager who lived inside of AOL." Eric's also that same person.

Adam: No!

Brian: It's the Reddit story. Yeah, so.

Adam: Wow, really? Oh God, I love everything about that now.

Brian: I've known him for years, and then I think there was like a, every year, on Reddit, it like comes back and someone like reposts it. And I was like, "Dude, this is you? And I've like known you for like a few years."

He's like, "Yeah, it comes back."

Adam: And that's the person who made all that money on bolt.new?

Brian: Yeah, so if you want to start-- Choose open source, but also try to live inside of a tech office for a couple months.

Adam: Talk about overnight success, you know what I mean? Like that's amazing. Brian, that was incredible.

Brian: Excellent. We'll link it in the show notes.

John: I'll go quick with one. I've been playing around with a model that's available on Ollama, and I'm definitely going to butcher this. I think it's QWQ.

Q-W-Q, you can run it with Ollama, run QWQ. And it's been a while since I've like played around with some AI thing and really felt like, "Wow," kind of had that kind of like impressive moment.

This is a research model that is sort of exploring this like recursive reasoning that I think the ChatGPT o1 preview model kind of does something similar with like how it works with attention and things.

The idea is that you can give it some problem, like a puzzle or even like a math problem, and you can watch it kind of iterate and go through various answers. I gave it a puzzle, well, I guess kind of a puzzle, where it was like, "Here's a sequence, what's wrong with this sequence?"

And it was the Fibonacci sequence missing one of the numbers. And at first it's like, oh, like you're the numbers or you're subtracting the numbers, like, "Where is this thing?"

And obviously this is a model, it's not actually reasoning, but very cool that you can run that on some local hardware. It'll be in the show notes for some of the more in-depth white paper if people are interested in that.

But yeah, go check it out on Ollama. Very interesting stuff.

Adam: Yeah, mine was Jujutsu. So it's a virgin control system, sort of built on top of Git. Has been fantastic to use. Like just really a lovely workflow, makes sense, easy Git interop, you don't have to do a whole lot of shenanigans.

But it makes some really common operations, like rebasing or having like stacks of commits, really straightforward, and I think it's great and you should try.

John: Yeah, I saw that, it had some Mercurial-inspired features, which is always interesting.

Adam: Yeah, and I think we're kind of ready for a new version control system.

I know that that, again, we're straight back to me having gray in my beard, but like it smells like how it smelled when we got Git, like when we all like moved from CVS to Subversion, and then Subversion was everywhere, and then suddenly there was like SVK, and then there was like a Git bridge, and then Git ate the world.

And so like it feels like we're at a moment of experimentation where we're trying to find a path to what a better source control system would be. And I think that's such an interesting layer to play in.

John: What's the main driving factor you think? Like scaling or features?

Adam: I think it's fundamentally user experience.

John: Okay, okay.

Adam: Like version control, Git's a perfect example. Most people when they think of version control, it's the only thing they've ever used.

And the idea that that's technology that was built by people who made choices is like incredibly weird, do you know what I mean?

Like it's just the sort of thing where you're like, "Wait, why would you ever do that?" And like we talk a lot about like, people talk about like things being commoditized or whatever, and so it's like you shouldn't think about it anymore.

It's a good example of something that's really fundamental to how we think about work and how we work in general.

And that has really deep implications for everything sort of on top of it in the stack where if you could take your creativity and bend it toward, "Well, what would source control be like if it didn't work that way?"

Like what if it worked some fundamentally, like Unison, like in a completely different way, where suddenly what you could build opens up.

And so those sorts of places are where opportunity is starting to live because people who are willing to rethink some of those more fundamental concerns, they start to have the possibility of getting order of magnitude better outcomes, because by moving those fundamental underpinnings, suddenly you change the shape of the car.

You go from it's a '70s Pinto to a 2024 Audi, and like those changes, they could be dramatic. So that's what I think is driving it.

John: Ah, that's fascinating. Yeah, well, where can people find you these days on the internet?

Adam: Yeah, you can find me on Bluesky. I don't really post much on X 'cause I can hang out on Bluesky, and I would rather do that.

You can find me on LinkedIn surprisingly, which has turned into a useful little social network in a way that trips me out. And then I'm easy to find, you can always send me an email. I'm adam@systeminit.com.

John: Awesome, well, this has been absolutely incredible. You're a legend. Thank you so much for providing your thoughts. And listeners, remember, always stay ready.