WEBVTT 00:00:00.017 --> 00:00:03.857 Hi and welcome to another episode of the Option Plus podcast. 00:00:04.237 --> 00:00:11.217 As you can see, and here, maybe you have noticed that these episodes are quite irregular. 00:00:11.717 --> 00:00:18.657 I'm breaking the common wisdom of, you know, release podcasts regularly at the 00:00:18.657 --> 00:00:25.757 same time, like every day, every Wednesday, or every month, and so on. 00:00:26.457 --> 00:00:33.297 But I think it is a good idea to do an episode when I have something interesting 00:00:33.297 --> 00:00:40.577 to say and not to bore you with my rambling when I've already said it in the 00:00:40.577 --> 00:00:43.417 previous episode and I don't have anything new to add. 00:00:44.117 --> 00:00:51.397 This time I've been quite busy with writing a lot of code so that's why I have 00:00:51.397 --> 00:00:55.237 not been speaking to the camera as much. 00:00:56.157 --> 00:01:05.957 And what I've been doing is I've been writing really various programs, mostly open source. 00:01:06.237 --> 00:01:12.837 They're mostly on my GitHub, or at least I publish them when I finish at least 00:01:12.837 --> 00:01:15.797 a version that can compile and run and do something. 00:01:16.917 --> 00:01:24.057 And this is what this episode is also about. With AI, with new models, 00:01:24.317 --> 00:01:30.937 both open weights and closed, we have been given a superpower to create software. 00:01:31.277 --> 00:01:35.157 And by we, I'm expanding the definition of programmer. 00:01:35.777 --> 00:01:38.857 And there are many people who 00:01:38.857 --> 00:01:44.077 were not traditionally coders who can now write code using these tools. 00:01:45.337 --> 00:01:49.697 And in this episode, I'm asking a question today. 00:01:49.937 --> 00:01:53.717 Should we share the vibe-coded stuff? 00:01:53.917 --> 00:01:59.797 Should we write and publish software that is maybe not neat, not perfect? 00:02:00.077 --> 00:02:03.317 Maybe we don't even know how it works? 00:02:04.077 --> 00:02:10.057 And does it make sense to publish it if it will have maybe no users, 00:02:10.057 --> 00:02:18.137 maybe a few users, maybe it's just a buggy piece of code and people will laugh at us and so on? 00:02:18.137 --> 00:02:23.457 So I think writing software, open source hacking is about tinkering, 00:02:23.697 --> 00:02:26.697 it is about playing, experimenting, trying new things. 00:02:26.997 --> 00:02:31.777 It doesn't mean that these things should be perfect from start to end. 00:02:32.960 --> 00:02:37.580 But we can play with them and this talk 00:02:37.580 --> 00:02:41.720 that i will just roll in is 00:02:41.720 --> 00:02:49.380 a recording from devhack day which is a part of btc prague for nerds and last 00:02:49.380 --> 00:02:56.580 year i noticed that not all participants are actually coders which is fine because 00:02:56.580 --> 00:03:01.140 we need many more people than just coders you know the common saying, 00:03:01.420 --> 00:03:04.880 yes, cypherpunks write code, but they also write documentation, 00:03:05.320 --> 00:03:07.120 they make design, they make art. 00:03:07.800 --> 00:03:11.500 They promote, they figure out new ways to use these technologies. 00:03:11.620 --> 00:03:18.260 They participate in social media, organize events, and then they apply and use 00:03:18.260 --> 00:03:23.560 these technologies in various businesses and give us legal help and many, 00:03:23.560 --> 00:03:25.220 many other important things. 00:03:25.600 --> 00:03:28.720 So for hacking and tinkering there is i 00:03:28.720 --> 00:03:31.800 would say i i would expand the definition 00:03:31.800 --> 00:03:34.880 of who is the 00:03:34.880 --> 00:03:37.600 cypherpunk's not only write code but do 00:03:37.600 --> 00:03:42.800 many of these other things so i hoped that there will be some people who are 00:03:42.800 --> 00:03:50.940 not necessarily hardcore coders and i spoke spoke about tinkering and this process 00:03:50.940 --> 00:03:55.280 of experimenting and creating new things. So I hope you like it. 00:03:55.600 --> 00:04:01.340 Let's dive in. I always liked to write code and release it. 00:04:01.420 --> 00:04:07.220 I was an early member of an open source movement, a free software movement. 00:04:08.240 --> 00:04:12.380 And I was coding since I was very young. 00:04:12.600 --> 00:04:19.060 I got my first computer when I was seven years old and the games were shitty. 00:04:19.060 --> 00:04:21.180 So I started to code instead. 00:04:22.240 --> 00:04:30.740 And then sometime I kind of was doing some other things, mainly Linux servers and so on. 00:04:30.860 --> 00:04:34.720 But I got back to playing, tinkering and open source. 00:04:35.380 --> 00:04:38.880 And this talk is about a long tale of open source. 00:04:39.020 --> 00:04:45.900 And I want to encourage you to create software, contribute, write and do it, 00:04:45.900 --> 00:04:51.300 especially if the software is shitty, if you're not proud of it and. 00:04:52.240 --> 00:04:53.920 It has some benefits. 00:04:55.000 --> 00:04:58.120 So, I don't know. Okay. 00:04:58.800 --> 00:05:03.560 So, you can become part of something that I call the GitHub graveyard. 00:05:04.100 --> 00:05:10.180 So, most people, when we speak about open source, we talk about very successful software. 00:05:10.360 --> 00:05:15.160 The Linux kernel is used definitely on more than 3 billion devices. 00:05:15.460 --> 00:05:20.700 It's by far the most used operating system kernel in the world. 00:05:20.700 --> 00:05:28.460 Android phones run Linux kernel, routers, servers, many desktop computers and so on. 00:05:29.873 --> 00:05:36.213 The second most popular kernel, which is Windows NT kernel, is not even close. 00:05:36.373 --> 00:05:39.113 It's like way farther in use. 00:05:39.293 --> 00:05:44.593 But I'm not talking about this widely successful open source software like the 00:05:44.593 --> 00:05:52.593 Linux kernel, Bitcoin Core, or many other applications that are used by many people. 00:05:52.593 --> 00:05:56.733 And when you read traditional traditional 00:05:56.733 --> 00:06:03.113 like like cathedral and the bazaar and traditional philosophy of free and open 00:06:03.113 --> 00:06:09.753 source software it usually talks about it is higher quality because given enough 00:06:09.753 --> 00:06:16.613 eyeballs all bugs are shallow people look at it they fix things they contribute things and so on. 00:06:17.153 --> 00:06:20.733 But that's the core. But then there's the tail. 00:06:20.953 --> 00:06:26.133 These are the projects on GitHub that have no users and less than three stars. 00:06:26.573 --> 00:06:32.233 And they're basically failed by all conventional metrics. 00:06:32.753 --> 00:06:39.493 And I'm going to try to convince you that it's still a good idea to publish code like this. 00:06:40.393 --> 00:06:43.253 Maybe we are measuring it wrong maybe this is 00:06:43.253 --> 00:06:46.273 not a success maybe a success is a weekend 00:06:46.273 --> 00:06:50.033 hacking project that you share and this 00:06:50.033 --> 00:06:56.733 also gives information to the community because even if the project fails if 00:06:56.733 --> 00:07:01.473 you tried something and then realize okay this is not a good way to do it if 00:07:01.473 --> 00:07:05.853 you don't share it there will maybe be 10 other people that will try the same 00:07:05.853 --> 00:07:09.553 thing and also fail while they could have just, you know, 00:07:09.753 --> 00:07:13.113 looked at your software or looked at your experience. 00:07:13.953 --> 00:07:16.473 So in science, this is called the dark data. 00:07:17.603 --> 00:07:23.163 So many, many scientific papers try to prove some theory. 00:07:23.403 --> 00:07:28.243 They gather data, they have a theory, they gather data. And with the data, 00:07:28.463 --> 00:07:29.963 they reject their theory. 00:07:30.483 --> 00:07:36.183 So this is basically, people call it a failure of science because they set out 00:07:36.183 --> 00:07:39.143 to prove something and they proved it's not true. 00:07:39.703 --> 00:07:45.043 But if they don't publish this information, then there may be other scientists 00:07:45.043 --> 00:07:48.663 that will do the same thing and they will also prove that it's not true. 00:07:48.903 --> 00:07:52.323 So it's better if they can just read the paper and say, ah, okay, 00:07:52.503 --> 00:07:58.743 someone thought this about reality or about whatever, medicine, 00:07:59.103 --> 00:08:06.963 about something they were investigating in this scientific search and they realized, 00:08:07.163 --> 00:08:10.803 okay, it doesn't work this way, but it's useful information. 00:08:10.803 --> 00:08:12.583 It saves time to other people. 00:08:13.603 --> 00:08:16.603 On the other hand open source 00:08:16.603 --> 00:08:20.323 and hacking is not scientific it's 00:08:20.323 --> 00:08:23.683 about tinkering it's about you know playing with things 00:08:23.683 --> 00:08:26.783 trying to do small changes maybe 00:08:26.783 --> 00:08:32.783 do some some experiments and you can you can see this in a 3d printing community 00:08:32.783 --> 00:08:38.323 oh okay there's a cup what if i do the handle this way and people play and do 00:08:38.323 --> 00:08:44.443 small adjustments and if they share it it's maybe useful to someone. 00:08:46.183 --> 00:08:49.963 But then people say it's the fear of the ship. 00:08:50.323 --> 00:08:53.543 You know, this code is not polished enough. 00:08:53.723 --> 00:08:59.103 I will release it when I write the documentation or when I remove all the to-do 00:08:59.103 --> 00:09:05.643 lines or to-fix lines or whatever very common is. I will rewrite this in Rust. 00:09:05.863 --> 00:09:11.643 Now it's in Python, but I will release it when it will be rewritten in a proper language. 00:09:13.057 --> 00:09:17.517 Then another reason why people don't ship is nobody will use this, 00:09:17.777 --> 00:09:20.697 which might be true. All of them might be true, actually. 00:09:21.757 --> 00:09:25.577 Maybe better projects exist, but then why did you write it? 00:09:25.717 --> 00:09:30.457 Maybe you just wanted to learn something, but maybe you wrote it because something 00:09:30.457 --> 00:09:33.357 was wrong about the original project. 00:09:33.957 --> 00:09:39.157 So maybe there's a better project that exists, but that didn't meet your needs. 00:09:39.157 --> 00:09:45.037 So you wrote the project because there was something wrong about the big project. 00:09:45.877 --> 00:09:50.317 Sometimes also might be true. It's too specific to my workflow. 00:09:50.637 --> 00:09:54.337 For example, I have a software that does one thing. 00:09:54.477 --> 00:09:59.337 It's one button and it opens the gate in our building, which is a relay. 00:09:59.777 --> 00:10:05.397 Very specific to my workflow, but maybe other people want to open doors as well. 00:10:06.517 --> 00:10:10.017 So does it make to release anyway even 00:10:10.017 --> 00:10:14.297 if any of this is true maybe maybe it's 00:10:14.297 --> 00:10:21.357 it's a good idea to release anyway so this is very controversial i i have to 00:10:21.357 --> 00:10:26.077 say i'm a proper programmer i even studied it i have paper with a stamp that 00:10:26.077 --> 00:10:32.597 i studied computer science so when i when i say this among my peers. 00:10:32.937 --> 00:10:36.177 They usually don't agree, but I love ugly code. 00:10:36.717 --> 00:10:43.057 Like it's a common story. There's a new engineer that comes to a company that 00:10:43.057 --> 00:10:44.757 had like an internal system. 00:10:44.977 --> 00:10:50.377 They've been developing it in PHP spaghetti code for 15 years and they come 00:10:50.377 --> 00:10:57.817 and they like hold their head and they say, okay, well, this is the worst code ever. And, 00:10:58.716 --> 00:11:02.736 And the only thing you can do with this code is rewrite it. 00:11:03.676 --> 00:11:06.976 When I hear this, first of all, I love the ugly code. 00:11:07.156 --> 00:11:12.376 Like if someone has been working on it for 15 years, even if it's ugly code, 00:11:12.516 --> 00:11:19.336 if it's spaghetti code, it kind of absorbs all the needs over the 15 years, all the specifics. 00:11:19.536 --> 00:11:23.656 And if it works, if it's useful, it doesn't matter that it's ugly. 00:11:23.836 --> 00:11:29.836 I think what matters is if it solves the problem and maybe have ugly user interface 00:11:29.836 --> 00:11:34.216 or, you know, it's not properly abstracted, there are no modules or something. 00:11:34.576 --> 00:11:40.476 But if it solves the problem, I think we should kind of praise the code. 00:11:40.736 --> 00:11:46.296 And what people don't see when they look at the ugly code is that it's quite 00:11:46.296 --> 00:11:49.936 hard because the specifications are changing over years. 00:11:50.096 --> 00:11:55.236 If it's something that's been developed over 15 years, it has a lot of knowledge in there, 00:11:55.236 --> 00:11:59.536 even though it doesn't have beautiful architecture so often 00:11:59.536 --> 00:12:02.376 so i draw here with a with 00:12:02.376 --> 00:12:05.556 a car that has a comma ai which is open source 00:12:05.556 --> 00:12:08.236 free and open source self-driving kit i don't know 00:12:08.236 --> 00:12:11.376 if any of you know comma it's basically 00:12:11.376 --> 00:12:16.956 an ai based self-driving so i don't have to drive it is actually on par with 00:12:16.956 --> 00:12:22.516 tesla self-driving but you can install it in many normal standards cars so i 00:12:22.516 --> 00:12:30.656 have a volkswagen so I draw here and about a month ago I was kind of curious okay this is a. 00:12:31.176 --> 00:12:33.836 Fastly developing field and I didn't get 00:12:33.836 --> 00:12:36.916 an update for the software for one year so I 00:12:36.916 --> 00:12:39.976 looked at their github and they said that they are rewriting the 00:12:39.976 --> 00:12:42.936 code so then I realized okay this 00:12:42.936 --> 00:12:45.876 project is dead so let's look for the forks there is 00:12:45.876 --> 00:12:48.636 a fork that is called frog pilot which is 00:12:48.636 --> 00:12:52.396 up to date released monthly it's not being rewritten 00:12:52.396 --> 00:12:55.116 and it has jumping frogs on the 00:12:55.116 --> 00:12:58.356 screen so it looks like kind of silly makes the 00:12:58.356 --> 00:13:03.476 frog sounds when you engage the autopilot so i 00:13:03.476 --> 00:13:06.156 really hate when someone says okay this is 00:13:06.156 --> 00:13:09.196 all wrong let's just rewrite it because usually people 00:13:09.196 --> 00:13:11.936 will waste a year of their 00:13:11.936 --> 00:13:14.796 time into rewriting and then 00:13:14.796 --> 00:13:17.756 the new requirements start coming in 00:13:17.756 --> 00:13:21.676 and the nice new codes will again not 00:13:21.676 --> 00:13:25.496 be so nice so it's not a permanent solution that's 00:13:25.496 --> 00:13:28.676 just you know freshly graduated engineer 00:13:28.676 --> 00:13:31.736 had a good idea we need to rewrite this because this 00:13:31.736 --> 00:13:34.636 is shit so the problem is the rewrite 00:13:34.636 --> 00:13:40.936 is often never finished so it's just a waste of time and the project dies the 00:13:40.936 --> 00:13:45.816 only thing i hate more is when they say we are going to rewrite it in a rust 00:13:45.816 --> 00:13:50.236 because that means that even if they rewrite it it will have zero contributors 00:13:50.236 --> 00:13:52.956 because no one likes to write rust code or. 00:13:54.270 --> 00:13:57.570 Five percent like compared to python or php 00:13:57.570 --> 00:14:01.190 it's very hard to write in rust so and 00:14:01.190 --> 00:14:03.950 embrace the shitty code doesn't matter that it's 00:14:03.950 --> 00:14:07.090 written in a shitty language if it works and please 00:14:07.090 --> 00:14:10.070 don't rewrite for like maybe there 00:14:10.070 --> 00:14:13.170 are some cases when it's good idea to rewrite maybe some 00:14:13.170 --> 00:14:17.870 cases even when it's good idea to rewrite in rust but just 00:14:17.870 --> 00:14:21.150 you know it's okay don't don't 00:14:21.150 --> 00:14:24.750 worry that it's ugly code just publish it if it 00:14:24.750 --> 00:14:27.750 has users maybe you can fix the 00:14:27.750 --> 00:14:30.670 ugly parts later maybe you will get some pull 00:14:30.670 --> 00:14:33.450 requests but most probably you will not 00:14:33.450 --> 00:14:36.210 get many users and you don't have to make it 00:14:36.210 --> 00:14:39.550 polished and nice now with 00:14:39.550 --> 00:14:42.910 ai coding there's this thing that happened and 00:14:42.910 --> 00:14:47.790 it happened before in other fields so now everyone who has a smartphone is a 00:14:47.790 --> 00:14:54.510 photographer and everyone everyone one who sets up a blog on Medium or Habla 00:14:54.510 --> 00:14:59.710 or Nostra or X or wherever is basically a journalist. 00:14:59.710 --> 00:15:03.070 We can write about our opinions, about our things. 00:15:03.310 --> 00:15:06.170 We can take pictures of our holidays and so on. 00:15:07.170 --> 00:15:12.510 And like these hobby photographers, they're not afraid to, you know, 00:15:12.750 --> 00:15:18.750 show their photos from their travels to their friends just because they're not 00:15:18.750 --> 00:15:19.850 professional photographers. 00:15:19.850 --> 00:15:25.790 Yes, the photos are not as nice as like super artsy photographers with fancy 00:15:25.790 --> 00:15:29.130 cameras and knowing how to do composition, 00:15:29.130 --> 00:15:33.950 but the professional photographers are not traveling with you all the time, 00:15:33.950 --> 00:15:37.810 so they couldn't have made pictures from your travels, so only you could have. 00:15:38.070 --> 00:15:42.450 And I think it's the same case with the code. Many people are going to write 00:15:42.450 --> 00:15:47.430 code, especially using AI, even people who don't know how to code really well. 00:15:47.530 --> 00:15:50.450 Now it's literally wipe coding. you write a prompt and 00:15:50.450 --> 00:15:53.870 you have a software and like 00:15:53.870 --> 00:15:56.810 if i take my degree out 00:15:56.810 --> 00:15:59.550 of my pocket and i i would look at the code i would say oh 00:15:59.550 --> 00:16:02.570 okay i see this was written by ai there are 00:16:02.570 --> 00:16:05.910 five bugs here and it's a shitty code but i 00:16:05.910 --> 00:16:08.950 don't live your life i would not have written the 00:16:08.950 --> 00:16:11.930 code for you because you wanted to write 00:16:11.930 --> 00:16:15.010 it so why not publish it it's like it doesn't 00:16:15.010 --> 00:16:17.990 matter the same same way as my photos 00:16:17.990 --> 00:16:21.010 when I want to publish them I publish them because 00:16:21.010 --> 00:16:26.150 there's no professional photographer following me around so don't worry about 00:16:26.150 --> 00:16:33.890 ugly code so maybe your hacky script is someone's missing piece maybe your failed 00:16:33.890 --> 00:16:39.130 experiment is someone's free experience and maybe your thinking process is someone's inspiration. 00:16:40.460 --> 00:16:45.860 This one is important because especially with permissionless technology such 00:16:45.860 --> 00:16:50.600 as Bitcoin, Nostra and so on, we can do many experiments. 00:16:51.260 --> 00:16:56.360 And it's maybe not the software that implements the experiment, but the idea. 00:16:56.800 --> 00:17:05.560 But people can see, okay, I will give you a few examples actually. 00:17:06.300 --> 00:17:09.280 Maybe this is a way to do it. maybe I 00:17:09.280 --> 00:17:12.240 won't use the software I will play with the software 00:17:12.240 --> 00:17:16.380 and then I will rewrite it in a way but the 00:17:16.380 --> 00:17:21.340 idea is expressed in the software so that's how you how you can write a blog 00:17:21.340 --> 00:17:25.440 about it but if you have an implementation people can play with the implementation 00:17:25.440 --> 00:17:31.460 and then write it properly so that's another reason to release it so I give 00:17:31.460 --> 00:17:33.300 you first example I have a few. 00:17:34.040 --> 00:17:37.020 So i have this device called pulsetto it's 00:17:37.020 --> 00:17:39.920 a vagal nerve stimulator vagal nerve is the 00:17:39.920 --> 00:17:42.720 longest nerve in the body that relaxes your body 00:17:42.720 --> 00:17:45.400 and i went to a 00:17:45.400 --> 00:17:49.140 music festival and there was a giant storm like portable 00:17:49.140 --> 00:17:52.640 toilets were flying around it's really windy so 00:17:52.640 --> 00:17:55.800 i went into my camper van electricity was off 00:17:55.800 --> 00:17:58.580 cell towers were off i was a little 00:17:58.580 --> 00:18:01.400 bit stressed you know i was looking if my friends 00:18:01.400 --> 00:18:04.840 are safe i couldn't contact them because the mobile network 00:18:04.840 --> 00:18:08.500 was off so i said oh perfect time for vagal nerve 00:18:08.500 --> 00:18:12.280 stimulation i need to calm down so i put it on i started 00:18:12.280 --> 00:18:15.700 the app the app basically sends the comment via bluetooth 00:18:15.700 --> 00:18:18.520 to the device to turn on uh but it 00:18:18.520 --> 00:18:23.060 needs to connect to internet to verify my account there's no subscription but 00:18:23.060 --> 00:18:27.220 it just didn't start because i was not online so i said okay this is fucking 00:18:27.220 --> 00:18:32.080 stupid i don't want this you know device to connect to the internet just to 00:18:32.080 --> 00:18:36.680 send signal to start electric stimulation so i released. 00:18:37.300 --> 00:18:40.960 An open source software i think it has three 00:18:40.960 --> 00:18:43.940 users me my wife and maybe one more user i 00:18:43.940 --> 00:18:47.080 i've heard someone used it for me it's 00:18:47.080 --> 00:18:50.320 not about how many users it has it solves my problem it runs 00:18:50.320 --> 00:18:53.240 on my phone it's published online so if i 00:18:53.240 --> 00:18:56.480 buy a new phone i i can just get it through obtainium 00:18:56.480 --> 00:18:59.440 on on my phone again and i learned how 00:18:59.440 --> 00:19:02.420 to reverse engineer bluetooth lower energy so that 00:19:02.420 --> 00:19:05.680 that was a learning experience for me for me the code 00:19:05.680 --> 00:19:08.820 is very shitty it is written in some framework that 00:19:08.820 --> 00:19:11.820 i don't even know it was completely wipe coded i 00:19:11.820 --> 00:19:17.380 did only the reverse engineering part but then everything else the gui was written 00:19:17.380 --> 00:19:23.060 by i think chat gpt i don't even know so yes it's a software doesn't have many 00:19:23.060 --> 00:19:30.760 users but i put it out there if anyone has this problem they can they can use it they can google it and. 00:19:31.891 --> 00:19:39.131 Do it. So I think we should be building for the joy of building the same way 00:19:39.131 --> 00:19:43.591 as people are taking pictures even though they're not professional photographers. 00:19:43.891 --> 00:19:50.511 I encourage you to write software just because it's fun and these days it's 00:19:50.511 --> 00:19:54.491 incredibly fun even if you're not very tech-savvy. 00:19:55.351 --> 00:19:59.151 So what should you release? Gists. 00:19:59.411 --> 00:20:03.491 Gists are a special thing on GitHub that is just one file, usually one script 00:20:03.491 --> 00:20:07.791 with documentation in comments for things you do daily. 00:20:07.931 --> 00:20:12.431 So you solve something once and you share it and people can use it. 00:20:12.571 --> 00:20:15.711 They just install it usually as a shell or python script 00:20:15.711 --> 00:20:18.691 for example i have just called whisper 00:20:18.691 --> 00:20:22.271 file so you get any video or audio 00:20:22.271 --> 00:20:25.451 file and it will transcribe it 00:20:25.451 --> 00:20:28.611 locally using open source whisper speech to 00:20:28.611 --> 00:20:33.111 text models but it does all the auto conversion so it doesn't matter what the 00:20:33.111 --> 00:20:38.211 format the the file is in you just you do just whisper file and then path to 00:20:38.211 --> 00:20:47.091 to an mp3 or m4a or m mp4 or whatever and you will get txt file and. 00:20:48.151 --> 00:20:52.691 Srt file with the subtitles which includes timing and you have a transcription 00:20:52.691 --> 00:21:00.091 then i've another one is yt whisper which is the same thing but it takes a youtube 00:21:00.091 --> 00:21:04.931 url so it downloads the video extracts audio and then transcribes it. 00:21:05.071 --> 00:21:11.831 So if you want just to look in some video, see what they've been talking about and so on. 00:21:12.611 --> 00:21:16.231 Venice IMG was used to generate all these images. 00:21:16.571 --> 00:21:20.011 So basically when you have these AI image generation models, 00:21:20.111 --> 00:21:22.991 you need to type the prompt, you press enter and you wait for, 00:21:23.031 --> 00:21:25.311 I don't know, half a minute, one minute. 00:21:26.264 --> 00:21:33.364 This one uses API and does it in parallel. So what I did, I took the PDF of this presentation. 00:21:33.704 --> 00:21:38.604 I gave it to four different AI models and I said, describe me cool pictures 00:21:38.604 --> 00:21:41.624 that could go along with this presentation. 00:21:41.724 --> 00:21:47.084 I just set the aspect ratio for vertical because I have the space on the right side. 00:21:47.844 --> 00:21:51.884 And then I just generated all of them in parallel. So in two minutes, 00:21:52.124 --> 00:21:56.844 I got 400 images and I just chose the nice ones. 00:21:57.024 --> 00:22:01.304 So you can do it. It's just literally one short Python script. 00:22:02.044 --> 00:22:06.904 I've been publishing my .files. It's all the macros in the shell and things 00:22:06.904 --> 00:22:09.224 that makes your shell more useful. 00:22:09.584 --> 00:22:13.524 Another thing you can release are weekend experiments. 00:22:14.284 --> 00:22:18.304 What if? So I will show you this one a little bit later. But for example, 00:22:18.584 --> 00:22:24.764 I'm in many signal groups with a lot of messages and sometimes I want to have 00:22:24.764 --> 00:22:26.724 an overview of what is happening. 00:22:27.004 --> 00:22:32.884 So I have a software, I give it one month of messages from a signal group. 00:22:33.044 --> 00:22:37.884 It uses AI to basically summarize it, extract all the links, 00:22:38.044 --> 00:22:43.724 describe images, transcribe voice messages, and it basically creates one nice 00:22:43.724 --> 00:22:45.924 summary of what has been going on. 00:22:45.924 --> 00:22:48.844 So my friend is using this for a newsletter. He's writing... 00:22:50.049 --> 00:22:55.169 It's a Liberation Travel newsletter, and it basically summarizes discussions 00:22:55.169 --> 00:23:02.209 from many groups that this company, Liberation Travel, has for its clients. 00:23:02.229 --> 00:23:09.349 I think we have, I don't know, 10, 12 signal groups, and that's the source of the content. 00:23:09.549 --> 00:23:14.829 So it's a nice weekend experiment. This weekend lasted for four months. 00:23:14.829 --> 00:23:18.669 Actually, it's not as easy as it sounds, but it's out there. 00:23:18.669 --> 00:23:27.149 I will talk about this one later personal automation basically your workflow is someone's. 00:23:27.809 --> 00:23:30.869 Inspiration so for example rss to 00:23:30.869 --> 00:23:33.689 podcast i want to read blogs but i don't have 00:23:33.689 --> 00:23:36.429 time to look in the on the screen so when 00:23:36.429 --> 00:23:39.489 i'm driving or actually when my car is driving 00:23:39.489 --> 00:23:42.569 itself i have time to read blogs but 00:23:42.569 --> 00:23:46.329 i don't want to look because i want to look on the road so this 00:23:46.329 --> 00:23:50.069 software will create personal podcast for 00:23:50.069 --> 00:23:52.929 for me from the blogs that i want to read i also 00:23:52.929 --> 00:23:55.829 release it online as a service which is called loaded for 00:23:55.829 --> 00:23:58.869 dot me so if you want to read some blogs 00:23:58.869 --> 00:24:02.129 in audio form this will this will basically do 00:24:02.129 --> 00:24:05.649 text to speech in a very fancy way it's also 00:24:05.649 --> 00:24:08.609 quite nice inspiration of how it's done it's not 00:24:08.609 --> 00:24:12.189 it's quite fancy actually then there's 00:24:12.189 --> 00:24:15.509 like video summarize so if someone like 00:24:15.509 --> 00:24:18.449 wants to hear a summary of this talk they can 00:24:18.449 --> 00:24:21.649 take the transcript and get the main 00:24:21.649 --> 00:24:24.589 points so it's a command line utility i use oh there's 00:24:24.589 --> 00:24:27.689 a nice talk i want to just see what they're talking about 00:24:27.689 --> 00:24:31.149 so i download it transcribe it summarize it 00:24:31.149 --> 00:24:33.949 and i have like a one pager to see if i want. 00:24:33.949 --> 00:24:37.929 To actually see the video some stupid 00:24:37.929 --> 00:24:40.769 ideas so so i i 00:24:40.769 --> 00:24:43.909 wrote a software that sends cashew tokens over 00:24:43.909 --> 00:24:46.849 a long-range radio over reticulum it's called 00:24:46.849 --> 00:24:49.949 not bent i have a very nice video showing 00:24:49.949 --> 00:24:53.169 how you can contact your 3d printer filament 00:24:53.169 --> 00:24:56.549 dealer in case of apocalypse and you can you 00:24:56.549 --> 00:25:02.949 can play with this my very recent project i think started working about a week 00:25:02.949 --> 00:25:09.469 ago was creating any cashew tokens for any fiat currency so if it has a price 00:25:09.469 --> 00:25:14.069 feed you can do check rounds over cashew or anything like that on my github 00:25:14.069 --> 00:25:16.449 or read about it on my nostril. 00:25:18.289 --> 00:25:24.029 So then we have the discovery problem, and this is something that is not solved. 00:25:24.769 --> 00:25:34.129 So GitHub search is broken for small projects. If you are trying to look for something, 00:25:34.349 --> 00:25:41.809 it is very difficult to see if someone has written this software or something like that. 00:25:41.809 --> 00:25:48.269 There is almost no curation for niche tools like my Pulseto software. 00:25:48.709 --> 00:25:55.069 There's a lot of signal lost in even more of noise. So what works? 00:25:55.409 --> 00:26:00.969 Personal blogs, notes, you can just share, okay, I built this and people can play with it. 00:26:02.009 --> 00:26:06.929 Then there should be, I have a friend and he used to do this thing. 00:26:06.929 --> 00:26:09.909 He would open facebook and he would say oh 00:26:09.909 --> 00:26:13.489 let's let's show the our female friends 00:26:13.489 --> 00:26:16.369 to each other but i never got to show him mine because 00:26:16.369 --> 00:26:19.269 he was showing me his so i started doing 00:26:19.269 --> 00:26:22.229 this on github i was just showing people oh these 00:26:22.229 --> 00:26:25.569 are my 50 repositories that i that i've 00:26:25.569 --> 00:26:28.829 done and people just sharing this which with 00:26:28.829 --> 00:26:32.589 each other crowdfunding is very useful 00:26:32.589 --> 00:26:35.529 for discovery actually because with crowdfunding you 00:26:35.529 --> 00:26:38.669 have to do marketing it gives you users buying 00:26:38.669 --> 00:26:41.849 so basically what i did i i've done one software 00:26:41.849 --> 00:26:45.629 which is called llama like venice i did 00:26:45.629 --> 00:26:50.789 crowdfunding basically if you if you supported the project you could get the 00:26:50.789 --> 00:26:55.929 source code right away and then it was released as free and open source when 00:26:55.929 --> 00:27:00.849 it reached the funding goal so this gives makes it more marketing intensive 00:27:00.849 --> 00:27:04.909 you have to focus okay you need to promote your crowdfunding campaign it's not easy. 00:27:05.489 --> 00:27:08.869 But you will write much better code because people paid 00:27:08.869 --> 00:27:12.389 for it so you're not gonna you know give them of course 00:27:12.389 --> 00:27:15.229 uh but also when people invest it 00:27:15.229 --> 00:27:18.009 in your code they will at least try it maybe they 00:27:18.009 --> 00:27:20.969 will not use it but that it's this notion of oh 00:27:20.969 --> 00:27:23.829 okay i paid i don't know 10 000 sats 00:27:23.829 --> 00:27:26.529 for it so at least i'm going to download and play with it 00:27:26.529 --> 00:27:29.729 so so it's actually creating this 00:27:29.729 --> 00:27:34.349 barrier was very helpful for for this then 00:27:34.349 --> 00:27:37.409 if people are doing something and and you 00:27:37.409 --> 00:27:40.589 have solved it in a software that is like 00:27:40.589 --> 00:27:45.429 when i see something doing something on a linux machine and and they're not 00:27:45.429 --> 00:27:50.889 very good i i have this urge to kind of punch them and you know steal their 00:27:50.889 --> 00:27:55.789 keyboard and show them how it's done so share your software like this oh i why 00:27:55.789 --> 00:27:59.969 are you transcribing a video so So, so complicated way, 00:28:00.069 --> 00:28:02.449 just run the script and it does everything for you. 00:28:02.889 --> 00:28:04.809 And lightning talks are really nice. 00:28:06.379 --> 00:28:11.039 Okay, and now we are approaching the end. This is one of the projects. 00:28:11.439 --> 00:28:14.759 This generates, it's basically an Oster client. 00:28:15.099 --> 00:28:21.659 This is the NPUB version for things when you want to share something, 00:28:21.659 --> 00:28:24.399 but you don't want to share it under your own name. 00:28:24.939 --> 00:28:30.559 But creating one-time use NPUB doesn't make sense because no one will read it. 00:28:30.759 --> 00:28:33.939 So this is a shared identity. the private 00:28:33.939 --> 00:28:37.459 key is derived only from the password so basically 00:28:37.459 --> 00:28:40.419 here you write what you want to write when you 00:28:40.419 --> 00:28:44.979 write the password the private key is is derived you press this button and it's 00:28:44.979 --> 00:28:49.359 published under this shared identity and then you just spread the password among 00:28:49.359 --> 00:28:55.459 the cool people and you have a community of anonymous microblog of people no 00:28:55.459 --> 00:28:57.979 one knows who it is and so on, 00:28:58.719 --> 00:29:01.219 So wipe coding is very cool. 00:29:01.719 --> 00:29:05.539 Please ship. I will skip over this a little bit. 00:29:06.399 --> 00:29:10.719 But basically, we are all going to run our own AI inference. 00:29:11.319 --> 00:29:16.379 And the only challenge I only ask is when you wipe code something, 00:29:16.879 --> 00:29:19.659 please write how to use it and how to deploy it. 00:29:19.859 --> 00:29:24.239 No one knows how to install your specific version of Node.js. 00:29:24.459 --> 00:29:28.939 It can be actually generated also. So write good documentation. 00:29:29.699 --> 00:29:35.579 So building in public, why? Code diary, you enable serendipity. 00:29:37.179 --> 00:29:42.819 You create breadcrumbs for other people. And the thinking process is also interesting. 00:29:43.299 --> 00:29:49.299 Maybe you learn about some better and bigger software that solves the same problem. 00:29:50.979 --> 00:29:54.059 And how to do it? Write a good readme. 00:29:54.279 --> 00:30:01.119 Why I build this? How to deploy? And so on. share failures and the ask you are 00:30:01.119 --> 00:30:06.659 a coder now if you have ai ship at least one useless project per month and share 00:30:06.659 --> 00:30:10.999 it on nostr thank you very much this is how you can contact me and what you 00:30:10.999 --> 00:30:14.219 can talk with me about so thank you very much. 00:30:19.739 --> 00:30:24.179 Yeah we could do i didn't see you were there we could do one extra questions 00:30:24.179 --> 00:30:27.539 maybe if you're still around for five minutes we could take that out the break 00:30:27.539 --> 00:30:34.999 if you're interested anyone has any questions for you man there's one there's 00:30:34.999 --> 00:30:37.499 two down here there oh i'll take it. 00:30:40.903 --> 00:30:45.043 As you mentioned earlier, there's a search engine in the GitHub is broken. 00:30:45.423 --> 00:30:47.243 And what about the white coding? 00:30:47.723 --> 00:30:53.563 There should be a tool to filter the repository that has been white put because of the searching. 00:30:53.883 --> 00:30:59.403 I think maybe the next step should be there should be a tag, 00:30:59.523 --> 00:31:01.923 at least I white put it in this record. 00:31:02.223 --> 00:31:09.303 So maybe don't spend maybe 12 hours to just look at the code because maybe even 00:31:09.303 --> 00:31:12.003 the owner doesn't know what to deploy. 00:31:12.323 --> 00:31:16.123 I think that's not the problem. The problem is finding. Like you are doing something 00:31:16.123 --> 00:31:19.343 super niche like my, you know, Pulseto device. 00:31:19.663 --> 00:31:25.483 And how do you even find for it? Because when you search, you find 50 projects 00:31:25.483 --> 00:31:27.583 that don't solve the problem that you are solving. 00:31:27.823 --> 00:31:34.483 So one of the solutions we were trying was using this kind of software to just 00:31:34.483 --> 00:31:38.223 share cool projects. But the problem is that the domain is too wide. 00:31:38.863 --> 00:31:44.143 So it is very hard to find if someone has even tried to wipe code something. 00:31:44.583 --> 00:31:48.443 As I said, you know, I'm not against shitty code. It's okay. 00:31:48.663 --> 00:31:54.083 Like I can look at the code. You can ask AI to see if there are any backdoors and so on. 00:31:54.203 --> 00:31:57.083 So run it in a virtual machine. It doesn't matter. 00:31:57.383 --> 00:32:03.943 The problem is to actually see if someone has done anything related to what 00:32:03.943 --> 00:32:05.383 you want to do to save you time. 00:32:05.783 --> 00:32:09.003 Maybe you will not use the code, maybe you will just use the ideas, 00:32:09.243 --> 00:32:13.683 maybe you will just take the README and wipe code it yourself or something like 00:32:13.683 --> 00:32:15.103 that. So that's not a problem. 00:32:15.283 --> 00:32:20.903 The discovery problem is there are hundreds of thousands, millions of projects 00:32:20.903 --> 00:32:25.703 and there will be many more because of wipe coding and how do you find if you 00:32:25.703 --> 00:32:29.823 have to actually code something or just use someone else's software and improve it maybe. 00:32:31.325 --> 00:32:36.885 All right, we got one more question here. So I really like this project. I'm not a developer. 00:32:37.245 --> 00:32:40.265 I'm a marketer, but I really like the idea of shared identity. 00:32:40.605 --> 00:32:44.465 So how can I actually use it in real life? 00:32:44.585 --> 00:32:48.465 Can I deploy this to Instagram, to X, or some social media? 00:32:48.705 --> 00:32:53.045 So this is Nostro only, but this website is called NPUB Pro. 00:32:53.225 --> 00:32:56.925 You get it for free from any Nostro identity. and this 00:32:56.925 --> 00:33:00.705 is one html file where you basically say you 00:33:00.705 --> 00:33:03.565 basically just paste the npub so in 00:33:03.565 --> 00:33:06.625 in the readme of my project called anon microblog there's 00:33:06.625 --> 00:33:10.765 a how-to it's very easy to set up it takes like five minutes so if you want 00:33:10.765 --> 00:33:16.825 i can help you even write me and i i've set up there are i think three yes at 00:33:16.825 --> 00:33:21.745 least three instances of this no four that i know of that are that are running 00:33:21.745 --> 00:33:24.005 and it's literally super easy. 00:33:24.165 --> 00:33:27.285 So you need a profile picture and an idea. 00:33:28.405 --> 00:33:32.865 And basically, Enpa Pro gives you a link up there for posting. 00:33:33.205 --> 00:33:37.225 And it can be used from the web. So this is a Nostre account. 00:33:37.225 --> 00:33:41.425 So you can follow this account on Nostre and repost, reply, everything. 00:33:41.985 --> 00:33:46.185 But most users even don't know that there is Nostre behind because this looks 00:33:46.185 --> 00:33:48.965 like a blog and there's a website where you have, 00:33:48.965 --> 00:33:51.865 okay this is where you write what you want to post 00:33:51.865 --> 00:33:54.585 and this is where you write your secret password and and you 00:33:54.585 --> 00:33:57.605 click publish there's no back end everything is on nostril 00:33:57.605 --> 00:34:00.725 relays so you don't have to worry about any 00:34:00.725 --> 00:34:06.225 kind of infrastructure if you can host one html file which i can host for you 00:34:06.225 --> 00:34:11.805 which is this one then you get everything else in the ecosystem for free so 00:34:11.805 --> 00:34:16.265 i'll i'll gladly help you just write me email or whatever and I'll set it up 00:34:16.265 --> 00:34:18.305 for you. It's super easy. Thank you. 00:34:19.625 --> 00:34:21.225 Is there any other questions? 00:34:22.805 --> 00:34:27.045 No? There's, there's, there was one here. It is. 00:34:29.534 --> 00:34:33.834 You mainly talk, I think, about like personal projects of yours. 00:34:34.114 --> 00:34:38.514 Does your mentality change when you actually come to an existing project, 00:34:38.674 --> 00:34:41.094 even if it's also a very small one? 00:34:41.234 --> 00:34:47.334 Because there are other people involved who maybe don't like the idea of a key or a bike code. 00:34:48.714 --> 00:34:53.674 Yes, definitely. The first thing that you need to do is to understand how to 00:34:53.674 --> 00:34:56.394 wire it in and how the people do. 00:34:56.394 --> 00:35:02.054 So, for example, the fiat with Cashew that I was doing, first I had to learn, 00:35:02.194 --> 00:35:04.634 okay, how are actually tokens issued here? 00:35:04.834 --> 00:35:06.894 What's the structure of the project and so on? 00:35:07.314 --> 00:35:12.374 And you are aiming for minimal changes. So, if you want to incorporate in there, 00:35:12.474 --> 00:35:19.374 you don't want to change too much of their code because then it's a maintenance hell and so on. 00:35:19.574 --> 00:35:25.074 So, you want to concentrate your code in like a separate module and do the minimal 00:35:25.074 --> 00:35:28.374 changes there. So it definitely changes how you look at the project. 00:35:28.674 --> 00:35:33.914 You can use AI to understand the project itself as well. It can look at the 00:35:33.914 --> 00:35:36.074 code and tell you, okay, how things are done. 00:35:37.054 --> 00:35:43.014 You should also include the documentation in the prompts so the AI understands 00:35:43.014 --> 00:35:44.914 what the project is about. 00:35:46.274 --> 00:35:50.934 So I'm not saying you should wipe code everything. I also write some code. 00:35:51.214 --> 00:35:54.694 So mainly the boring things like parsing command line options. 00:35:54.694 --> 00:35:58.174 I don't want to write the stupid parser, you know, I just say, 00:35:58.314 --> 00:36:02.634 this is how I want to call it, please parse it and put it into variables. 00:36:02.634 --> 00:36:07.034 And that part is great, like I wouldn't have written it any better. 00:36:07.354 --> 00:36:13.774 So it doesn't like, I think AI writes better code than 90% of programmers. 00:36:14.494 --> 00:36:19.054 So I'm not bashing the output of the AI is actually quite nice. 00:36:19.374 --> 00:36:24.014 So yeah, but it changes like if you're interacting with some other code, 00:36:24.014 --> 00:36:30.394 and you eventually want it merged, then you need to do it the way they do it, basically. 00:36:31.354 --> 00:36:32.514 Thank you very much. 00:36:33.654 --> 00:36:39.094 Thank you for listening. I hope you enjoyed this episode of the Option Plus podcast. 00:36:39.474 --> 00:36:42.914 This is a value-for-value production. If you liked it, 00:36:43.134 --> 00:36:50.734 I hope you used service such as Fountain.fm or any other app or podcast player 00:36:50.734 --> 00:36:56.094 that supports giving value back and that you were streaming some sets. 00:36:57.054 --> 00:37:03.454 I appreciate feedback and in the value for value ecosystem you can attach feedback 00:37:03.454 --> 00:37:12.094 to something that is called a boost so you can send however many sets you like and give comments, 00:37:12.434 --> 00:37:13.914 questions and so on. 00:37:14.094 --> 00:37:16.874 Thank you very much for listening. Bye.