WEBVTT 00:00:00.005 --> 00:00:02.605 Oh, and I just got hungry. Right on time. 00:00:02.725 --> 00:00:03.065 Right on time. 00:00:03.325 --> 00:00:04.205 Isn't that something? 00:00:04.345 --> 00:00:05.325 That's how we know it's time to stop the show. 00:00:05.325 --> 00:00:06.705 Now I'm hungry. How dare you? 00:00:07.665 --> 00:00:10.065 You guys are going to have to share oatmeal. It's going to get rough. 00:00:10.145 --> 00:00:10.965 Sorry about that. 00:00:12.165 --> 00:00:13.665 No, I got my... 00:00:13.665 --> 00:00:15.985 Don't put Matt over the soundboard. It's so risky. 00:00:16.285 --> 00:00:17.305 Can you believe this guy? 00:00:17.345 --> 00:00:18.845 He's a wild man this morning. 00:00:19.345 --> 00:00:19.665 Chaotic. 00:00:19.665 --> 00:00:20.205 What happens? 00:00:20.405 --> 00:00:20.765 Chaotic. 00:00:20.885 --> 00:00:22.205 As it shakes him up. 00:00:22.865 --> 00:00:25.865 He's trending to chaotic. Neutral. I mean, I don't know. 00:00:26.185 --> 00:00:27.105 Uh-oh, D&D reference. 00:00:27.225 --> 00:00:27.925 It's trending here. 00:00:28.045 --> 00:00:28.525 Oh, come on down. 00:00:40.297 --> 00:00:44.837 Hello, friends, and welcome back to your weekly Linux talk show. My name is Chris. 00:00:45.077 --> 00:00:45.877 My name is Wes. 00:00:46.077 --> 00:00:46.777 My name is Brent. 00:00:46.917 --> 00:00:47.917 And my name is Jason. 00:00:48.197 --> 00:00:52.017 Hello, gentlemen. We are live from beautiful Pasadena, California, 00:00:52.037 --> 00:00:57.077 and it is Planet Nix and Scale 2026 going on right now. So you guessed it, 00:00:57.157 --> 00:00:59.257 that's what we are talking about this week. 00:00:59.397 --> 00:01:02.457 We have really seen the future. 00:01:02.717 --> 00:01:07.497 I mean, this is probably one of the biggest trendsetting scales we have ever been to. 00:01:07.637 --> 00:01:10.677 So we'll tell you all about that. Then we'll round out the show with some great 00:01:10.677 --> 00:01:13.517 boosts and picks and a whole lot more. 00:01:13.697 --> 00:01:17.317 So before we get to that point in the show, we got to do a little good morning 00:01:17.317 --> 00:01:20.417 to our friends over at Managed Nebula. 00:01:20.537 --> 00:01:24.317 Go check out Defined Networking and grab yourself some Managed Nebula. 00:01:24.857 --> 00:01:27.277 Now listen, Nebula is one of our 00:01:27.277 --> 00:01:31.097 favorite open source project and Defined Networking is doing this right. 00:01:31.277 --> 00:01:33.957 They have built a decentralized VPN on top of the 00:01:33.957 --> 00:01:36.717 open source Nebula platform that you can get set up in just 00:01:36.717 --> 00:01:39.737 minutes no credit card required you go to define a.net slash 00:01:39.737 --> 00:01:43.077 unplugged it's built for simplicity it's serious security 00:01:43.077 --> 00:01:46.217 and if you're building a home lab with just a couple of machines or you 00:01:46.217 --> 00:01:49.857 have an entire global enterprise infrastructure it will 00:01:49.857 --> 00:01:52.717 manage the job no fragile hub and spoke choke point 00:01:52.717 --> 00:01:56.097 no control plane that you don't manage no particular 00:01:56.097 --> 00:01:58.817 big tech lock-in at all you build it how you want 00:01:58.817 --> 00:02:02.697 it if you want to self-host the network you can or if you want to take advantage 00:02:02.697 --> 00:02:08.237 of their whole setup with 100 hosts absolutely free go to defined.net slash 00:02:08.237 --> 00:02:12.477 unplugged support the show no credit card required and check it out see why 00:02:12.477 --> 00:02:17.917 we love it and a big thank you to defined.net for sponsoring this here show you guys are the best. 00:02:17.917 --> 00:02:22.077 You know you can tell we love it because we were troubleshooting some networking 00:02:22.077 --> 00:02:25.137 issues for our pal brentley here and what was the first thing that we said you 00:02:25.137 --> 00:02:26.717 know if we were using nebula it's. 00:02:26.717 --> 00:02:28.177 So true yeah we can we how can. 00:02:28.177 --> 00:02:29.497 We use nebula to test this. 00:02:29.497 --> 00:02:33.117 We'd have better insights into this problem if we were using Nebula. That is very true. 00:02:35.755 --> 00:02:39.235 Before we do chat about everything, I just want to say one more time, 00:02:39.395 --> 00:02:41.835 thank you to Phlox for sending us down here. 00:02:41.975 --> 00:02:44.875 And then honestly, for putting on a hell of a planet, Nick's. 00:02:45.355 --> 00:02:45.735 Oh, yeah. 00:02:45.915 --> 00:02:49.735 They really do a great job. They know the people in the industry to bring in. 00:02:49.815 --> 00:02:51.555 So you really get something valuable out of it. 00:02:51.695 --> 00:02:55.195 And then they bring everybody together with a couple of different after parties, too. 00:02:55.675 --> 00:02:58.815 And they provide. It has really been a fantastic event. 00:02:58.955 --> 00:03:03.975 And Phlox made it possible for us to make it. And they really are taking something 00:03:03.975 --> 00:03:06.295 that is very complex in corporations. 00:03:06.615 --> 00:03:09.495 You know, you get somebody in maybe your business or your enterprise, 00:03:09.535 --> 00:03:10.535 and they're the next person. 00:03:11.106 --> 00:03:15.486 And they become the next expert and then they move on and companies sometimes struggle with that. 00:03:15.626 --> 00:03:18.906 And there's ways to solve that. And Phlox is one of them that is trying to make 00:03:18.906 --> 00:03:22.506 it possible for people to create dev environments quickly that other people 00:03:22.506 --> 00:03:26.146 can address where you know the S-bomb, you have everything right there onboarded 00:03:26.146 --> 00:03:28.646 instantly across the entire software lifecycle. 00:03:29.126 --> 00:03:33.186 That's what Phlox is doing. It's phlox.dev and go check them out because they 00:03:33.186 --> 00:03:36.386 really, really made it possible. And they threw a hell of a party. 00:03:36.686 --> 00:03:37.826 And approachable too. 00:03:38.106 --> 00:03:38.326 Yeah. 00:03:38.326 --> 00:03:42.186 Like really, you know, happy to, you know, chat with you, give you a great time 00:03:42.186 --> 00:03:45.006 and just down to earth, which is actually really nice. 00:03:45.226 --> 00:03:47.146 We had a great meetup as well. 00:03:47.226 --> 00:03:47.826 Yeah, we did. 00:03:48.606 --> 00:03:54.466 I mean, it was, we were just non-stop, just shaking hands and saying hi and 00:03:54.466 --> 00:03:55.266 getting in conversations. 00:03:55.866 --> 00:03:57.266 Great to see Cessna Mike there. 00:03:57.726 --> 00:03:58.026 Yeah. 00:03:58.146 --> 00:03:59.486 Got some great stories. Good to see him. 00:03:59.606 --> 00:04:03.746 I didn't confirm whether he flew in or not. I guess he lives near it here. So why would you fly? 00:04:04.266 --> 00:04:06.566 Sounds like his plane might be in the shop for a bit. 00:04:06.706 --> 00:04:06.966 Oh, no. 00:04:06.966 --> 00:04:10.806 He showed me his avionics layout that he had replaced in his Cessna, 00:04:10.886 --> 00:04:14.106 and I was so impressed. I was like, oh, man, that is gorgeous. 00:04:14.486 --> 00:04:17.246 He's also, I won't say too much because I don't think it's fully public yet, 00:04:17.286 --> 00:04:21.626 but he's working on a really cool project in the avionics industry that brings 00:04:21.626 --> 00:04:26.846 a modern tech solution to a problem that has been in this industry for 50 years. 00:04:27.406 --> 00:04:29.426 Of course, it was great to see Carl from Texas. 00:04:29.866 --> 00:04:33.026 He brought a new type of pocket meat for us. 00:04:33.106 --> 00:04:34.106 Yeah, it was good, too. 00:04:34.246 --> 00:04:34.826 I was afraid. 00:04:35.026 --> 00:04:36.386 It was your first pocket meat. What did you think? 00:04:36.386 --> 00:04:41.886 But it was actually quite delicious. I mean, savory and at the same time sort of warm and cozy. 00:04:42.046 --> 00:04:45.046 It was like love in small meat chunk form. 00:04:45.366 --> 00:04:47.026 Harvested by Carl himself, as they say. 00:04:47.166 --> 00:04:50.686 It's true. Little deer. It's kind of like Bambi. Never mind. 00:04:50.886 --> 00:04:51.946 Nope, that didn't make me afraid. 00:04:52.186 --> 00:04:57.546 He also puts in the work. Not only was he there at the booth talking about Fedora 00:04:57.546 --> 00:05:02.366 and CentOS and all the great stuff he's up to there, but he also led a workshop. 00:05:02.366 --> 00:05:05.246 So you get hands-on stuff if you want to try your hand at packaging. 00:05:05.766 --> 00:05:10.286 Also, there's the companies that don't necessarily have an official presence, but they're here. 00:05:10.849 --> 00:05:16.269 you know um anthropic is one of them shopify was another system 76 they were 00:05:16.269 --> 00:05:19.749 they had some booth stuff here with the amper folks but uh like emma was here 00:05:19.749 --> 00:05:21.029 we got to hang out with emma always. 00:05:21.029 --> 00:05:21.969 Good more lasers. 00:05:21.969 --> 00:05:27.189 And we saw listeners from all over the place anybody stand out anybody i mean i. 00:05:27.189 --> 00:05:31.349 Always try as you know to figure out who came for the furthest and uh we had 00:05:31.349 --> 00:05:34.149 a listener who came all the way from north california. 00:05:34.149 --> 00:05:36.389 North carolina. 00:05:36.389 --> 00:05:37.129 That's the place. 00:05:37.129 --> 00:05:37.769 Nice uh. 00:05:37.769 --> 00:05:41.929 Which is pretty much the opposite i think it was dewey you're right. 00:05:41.929 --> 00:05:42.569 Pretty much the. 00:05:42.569 --> 00:05:44.789 Opposite side of the continent distros. 00:05:44.789 --> 00:05:45.289 Stew was there. 00:05:45.289 --> 00:05:49.789 Distros stew yeah yeah it was great it was great i mean i i didn't get everybody's 00:05:49.789 --> 00:05:52.389 handle so i don't i can't shout at everybody but true we're sorry we. 00:05:52.389 --> 00:05:52.769 Love you though. 00:05:52.769 --> 00:05:56.869 And um um some people that were new too we got to meet some new folks it's always 00:05:56.869 --> 00:05:58.829 a chance to network so you. 00:05:58.829 --> 00:06:01.909 Know i stopped you on the stairs right before you walked down to planet nix 00:06:01.909 --> 00:06:05.489 and they were like hey hey it's you and one of them literally said, 00:06:05.994 --> 00:06:07.214 You guys just made my weekend. 00:06:07.474 --> 00:06:07.754 Yeah. 00:06:08.094 --> 00:06:08.974 It was so sweet. 00:06:08.974 --> 00:06:11.694 It was so sweet and beautiful to behold. As like, as an outsider, 00:06:11.714 --> 00:06:13.794 I was just like, look at them guys shine. 00:06:13.954 --> 00:06:14.354 It was fun. 00:06:14.494 --> 00:06:17.194 You know, multiple times, I do think we ran into people, or at least I did, 00:06:17.294 --> 00:06:21.014 where, you know, I knew one of the group, but they drug along a bunch of other 00:06:21.014 --> 00:06:24.774 people for this Linux thing or the Nix part or, you know, whatever nerdery they were sharing. 00:06:24.914 --> 00:06:27.494 The other thing that's fun is, you know, you walk up to somebody like, 00:06:27.574 --> 00:06:29.574 oh, you do that project? I like what you make. And they're like, 00:06:29.634 --> 00:06:32.254 oh, I like what you make. And it's, you meet, oh, you're the person who does that. 00:06:32.634 --> 00:06:35.214 And that's really the best thing about these conferences, right? 00:06:35.214 --> 00:06:39.374 the talks and stuff are good but like they say the hallway track or for us the 00:06:39.374 --> 00:06:41.814 stairway track is really where all the magic happens. 00:06:41.814 --> 00:06:47.154 And it is if you missed your opportunity linux fest northwest is just around 00:06:47.154 --> 00:06:52.014 the corner april 24th to the 26th at bellingham technical college in the beautiful 00:06:52.014 --> 00:06:54.974 pacific northwest at a beautiful time of the year rumor. 00:06:54.974 --> 00:06:56.514 Has it we'll be doing a live show. 00:06:56.514 --> 00:07:01.574 Oh god and right snap the big venue's back to the big room is back so we're 00:07:01.574 --> 00:07:04.334 back in our original location, well, what has become our original location. 00:07:05.734 --> 00:07:08.194 We'll tell you more about it as we get closer. I won't nag you about it, 00:07:08.254 --> 00:07:10.374 but I just want you to know, if you're feeling a little FOMO after listening 00:07:10.374 --> 00:07:11.714 to this episode, you do have another opportunity. 00:07:11.934 --> 00:07:16.334 I will say I've committed to bring the van to the venue. Here we go. It's on the record. 00:07:17.254 --> 00:07:17.574 Committed. 00:07:17.794 --> 00:07:18.134 There it is. 00:07:18.234 --> 00:07:18.854 Can I get a lock? 00:07:19.094 --> 00:07:20.154 I think it is locked. 00:07:21.094 --> 00:07:23.554 Drew, don't edit that out no matter how much he ends. You're doing a good job. 00:07:23.674 --> 00:07:26.254 You know what? Let's officially lock that in. There it is. Thank you. 00:07:26.394 --> 00:07:26.774 Well done. 00:07:28.094 --> 00:07:30.274 Start betting on whether that'll happen or not. 00:07:30.374 --> 00:07:34.094 Yeah, right. Okay. So let's do this in chronological order. 00:07:34.214 --> 00:07:39.094 The way this worked is Scale has had an expansion over the years where they 00:07:39.094 --> 00:07:41.814 have what you could consider parallel tracks or side tracks. 00:07:42.394 --> 00:07:43.254 Subconferences, maybe. 00:07:44.074 --> 00:07:48.034 There's a Postgres subconference, there's a VoIP subconference, 00:07:48.254 --> 00:07:50.574 there was obviously a Nix, a Planet Nix subconference. 00:07:50.594 --> 00:07:53.014 Yeah, I think DevOps Days LA is part of this. 00:07:53.174 --> 00:07:59.874 Yeah, yeah. And so those subconferences start a couple of days before the actual Scale event. 00:07:59.994 --> 00:08:03.034 And then there's a tiny bit of overlap there. So we showed up. 00:08:03.357 --> 00:08:07.477 a couple of days before scale to attend planet nix and we got ourselves all 00:08:07.477 --> 00:08:11.437 situated and arrived on day one with uh bells on. 00:08:11.437 --> 00:08:15.717 Well it's the morning of planet nix just about to get going and we've replaced 00:08:15.717 --> 00:08:19.897 brent i think this is a great idea your idea really well i wanted to use your 00:08:19.897 --> 00:08:23.997 personal openclaw agent but you broke it so we couldn't but the next best thing 00:08:23.997 --> 00:08:26.697 is we brought jason with us hi jason can you describe the scene for us, 00:08:27.520 --> 00:08:34.380 Right now, we're inside a room, inside a larger room, inside another room, here at scale. 00:08:34.860 --> 00:08:38.100 The planet nixifying itself as we speak. 00:08:38.400 --> 00:08:43.240 That is actually the best description ever. It's a room, in a room, that's in a room. 00:08:44.280 --> 00:08:47.320 Instead of like, in a world, it's in a room. 00:08:48.360 --> 00:08:50.860 Okay, I lied, we brought Brent too. You made it, Brent, Lee, 00:08:50.940 --> 00:08:54.180 you made it. Oh, hi, I'm here. Hey, guys, you're looking sharp. 00:08:54.580 --> 00:08:59.480 Well, we dressed up for you. So they're going to give us the introduction and all that jazz. 00:08:59.720 --> 00:09:03.460 And then a little bit later in the day, Wes has a talk. And I'm already thinking about lunch. 00:09:04.060 --> 00:09:07.060 Yeah, well, I always do, of course. Because as soon as these things start, 00:09:07.120 --> 00:09:07.880 that's when I get hungry. 00:09:08.700 --> 00:09:13.320 And I thought it was good to sit there and kind of get the latest update from Ron. 00:09:13.540 --> 00:09:18.280 Ron, not only the CEO of Phlox, but also runs the, he's the head of the Knicks Foundation. 00:09:19.020 --> 00:09:22.400 And there was a good conversation between him and Kelsey Hightower, 00:09:22.500 --> 00:09:24.080 which we had on the show just recently, Kelsey. 00:09:24.800 --> 00:09:28.220 And I thought this point was worth capturing for you. And that is sort of this 00:09:28.220 --> 00:09:31.420 discussion with if vibe coding becomes more and more of a thing, 00:09:31.500 --> 00:09:32.580 and then people have these agents 00:09:32.580 --> 00:09:35.600 that can just crank out software as they need it to solve solutions. 00:09:36.080 --> 00:09:42.020 Do we, as an open source community, start to see a lack of contributions back? 00:09:42.200 --> 00:09:46.540 Because these things are just one-shotting or creating a custom bit of code for the user. 00:09:46.740 --> 00:09:49.420 The user implements it, and they never contribute anything back upstream. 00:09:49.520 --> 00:09:52.300 And that was the sort of context for this conversation. 00:09:52.300 --> 00:09:57.140 There's no guarantee there's going to have to be anything. That's the hard truth. 00:09:57.440 --> 00:10:01.580 If you all believe that it's just going to be there because we want it, that's not how it goes. 00:10:02.100 --> 00:10:06.920 A lot of us don't own these companies. A lot of us are employees. We're not. 00:10:07.529 --> 00:10:12.189 at these companies. They're given RSUs that you probably sell pretty close to 00:10:12.189 --> 00:10:13.869 immediate because it's deferred comp. 00:10:14.309 --> 00:10:18.409 A lot of us aren't that invested in this. And there are pressures, right? 00:10:18.489 --> 00:10:22.369 I'm watching people say, hey, I'm anti-AI until you see the layoffs. 00:10:22.709 --> 00:10:25.709 Changes your tune real quick if you need that job. 00:10:25.909 --> 00:10:32.949 So I think the reality is the tools of the trade will dictate what the hiring process looks like. 00:10:33.129 --> 00:10:35.709 I'm pretty sure someone somewhere is like, I don't believe in typewriters. 00:10:36.817 --> 00:10:41.777 It's going to be real hard to get a job today if you don't believe in mechanical keyboards. 00:10:42.177 --> 00:10:44.317 It's not going to work. The team will look at you crazy. Like, 00:10:44.357 --> 00:10:47.397 what do you mean you don't believe in email and sending electronic messages? 00:10:47.717 --> 00:10:51.737 That will change. It probably will. And where the inertia is right now, 00:10:52.217 --> 00:10:55.397 a lot of people are trying their very best to make that happen. 00:10:55.397 --> 00:10:57.057 So we've got to be honest about it. 00:10:57.217 --> 00:11:00.817 And if those tools are actually any good, would you not use them? 00:11:00.957 --> 00:11:02.337 Who uses a compiler here? 00:11:03.077 --> 00:11:05.757 Okay. There was an argument. If you look back in the archives, 00:11:05.857 --> 00:11:09.017 30 years ago, compilers are for people that don't know how to write code. 00:11:09.397 --> 00:11:12.257 What are you using a compiler for? Are you going to let the compiler decide 00:11:12.257 --> 00:11:14.177 what instruction sets you use? 00:11:15.117 --> 00:11:20.657 People are still arguing about garbage collection. Garbage collection is still a contentious. 00:11:23.217 --> 00:11:27.537 Garbage collection? Nah, real people manage their whole memory. 00:11:29.097 --> 00:11:32.377 Buffer overflows and all. So the way I think about it right now, 00:11:32.497 --> 00:11:35.597 that's what we're faced with, the reality of it. So I think a lot of people 00:11:35.597 --> 00:11:37.077 here are practitioners. 00:11:37.737 --> 00:11:40.717 You're at a NICS conference, Flux conference. 00:11:40.977 --> 00:11:45.297 You are probably in a fringe part of the technology where you know what's going on. 00:11:45.497 --> 00:11:48.637 You've turned that knowing what's going on to leverage. And also, 00:11:48.797 --> 00:11:52.737 I would really hope that people who are in charge of the core layers of our 00:11:52.737 --> 00:11:54.497 infrastructure actually know what's going on. 00:11:54.617 --> 00:11:58.917 So I'm not surprised that people in this community need to know how software is actually built. 00:11:59.177 --> 00:12:02.557 We need to know how software can be exploited. And we need to understand how 00:12:02.557 --> 00:12:06.757 these tool chains go together. So I think this group is very special in that regard. 00:12:06.837 --> 00:12:09.517 But if you all zoom out a little bit, that's not the norm. 00:12:10.077 --> 00:12:12.217 A lot of people can care less where the software comes from. 00:12:13.160 --> 00:12:16.300 Appgit install the internet until this thing works. 00:12:16.880 --> 00:12:19.660 Not sure who packaged it. I don't even look at the author's name. 00:12:19.980 --> 00:12:24.860 I don't look at the checksums. If it works, it works. That's the reality. 00:12:25.140 --> 00:12:31.300 And one thing you and I were talking about was, do you need package management in this day and age? 00:12:31.840 --> 00:12:34.920 And you all would say, Kelsey, of course, this is a silly question. 00:12:35.040 --> 00:12:37.080 But have you seen people use some of these tools? 00:12:37.520 --> 00:12:41.520 Give me a thing that does X. And I don't honestly see these things go out and 00:12:41.520 --> 00:12:42.660 go find the right library. 00:12:43.020 --> 00:12:46.900 Sometimes it just generates the snippet. You remember the left pad debate? 00:12:47.040 --> 00:12:51.140 Should you use things like left pad where you can just write a three-line function 00:12:51.140 --> 00:12:53.260 instead of taking a third-party dependency? 00:12:53.800 --> 00:12:57.980 Well, every prompt is having that same discussion except for the developer isn't 00:12:57.980 --> 00:12:59.360 really making that decision anymore. 00:12:59.600 --> 00:13:03.660 So what happens to all this package management work that we're doing underneath the hood? 00:13:03.920 --> 00:13:08.860 Do we become the guardrails or do we get replaced by people creating software 00:13:08.860 --> 00:13:11.140 for themselves and not sharing it back. 00:13:11.440 --> 00:13:15.680 I think this is an interesting discussion to be having right now. 00:13:16.280 --> 00:13:19.560 And it's probably something we're going to have to wait and see to a degree. 00:13:19.780 --> 00:13:22.580 But I do think maybe it's the wrong framing for the conversation. 00:13:22.780 --> 00:13:27.800 I think the right framing is, will operators be trained to submit that code 00:13:27.800 --> 00:13:30.400 back upstream, right? Because it's an LLM that's being operated. 00:13:30.520 --> 00:13:34.260 The LLM isn't choosing on its own initiative to build a piece of software, 00:13:34.360 --> 00:13:35.320 to fork a piece of software. 00:13:35.460 --> 00:13:40.900 The LLM is being instructed to do so. So will the operator remember to contribute backup code? 00:13:41.000 --> 00:13:43.540 Can we train that back in for people that maybe don't have that culture? 00:13:43.700 --> 00:13:48.740 Or will the agents that get built, this be part of their instruction set? 00:13:49.080 --> 00:13:52.860 Because if we get more autonomy a little bit, the agents are off and they're 00:13:52.860 --> 00:13:56.860 doing things on their own, is part of that mindset that you give the agent, 00:13:57.000 --> 00:13:59.140 quote unquote, that it goes and contributes? 00:13:59.460 --> 00:14:02.440 Because that's a thing it can do while you're asleep. So... 00:14:03.042 --> 00:14:06.122 I think the incentives line up, and I'm curious to know what you think, 00:14:06.262 --> 00:14:11.102 Wes, but I think the incentives line up to reuse existing open source code and 00:14:11.102 --> 00:14:12.842 to contribute back upstream, because, 00:14:13.322 --> 00:14:17.402 there is ultimately an incentive to use the least amount of resources, 00:14:17.602 --> 00:14:21.762 the least amount of tokens, and the least amount of time and produce the most stable result. 00:14:21.882 --> 00:14:25.502 And the way you do that is by taking advantage of things that have already been built and are solid. 00:14:25.982 --> 00:14:28.762 And it doesn't matter if you're an LLM or if you're a human, 00:14:28.782 --> 00:14:32.162 that is just sort of a natural incentive you often arrive to, 00:14:32.162 --> 00:14:33.622 unless it's some small piece, right? 00:14:33.982 --> 00:14:36.982 Well, I think that the size of that might now be able to be changed, right? 00:14:37.042 --> 00:14:42.622 Because I think the cost of having your own fork is less because the LLM can maintain it. 00:14:42.722 --> 00:14:45.682 And the LLM is pretty good about managing patch set, especially if you're just 00:14:45.682 --> 00:14:48.662 adding some stuff or making a few changes, not trying to re-architect the core. 00:14:49.502 --> 00:14:55.442 And it costs tokens to file the issue and to respond to the request to change your code. 00:14:55.762 --> 00:14:59.262 And I think you're right, it is still worth it. But I think there could be some 00:14:59.262 --> 00:15:02.462 shorter term incentives And maybe people that aren't yet don't fully get it 00:15:02.462 --> 00:15:03.342 because they haven't done a 00:15:03.342 --> 00:15:07.202 few cycles of this where it is just very convenient to have it, you know, 00:15:07.622 --> 00:15:10.842 copy the code upstream, maybe template it out, fork it, make your tweaks on 00:15:10.842 --> 00:15:14.202 top, carry your own changes, and you can push it back up. 00:15:14.282 --> 00:15:17.122 But you might not be material harmed in the short term if you don't, 00:15:17.162 --> 00:15:18.102 or you might not think you are. 00:15:18.602 --> 00:15:22.402 In the context of this conversation, I'm curious what's going to happen with 00:15:22.402 --> 00:15:28.362 licensing, because we've seen a lot of these LMs just suck in basically the world's content. 00:15:29.302 --> 00:15:34.182 but what has been really important for open source projects is to follow licensing and try to make, 00:15:35.002 --> 00:15:39.142 modifications available again is that we're just going to ignore all of that 00:15:39.142 --> 00:15:41.102 now and we're not going to get 00:15:41.102 --> 00:15:45.062 open source projects to have the same momentum that they have in sort of. 00:15:45.062 --> 00:15:48.082 Well I don't think necessarily I mean we can talk about the training but the 00:15:48.082 --> 00:15:51.262 use case could totally be compliant right you only have to distribute stuff 00:15:51.262 --> 00:15:55.842 if it's you know maybe a GPL or GPL and you're trying to distribute some actual 00:15:55.842 --> 00:15:58.602 end user thing you can carry your own internal forks as long as you want. 00:15:59.354 --> 00:16:03.694 Yeah, although it's easier said than done, especially for people that have never done this before. 00:16:03.954 --> 00:16:08.234 People that are like yourself, that are familiar with software development, yes, it's very trivial. 00:16:08.414 --> 00:16:13.894 I maintain it is not a trivial thing for regular average users. 00:16:14.074 --> 00:16:18.014 And to that end, I think this is speaking to sort of a meta theme. 00:16:18.134 --> 00:16:23.574 It wasn't implicitly stated at Planet Nix directly, but it definitely kept being inferred to. 00:16:23.574 --> 00:16:29.254 It was the sense that NICs might become a little more inevitable when agents 00:16:29.254 --> 00:16:35.054 and LLMs are building software simply because these things, these LLMs need guardrails. 00:16:35.174 --> 00:16:39.114 And they need the ability to know when they've broken the system or when the system is operational. 00:16:39.274 --> 00:16:43.994 And Ron kind of touched on that on stream or on the stage during the keynote. 00:16:44.014 --> 00:16:45.334 So I'll just play a moment of that. 00:16:45.334 --> 00:16:49.134 The way that I've been looking at it, and I've been talking to folks inside 00:16:49.134 --> 00:16:53.854 of Nix, outside of Nix, is I think the more I'm seeing where the industry is 00:16:53.854 --> 00:16:57.774 going is that I would love for it to be a fact, right? 00:16:57.914 --> 00:17:03.074 So to your point about the vision for Nix in the modern SELC is that Nix can 00:17:03.074 --> 00:17:08.174 be the pure way that we actually are able to derive the packages and the baseline 00:17:08.174 --> 00:17:10.114 infra for what we're building, 00:17:10.354 --> 00:17:14.614 regardless if it's led by a human, by an agent, by a model, it doesn't matter, right? 00:17:14.614 --> 00:17:18.974 So when you need something to be reproducible and you care about the speed, 00:17:19.114 --> 00:17:22.214 the infrastructure, the security, the underlying factors of it, 00:17:22.574 --> 00:17:26.754 I think Nix is uniquely positioned to be that platform. 00:17:27.414 --> 00:17:31.254 And again, folks in this room kind of understand the underlying tech for it 00:17:31.254 --> 00:17:32.654 and why it's uniquely positioned. 00:17:32.814 --> 00:17:36.854 But I think that's, you know, when you want to be pure, there's Nix. 00:17:37.014 --> 00:17:40.134 And I think it's also fine to tell the industry that it's fine to be impure. 00:17:41.496 --> 00:17:45.636 If you want to just vibe code an app that you don't care if you throw it away, 00:17:45.776 --> 00:17:49.176 right? Like there is still a reason to use paper plates. It's bad for the environment. 00:17:49.336 --> 00:17:53.096 You shouldn't do it. But, you know, when you're tired at the end of the day 00:17:53.096 --> 00:17:55.856 and you have like four people over and the last thing you want to do is dishes, 00:17:56.196 --> 00:17:58.836 throw the paper plates on there and then throw them in the trash. It's totally fine. 00:17:58.956 --> 00:18:01.816 I don't love the message, but I actually think I agree with that framing. 00:18:01.956 --> 00:18:04.296 A lot of this vibe coded stuff is paper plates. 00:18:04.496 --> 00:18:07.156 It's stuff you can throw away after you're done with the meal. 00:18:07.376 --> 00:18:10.016 And that stuff, I don't know if it really matters if it gets contributed back 00:18:10.016 --> 00:18:13.616 up because the person's not putting a lot of intention or design into it and 00:18:13.616 --> 00:18:16.276 the people that do put intention and design into it i think are inclined to 00:18:16.276 --> 00:18:19.476 contribute back upstream because i maintain it's a personal operator thing. 00:18:19.476 --> 00:18:22.236 And it's not like if you trolled all of github and 00:18:22.236 --> 00:18:25.976 looked that the code quality is the same across the entire thing guys let's 00:18:25.976 --> 00:18:29.996 not pretend i think this is a substrate difference it's it's not like before 00:18:29.996 --> 00:18:34.436 this humans didn't look at stack overflow and look at a bunch of stuff and then 00:18:34.436 --> 00:18:38.756 amalgamated in their mind and then spit out something else when a human does 00:18:38.756 --> 00:18:40.676 that were like, oh, look at them craft that stuff. 00:18:40.836 --> 00:18:45.676 But they're doing what the LLM was doing and they just did it in a human substrate with their brain. 00:18:46.356 --> 00:18:49.076 There's also maybe different kinds of open source and sharing. 00:18:49.536 --> 00:18:53.716 What might change is it's easier to have whatever you make be shared, 00:18:53.736 --> 00:18:55.516 especially if you have a flake.nix in there. 00:18:56.156 --> 00:19:01.036 Immediately any other system that has Nix, be it human or LLM or whatever, can consume it. 00:19:01.136 --> 00:19:05.236 So whether or not you take the time to do the stuff at the core root upstream 00:19:05.236 --> 00:19:07.296 of doing it, maybe it's a whole discussion. 00:19:07.536 --> 00:19:11.576 But I think there's a ton of incentives to just go put it out there for other 00:19:11.576 --> 00:19:13.976 things and yourself to use, because especially having it out there makes it 00:19:13.976 --> 00:19:15.276 easy to consume for yourself as well. 00:19:15.576 --> 00:19:18.056 Yeah, I think that's what I've been noticing is it seems like the first step 00:19:18.056 --> 00:19:20.896 is on the readme for a GitHub project. 00:19:21.096 --> 00:19:24.996 It's the readme now has, if you're a human, do this. If you're an LLM, 00:19:25.356 --> 00:19:28.776 do this. And they're like, there's actually writing the readme for an LLM audience. 00:19:29.056 --> 00:19:31.856 And you're seeing that more and more. And I think that's step one. 00:19:31.996 --> 00:19:33.556 And then I think step two is, 00:19:34.412 --> 00:19:38.252 And here's the flake. And because it tells the LM and because Nix, 00:19:38.392 --> 00:19:42.552 Nix can run on Mac OS, it can run on any Linux and it can run, of course, on Nix OS. 00:19:42.852 --> 00:19:47.652 It is sort of this. It could be this universal agentic package manager or something like that. 00:19:48.232 --> 00:19:50.672 That is the human never needs to worry about it. 00:19:50.752 --> 00:19:55.812 But what's interesting is humans can. It's sort of at this sweet spot where 00:19:55.812 --> 00:19:57.832 it is readable enough for humans. 00:19:58.232 --> 00:20:02.472 The four of us in this room have proved that. But yet it also lends itself towards 00:20:02.472 --> 00:20:05.532 machine learning, having the capacity to deal with it. 00:20:05.672 --> 00:20:08.592 And that's kind of a nice spot on the seesaw. 00:20:08.792 --> 00:20:11.992 Yeah. Yeah. It's in a way it's, it's nice that we were able to get in and learn 00:20:11.992 --> 00:20:13.152 it before the machines took over. 00:20:13.392 --> 00:20:16.392 Yeah. And I don't want, I don't want it to sound like morality based when I 00:20:16.392 --> 00:20:19.012 say it's just a substrate difference that there isn't, you know, 00:20:19.092 --> 00:20:20.892 some moral quandary about that. 00:20:21.112 --> 00:20:24.732 There is, there's a big difference between, you know, making all these data 00:20:24.732 --> 00:20:27.052 centers take up lots of natural resources and everything. 00:20:27.252 --> 00:20:30.052 There are plenty of differences in the end. I only meant 00:20:30.417 --> 00:20:33.817 If you take out the morality, it is what code gets generated. 00:20:34.137 --> 00:20:37.077 What you're talking about, like a human's copying from Stack Overflow versus an LLM. 00:20:37.077 --> 00:20:38.137 Versus an LLM doing it, yeah. 00:20:38.177 --> 00:20:43.177 Yeah, yeah, I agree. So after the talk, Wes, we actually, I kind of liked this. 00:20:43.277 --> 00:20:46.517 Wes's talk was up early in the day, early in the event. 00:20:47.497 --> 00:20:50.397 And Brent and I were, I mean, we were hyped. 00:20:50.497 --> 00:20:50.797 Oh, yeah. 00:20:50.897 --> 00:20:51.477 Oh, we were hyped. 00:20:51.477 --> 00:20:54.637 I mean, we've seen Wes talks at previous conferences like Texas Linux Fest, 00:20:54.677 --> 00:20:56.057 and they're always kind of wonderful. 00:20:56.277 --> 00:20:57.797 I'm so lucky to have you as my booster. 00:20:57.817 --> 00:21:00.757 You know, it was so good, though. I really felt like, I was like, 00:21:00.877 --> 00:21:02.057 you were right in the pocket now. 00:21:02.057 --> 00:21:06.377 What's great about a Wes talk, and Brent and I, we don't, he has it handled, 00:21:06.677 --> 00:21:08.897 so we don't have to worry and focus on that. 00:21:08.897 --> 00:21:09.177 We're not stressed. 00:21:09.457 --> 00:21:14.697 Okay, well, we socialized, and now it's time for Wes's talk. 00:21:15.017 --> 00:21:21.057 And he is doing Nix Anywhere Else, replicatable binaries via elf, 00:21:21.197 --> 00:21:23.177 oh, relocatable, sorry, my eyes are a little blurry. 00:21:23.817 --> 00:21:29.057 Relocatable? Yeah. Relocatable elf on the shelf binaries during the holidays. Did I get that right? 00:21:29.397 --> 00:21:32.557 Something like that. I think Wes might clear it up for us, but I think you're 00:21:32.557 --> 00:21:35.557 on the right track. Ho, ho, ho, lad. Where is your elf binary? 00:21:37.057 --> 00:21:39.897 I think Wes has it. I think it's going to be a good talk, though. 00:21:40.077 --> 00:21:41.037 My kids would love this one. 00:21:41.537 --> 00:21:43.537 It was a good talk, right? Good talk. 00:21:43.597 --> 00:21:44.677 Clean, smooth. 00:21:45.397 --> 00:21:48.217 I felt like I learned something while you were learning something. 00:21:48.337 --> 00:21:52.877 Because you not only took it to the place where I want to try this to solve 00:21:52.877 --> 00:21:56.217 this problem, but after you solved it, you were like, wait a second. 00:21:56.477 --> 00:22:00.757 What if I also do this? and it was so lovely. You guys have to watch it online. 00:22:01.037 --> 00:22:05.357 There was a moment that I was a little like, hold my breath and because Wes 00:22:05.357 --> 00:22:11.237 decided to do a live demo of taking the bcachefs Nix package and translating 00:22:11.237 --> 00:22:14.757 it over to a dev package and getting it actually installed and running and getting 00:22:14.757 --> 00:22:16.677 bcachefs on an Ubuntu system. 00:22:16.697 --> 00:22:17.237 So bonkers. 00:22:17.297 --> 00:22:19.277 Even though the Debian base has removed support. 00:22:20.317 --> 00:22:24.757 And now we can actually format a BcacheFS file system on Ubuntu, 00:22:25.137 --> 00:22:27.057 entirely built from Nix. 00:22:34.257 --> 00:22:37.197 So just to kind of emphasize what happened, it was the same tool. 00:22:37.597 --> 00:22:42.977 All these binaries have the metadata anyway. We were blessed from Ilko and others 00:22:42.977 --> 00:22:46.937 to come up with all of this crazy tooling to help Nix work that it turns out 00:22:46.937 --> 00:22:50.057 can do more than just make one of the best Linux distributions possible. 00:22:50.317 --> 00:22:53.837 It can do surgery in to bring whatever tools you need to Nix, 00:22:54.077 --> 00:22:55.917 even if it's not maybe the right way to do it. 00:22:56.057 --> 00:23:00.297 And it turns out it can be used for surgery out to take all of the wonders we 00:23:00.297 --> 00:23:05.837 have in the Nix ecosystem, make them portable, and bring them with you wherever you are. 00:23:06.457 --> 00:23:10.177 Patch Elf is really, it's setting metadata, it's writing some strings, 00:23:10.417 --> 00:23:13.497 and once you see that, you might start seeing it everywhere. 00:23:14.657 --> 00:23:18.317 Attaboy, Wes Payne, attaboy. It was a good talk. We enjoyed it, 00:23:18.417 --> 00:23:19.737 and like Jason said, we learned. 00:23:19.737 --> 00:23:23.137 All right. That was pretty good. The Pat your elf on the shelf had a little 00:23:23.137 --> 00:23:24.497 demo that I thought was pretty good. 00:23:24.637 --> 00:23:30.177 He managed to get Bcash FS working on an Ubuntu box by using Nick's tools. That was neat, right? 00:23:31.437 --> 00:23:35.817 I did not. I'm still kind of speechless. I did not expect that to be a possibility. 00:23:36.137 --> 00:23:39.597 Yeah. Right there on stage. It was good. It was a fun talk. 00:23:40.497 --> 00:23:45.477 And now it's lunchtime. That's my favorite, favorite time of the day. Give it. 00:23:45.637 --> 00:23:47.837 You know, what's funny is we don't have any clips after lunch. I don't know. 00:23:49.537 --> 00:23:51.677 Wes Payne you're doing a good job. 00:23:51.677 --> 00:23:58.157 Yeah so day one was great lunch was one of those things where on our way to 00:23:58.157 --> 00:24:02.177 lunch we got stopped 15 times and then we talked with folks and it was good 00:24:02.177 --> 00:24:03.537 it was a lot of socializing we had. 00:24:03.537 --> 00:24:05.317 A lot of people to see that we haven't seen in a while. 00:24:05.317 --> 00:24:10.457 But we stayed out a little too late and maybe we should have realized this was 00:24:10.457 --> 00:24:13.577 going to be a trend that we needed to get a handle on but at this point we were 00:24:13.577 --> 00:24:15.397 still naive and thought this would be the only late night. 00:24:15.957 --> 00:24:17.297 Planet Nick's day two you. 00:24:18.113 --> 00:24:22.333 Thankfully, it starts a little later today. We're still at the Airbnb, if you can't tell. 00:24:23.013 --> 00:24:27.173 Moving a little slow. Yeah, and I think I talked a little too hard yesterday. 00:24:27.453 --> 00:24:32.633 Yeah. Too much socialized. Lunch was good, and then we went and just talked 00:24:32.633 --> 00:24:37.453 and talked and talked, and then we went to the after party and talked and talked. 00:24:37.933 --> 00:24:40.493 Yeah, how you doing? You all right? Yeah, actually. Well, I didn't have to walk 00:24:40.493 --> 00:24:44.373 as much as you guys because I got a piggyback ride. That's true. That is actually true. 00:24:44.553 --> 00:24:45.733 Yeah, and it was a good piggyback ride. 00:24:45.833 --> 00:24:46.333 Too. It was, yeah. 00:24:46.873 --> 00:24:48.373 I saw pictures. it was legit. 00:24:48.893 --> 00:24:50.373 I got some free sunglasses. 00:24:50.833 --> 00:24:54.193 That's true. Yeah, so it was, 00:24:54.813 --> 00:24:59.493 really, in a way, Day 2 had the big heavy hitter keynote because it's the State 00:24:59.493 --> 00:25:03.853 of the Union for Nix OS which we're always kind of interested because you tend 00:25:03.853 --> 00:25:09.453 to get a few, at least one or two tidbits that you kind of hang on to for the rest of the year. 00:25:09.713 --> 00:25:15.433 You can hear the hubbub of Planet Nix. The secure Nix State of the Union is about to kick off. 00:25:16.093 --> 00:25:18.233 It should be pretty interesting. It's going to be both a State of the Union 00:25:18.233 --> 00:25:22.153 and a talk about security in one, which probably should go together more often than they do. 00:25:22.713 --> 00:25:25.633 And there's already people with laptops. Look at all those people, 00:25:25.693 --> 00:25:28.933 actually, with laptops open getting work done. I guess they didn't party as 00:25:28.933 --> 00:25:29.893 hard as we did last night. 00:25:30.333 --> 00:25:33.553 Yeah, people getting work done is a big trend at Planet Nix. 00:25:33.833 --> 00:25:38.653 And once the keynote started, Ron touched on something that I think was on a lot of people's mind. 00:25:39.073 --> 00:25:43.113 Bottom line, at the end of the day, the order and solution is Nix. 00:25:44.217 --> 00:25:49.257 Nix comes in and, simply put, because everyone in this room knows that it's 00:25:49.257 --> 00:25:56.537 a lot more complex than that, unifies the baseline architecture of a few steps of infrastructure. 00:25:57.937 --> 00:26:01.957 Now, simply put, what happens when you have one step instead of five? 00:26:01.977 --> 00:26:03.857 What happens when you have one step instead of six? 00:26:04.477 --> 00:26:08.237 It's just faster. It doesn't matter what's happening on the top. 00:26:08.237 --> 00:26:12.517 It just means that whatever that solution ends up being, that solution is able 00:26:12.517 --> 00:26:18.917 to be more reproducible, more secure, more deterministic, and in general, fast. 00:26:19.777 --> 00:26:25.037 And we always listen for some of the financial details because last year at 00:26:25.037 --> 00:26:28.977 the first planet Nix, Ron said that the goal was to have two years of runway. 00:26:29.057 --> 00:26:32.517 If any sponsor were to pull out, they want to have two years of runway to run 00:26:32.517 --> 00:26:33.657 the NixOS infrastructure. 00:26:34.157 --> 00:26:38.257 And what was the number? He said their costs are going up by $25,000 a day. 00:26:38.457 --> 00:26:40.317 Is that what he said for the S3 cash? 00:26:40.477 --> 00:26:41.797 I think it was gigabytes. 00:26:42.177 --> 00:26:42.457 Gigabytes. 00:26:42.737 --> 00:26:42.997 25. 00:26:43.657 --> 00:26:45.737 It was 200 gigabytes a day. 00:26:45.857 --> 00:26:47.277 200 gigabytes a day increase? 00:26:47.497 --> 00:26:48.177 Yes. Okay. 00:26:48.337 --> 00:26:52.517 And I think he said it's costing them $25,000 a day on average right now. 00:26:52.617 --> 00:26:55.677 So you can imagine that expense. That's just for the NixOS cash. 00:26:55.857 --> 00:26:58.657 And so they really want to get to a point where they have two years of runtime. 00:26:58.957 --> 00:27:01.517 And Ron gave it to us straight. They're not quite there yet. 00:27:01.517 --> 00:27:05.257 We like to keep finances very, very transparent. That's what we do. 00:27:05.417 --> 00:27:06.697 We started this a few years ago. 00:27:07.277 --> 00:27:10.937 Everyone here will have access to the slides and you'd be able to see. 00:27:11.637 --> 00:27:15.617 Currently, the foundation still does not receive enough funds to be able to 00:27:15.617 --> 00:27:19.137 do what I think is the sustainable, I would say, P&L. 00:27:19.697 --> 00:27:23.677 I want the foundation to be in a position that we can fund and continue the 00:27:23.677 --> 00:27:28.517 infrastructure and keeping the lights on for at least two to three years without relying on anyone. 00:27:29.309 --> 00:27:33.489 And, um, unfortunately we're not there yet. I think some good progress has been made. 00:27:33.709 --> 00:27:36.309 Um, they, I don't think the numbers are quite final, right? But, 00:27:36.409 --> 00:27:39.549 uh, we did get a little preview on some of the slides here and, 00:27:39.729 --> 00:27:43.049 you know, donations, there's nearly 46,000 euros a year in donations, 00:27:43.149 --> 00:27:46.069 a bunch from open collective also on NixCon 2025. 00:27:46.249 --> 00:27:50.049 It looks like they were able to make some money, uh, nearly 33 grand, right? 00:27:50.209 --> 00:27:56.429 Always good when incomes, uh, more than expenses, but still down estimated profit 00:27:56.429 --> 00:28:00.229 of negative 24 000 euros it's. 00:28:00.229 --> 00:28:03.409 Amazing it's only that low actually when you think about the cost but it makes 00:28:03.409 --> 00:28:05.189 it unsustainable and they don't have that two-year runway. 00:28:05.189 --> 00:28:07.889 And ron did a great job calling out the talk some of the sponsors right like 00:28:07.889 --> 00:28:11.549 working with fastly working with aws um corporations are willing to help because 00:28:11.549 --> 00:28:12.909 they know the the stuff is good and. 00:28:12.909 --> 00:28:16.109 They seem to be getting a little bit more and more the conversations are beginning 00:28:16.109 --> 00:28:18.089 to be elite uh more productive. 00:28:18.089 --> 00:28:20.589 Yeah i think right everyone acknowledges were not there yet, 00:28:20.789 --> 00:28:24.389 but I think maybe that's just me, my read, but it felt like Ron was maybe a 00:28:24.389 --> 00:28:26.929 little less existentially freaked out about it. 00:28:27.589 --> 00:28:28.789 Yeah, it was a good State of the Union. 00:28:28.989 --> 00:28:30.389 All right, so the State of the Union 00:28:30.389 --> 00:28:33.949 was good, and I want to capture some impressions. What'd you think, Wes? 00:28:34.449 --> 00:28:38.409 You know, I think Ron did a good job of bringing up a lot of different people 00:28:38.409 --> 00:28:40.789 from the community, and you get a sense that, like, 00:28:41.429 --> 00:28:44.889 we have a steering committee now, we have a constitution, we have a foundation, 00:28:44.889 --> 00:28:48.649 and some of those people seem pretty professional, and they do professional 00:28:48.649 --> 00:28:49.809 jobs, but at the same time, 00:28:49.889 --> 00:28:53.569 they're all wearing many hats, and you kind of get a sense of what it takes 00:28:53.569 --> 00:28:58.429 to keep something as diffuse and large and exciting as Nix and NixOS hanging together. 00:28:58.669 --> 00:29:02.549 Yeah, and some of it's still new. They're still experimenting a bit with the structure. 00:29:03.009 --> 00:29:06.709 Yeah, I noticed the steering committee, they're getting their legs. 00:29:06.709 --> 00:29:10.689 It felt like they still are getting their sea legs on what it is like to be 00:29:10.689 --> 00:29:17.129 a public-facing thing that is responsible for shepherding the needs and wants 00:29:17.129 --> 00:29:18.529 of a community, that's not easy. 00:29:20.687 --> 00:29:26.887 Human interaction. As part of a steering committee, you have to take human needs 00:29:26.887 --> 00:29:30.487 and deal with those, and at the same time, the other five hats they're wearing. 00:29:30.707 --> 00:29:33.707 So kudos to them for getting up there, and it does seem new. 00:29:33.907 --> 00:29:38.187 It does, you can sense that it's new, but something I think Nick sorely needs. 00:29:38.867 --> 00:29:46.067 So I'm hoping that they get a little strength. I think they could use a little inner strength. 00:29:46.467 --> 00:29:49.647 Yeah, I think, well, don't you think, as they have to make more decisions as 00:29:49.647 --> 00:29:51.847 time goes on, right? Because the steering committee is like the committee of 00:29:51.847 --> 00:29:54.747 last resort when nobody knows what direction to go. 00:29:54.927 --> 00:29:57.727 Because they don't want it to be authoritarian dictatorship. 00:29:58.407 --> 00:30:01.907 Because then, no matter how benevolent a dictator is, it's still a benevolent 00:30:01.907 --> 00:30:06.167 dictator. So you want to move to something that seems to be on more sure footing. 00:30:06.607 --> 00:30:10.507 Where you get multiple opinions about what the right way to go forward is. 00:30:10.647 --> 00:30:16.887 And it is the correct model. I think it's just a muscle that hasn't been exercised. So you sense that. 00:30:18.200 --> 00:30:23.100 And speaking of opinions on how the thing went and forming those, what were yours? 00:30:23.720 --> 00:30:29.080 I noticed just a lot of passion. Like every team member that was there got to 00:30:29.080 --> 00:30:33.360 dive into their niche and be able to say, hey, I'm really interested in infrastructure, 00:30:33.360 --> 00:30:35.760 and that's where I'm going to do my work. 00:30:35.880 --> 00:30:39.900 And there's enough people with different interests that it seems to be going 00:30:39.900 --> 00:30:41.820 in a positive direction. That was what I noticed. 00:30:42.080 --> 00:30:45.180 Yeah, I mean, we have a formatting team, right? It's like, that spells it out. 00:30:45.180 --> 00:30:50.000 And you could actually sense when a person was done talking up there, 00:30:50.060 --> 00:30:53.600 how much the people in the room, and I imagine the people who are watching this, 00:30:53.880 --> 00:30:55.640 appreciate what they're doing. 00:30:56.180 --> 00:31:02.380 It's kind of like everybody was so glad they were doing those CUDA things that are difficult. 00:31:02.580 --> 00:31:06.800 And you really get a sense of gratitude from this community because they know 00:31:06.800 --> 00:31:10.280 that Nix makes things so much easier than it would be without it. 00:31:10.380 --> 00:31:12.440 And then without those people, it would be infinitely harder. 00:31:12.640 --> 00:31:13.660 So I felt that gratitude. 00:31:14.540 --> 00:31:18.320 Yeah, I think there's like a certain amount of excitement. We know that this 00:31:18.320 --> 00:31:21.980 community can have a lot of, let's say, active discussions on things. 00:31:22.080 --> 00:31:24.220 And maybe it feels like between, you know, flake stability. 00:31:24.440 --> 00:31:27.180 There's just a lot of outstanding efforts because it is a giant scope. 00:31:27.600 --> 00:31:31.540 And I think it's easy to see, you know, hot discussions on the discourse or 00:31:31.540 --> 00:31:33.600 feel like things are stagnant and not necessarily moving. 00:31:33.840 --> 00:31:38.580 And they get updated in a very diverse way. So it's cool to have all of those updates in one spot. 00:31:38.700 --> 00:31:41.880 Because it turns out, even if you're not closely watching, there is a lot of 00:31:41.880 --> 00:31:45.960 work happening. Yeah, I think the other takeaway I had is a lot of it's being done in the public. 00:31:46.180 --> 00:31:48.980 You just have to know where to look. And that was one of the things that we 00:31:48.980 --> 00:31:50.400 got was resources of where to look. 00:31:50.820 --> 00:31:54.660 Now, you had a chance to sit down with Ron a little bit after his keynote. 00:31:55.340 --> 00:31:57.400 And I grabbed a couple of moments from that. 00:31:57.560 --> 00:32:00.160 For humans, it was like, what is this? 00:32:01.370 --> 00:32:05.590 Like, am I looking at the ingredients of, like, my, I don't know, 00:32:05.750 --> 00:32:06.950 my, like, multivitamin? 00:32:07.050 --> 00:32:11.930 Like, why do I need to know what type of B12 complex is included in this, right? 00:32:12.010 --> 00:32:15.750 So, for us, I mean, for a subset of us, the folks that were here for a long 00:32:15.750 --> 00:32:16.770 time, they're like, yeah, that's cool. 00:32:16.870 --> 00:32:20.430 I want to know, like, you know, that the B12 complex came from exactly that 00:32:20.430 --> 00:32:23.630 extractor, that, like, laboratory function that made it happen. 00:32:23.990 --> 00:32:28.090 But I think that's also what's making it so in tune with agents, 00:32:28.530 --> 00:32:32.950 right? Because Nix was one of the purest ways to communicate with computers that you can find. 00:32:33.530 --> 00:32:36.610 And guess what? Now, agents, the computers are able to do more. 00:32:36.750 --> 00:32:40.730 So I don't think it's surprising that the thing that can communicate the most 00:32:40.730 --> 00:32:42.890 with the computer and is the most, I would say, 00:32:43.010 --> 00:32:45.790 perpendicular to the way that computers are thinking in a descriptive format 00:32:45.790 --> 00:32:51.230 is able to be utilized and pretty much leveraged very effectively by those very 00:32:51.230 --> 00:32:52.790 computers now that they have more capabilities. 00:32:53.050 --> 00:32:57.190 I'm going to tell it to you straight, audience. I'm not hyping this up. 00:32:57.190 --> 00:32:59.570 I'm not making it up at all. 00:32:59.650 --> 00:33:03.650 The people that we talk to that are really heavy hitters that are in industry 00:33:03.650 --> 00:33:04.670 that are building things. 00:33:05.949 --> 00:33:09.329 Every single one of them is working with agents. 00:33:09.529 --> 00:33:09.689 Yes. 00:33:10.209 --> 00:33:12.789 Every single one of them. None of them were like, oh, we're not doing that. 00:33:13.209 --> 00:33:16.089 And then there were people that were like, oh, the gal we talked to from France. 00:33:16.269 --> 00:33:16.629 Roxanne. 00:33:16.869 --> 00:33:18.609 Yeah. She's like, that's her whole business. 00:33:18.929 --> 00:33:19.289 Fisher with a C. 00:33:19.469 --> 00:33:20.449 Yeah. Fisher with a C. Roxanne Fisher. 00:33:20.449 --> 00:33:23.409 There was some diversity. Some folks are using it more or less or for different 00:33:23.409 --> 00:33:25.789 things. But yeah, that was a consistent name. 00:33:25.789 --> 00:33:27.089 I mean, that was her whole focus. 00:33:27.229 --> 00:33:27.429 Yeah. 00:33:27.869 --> 00:33:29.929 Agents. Like her company's whole focus. 00:33:30.069 --> 00:33:33.669 But my point here, too, is even the companies where that's not their focus, 00:33:33.889 --> 00:33:37.449 where they're where they're delivering other things, they are all experimenting with agents too. 00:33:37.709 --> 00:33:41.669 Now, I'm not saying 100% of people. I'm saying the people that are established 00:33:41.669 --> 00:33:45.569 heavy hitters that build things you use every day that are out there shipping 00:33:45.569 --> 00:33:52.609 things to end users, 100% of them are messing with stuff and building agents. 00:33:52.749 --> 00:33:56.229 And some of them are seemingly building harnesses and test suites, 00:33:56.689 --> 00:34:00.129 including, sounds like even Ron and the folks over at Phlox. 00:34:00.509 --> 00:34:03.569 We've been able to build up something that I'm hoping to open source completely 00:34:03.569 --> 00:34:05.529 unrelated to Phlox inside of Phlox. 00:34:05.709 --> 00:34:09.949 We pretty much made a mandate inside of the firm around AI utilization. 00:34:11.478 --> 00:34:17.138 three, four months ago, and pretty much we built an engine that is able to take 00:34:17.138 --> 00:34:21.998 all of our context from across the entire bit, especially with Nix, 00:34:22.298 --> 00:34:27.358 and almost be like a sidekick assistant to everything we deploy and actually deploy things for us. 00:34:27.598 --> 00:34:31.618 So I think our engineering pace inside of the company, and I was a skeptic, 00:34:31.738 --> 00:34:35.718 I was like, no, no, no, this Nix stuff, it's going to take way longer for us 00:34:35.718 --> 00:34:40.098 to actually be able to automate any part of that build through the roof. 00:34:40.218 --> 00:34:44.258 Like We're seeing about 10 to 20x across the band on engineering output, 00:34:44.518 --> 00:34:46.518 velocity, product response. 00:34:47.078 --> 00:34:49.598 Customers are coming in with ask. We're able to split that up, 00:34:49.898 --> 00:34:52.818 put that into roadmaps at a pace that I've never seen before. 00:34:53.138 --> 00:34:56.458 So I guess it's kind of related to Flex, but I'm more personally excited about 00:34:56.458 --> 00:34:59.938 that because I think that's going to set the entire industry standard across 00:34:59.938 --> 00:35:03.478 the band of what we as humans are able to ship out and create value around. 00:35:04.098 --> 00:35:06.458 Totally game-changing for the market. 00:35:06.678 --> 00:35:10.358 He said a 10x increase in some of what they've been able to deliver. 00:35:10.478 --> 00:35:10.998 That's crazy. 00:35:11.198 --> 00:35:11.338 Yeah. 00:35:11.658 --> 00:35:12.638 And these are still early days. 00:35:13.756 --> 00:35:16.096 And he's so excited about it, he wants to release it. 00:35:16.236 --> 00:35:17.676 They're all so pumped about it. 00:35:17.756 --> 00:35:17.936 Yeah. 00:35:18.056 --> 00:35:22.656 Yeah. One of the conversations we were having before the event is it's going 00:35:22.656 --> 00:35:25.956 to be really interesting to see if we walk in and this is a crowd of skeptics. 00:35:26.416 --> 00:35:29.836 This is a Linux community. It could be a total AI skeptic crowd or if it's going 00:35:29.836 --> 00:35:31.536 to be a crowd of enthusiasts. 00:35:32.576 --> 00:35:35.776 And it wasn't 100% alignment on any particular issue. 00:35:35.876 --> 00:35:38.416 And there's a lot of people that are grappling with some of the wider issues 00:35:38.416 --> 00:35:42.196 that we often discuss when it comes to AI and big tech AI in particular. 00:35:42.196 --> 00:35:44.676 We definitely heard that from our audience there and folks we're talking to. 00:35:45.076 --> 00:35:49.776 But I mean, it was overwhelmingly, they are working on these things. 00:35:49.916 --> 00:35:53.056 But like anything that is this important and this impactful, 00:35:53.356 --> 00:35:55.416 it's people working through it in real time. 00:35:55.776 --> 00:35:58.756 It's humans being like, oh, I think this is, oh wait, no, that's not. 00:35:58.916 --> 00:36:02.576 But maybe, and they're literally grappling with something brand new that has 00:36:02.576 --> 00:36:04.836 long-term consequences in real time. 00:36:04.876 --> 00:36:07.376 And that's how we work it out. We need to talk about this stuff. 00:36:07.496 --> 00:36:12.116 We need to just say, look, let's not boogeyman this. let's just rationalize 00:36:12.116 --> 00:36:15.316 that there are good there are bad there's all kinds in this in this. 00:36:15.316 --> 00:36:19.196 It also didn't feel like humans were excluded right like i i didn't you know 00:36:19.196 --> 00:36:22.076 we heard about that from flox but flox was just bringing on new team members 00:36:22.076 --> 00:36:24.936 who were kicking butt helping the event go and you know everyone's talking about 00:36:24.936 --> 00:36:28.616 jackie how um you know how it's sort of enabling them to do more with like a 00:36:28.616 --> 00:36:31.376 distributed team that they have that you know now they can do more with these 00:36:31.376 --> 00:36:34.056 tools so it's interesting to watch. 00:36:34.831 --> 00:36:38.691 I have to also say as an observation, I didn't hear anyone talking about how 00:36:38.691 --> 00:36:41.411 any of this would lead to employment loss. 00:36:41.651 --> 00:36:46.091 If anything, there was like just this buzzing about new ideas for new projects 00:36:46.091 --> 00:36:48.071 and new ways to solve problems. 00:36:48.411 --> 00:36:53.071 Well, you know, I think the biggest trend was is, you know, Bob's totally tapped out. 00:36:53.251 --> 00:36:56.771 He's been so busy. And so he took three days and he built this system and now 00:36:56.771 --> 00:36:59.551 Bob's got so much free time. Like we just kept hearing that over and over again. 00:36:59.771 --> 00:37:02.351 Or so much new excitement for the problems he's tackling. 00:37:02.851 --> 00:37:06.811 I had a chance to sit down with Graham from Determinant Systems, 00:37:06.991 --> 00:37:08.231 and he's the CEO over there. 00:37:08.411 --> 00:37:10.571 I'm sure I'll say that in the clip here that I'm about to play, 00:37:10.631 --> 00:37:13.991 but they also have been working on this stuff. 00:37:13.991 --> 00:37:16.431 And I won't say too much because I don't know how much he wanted public, 00:37:16.471 --> 00:37:22.051 but they have figured out a way that just directly enables them to really ramp 00:37:22.051 --> 00:37:26.291 up how fast they're building on ARM systems and testing on embedded ARM systems 00:37:26.291 --> 00:37:31.291 and coming up for solutions for some of the limitations on totally bootstrapping 00:37:31.291 --> 00:37:33.971 a ARM device that maybe doesn't have device drivers. 00:37:33.971 --> 00:37:38.951 And it needs a complete strapping from the base to all the way up to booting Nix OS. 00:37:39.231 --> 00:37:42.231 And they are working on these things. And it's just really awesome. 00:37:42.231 --> 00:37:45.831 So I had a great chat with Graham, and it was good to see them there at Planet Nix. 00:37:46.331 --> 00:37:50.211 Walking around, we found Graham Christensen, CEO of Determinant Systems. 00:37:50.371 --> 00:37:54.771 We're sitting down chatting, and I got wind of some new software releases, 00:37:54.971 --> 00:37:59.091 a couple of big things, including a new version of the Determinant Systems Nix installer. 00:37:59.291 --> 00:38:01.351 I think it was version 3.17. Is that right? 00:38:01.911 --> 00:38:05.651 Yep, 3.17. We actually just finished rolling it out yesterday. 00:38:05.991 --> 00:38:09.951 It ships a whole bunch of new stuff. So, like, last year at Planet Nix, 00:38:09.991 --> 00:38:13.171 we talked about Flake Schemas. This is the first release that actually contains 00:38:13.171 --> 00:38:14.331 Flake Schema's first party. 00:38:15.071 --> 00:38:17.751 We've just shipped some providence improvements, which is huge. 00:38:17.931 --> 00:38:22.331 And so what that does for you is it lets you bundle and query a store path itself 00:38:22.331 --> 00:38:24.231 to see what the license is for that software. 00:38:25.071 --> 00:38:30.171 We're using it to attach providence data, like CPE matches, so you can see what 00:38:30.171 --> 00:38:31.611 vulnerabilities apply to it. 00:38:31.871 --> 00:38:35.611 Instead of just a rough package name, package version, which a lot of SBOM tools 00:38:35.611 --> 00:38:37.071 do, this is a very precise application. 00:38:37.656 --> 00:38:41.756 Tons more. We shipped Wasm support. That's a collab we did with Shopify, 00:38:41.756 --> 00:38:45.156 and it's huge. We got a sick Mandelbrot demo. 00:38:45.856 --> 00:38:49.036 Yeah, we got to see it. It's a little bit of art, actually. But then you were 00:38:49.036 --> 00:38:52.156 also mentioning there's secure packages now, and that seems like a big deal. 00:38:52.216 --> 00:38:52.836 Can you tell me about that? 00:38:53.196 --> 00:38:58.836 Yeah, and it's really been surprising, the turnaround and what that's done for us and our users. 00:39:00.016 --> 00:39:04.436 So basically, we take Nix packages, and we've picked out a subset of it, 00:39:04.436 --> 00:39:08.616 which is like, essentially we promise to cover everything our customers depend on. 00:39:08.716 --> 00:39:11.756 But it is not all of NICs packages. It's huge. You couldn't do it. 00:39:12.236 --> 00:39:15.216 But so what we do is for everything that's in our covered set, 00:39:15.436 --> 00:39:18.656 we continuously monitor it for CVEs. We reevaluate that twice a day. 00:39:18.956 --> 00:39:22.876 We're actually working to annotate all of NICs packages with CPE identifiers 00:39:22.876 --> 00:39:25.196 so we can be really complete on that front. 00:39:25.416 --> 00:39:29.276 And then have an SLA on shipping vulnerability fixes for it. 00:39:29.696 --> 00:39:34.716 Really the goal here is to, Well, we've had customers over the years that said 00:39:34.716 --> 00:39:39.176 we ship Nix, we use Nix packages in critical spots. We want to keep using it. 00:39:39.456 --> 00:39:43.496 Our auditors are kind of scared of it. And this is really helping a lot of places 00:39:43.496 --> 00:39:46.276 to take advantage of it and be happier and comfortable with it. 00:39:46.756 --> 00:39:49.556 Yeah, as someone who's had to do compliance work at past jobs, 00:39:49.716 --> 00:39:52.296 like that's either stuff that you get told no because it doesn't exist, 00:39:52.456 --> 00:39:55.596 or you maybe have willing management who's willing to let you build it in-house. 00:39:55.796 --> 00:39:58.476 And then you're probably going to build just a worse version that no one else 00:39:58.476 --> 00:40:00.176 gets to use. So it seems a lot better. 00:40:01.759 --> 00:40:05.559 That's actually, so we had one of our customers come to us and say, 00:40:05.799 --> 00:40:09.699 you know, our auditor flagged this concern that like Nix packages, 00:40:09.899 --> 00:40:13.979 they're not, you know, it's not the most like professional, like ultra, 00:40:14.019 --> 00:40:15.639 like conservative ecosystem. 00:40:16.059 --> 00:40:19.799 Is this safe at all? Like, is it actually safe at all to depend on this? 00:40:19.919 --> 00:40:24.059 And the work we're doing with the CPEs and the Providence and secure packages 00:40:24.059 --> 00:40:29.119 is letting them continue to serve their customers with Nix and NixOS. It's really exciting. 00:40:29.539 --> 00:40:33.899 Graham, I think maybe listeners that just use Nix privately in their home lab, 00:40:34.019 --> 00:40:37.419 they might be surprised to hear there's demand for enterprise features like 00:40:37.419 --> 00:40:40.899 this, because this sounds like stuff maybe RHEL has to incorporate for compliance purposes. 00:40:41.299 --> 00:40:44.299 And one of the things I picked up on here is there's a surprising amount of 00:40:44.299 --> 00:40:46.619 people that are using Nix in the workplace. 00:40:47.279 --> 00:40:51.199 But maybe they're not even disclosing it to their management, 00:40:51.199 --> 00:40:54.899 or they're not talking about it, they're not public about it. 00:40:55.139 --> 00:40:56.999 Are you seeing that at Determinant Systems too? 00:40:57.499 --> 00:41:02.139 Yeah. So something we've noticed over the last 10 years or so is the people 00:41:02.139 --> 00:41:05.019 we're talking to are getting older and more mature. 00:41:05.619 --> 00:41:10.299 In the beginning of consulting, so often we would talk to somebody who was fresh 00:41:10.299 --> 00:41:12.899 in their career and like, I want to get Nix at work. 00:41:12.979 --> 00:41:15.239 And I was like, cool. I hope you succeed. 00:41:15.559 --> 00:41:19.479 And now we're starting to talk more to people who have been working at that 00:41:19.479 --> 00:41:23.519 company for 20 years. And I'm saying, Nix is the next step for us and we want your help to do it. 00:41:23.799 --> 00:41:27.559 And so that's happening. but also we're hearing from, you know, 00:41:27.799 --> 00:41:31.139 people who've been building on top of RHEL for their entire career. 00:41:31.359 --> 00:41:36.159 And now they're shipping and using Nix and school systems and the government 00:41:36.159 --> 00:41:40.379 agencies and state agencies and healthcare tech, like, and not just, 00:41:40.399 --> 00:41:43.479 not just upstarts, but like major established players. 00:41:43.659 --> 00:41:47.339 They're truly, I would say Nix is approximately everywhere. They just haven't, 00:41:47.379 --> 00:41:48.279 they just haven't told you yet. 00:41:49.580 --> 00:41:53.480 I'm just curious, do you have any insights or your own theories around why folks 00:41:53.480 --> 00:41:56.800 are so reluctant sometimes, either individually or corporations, 00:41:57.060 --> 00:41:58.960 companies, to talk about Nix as a tool? 00:41:59.560 --> 00:42:01.440 This one's a little touchier, a little spicier. 00:42:02.600 --> 00:42:05.100 And I think we're sort of getting over this as an ecosystem. 00:42:05.400 --> 00:42:10.860 But at least over the past several years, there's been an explosion and growth of usage in Nix. 00:42:11.000 --> 00:42:13.700 And I think the community was kind of jarred and upset by that. 00:42:13.860 --> 00:42:17.260 And so a little skepticism and a little fear about talking about the places 00:42:17.260 --> 00:42:21.040 that they're being used in agencies that are not popular, right? 00:42:21.200 --> 00:42:25.100 And not exciting or like clearly like a moral good. 00:42:25.300 --> 00:42:29.040 You know, those agencies exist and they're using NICs too. And they don't want 00:42:29.040 --> 00:42:30.560 to be like get in trouble, right? 00:42:31.000 --> 00:42:38.340 But they are using NICs. And personally, I feel like they exist and if they're 00:42:38.340 --> 00:42:39.840 going to use software, which they are, 00:42:40.160 --> 00:42:45.380 they are better served by software that does the stuff correctly versus might 00:42:45.380 --> 00:42:47.700 be working incorrectly or less reliably. 00:42:48.220 --> 00:42:49.840 I don't know. It's complicated. 00:42:50.660 --> 00:42:54.860 Yeah, I think we see this a lot in Linux and open sources. It's sort of a sign of success, right? 00:42:54.940 --> 00:42:58.860 Like early adopters maybe, oh, it's pure, it's functional, all the nerdy details, 00:42:58.900 --> 00:43:00.500 or maybe it's the open source philosophy. 00:43:00.740 --> 00:43:03.740 And then you get to the point where it's just so clearly a useful tool that 00:43:03.740 --> 00:43:05.140 the pragmatists come, right? 00:43:05.200 --> 00:43:09.040 They don't need to buy into all those other layers. They just see that it builds a better Docker. 00:43:10.023 --> 00:43:15.323 Yeah, I mean, that's totally true. I mean, that's the history of basically technology, right? 00:43:15.503 --> 00:43:21.623 Starting with ARPA and DARPA, like it has always been very, so much of like 00:43:21.623 --> 00:43:26.163 the advanced technology we have today comes from the military, right? 00:43:26.263 --> 00:43:32.123 And the military being willing to spend heavily and invest heavily in making this tech better. 00:43:32.643 --> 00:43:35.923 And I've talked about this before. I have complicated feelings about the military, 00:43:36.283 --> 00:43:41.963 but I feel like it's better for them to run Nix than for them to run something else. 00:43:41.983 --> 00:43:44.623 So I am excited to help them do that. 00:43:45.143 --> 00:43:47.963 I think I sleep a little better at night knowing that they're running Nix OS, 00:43:48.083 --> 00:43:49.483 or more and more so, at least. 00:43:49.963 --> 00:43:53.043 So one of the other, I'd say, big trends, and I'm surprised it hasn't come up 00:43:53.043 --> 00:43:55.743 yet in our conversation, it seems to be here, is a lot of people are talking 00:43:55.743 --> 00:43:57.783 about AI and agents in particular. 00:43:58.243 --> 00:44:02.283 Are you seeing much of that materialize from a customer demand or anything like that, or workload? 00:44:02.883 --> 00:44:05.703 Yeah, so actually this is something we've started doing internally. 00:44:06.043 --> 00:44:11.643 So we've been working on a real small operating system to run on embedded targets, 00:44:11.683 --> 00:44:15.143 and we're going to talk about that more another time, but one of the things 00:44:15.143 --> 00:44:19.503 we've been able to do with that is we have a hardware test lab, and it's pretty neat. 00:44:19.603 --> 00:44:22.843 One thing we can do is we've instrumented the hardware, and we can say, 00:44:23.143 --> 00:44:27.003 you know, hey, Claude, this is how you talk to the hardware. Here's the Nix flake. 00:44:27.283 --> 00:44:31.283 Go make it work, right? And then it can do Nix build. it can go reboot the device 00:44:31.283 --> 00:44:35.323 and see if it booted successfully, and take that input and create basically 00:44:35.323 --> 00:44:37.223 the board support until it can boot. 00:44:37.483 --> 00:44:40.783 And you know, that's like an extreme example, but it's really compelling. 00:44:41.543 --> 00:44:48.623 And actually, this is something that, totally tangential, but the rise of LLMs 00:44:48.623 --> 00:44:51.583 has made it so much easier for people to engage with and learn Nix. 00:44:53.373 --> 00:44:58.033 I feel pretty vindicated. It's been scary over the last... Well, 00:44:58.093 --> 00:44:59.273 I don't really want to say scary. 00:44:59.633 --> 00:45:04.873 But anyway, so it's been hard to think about the adoption and the education process for Nix users. 00:45:05.193 --> 00:45:09.633 And maybe it is right to try and wrap Nix up and make the interface easier and 00:45:09.633 --> 00:45:11.913 sort of hide Nix behind it. 00:45:12.013 --> 00:45:16.093 But the rise of LLMs and the ability for people to more quickly adapt and get 00:45:16.093 --> 00:45:19.973 used to Nix, I feel completely vindicated about the right investment for us 00:45:19.973 --> 00:45:22.533 is Nix itself. That is a great point. Right. 00:45:23.093 --> 00:45:26.093 And I've just amongst ourselves seen it too in the audience as well. 00:45:26.533 --> 00:45:29.293 Graham, it's always great to catch up with you. Where should we send people 00:45:29.293 --> 00:45:31.553 to check out like the new installer and stuff like that? 00:45:31.973 --> 00:45:35.533 Yeah. So we've published all our things on determinate.systems. 00:45:35.793 --> 00:45:37.573 And that's probably the best place to get started. 00:45:38.413 --> 00:45:40.353 Thank you. Thank you, sir. Yeah. 00:45:40.793 --> 00:45:44.433 You know, we really couldn't fit everybody in, but it is the friends you make 00:45:44.433 --> 00:45:46.673 along the way. But you were pointing out there is one group that would have 00:45:46.673 --> 00:45:48.093 been particularly great to get on the show. 00:45:48.353 --> 00:45:52.373 Yeah. I mean, Anthropic had nine representatives there, which when you think 00:45:52.373 --> 00:45:56.953 about, well, you were talking about how the salary for watching one presentation 00:45:56.953 --> 00:46:00.213 that was, it was like a $60,000 presentation that was an hour long. 00:46:00.353 --> 00:46:06.153 And for them to think this is important enough to go down to the basement with 00:46:06.153 --> 00:46:08.573 the rest of us to send nine people down there. 00:46:08.653 --> 00:46:11.473 Giving talks and they had a big booth there on the floor. 00:46:11.753 --> 00:46:15.553 That speaks to what they think is important. And Nix is a big part of their infrastructure. 00:46:15.933 --> 00:46:18.353 And last year, Anish gave a talk. 00:46:18.726 --> 00:46:22.886 And that was new. He was working with Nix, but he said, oh, yeah, 00:46:22.946 --> 00:46:23.986 that was my first talk ever. 00:46:24.366 --> 00:46:28.246 And then he came back this year as almost like an ambassador for Nix, 00:46:28.346 --> 00:46:32.466 but also an ambassador for bringing Anthropic folk into the ecosystem. 00:46:32.846 --> 00:46:35.446 Yeah, he's a good shepherd for his team. And Nish is a great guy. 00:46:36.186 --> 00:46:39.286 And, you know, we hear so much about Anthropic from the outside. 00:46:39.286 --> 00:46:42.866 But then when you see the people that are building, they're engineers and they're 00:46:42.866 --> 00:46:46.006 nerds and they're solving big problems that people haven't solved before. 00:46:46.006 --> 00:46:50.066 and they're then upstreaming some of those solutions and then they're bringing 00:46:50.066 --> 00:46:53.486 that knowledge to these events and telling people in a couple of years you're 00:46:53.486 --> 00:46:56.506 going to need to solve this problem here's how you could start thinking about it today. 00:46:56.506 --> 00:47:01.566 And they're very intent on getting rid of those little roadblocks that maybe 00:47:01.566 --> 00:47:05.586 nix has or that they've encountered because they're on the cutting edge they're 00:47:05.586 --> 00:47:09.406 going to hit it first and they're very intent on being like hey hey let's let's 00:47:09.406 --> 00:47:12.566 fix this let's make sure this isn't a roadblock which makes it better for all of us you. 00:47:12.566 --> 00:47:15.106 Know that was one of the things we talked we thought we kind of would see that 00:47:15.106 --> 00:47:18.366 theme of Planet Nix this year, but I think it pans out. 00:47:18.446 --> 00:47:21.866 You see just people from all diverse set of companies, workshopping ideas from 00:47:21.866 --> 00:47:25.246 like small firms who are just adopting it, getting to pick from the brains of 00:47:25.246 --> 00:47:27.746 folks who've maybe been doing it at scale for a couple of years now. 00:47:27.906 --> 00:47:31.006 It's really fun to be a fly on the wall to watch that happening. 00:47:31.206 --> 00:47:35.546 It's such a positive feedback loop, it seems. And some places don't have that. 00:47:35.686 --> 00:47:39.526 I think sometimes certain technologies engender bringing the best out of people. 00:47:39.706 --> 00:47:42.946 And it seems that Nix tends to do that out of really smart people. 00:47:42.946 --> 00:47:46.346 From all different backgrounds. Solving huge problems or just even just... 00:47:46.346 --> 00:47:49.166 Or like, you know, folks who do it who've been longtime community members because 00:47:49.166 --> 00:47:51.766 they love open source and they love Nix, meeting people who maybe just aren't 00:47:51.766 --> 00:47:54.566 learning it because they joined a company who's already using it. 00:47:55.226 --> 00:48:01.246 So that was, you know, part one, Planet Nix. And then it was shift gears and you go to scale. 00:48:01.406 --> 00:48:04.346 So Planet Nix is an intimate setting, you know, six, seven hundred people now. 00:48:05.326 --> 00:48:10.306 Last year, maybe 300 people. Scale is a whole other deal. 5,000, 00:48:10.306 --> 00:48:15.166 6,000, 7,000 people, it's the largest Linux event in North America. 00:48:18.072 --> 00:48:20.252 We don't have a sponsor, so... 00:48:20.252 --> 00:48:22.612 Can you give us a sponsorship here? 00:48:22.792 --> 00:48:23.292 Yeah, sure. 00:48:23.652 --> 00:48:24.512 Yeah, what do you got for us? 00:48:25.632 --> 00:48:30.072 This is brought to you by Ollivander's Wands. 00:48:30.432 --> 00:48:35.192 As we violate copyright like an LLM, just sucking up random novels, 00:48:36.392 --> 00:48:39.372 Linux Unplugged is an incredible, wonderful show. 00:48:39.552 --> 00:48:42.412 It is magical. We can make things happen for you. 00:48:42.512 --> 00:48:45.912 Wait, can I have a request? What if the show is brought to you by Chris's Chase 00:48:45.912 --> 00:48:50.992 Reserve credit card? Chris's Chase Reserve credit card. That's what the show is brought to you by. 00:48:51.432 --> 00:48:57.112 In that case, it's my precious likes to spend his money. Yes. 00:48:57.632 --> 00:49:01.332 Your bank account is empty and I's in negatives. Yes. 00:49:01.612 --> 00:49:02.192 That's about it. 00:49:02.652 --> 00:49:07.472 Or when you reach the limit, it's you shall not run this card. 00:49:11.463 --> 00:49:16.503 So it was time to get ourselves put together, stop partying so hard, 00:49:16.543 --> 00:49:20.263 and go to the Southern California Linux Expo. 00:49:20.523 --> 00:49:24.143 Well, this is actually day one for scale. We start all over again. 00:49:24.143 --> 00:49:25.523 Boys, it's a whole new event. 00:49:25.743 --> 00:49:28.203 We poked our head in some of the rooms. Everybody's sitting down. 00:49:28.403 --> 00:49:30.363 But we haven't yet gone to the Expo Hall. 00:49:30.983 --> 00:49:35.243 That's usually where a lot's going on. Yeah, you get to see who took the time 00:49:35.243 --> 00:49:38.723 to come out, who's trying to sell you something, or show you an exciting project. 00:49:38.723 --> 00:49:41.843 and I suspect we have more than a few friends around in there. 00:49:42.223 --> 00:49:45.943 It's funny because you can hear there's large crowds inside the rooms as you 00:49:45.943 --> 00:49:50.143 walk by. They're like little honey hives of buzzing activity. Yeah, yeah. 00:49:50.543 --> 00:49:54.923 So everybody's in sessions right now, so I think this is our moment to go check out the Expo Hall. 00:49:55.263 --> 00:49:59.363 There's also a special energy on that first talk that kicks off. 00:49:59.443 --> 00:50:01.983 So as we're going to the Expo Hall, we've got to poke our heads in. 00:50:02.423 --> 00:50:07.023 And the EFF, the executive officer of the EFF, was actually giving the keynote 00:50:07.023 --> 00:50:09.863 talk, and I didn't really understand it first, but I learned very quickly that 00:50:09.863 --> 00:50:12.883 the EFF has been one of the longest supporters of Scale. 00:50:13.503 --> 00:50:16.203 Elan sort of kicks everything off, and then we meet her. 00:50:16.663 --> 00:50:23.243 Welcome to the 23rd Annual Scale Conference. We're excited to have you all with us again. Thank you. 00:50:30.548 --> 00:50:33.628 And so without further ado, I'm going to let Cindy take the stage. 00:50:33.828 --> 00:50:35.428 Thank you. And we'll have some opportunity. 00:50:36.528 --> 00:50:41.628 Thank you so much. Thank you to Scale for inviting me. This is my first time 00:50:41.628 --> 00:50:43.008 at Scale. I'm so excited. 00:50:43.508 --> 00:50:49.048 You know, EFF and the open source community, it's a long-term relationship we're in, friends. 00:50:49.368 --> 00:50:51.748 I thought that was actually a fair point. And she actually had a pretty good 00:50:51.748 --> 00:50:55.088 keynote. And there is definitely some fans in the audience. 00:50:55.448 --> 00:50:59.828 So what we do is we vary strategically. We pop in, we get a sense, 00:50:59.888 --> 00:51:02.488 and then we keep going because while the people are in the keynote, 00:51:02.688 --> 00:51:05.788 that's the time to check out the expo hall, especially if you have a microphone 00:51:05.788 --> 00:51:07.868 and you want to have any shot of being able to hear somebody. 00:51:08.268 --> 00:51:12.488 And we walked in and you were hit with some fancy booths right away. 00:51:12.808 --> 00:51:17.368 Well, I can confirm the expo hall is definitely buzzing. 00:51:18.236 --> 00:51:21.796 Big Microsoft presence right there at the entrance with Microsoft and GitHub. 00:51:22.656 --> 00:51:26.336 Hello. Hello. I'm here. Thank you so much. Thank you. Thank you. 00:51:26.616 --> 00:51:29.056 Okay. So there's a lot of people. All right. Here's the Ubuntu booth. 00:51:29.156 --> 00:51:30.696 Oh, they got some music going on. 00:51:31.216 --> 00:51:34.136 Framework's got a booth right next to Ubuntu. Then RISC-V does as well. 00:51:35.056 --> 00:51:39.036 This is a good group right here. I'm seeing some AI and science combination 00:51:39.036 --> 00:51:42.496 over here. Yeah. Okay. All right. We've got to just soak this in. 00:51:42.616 --> 00:51:49.676 A lot of those. But I think what we sort of honed in on is the vibe around RISC-V this year. 00:51:49.856 --> 00:51:53.736 It was really good. And people were excited. 00:51:53.936 --> 00:51:56.016 They had like a double booth set up. 00:51:56.676 --> 00:52:00.196 And Brent had a chance to talk with the Deep Compute CEO briefly. 00:52:00.216 --> 00:52:03.976 And he is working on an entire RISC-V stack. 00:52:04.176 --> 00:52:07.316 I mean, everything from the CPU to the soundboard and Ethernet. 00:52:07.496 --> 00:52:12.376 I'm from Deep Computing. and the founder of Deep Computing and specialized to 00:52:12.376 --> 00:52:15.996 make all the RISC-V consumer electronics. 00:52:16.236 --> 00:52:18.916 And he's spending like $2 million a year to try to make it happen, 00:52:18.916 --> 00:52:20.936 to build these things and try to make this a reality. 00:52:21.116 --> 00:52:23.776 And using a bunch of nicks under the hood to do it. 00:52:23.896 --> 00:52:26.736 Yeah, and so is Carl. And Carl was a different Carl. 00:52:26.956 --> 00:52:29.476 He's one of the community, he's the community representative. 00:52:29.476 --> 00:52:32.776 And he really, really knows his stuff. 00:52:33.136 --> 00:52:36.416 And he had all kinds of cool hardware, you guys. I mean, stuff you would want 00:52:36.416 --> 00:52:39.176 to own right now. and of course you had to tell us a little bit about it. 00:52:39.316 --> 00:52:42.176 Thank you, hi, my name is Carl. I am with Works with RISC-V. 00:52:42.376 --> 00:52:47.876 We're open community ecosystem for getting things built with the RISC-V instruction 00:52:47.876 --> 00:52:49.616 set architecture, both hardware and software. 00:52:50.343 --> 00:52:54.203 And then you have a table of gadgets here. Give us a tour. So this is using 00:52:54.203 --> 00:52:56.103 a newer processor. Came out about a year ago. 00:52:56.403 --> 00:53:00.343 It's very fast. Operates about Raspberry Pi 3 to 4 level performance. 00:53:01.203 --> 00:53:06.123 Unfortunately, does not have vector. And then this stuff here in the middle is the future. 00:53:06.363 --> 00:53:11.203 This will be available about 3 to 4 months. It is running the Spacetime K3 processor. 00:53:12.203 --> 00:53:16.103 This uses the RVA23 standard, so it will work with the new version of Ubuntu, 00:53:16.283 --> 00:53:17.463 which is across the way from us. 00:53:17.823 --> 00:53:21.963 It is fast. It has vector. it has hypervisor, and it has AI cores. 00:53:22.423 --> 00:53:26.983 I notice this in more of a server case as well. Yeah, so you have options of how to do it. 00:53:27.083 --> 00:53:30.743 The way that these products are gonna be sold when they're available is they 00:53:30.743 --> 00:53:31.643 come in two form factors. 00:53:31.823 --> 00:53:37.943 One is this Jetson Orion NX compatible SOM, and then the other one is a Pyco 00:53:37.943 --> 00:53:41.343 ITX board that kind of fits in like a NUC-like mini PC. 00:53:41.903 --> 00:53:46.843 Because of the SOMs, someone has built this ridiculous example where you can 00:53:46.843 --> 00:53:49.023 shove 10 of them into a 1U rack mount case. 00:53:50.061 --> 00:53:52.921 So we're going to work on something that's a little more flexible than this 00:53:52.921 --> 00:53:55.161 one because we actually want to be able to have not just these, 00:53:55.301 --> 00:53:56.461 but other architecture systems. 00:53:56.721 --> 00:54:00.881 And then we're making build farms available to the general open source community 00:54:00.881 --> 00:54:02.901 for getting stuff built on RISC-V. 00:54:03.161 --> 00:54:04.121 That's cool, right? 00:54:04.321 --> 00:54:06.761 Oh, boy. They are doing a lot with RISC-V these days. 00:54:06.981 --> 00:54:10.141 That's right. I mean, the build farm for the community access for developers 00:54:10.141 --> 00:54:12.721 that are doing open source projects is a really nice touch. 00:54:12.861 --> 00:54:15.141 Yeah, because, you know, people don't have access to this yet. 00:54:15.241 --> 00:54:16.681 But that NUC-sized machine. 00:54:16.761 --> 00:54:17.181 Oh, wow. 00:54:17.441 --> 00:54:17.781 Oh. 00:54:18.121 --> 00:54:22.201 And I love the fact that they're doing almost everything around SBC and that 00:54:22.201 --> 00:54:25.921 the one you had 10. They were like, well, if we're going to make a big server, we'll just put 10. 00:54:26.041 --> 00:54:26.801 Slut in more. 00:54:27.161 --> 00:54:29.961 Yeah. And they were kind of laughing like the deep compute CEO. He's kind of laughing. 00:54:30.061 --> 00:54:32.361 He's like, you know, these companies that are investing these billions of dollars 00:54:32.361 --> 00:54:35.961 on these proprietary SOCs, we're coming for him. We're going to come for him 00:54:35.961 --> 00:54:37.901 with an entire open stack top to bottom. 00:54:38.041 --> 00:54:40.681 He might have been one of the most excited and driven people I've ever met. 00:54:40.861 --> 00:54:41.541 No kidding. 00:54:41.701 --> 00:54:43.921 Yeah. And every time you saw him, he was buoyant. 00:54:44.041 --> 00:54:44.241 Yeah. 00:54:44.381 --> 00:54:48.821 Which I loved. I'm like, this guy's been deep in these trenches of a seemingly insurmountable thing. 00:54:49.041 --> 00:54:51.081 Not when people look at that from the outside. They're like, 00:54:51.241 --> 00:54:54.761 that's a hill that's really hard to climb. And he's not deterred by that. 00:54:54.861 --> 00:54:57.641 No, he was also bringing around gadgets because he was also at Planet Nix. 00:54:57.701 --> 00:55:02.261 And he had like a little risk tiny computer in a transparent box. 00:55:02.681 --> 00:55:05.501 Yeah, it was so neat. The whole thing. And it really makes you feel like something's 00:55:05.501 --> 00:55:08.261 happening because, of course, across from him is the Ubuntu booth. 00:55:08.341 --> 00:55:10.941 And they're working to make sure that they get more risk five support. 00:55:11.581 --> 00:55:14.481 And it's just a general sense around. The only thing that we never really touch 00:55:14.481 --> 00:55:16.681 on, but it's kind of neat to see, it's kind of become tradition, 00:55:17.041 --> 00:55:22.821 is the NextCloud and the OpenSUSE folks and the KDE people. 00:55:22.941 --> 00:55:26.821 They kind of have like an area and they all kind of help each other and support 00:55:26.821 --> 00:55:29.141 each other's booths and stuff like that. It's always fun to go over and chat with them. 00:55:29.141 --> 00:55:30.621 Good open source vibes over there. 00:55:30.761 --> 00:55:35.921 Yeah, yeah. They were really cool people. And I love the fact that KDE has a 00:55:35.921 --> 00:55:38.421 presence at something like this. Like they're, you know... 00:55:39.480 --> 00:55:44.140 out there hanging out. Some volunteer from the community from this area decided 00:55:44.140 --> 00:55:47.380 to set up a booth and that they have their representations. 00:55:47.680 --> 00:55:51.260 And they're dedicating time to that. Serious time. We, you know, 00:55:51.340 --> 00:55:55.000 we, so it's always really great to have easy conversations with these really 00:55:55.000 --> 00:55:56.180 community focused folks. 00:55:56.480 --> 00:56:00.280 It was a little trickier to try to get some of the bigger names on the record. 00:56:00.320 --> 00:56:03.340 And we thought we don't normally talk to them, but we thought it would be great 00:56:03.340 --> 00:56:08.360 to get Microsoft or meta or GitHub on the record just to kind of flush out the 00:56:08.360 --> 00:56:11.480 scale coverage for you all and this would be the point where i'd be playing 00:56:11.480 --> 00:56:13.160 conversations with those companies but. 00:56:13.160 --> 00:56:14.200 So chris how did that go. 00:56:14.200 --> 00:56:17.400 Well how did it go brent let's see hey you were the one that had to take the 00:56:17.400 --> 00:56:18.940 rejection directly a few times your. 00:56:18.940 --> 00:56:20.000 Feedback on this experience. 00:56:20.000 --> 00:56:23.080 Have you had that happen before because i'm pretty used to it because every 00:56:23.080 --> 00:56:24.900 time i go to these events they say no. 00:56:24.900 --> 00:56:29.380 Well i have a little bit of a secret superpower which is that like i seem to 00:56:29.380 --> 00:56:31.820 be able to charm almost anybody in that conversation was gonna. 00:56:31.820 --> 00:56:33.080 Do it it would probably be you oh yeah. 00:56:33.080 --> 00:56:35.880 And sometimes they're like you know uncomfortable about it and it's 00:56:35.880 --> 00:56:38.580 like no no it's gonna be it's five minutes it's only five minutes that's all 00:56:38.580 --> 00:56:41.220 i need uh i was i actually lost a little 00:56:41.220 --> 00:56:43.900 bit of sleep over this i was shut down in a 00:56:43.900 --> 00:56:47.500 way uh specifically by microsoft too 00:56:47.500 --> 00:56:51.360 that they just didn't want to make time for us and i 00:56:51.360 --> 00:56:54.360 approached them very kindly and said hey like you're here you 00:56:54.360 --> 00:56:57.280 have the fanciest booth right at the the door of the exhibition 00:56:57.280 --> 00:57:01.800 hall you have the director of open source here that seems like a perfect fit 00:57:01.800 --> 00:57:05.340 we'd love to have a conversation why you're excited to be here at scale what 00:57:05.340 --> 00:57:09.720 you're doing with open source these days and uh they were just more interested 00:57:09.720 --> 00:57:13.080 in the raffle that they were going to give and didn't want to make time to have 00:57:13.080 --> 00:57:14.600 a five minute chat with us you. 00:57:14.600 --> 00:57:18.640 Are so rational i love the way that you approached it you're like why wouldn't 00:57:18.640 --> 00:57:22.300 they want to and like i've been in hollywood for a long time and i could see 00:57:22.300 --> 00:57:23.360 them giving you the brush off. 00:57:23.360 --> 00:57:23.780 And every. 00:57:23.780 --> 00:57:26.780 Time you'd come back to me and tell me what their excuse was in the back of 00:57:26.780 --> 00:57:29.800 my mind i'm like this is a brush off this has happened to my friend. 00:57:29.800 --> 00:57:31.900 I am so sad why bother show up. 00:57:32.974 --> 00:57:33.794 That's a good question. 00:57:33.954 --> 00:57:34.834 Yeah, it's... 00:57:34.834 --> 00:57:35.634 Because they're a raffle, mate. 00:57:35.854 --> 00:57:40.254 Well, they're there, you know, to chat with the community, which is good. 00:57:40.534 --> 00:57:41.334 It's a raffle. 00:57:41.514 --> 00:57:45.694 They're very afraid of media exposure going wrong. And so they're very, very safe. 00:57:45.934 --> 00:57:48.554 And of course, to us, it seems silly. Because you're there, you've spent, 00:57:48.554 --> 00:57:50.994 you know, 20 grand on the booth. 00:57:51.174 --> 00:57:55.474 You've probably spent 60 grand in payroll because you've got seven, eight people here. 00:57:55.594 --> 00:57:56.914 Which is a pack of peanuts to Microsoft. 00:57:57.154 --> 00:58:00.814 Yeah. And then you're going to spend 15 grand on the hotel stay for all of those 00:58:00.814 --> 00:58:03.194 people. You're going to spend another 15 grand on the flight. 00:58:03.374 --> 00:58:05.694 And then we come here and we're like, so you're going to talk to, 00:58:05.694 --> 00:58:07.214 you know, a thousand people today. 00:58:07.474 --> 00:58:09.774 Would you like to talk to another 60,000 people today? 00:58:10.074 --> 00:58:13.054 And they say no. And to us, it just doesn't, it just doesn't compute. 00:58:13.234 --> 00:58:17.734 Why would you say no to additional exposure? You've put all this money and effort to get exposure. 00:58:17.874 --> 00:58:19.374 They could have read off a document. 00:58:20.067 --> 00:58:24.667 They could have prepared some form of thing that they could just read to us. 00:58:24.787 --> 00:58:26.147 And to be clear, it's not just Microsoft. 00:58:26.147 --> 00:58:28.567 It's fine. I just want to hear what they want to say. 00:58:28.827 --> 00:58:33.727 But for them, it's too dangerous. And so they don't want to do it. 00:58:33.787 --> 00:58:37.487 But it's funny because it doesn't come across as being part of the community, right? 00:58:37.607 --> 00:58:41.887 Because the booths that are actually part of the community are thrilled to talk to us. 00:58:42.087 --> 00:58:44.847 You can tell. It has the anti-community stank on it. 00:58:44.987 --> 00:58:45.227 Yeah. 00:58:45.367 --> 00:58:48.427 And you may get to a certain... But let's compare. 00:58:49.347 --> 00:58:52.707 Anthropic has a lot to lose by saying the wrong thing. Everyone here at this 00:58:52.707 --> 00:58:55.267 table agrees that, especially with everything that's going on with them. 00:58:55.687 --> 00:59:00.427 But yet, Anish came right up to us, warm, welcome, hi, right? So that's different. 00:59:01.087 --> 00:59:03.507 Those companies are on par in the AI space. 00:59:03.887 --> 00:59:04.247 Yeah. 00:59:04.407 --> 00:59:06.647 And the behavior was different. That's all you have to say. 00:59:07.047 --> 00:59:10.007 Yeah. And so it means we don't have those clips. We did get to chat with them. 00:59:10.187 --> 00:59:12.647 You know, I'll tell you one of the things I observed that I thought was interesting 00:59:12.647 --> 00:59:16.367 is the Meta booth had a whole bunch of Asahi machines. They had about the whole 00:59:16.367 --> 00:59:21.827 lineup of Asahi workstations. you can possibly have, including a Mac studio at the Meta booth. 00:59:22.047 --> 00:59:24.847 And it seems that there is some Asahi Linux usage in Meta. 00:59:25.087 --> 00:59:29.287 How much we can't really say, but it's neat. And they were happy to be pointing 00:59:29.287 --> 00:59:31.227 that out to a degree and talk about it off the record. 00:59:31.367 --> 00:59:36.567 I really wanted them to just open up and chat with us about what was going on inside. 00:59:36.807 --> 00:59:39.847 Yeah, and we're not trying to do a hit piece. Like, hey, I see you're using 00:59:39.847 --> 00:59:41.367 Asahi Linux. You're excited? 00:59:41.487 --> 00:59:41.927 We're excited. 00:59:42.047 --> 00:59:44.187 We want to celebrate open source. That's our jam. 00:59:44.287 --> 00:59:51.247 At one table, you You have Linux and meta and Apple. That's actually pretty freaking cool. 00:59:51.327 --> 00:59:52.887 It is an interesting combination. Yeah. 00:59:52.927 --> 00:59:53.367 That's a soup. 00:59:53.547 --> 00:59:57.167 And the answer would have been, yeah, you know, a dozen or so or more of our 00:59:57.167 --> 01:00:00.267 developers or however many it is are using it on the workstations. 01:00:00.267 --> 01:00:03.087 That would have been, that would have been the extent of the big reveal. 01:00:03.307 --> 01:00:03.547 Yeah. 01:00:03.847 --> 01:00:05.127 It is what it is. 01:00:05.127 --> 01:00:07.067 They were doing it there anyway. 01:00:07.467 --> 01:00:10.727 And then, you know, I guess what I, what I, if, if, 01:00:11.079 --> 01:00:14.759 honestly they're probably going to be the lesser of the interesting chats 01:00:14.759 --> 01:00:17.619 we would have had i just wanted to round it out and i 01:00:17.619 --> 01:00:20.479 just thought it would be good to represent them and the other thing is is they are making 01:00:20.479 --> 01:00:23.239 you know by coming there and spending that money they're not 01:00:23.239 --> 01:00:26.599 only adding a bit of legitimacy to the roster when you look at it but they're 01:00:26.599 --> 01:00:30.399 also you know helping pay the bill a little bit to make scale possible and so 01:00:30.399 --> 01:00:34.179 i'm grateful they're there and i just wish that this was something that could 01:00:34.179 --> 01:00:36.959 be sorted out but i'll tell you this is not a new thing this has been an issue 01:00:36.959 --> 01:00:40.819 since we've been doing linux action show it's just it's media stuff and now there are, 01:00:41.479 --> 01:00:44.339 more of these companies at these events than there were before. 01:00:44.479 --> 01:00:46.779 It might've been one or two companies before, and now there's like a dozen, 01:00:47.039 --> 01:00:48.879 you know, and they spend big money. 01:00:49.099 --> 01:00:51.899 And so I guess in one way, I'm glad they're there. I just wish they'd talk to us. 01:00:52.619 --> 01:00:56.159 But we can, you know, we can relay that they are there and they do seem to be 01:00:56.159 --> 01:01:00.219 interested in hiring still. And they do seem to be trying to show off what they're working on. 01:01:00.339 --> 01:01:02.939 Yeah, about the message we got was that they had a sticker that said Microsoft 01:01:02.939 --> 01:01:05.819 loves open source. So, you know, make of that what you will. 01:01:08.199 --> 01:01:08.519 Sticky? 01:01:12.799 --> 01:01:14.839 Jason, can you do an ad about your new laptop? 01:01:15.319 --> 01:01:15.859 Oh, sure. 01:01:16.899 --> 01:01:18.299 Can you do it in Grimm's voice, though? 01:01:18.519 --> 01:01:25.339 This is the T480. It's quite a laptop. I got it for $289, and I decided I'd run NexOS on it. 01:01:25.659 --> 01:01:29.479 Had no issues at all installing it. Just stuck that USB stick in, 01:01:29.579 --> 01:01:32.519 booted it right up, and this thing is flying like the wind. 01:01:32.879 --> 01:01:36.159 As a matter of fact, it might be a little quieter than Brent's framework laptop. 01:01:36.399 --> 01:01:38.739 Oh, it's only a Gen 1. That's fine. I'm not throwing you under the bus. 01:01:39.439 --> 01:01:44.599 There it goes. and I'm running Plasma on here, Plasma 6. Just a beautiful next 01:01:44.599 --> 01:01:45.779 config, nice and simple. 01:01:46.019 --> 01:01:47.999 This thing is golden toast. 01:01:48.619 --> 01:01:49.519 Now I want one. 01:01:50.119 --> 01:01:53.559 And by the way, listen to Linux Unplugged and support them and become a member if you're not. 01:01:57.479 --> 01:02:02.219 Well, gentlemen, we have some boost to get to because this was a audience-supported 01:02:02.219 --> 01:02:04.359 episode. So what do you say? Should we kick it off? 01:02:09.631 --> 01:02:14.191 All right. And I believe our first, is this a new booster? 01:02:14.411 --> 01:02:14.931 Oh, wow. 01:02:15.591 --> 01:02:18.411 Biggles? Have we heard from Biggles before? I love that name. 01:02:18.491 --> 01:02:19.451 So I feel like it would stick with me. 01:02:19.591 --> 01:02:21.391 Maybe a new booster. 01:02:21.691 --> 01:02:26.411 Biggles is our baller booster with 111 and 111 sets. 01:02:26.871 --> 01:02:28.451 Hey, rich lobster. 01:02:30.791 --> 01:02:34.831 He says, have fun at scale and planet Nix. Here's a small contribution to the 01:02:34.831 --> 01:02:36.391 cost. Not so small. We really do. 01:02:36.511 --> 01:02:37.971 You are our baller. 01:02:38.151 --> 01:02:39.131 We do very much appreciate that. 01:02:39.131 --> 01:02:41.231 You make me want to be a better man. 01:02:41.531 --> 01:02:44.791 Thank you very much, Biggles. And sent that in from Fountain. 01:02:45.411 --> 01:02:50.651 Well, K.R. Hill 94 comes in with 100,000, Zad. 01:02:50.671 --> 01:02:52.371 Well, that's a baller booster as well. 01:02:52.751 --> 01:02:53.311 Ah, fly! 01:02:53.611 --> 01:02:55.231 There you go. Get a big old duck. 01:02:55.271 --> 01:02:56.571 You're doing a good job. 01:02:57.131 --> 01:03:01.491 For the trip, I assume the full 100,000 made it through, yet still not sure 01:03:01.491 --> 01:03:03.291 about that note channel in Albie Hub. 01:03:03.691 --> 01:03:04.131 Brent! 01:03:04.491 --> 01:03:06.611 Oh, I'm sorry. I delegated that one to Jeff. 01:03:06.811 --> 01:03:10.931 That's okay. He delegated to Fountain and Fountain looked into it for us and 01:03:10.931 --> 01:03:11.791 determined it was Brent's fault. 01:03:12.480 --> 01:03:15.540 Let's see if that is the official fact of my team. At least we have the answer. 01:03:17.360 --> 01:03:20.580 We're doing a lot with agents these days. Whose fault is it? 01:03:21.740 --> 01:03:22.580 Hashtag Blank Brant. 01:03:23.400 --> 01:03:23.880 I love it. 01:03:24.000 --> 01:03:26.380 We did get a boost from our dear adversaries. 01:03:26.740 --> 01:03:27.300 Hey, adversaries. 01:03:29.440 --> 01:03:30.340 67,514 Satoshis. 01:03:31.200 --> 01:03:32.480 Stay a while and listen. 01:03:33.000 --> 01:03:35.180 This simply says advertiser boost. 01:03:35.440 --> 01:03:36.020 Appetiser boost. 01:03:36.320 --> 01:03:37.220 Oh, I'm so sorry. 01:03:37.500 --> 01:03:38.380 Appetiser, advertiser. 01:03:38.380 --> 01:03:39.160 Why are you pulling it, Chris, over there? 01:03:39.540 --> 01:03:41.460 This advertisement's full of Froot Loops. 01:03:42.040 --> 01:03:45.880 Listen, in the lack of advertising, this might as well be an advertising boost. 01:03:46.000 --> 01:03:48.020 Thank you very much, adversaries. Really appreciate that. 01:03:48.080 --> 01:03:49.260 How were the appetizers? Good? 01:03:49.560 --> 01:03:54.460 Well, I ate the fries. Actually, the fries were good. They had a garlic olive oil on them. 01:03:54.520 --> 01:03:56.460 I ordered mine with garlic aioli on the side. 01:03:56.460 --> 01:03:58.400 That was quite good. That was good. That was good. 01:03:58.860 --> 01:03:59.440 Fair good. 01:03:59.580 --> 01:04:02.460 And then, you know, the next best thing I had, actually, wasn't anything we bought there. 01:04:02.640 --> 01:04:06.040 It was Emma brought me, she wanted to know my thoughts. She brought me a little 01:04:06.040 --> 01:04:07.920 bit of brisket from like a barbecue place nearby. 01:04:08.780 --> 01:04:13.680 And I'll say- I have multiple people bringing you strange meat at this conference. How do I use that? 01:04:14.120 --> 01:04:17.200 It's great. I will say it was pretty fantastic. I was also a couple of drinks 01:04:17.200 --> 01:04:18.320 in and I needed that protein. 01:04:19.120 --> 01:04:24.260 Speaking of producer Jeff, PJ comes in with 55,555 stats. 01:04:24.580 --> 01:04:28.640 It's over 9,000. I have no idea when that's appropriate. 01:04:28.840 --> 01:04:33.400 I was scrolling down to see if anybody broke the threshold, but I'm pretty close, right? 01:04:33.860 --> 01:04:36.900 Well, he says this might help feed Brentley for a few hours. 01:04:38.280 --> 01:04:44.260 Well, PJ, we got some pizza from, what was it? Big Toast or Big Slice? 01:04:44.400 --> 01:04:46.560 Fun will now commence It was gluten-free. 01:04:46.600 --> 01:04:49.340 To be clear I was, thank you And before that, we got them some. 01:04:49.340 --> 01:04:52.540 Indian food They threw olives on mine for no reason, which I always appreciate. 01:04:54.394 --> 01:04:57.154 I do, too, actually. You know, who doesn't love a free olive, 01:04:57.254 --> 01:04:58.994 especially when they're usually $7 to add? 01:04:59.254 --> 01:05:01.634 Right. I was like, wow, that's a plus. 01:05:02.194 --> 01:05:05.614 Well, A-A-R-on comes in with 50,000 sets. 01:05:06.154 --> 01:05:09.614 A-A-R-on! Is there an A-A-R-on here? 01:05:10.274 --> 01:05:14.394 Let's be clear. Nobody likes listening to ads. If you say you do, 01:05:14.634 --> 01:05:16.814 I've got a good therapist I can introduce you to. 01:05:16.974 --> 01:05:19.334 Hey, sometimes I like to hate listen. 01:05:20.114 --> 01:05:25.574 That being said, I will listen to thousands of hours of ads if it means that you get paid. 01:05:25.794 --> 01:05:26.474 Oh, that's... 01:05:26.474 --> 01:05:29.094 Even in the membership feed, if need be. 01:05:29.274 --> 01:05:30.554 No, no, that's never happening. 01:05:30.854 --> 01:05:31.254 No, no, no. 01:05:31.314 --> 01:05:35.334 My only ask is that they don't just randomly pop up in the middle of a segment. 01:05:35.474 --> 01:05:35.754 That's fair. 01:05:35.854 --> 01:05:40.074 Yeah, yeah. I think that's pretty fair. I suppose we would have the technology. 01:05:41.214 --> 01:05:45.214 One of the reasons we went with the... Even though they're not great, 01:05:45.374 --> 01:05:46.074 you know, for some people. 01:05:46.174 --> 01:05:48.714 Actually, think, though, if you think about it, just get it out of the way. 01:05:48.794 --> 01:05:51.234 It's kind of nice. But while we don't have any in-show sponsors, 01:05:51.474 --> 01:05:55.154 we have occasionally like a pre-roll that is playing for some people. 01:05:55.294 --> 01:05:56.874 It's the first time I've ever done something like that. 01:05:57.054 --> 01:06:00.454 And what I do kind of like about it is it just kind of gets it out of the way. 01:06:00.514 --> 01:06:03.334 And then we don't put it in the middle of the show. And they just. 01:06:03.454 --> 01:06:04.434 Or I'll just read them all. 01:06:04.874 --> 01:06:06.034 I would listen to that. 01:06:06.034 --> 01:06:06.854 I'll just do your ads. 01:06:07.014 --> 01:06:09.494 But thank you very much, A.A. Ron. We really do appreciate that. 01:06:09.494 --> 01:06:14.874 Thank you, A.A. Ron. Don't mess with me up in here. You give me a boost. I appreciate it. 01:06:16.094 --> 01:06:16.734 Is it my turn? 01:06:16.934 --> 01:06:17.094 Yeah. 01:06:17.674 --> 01:06:19.254 A Monday. A Monday? 01:06:19.674 --> 01:06:20.234 A Monday. 01:06:20.654 --> 01:06:21.074 A Monday. 01:06:21.274 --> 01:06:21.794 A Monday. 01:06:21.834 --> 01:06:22.754 A Monday, Monday. 01:06:23.294 --> 01:06:26.314 Boosted in through fountain 45,000. 84 satoshis. 01:06:27.834 --> 01:06:28.214 Engage. 01:06:28.914 --> 01:06:29.694 Boosty, boost, boost, boost. 01:06:29.734 --> 01:06:30.794 Jason, I think you read this one. 01:06:30.994 --> 01:06:31.594 I do? 01:06:31.874 --> 01:06:32.374 Sure, why not? 01:06:32.454 --> 01:06:34.314 Where's it? Where's it right there? Boosty. Ah. 01:06:35.377 --> 01:06:36.897 Boosty, boost, boost, boost, boost, boost! 01:06:39.037 --> 01:06:42.037 Also, boost number two. But that one... 01:06:42.037 --> 01:06:42.777 Oh, that was... 01:06:42.777 --> 01:06:46.257 I keep getting partial figures boosting via fountain. I'm not sure why you guys 01:06:46.257 --> 01:06:47.237 need to clear that up, over. 01:06:47.357 --> 01:06:48.137 We're all working on it. 01:06:48.397 --> 01:06:50.077 Yeah, we're getting Brent on it. We'll get him fixed. 01:06:50.137 --> 01:06:50.577 It's Brent's fault. 01:06:50.657 --> 01:06:51.677 I'll just randomly read it. 01:06:51.837 --> 01:06:53.137 Well, he's captive now, so... 01:06:53.677 --> 01:06:56.197 Read it in there. I'll read it in their voice. That's what I'll do. 01:06:57.557 --> 01:07:00.957 WLP2SO comes in with 20,604 cents. 01:07:00.997 --> 01:07:03.857 Also known as our favorite Wi-Fi interface. 01:07:03.857 --> 01:07:05.557 You're doing a good job. 01:07:05.697 --> 01:07:08.977 Some value, he says. Thank you. Thank you. Appreciate you. 01:07:09.137 --> 01:07:09.397 Thank you. 01:07:10.137 --> 01:07:12.397 Hey, we got another Adversaries here. Wait, is this Twice from Adversaries? Yeah. 01:07:12.837 --> 01:07:13.357 That is nice. 01:07:13.537 --> 01:07:15.317 That means Wes's script is broken? 01:07:15.997 --> 01:07:17.637 No. Well, one of them... 01:07:17.637 --> 01:07:19.297 It's over 18,000! 01:07:19.597 --> 01:07:22.697 I mean, it is broken, but one of them, the username has at fountain, 01:07:22.797 --> 01:07:25.497 and one of them doesn't. So that's the naive script. 01:07:25.697 --> 01:07:28.357 It did not figure out they were the same. We should strip the fountain stuff, I guess. 01:07:29.217 --> 01:07:32.497 He's commenting on a crash out I had on last week's live stream. 01:07:32.497 --> 01:07:37.417 and points out that Quinn 3 Coder's 30 billion parameter model also runs pretty fast. 01:07:37.577 --> 01:07:41.997 He says, I've had a little run Rust app that I built in just 20 minutes, 01:07:42.017 --> 01:07:44.217 and I mostly use it to write integration and unit tests. 01:07:44.337 --> 01:07:50.257 I'm using the Zed editor's built-in agent and recently added support for OpenAI's 01:07:50.257 --> 01:07:53.597 API-compatible servers for both the prediction and agent use. 01:07:53.597 --> 01:07:56.077 It is interesting because we were chatting about this. I was like, 01:07:56.137 --> 01:08:00.077 hey, where are the smaller models here that you could compartmentalize for jobs 01:08:00.077 --> 01:08:01.397 like coding and things like that? 01:08:01.537 --> 01:08:02.097 Quinn 3. 01:08:02.097 --> 01:08:04.897 Like in this thank you for the shout out for that specifically. 01:08:04.897 --> 01:08:09.017 Quin3 coder also seems to be using devstrel small 2 24b yeah. 01:08:09.017 --> 01:08:09.577 Thank you very. 01:08:09.577 --> 01:08:12.257 Much for those of you who don't know what that means that's actually a formula 01:08:12.257 --> 01:08:13.997 for vitamins that ron mentioned earlier now. 01:08:13.997 --> 01:08:18.077 To be clear he is doing this with 256 gigabytes of ram. 01:08:18.077 --> 01:08:18.997 And an. 01:08:18.997 --> 01:08:20.037 Epic cpu yeah. 01:08:20.037 --> 01:08:22.497 So i can't do that at home not yet not. 01:08:22.497 --> 01:08:23.117 Yet well. 01:08:23.117 --> 01:08:23.997 You know if you ask. 01:08:23.997 --> 01:08:29.437 Adversaries you make me want to be a better man with your video card and your Big Fancy Ram. 01:08:30.318 --> 01:08:34.338 I-Quest comes in with 9,001 sats. It's over 9,000. 01:08:34.438 --> 01:08:38.618 It's over 9,000! I know what your neighbors are thinking as I yell this out 01:08:38.618 --> 01:08:39.798 in this garage, like, repeatedly. 01:08:40.158 --> 01:08:41.218 No, we're just doing sports betting. 01:08:41.298 --> 01:08:43.438 Yeah, they're gonna be like, FBI, open up. 01:08:43.738 --> 01:08:46.598 Yeah. But listen, you finally got to legitimately do it. He says, 01:08:46.678 --> 01:08:49.898 I really appreciate the show. The ad did surprise me, but I'm taking it as a 01:08:49.898 --> 01:08:51.118 sign to do my part to contribute. 01:08:51.358 --> 01:08:52.518 No, see, that's great. Thank you, I-Quest. 01:08:52.558 --> 01:08:52.938 That's nice. 01:08:53.178 --> 01:08:53.798 We appreciate that. 01:08:53.858 --> 01:08:54.498 Especially on this trip. 01:08:54.718 --> 01:08:55.498 It's actually quite sweet. 01:08:56.078 --> 01:08:59.278 Spooky sat, come, Hanson, with 10,000 sats. 01:09:03.664 --> 01:09:07.224 long time listener first time hey really, 01:09:09.944 --> 01:09:16.144 have y'all ever seen or heard of mango wm could be a great replacement to hyperland 01:09:16.144 --> 01:09:19.984 has been my window manager for about four months now and i thought it would 01:09:19.984 --> 01:09:23.504 be worth sharing keep up the great work and we got a little kit hub link here. 01:09:23.504 --> 01:09:27.724 Well you're probably never going to get wayland or i mean a hyperland and wayland 01:09:27.724 --> 01:09:31.024 out of my uh cold dead hands but i do like the name. 01:09:31.024 --> 01:09:31.664 It is Wayland. 01:09:31.784 --> 01:09:35.324 It is Wayland-based. Yep. It's Mango is a great name. 01:09:35.504 --> 01:09:36.004 Very fruity. 01:09:36.344 --> 01:09:36.984 The Mangos. 01:09:37.024 --> 01:09:40.164 It can be built completely within a few seconds. So, you know, 01:09:40.244 --> 01:09:42.644 I know you know how it takes to compile stuff. 01:09:42.744 --> 01:09:45.084 There's a Knicks flake in this repo. 01:09:46.124 --> 01:09:49.524 In that case, I've got the same combination on my luggage. 01:09:49.764 --> 01:09:51.144 I think that's a winner. Thank you. 01:09:51.264 --> 01:09:51.524 Knicks. 01:09:51.624 --> 01:09:52.304 Check out Mango. 01:09:53.104 --> 01:09:55.424 So Ham sent in 5,000 sets. 01:09:56.144 --> 01:09:57.704 Live long and prosper. 01:09:57.884 --> 01:09:59.344 All right. What do we got? What's our boost? 01:10:00.124 --> 01:10:01.744 Oh, right. I'm reading this one. No, I'll do it for you. 01:10:03.024 --> 01:10:06.364 Just wanted to tell the non-members that they are missing out big time with 01:10:06.364 --> 01:10:09.504 this one. I'm talking a certified Chris Red. 01:10:10.284 --> 01:10:13.424 And at this rate, they're going to require age verification for this podcast. 01:10:14.244 --> 01:10:16.104 Oh, snap. I see they snuck it in there. 01:10:16.184 --> 01:10:16.764 That's good. 01:10:16.924 --> 01:10:17.404 Oh, yeah. 01:10:17.564 --> 01:10:18.644 Someone might have gotten excited. 01:10:18.804 --> 01:10:20.004 Yeah, I did get a little worked out. 01:10:20.024 --> 01:10:20.584 That's a good one. 01:10:20.704 --> 01:10:22.664 Every now and then. You know, every now and then. 01:10:22.724 --> 01:10:23.684 I actually really like that. 01:10:23.864 --> 01:10:27.684 As I get older, they happen less and less. So I think it's a good sign now. 01:10:27.864 --> 01:10:30.604 Before, I used to not like it. and now I feel like hey I still got it I still 01:10:30.604 --> 01:10:31.844 got the energy every now and then what. 01:10:31.844 --> 01:10:34.824 I like though is like you're doing them less but when you do do them. 01:10:34.824 --> 01:10:36.324 Are these zingers oh. 01:10:36.324 --> 01:10:37.544 You just lean right in. 01:10:37.544 --> 01:10:40.384 Because like you haven't you know what are. 01:10:40.384 --> 01:10:41.244 They gonna do fire me. 01:10:41.244 --> 01:10:44.484 At this point I got a little steam built up too I suppose yeah yeah, 01:10:45.293 --> 01:10:48.813 Kiwi Bitcoin Guide comes in with 10,000 sats. 01:10:49.473 --> 01:10:54.753 10,000 sats? I don't know. It's possible that is right in the pocket. 01:10:55.173 --> 01:10:58.873 Boost for the road. I'd like to hear your opinion on the open source LLMs. 01:10:59.013 --> 01:11:03.593 If an LLM is open source, how much can we actually know about the security and 01:11:03.593 --> 01:11:04.773 telemetry of that model? 01:11:04.953 --> 01:11:08.513 If the source code is visible and we can audit, can we actually be reassured 01:11:08.513 --> 01:11:10.253 of what's being logged or used for training? 01:11:10.413 --> 01:11:14.273 Or is it just still a black box for us? I am having a debate with a friend at 01:11:14.273 --> 01:11:17.753 the moment, and he says that LMs are complete black boxes, even when they're 01:11:17.753 --> 01:11:20.393 open source. What say you? 01:11:20.593 --> 01:11:24.933 I think the key for this one to start with is just to sort of have some agreed 01:11:24.933 --> 01:11:29.573 defined set of terms, because on open source LMs, that can mean a lot of things. 01:11:29.713 --> 01:11:33.113 And the black box, are you talking about understanding the model and sort of 01:11:33.113 --> 01:11:35.433 like a research and AI theory standpoint? 01:11:35.593 --> 01:11:39.013 Or are you just talking about understanding the harness, and is it running locally, 01:11:39.113 --> 01:11:42.493 and is it talking over the network? And those are all valid things to discuss 01:11:42.493 --> 01:11:43.373 and have different answers. 01:11:43.573 --> 01:11:46.373 And do you object to how they were training is the question. 01:11:46.533 --> 01:11:49.653 So is it really front end what goes into the black box that we're trying to 01:11:49.653 --> 01:11:52.533 discuss? Is that something that you object to, how it was trained? 01:11:52.813 --> 01:11:57.013 Is that developing bias? So, so many perspectives to find. 01:11:57.293 --> 01:12:00.933 And then there's the weights. Are the weights open? Is the data set open? 01:12:01.613 --> 01:12:04.173 And the answer is it's a gradient. 01:12:04.933 --> 01:12:09.313 The different models are a gradient of this right now. And that's not necessarily 01:12:09.313 --> 01:12:11.573 a bad thing. It's a marketplace at the moment. 01:12:11.753 --> 01:12:16.633 And we're actually seeing the ones that are more open, have more longevity and 01:12:16.633 --> 01:12:21.293 more integration into various use cases than I thought was initially maybe expected. 01:12:22.875 --> 01:12:26.415 Six months ago, a lot of us thought this was going to go all locked up to OpenAI 01:12:26.415 --> 01:12:29.215 and Anthropic and maybe a little bit of Microsoft and Google. 01:12:29.575 --> 01:12:32.275 And now it seems more like, yeah, they're still going to be around, 01:12:32.335 --> 01:12:36.415 but these open source models are getting smashed into things like Zed editors 01:12:36.415 --> 01:12:39.015 or little embedded LLMs for people. 01:12:39.335 --> 01:12:42.355 Yeah. And, you know, it's also I think you may have a lot of concerns about 01:12:42.355 --> 01:12:45.695 how are they trained and what went into them and how much control do I have over that. 01:12:45.695 --> 01:12:48.255 But if you do at least have the weights in like a reasonable, 01:12:48.455 --> 01:12:51.995 say, safe tensor style format where it's not like can't execute stuff, 01:12:52.115 --> 01:12:54.795 it's not its own programming language, it's just a specification for weights, 01:12:54.815 --> 01:12:58.075 and you run some sort of open source inference server and harness, 01:12:58.275 --> 01:12:59.655 you do still get the benefits. 01:12:59.655 --> 01:13:03.155 Even if you don't totally have understanding of like why which weights and neurons 01:13:03.155 --> 01:13:06.475 activated and all that kind of stuff, you do at least know you have control 01:13:06.475 --> 01:13:09.675 over what it's running, how it gets sent, what gets sent over the network. 01:13:10.015 --> 01:13:13.455 And it's not going to use any of that data for training. If you lock it down, 01:13:13.615 --> 01:13:14.595 there's no way for it to send it. 01:13:14.815 --> 01:13:15.295 Yeah, exactly. 01:13:15.695 --> 01:13:20.155 There's also not quite a definition of open source AI that's been agreed upon. 01:13:20.435 --> 01:13:24.115 I mean, the open source initiative came out with their definition that a lot of people didn't like. 01:13:24.315 --> 01:13:29.935 So we're not even, you know, saying open source for AI stuff is actually feeling 01:13:29.935 --> 01:13:32.475 quite different than saying open source just for source code. 01:13:33.155 --> 01:13:36.415 We're using words that don't necessarily always mean the same thing in this case. 01:13:36.635 --> 01:13:41.515 So if you have desires for a particular LLM, I think you probably at this point 01:13:41.515 --> 01:13:43.255 need to use more words for what you want to have. 01:13:43.395 --> 01:13:48.275 I'll take the next couple of ones. Anonymous came in with 2021 sats. No message. 01:13:49.135 --> 01:13:49.735 Just a quick one. 01:13:49.815 --> 01:13:53.315 Just a quick one. People who like to mess with computers. 01:13:53.635 --> 01:13:57.275 There you go. Leo, shout out. Okay, Groovy comes in with 2,000 sats. 01:13:57.415 --> 01:13:59.195 Also, no message, just the value, though. 01:14:00.055 --> 01:14:00.575 Let's pull up. 01:14:00.715 --> 01:14:03.275 Let's just help you. It will help us all. I don't know. 01:14:03.315 --> 01:14:04.255 I think we pull Moon Knight up. What do you think, right? 01:14:04.255 --> 01:14:04.675 Sure, go ahead. 01:14:04.815 --> 01:14:06.035 Because one sat below the cutoff. 01:14:06.115 --> 01:14:06.475 I like it. 01:14:06.775 --> 01:14:12.675 Yeah, well, Moon Knight boosts in with 1,999 sats, or an honorary 2,000 sats. I'll chip one in. 01:14:13.155 --> 01:14:17.355 I have to echo the sentiment that the light theme in KDE Plasma is great. 01:14:17.495 --> 01:14:17.615 Yeah. 01:14:17.815 --> 01:14:18.135 Same. 01:14:18.135 --> 01:14:22.575 I'm an always dark mode guy, but I've been running light mode on my system, and I love it. 01:14:22.855 --> 01:14:22.975 Gross. 01:14:23.255 --> 01:14:27.275 1, 2, 3, 4, 5? I got the same combination of my luggage. 01:14:27.555 --> 01:14:28.595 It is really nice. 01:14:28.655 --> 01:14:29.715 There was no 1, 2, 3, 4, 5. 01:14:29.835 --> 01:14:33.135 They're getting close, though. But we do have some stats for you. 01:14:33.155 --> 01:14:34.655 Thank you, everybody who did 01:14:34.655 --> 01:14:37.715 support the show with a boost or streamed those stats as they listened. 01:14:37.975 --> 01:14:42.515 22 of you streamed stats, and collectively, y'all stacked 30,193 stats. 01:14:42.635 --> 01:14:45.935 Thank you very much. When you combine that with our boosters, 01:14:45.955 --> 01:14:51.715 we stacked a grand total of 540,083 sats this episode. 01:14:51.875 --> 01:14:54.715 Thank you, everyone who boosted us. We appreciate you very much. 01:14:55.215 --> 01:14:56.695 You're the best. Whoa. 01:15:02.080 --> 01:15:04.980 Sometimes I just sing that in my car. It's Karate Kid. I'm all over it. 01:15:05.520 --> 01:15:08.960 And, of course, thank you to our members who support us. We appreciate that 01:15:08.960 --> 01:15:11.240 very, very much. And you made the show possible. 01:15:11.660 --> 01:15:15.060 And with that, we'll treat you, we'll regale you, if you will, 01:15:15.180 --> 01:15:17.060 with a pick. Are you ready, boys? 01:15:17.360 --> 01:15:17.880 I'm ready. 01:15:19.060 --> 01:15:23.460 All right. We got two, I believe, for you this week. Two picks to round out the show. 01:15:23.520 --> 01:15:29.020 And the first one is one that Editor Drew sent our way. It is a native Tidal 01:15:29.020 --> 01:15:30.320 desktop client for Linux. 01:15:30.460 --> 01:15:33.280 You know, the Tidal music service. It sounds super good. It's got a modern UI, 01:15:33.500 --> 01:15:36.440 custom themes, and bit perfect lossless audio. 01:15:36.680 --> 01:15:39.540 And I'm talking 192 hertz, boys. 01:15:39.920 --> 01:15:41.300 Delicious. Spicy. 01:15:41.760 --> 01:15:47.400 It's called Sone. S-O-N-E. And if you're like me, like to go with the lossless 01:15:47.400 --> 01:15:49.320 flack with your Tidal, it does that. 01:15:49.460 --> 01:15:55.920 It can bypass pipe wire and pulse audio and claim direct hardware access if you want the full fancy. 01:15:56.160 --> 01:15:56.460 Ooh. 01:15:56.460 --> 01:15:59.980 And it has an autoplay mode to discover and play similar tracks, 01:16:00.020 --> 01:16:05.560 as well as a volume normalization mode with automatic context switching between 01:16:05.560 --> 01:16:07.260 albums and tracks for gain. 01:16:08.557 --> 01:16:12.737 Looks like it's a 57% typescript and 39% rust. 01:16:13.077 --> 01:16:14.417 Just a little rust. 01:16:18.937 --> 01:16:21.257 See, and nobody believed us when we said we had a soundboard guy. 01:16:21.397 --> 01:16:23.297 They didn't believe that Wes hired a soundboard guy. 01:16:23.437 --> 01:16:25.017 And hey, GPL3. 01:16:25.277 --> 01:16:29.537 Yes, yes. Okay, now this one, hard shift. Hard shift, but I want to put it out 01:16:29.537 --> 01:16:32.477 there. I just want to get in your thought space so you know about this. 01:16:32.677 --> 01:16:37.457 It's called QMD. And it's a mini command line search engine for your docs, 01:16:37.617 --> 01:16:40.437 your knowledge bases, your meeting notes, your markdown files, 01:16:40.577 --> 01:16:42.757 whatever it is sitting on your file system like an animal. 01:16:43.057 --> 01:16:48.257 You can now search on it with the QMD on device search engine for everything you need. 01:16:48.737 --> 01:16:53.337 Like for me, I just have a crap ton of old markdown notes and documentation, 01:16:53.337 --> 01:16:56.177 even have some knowledge bases on my file server, things like that. 01:16:56.297 --> 01:17:00.637 You can point QMD at it and it has a BM25 full text. 01:17:00.637 --> 01:17:04.537 There is so much going on here. You take your query, it does a query expansion 01:17:04.537 --> 01:17:07.517 on it, plus the original query, so it has two alternate queries, 01:17:07.577 --> 01:17:12.397 and then it sends that down for both BM25, like keyword search and vector search, 01:17:12.517 --> 01:17:13.737 on all three of those queries. 01:17:13.937 --> 01:17:19.097 And then it has some kind of RFF fusion plus bonus algorithm that sort of tries 01:17:19.097 --> 01:17:23.237 to smartly rank the different results it's going to get before it can synthesize a final answer. 01:17:23.417 --> 01:17:28.137 And you can run it through an LLM for ranking, too, to have a decent rational ranking of the results. 01:17:28.697 --> 01:17:32.417 And it's got a nice CLI that agents can just work with or you can work with 01:17:32.417 --> 01:17:36.217 yourself but it's also got an MCP server and huge shout out for them. 01:17:36.397 --> 01:17:40.297 It does have standard IO support but it's also got streamable HTTP. 01:17:40.737 --> 01:17:41.337 It's about time. 01:17:41.497 --> 01:17:43.397 Wow. You can never have enough QMD. 01:17:43.657 --> 01:17:44.857 That's true. Let's check it out. 01:17:44.957 --> 01:17:47.117 Or is it QMID? It's a little QMID in here. 01:17:47.117 --> 01:17:49.437 I think I've always said it. I've always read QMD. 01:17:49.617 --> 01:17:49.897 Yeah. 01:17:50.478 --> 01:17:53.218 And it looks like it's, yeah, it's using via Node, Lama, CPP, 01:17:53.618 --> 01:17:57.498 GGUF models to do a lot of this stuff. So it just runs locally. 01:17:57.638 --> 01:18:00.178 Runs on your CPU. Ask me how I know. 01:18:00.218 --> 01:18:04.058 You can also specify your own custom embedding model if you want to get tricky with it. 01:18:04.098 --> 01:18:07.558 Or if you need to use a cloud model because you're running on like a Pi or something like that. 01:18:07.698 --> 01:18:10.038 And the fan manufacturers are very glad it runs locally. 01:18:10.218 --> 01:18:14.698 Yeah, that's true. And it's MIT licensed and nice to see. 01:18:14.958 --> 01:18:18.558 Now, Brentley, you want to recommend a tried and true tool that comes in handy 01:18:18.558 --> 01:18:20.298 when you're, say, at an Airbnb with horrible internet. 01:18:20.478 --> 01:18:26.118 It's true. And this is a new, fresh discovery for me. And you all seem to be 01:18:26.118 --> 01:18:28.218 like, oh, yeah, I used this tool like decades ago. 01:18:28.238 --> 01:18:30.118 For about 15, 20 years. 01:18:30.158 --> 01:18:33.218 Yeah. Thanks for telling me. Really appreciate it. 01:18:33.878 --> 01:18:35.938 Friends let friends discover stuff on their own. 01:18:36.038 --> 01:18:38.638 That's right. You know, maybe other people haven't discovered this. 01:18:38.878 --> 01:18:41.938 I hope I'm the only one who hasn't, because then this is useless. 01:18:42.118 --> 01:18:45.278 But I discovered on my own MTR, 01:18:45.278 --> 01:18:48.458 which I was like an animal using ping 01:18:48.458 --> 01:18:51.358 to try to solve these problems and here comes 01:18:51.358 --> 01:18:54.198 mtr it basically combines the functionality of trace 01:18:54.198 --> 01:18:59.658 routing and ping programs all in one network diagnostic tool and it turns out 01:18:59.658 --> 01:19:05.798 it's really nice to use it has colors and stuff and it makes diagnosing these 01:19:05.798 --> 01:19:12.398 kind of let's call them bnb internet issues way easier so how dare you guys not 01:19:12.783 --> 01:19:13.743 Not tell me about this. 01:19:13.983 --> 01:19:14.123 Sorry. 01:19:14.283 --> 01:19:15.023 Look at the colors you had, too. 01:19:15.023 --> 01:19:15.943 You've been friends for so long. 01:19:16.123 --> 01:19:19.903 It's so beautiful. It's old enough that, you know, like some of the primary 01:19:19.903 --> 01:19:23.603 history events was a changeover from the original author to like the current maintainer. 01:19:23.703 --> 01:19:25.863 And that happened in October of 98. 01:19:26.463 --> 01:19:32.803 Yeah. You know, I'm running that right next to my FVWM 95, just so everybody knows what that is. 01:19:32.903 --> 01:19:33.003 Good. 01:19:33.663 --> 01:19:37.423 I love it because I actually saw you run it on my laptop when you first grabbed it. 01:19:37.583 --> 01:19:40.503 And I was like, what is that? And then because you played with it for a while. 01:19:40.503 --> 01:19:42.443 And then I picked up my laptop and saw it running and I was like, 01:19:43.223 --> 01:19:46.123 oh, that's right. That's the good old MTR. That's nice. 01:19:46.503 --> 01:19:47.643 How am I so late to this one? 01:19:47.823 --> 01:19:49.223 You know, there's a GUI version too now. 01:19:49.523 --> 01:19:50.383 You're a young'un. 01:19:51.023 --> 01:19:55.183 Honestly, can we just shout out the scale networking infrastructure team? 01:19:55.403 --> 01:19:56.323 Yeah, one more time, please. 01:19:56.503 --> 01:20:04.243 Because the Wi-Fi was more solid at scale with 6,000 people than the Wi-Fi is in our Airbnbs. 01:20:04.543 --> 01:20:06.903 We probably could have just streamed out of scale. It might have been easier. 01:20:06.903 --> 01:20:10.283 And when you think about it, this is permanent infrastructure at this Airbnb, 01:20:10.543 --> 01:20:12.043 and that was temporary infrastructure. 01:20:12.303 --> 01:20:15.843 It's amazing. It really is a shout out to the infrastructure team at Scales. 01:20:15.903 --> 01:20:18.583 Especially because they have to show up early to like re-take over, 01:20:18.663 --> 01:20:21.303 right? Like a lot of conferences are just using whatever the vendors provided, 01:20:21.383 --> 01:20:23.103 and they have to like tap into the backhaul and all that. 01:20:23.203 --> 01:20:26.363 But they put all their own APs down and rig things up. 01:20:26.583 --> 01:20:29.423 You can see those Ethernet backhauls running along the floor with the leader. 01:20:29.443 --> 01:20:30.783 Duct tape to the wall. Duct tape to the wall. 01:20:31.003 --> 01:20:32.143 I was like, yeah, boys. 01:20:32.343 --> 01:20:32.843 Good for them. 01:20:32.863 --> 01:20:33.943 Those are my kind of fellas. 01:20:34.283 --> 01:20:38.523 Yep. So we have links to these picks and everything else, which wasn't much 01:20:38.523 --> 01:20:43.663 that we can link this week, but that'll all be over at linuxunplugged.com slash 657. 01:20:43.823 --> 01:20:45.023 Or you know what the power move 01:20:45.023 --> 01:20:48.923 is? If you want to shift into power move mode, which I mean, that's... 01:20:48.923 --> 01:20:50.263 I hardly ever use power mode. 01:20:50.403 --> 01:20:54.563 Yeah, be sparing with that. But if you do want to shift into power mode, 01:20:54.703 --> 01:20:56.683 you go over to jupiterbroadcasting.com. 01:20:56.823 --> 01:21:00.023 That's where the OG source is for everything and other fine shows. 01:21:00.203 --> 01:21:03.583 Yeah, and there's a GitHub repo behind it if you want to get access to the raw details. 01:21:03.583 --> 01:21:07.103 Or, you know, make it even better if it's if something about our website bugs 01:21:07.103 --> 01:21:08.743 you, you can take action directly. 01:21:09.343 --> 01:21:12.723 We always do appreciate that. If you want to get involved, too, we have a website team. 01:21:13.662 --> 01:21:17.502 chat on matrix that you can sneakily find and always participate in the conversation 01:21:17.502 --> 01:21:20.742 there and shout out to the web team just keeping it going right keeping it going up. 01:21:20.742 --> 01:21:21.682 With our crazy changes. 01:21:21.682 --> 01:21:23.502 Our web team doesn't skip leg day. 01:21:23.502 --> 01:21:26.502 Yeah and you know that's right that's right and also just appreciate them a 01:21:26.502 --> 01:21:29.282 little bit extra when we're on the road and they're just kind of keeping their 01:21:29.282 --> 01:21:30.962 eye on things and everything's humming along things. 01:21:30.962 --> 01:21:32.282 Always break when we leave home. 01:21:32.282 --> 01:21:35.822 Uh-huh so it's really nice so thank you everybody over at the web team it's 01:21:35.822 --> 01:21:39.122 not a huge team just a couple guys really but they're doing a great job and 01:21:39.122 --> 01:21:41.302 uh an occasional west pain and Occasional brand. 01:21:42.002 --> 01:21:44.242 Mostly I just come in when I have issues. I roll out. 01:21:44.422 --> 01:21:47.342 Usually you made some new change in how you want to specify something. 01:21:47.662 --> 01:21:49.142 I think you meant steamrolling. 01:21:50.142 --> 01:21:53.522 It's usually like, oh, yeah, I published this show with a completely new method. 01:21:53.982 --> 01:21:55.302 You guys don't mind fixing that. 01:21:55.402 --> 01:21:56.502 Do you? Yeah, we fixed that today. 01:21:58.642 --> 01:21:59.362 In 20 minutes. 01:21:59.382 --> 01:21:59.942 And somehow they do. 01:22:00.182 --> 01:22:04.362 Yeah, really. It's really impressive. So I guess I'm feeling when we're on the 01:22:04.362 --> 01:22:06.582 road, I'm really appreciating the people that do the infrastructure out there, 01:22:06.942 --> 01:22:08.702 including the listeners that are involved with their infrastructure. 01:22:08.702 --> 01:22:11.762 So, with that, we will be back in the studio next week. 01:22:15.142 --> 01:22:19.682 Of course, we'll be in the Mumble Room again. That'll be going details at jupiterbroadcasting.com 01:22:19.682 --> 01:22:21.422 and on our website for that. 01:22:21.442 --> 01:22:23.702 Mumble Room is a low-latency opus way to listen. 01:22:23.842 --> 01:22:27.582 And then we have our Matrix Chat, which has not only our live chat, 01:22:27.602 --> 01:22:32.222 but chatting going all throughout the week, sharing geeky details and projects people are working on. 01:22:32.242 --> 01:22:35.202 Yeah, talk about Mestastic or organize a meetup in your area. 01:22:35.322 --> 01:22:35.582 That's right. 01:22:35.582 --> 01:22:36.302 You can do all kinds of stuff. 01:22:36.362 --> 01:22:41.342 See it next time. Same bat. I was actually going to try to do it with the old-timey, 01:22:41.342 --> 01:22:43.222 you know. Same bat station. 01:22:43.442 --> 01:22:43.902 That's not bad. 01:22:44.162 --> 01:22:44.982 I was going to try to get the cup going. 01:22:45.082 --> 01:22:48.362 Also, thank you everybody who came out to Scaler Planet Nix and said hello, 01:22:48.542 --> 01:22:51.982 shook a hand, or just attended. We really, really appreciate it. 01:22:52.022 --> 01:22:54.582 And you made it a blast. And we'll see you all next week. 01:22:54.602 --> 01:22:55.522 Chapter Transcripts! 01:23:38.422 --> 01:23:43.182 Thank you. This mask smells like fart.