Jerod Santo: ChatGPT is the biggest news since Stable Diffusion broke the internet a few months back. If you haven’t heard, ChatGPT is the latest release from OpenAI. It’s an improvement in a few big ways over GPT3. Notably, ChatGPT has long-term memory of up to 8,192 tokens, which means it can recall the contents of conversations. This feature helps enable the killer feature of ChatGPT: a very good chat interface where you can talk back and forth with it, instructing it to modify what it generates.

The chat experience plus the fact that OpenAI is allowing free use of ChatGPT for now has generated countless tweets & toots demonstrating its abilities over the weekend, some of which are truly astounding. The most impressive to me, and the one we’ll link to in the show notes, is by Jonas Degrave who discovered that you can build an entire virtual machine inside ChatGPT. From there, he uses curl inside ChatGPT to access OpenAI’s API… which is where things start to get really interesting. Here’s a quote from the article:

“So, inside the imagined universe of ChatGPT’s mind, our virtual machine accesses the url https://chat.openai.com/chat, where it finds a large language model named Assistant trained by OpenAI. This Assistant is waiting to receive messages inside a chatbox… We can chat with this Assistant chatbot, locked inside the alt-internet attached to a virtual machine, all inside ChatGPT’s imagination. Assistant, deep down inside this rabbit hole, can correctly explain us what Artificial Intelligence is.”

I’ve seen this movie before…

Luis Ceze: Yeah, absolutely. So the answer to how you keep up is really having a strong community and nurturing it, and being [unintelligible 00:35:37.25] and encouraging more folks to participate. Looking back, I’m very grateful for the community, and I think we were lucky to have been involved in helping catalyze that community, because somehow, luckily, TVM was able to capture the interest of folks that build the frameworks, folks that build new classes models. [unintelligible 00:35:56.21] but it’s essentially different aspects of your model that makes them more powerful and more general needs to be supported in TVM. All of that were actually contributed by community members, and we help make that happen; we put a lot of work ourselves, too.

[36:18] But then the hardware vendors are the ones that actually really fuel the pain, to be honest, of this growing complexity of the ecosystem that you’ve put very well, Chris. Hardware vendors today, they have to write low-level libraries that are tuned for their hardware targets, for each one of the major operators in these models. Every time models change, they have to go and tune it again… So they’re always having to update it, and that’s not sustainable. That’s why TVM automating a lot of that made it very attractive for them to contribute to TVM. And they wanted it to be open source, because they also want to enjoy the simplification effect that the community has.

So since we incubated Apache TVM into the Apache Software Foundation, there was even more interest, and industry became more comfortable in contributing, because now there is professional independent governance for the project. Before it was a few grad students and a couple folk, some contributors sitting in a virtual room, or folks at the University of Washington, too.

Anyway, so that was a long answer to your question, but basically, it is by having an open source community and having the technical incentives for folks to contribute to it. That’s how we deal with the growing diversity.

Yanis Caritu: I will try to not be too much technical, but as you pointed - we need to find something which is equivalent to an RGB image, or like the frequency in the odor, and so on. So what we are measuring are quantum of molecules with different properties, and so it’s difficult to have sensors that are on a unique dimension. So collecting all this data within the specific universe of odors to limit a bit the complexity - you can find what we call orthogonal axes in representation of data, so that you have virtual sensors, which are not exactly the sensors you have, but you try to find the different axes which are not correlated, and which put you in a world which will provide you with something that is not so one-dimensional.

Justin Searls: So DevOps, just like so many things in open source, became a hot and trendy buzzword that was heavily marketed and associated with either products or sort of halo projects when it comes to recruiting in like big tech companies. And the original idea that DevOps would be like test-driven development, and if you just gave developers testing, they would incorporate it into their team room, they would automate away a lot of the pain around testing and quality assurance, and then the intrinsic quality would increase at a marginal decrease in that team’s ability to deliver things quickly… And in part, accelerated by the fact that they no longer had other people to have to communicate requirements to, so that things can be tested. So like the theory went, if we just did that with operations, we would get the same lift.

And to me - I had that experience, and it was called Heroku. You know, it was the most DevOps thing that I had ever used in my entire life, was being able to say “git push heroku”, not have to think about my operations at all, but know that it was like taken care of, that I had answers to every question about scale, and about adding on additional components, without necessarily having to turn it into my side hustle or my day job or my identity.

But DevOps as a term has changed, as I think the Agile era of the aughts sort of undervalued and played down the importance of operations as a practice. I think a lot of the people who are the Linux sysadmin archetype of the late ‘90s might be seen as sort of getting their comeuppance now or their day in the sun of lots and lots of new innovations and technology that are focused on meeting the same kind of just core desires… You know, some of it’s like “Hey, how big can we make things? How fast can we make these? How can we automate all of these very fancy and cool, but maybe a little arcane and unnecessary at small volumes and scales, like orchestration of like lots and lots of real and virtual systems up in the cloud?” So DevOps and automation tools have enabled and empowered lots and lots of really cool stuff.

And my experience, of “I just want to be able to “git push heroku” and have my app work in the cloud and not have to worry about it ever again” is, I think, still the pinnacle of what I would want as a developer. And of developers that I’ve talked to that have had that experience in real life, they all wish that we could still have that.

[28:10] And Heroku still exists and it’s still a thing, and I love the people there and I love the product, but clearly, it’s not a flavor of answer that the market is searching for, because everyone thinks that they’re going to need Google scale and Facebook scale kind of tools for the job that’s in front of their very straightforward CRUD app, with very few users. And this is all of a piece with sort of startup culture that everything needs to be a billion dollar unicorn to be valuable, and so you have to presume the conclusion that of course they’re going to reach that scale, so then you may as well just on day one reinvent the universe in AWS through all this automation.

So DevOps as an overall meme in the industry I think has been net negative, and slowed down a lot of teams by way of distracting them, where the fact that teams now have to hire a certain number of DevOps people, quote unquote, “to full time just keep the hamster wheel spinning of their cloud-based computing”, whereas before you might even have had an on-premises server that was just sort of sitting there and was just on and worked…

That’s what I, in spite of the poll results - I think like 44% of the people saying sped up… I think some percentage of those people are just people who like really geek out about DevOps technologies and kind of don’t care and are just team pro-DevOps… And some percentage are just people who like living in the ideology that we live in and probably just never had the experience of what if you could just set it and forget it and not have to worry about it again? Because if it’s a means to an end, why would you want the thing that required a ton of effort and thought and complexity and specialized skills and so forth, and constantly having to read up?

So I’m coming across as pretty anti-DevOps here, but I think that when you look at the replies, the number one point of contention is that no one has a shared understanding of like what we mean by the word “DevOps”. And so just to focus on automation here, it’s - yes, I love real automation, but I don’t think that what we’re typically describing around DevOps related activities is like actually automating anything, in terms of actually automating away a problem.

Ryan Carniato: I mean, yeah, how do you compete with that? I mean, you don’t. When I first started out, I wasn’t worried about adoption, so it was very easy for me. I just built what I thought was the best possible thing I could build. And then over time I just kept on building it better and better, hopefully becoming more obvious to people. But the truth of the matter is, even if people would want to, so much of the ecosystem is already locked in. Even if you could picture perhaps that every aspect that I’m telling you today improves on what you’re used to in every possible way, I know that won’t be enough.
This is one of those things that you know it’s unfortunate, you know it’s like a David and Goliath type thing… We just continue doing what we’ve always done, which is to innovate. And the more people who come on, slowly, surely, as long as they’re enjoying what they’re building and they’re getting what they want out of it, that makes me happy… Because as long as I get to keep on doing this, as long as I get to continue to push the boundaries… What we’ve done as a community, the way that Solid has impacted the frontend, I would say this changes Signals. We haven’t seen something so revolutionary since the virtual DOM first showed up 10 years ago. And this wasn’t coming out of Meta, or some large company. This was completely grassroots. And it went on to change the whole shape of the frontend ecosystem. So we did that. And even if big corporations will take their time and be careful, because they should, that’s just the kind of reality.

Because for a while there, I was thinking – you know, there’s always that checklist… And I worked in b2b for a while, so I’m familiar with that on like the enterprise - or I was in the education side - of where there’s just this neverending checklist basically to keep the big players in. That checklist never finishes. When Solid first came out, it was like “Oh, maybe when it’s 1.0. Maybe when it has this feature”, or “Maybe when there’s a meta framework”, or “Maybe when it can do mobile, or universal rendering.”

[01:20:12.17] And the truth of the matter is we’ve checked out all the boxes, one after another, as things have come on, and people will be able to move to do new things, exciting things when they can, or they won’t. So yeah, I think the biggest thing for us in the community was it’s never been about that. It couldn’t be. We showed up way too late if we were going to win people over on popularity. But if our ideas succeed - well, that’s something.

Aiden Bai: Hey, I’m Aiden Bai, I’m currently a high school senior up in Washington State. Next year I’ll be going to the University of Washington for Computer Science. But what we’re here for is Million.js. So for the past few years I’ve been working with several other contributors on this project. Essentially, it’s a virtual DOM replacement for React. By wrapping your components in a block function, kind of how you do with Nemo, you can speed up the virtual DOM. So essentially, imagine if you wrote React Components, but they were actually fast.

Steve Guggenheimer: It is that ability to amplify human ingenuity or amplify human capability that is so powerful… And I feel comfortable because we’re early enough in it that I worry less about the Skynet scenario. And the truth of the matter is the kids coming out of university today - they’re gonna grow up with a new set of tools; they’re gonna grow up with AI, blockchain, virtual or mixed reality, IoT… So they’re gonna grow up with a new generation of people–

Benji Vesterby: [00:06:23.05] Yeah, so security as a whole - if we take and look at the breadth of the security community and everything that it entails, the scope is fairly large. If we just take a broad look, you have infrastructure security, you have application security, you have network security, you have… And when you start breaking down into smaller pieces, then you start looking at “Okay, well, what are the subcategories of infrastructure and network and application? And how are those things intertwined?” I mean, you start going down the rabbit hole of “Okay, we’ve got infrastructure that’s hosting virtual machines, which have containers running on them, that are hosting Kubernetes clusters”, or - well, not containers running Kubernetes clusters, but VMs that are running Kubernetes clusters… Which were then hosting pods, that are running containers, that these mesh networks inside of them… And each of these different layers has a different layer of security. You’ve got networks and networks, you have all of these different levels of different security.

So what’s happened is you have this really, really large scope of necessary, essentially training and knowledge that has become necessary to know in the security community. And one of the ways the security community has gone about trying to solve this problem of a lack of knowledge or training has been to develop these Capture the Flag events… Where people can challenge themselves to learn new skills in a more engaging way than just sitting down in a classroom - which most people in the security community are not great at. And it allows them to engage more with the actual knowledge, and do it in a more real-world scenario where they get to actually attempt to attack a system, or pivot into a network, or use a multi-layered vector in order to actually infiltrate a specific application, or something like that.

It’s not specific to security, because the idea is just problem solving. It’s just using a different mentality to problem solve. And in fact, a security expert dealing with something like a capture the flag actually falls more into the mindset of somebody that you would normally meet in the QA. Somebody whose primary role is to actually break an application, if you start comparing that to security. And then if you start looking at like challenges, or problem-solving type events like this, the Association of Computing Machinery has had this for a long time, through universities and things like that, where they’ll actually host these events. And you have students competing against each other, in competitions and in Academia, solving computer science-based challenges for awards. So it’s not that it’s a new idea, but the security community has kind of really popularized it, and made it more of a generalized conference type event. And I think it really pairs well with the software community.

Gavin Wood: Thanks for having me. So I’ve been coding since probably around the age of 7 or 8 - I can’t remember exactly when - on an old 8-bit barely supported computer with no games, which kind of lead me to the necessity for actually coding.

I did school, I actually made a few games then, and I actually ended up publishing one of them in a magazine at the time, which was kind of nice. I got through university, and then I did a masters in computing and eventually a PhD, which was specialized in music visualization, an interesting subject in and of itself.

I spent a short stint in the games industry, working with a company called Frontier Developments, which some of the older listeners probably remember as Elite, one of the original 8-bit space trading game, which was pretty enjoyable. I left there to go and do some consulting work on and off. I worked for Microsoft Research for a while, doing some of their more advanced and API-driven projects, working on things like video synthesis and embedded them in specific languages. After a couple of startups I ended up getting to Ethereum. I met up with Vitalik back in late 2013, and we started coding the C++ implementation, which was kind of more or less operational about a month later, and is probably fairly stated as being the first working implementation, in that it’s the first one that could actually send transactions between machines and allow you to run programs on the Ethereum virtual machine.

Zach Latta: [12:03] Yeah, sure. So by way of background, I’m Zach, I’m the founder of Hack Club. And I grew up in Southern California, where both my parents were social workers. My mom worked in foster care, and my dad in homelessness, and I went to public schools, that like most schools in America still today didn’t offer any coding classes. And I was really lucky enough to be part of, I think, one of the first generations that really didn’t know a world without the Internet. And when I would get home from school, starting in like third grade, I would just – like, I could not pull myself away from the computer. It felt like “Oh, my God. This is where the secrets of the Universe lie.” And when I realized that you could learn how to code and not just consume stuff from the computer, but be one of the creators, that was the most exciting, interesting idea, and I’m like “Somehow I have to figure out how to be one of these wizards that knows how to do this.”

And I, got involved, I taught myself after school on the internet, and when I made it to high school, I felt so incredibly lonely, because it felt like the one thing I wanted to do with all my time, which was make things with code, was also the one thing I couldn’t do at the one place where I had to spend all my time, which was school. And I think generally – I kind of had felt like there’s this whole path that’s set up for young, ambitious people. First you do x, then you do y, then you do z. And I always felt like a bit of a misfit within that. And I ended up dropping out of high school after my freshman year. I had moved to San Francisco when I was 16 to become a programmer. I helped make one game that became the most popular game at the App Store at the time. It’s called Football Heroes, you can still download it. I was like a junior programmer on the team, and probably held us back more than I contributed… And that was like an incredibly meaningful chance to work on a real piece of software for the first time.

And then I helped build an app called Yo!, which was like Facebook Messenger but the only word you could send to people was the word yo. And the idea was like “What if we build an app that’s like so silly, so ridiculous that it can become viral just from that premise?”

[14:07] “Guys, something interesting just happened… So I downloaded Wajeed’s Bro app out of curiosity, and found it very sticky. I’ve never felt like I was anyone’s bro before. The only people who have used that term with me were assailants… But I started bro-ing people, and getting bro back. And all of a sudden, I’m bros with all kinds of people, including a guy from Branscomb Ventures.

Branscomb? That’s a solid shop.

So we bro-ed about this and that, and then when he heard I worked at Pied Piper, he got excited, he tripled-liked my bro, and he asked about meeting us.

Jared, what did you tell him?

I was waiting a bit to bro him back, so that I don’t seem overeager…

Bro him back, bro him… We’re not dead yet, guys.”

And that just absolutely blew up, and became the number one app on the App Store.

José Valim: [08:46] That’s a great question. To give some context - one of the things… Like, going way, way back, it always started because of the Erlang Virtual Machine; the only reason that Elixir as a programming language exists is because of the Erlang Virtual Machine… And the Erlang Virtual Machine was built by Ericsson, which is a telecommunication company for building concurrent, distributed and fault-tolerant software. I don’t want to expand on that; you can check Elixir on the website… But all of these things, my love for the Erlang Virtual Machine – so when I created Elixir, I was like “I want to have as many people as possible building on this platform, because I love it, and I think other people are really going to love it and enjoy it, too.

So I’ve created Elixir, and I’ve always thought – in terms of programming languages, I really think that Python is a really stellar example of tackling a bunch of different problems. I always had in mind that I want that for Elixir and for the Erlang Virtual Machine, for the Erlang ecosystem. I think we can grow diverse to solve all the different kinds of problems. So I come from a web background; I was a member of the Rails core team almost a life ago… When I started with Elixir, I had this obvious web background, and that was the first dimension that Elixir took off, with the Phoenix web framework… People started using Elixir more and more for the web. Elixir was already a good natural fit for building distributed systems, or anything regarding the network due to the Erlang heritage…. But it was like, “I’ve always wanted to try to expand this.”

The first time I expanded this was back in 2016 we released abstractions for data pipelines and data ingestion. So if you need to consume queues, and you need to do that very efficiently, we released libraries for that, and that brought Elixir to a new domain, which was like data processing, and there are some very nice use cases on our website… For example how change.org is using data abstractions that we wrote back then to process – because if you have a petition that one million people signed, you need to send them an update; now you have to send an email to a million people. How are you going to do that?

So we started that segment, and then the community started to grow, so people started bringing Elixir in the Erlang Virtual Machine for embedded. So there is the Nerves framework; people started bringing that to [unintelligible 00:11:12.25] streaming… And then there’s always the question, “Why not numerical computing? Why not machine learning?” So I always had this interest; I feel like it’s part of my responsibility, part of my job to try to broaden the domains and the areas of the language. The community is also doing that a lot for a bunch of areas… But you know, if there is something where I feel like “Hey, this is a good opportunity. We can do it”, then why not? Let’s do it.

And this always started - just to finish giving more context - when PragProg… I always had this interest. Actually, my thesis, my master thesis was in task classification. But that was 11 years ago… So we were not talking about deep learning at the time yet; I think everything was still support-vector machines were kind of state of the art. I never fell back, but I always had this interest.

[12:08] So in October last year, PragProg announced a book, which is “Genetic Algorithms in Elixir.” And then I was like, “Hey, apparently there is somebody who knows things about AI and machine learning in the Elixir community”, and he is Sean Moriarity. I sent him an email and I was like “Hey, I think the platform could be good for us to do everything in machine learning”, and he said “I agree. Let’s work on it.” And we started working on it. So it’s kind of like “Why not? If we can make it happen, let’s make it happen. Let’s build this, and then later we will continue working on how to package and how to sell this to people and say like “Hey, what are the benefits of having those two worlds joined together and working together?”

José Valim: [08:27] That’s a great question. To give some context - one of the things… Like, going way, way back, it always started because of the Erlang Virtual Machine; the only reason that Elixir as a programming language exists is because of the Erlang Virtual Machine… And the Erlang Virtual Machine was built by Ericsson, which is a telecommunication company for building concurrent, distributed and fault-tolerant software. I don’t want to expand on that; you can check Elixir on the website… But all of these things, my love for the Erlang Virtual Machine – so when I created Elixir, I was like “I want to have as many people as possible building on this platform, because I love it, and I think other people are really going to love it and enjoy it, too.

So I’ve created Elixir, and I’ve always thought – in terms of programming languages, I really think that Python is a really stellar example of tackling a bunch of different problems. I always had in mind that I want that for Elixir and for the Erlang Virtual Machine, for the Erlang ecosystem. I think we can grow diverse to solve all the different kinds of problems. So I come from a web background; I was a member of the Rails core team almost a life ago… When I started with Elixir, I had this obvious web background, and that was the first dimension that Elixir took off, with the Phoenix web framework… People started using Elixir more and more for the web. Elixir was already a good natural fit for building distributed systems, or anything regarding the network due to the Erlang heritage…. But it was like, “I’ve always wanted to try to expand this.”

The first time I expanded this was back in 2016 we released abstractions for data pipelines and data ingestion. So if you need to consume queues, and you need to do that very efficiently, we released libraries for that, and that brought Elixir to a new domain, which was like data processing, and there are some very nice use cases on our website… For example how change.org is using data abstractions that we wrote back then to process – because if you have a petition that one million people signed, you need to send them an update; now you have to send an email to a million people. How are you going to do that?

So we started that segment, and then the community started to grow, so people started bringing Elixir in the Erlang Virtual Machine for embedded. So there is the Nerves framework; people started bringing that to [unintelligible 00:10:53.25] streaming… And then there’s always the question, “Why not numerical computing? Why not machine learning?” So I always had this interest; I feel like it’s part of my responsibility, part of my job to try to broaden the domains and the areas of the language. The community is also doing that a lot for a bunch of areas… But you know, if there is something where I feel like “Hey, this is a good opportunity. We can do it”, then why not? Let’s do it.

And this always started - just to finish giving more context - when PragProg… I always had this interest. Actually, my thesis, my master thesis was in task classification. But that was 11 years ago… So we were not talking about deep learning at the time yet; I think everything was still support-vector machines were kind of state of the art. I never fell back, but I always had this interest.

[11:49] So in October last year, PragProg announced a book, which is “Genetic Algorithms in Elixir.” And then I was like, “Hey, apparently there is somebody who knows things about AI and machine learning in the Elixir community”, and he is Sean Moriarity. I sent him an email and I was like “Hey, I think the platform could be good for us to do everything in machine learning”, and he said “I agree. Let’s work on it.” And we started working on it. So it’s kind of like “Why not? If we can make it happen, let’s make it happen. Let’s build this, and then later we will continue working on how to package and how to sell this to people and say like “Hey, what are the benefits of having those two worlds joined together and working together?”

Emma Bostian: Yeah, thank you. I remember I woke up, it was like 6 AM, and I was like “Oh my god, Quincy messaged me.” I was so excited. So thank you for having me. I think it was an absolute blast, and I also think it was very successful, just to hear these diverse viewpoints from people all over the world.

So we kind of touched on some of the benefits and drawbacks of virtual conferences, but I would love to hear from you, Quincy, as a conference organizer, what were some of the benefits and drawbacks, in your opinion, of hosting a virtual conference?

Tim Triemstra: Yeah, I mean - every project is a little different. What we generally see is that if people completely own their own code - so they have all the source code, they can rebuild it, usually within days they are up and running. I talked with, like I said, some of the developers here recently. Pretty much everybody has told us that’s the easiest transition they’ve had to deal with. And so they’re just excited about doing the transition a few days or a week, and they move on.

We do understand that some people pull in third-party code, and if that third-party code is just in binary, they have this dependency tree where they wait for that to get done… We introduce new things called XCFrameworks, that makes it easier for people that do vend third-party code to vend it in multiple platforms, just like a universal app. It’s essentially a universal framework that you can ship. So it’s become a lot easier to do that.

And then there are a few edge case things where people were writing to very specific Intel instructions. There was some floating point math, or something like that… Very uncommon, but there are a couple of people that may have a little bit of assembly language that’s using that, and then that’ll do a little bit of work. But we’ve worked with big and small developers, and everybody is telling us that this is about the easiest transition they’ve ever had to deal with. And the fact of the matter is that I think you’ll be seeing a lot of software in the App Store that’s universal very quickly, which will attest to how good it is.

I think I saw Visual Studio Code announce that they are already building an Apple Silicon version on their nightlies. They’ve tweeted about that the other day. You saw the Adobe announcement; all the Omni apps will be in the store at launch. Really, it’s been an extremely good pickup.

Johnny Boursiquot: Okay. Well, I mean, not an extremely radical take… I can parse some sensibility in there. Alright, alright… Yeah, I have - perhaps not an unpopular opinion, but I recently… As we’re recording this, I think a few days ago there was a keynote that I watched. There was a GitHub Universe conference going on, and there was a keynote. And in the keynote, they were talking about Copilot and its ability to - like the whole Spark system, whatever; its ability to generate apps from scratch, and all that stuff… And that’s usually one of the selling points, the marketed stuff that says “Hey, you can build an app from scratch”, blah, blah. And it’s always like – the pitch is always to generate these new, one off, usually very small, very low-feature applications just to show that “Hey, look at this thing. You don’t even need a program anymore.” But these things are usually very simplistic applications. Not to say that some of them aren’t sophisticated, but when you’re generating things from scratch, like a greenfield, you have the benefit of basically writing everything from scratch. But for existing systems, for existing large systems, that have been around for years, decades even, those are the things that I need help with. Those are the things that when I’m tasked with maintaining a system that I’ve – say I was hired by a company and they’ve been running this software that runs their business for well over a decade, and here I am, tasked to make a change, or add a feature, or whatever… Those are the things that I need help with. And I’m not generating CRUD apps, or to do apps with an LLM from scratch. I’m trying to work into an existing codebase.

So one of the things that they showed was - I think it was just available for Java, but there was something that could help you upgrade your application. Maybe like you have an old school Java app, maybe you need to upgrade which virtual machine version it uses, or whatever the case may be… And it’s going through all the different files in the project, and basically changing the syntax… Maybe you used old school syntax; it uses new syntax to represent something. Maybe you want to use lambdas, or whatever it is. It’s going through and allowing you to change all these different files, basically upgrading an existing application. But I think even that doesn’t go far enough.

[01:00:05.19] And I think there’s always going to be a need for a human to provide some additional context, some backstory for an LLM that wants to change code… Because so much technical debt I’ve seen in my time, a lot of people who no longer work on an organization, they walk out the door with institutional knowledge in their heads… It’s not documented. You see a piece of code and you’re probably in a situation; you change something and you think you are doing the right thing, and then all of a sudden you broke something else somewhere else, not thinking that “Oh wow, I didn’t know why this code was there. It just seemed like a wrong thing, so I just “fixed it.” And then you end up breaking something else you didn’t know was relying on the code to work a particular way. So this technical debt stacks up over time, and it’s not like people are keeping track of the reasons why they can’t change, or shouldn’t change something.

So when you throw an LLM at these kinds of situations – not to say that one day we won’t be able to have models powerful enough to fix those things, but these are the kind of… Like, when they’re doing those marketing presentations and getting everybody excited about large language models and gen AI, those are the kind of demos I want to see. Take an existing app, with a lot of assumptions baked in, and refactor that. Because that’s my job. I’m not generating a 3D chess program for work. That’s not what I’m doing. That’s not how I earn my living. I work on old, legacy systems, with a lot of assumptions, a lot of decisions that were made prior to me getting in front of this codebase. Help me deal with that mess, not the trivial stuff.

So perhaps not an unpopular opinion, but more like a, I guess, a request. Well, I don’t even know what it is. It’s more of a – perhaps letting everybody know that yes, the demos look cool, but we’re still some ways away from truly being able to leverage LLMs to do the things that we really, really care about… Unless you’re building something trivial and greenfield, which is not most of us.

Dennis E. Taylor: Well, I think one of the reasons that the VR works in the series and isn’t a problem - it’s a situation similar to multiverses. There’ve been a lot of movies where people get stuck in a VR, and they have to fight their way out, or something like that… And you always have this problem of trying to justify real stakes. “Oh, if you die in the VR, you die in real life.” Uhh, you know…

So when you have a movie or a story where the VR is the main combat arena, if you will, and you’re trying to give people a reason to care about, or the reader a reason to care about what happens in the VR, that can be difficult, because it is by definition virtual. So in the Bobiverse books, the VR is a supporting structure, but it’s not plot-determinant. It doesn’t save the day. It doesn’t cause life or death situations. It’s a way for Bobs to communicate, and all the real action, all the real stakes-based stuff happens in the real universe.

Sherol Chen: “Now let’s take this down the practical route…” So we have machine learning, we have AI - how does that impact the world? How does it affect the everyday person? A lot of people in the U.S. use smartphones, so now if they’re traveling abroad, they could take a picture of a menu and get that translated on the spot. Or they don’t even have to take a picture; within the computer screen, they can translate from Spanish to English… And a lot of this image recognition or language understanding comes from advanced [unintelligible 00:25:10.22] machine learning.

So on the user side there’s already a lot of people using it, and what they’ve found is that a lot of these tools are useful if you want to be able to use voice commands. You don’t need to necessarily retrain – I don’t even know how much data they used, but all the data that they require to train that voice model, that voice speech-to-text model, they don’t need to necessarily have to go through… Like, every company that needs that doesn’t need to go through and do that again. So Google having worked on this technology and using it in its own products makes this available for people.

So you can come in, you have an app that you wanna build, you want it to be voice-activated, or you wanna be able to take images and figure out if they are a kid-friendly image… There are these models that Google has pre-trained, and Google is ensuring that these models work well. So without any data, with just the desire to have this feature, you can already plug and play certain products that Google has in machine learning. Anything that requires vision, or audio processing, or natural language - I think these are very common use cases.

And then on the other end of things, like you were mentioning, you could set up your own Python environment, you could be coding in TensorFlow, you could train your own model, you could feature-engineer on your own, you could hyperparameter on your own, you can look at the models and the precision and recall, and assess how good these models are doing, and then you could do it all over again. And maybe you have something running in production - you can have all the control over every aspect of that. So that would be the other end.

Of course, enterprise cloud technologies - they provide the environments. You have Docker containers, you have virtual machines that can also satisfy that need. What AutoML does is it focused on the middle piece, or at least trying to figure out how you go from one end to the other. So in one end you have pre-trained APIs, where you have basically no control over how that model is making decisions, and then you have build-it-from-scratch, do it all yourself. For a lot of businesses, it’s kind of hard to go sometimes from research directly to production, and so trying to find that middle point is what people working in AutoML are trying to figure out. Things like hyperparameter tuning; tuning these weights around to kind of get a feel. This is a highly intuitive thing, where you get a sense of whether this number should be bigger or lower. AutoML, one of the aspects is trying to figure out how to just simply do that.

[28:02] With the tools that we have now, it could just be a matter of thinking – you might be able to figure out safe search, or kid-friendly images; it’s something pretty universal for the most part, but maybe not for every culture. You might wanna label your own data, as like “This image isn’t safe.” Maybe it’s not enough to know which pictures have cats in them. You can go to your Google Photos right now and do that. You can go into your Google Photos, type in “cat pictures”, get all the cat pictures. But what if you wanna know what type of cat, or what if you wanna know different colors of cats? The specificity that these pre-trained, more general use cases may not provide.

So AutoML, in the most general use case, is basically giving you the opportunity to label your own data, and then feed that in. So you’re getting customization without having to build everything out from scratch.

Kevin Ball: Interesting. So if we kind of explore that direction then - it sounds like what you’re saying is we’re calling it WebAssembly, but really the web is like Basecamp. Everest is some place that is essentially a universal virtual machine.

Nathan Sobo: Yeah. Yeah, we execute – and we’ve experimented with a couple of different approaches, but what we’ve landed on now is we actually have the agent run in your shell. The assumption being that whatever shell you’re using, you probably configure it with your Python virtual environment… I don’t even pretend to understand that world [unintelligible 01:15:16.12] They need to understand it better than I do. People on our team have improved it a lot in Zed, but there’s a lot of language-specific, situation-specific stuff in your shell. So we just tell the LLM, “Here’s the user’s shell.” We set it up to run in that setup, so that whatever you’ve configured is there for it. And then inside that shell, it can run commands. It can do things. So in some sense, you can do whatever. Then there’s some dedicated tools, but they’re not that many, honestly. And we’ll be adding to it over time, I think… But just the ability to do glob matching on paths… Grep… And we’ve done some work with Tree-sitter around Grep to make sure that when we find a match, we’re giving the LLM a coherent piece of text, that stops at reasonable syntactic boundaries etc. And then editing. But then of course, you can bring model context servers… And so as part of this launch, we’ve put a lot of work into just making that easy.

[01:16:26.07] So there’s a couple different ways. The most universal way is just in Zed settings, you can configure a model context server, say where it’s listening, and it’ll work. For a curated set that we hope to grow, and the community can help us grow, there’s also in the Zed extension store just a section for model context servers, basically. Context servers, that speak the model context protocol. And we’re not another registry. I guess we are. We’re kind of a meta registry, in the sense of “This is how you install this model context server and connect to it from Zed”, basically. It’s just like a little recipe, basically, that pulls it from wherever… And then we put some work into making them easy to configure etc. Like, okay, you’ve installed it. Does it need auth? Does it need some other environment variables to be set for it, or whatever? Streamlining that. So that’s a way to kind of build your own tools, or bring in… There’s like a tool that will pull in Postgres schemas, and there’s all –

Natalie Pistunovich: [43:56] As a person who’s not native in English - as you might know this by my accent - a teaching assistant is a term that I first learned in university. This was somebody who was giving classes about whatever the professor was teaching; so the teaching assistant was teaching the hands-on or even kind of workshop equivalent to what the professor was theoretically teaching. But then when I started teaching workshops, I learned the teaching assistant in this context means something pretty different; this is somebody who has some technical knowledge, like “My Docker will not run. I don’t know what this error means.” But it’s not necessarily somebody who’s as experienced as you are in the content. It’s not somebody who’s able to answer all the questions; maybe some of them yes, but many times it could be a person who has more housekeeping duties, let’s say, than I in the beginning expected from a teaching assistant to have… In the sense of this is a person who would tell you “Hey, this is time to pause. Many people are asking you questions. You are on mute.” So this level also.

And it’s just worth pointing out I guess that “teaching assistant” is such a context-rich word, for me at least, and definitely valuable, because - we talked about this, but reading the room when you’re in virtual versus in-person is very different, and you definitely need a second pair of virtual eyes in your virtual crowd, and that’s like an extra duty for the teaching assistant… Which is also different for a teaching assistant from in-person versus to a virtual workshop.

Natalie Pistunovich: Yes, definitely. I try to bring as much to the table as I can from different experience… Not to bore the people who have heard the previous episode where I was, but I will say in short that I spent a year living in Kenya, and when I went back to Germany, I maintained as many connections as I can, and I use this for things like asking all the local user groups to always participate in meetups and in conferences, and submit to talks, and so on. And now that everything is virtual, this makes this very easy.

This is, of course, becoming a stretch, because it’s been a couple of years ago, and the last time I was there it was with Bill. We gave a workshop with Bill Kennedy; we gave a Go workshop at an amazing institute that is called Tunapanda, which is a one-year training, and then people can get jobs as developers… Not just in Go, but also in Go.

And all this keeping in touch with communities that are very much outside of my network is a hard work, but it’s also very much worth it. And of course, there’s the easier (if you will) diversity work of also making sure to work with the local community, for example, being inclusive more towards women, which is also not a big part of the Berlin tech scene… Working with refugees as much as possible, talking to organizations and asking them to invite the people who [unintelligible 00:40:16.17] school kids, or if it’s people in university age, and so on - so invite the relevant groups to all sorts of events that we’re organizing… Yes, diversity is a hard work, and it’s definitely paying off.

Dustin Kirkland: Yeah, so that’s exactly where I was going with this, it’s that a lot has changed in the last 12 years. Canonical, we found this huge focus and success on Ubuntu as a server platform. We haven’t abandoned the desktop, our desktop is now better than ever and we are bringing it to mobile devices. But the cloud did not exist - in that name anyway - in 2004 and Ubuntu server didn’t either. Microsoft was led by Steve Ballmer at the time, and the thought of Microsoft doing with anything open source was just unthinkable. He called Linux a cancer and had essentially banned any Microsoft team from working with open source. But there has been a regime change at Microsoft. Satya has lead a real revolution at Microsoft in the adoption of open source. Over the last several years Canonical and Microsoft have worked together, most notably on Azure, Azure being Microsoft’s cloud. It’s sort of in the big three clouds of Amazon’s AWS, Google Compute Engine and Microsoft Azure, and Microsoft through Azure has been a tremendous partner of Canonical’s, and we have helped deliver the best Linux experience inside of Microsoft Azure. The vast majority of instances running in Azure that are not Windows instances, like 80% of the Linux instances are Ubuntu. We’ve worked very successfully with the Azure team in doing so, and that really fostered an interesting set of business relations, as well as - and this is extremely important - technical developer relations with Microsoft. [24:08]

Now over the last couple of years I have gotten to know a couple of developers at Microsoft, especially the Windows kernel team and the Azure team, and these guys are a lot like me. They went to college 5, 10, 15 years ago and universally, their college experience was much like yours, running a Linux desktop or certainly running a Linux on servers. The newer kids are running it on Raspberry Pis or in a data center on campus. That was absolutely my experience, I was at Texas A&M from 1997 to 2001. That was when I really went all in on Linux. I used it across the board, stood up web servers on Linux, used it on a desktop, and then I sort of made the move wholesale, in about 2000. Microsoft has totally taken note of that, that the developer environment, while it certainly used to be very closely tied to Windows desktop, has moved away from it. And to neither Canonical nor Microsoft’s benefit, a lot of that desktop environment has actually moved towards Macs. I think you can walk into a coffee shop almost anywhere, certainly in Austin or San Francisco where I spend most of my time, and in that coffee shop there is a developer with a Mac, writing some code, and that code is almost always inside of a virtual machine, in a Vagrant or a virtual box - a shell essentially, where they are running Ubuntu. And Microsoft actually commissioned a study to sort of understand what kids in college are using as their developer environment. One thread that was constant through that entire study was that regardless of whether someone was running Windows or mac or Ubuntu on their desktop, they were almost certainly doing their development in a Linux shell, and the vast majority of those Linux shells were Ubuntu Linux shells. It’s something that I think we have all seen, and whether or not Forbes or Wall street Journal or anyone has taken note of that, you could talk to almost any developer in the Valley or anywhere across the world in fact, and I don’t think any of them would be very surprised by that story. You know, most web development being done in a Linux shell isn’t exactly a headline that - you guys are in the media business, that’s probably not gonna grab a whole lot of clicks, right?

Daniel Whitenack: Now, they also bring in the Omniverse to this. So I don’t know if you saw that reference, but it’s like – so obviously, there’s the Metaverse, which I assume a lot of people know what that is out there, but it’s like the virtual reality world that Facebook or Meta is designing and running. I don’t know, have you been in the Metaverse, Chris?

Kyle Daigle: I believe it’s currently in a single flow. In all her blog posts, Jess and some other folks on the Actions team are also looking at ways that we can make that be a bit more flexible, like you mentioned, so you’re not sort of doing everything on a single CPU… So looking at what our options are to allow for the parallel tracks that have their own VMs, maybe each action should have its own, so that way it has its own compute, depending on what you’re doing… But yeah, you should be able to run the parallel tracks semi-independently, depending on exactly how the dependency tree maps. That’s one of the very interesting things in this - if you watched the GitHub Universe stuff, it gets a little tricky to describe, but ultimately if you’re branching off and both of those things need to work for this last thing to ultimately run, there’s this dependency tree that we have to build in order to make it run, and ultimately that tree is what’s telling us sort of where things are able to execute in these VMs.

Man, I’ve forgotten how difficult it is to describe visual things… I have my hands here, which aren’t helping anyone but me at the moment.

Karl Fogel: A few things. One, the adjustments that I had been doing in the years from 2006 roughly to 2013, they weren’t that trivial. I mean, there were a lot of small scale changes that went in. I think most sections of the book got touched, some of them pretty heavily, but I was never thinking of it as a full rewrite. And then it was really partly my own feeling about certain things that were out of that and partly feedback I was getting from other people. One thing everyone noticed - and I noticed too, because I also use Git for my coding work, although I use Subversion for non-coding version control - was that all the examples used Subversion, which was totally the right thing to do in 2005, because that was the thing that you stored your open source code in, but it just wasn’t by 2013; Git was the obvious answer, and frankly even though the site itself is not open source, GitHub was clearly the thing to use. For example, most active open source code is just on GitHub, and if the book doesn’t acknowledge that fact, then it’s just not reflecting reality and it’s not giving people the smoothest entry into open source that they can have.

So one obvious thing was the revamping of all the examples to use Git instead of Subversion and to talk about GitHub. And also in general, the project hosting situation had changed. I’m sorry, I just don’t consider SourceForge a thing anymore. [laughter] So many ads, too much visual noise, not compelling enough functionality, and that’s despite the fact that the SourceForge platform itself finally went open source, as the Allure project - which is great, I’d love to be using it, but I’m afraid I just have a much better experience with GitHub and Git, so that’s what I use.

So the recommendations about how to host projects really needed to change to be oriented more around the Git-based universe and to at least acknowledge and recommend GitHub, while acknowledging that it itself is not open source… Although I hope that they see a grand strategic vision whereby opening up their actual platform makes sense some day; I think that the real secret sauce there is the DevOps, it’s not the code, so I hope they do that someday.

[19:53] The other thing that changed kind of in a big way was what I think of as the slow rise of business-to-business open source, which is… The old cliché was “Open source always starts when some individual programmer needs to scratch an itch”; she needs to analyze her log files better, so she writes a log analyzer and then she realizes that there are other sysadmins who need to do the same, so they start collaborating, and now you’ve got an open source log analyzer, and she’s the de facto leader of this new open source project. Well, that did happen a lot, but now you have things like Android. You have businesses putting out open source codebases like TensorFlow. I don’t mean to pick on Google examples only, it’s just that those are the first things that come to mind, but Facebook also does this, Hewlett Packard does it… Lots of companies are releasing open source projects which are - I guess you could call them it’s a corporation scratching a corporation’s itch, but it is not a case of an individual developer; it’s a management move, it is done for strategic reasons which they can articulate to themselves and sometimes they also articulate to the world.

And I thought that the rise of that kind kind of project needed to be covered better, and that that was a trend that if the book could explain it better to other managers in tech or tech-related companies, that perhaps it would encourage some of them to join that trend.

And sorry, I’m realizing that there’s one more component to the answer - the other thing that changed was that I expected governments to be doing more open source by 2013 than they were, and I had at that point been very active in trying to help some government agencies launch technical products as open source, because they were gonna need that technology anyway. It’s taxpayer-funded, why not make it open source? And they were just really culturally not suited to it. There were just many, many things about the way governments do technology development, in the way they do procurement, in the way they tend to be risk-averse to the exclusion of many other considerations, really made open source an uphill struggle for them, and I wanted the book to talk a lot more about that, because I wanted it to be something that government decision-makers could circulate and use as a way to reassure themselves that they could do open source and that it could be successful, and that they didn’t have to look at it as a risky move.

So there were some new trends that I wanted to cover and there were some new goals that I had for the book, and they just required ground-up reorganization and revamp.

Emma Bostian: Yeah. So if y’all wanna learn more about Vue, we’ve got a couple of episodes; we’ll link them down. But we wanna focus on some other best and worst moments in web dev… But not just in web dev; I think as an industry. Or as a human population, I guess. Because for me, I actually didn’t do any conferences this year. I did one before Covid got bad and there was quarantine. It was in February. That was in-person. I didn’t do any virtual conferences, but they were rampant… Which is good and bad.

I think that they are – I’d like to see the good in it, because it makes them accessible to more people, and it allows us to have more diverse speakers when we don’t have to factor in visas, and travel, and things of that nature… But how do you all feel about virtual conferences? Did you do any this year?

Beau Perschall: No, this is very practical. As a matter of fact, we have big and small customers that are using it, over 200,000 downloads for Omniverse as a platform that you can get from the NVIDIA site. You’ve got companies like BMW that are using it to plan their factory of the future, and part of that is worker safety, so they have to have legs. You can’t simulate the ergonomics of if you’re doing a repetitive task, are you going to hurt somebody by doing it? Or are they in danger of getting hit by something in a work cell, or something on the assembly line? So there’s all sorts of simulation around that kind of information as part of Omniverse.

[00:04:17.28] But it’s a really broad platform. It’s designed to be extendable, so that customers can come in and write their own tools and connectors. It’s not supposed to be just its own endpoint. In other words, we have connectors, which are basically bridges to other applications, whether you’re coming from the manufacturing side, like Siemens, or you’re coming from architectural software, like Revit, or you’re coming from animation software, like Blender, or Houdini, or Maya, or Unreal, for that matter. All of that data can be aggregated through USD, Universal Scene Description. That’s the file format that Omniverse is based upon, which was a Pixar open file format. It is very robust. And basically, we figure, we’re the kind of the connective glue between all of these platforms, so that simulations can be run inside of Omniverse, but all the data can move in and out. It’s not like captive data. Hopefully, that gives you a little bit of a background of Omniverse in and of itself. It is a visual platform.

Caitlyn Ralph: Yeah, definitely. So I started, like Russell said earlier, by making articles at The Pudding, but the reason I was the first person on our very small team to have a specialized role focused at Polygraph… And the reason I kind of gravitated in that direction - I had my personal career goals, that I liked management and things like that, and I wanted to go in that direction… But what I’ve found exciting about the visual and data storytelling work we do at Polygraph is that – I know you said at the beginning we’re not gonna pit The Pudding and Polygraph against each other, and now it sounds like I’m doing that… But what I find exciting about Polygraph is that right now it’s set up that we can actually go outside of the page and do this kind of work… So all of The Pudding articles right now live on The Pudding’s website.

What we do at Polygraph lives on all of the different brands’ and organizations’ we worked with sites, but we actually do installation work, we’ve done work with the Smithsonian National Portrait Gallery, and stuff like that… So I think the first part of the answer to that question, what I find really exciting is that there’s a lot more variety in what we’re doing… And Amelia, you remember a lot of these conversations; because there’s a lot more variety, we find ourselves straddling very interesting lines in between being more editorial with our client work, or being more agency, flashy, pulling in some crazy JavaScript work to create very immersive experiences with data, things that we’ve done with some of the installation work, and stuff like that… Or to really replicate our editorial nature for clients.

[27:59] What I’ve decided to do strategy-wise this year is dabble in both, and show that in our portfolio we’ve done both. Again, we do see ourselves as data journalists, so that is our expertise, that storytelling… But we can’t do a 40-foot installation for Universal Music Group’s lobby offices of their artists and musician data, if that’s something someone wants us to do.

And I think it also ends up serving as a creative outlet for people at the pudding to do different things as well, but that’s a whole other conversation thread as well, with how we set up our company. So go ahead, Russell; I see your face.

Daniel Whitenack: That’s what I was just thinking… So I can ride regular rollercoasters all day, but I remember we went to Universal Studios recently, and we were in one of those 3D rides, where it also shakes you around, and stuff… Man, I got so sick. Something about that virtual motion just really does not agree with me.