Stefano Maffulli: [00:33:23.02] Luckily, we were already into this two-year process of defining open source AI. Actually, I was already in conversations with Meta to have them join the process and support the process to find the sheer definition of open source AI. And in fact, they’re part of this conversation that I’m having with not just corporations like Google, Microsoft, GitHub, Amazon, etc. but also, we’ve invited researchers in Academia, creators of AI, experts of ethics and philosophy, organizations that deal with open in general, but knowledge open data like Wikimedia, Creative Commons, Open Knowledge Foundation, Mozilla Foundation… And we’re talking also with experts in ethics, but also organizations like digital rights groups, like the EFF, and other organizations around the world who help into this debate. Like, we had to first go through an exercise to understand and come to a shared agreement that AI is a different thing than software. Then we went through an exercise to find the shared values that we want to have represented, and why we want to have the same sort of advantages that we have for software also ported over to the AI system.

And then we have identified the freedoms that we want to have exercised, and now we’re at the point where we are trying to name the list of components of AI systems, which is not as simple as binary code, compiler, compiler and source code… So it’s not as simple as that. It’s a lot more complicated. So we’re building this list of components for specific systems. And the idea is by the end of the end of spring, early summer, to have the equivalent of what we have now as a checklist for legal documents, for software, and have the equivalent for AI systems and their components, so that we will know… Basically, we have a release candidate for an open source AI definition.

Zac Smith: Well, I would say, I’ve always been a small company person. I’ve had the pleasure of building a couple of startups, and being part of, I think, the biggest company I was ever at, for any substantive period of time, was no more than 100 people. Packet at its heyday was about 120. So I was really used to small organizations that didn’t really have advantages in terms of global scale, or amount of money, or brand, or anything else like that, but you had agility. So you kind of organized around speed, organized around ability to change. And that’s what I love about startups, because I love to experiment. I love to lean in with new ideas, and I love to go fast. I’m a person of action. And when we became part of Equinix, I would say after about one year where we had a little bit of a grace period, we were off to the side, and kind of got our bearings… You know, this is a 15,000 person company that had a very different kind of culture. They were a market leader, they had decades of success doing their business, they had a lot of kind of experience and leadership there… But that comes with bureaucracy. It comes with more people, more checks and balances. They had a matrix structure, which I’d never heard the word of, but basically, it took me a while to figure out “Well, who makes a decision here?” And the answer was “Well, everybody has to get aligned, and then decisions can be made”, or something.

So for me, that was super-different. And I will admit, at first it was very, very frustrating, and I had to take a step back and really learn some new skills… Which once I took kind of a learner’s mindset or a growth mindset, I really enjoyed – I’m not going to say I enjoyed all the experiences of cross-functional work, because it’s a little bit different than my personality. But I enjoyed the learning of how to figure that out, because you basically had to think “Hey, how do you get 15,000 people at a company to march in a direction together?” Or to do things that might feel uncomfortable, or to try new stuff. And the answer is not that much different than marketing that you would have to do to go convince customers to try your thing. So a lot of those skills I started to use internally, and figure out how to explain, and communicate in different channels, and in different ways… And that was super-powerful. I think those are skills that I’ll take away with me in other things that I do.

And then you also get a little bit of a different lens on the world. Running the kind of the digital side at Equinix, I was more of the technology-focused person at a big, global real estate company. And so I had a first-person view working with our partners like NVIDIA, with Intel, with Dell, with VMware… Those were all very different conversations than I would have had as a startup, because we had something that they really wanted and needed. It was very different than the conversations that a startup has, with those kinds of big organizations. So that was super-cool, because we did some awesome stuff and we were able to unlock things that would have never been possible, I think, for a smaller organization.

But then, after three years, which was, I would say, a very measured process that we had with Equinix, they wanted to acquire the business to help make their platform, which is hundreds of data centers all around the world, more digital. And so the first two or three years were about kind of working through at the pace to integrate some of those capabilities from the packet acquisition into that broader company. But we always knew that there would be kind of an integration process, and then, for me, after three years, it was the right time to go and, I would say, use my skills elsewhere… Primarily because that kind of integrate and scale function as a leader is not actually my best quality. I’m more of a driver and creator, rather than a kind of a manager, a process person.

[00:20:26.24] So for me, that was a good transition. I think it was right for the company as well. I was a little bit more mentally prepared than my first startup - in 2012 I sold, with Raj. That was really emotional, because he was like your baby… And you’re like “Don’t touch my baby.” But I was much more mentally prepared going into the transaction with Equinix around “This is your company now” and “How can we make it work for you, and how can it last well beyond me?”, which is the point. And so it was easier to step away when the time was right. I still had a lot of relationships and touch with the customers, that I obviously miss, but it was the right time, and it was a good, healthy transition, I think, for all parties.

Kris Brandow: Yeah… It’s sort of painful, I will say. I see that “Export to a Premiere” button or menu item in Audition, and I’m like “I should just be doing this…” But I’m on a Mac, and there’s a lot of really deep integration that Final Cut has with the hardware that I wanted to take advantage of if I could. And unfortunately, Apple does not have a similar level DAW, digital audio workstation, for podcast editing. Like, they have Logic, but that’s definitely more of a music-based DAW, that you can sort of squint at and turn it into like a spoken speech kind of thing. But it’s like, Audition is definitely the gold standard for this type of audio content. And so it’s just like “You know what? I’m just I’m just going to use Audition, and I’ll just deal with this weird flipping between Final Cut and Audition.” I think I’ve gotten it down after editing a few episodes now. It was kind of a mass jumping back and forth between the two… But I think I’ve figured it out now. Gotten it down.

Amal Hussein: Changes, yes. both with the App Store… They’re like “Oh, fine. Sure, people can create their own third-party digital marketplaces, but you have to basically still go through us, you still have to give us some money… If your app gets downloaded over a million times, we’re going to come after you for more money…” Basically, everything has to still get notarized by them. It’s basically just lip service. They’re like “Okay, sure, we’ll comply, but here’s how we will comply”, in the most douchey way… And then on the flipside of that with browsers - so they’ve basically now been forced in the EU to allow for browser choice. So hopefully now we’ll have other browser engines starting with iOS 17.4, which I think is set to release in March… However, the big kind of hoopla this week is that two weeks ago they released a beta of 17.4, and basically that broke existing PWA support, where you had an app on the homescreen, it would launch in like full-screen mode - now that same app is basically just like a bookmark. It’s like just a regular browser app. It’s not like a – it doesn’t take advantage of all the other rich APIs. And they came out with an excuse today, where they said that “Oh, yeah, that wasn’t a bug. We intentionally broke that because of some–” They were handwavy about security, and a bunch of other things that really just don’t make sense. They’re “Yeah, we have to kind of protect users, and if we lift that homescreen thing in, users will be able to see each other’s data, and this and that…” I’m like “Oh, really, Apple? You multibillion dollar company, you really couldn’t find a way to design this in a secure way?” You’re basically just intentionally breaking the PWA experience, because you now are forced to sort of not only support it, but now there’s going to be other browsers that are also going to enable this experience for users.

So long story short, they’re just being really problematic, and they’ve been silent, they’ve haven’t really – there’s been journalists and all these people trying to get them to comment, explain, and all they came out with today is these few paragraphs on the developer.apple.com, under a section called titled “Why don’t users in the EU have access to homescreen web apps?” So yeah, Apple’s really – I mean, this is just like a fight for… This is a fight where it’s about money, clearly. They’re not gonna just give up their 30% without a fight. But I just didn’t think it would be this dirty of a fight. And I just didn’t think that they would consider – do they think people are stupid? I just really don’t – I don’t get that. So yeah, so anyways… I’m done talking now, but yeah…

Nader Dabit: Yeah, we have a way to actually get around that by programming our own digital scarcity. All you have to have is an internet browser. You can now receive and send payments between parties.

The other thing is state. So to save information - how do we do that today? Well, we have a centralized database somewhere, like AWS, that if we don’t pay the bill, goes away forever and we lose all of our information. Maybe the data leaks, maybe it gets hacked, or maybe we just – I don’t know, there’s all types of stuff… Maybe it goes down because AWS has an outage in U.S. East 1. But there are a lot of things that are not ideal, and I think that with blockchain technology we’ve unlocked native payments and native state… And early on, with Bitcoin, it kind of jsut showed that these types of decentralized protocols were possible, but you couldn’t really write a program in Bitcoin. And I think the reason that you’re seeing it start to become a thing now is that the evolution of that technology and the improvement - you’ve had increased transactions per second, so higher throughput per second, reduced costs…

[40:07] For instance, Solana is a good example of a blockchain that has fractions of a penny for a transaction… So you have all of these things that are kind of happening, that are now starting to become mature enough for people to build on, that people are taking advantage of them… And the things that they’re unleashing are also, I would say, really interesting to people that are developers already, but they also enable people to make money. In the United States, where we can make like a couple hundred grand a year for a senior developer, we might not be that interested in this stuff… But let’s say you’re making 10k as an intermediate or senior developer somewhere else in the world, and you can partake in a protocol and increase your wage by 10x because all these projects are international and they’re completely decentralized, and they’ll hire anyone, anywhere, it doesn’t matter where you live… It’s just very, very interesting to a lot of people.

Harry Stevens: Yeah, it’s crazy. It’s kind of like catching lightning in a bottle a little bit. I definitely couldn’t have anticipate that it would do so well. If I published it a week later, maybe it wouldn’t have. It was just kind of like the right time for it. When I published it, you were seeing videos of college kids in Florida on spring break saying “I’m gonna party no matter what.” We’re kind of moved past that now, I think, for the most part. So I think it was what people needed at the time.

But also, in general, I would say that the news business across the board is going through a crisis right now, because our sources of revenue have really dried up. Online advertising is not nearly as lucrative as it was to advertise in print… So newspapers across the country, across the world are cutting staff. And it’s kind of sad, because – so people who do what I do, a lot of times, are seen as like a frivolous extra expenditure. And in a way, I can relate to that view, because you know, if you don’t have somebody who’s covering city hall, or you can’t afford someone to cover the local sports team, you’re not gonna get a graphics reporter, so I totally get that. And a lot of newsrooms are going through really hard decisions right now.

But on the other hand, it is possible for a newsroom to pursue this kind of storytelling, where you’re leveraging the interactivity of the web, and you’re trying to take advantage of new ways of telling stories that can be more engaging to readers, I think it’s a good idea to pursue that.

[44:19] I’m not a business person, I don’t look at profit and loss sheets, and so it’s probably easier said than done, but I do know that the places that have early on embraced interactive storytelling and digital storytelling have been more able to weather the economic storm that has befallen the journalism industry in general. You look at the New York Times - they were one of the really early adopters, and a lot of their most successful and popular pieces have been interactive graphics; it’s the same thing with the Washington Post and a number of other places.

So it’s not just my story… Like I said, I kind of caught lightning in a bottle, and this one happened to do really well, but I think a week after my story was published, different people on the graphics team published a story that was like a calculator where the 2.2 trillion dollar stimulus bill was coming out… So you could go to this calculator and tell it a few things about yourself and it would tell you how much money you could likely claim from the stimulus package. And that’s just news you can use. Everybody is wondering this, it’s a really simple design, and it tells you important information…

That one also just got a ton of traffic, which is great for our readers, because obviously they’ve found it useful, and it’s also great for the company, because it’s subscriber-based and advertiser-based, so you wanna have more of a readership. So I think that these sorts of interactive stories can be really, really successful.

Eric Berry: [55:51] I think mentioning the ones on the post – I wrote this when I was very much deep in reminiscing about the past year… I didn’t even realize it’s been a year, and all of a sudden somebody on LinkedIn said “Hey, congrats on one year!” “Oh my gosh, it’s been a year…” So to touch on these, I think these are the most important to me.

When I started the business off, and over the past year, I think one of the things that I’m most proud of is the honesty that I’ve been able to approach us. Digital advertising is kind of a messy, dirty, yucky thing. It can be. And advertisers that I’ve talked to over the past, where I basically open up my data and I say “Here, you could see exactly what we’re doing. This is my overall click-through rate. Here are all of my publishers that I’m working with. Here’s how much money we’ve made as a company this month, and this is how much money we’re giving back to developers”, and all this stuff. I’ve been so open and honest with them, and I think that’s paid off a lot.

I don’t do it to pay it off, but I’m not a good poker player, I’m a developer… Logically, in my mind, if I’m open and honest with them, then they will appreciate it and work with me, or they will say “Okay, thank you for being open and honest. You’ve given me enough information to know that this won’t work for me, but I’ll think of you in the future.” Never has anybody said “Oh, they’re so open and honest… I’m gonna take advantage of this and see if I can…”, whatever it might be.

That’s something that I’ve just been really proud of, I continue to be proud of. In all openness, I posted what our earnings was for last month, July - we had 2.4 million impressions, average click-through rate of 0.33%, we had $7,200 come in the door, and $4,400 went right back out to developers. So we made $2,800 in profit. Again, we’re far, far from being able to support myself and the other developer that works with me, but we’re in the right direction, and we’re making a difference.

So far, in the whole, Code Fund has generated over $36,000 for open source. I have literally been able to pay out $36,000 over the last year, to developers all over the world… And that is just such an amazing feeling for me. And the best part is I know I’m just at the very beginning. It’s gonna grow, and it’s gonna grow, and it’s gonna grow.

I think part of the reason why I was able to pivot and not topple is because of the mission. I look at some companies’ missions, and their mission is to make revenue, or “I want everybody to use this brand of soap”, or whatever. That stuff to me is not very important. But having a mission that actually has true meaning, such as “How can we help fund open source? How can we help get money to people?” and as long as that is the rod that I hold on to, that is the goal post that I’m aiming at, it’s really easy to make decisions… And every decision that I’ve had to make is like “Well, how is this gonna help open source? If it is - okay, great. If not, I’m definitely gonna do it.”

Julian Shapiro: Libscore came about from two things. One was after marketing NameLayer I realized how important it was to actually assess how effective your marketing is. It seems self-evident, but when it comes to open source, there was actually no way to do that. Even if you are not marketing your open source project, let’s say you just get it out there on GitHub - you have no idea who’s using it. If you scroll through the list of everyone who’s starred it, maybe you can find in their bios they’re the employee of Yahoo!, but that’s not a representative metric, that’s not enough. A lot of people don’t star, a lot of people use npm, so in open source in general, I just had no way of figuring out how effective was my marketing. All I could track were page views on VelocityJS.org. So I knew I needed to close that feedback loop that any growth-minded entrepreneur would wanna close. That was the first thing. That was like the seed of the idea.

At some point, I heard that Stripe was having this open source grant program which hadn’t been done before. It was a for-profit company saying, “We wanna sponsor interesting open source developers, working on projects that we think will have an impact, and maybe tangentially related to the work Stripe does.” I got the grant, and they took me out to San Francisco.

On that grant I realized I’m actually mostly done building Velocity. It was at the point where it worked well. It wasn’t encountering much bugs, and so I thought, “Well, maybe I could actually use this as an opportunity to pitch Stripe on a brand new project, something else.” And I told them the story about how I marketed Velocity and how I wanted to track its usage across the web, but doing so seemed very intimidating. It didn’t seem like something I could do in my room.

[40:09] I felt like I needed to be with a bunch of very technical people; I needed the processing power, I needed some expertise in scraping, so I took advantage of all these people around me at Stripe to put together a small team and a little bit of sponsorship to build Libscore, with the intent of scraping the top million most trafficked sites on the web, to determine which JavaScript libraries those sites were using.

It turned out that part was actually very easy. At that point in time module loaders weren’t very heavily used to the point where they were obscuring the footprint in the browser of which libraries were in use, so I could just sniff all the top-level variables and run a bunch of heuristics to figure out, “Is this variable random-leaked stuff, from code that was messily written? Or is this the exposé endpoint of an intentionally installed third party JavaScript library?” So I would try to figure out how to do that, and then I would extract that list from the web page, and then index that across a bunch of things, put that into this big database and then make that searchable.

That was the intent, but all I knew how to do myself was that scraping portion to figure out which libraries were in use on the page. So I got together with this incredible designer named Jesse Chase. He was the creative director of Digital Ocean at the time, who built the whole UI and the graphing system. And then Jason Chen, founder of another open source project called Quill, which is a rich text editor. He built the actual backend architecture and the system that scraped the web. And collectively we built this thing that finally gave a footprint to the impact of open source projects, whereas it prior did not exist whatsoever. You didn’t know who was using the project. So the outcome was not just an idea of, “Okay, great, people are using my project”, but also “Oh, wow, Uber.com, WhatsApp, Twitter - these companies that I use, these sites that I go to every day are using my thing. That’s really cool.”

So what I saw was Libscore motivating developers who may not have had that insight into who is using their stuff to be like, “Oh wow, I really wanna make sure I continue maintaining this. I want to merge these PRs, there’s a lot of people depending on this, and now I know who that is.” And it provided that tangible footprint, which was pretty exciting.

Chris Benson: The next thing - we’ll take a little bit of a turn… I’m often really interested in some of the non-technical posts as well about how AI is affecting the world in different ways, and I came across one that is from Harvard Business Review, and it’s called “What’s the purpose of companies in the age of AI?”

I’m often talking about digital transformation in general and how AI affects that, so that caught my attention… It was interesting to see Harvard Business Review seriously considering with these new tools how does that literally change the business functions - technology aside - that AI is introducing.

To summarize it, they kind of came down to that from a business strategy standpoint, AI was really gonna have an effect in one of the four following ways. One is really AI being used to exploit existing advantage, and to make that existing advantage more pronounced.

They also talked a lot about tradeoffs in terms of long-term and short-term, and we’re all used to hearing about companies that get caught in the short-term for stock market, for quarterly reporting and stuff like that, but that AI can be used to figure out what tradeoffs make sense for moving companies more into a long-term perspective, and that there may be some insights there that would not otherwise be available.

The third one that they mentioned really appealed to me, and I know it’s gonna appeal to you; they’re talking about – they define it as a moral or spiritual call to action, and I know you explicitly always talked about AI for good, and that’s a big, important thing for me as well… As I think my future and the organizations that I wanna be part of, knowing that that organization stands for something more than just making a profit is a motivating factor… So I found it really interesting to think of AI being used to promote that bettering the world approach.

And then the final thing was really being the – they use Steve Jobs and Elon Musk as examples… Is really people who are going to really meld the world into their own view, rather than trying to fit into the world. You might say that that is innovation for the purpose of creating value… But I just like seeing this conversation where companies are recognizing how important this is and recognizing it’s not just another tech coming in, but it’s actually something that is gonna affect the core way they operate.

Jerod Santo: It’s very cool. Yeah, it’s funny, because I looked into Erlang probably way back when you were starting DNSimple, and took a few tutorials, and I just couldn’t get the syntax to stick into my brain… But I liked, for instance, the pattern matching that you’re describing, and a lot of the properties of it just seemed really smart. And I couldn’t quite get over the hump. And then when Elixir came out, it actually – because I was a Rubyist before that, and I knew Jose and everything, it all kind of connected. And so I’ve been writing Elixir for a long time, and enjoying some of the advantages of the BEAM, with a language that for me just more syntactically aligns with the way I think. But I think once you get past that, if you’ve written some Erlang for long enough, it’s like the matrix; you don’t even see the ones and zeros anymore.

Beyang Liu: It’s really cool that you can one-shot that now. But our mission, the thing that gets us out of bed every morning is really to enable the software that drives the world to be built much more quickly and much more robustly. And so that’s the thing that we’re targeting, and that’s the reason why the customers that have chosen Sourcegraph have chosen us… Because they have large, complex codebases and they appreciate the other technologies that we bring to the table, that help us be effective in that setting.

And that brings me to this kind of second dimension, which is - our technical strategy here is transformers, LLMs, AI… That’s all really useful, it’s game-changing technology, but we think that there is a step function more capability and power you can unlock by combining that technology with some of the other technologies that we’ve built around information retrieval, code search, mapping out the reference graph and the knowledge graph of the code… We think these technologies are highly complementary. In some sense, they represent the two sides of what traditionally has been called artificial intelligence. Before the current incarnation of what AI means, there was this classic debate between the formal methods, like context-free grammar, Chomsky kind of party of artificial intelligence, where everything’s kind of rule-based, and there’s some philosophy behind it that these grammars emerged from the mind of maybe some higher order being in our goals to discover the platonic ideal, the hidden structures that connect all these things… You know, very symbolic-based.

And then on the other end is the statistical approach, which is like we’re not going to be opinionated about defining down a complete set of axioms or rules for describing what language is; we’re just going to try to model it statistically and learn it from the data. I think the age we’re living in, everyone sees clearly the limitations of the Chomsky approach, and it’s very apparent to everyone the advantages of the data-driven approach now, with modern technology. But I think there’s still actually highly complimentary. There’s a bunch of productivity-oriented use cases that essentially combine a knowledge graph or some sort of symbolic representation with generative models. So you have the generative model providing some notion of reasoning, but you combine that with a symbolic representation to be very precise about how things are connected. And the composition of those two yields way more useful use cases than if you were to just use one or the other.

Arun Gupta: That is correct. I mean, different foundations operate very differently. They play a very critical role in terms of growing these open source projects. They provide this central functionality like CI/CD, security audits, legal support, marketing support, event support… So a lot of advantages of these open source foundations.

Some foundations give you a seat at the governing board, depending upon what tier of membership you take. Most of the foundations that I’m aware of, they usually have their technical oversight committee or technical advisory committee. Those seats are all elected. Those are not pay-to-play at all. The maintainership is, again, meritocracy, so douocracy, and how do you rise up to become a maintainer. So in that sense, the administrative part and the technical parts are separate. Like, I’m on the governing board, so we are responsible for the administrative, financial, legal aspect of it, but we have zero say in terms of what the technical roadmap of the foundation is going to look like…

[01:12:23.13] And I like that separation of concerns in that sense. But if the technical folks need some budget support, for example in OpenSSF, they come over to the board. But we’re also trying to create swim lanes over there, where a certain level of funding is not required to be seeking permission of board every time, because we want to create swim lanes where people can go faster on their own.

Kent Quirk: [00:20:08.15] I mean, that drove an awful lot of Node adoption, because people are like “I’ve been coding in frontend in JavaScript, so why can’t I code the backend in JavaScript?” And that was in large part why the team that I had worked on was using it. And once you realize that there is a bright line separation between those two, there’s no particular reason that doesn’t justify, in my view, wanting one language. The languages are pretty fungible. You can you can switch between different languages and get the thing done. It’s the systems around those languages and what they give you, and that’s one of the big benefits of Go. I think, from my point of view, the most enormous benefit of Go is the fact that if I went back to the code I wrote in 2014, I could even compile it with the current Go compiler. I can read it, I can maintain it. And Go, beyond – like, the JavaScript code we had at that time wouldn’t even run today. And it would be considered five generations old. And one of the major, major, I think, advantages of Go is its long-term maintainability.

The Go 1.0 compatibility promise has been really, really important. And the fact that that code - I can still read and write the same code, in the same styles, for a decade… It turns out software lives a long freaking time; a lot longer than you wanted it to. I’ve got stories about that too, but man, you’ve got to be able to keep working on it. And the pain of – you know, we just watched the team switch from React 16 to 18… That was a painful multi-month effort, by multiple people.

Jerod Santo: I certainly think it is. Obviously, you can have more or less success with open source. And yet, it’s a great way to be able to point at something that you’ve done, or built, which is not proprietary, or needs permission from your previous employer… So there’s a lot of advantages, even just in that. And of course, there’s opportunity to raise your clout in the community by having a successful project or participating in one. So I think it absolutely is.

Where I thought you were going to go with that - maybe you’re headed there - is that content creation. Hm. Is open source creating content? Because –

Jerod Santo: Yeah, it really speaks to the advantages of going all-in on a technology, and really leveraging it for what it’s good at. And I freely admit that that’s one of the nice things about TypeScript, is that.

Adithya Krishnan: I can give a brief overview of this. Some of the biggest users of DuckDB come from the Python ecosystem, which means that it’s being a stand-in for a data frame, for example. And one of the advantages of using DuckDB is that it’s really fast on aggregates, and for the Python ecosystem it helps with standing in for a data frame to be used with other ML libraries, for example.

So that’s one part of the ecosystem. And the other part of the ecosystem is for a data engineer to be able to pull in data from different sources - like you said, Postgres, from CSV - and to be able to join those different datasets. Joins are really good with DuckDB as well, and to create transformed datasets is also pretty useful.

And on the third ecosystem, for a data analyst who is writing SQL, and one of the really nice aspects of DuckDB is the SQL dialect itself. It’s pretty flavored, that you have a lot of DuckDB functions that makes data cleaning easy, data transformation easy. For example, we also have a dialect that says FROM TABLE, and that’s just going to show you the table. Instead of going SELECT STAR FROM TABLE, you can go FROM TABLE, and that will just fetch data from that table. So there are these flavors of dialect for DuckDB that makes it nice.

Erez Zukerman: Two main reasons. Posture and movement. Posture, first of all, is - let’s say I were to separate the parts and leave them in one place; just stretch them way out and leave them in one place. Actually, I’m going to go with posture, movement and mobility. I’ll get to that. But posture. If I were to separate the two parts and just leave them fixed in their new positions, I already opened up my chest. I dropped my shoulders. The whole neck area. Oeople get stress and tension headaches because of - this has to do with the monitor placement, but also the keyword placement and the interaction between the shoulders and the neck opens up. And then once you open up, that also has implication for breath. It’s actually easier to breathe, and that has a whole host of other health repercussions, or I guess benefits. To take the positive here. If your breath is working, many things in your body work right. So just open up, calm down, and it reduces stress across many parts of your body, not just your hands.

That’s one benefit of split. But the other benefit is being – well, and I guess actually to go more in position, another side benefit there is that you can tend to them. You can control the angle relative to the desk, and actually make a little roof. And it doesn’t have to be the same angle on each half. So one can be like really severely angled to an almost handshake position, the other one much flatter. That’s fine. The keyboard can do that. That’s not a problem, because it’s split. It’s two totally independent halves, right?

[23:57] And then you can also rotate them as whatever works for you. You can have them rotated o you’re coming in from the side like this, or you can have them rotated outwards… And again, you can change each of them independently. But that brings me to movement, which is the second one. You can actually - you can, and I’m going to say you should make those changes throughout the day. Move around. Like, one thing that really gets me is those ergonomic posters and infographics that show you like the right typing posture… And you have this robot-looking person that’s all right angles, and the head is exactly straight, and the forearms are exactly straight, and the knees… Okay, that’s one way to do it, sure; for 5-10 minutes. But really, what we are meant to do as humans is move. We are meant to shift our position throughout the day. And I definitely think you should take movement breaks when you work. You should stand up, you should sit down.

I have this very cool little thing here called the MoveMate, actually, that’s like this wobbly platform that you can stand and kind of play around on… But a part of that is being able to just shift your keyboard around; draw it in closer, push it far, all that. And that also has other advantages, of course… Less movement-based, but let’s say you’re gaming. No problem. You just unplug the right half entirely, chuck it out. Lots of room for your mouse, and the left half can now be this super-powered macro pad just for your game, if you’re like an MMO type of guy. So that’s the second one.

And then the third one, the third good reason for a split - it packs down. It’s really easy. When you type, it’s a full size, spacious keyboard, because you have all this room to spread out… But when it’s time to take it somewhere, there is an included carrying case which is small, because it’s like a sandwich. It’s basically the footprint of one half. So you’re just going to smoosh it together…

Steven Wu: [00:20:07.00] Yeah, we’re basically fully focused on the like kind of freeform text search with aggregations use case. So our product is not schemafull. So you can dump arbitrary text files or JSON files into it, and it is aware of the concept of columns, but there’s no – you don’t have to pre-specify “These are the columns or you’re we want to index.” We index pretty much everything you dump in. And the columns end up just kind of being like metadata information on that data, where – I mean, I don’t want to say it’s like NoSQL, because it’s not a database in the traditional sense… But it is NoSQL in that there’s no tables, no schema… And kind of the goal is just to make it as easy as possible to get your data in there and have it be usable, and then to have the right tools so that you can still kind of treat it as if it were a columnar… Which is an interesting technical problem, I think, and I think a lot of our resources have been spent on “How do we make this system fast?” And how do we make these particular aggregation operations feel good? And what do you do if someone does aggregations on columns in weird ways, where that column has exactly one value in the entire database, and it’s like somewhere over here in this corner, [unintelligible 00:21:10.06] And I don’t think we solve it perfectly, but I do kind of think that, for the most part, there are notable advantages of using something like what we do over using something that requires a schema. And I think it might be true that if you had like a really, really, really, really big cluster of whatever database system is your preference, and you sharded it, and you have like a guy or like two guys full-time running it, that you could get comparable or better performance, but it costs you a lot more, both in terms of having two guys running it, but also just in terms of you would need a lot of beef to handle that much MySQL or that much Postgres, or that much Snowflake even.

And I just I brought up Snowflake because Snowflake is one of the direct – when we sell to a lot of these security teams, Snowflake is the preferred data store. And a lot of companies – I don’t know how familiar you guys are, but like Panther Labs, for instance, is like one of the larger SIEM tools, and their entire product is built on top of Snowflake. So they basically – they just give you access to like querying on Snowflake, which is very powerful. But then also, it costs the same as Snowflake, so that ends up being a blocker for lots of people.

Ian Lopshire: Yeah, you’d almost have to hide it in plain – well, not almost; you would have to hide it in plain sight. I think that’s one of the advantages of how Go does things. For, say, a C program, you’re often shipping a tarball with compiled stuff in it. So you can hide whatever you want in there.

Denis Yarats: [00:33:55.28] I believe and – I mean we were very confident early on chat interfaces is a temporary thing. It’s just too limiting. It has a lot of constraints. And that’s why we didn’t follow the usual road of all of the chatbots. They literally copied ChatGPT and put like a chat interface. We thought a little bit more about this, and we decided – I feel ultimately, right now we’re still in this early stage where people care about the model itself, so the model is the theme… But as this thing gets more advanced, as more people start using gen AI products, I feel like the main thing is going to be the product itself. What kind of things can the product do? Do you do it better than this? Do you have the best UI? Do you have the best UX? And that’s why we early on we were thinking about those things, and designed our product in a way that is the most suitable for the things that we wanted to do. If it’s search, we knew that chat doesn’t make sense for search. It’s just like, that’s not how people search for information. That was a very big factor, I think, in our success.

The other thing I guess we also – even last year, we started prototyping and experimenting with this concept of generative UI… So something where the LLM can guide what UI elements you can generate… Sometimes one of the things in a chat interface, if you want to ask a follow-up question, sometimes it doesn’t make sense to ask it as a sentence. So maybe you want to show a checkbox, or a button, or whatever. Especially on mobile - everybody uses phones - it’s just not very convenient to type, especially if you’re on the run, so you’d rather press a button. That’s why maybe speech, and I guess voice technology is going to be one of the interesting modalities, for sure, for an interesting interface, because it has a lot of advantages. Obviously, it has lots of a disadvantages too, but it’s definitely going to be interesting. And I think going forward, as we go towards agentic behaviors and more things are going to become possible, I think it’s definitely not going to be chat interfaces. It has to be something else.

Daniel Whitenack: I have a selfish question, which now that – this is one of the advantages of doing the podcast, I get to talk to all the amazing people doing amazing things, and learn from them. But I’m wondering, as a person who is also trying to fine-tune some models, either just for my own enjoyment and learning, but also fine-tuning models for specific tasks, and in specific customer use cases and that sort of thing… There’s a lot of people out there, I think many of our listeners who are thinking like, since you being part of this collective have worked since the sort of dawn of these many proliferation of fine-tunes, from LLaMA etc. and as you’ve seen all that, as you’re doing more and more fine-tunes, now as you’re looking towards the future, do you have any kind of good advice or things to keep in mind for all those fine-tuners out there that are thinking about grabbing something off of Hugging Face, creating their own versions of these models, maybe they have their own ideas about a specific take on a model? Any general tips that you’ve found to be really useful over time, or like pitfalls that you’d like to highlight?