Python documentary companion pod (Interview)

24:35And then a consumer of that come in and say, oh, to use this, you've got to buy this proprietary package before you can even understand my science. Like science should be reproducible from. I didn't want that behind licensing and IP barriers. Right. So I'm very much for open science, open production. I'm not a big fan of actually the, you know, I understand journals. They have to be funded, too, but I want papers to be available, easily accessible. So that was really what drove me in this direction was the desire to be a scientist and share. And then I started with software. And then I learned a ton about how software works generally beyond just doing scientific software.

25:16But that journey was because of my association with the Python community, because I was a young scientist writing scientific research code, essentially. and maybe some libraries and learning about, okay, data structures and packaging systems and algorithms and high-level compilers. That came later. And when I became what I would call one of the ambassadors for science to the Python community, some of the early ones were Conrad Hinson, Jim Huguenin, Paul Dubois. They were the ones who helped Guido add key features to the language. And so I go into this in the documentary. I don't know how much of it's going to be in the documentary when it comes out, but I talk a lot about that, about how critical it was for those early science ambassadors to the Python community to have a voice, to be able to be heard.

26:03Because Guido's not – he didn't always know, like, why. He wasn't himself deeply involved in scientific computing, but he was open to hearing the input. He was open to changing things like, okay, let's add extended slice syntax. Okay, let's add tuples can be constructed without formally creating the parentheses. You can create a tuple without just having the parentheses sitting there. And that was really critical for multi-index selection. And extended stride syntax enabled you to skip. If you had a big array, you wanted to skip elements evenly. You could represent that simply. Things like complex numbers.

26:40Like a lot of languages, they just say, oh, yeah, complex numbers are implementation detail. Now, the problem is complex numbers are everywhere in science and engineering. They're so fundamental that, you know, really, when they're in the language, it matters. Otherwise, you end up with 15 competing implementations of them and nobody can agree. So having them in the language meant that they were there from the beginning. So I likely wouldn't have dropped Python if complex numbers hadn't been there. And as I looked into it, how did that happen? It was Conrad Hinson, specifically, like him and other people who joined in the community and said, lovely language here, but we need a complex type.

27:16And then back and forth conversations with Guido said, okay, well, what letter do we pick for the complex representation, I or J, right? And the engineers won with the J. So, you know, that, those early conversations were, that was possible by what I'm calling like scientific ambassadors to the Python ecosystem. And there's actually been very few of them, unfortunately, like we need more, honestly, I played a role, Nathan, We had the matrix operator added finally. That was a missing element for many, many years. So that's a bit of drama around how did matrix operator get added. And Nathaniel Smith made that happen, right?

27:51And he was as an ambassador. What do you do, sneak into the repo at night and slip it in there? No, no, no, no, exactly. He didn't. So I'll tell you how it works. You have to actually go into these communities, whether it's a Discord channel, a public discuss channel now. Now, you spend hours basically talking to people, helping them understand why it's important, answering their questions, because a lot of them are coming from a space of, I don't know why this is important. And then kind of, so therefore, I'm negative about it. Yeah. And as the language matures, of course, it gets harder and harder.

28:24Early on, the affect was, okay, I don't know about this, but I'm a plus zero. So Python, this sort of story of negative one, negative zero, plus zero, plus one emerged. the early community said, okay, here's how we're going to vote on things. Right, plus one or minus one, but you can also do a plus zero, minus zero. Okay. Which is kind of, I don't really care, but I'm leaning one way or the other. So help differentiate the different ways people might feel about it. That's interesting. That's still, it is actually very, it's very useful. Yeah, I like that. A lot of people use that actually to help kind of navigate this.

28:55You know, a lot of people paying attention, a lot of bike shed people, but like who really cares? Right. If you don't really care, why are we listening to your voice? Right. Because you're there. Plus zero. Yeah. Okay. Well, plus zero. Great. That's good to know, but okay. Let's figure out the plus ones versus the minus ones. Understand that, understand that dichotomy and see if there actually is a compromise somewhere. Maybe it's a misunderstanding. A lot of times it absolutely is. A lot of times it's a misunderstanding. We get so deep in our awareness of something that's hard to be aware of another thing.

29:27I personally experienced that bring as a scientific ambassador coming to the Python ecosystem and getting some PEPs accepted. So I became a Python core contributor by adding some Python enhancement proposals and going through that work on a few fundamental cases. And that was great. It was exhilarating. It's also difficult, right? Because I was dealing with really smart people with knowledge in areas I was weaker in, and I had to understand where they're coming from and then respond so they would listen and then understand how they needed to be communicated with so they would understand where I'm coming from.

30:01Right? And we got, so one of the things about NumPy, so that was, that's how I got involved. And then 2004, 2005 timeframe. So this was the state of the world until 2004, 2005 with SciPy on top of numeric with a bunch of libraries people are installing. Usually installation meant either build it yourself with a tarball or rely on a guy named Christopher Golke who was handling, like at that point, probably several hundred windows.exe files on his own personal page. So there was no IPI. There was no, like if you want to install this stuff, you've got the, you've got the source code you built from scratch yourself, or you installed from this exe page.

30:42And one time I, in about, this is about 2000, when did we have this conversation? 2013, 2012, 2013. I talked to Guido at a PyCon and I was kind of just saying, so, okay, the packaging story, like it's really bad. Right. Like, and he goes, yeah, I don't really care about it. It's sort of, it's not something he really worried about or thought about until later he did. But at the time he was like, Oh, you know, if I have a, if I have a project package I need, I just put it in the standard library. If it's something I care about. That's gotta be nice. Yeah. Well, yeah, exactly. That doesn't really work for everybody.

31:17So anyway, that's kind of illustrative of part of the problem is that that wasn't sort of tackled from the beginning. And we had a lot of challenges in trying to just, how do you, we have this great language for getting the head of the person who just wants to solve a problem. And then they can think about their problem instead of the syntax. And super extensible. Like I can actually take Fortran code and make it accessible in Python very quickly. In fact, a guy named Piero Peterson, who collaborated with me on SciPy, he wrote a tool called f2py f2py would take fortran code analyze it and automatically create the python extension to to allow that fortran code to be called from python easily so he took what i was doing manually because i was basically manually doing that on a bunch of fortran code and he looked and said what are you doing that's idiotic and he basically built a tool to do that automatically so he's he's far more of a computer developer than i am i feel like i'm a integrator an implementer, an agitator perhaps, you know, someone who can try to facilitate.

32:21And that's why it led to me doing more entrepreneurial activity than development activity over my career. But anyway, that's the story of SciPy. We can get into NumPy if you'd like, or if you have questions about the SciPy story and the ambassador to Python and the kind of how did Python become a space where science is welcome. I kind of have a question that may not be directly related to Python, but more of the era because Python and Ruby kind of came out roughly in the same space yes they did and you said that you came for the language and state of the community and i think that ruby is just as welcoming as python has been to me personally and i'm kind of curious i've always been kind of curious why science related folks gravitated towards python the language and then even the community versus ruby because that's a great great question both perfectly capable yeah that's a great question a lot of this is timing i think like i thought about Ruby later after, you know, had already invested some in creating the Cypa ecosystem, right?

33:17So, and then new people essentially are just, well, what's there already, right? And like I said, I was there when numeric already existed. With Ruby, you know, where is the numeric Ruby, right? Who wrote that? Who wrote a numeric Ruby? Ruby had a really, really good community around web apps, and they're Ruby on Rails, for example, phenomenal story for building, you know, quality, in fact, they're way ahead of the Python community in terms of helping people build quality web applications on a Python back, on a Ruby backend. But there was no numeric Ruby. If there was, it was very nascent. And so it's kind of that critical mass question.

33:57And I think part of it being, I think part of it also was the language. I mean, Ruby had a, it was in Japan, a lot of, a lot in Japan, right. And I don't remember the name exactly, but Mots, Motsaki, Motsa, Motsa. All of them is Mats. That's all I'm calling Mats. Yeah, I just call them Mats, right? M-A-T-Z, Mats. Perfect, Mats, yes, yes. That's what I remember. So he was really, you know, really talented. Then the community in the United States around Ruby was in a couple of companies, right? But they're like this, I think what happened with Python is Python, because of this early cooperation impact from the Conrad Henson, the Jim Huguenin, the Paul Dubois, the David Asher, like that's what was needed for Ruby to really get that critical mass.

34:39And it just didn't happen for some reason or another. I don't know. Maybe it did. Maybe those people got tired quickly because it's kind of people in academic settings that were able to pull that off. Like, who's going to do this, right? It's going to be someone like me who's a grad student at the Mayo Clinic who's got more time than money and then kind of ambition than sense to try to do stuff myself. And there were people like me too. Eric Jones, a grad student at Duke University. P.R. Peterson was an academic. So it's kind of an academic mindset. Ruby did a good job of tracking the scripting, you know, the web app builder, but didn't appeal to the science mind.

35:17Syntax is part of it. I think what appealed to us in the Python side is it didn't have extraneous syntax. We did, you know, there's all this white space arguments, right? Well, the white space argument appealed to the scientists, right? Just, you know, putting braces where you didn't need them, putting extra line noise that the principle that Guido had. It really made a difference for people like me. right who were like i don't want this extra lime noise because partly because of maybe the ptsd from the pearl experience pearl like had all kinds of special characters you couldn't remember what they did now you know ruby's better for sure but i think that's a big part of it but honestly i think the question would be why didn't the numeric ruby get written right because that's what was there in 1994 95 and or did ruby does ruby have complex numbers like i don't think it does last I checked.

36:04Right. And a lot of computer scientists go, yeah, who cares about complex numbers? And that's typically what happens. You get computer science building a language and then the scientist comes in. Cool. Can we add complex numbers? And they go, you do it. It's on your own. It's like, well, okay, thanks. And then 15 people do it. And then there's like, okay, which one am I using? Who knows? Like there's a, there's an aspect of leadership from the center that is required to ensure that certain things get rooted. Now I can put a laundry list of things that didn't happen in the Python community that I wish would have that would have helped it even further.

36:35You know, packaging being one of them, having a really strong packaging story that's led from the early community instead of like an afterthought later. I think self-macro, like a way to not execute code, like to pass on, you know, back tick operator or something that lets you put code that is not immediately executed, like delayed execution on code blocks. with python you can only do that if you use like the width operator or you have to you have to write you have to write you're at a class or at a function like just being able to do that in line would have really helped in a lot of ways a lot of places and then i would love to see extended slice syntax available to be used as an argument to a function instead of only within the list operator anyway sorry that's just a couple of feature requests yeah a couple of feature requests that are really hard now to get in because you've got such a massive community yeah you got to convince a lot of people yeah it's a lot of people i looked it up while you were talking talking ruby has complex numbers but not until 2007 so back when you're getting involved there you go yeah a little bit late to that particular game and probably because there wasn't demand for it i mean a lot of times you just build what people are asking for or what you personally need you're not gonna go out and say you know what this needs is something i don't care about you know so you need the people to show up and say we want complex numbers then you know gato He just says, okay, let's do it.

37:57So it's a bit of a random walk. That's why I'm thinking. Yeah, I mean, you don't know what's going to be important, right? You know, because that's why I say, look, it's because Conrad Hinson showed up and found the language and said, yeah, I want this and started to participate. And that wasn't driven by, like, nobody budgeted for that. Nobody articulated that. Conrad did, right? Yeah. And so, and then other people like that. Paul DeWa got involved. You know, then Jim Hugin is a grad student. He was a master's degree student at MIT. He saw the matrix SIG. He saw someone built a matrix.py in Python.

38:25He said, huh, I'll try my hand at building an array object in Python. And he did. He wrote kind of, and it was good. It was like a pretty, really nice array object with multidimensional arrays and some compute infrastructure. And he pushed that out there in 94. And that was the foundation of, so that's what I would say. Where was the Jim Huguenin? And then later I came in and I felt like, you know, I definitely added a lot, but I mean, I wasn't even been there if I hadn't been for Jim Huguenin. And some of these early people that have come on. Snowballs. Yeah. Exactly. Yeah. NumPy came about because numeric was, was there and SciPy was like, that was my baby.

39:03I wanted SciPy to get usage and adoption and people and more, more because I want collaborators. It wasn't so much the users I cared about is I wanted collaborators. I like my people to jump in and help build some really good computing tools. I really got into loving the numeric computing, like solving science problems with computers. Well, right. That was a thing I really enjoyed. So I want collaborators. That was for me the driver. It was collaborators I could lean on. And then the Space Telescope Institute, Hubble, the people that shipped Hubble, they were starting to use Python, and they're using numeric, and they're using scipy modules.

39:36And they thought, how do I need a better array? They needed a better array, an array with more types, an array that handled memory a little better, was faster in certain circumstances. They had a couple of needs, right? And we were talking about that early as 2000. And so fast forward two years, they started work on a project called Numerae. And they started to sponsor a project called Numerae, and they started to build it. And they started to release it, probably 2003, 2004. So Numerae, it had some really good features. It was nice. It did some things that were better than Numeric did. And that's fine.

40:10Okay, you could use it optionally. And then somebody released a medical imaging library called NDImage that had – and I was jealous, basically. they released in the image and it worked for number a, right? And so, and the image had something called morphology and then end dimensional convolution. They did a great job of implementing it and it did it better than the Nd convolution algorithm I'd written. It was in sci-pi. And so, but it used numeric, I'm sorry, it used number a. And so now there's sci-pi on number on numeric and it's in the image library on number a. And the two communities were basically like it was competition.

40:48It was competition, but it wasn't collaborative either. Like, if you were using numeric, you had to copy your data over to a numerae. Like, it was – and for a scientific programmer, you end up using memory tons. Like, because a whole other stream of thought would be, what about PyPy? What about the JIT compilers? All that story, right? And we can get into that because I ended up in the middle of that later. But, right, in this point, it was, oh, how do I actually connect these two arrays? And I have to copy it. And I'm already struggling with memory. I'm using gigabytes of memory, and I can't use another gigabyte to put it in a numerae.

41:19so it's going to split the community and cause people to interrupt ability not to work very well so that's the problem i set out to solve in writing numpy well friends i'm here with damien shingleman vp of r &d at oz zero where he leads the team exploring the future of ai and identity so cool so damien everyone is building for the direction of gen ai artificial intelligence agents agentic what is off zero doing to make that future possible so everyone's building building Gen.AI apps, Gen.AI agents. That's a fact. It's not something that might happen. It's going to happen. And when it does happen, when you are building these things and you need to get them into production, you need security, you need the right guardrails.

42:03And identity, essentially authentication, authorization, is a big part of those guardrails. What we're doing at OZERO is using our 10 plus years of identity developer tooling to make it simple for developers, whether they're working at a Fortune 500 company and we're working just at a startup that right now came out of Y Combinator to build these things with SDKs, great documentation, API-first types of products and our typical Auth0 DNA. Friends, it's not if, it's when. It's coming soon. If you're already building for this stuff, then you know. Go to auth0.com slash AI. Get started and learn more about Auth4GenAI at auth0.com slash AI.

42:48Again, that's Auth0.com slash AI.

42:56So that's the problem I set out to solve in writing NumPy. And in 2005, I basically was a professor at BYU at the time. And I said, ah, this is a problem, but who's going to solve it? Like, nobody's going to do this because, you know, it's complicated. And who knows enough to be able to do something like this? So there was a semester, a class fell through, and so I didn't have any teaching load. And so I said, well, I'll just do this. I'll take the next three months and do this. Why not? Okay. I'd written a bunch of modules for Sci-Fi. I'll do this. That was in 2005. So that's where NumPy came from.

43:27That's where NumPy came from. Okay. And a big part of the strategy was actually, well, how am I going to get an option? Well, let me go talk to Guido. Maybe we make NumPy and we put it into Python. So I started to talk to Guido around that time, around 2005, about how do we make the Python community more aware of the NumPy ecosystem, and maybe by putting NumPy into Python itself. And I met with them about 2005. We talked about it, decided against it initially, but that is where the idea of the buffer protocol and the memory view object came. So the memory view object and the extended buffer protocol, there was one already, we made a better one.

44:04That came because of those conversations. And I would say we got the data structure of numeric of NumPy into Python, if not the library. Right. So that was that was the and that's how I became like this deep embedded ambassador for a few years, probably about seven years. I did that for a while. Did that solve the problem of this division and this competition? Yes. It solved the problem. It absolutely solved the problem. So NumPy was wildly successful, way more successful than I thought it would be. Honestly, it blew me away. It was it was work. And I actually ended up losing my tenure track position at university because I spent too much time on NumPy.

44:41Oh, really? Yeah. Now I know why you're an entrepreneur now. That's what I'm – yeah, exactly. No, there's a deeper story to that. But it depends on which version of the story you want to hear. It's a little soundbite. You never get to the full story. But it definitely had an impact on my academic career. So – but it was okay. It was okay. It was something I really wanted to do. Hey, you're a Python martyr. You know you're a Python martyr. Python martyr, exactly. Exactly. Hey, if that helps people, if that helps people launch their careers, fine. Right. Now what I do is I try to say, look, if I can be helpful to you to build a company, to build a Python organization, to build a project, I would love to help.

45:18Let me know because I've been so benefiting from other people as well. I want to help you do the same thing. Let me know if I can help. And that's what I do now. Right. So it's a long history of all kinds of stories. You can see there's all kinds of avenues and places where you can literally you could have you could have like a Netflix series. Yeah, totally. There's lots of different ways to go. With the right writers who can begin the drama, you know, because there is. There's drama. There's people. And there's conflict. And there's agreements. And there's, you know, I don't know if there's any love interest.

45:44I don't think we've had enough romance, actually. You can always write that into the script. You can write that into the script. Based on a true story doesn't mean it is a true story, you know. You've got lots of liberty. I mean, I've had romance. I've got a wife. And we've married 30 years. We have a great time. Oh, congrats. That was a long time. Yeah. So that's a bit of the scientific side of Python, your subgenre, so to speak, and its rise to prominence. Or why so many people adopted Python? Because the early adopters came, they laid the groundwork, people like you came after them, built tooling.

46:19Really, it was like, this has the tools I need, it's a nice language, and it grows and it establishes, and people just start to choose it. What about the industry side? Why did Python grow through industry? My only guess is like because Google used it and like people want to work for Google. That's a great question. And it has this similar multifaceted story narrative that also coincides with individuals using it. And those individuals then go get hired by the companies. So one of the very first was Google because Google started to very early on, they used Python as kind of a, oh, this is a way for us to find great developers.

46:55Because the only people using Python are people who care enough to go and do this on their own. And so they kind of used it as a selection filter to find great developers. And then early on, okay, you're here. All right, you want to use some Python? Okay. And they started using Python internally. But it was more because of the people they pulled in who wanted to help Python grow. Gotcha. But they built it. They've used it quite a bit. But lots of languages are supported inside of Google, right? They've supported lots of this open source growth. I think initially it was the people that drove it.

47:27And then my experience personally has been in the financial industry and in the scientific use of coding industry where it was they were trying to solve problems and the scientists need to solve a problem. One character, for example, is a guy named Karat Singh who had made the rounds between Goldman Sachs and then J.P. Morgan and then Bank of America until they went to his own company. And every time he was basically trying to build a system for derivatives management. You know, and many people will understand and know a little bit at least about the derivatives empire, the fact that, you know, we have these underlying equities, then you have derivatives on them, the derivatives value get ballooned up and it led to the 2008 crisis, all that stuff.

48:11Well, about 2008, suddenly that's when, you know, all the investment banks said, we got to get our handle around this. And so there was requirements to build a system that would manage their risk assessment or their understanding where they are. And Karat Singh was there helping them build that. And he went, oh, well, let me just use Python. rather than Goldman Sachs have been in their own language. In fact, if Goldman Sachs had actually released their language as open source back in 92 or 93 when they built it, we might all be using slang, right? Because they had something pretty good inside of Goldman Sachs way back then.

48:41It's called slang. They didn't release it. It's called slang. Slang and SecDB is a database. Never heard of it. No, exactly. Neither is anybody else. Yeah, they should have released it. You might have if they had a list as open source. It's actually a good story to enterprise. And so enterprise is starting to learn that story, right? This is also part of why I think companies start to come to Python because companies, so part of it is just the developers and part of it is companies realizing that they need to rely on these open source projects. There's been building dependencies on it and they should start participating in it, right?

49:11And how do they do that? And I think that journey is still with us. I think companies haven't figured that out yet. And, you know, that's one area that I love to talk to people about because I think I have some insight, Like having seen so many things in this direction, both inside of big companies as well as part of the communities that are there, there's some, okay, where's the friction and where's the real opportunity? But, you know, these financial communities, they all sort of build these tools around Python, right? Because really of these key connected people, again, and people have written books about this.

49:43You know, the Mavens, I think, is the book that is the word they use to these these key connectors, these influencers, I call it technology influencers. Yeah. And it's kind of, you know, it's people that do a bunch of work. And then because they do the work, other people go, cool, you've done the work. I'll use it. OK, right. And so then they build from there. It's how open source is diffused throughout throughout the world. But companies start to rely on it for risk assessments and risk assessment is mathy. Like there's a bunch of simulation, there's a bunch of array math you're doing. And so all of that leveraged the NumPy, SciPy, and related ecosystems that were emerging pandas.

50:20So Wes McKinney was at AQR, one of the hedge funds that was heavily on Python. He and every single other hedge fund, heavily on Python, built their own data frame object or something like a data frame object themselves. Wes McKinney was able to convince AQR to let him open source it. and then he spent further time in 2011 2014 shepherding that early release to make pandas what it is right so and he was a friend connection part of the sci-fi community we knew him he was doing this thing there are other libraries are also similar for a while but pandas ended up dominating because of his his effort and influence but lots of other companies now are going ah darn we got to convert internally to pandas because of that same issue like so companies have learned Oh, either I open source what I'm relying on or I risk technical debt of having a code base to maintain that nobody cares about.

51:15And the new people are coming in with different infrastructure, different story. So that's something I think has happened with the AI world. And I know specifically what happened in the AI world because of TensorFlow and PyTorch. And then in 2018, when I left Anaconda, which is a company I founded to help make packaging work better and then data science better, actually. Basically, I saw, oh my goodness, I wrote NumPy to bring the array communities together. In 2018, there were literally 20 others, but they're all around this AI story. There's MXNet, there's PyTorch, there's Torch, there's TensorFlow, there's so many of these other array libraries, effectively.

51:55Now fortunately, we'd fixed it twice in NumPy. Not only did I create NumPy, the array library, but also created a buffer protocol so that even if new array libraries existed and showed up, you could still share data. So there's a fix it twice concept. And then that's what we did get in there so that, okay, great. There are all these array libraries, but at least you can share data now. You don't have to copy data from one to another. So that's at least a saving grace, but they're all over the place. And each one of these companies, they did it. Primarily, PyTorch started as a Lua library. There's a C++ Lua library, and there's three versions of it.

52:29TensorFlow was a C++ library called Disbelief. Each of these have their own internal organizational C++ stories. They started to launch them in 2015 timeframe, 2014, 13, 14, 15, started to put these libraries out there. People started to use it. Well, at this time, there had already been 13 years of the SciPy ecosystem and the SciPy conference. And a lot of academics in the very first machine learning library, Theano, was right in the middle of the SciPy ecosystem. So a lot of these first early array, these early papers on machine learning were Python-based. And so when these big companies came and shipped their libraries, the community said, where's your Python interfaces?

53:05And demanded their Python interfaces. And then, oh, okay. I mean, I heard this story specifically from several of these people inside these organizations. So they went back, okay, and they built the Python interface. Sometimes badly, TensorFlow. Sometimes better. TensorFlow fixed their problem by buying Keras. Keras was a good interface. And so that's the interface for TensorFlow now. that's great. But PyTorch from the beginning, because they were in the research lab, they, from the beginning, tried to build community around what they were doing. And so because they did that, they were actually able to get more scientific embedding.

53:42So a lot of this is how a company goes about doing open source matters a lot in terms of its longevity and how it's going to be, I think who's going to win. You can predict based on how they're approaching the open source ecosystem. And PyTorch did a better job than TensorFlow. Even though Google invested in a group to try to make that happen. But the problem is it's not just having a few people do some podcasts and do some DevRel. It's actually how are you managing internally? Like, what is your story internally? Like, do you like on the meta side, they had PyTorch as a public thing. And then they had the meta application of PyTorch that was separate.

54:15On Google, the difference between the TensorFlow development and their internal usage of TensorFlow was not very separate. So they really couldn't open TensorFlow to wide scale, input because they depended on it heavily. And therefore, they couldn't just allow some random pull requests from a community upset their own internal usage. So you have to manage that. You have to manage your, if you're going to be serious about engaging with others in your open source community, you have to put a bit of separation between your community-driven sponsorship or community engagement and your internal development processes.

54:50If you don't, you're going to, you're not gonna be able to manage it. You know, some version of Conway's law, you know, where the, the software resembles the organization. The corollary of that is if your organization relies on the software structure, you're going to hurt your organization. Cause the, you know, you can't have your development teams and your delivery teams dependent on a random pull request from somebody in Estonia that you can't, that isn't working for you. Like there has no, right. You have no way to kind of keep, hold them accountable other than just stop using their code. So anyway, there's a lot to be said about that actually, as corporations started adopting open source, this is not well understood by the open source communities often, or it's better understood by the internal development.

55:32teams because they have to, but open source communities sometimes don't understand that. And then they get a little confused by corporate actions. They make the wrong assumptions about what companies are doing and why. And that can lead to some friction, misunderstandings, and people feeling like they shouldn't support something because they're reading into a corporate action that isn't true. It's just a consequence and the lack of a PR statement from somebody in the organization. Anyway, that's a whole other topic, but we often get involved in conversations like that with the corporate side. But the open source community side, it's just a massive ecosystem.

56:15And corporations adopting open source because it helps them. Actually, Zuckerberg probably said it really well most recently. I don't know how he feels about it right now with the recent AI changes. He's kind of, but previously he said things like, and I think it's valid, By releasing open source, you can actually influence the direction that the industry is going to more align with your internal choices. So you don't have the slang moment at Goldman Sachs or the Pandas moment at any number of unnamed hedge funds that had to adopt Pandas instead of their own data frame because they actually recognize the open source, the way to align industry to your roadmap.

56:55That's certainly one way to contribute it for sure. I mean, if you're that far ahead to be a leader, I think a good example is probably like Borg and Google and Kubernetes. Yes. Borg was great internally but didn't make sense externally because Google is Google, right? But Kubernetes, an orchestrator engine made a ton of sense. Yes. And the whole story is different now because of that. Yeah. And actually, great point because I actually knew the person who was in charge of both TensorFlow and Kubernetes at the time. And they wanted to kind of figure out how to, you know, get community involvement in those two.

57:28Kubernetes succeeded. TensorFlow did not in that regard. And I think you have a great point. And part of it was because they didn't really depend on Kubernetes internally. Right. It was just a thing that came out of their story, whereas they did depend on TensorFlow. Yeah. Yeah. Is part of what you're saying, and maybe this isn't the point of this conversation in a grand scale, but it's probably saying is when corporate control permeates into the development life cycles of the software, it begins to mimic what the corporations want versus what the community and the technological folks need. Yes.

58:09Yes. It's actually that juxtaposition of corporate constraints and therefore corporate control and community needs, which is broader. Because the community needs includes other corporations, but it's basically the broader set of infrastructure needs that everybody needs. And how much of that is the corporate willing to give up for the benefit of other people contributing? Right. Because, you know, why would a company do this? It's the 80, 20 rule. Exactly. Exactly. Because a company could just do it themselves. Great. Just do it yourselves. They have their own story. Who cares? In fact, a lot of companies have often, I've given talks at organizations where they want me to talk about how do we adopt some of this open source ethos in our 100 ,000 developers?

58:53I mean, we have 100 ,000 people here. And we would love some of the energy that happens in the open source ecosystem to happen here in like a private mini, you know, open source enclave. Just even if it's not to the full world, but at least it's inside of our company, people are more aware of it. Like, how do you do that? So we've gotten into some of those conversations about what are the incentives that drive open source? How do you mimic those inside of an organization? Can you? One of the critical things to me, and this is also why it goes back to the people stay for the community, it really comes down to respect and ownership and accountability.

59:26And do people feel that in the community? Do they feel like their voice matters? They feel like when they contribute, it's not just ignored, but it's heard. And then furthermore, do they do something that they feel pride in? Because this is the thing I built and I show it to the world. And that's what open source enables for engineers that inside of a big company doesn't happen as well. You go to get another job and you're like, cool, I did all this great work. Well, can I see it? No, it's over there on their repo I can't show you. Whereas if it's open source, here it is, I can show it. And so a lot of developers, a lot of engineers have felt like, oh, this matters to me in my career.

1:00:03I need to do that. Now, what I'm finding is that it's actually some of that spirit, that energy, that ethos is not there among the early, among the new developers coming in that there used to be. And so, you know, a lot of people new to programming are like, their GitHub repos aren't great. They haven't really contributed to open source communities. They're kind of, let me just get that job. And my advice to a young developer is go find a few open source communities you love, you get excited about and start participating in them. Just show up, hear what the problems are, contribute something to it.

1:00:38If it's documentation, even if it's just, oh, read me for a newbie. Like everybody, if you're new, great. You have a power nobody else has in the community is you're new to it. Everybody else has been there a while. They someone new adopting it. So great. Document your journey and write a story for a new person, how to make this more accessible. There's a lot you can do, anybody can do to participate in open source and do that. Okay, maybe you can't do it 20 hours a week, but do it for five hours a week. Do it for an afternoon. Do it for a weekend. That'll have more impact on your career than sending out resumes because it just helps.

1:01:13There's a correlation between your participation in open source communities and your ability to contribute to a corporate community. Now, it's not one-to-one. It's not identical. And the reverse isn't always true. There are also people who struggle to participate in open source communities because it takes a bit of thick skin, a bit of I can work remotely with people, I can communicate via text or email, and maybe you're not good at that. You're better in person. You're better at that. And so it's not the only way. But if you do, if you can participate, it's easy. Well, you can always just go practice that leak code problem.

1:01:49You know, that's what they're doing with their time too. Yeah, that's true. That is true. It's like, which one do I do? I don't know. No, I've had that conversation with someone on LinkedIn who was asking me, how do I do it? I said, well, okay. But, you know, I can just share my experiences, right? I mean, you know, some of my experiences might be relevant for the future and some might not. You know, it's because things do change. But I'm happy to share my experiences. and if anything is helpful to somebody, I'm more than happy to share. I'm super eager to help people. Like I want to see a world of more owners.

1:02:21I want to see a world where there's distributed opportunity to collaborate. What I've loved about the ecosystem I participated in is when people come to the table, like it's a round table. Like I like these round tables. I understand the role of hierarchies. They're helpful for certain things. But I love when you're in a room where people feel like they can contribute and they have something they have skin in the game and they have something they're trying to contribute i've definitely seen the decline of the github resume yeah i wondered about that i don't think it's gone i think it's declining and but that being said i don't actually disagree with you i think that people should do that um but maybe do it a little bit more i'm not saying you have applied intention but like maybe do it for the love of the game and i think maybe like your your portfolio, your resume, whatever could be a nice side effect of you, but actually it's fun.

1:03:13It's fulfilling. Thank you, Jared. Yes. Pay it forward there. That really is satisfying. Yes. Nothing I do, nothing I've done would have been possible if I didn't love it. Yeah. The only reason I did it because I loved it. It was actually something I really loved doing. I didn't know if I, I mean, it wasn't because, oh, I get a job. If I do this, the opposite sometimes, like actually I lose my job. I do this. Yeah, I've seen a lot of people that they're like asked about their open source work and they don't really have any. And so they think, oh, I need to go have some. And if the goal is to like have, you know, the green dots on your contribution graph, well, of course that can be gamed.

1:03:48But if your goal is like to have some open source because you need to have some open source, that's ultimately going to be empty and not very fulfilling. That's kind of the same thing. And not very impressive either because you're like, oh, cool. I opened up a PR on some meaningless thing. and so like if the right intentions are there and stuff it all works out if not and then it is you might as well practice your code yeah thousand percent agree jared yeah yep very well said side tangent you were talking about all of these the tensor flows and the pandas and then like all these things that were like kind of competing or burgeoning out of in this messy way on the scientific side of Python because of machine learning needs and because of industry driving these things.

1:04:33And that got me thinking about Mojo. And I just wonder your opinion on Mojo, which is the Pythonic language that's like right in that wheelhouse. Yes. Yes. So I'm excited by this kind of innovation, right? In 2012, when was it? 2011, 2012. 2012, I started a project called Numba. At the same time, there was a language called Julia that was emerging. And Numba was a Python compiler that took a subset of the Python language and made machine code to make things fast. And we actually, by 2013, were targeting GPUs. So you could actually write Python code that ran faster than any C++. Like we were showing that in 2013 at the GTC conference at NVIDIA because you could basically write Python code that would run on GPUs directly.

1:05:23So I love that whole space. Like this is something that it took me a while because I was – this is when I finally learned how to write compilers. Now, I wrote the first version and then quickly found other people to make it better. But I actually took a pilot class from my friend David Beasley in Chicago to try to understand – at least have something besides just what I picked up from the Internet. right and that helped me see the vision for oh python can actually be orchestrating compiler tools and there's no reason you can't just write syntax that's pythonic looks pythonic and then have it compile the machine code there's zero reason that can't happen so there's no reason you have to use c++ or russ to write machine code you can totally write a subset of python and have very fast code and number proved that and proved it pretty well actually and so and then since then there's been like 20 other versions of that like right now there's a if you go to lpython.org lpython stands for llvm python it's an example of a compiler for python lpython.org and a table there shows a list of other similar projects some of them are a little bit more like mutica where they're translators from python to c++ others are like python is another one codon's another one that's out there.

1:06:37Like there's literally a dozen, all right? And so - Yeah, I'm looking at the table. It's probably two dozen. Two dozen, yeah. Now, all of them are kind of this example of scratch your own itch, something that's built. You know, Numba has a team that kind of is still working on it. And so it's actually received a lot of support. And so it's kept up to date with the latest releases. And there's a lot of, every time the byte code changes, Python, Numba's got to change because it goes from the byte code on. And you can actually write, There's also a tool called Cython out there. Cython was in the same – it's not exactly the same, but it was like, oh, write it in this language, and then you can write extensions quickly.

1:07:13And even something like F2Pi might be considered in that category. But Cython, Numba – so I love this space, right? I would say – so Mojo is great because Christian was one of the original creators of LLVM, and Numba depended on LLVM. Like, so I've been an LLVM fan for a long time. So I'm like, well, this is awesome. We get kind of the OG kind of bringing Python to compiled languages. Awesome. What's he going to do? Let's make that happen. I think what I'm hopeful for, and I know he's got funded and there's a company structure there. So some of the same concerns about, OK, corporate capture of communities and how does that work?

1:07:51I don't think he's actually even had those conversations yet. I think those are, you know, so right now I'm very, very hopeful and excited by the technology innovation that will occur. And I'm cautiously optimistic about what that might mean. And I know that he wrote, I think one, very concretely, he transformed Objective-C to Swift, essentially, with LLVM. And so I think part of his thing, and I talked to Christian, so I know a little bit, but I don't know him 100%. But I've had a couple of conversations with him about what does he think, where is he trying to go with this. I think his experience making Objective-C users move to Swift was pretty easy, right?

1:08:30Because Objective-C is kind of all Apple community, right? Right. Python's much, much different. You're not going to have the same experience. You're not going to be able to take the whole Python ecosystem and move them to Mojo in maybe a decade. You could pull that off or billions of dollars. Money might be able to, but that money has also got to be able to hire the right people to have the right connection points so you're interacting with the right communities because you can't just have great tech and then push it on people. And then, but so I'm eager, I'm open. I'm really eager to see it, but I'm, okay.

1:09:04Is Mojo open source yet? Is it, you know, is it community driven? How are decisions made? And it's all early days. So I don't, I don't, I don't, I don't, these are hard things. So I'm right now, I'm of the attitude. I'm excited. I love seeing what he's going to come up with because he's a brilliant guy. And I want, I want to see an explosion of languages that are Python-like that enable the Python ecosystem to actually make fast runtimes starting from Python syntax. I know there's a lot of enthusiasm around Rust, and I don't need to dampen that enthusiasm. I like Rust, too. But I think there's far more low-hanging fruit in just having statically typed Python.

1:09:45In fact, I call a little project we're working on at OpenTeams, but it's very early, and it's not a competitive mojo by any means. it's called I call it post Python performance optimized statically typed Python okay and it's basically saying oh Python now has type annotated types optionally annotated types right now you can use them for lots of things but if we just with this post Python language definition spec so post Python is about creating a spec and a definition with lots of runtime implementations So in my ideal world, you know, Mojo, Numba, Cython, Kodon, all these, and then Jax, PyTorch, these are also compiling subsets of Python.

1:10:29PyTorch does it directly. Jax does it directly. If these could all have, oh, here's the syntax for a statically typed sublanguage that if you format this way, then you can compile it. And then these become compilers. So in my world, you have a standard that allows people to, and then you have implementation of that standard. You can almost think of like SQL. So will we get there or will Mojo become that by de facto or will PyTorch? That's a good question. I don't know. Yeah. I would love to see more cooperation here. One of my concerns is these are challenging problems. And so the communities in them end up being siloed pretty quickly.

1:11:04And they don't look at each other very often. And so because you've got – I'm just naming a few. There's many, many more that I've seen over the years. And they're all siloed. And they all don't really talk to each other. So hopefully we get cooperation between them. Do you think VC funding is the reason or promotes the silos or no? I think it's been promoting the silos primarily. I think where we've gotten the funding from the corporates have helped break them down, like the big funding. But VC funding has encouraged silos. I've actually been an outspoken critic, not that outspoken, but I am a critic of the GitHub star funding model of VCs.

1:11:41right for a while there it was like oh your github i see your github here has many stars here's five million dollars right and the trouble with that is there is not a one-to-one github repository to company mapping you can't just trade those in for dollars yeah i mean i don't mind and in fact we could have another conversation i i think i've got a way to actually create a market for that but it's not as simple as because i have a vc fund too i've started a fund, right? For this purpose of helping entrepreneurs create. And we've got 14 companies invested in. I've got an idea of how to actually create a marketplace where you can actually have GitHub stars be a market driving question.

1:12:25And investors are actually looking at, okay, this project's going to be used. Great. I'll invest there. And then have that investment dollar go to support the development of the project while also having to return some at some point from somewhere. Because that's the thing you got to figure out. You got to build an instrument that drives the return of the investor from the risk they took to invest in that project early? And a company is one way. Yeah, just have a company and ownership in the company and the company sponsors it. That is. But can we do it in a way that actually allows many companies to collaborate and still have the benefit from the time that somebody spent investing on that project?

1:12:59Inure to them as the companies that rely on it develop. I think I've got an answer to that, actually. It's my BHAG, my big hair audacious goal is to bring that to market in the next five to 10 years. Really? BHAG? Can you say that again? BHAG, big hairy audacious goal. Oh, BHAG. BHAG. BHAG, it's a dumb acronym, but it's kind of like dreaming. It's what I'm dreaming about now. It's like, how do you actually make open source intimately connected to the growth of companies? And how do I make more owners? How do I contribute to making more owners? Obviously, I'm not going to do it myself, but how do I contribute to a world where there become more owners?

1:13:41That's one of the problems of my lifetime is our Western societies, especially the United States, have decreased ownership. There's more and more money in fewer and fewer hands. And that's not great for society. I don't say your phrase. You'll own nothing and be happy, Adam. Yeah, you say that, Adam. You'll own nothing and be happy. He brings that one up a lot. No, it's true. You'll find yourself owning nothing and being happy. And okay, yeah, there's a pill you can take for that happiness. But I mean, I would love to hear this idea. What do you think, Adam? I mean, are you sharing this idea right now?

1:14:16Are you just building it privately? No, it's out there. It's out there, but I'm building it privately. It's under, but it's been incubating. The challenge is I can't, it's, it's, you go to FAIROSS.org. you can see the beginnings of that idea which we we we started a couple of several years ago and now we're just resting on because i don't i know what to do but i don't have the money i need about five million dollars right right so if if somebody but that five million dollars can't have a bunch of strings attached that they got to basically say here it is i'll check check back with you in five years right that's what i'm looking for okay um yeah it's not how investment for.

1:14:51I'm also looking forward for that. Check back in five years. Yeah, exactly. This is why, well, you can check back in there. You know, we can read reports, but I know what you mean. We're just, you know, it's like, you know, this is going to like, and really for the 5 million to return money on it is going to take five years. It'll have impact before then, but like, you know, it's seeding an ecosystem and seeding a story. The idea is pretty simple, right? Effectively, you just have an organization that, on the one hand, works with open source communities to document their dependencies, both their dependencies on code and their dependencies on people.

1:15:32So essentially, I call it a it's their end dips, their millibips table. Right. And they basically have a basis point. And if you do a thousandth of a basis point, you basically have 10 million units to give out for every project. And a project governance essentially gives out basis points for dependencies. Like, oh, I'm NumPy. I depend on Python a lot. So we're going to give, you know, 40 % of our cap table to Python. Right? And then from the rest of it, we're going to, you know, award it to people. And you have someone reserve. And as people come in and make contributions, the governance can actually go, oh, here you go.

1:16:08Here's, you know, here's ownership. Here's, so you basically have one side helping people organize that because there's a specification. You stick it in the GitHub table, then it's there. And you can say, oh, this is you, the project, have decided a cap table, basically. You've constructed a capitalization table for the project, right? And it depends on the governance. It's not our decision. Now, what we can do is to drive that as Fair OSS could essentially have a default approach to doing that. So it says, well, this is what we'll use if you don't tell us what not to use. This is how we're going to do it.

1:16:41And then, of course, encourage them to do that. So we don't wait. We're not gated by whether they participate or not. So you build that. And then to the companies, you go to companies and say, hey, you're missing from your cap table, your employee option pool, your investor group, the warrants you're giving out to partners, because this happens all the time. You're missing all the open source ecosystem from that group. And so let's get on your cap table. So we get on the ownership table of a company. And of course, you know, we can't get a million open source contributors on the cap table. Okay, great.

1:17:14We put an entity on the cap table. And the entity on the cap table represents the open source you depend on. So you have an entity on the cap table and Ferrosys maintains your dependency, like their millibips table. Like, what do you depend on? Right? And then as equity returns happen, which is distant usually, it's like as a venture capitalist, I know how this works. And you basically, over time, value will flow. Now, if you want quick value, it could be maybe you're not going to give up equity. You're a mature company. Equity is hard to come by. But we'll do a dividend agreement, like anacondas and dividend agreements.

1:17:50We'll do an agreement where we'll just allow some of this to come. Our venture fund, for example, we already do this with our carried interest. Like we have carried interest, which some of you know what that is. It's the motivation for the partners to find good deals to invest in for their LPs' money. And then they get a percentage of that growth. And so we, our general partner, some of that carried interest already goes to fund open source. Right? And so it's the same thing. We can formalize it through that process. You've got all these. Now you have all these inputs. You have equity deals. You have dividend deals.

1:18:21You have carried interest deals that are flowing through this instrument that becomes the funding source for all the open source dependencies. and then as opportunities arise you're just streaming it out distribution style to the to the cap tables and it has it flows down the cap table tree because oh you're you're dependent on pandas well pandas looks like oh it's dependent on python and numpy and it flows down until you end up at the leaf nodes where there's individuals who are basically getting you know participating so that's the basic idea and then what i did i tested that out with a few companies could i actually get these agreements could i get people to agree to that and yeah and i thought i'd have to have a venture fund, actually.

1:18:58The whole reason I started a venture fund in 2019, because I thought I would have to go to startups and say, yes, I'll invest in your company, but you have to do this. You have to put open source on your cap table, right? Because I knew they would say yes to that deal. What I found is I actually got people to say yes without the money. They understood the problem. Yeah. And so I went, oh, okay, that's good because it's like RegEx. You use RegEx to solve string parsing. Now you have two problems. You have the problem I was and then understanding regex same thing was starting a venture fund same thing was starting a company i started anaconda as a company to solve the problem how do i fund numpy how do i make how do i fund scipy how do i build this thing so my whole life has been like okay here's a problem how do i solve that problem here's the meta problem here's the next problem how do i solve that problem and they're just trying to learn along the way and just you know try to figure out ways that are you know learning from other people trying to organize because what drives me is I want to create a world where open source is a meaningful way to live a life.

1:19:55You can be an open source contributor and still pay for your kids and still have it in the, that's a job and it's meaningful. It's not just a thing you do on the weekends and nights and when your day job is doing something else. So this is why I've been thinking about this. And FAIR OSS is the epicenter of this? Yeah, it's the first version of the organization we've created to do this, right? And it's an idea of how it could be done, and we've tried it enough to know that I know it will work. But for it to work and to create the vision of where you can actually have investors, because eventually we'll have a ticker symbol for every open source project, literally.

1:20:32You can have a market where, oh, how's this open source project doing? That's going back to your stars. Your GitHub stars literally could translate to a value, a fair price, a fair market value for that open source project. And investors are looking and going, oh, yeah, I want to buy some of that. Right? Because they could. They could say, oh, I'll put in my portfolio that project because I think that's going to go somewhere. And how is it going to go? Because people are going to depend on it, and therefore enough of these agreements are in place where value will accrue because of those agreements in place.

1:20:59So FerroAssist puts the instrument together. It's responsible for creating the instrument that connects the dots between the investor money and the open source communities. What you need is a dependency graph versus the GitHub stars. I understand the analogy. No, no, you have to have the dependency graph. But that's also what FerroSS manages is the dependency graph of those millibit stables. But that's the evidence to show the, you know, should or should not be on the cap table, right? Like that's not a star, which is I like this or this is cool. Correct. That's not what gets you on the cap table.

1:21:32What gets you on the cap table is one, it's just one entity. So, you know, a company is not going to put a thousand people on a cap table, but they might put one entity on the cap table that represents those people. And you're saying that at some point for an open source project that may be of that caliber, they should have or might have either their own foundation representation or FeroSS is that representation. Maybe, yeah, FeroSS could be that representation. Maybe they could do their own. Ultimately, it's a market, right? So markets are very complex with lots of parties, right? But the project itself, all they do, all they have to do is just track their millibs table.

1:22:10They just have to put and document and record, well, who gets what? Like who's accountable for this? It's like a thanks file, but it's like a thanks file with a number next to it, like a percentage. Yeah, it's their own cap table. It's their own, and they manage that. That is used to build the dependency graph that allows the flow through to happen from the entry point, which is just the company that puts FaroSS in the cap table also submits their own high-level millibit table. Like, well, this is what we care about. And then FaroSS also, there's stuff around here to support that. What you need is to be in the IRS, you know, the tax code to make this have some serious incentive behind it because at that point, every company would actually do it.

1:22:56This is a separate question. Potentially, I agree. I'm trying to avoid creating a ministry of open source. like yes some of this could be facilitated with government as long as it's automatic and and rule-based and not and not bureaucratic well not even so much irs but like recognized by the business nature of the united states right like llc versus s-corp versus c-corp there's a reason why companies form one of those because they're recognized no 100 in fact in fact the way this would be implemented is actually through an ll series llc yeah like because essentially essentially what you'd end up with FerroSS doing, like what it does, it says, okay, we have an LLC representation for you project or your corporate or whatever it is that's representing your, because some products have one and some don't, right?

1:23:42And so anyway, yes, I agree with all that. I would say to your point about the IRS, what I would love to see, I would love to see the United States actually as a separate thread, have a sovereign wealth fund and allow deferred taxes to be paid. Instead of you have a tax, you just give a part of your equity to the sovereign wealth Fund in the United States. If they can allow that to be used for, there's a ton of places where it's the tax code. Like I've been looking at this a long time and so much of the inhibitions, the problems we face are because of outdated SEC code, outdated tax code that was meant for a different time.

1:24:14It's now stopping innovation in the business world today. There's so many, and so many lawyers, so many people get paid, so much effort is spent, you know, essentially preserving yesterday's infrastructure, yesterday's business infrastructure. sure it's hard to innovate in and there's no legal precedence for it like there hasn't been a case correct represented so you can't say oh yes we should or shouldn't do this because there's no case law that represents correct exactly how the courts may or may not you know lean in proceedings so a big part of the challenge is figuring out how do you fit in that existing case law but still do something innovative right which is which leads to a lot of challenges it really does so are you doing this yeah that's what that's why it takes five million dollars at least to get this going We've done it already.

1:24:56Look at this$5 million. Let's get this money. Well, if someone wants to buy anaconda shares, they can buy them from me, and I got the money. So that's what I'm doing with my next tranche of anaconda shares. You need some liquid, man. I can get liquidity if someone just wants to make a bet on anaconda. There you go. I'm happy to share that bet with them. So just, you know, if you hear this, contact me. And maybe you want to fund this, but you want to hedge. Great. You can own anaconda shares. Let me take the risk. I'll do it. Okay. So that's an option. I like that offer. Yeah. Anyway, so that offers out there to anybody who wants to talk.

1:25:26I like this idea. Thinking about it slightly more deeply, it seems like there's a bit of an impedance mismatch between the value provided by the open source projects and the companies that would then adopt. Because you're basically relying on, you're going to be adopted by winners. And you might not be. Yes, correct. But you're still valuable, but just to losers. You know what I'm saying? It's just one part of the funding fabric, first of all. I don't think it's the answer to everything. Sure. But it's something that should exist. And it's also the investor class. That's the part that we, you know, it's not just the companies.

1:25:58But when I realized there's at least$100 trillion of investment money out there looking for alpha, which is like looking for return. And that's in people's retirement accounts. That's in investment accounts. It's all over the world. All that money is looking for investing. Where are they going to put it? And so a lot of times, you know, property values go way up. Stock values get way overvalued. Like you need a place for that to go. So you have that investment value. then you have open source where innovation is happening because you look at the past 10 years and innovation that's occurred from ai to to cloud to to data right it's open source communities have been at the heart of that but yet there's not great connection between the investor class and the open source community class okay so imagine imagine a world where pharaoh ss is running okay yep and there is it's active and it's vibrant yep and i'm an investor and i want to put 10 grand on NumPy.

1:26:49Yep, exactly. What happens? There's a site you go to and it's like it's traded on a public market. Maybe it's crypto. Maybe there's a crypto token associated with it. That's what I was hoping that crypto would actually help this emerge quicker onto public markets. Like there should be a place you just call your broker or you go into your own online thing and go, yeah, I want to invest in NumPy. Cool. That's a ticker symbol. You just put it there. Great. Now at the moment, that's going to buy somebody else's position, but the project Like NumPy could say, you know what? We're going to make a release.

1:27:16We're going to authorize more of our millibips table to investors. And they funnel that up through and say, we've got now$10 million that can be purchased by new buyers. It's a kind of IPO. And you're buying their upside. And you're buying their upside. I see. So again, the NumPy community and FerroAssets would mediate the relationship between the community that has to essentially go, yes, we're willing to take investor money. because they're the ones that have to take it and do something with it through their own governance or however they're going to do it, right? There's somebody that has to take that money.

1:27:46FerroSS just facilitates the connection between that and the market. And actually, that particular entity would be a different company that would be actually the broker dealer between the projects and the market. That's a separate activity. The current thorough SS organization is intended to be the arbitrator of the milli-bips tables. The middleman, yeah. And then kind of the negotiation of the contracts with the companies, right? And then you have a broker-dealer that would connect the two and make a market. Because long-term, you have a marketplace. The Chicago Mercantile Exchange exists. Why?

1:28:21Because somebody put the idea of an option and a future out there and started selling them, right? It's just that. The New York Stock Exchange exists because the idea of a corporation was created, and then somebody started to sell them. So you make a market for it. This dog could hunt. How much have you been able to convince these companies? So I haven't really – the problem, when I left Anaconda, I had to build this whole other thing. And so my time has been spent building Quansight, OpenTeams Incubator, my funds, and then OpenTeams now. And so I've spent so much time there. But this has been in the back of my mind since about 2019, 2020.

1:28:54So I've had conversations on the side with multiple companies, enough to get dividend programs. Like Anaconda has a dividend program. Lots of companies are willing to the dividend program. That's an easy way to fund it. And then I've got a few conversations with funds to say, hey, here's carried interest. And then startups, I've had 100 startups willing to put Pharoah on the cap table. That's been super easy, actually. Well, it's a good question. So the highest is like 10 % of their company they'll put on the cap table. 10 %? Mm-hmm. That's a lot. It is. It must not be worth very much. it's startup land right because startup gets yeah right and it's like and the argument really goes like hey here's a way to get and for that they're going to want more what they're going to what they want is connection to the open source communities they want for their hiring pipeline and for improving their their because a startup your most important thing is the people you hire so if ferrocess can provide an opportunity to hire then it's worth that right so it's worth so So that is high.

1:29:55Mostly it's like one to 2%. It's pretty easy. And as the company matures, you know, maybe a big company like Anaconda, it ends up being like 0.1 % is all it is, right? Or 0.5%, right? But that's like, you add that up and that's, that's a lot. Like if you actually took, you know, and you only have, let's say, you know, a few basis points on every company. And a lot of companies they're, they're throwing, I mean, they're putting more of that in their option pool for employees. And that's kind of where you're, that's where you're arguing. You're saying, okay, here's your option pool for employees.

1:30:27You need to recognize that there's a portion of this that you're not either taking advantage of and you're missing. And that's where I've had the best conversation with people. They go, oh, you're right. That could help. Now what can, how, and then what they're looking for is, well, how do I, the company wants to know, well, can I, can I connect that to feature requests? Can I figure out a way to kind of connect this to community connections or hiring or, and that's where we can talk about that, right? Because it It could be a milestone-based vesting on that pool. There's ways to negotiate that.

1:30:55But that's what FerroAssess does is create the negotiation to increase the size of the instrument market. That's why I couldn't – my view is that has to develop, and then we can build the marketplace for the broker-dealer where you can actually get investors to come in. But you have to build this before the investors will come in. Anyway, that makes sense? Anyway, that's my big, hard-assist goal and what I'd really love to work on next. You just need some liquidity, man. Just need some liquidity, yep. Anaconda's not going public anytime soon? Not soon, but they just had their Series C. It's actually doing really well.

1:31:27And they're very, you know, it's a good example of a company. I can tell that story numerous times. A lot of people can tell that story. I, of course, have my own version of the story. It's doing well. It's doing really well. But I don't know when it's going to go public. I'm out of the loop on that. So you could ask Peter. You could ask the board. I could ask them. What I've heard is they're not really anxious to go public quickly, but they did just get Siri C and got a new people. Actually, I shouldn't announce that. I'm sure they're doing very well, but I'm not authorized to say that they've had that.

1:32:00Bandicata is doing very well. Anything else? This has been a wide range and an interesting discussion. Adam, you have any either follow-ups on FaroSS or something else? I mean, I'm just going to scrutinize it more, really. I want to see it happen, but I want to scrutinize it more because I'm like. Yeah, scrutinize. Give us feedback because it's still early. Right. And so the idea was, let's get it out there. Let's get talking. Let's see how to make it, how to adjust it. Well, I mean, like I'm I'm easy to please with it because I want it to happen. But then if you take Jared's example where he says, hey, I'm a venture capitalist, I got 10 grand, I want to put it on NumPy.

1:32:31What happens was this question. But like, what do I get? What is what is the incentive for that venture capitalist to put 10 grand on NumPy? They get the returns that later come. So remember, if you get on the cap table, have you ever you want a cap table? Have you been an owner of a company? You're like you basically thorough assess has ownership of a company, but that ownership is tied to flow down through the dependencies. Right. So so basically thorough assess is is the is the arbitrator helping as people use NumPy more and more. Now, all of a sudden, Netflix and Amazon and OpenAI and Google, they all basically depend on NumPy in some way.

1:33:11And even if it's like, you know, a millionth of a basis point of the value of that company is now flowing through FerroAssess to the NumPy community. In aggregate, it could be hundreds of millions of dollars, actually. People have told me, people said, you know, NumPy has had billions of dollars of impact on the economy. I'm like, great. Where do I see that? where do I measure that? How does that show up? We'll pause every second. So back to the venture capitalist with the 10 grand that puts on NumPy. Well, it would appear to be an investor. It might not be a venture capitalist would probably put more than 10 grand.

1:33:43They'd probably want to put a million in early, right? Okay, well, pick a number. Just tell me with the Robinhood app, you know? Yeah, that's later. The Robinhood app is what I want to get to, but that's going to take a little bit of time. But then you have 10 grand. You put it there. Where are you going? Well, one, you have other buyers, right? Because every market, you know, if you think about stocks in general, why do you put money in stocks? Well, maybe they'll give a dividend someday, but no, because you're going to sell to someone else. Right? So it's the same thing here. Is there a buyer for this later?

1:34:12And the answer is sure, if the market's real. And why would the market be real? Because there is an eventual value. What exactly are they buying? Help me understand that. They're buying ownership in the millibips table that NumPy is governing. And remind what the millibips table is. What does that consist of? So millibips stands for a millibips, a thousandth of a basis point. So there's 10 ,000 of them per percentage or 100 ,000 per percentage point. So it's a percentage where you're buying is a percentage of the NumPy project. So and then, OK, cool. I have a percent of the NumPy project. What does that get me?

1:34:47It gets you any value that flows from the FaroSS infrastructure to that project. So let's say in the future,$100 million is flowing to NumPy. from companies whose value either FaroSys sells the position it has and flows it through or it earns dividends. Basically, there's numerous ways FaroSys is creating actual transactions with the market that flows money back to the open source community. So it's aggregating that. And then downstream, some of that aggregation is coming to the NumPy project. And as an investor, you have a right to a percentage of that. Okay. That's what you're buying. It's like buying a stock in a company.

1:35:29It's the same thing. Why do you buy a stock in a company? I'm tracking now. I'm tracking the full cycle. We've had to create the instrument to enable the flow to happen. And so an investor could see, because that's what an investor is going to see. Before the market can develop, investors would have to look and say, okay, wait a minute, how is this going to work? And so that's what takes work to develop that market. It takes work to develop the relationship for the companies, the flow through from their value to the rest of the market, and then the aggregation. And it's a technology problem. There's stuff to do because you basically have, you know, what are they?

1:35:58There's like 5 million open source projects. I kind of think 500 ,000 or a million of them are probably, you know, so you're going to have to onboard them. You know, FaroSS will have, okay, we have 10 ,000 projects we're supporting. Now it's 100 ,000 projects we're supporting. You know, it's going to be a pipeline there. On the same time, FaroSS is going to have to build. We've got 1 ,000 companies participating. Now we've got 10 ,000 companies participating. Now we've got 100 ,000 companies participating. The market's growing in both directions, right? You've got to build both sides. So which side do you build first?

1:36:25It's a combination. Like you have to do both kind of, it's like every, every bootstrap problem you have to do, you have to attack both at the same time. And so it takes work and it takes, you know, a lot of conversation, selling, communication, you know, trust building. Let me, let me be more clear with Jared's question. What is the most clearest form of a good next step to make this happen? Oh, for me, for me is, for me is the money because we already did, I already spent as much as I could. I probably spent about somewhere between half a million and a million on this already. Just from the consulting company I had, and we were just taken from money we didn't send back to our investors.

1:37:10Instead, invested in this idea. So we've taken money already to invest, but hiring people. Because I hired people, and they did work on some things, and it takes resources to do that. So the next step is that money. That's why for me, the next step is to get that pocket of money. And$5 million is really$2 million is the minimum. But I'd like$500K to invest in this company for 10 years. That's why$5 million. It's got to be funded so that there's people who are hired to pay attention to this problem and are working towards it. And I want it for 10 years because this is not something that's going to be solved in three months or six months.

1:37:46It's something that will need time to develop. And the company is the FerroSS company. FerroSS is the FerroSS company. Yeah, it's investing in FerroSS. And why will it be valuable someday? Well, it's the broker-dealer, ultimately, that will create value of FerroSS. That's why FerroSS itself is a public benefit corporation. Because I don't know if it's going to have value or not. Its purpose is a public benefit. It's to build the market of open source. And so whether it has value intrinsically or not, who knows. But the broker-dealer will at some point. Because it'll be the same as any. It's a transactional.

1:38:19It's there because the brokering conversation with investors and their money. These public benefit corporations remind me of the structure of that. Is that a C corp that has a different classification? Correct. Correct. It's a designation. There's a few version versions of this. The public benefit corporation is in Delaware. You can basically tell in the articles of formation, this is a public benefit corporation, which means you've told the secretary of state, as well as every other investor who might invest in this company, This company exists not only to maximize shareholder value, but to also achieve some public benefit.

1:38:50So you can't later, a shareholder can't invest in the company and later come back and say, hey, you didn't have a fiduciary responsibility to me and maximize my investment, which is kind of the default corporate law. They go, well, actually, you have to prove that we weren't supporting the public benefit that was told you from the very beginning. It was in the corporate shotter. So it's an innovation. I like it a lot. It's actually a really good innovation because even though even companies that don't have it they still benefit the public like a lot of people have this the idea that some of corporations are somewhat inherently evil they're not they're just a there's an organization to kind of pool money together so you can do stuff do stuff their motivation is different than than a human that's the problem and it definitely is true definitely there's an agency capitalistic by nature i mean by design it's the whole point it's the capitalism by nature yeah exactly that's their focus so the public benefit corporation allows you to insert an incentive structure.

1:39:44It's a communication. So every shareholder goes and then management, you hire management and management can go, oh, we're attuned to this public benefit, not just maximizing shareholder value, right? It's actually this public benefit too. And so it's a nice innovation. There's also something called a B Corp. A B Corp is a different thing. You could have both. A B Corp is you get a license from a firm. So a B Corp is a branding thing. You basically pay a fee and get a license from a firm and the firm has ways you can show that you're a benefit corporation. Whereas a public benefit corp is just an institutional infrastructure thing.

1:40:17You publish, you record in the state of Delaware is where I've done it. Other states, you can do it too. That's the thing about corporations. They're state by state. Delaware is just the closest thing we have to kind of a, you know, everybody, a commonplace people do it. But that's what it is. It's a public benefit corp. And, you know, several companies have been doing this, particularly around open source, because people go, wait, as open source contributors realize, okay, how do we use the mechanisms of institutional capitalism to support our mission? And that's what I like to do. I'm not going to call myself a capitalist, but I definitely am a – I like people, and I like distributed ownership, and I like people to be able to be free to work with each other, and I like decentralization.

1:40:57And so I'm just like, how do we use this infrastructure to further the means of our open source desires? And so that's at the heart of what's basically driven me for the past 20 years. basically. And, you know, had some success, but some failures and I'm still, I'm still trying, right. Still working at it and trying to find other collaborators, other people that are, that are similarly interested. And, um, uh, you know, open teams is the latest one that's, that's, it's launching right now. And a condo was with them first. I did some consultancies before that. I've done some consultancies as well.

1:41:30And I didn't talk to a lot of people. I'm very eager to collaborate with anybody else thinking along these lines, because this is something I would love just to exist. I'm like you, Adam, basically. I'd love it to exist. So, okay, well, kind of like NumPy. I'd love it to exist. Well, nobody else is doing it. Maybe I should do it. Kind of the same way. Okay. I'd love it to exist. Nobody else is doing it. Maybe I should do it and we'll see. Okay. So I got to imagine that as part of this, we'll drum up some interest to some degree. We do have some variants of influence. So you need roughly$5 million to fund fair OSS.

1:42:07For the next 10 years. For the next 10 years. So half a million dollars. It'll need more than that, but that's enough to make sure that we can hire people because I'm going to allocate and I'm going to hire people and I'm going to commit to people. Then it'll self-fund. You'll hire some people to bring on projects. You'll have, let's say, 10 or 20 or 100 of your initial pool of projects that will show off there. What did you call it again, this table? the MIPS what was it called? Millibips Millibips the Millibips is there a different name for that? yeah please come up with one yeah I do like it it's just forgettable yeah I hear you and and easy to confuse with something else yeah fair enough fair enough I like the idea though because you're selling access to this table which is not truly ownership in the in the project but future value it can create it's future value it creates and so it's a separate file Like you have the thanks file.

1:43:04It's separate because, and it's the responsibility of the governance team to update it. Like Pharaoh assess when it onboards a company will have one. Right. And basically we're like, okay, here's our, here's the table we're going to use for your project and who gets to, and you can change it anytime. The governance of that project gets to change it. I think that's more important than that. I mean, it puts a heck of a lot more importance on like, it does really open source. And that's also, I'm aware of that. So I'm also like, huh? Cause I, cause I also seen projects when, when money starts to get involved, culture can change very quickly.

1:43:36Yeah. I mean, motivations totally change. I mean, it can, I feel like I put my work out there and somebody else is getting the dollars from my work. I'm feeling, then you're going to create, show up. Yeah. I feel like Pharaoh assess is going to need to have like almost a, like an omnibudsmon office, right. Whose whole purpose is to, is to, is to help resolve disputes between governance questions. It also feels like a license to fair opens, Fair OSS reminds me of like the license movement. Yeah. I kind of, that's why we kind of have a, we kind of have a groundwater. Exactly. If you look at the site, we've kind of, we've got the groundwater program.

1:44:14Like it's like, you have this idea and then you get a label. So the project could say, I'm participating in this groundwater program. And then companies could also say, I'm participating as a means to attract developers. Cause for companies who participated, they're basically like, why did they do it? Well, they want to hire great developers and they see there's a way to do that. They recognize the problem of they're going to bit dependencies on open source they can't maintain. And they want to figure out a way to, you know, pay ahead for that or figure out how to, how can I reliably rely on an open source ecosystem when that old XKCD graphic of the dependency that's in the middle, it disappears.

1:44:47Then what? Like, a lot of companies are realizing this is a problem. So there's, but there's more to do there. Like, this is going to, like, it's why I say it's going to take effort and consistent effort and the right kind of people. Like I'll have to have dev rels. I want to hire ambassadors. I want to, there's, you know, all the people involved in open source for multiple cycles. I want them to tell their stories and to come and talk about it. And then there's a podcast we'll have to promote. I mean, there's lots of things to develop here. But my perspective is, you know, if and when Anaconda actually becomes not just paper, not just paper value, but real money, what am I going to do with that?

1:45:25Build this, right? Build a way to give back to everybody else. it's got a shot it's got a shot i think well i appreciate that sharing that and that's kind of yeah it reminds me a little bit of the t.xyz conversation we have but not like exactly one-to-one but similar nature and it's similar nature and i've seen a few emerge i have i spoke to him yes and that's what's excited me recently is i've seen a couple of these were like oh okay we're on to something here this you guys have a similar concept and i would be happy to join forces with some of these folks too. Like, this is not a, like there is zero desire to be the only founder.

1:45:59Like I've been a founder. I know what that is and I'm fine doing it, but I don't, I'm not, I'm doing this to support a concept. If there's other people that want to take the, like I'm happy to be a spokesman. I'm happy to just be somebody that promotes it and that negotiates people and incurs them why. And I'd love other people to own it. I'm going back through our chapter markers. And when we were talking to Max, we talked to Max end of last year at All Things Open. We've got All Things Open every year. Big shout out to All Things Open. Yeah, yeah, yeah. Great, great group. Anybody there, I would love to talk to people there too because I did talk to T, and I think they were the closest of the ones I've seen.

1:46:35I need to revisit that again. That's actually who I'd love to talk to. Collaboration is key to this. I don't know what my idea was, but at one hour and 12 minutes of that conversation, it says Adam shares an idea. I don't know what the idea was. It may have been a really terrible idea. No, honestly, find out. Yeah. I would go back and listen to that. And I would definitely circle back to Max because you got similar desires, different implementations, but similar desires. I will. Thanks for reminding me about that. I will. Cause I do remember. I will. Thank you for that. That lead. Again, that reminder.

1:47:13You guys are great. I think what you do is actually very valuable because it gets people talking and then listening and reaching out. We listen a lot. We talk very little. sometimes sometimes a lot sometimes i share ideas that are bad apparently we'll see if it is the chapter title did not say adam shares a bad idea that's true if it was pretty sure i wrote that chapter title and i would have put that in there if i thought the idea was bad so it's probably good jared liked again she was a great idea i mean we care about funding open source and i think the mechanics around have always to some degree been uh ambiguous or flawed i think it's hard this idea did not come to me until I'd spent time in open source, time in Wall Street, working in investment banks, and then time as an entrepreneur, then finally as a fund manager.

1:47:58Essentially, all these different things have led to, oh, here's how we can put it together. Because it's like all these different facets of a big problem. Here's a big idea for you. Let's see this. This chapter is titled Adam Shares an Idea. It could be a bad idea. I'm thinking I'm in this mindset, personally where I'm thinking about my family bank and I'm not talking about like infinite banking. I'm talking about like ways I can set my family up that I can be the one who does the hard work to establish the foundation for which my future generations can borrow and lend against build wealth, have assets, et cetera.

1:48:36And I'm thinking like, gosh, I love software enough at this stage of my life. And at this point in my life for the future family and destination and the sentence would come from me that I want to say this family cares about software and so much so that it's open source software. I would want my family bank, if this was a thing, to invest into the pool of open source for these things. And it sure would be great if my foundation, I started as a part of it or whatever, can benefit because of the value that I invested in in open source. Like that'd be kind of cool. I think you need more of those kind of movements where you have people's long-term generational wealth or lack thereof starts to support open source.

1:49:22I totally agree. So I've actually – right this summer, I've been setting up a family office, which is just a company that holds the investments instead of them – Asset protection, et cetera. It's driven honestly by the fact that my taxes have gotten complicated and they're breaking me. And so I'm like, oh, I just need to push that to another organization that handles that. And so, but it's exactly that. And then that's gotten me in the family office community. There's a big community out there, family offices, family banks, right? We're in similar circles now because I'm doing the same thing. I'm trying to figure out that stuff.

1:49:53Exactly. Well, we can share notes on that on the side. I'd love to connect, share what we're trying to do. And there's just a lot of ways to do it. But I agree with you. There's a big world out there. And why not? As people become prosperous, have a way to kind of extend and help the world be better. The AI revolution is here. It's coming. the AI of transformations here, I want it to make more people wealthy. And not just a few people, but like a lot of people. And then pass that on. I want a world of peace and prosperity. I'm naive enough to still have that dream, but I actually believe it's possible with good ideas, with the right ideas.

1:50:25But there's also, you know, there's roadblocks and there's potholes and there's bad directions to go. And so that's why information matters. That's why cooperation matters. That's why I think open source communities can be a part of it. because in open source communities, you have real people working together to do something. And that's how anything great is done is real people, real communities. And how do you then help them and allow lots of them to flourish? So yeah, this is really good. And you're exactly right. It's like art, right? A lot of family offices invest in art for obvious reasons, right?

1:50:58Yes. Open source is art, right? It's an asset class we've never been able to invest in. Yes, you're 100 % right. You're 100 % right. 19 years of my life chasing it, exploring it, sharing it. Have you been to Italy and to the Florence and seen the great artists of that era? And you realize how – when I went two years ago, I finally realized, oh, it was concentration of these patrons that sponsored Michelangelo. They didn't come out of nothing. It came out of a concentration of wealth that wasn't distributed to the people to enable them. It was patronage that enabled that. And you're exactly right.

1:51:33We need the same kind of thing, but not just with a few wealthy people, but how do we do it for millions of people to participate as the patrons? Yeah. And in return, they're benefiting. It's got to function first. So I think that's where the hard part is. You need to get that half a million per year for the next five or whatever it is and do it. That's right. That's right. Figure out how to get past that. You need that first. I totally agree with you. That's what drives me right now is, okay, doing these other things, I still got to make a living myself. I'm not quite at the point where I can just focus on this.

1:52:01I've got to figure out how to, you know, I've got clients and other people I'm supporting so I can, you know, support my family. And then, but you see the idea of, oh, eventually I might have some means to invest. Cool. What am I going to do with that? And that's where this came from. Awesome. Super good to talk to you guys. You guys are awesome. This was great. Loved it. Let us know as you make progress. I will. Stay in touch. Be part of it. Help spread the word. Definitely stay in touch. Cool. Definitely stay in touch. Love talking to you all. All right, Travis. Thanks so much for being on the show, man.

1:52:31Okay, take care of my friends. You guys have a great day. Bye.

1:52:37Well, I don't know if Fair OSS will take off quite like Python did, but I appreciate anybody and everybody putting work into helping open source maintainers do their thing. So I hope Travis succeeds. If you are listening to this the day we ship it, the Python documentary comes out tomorrow. If you're listening to this anytime after August 27th, 2025, it's already out there on the YouTubes for your enjoyment. I've been previewing a rough cut and it's pretty awesome. Our friends at Colt Repo do amazing work. Speaking of amazing work, shout out to our partners at Fly.io and to our sponsors of this episode, depot.dev and authzero.com.ai.

1:53:19and thanks of course to our beat freak in residence the one the only the mysterious brake master cylinder that's all for now but we'll talk to you again on changelog and friends on friday

1:53:57Thank you.