Dancing elephants and upgraded Elves
This show is supported by you. Stick around till the ad break to hear more about that. Happy Valentine's. This is Cup o' Go for 02/13/2026. Keep up to date with the important happenings in the Go community in about twenty minutes per week.
Shay Nehmad:I'm Shay Nehmad. And I'm Jonathan Hall. Happy Valentine's. Hey.
Jonathan Hall:I love you, man.
Shay Nehmad:Yes. I love you, bro. We we have that tenacious disconnection. I'm just wondering who of us is Jack Black and who's Kyle Gas.
Jonathan Hall:Yeah.
Shay Nehmad:You don't know, but we have a tattoo on our like a birthmark or on our lower backs, half a go for each. And we discovered that they connect anyway. I really like Tenacious D.
Jonathan Hall:I can tell.
Shay Nehmad:Go one twenty six has been released, not on Valentine's, which in hindsight, I could have told you because Valentine's is Saturday.
Jonathan Hall:Yeah.
Shay Nehmad:But yeah. So let's go through the entire release.
Jonathan Hall:Yeah. Let's do it again. That was our best episode. Let's have another best episode. No.
Jonathan Hall:We're not gonna do that. We'll we'll put a link in the show notes, to that episode. You can go back and listen. We did a deep dive on that. It was a long one.
Shay Nehmad:What I am gonna do is I'm gonna ask our editor to put in the entire episode just really, really sped up. Filippo, can you put the episode, like, sped up 2,000 times right now?
Jonathan Hall:Fifteen seconds, the entire thing. There you go.
Shay Nehmad:Thanks, Filippo. I don't know if you know this meme, but sometimes on YouTube, there's like Shrek, but every time they take a step, the the movie is twice as fast and it's like a minute long. The b movie, but every time they say b, it it's twice as fast. Do you know that?
Jonathan Hall:Wow. I haven't seen that. That's an interesting concept.
Shay Nehmad:I don't understand the purpose of this content. I don't understand how it was created or came to be, but I found it a very American way to say it. I am here for it. Am here for it. What I will not be here for is the meetup at Stuttgart.
Shay Nehmad:I'm not I'm not gonna be there in time. Oh. But if you're gonna be in Stuttgart, they're having a event, Hello Stuttgart, hosted by Alex B. It's gonna be February 19. Have you been to,
Jonathan Hall:Stuttgart ever? I don't think I have.
Shay Nehmad:You're like a European European and sort of guy. Nearby?
Jonathan Hall:Yeah. I might have I might have stopped in at some point, but I certainly didn't, like, spend a weekend there or anything like that, but I might have driven through the area.
Shay Nehmad:I love the what they're gonna do in the meetup. It's they're just gonna go through sort of like a book club. The first chapter of learn go with pocket sized projects.
Jonathan Hall:Yeah. That's a nice one.
Shay Nehmad:Have you read that one?
Jonathan Hall:You're you're like the best I have. Well, I haven't read the whole thing. I read it before it was released and made a video about it. I liked the pre release version I read. I have a copy sitting around here somewhere, but I haven't made it through the whole thing yet.
Shay Nehmad:So they're gonna go through the first and generally, your recommendation is positive?
Jonathan Hall:Oh, yeah. I really like that book.
Shay Nehmad:So if you are in Stuttgart and listening to this podcast or know someone in Stuttgart who wants to start learning Go or just, like, you know, get together with a community, they're gonna take Hello World and make it into a testable program, which I think is pretty cool. And I guess they're just gonna hang out as well. In Thoughtworks Deutschland, I'm not gonna even try to maybe you can say the name
Jonathan Hall:of the street. No.
Shay Nehmad:I'll try. I'll try. Because they have the B letter, which is actually a S S. Strasse. Yeah.
Shay Nehmad:So it's The the Rupe Roopmans Strasse.
Jonathan Hall:Strasse. Yeah.
Shay Nehmad:If if the address was the thing that tipped you over, you're like, oh, no. It's on Roop Roopmans Strasse Street? I have to go now. Yeah. It does seem like a new kickoff for sort of a new community in of Go, so we really would like to highlight it.
Shay Nehmad:Eight people seem to be attending for the first time, and, you know, I I kicked the tires on the SF, Go meetup scene. It takes a while for people to understand these events are happening and come over, but the people that come for the first ones and then come for the other ones, they can become real cornerstones of the community. So get there. It's it's gonna be fun.
Jonathan Hall:Oh,
Shay Nehmad:yeah. I love how they say gopher curious folks are welcome. Are you gopher curious? Of course, link in the show notes. They also have, like, an x account and a meetup.com and everything you would expect.
Jonathan Hall:Awesome. Good luck to you guys in Stuttgart, on whatever Strasse. I hope that you have a great meetup. Let's talk about a couple other
Shay Nehmad:Don't stress out about this.
Jonathan Hall:Couple other highlights from 01/26. As we already mentioned, we're not gonna do a deep dive, but I wanted to call out one thing we did not mention in the deep dive that was brought up in our channel, and that I actually hadn't even realized was happening. I think we focused on standard library changes in our that episode. We didn't talk about tool changes. There is a new change to the way GoModule and Knit works.
Jonathan Hall:I like this one. It's not a fundamentally interesting change, but what it does is when you create when you do go mod init right now, what version of Go does it pin your module to? Do you know?
Shay Nehmad:Latest one. No?
Jonathan Hall:Yeah. That's what it does until now.
Shay Nehmad:Oh, wait. Why wouldn't it take the latest one?
Jonathan Hall:Now it's gonna do two versions before. So if you're running so and it's a phased it's a phased rollout. But what's gonna happen now so if you're on 01/28 and you do go modern nit, it's gonna tag your module for 01/26.
Shay Nehmad:Why would it do that? Why wouldn't it give me the latest? Seems like I would want the latest release all the time.
Jonathan Hall:For greater backward compatibility, so you don't unintentionally create modules that are not compatible with your colleagues across the hallway or the globe or whatever that need to pull it in. You can still bump it to the latest one if you want to, but it relaxes that default setting to be a little more friendly.
Shay Nehmad:It does have the other side of the coin is that your module will go out of support. Your new module will go out of support in six months. Right?
Jonathan Hall:I I suppose you could look at it that way, but there are many modules out there that, you know, you could still you could still compile because of the forward compatibility promise. You can still compile modules for Go 1.7. Not that modules existed then, but, you know, code for that old version can still be compiled. So I think it's a much less of a problem. If you are stuck using an old version of Go, and many people are for various reasons, you know, they need to support old hardware or whatever, having everybody and their and their dog update their modules as soon as Go one twenty six comes out can be problematic because that means you can't pull in libraries that no longer support your old version of Go.
Jonathan Hall:So this will help with that sort of situation. At least that's the goal. Also, in the Go one twenty six release world, Anton is doing a Go one twenty six release party. What does that mean?
Shay Nehmad:Yeah. This is, like, not sponsored, but we just Anton was if you don't know him, Anton Givanov puts out the release notes and was actually on the show.
Anton:Okay. My my name is Anton. I do some open source stuff, and, I write, interactive, maybe I can call them guides or books and, interactive articles on my blog. That's mostly what I do in my free time.
Shay Nehmad:I just was very happy to see JetBrains, the people behind the IDEs that for some reason my coworkers insist on using instead of VIMP.
Jonathan Hall:Why would you choose those two evil options as your as your
Shay Nehmad:You know why? Because I saw that Notepad has AI features now, so
Jonathan Hall:Wow.
Shay Nehmad:Can't use my Windows
Jonathan Hall:it NotePad comes with then. Clippy emoji. Right?
Shay Nehmad:It came out with a CV related to the AI features, like, last week. This is not related to Go, so I didn't bring it on the show. Yeah. But it just lives rent free in my head. So GoLand, the, Go ID for, Go by JetBrains, which amongst their products, I think is the, like, the second best after Datagrip.
Shay Nehmad:It's a really, really good ID. As much as I like to make fun of JetBrains, it's just a really strong choice. They're having a Go 01/26 release punt party with Anton, on February 19 as well. So if you're not at Stuttgart, here's your event. And they'll just walk you through what's new in 01/26, and, you know, it's gonna be live.
Shay Nehmad:You can save your seat. We put the LinkedIn link, in there. And, yeah, I just thought I don't know how it's gonna go, but it sounds like a super cool event and always fun to see people that were on the show doing other things as well. You know what
Jonathan Hall:I mean? Yeah. Yeah.
Shay Nehmad:Where else, can I go read about the 01/26 release? I can tell you one place it's not. Yeah.
Jonathan Hall:Yeah. So we had on the show in August 2025, we had George Adams from Microsoft on the show who talked about the Microsoft dev blogs and their new Microsoft for Go developers blog, which is just release notes.
Shay Nehmad:Yeah. It turned out to not include any content. I I signed up to their, like, feed to follow this blog, and it's just like, oh, we managed to build Go for, like, Microsoft.
Jonathan Hall:Okay. Hang on. We have to read it this whole segment because you were wrong. It is on there.
Shay Nehmad:Bro, it was literally released right now. Really? I I promise you that twenty minutes ago when I opened this page, it was closed. It didn't have it. Look.
Shay Nehmad:Look. I have the
Jonathan Hall:today's date. That's true.
Shay Nehmad:Look. I I have it open.
Jonathan Hall:Alright. Shai, I I owe you an apology. But for the audience, you heard it here first. Go one point twenty six point zero point dash one is available from Microsoft. You literally heard it here first because it happened as we were recording.
Shay Nehmad:Yeah. In media res. And, yeah, there are, you know, a lot of specific tool chains, specific things for Microsoft with crypto and whatever, if it's relevant for you. I wonder who it's relevant for other than internal devs in Microsoft, but still interesting.
Jonathan Hall:I'm curious, Si. Are you using Go one twenty six yet?
Shay Nehmad:I am not. I have my Go project, and I just realized that it's still on one twenty five. Mhmm. But I am, eager to grab my laptop and, just, try and upgrade and see if it works. Deployed, you know, very simply, like, Lambda's, so shouldn't require any, like, major config or servers or whatever.
Shay Nehmad:I'm not looking for any specific features from, 1.26, and I don't care specifically about performance, but I don't know. I feel like I have to because of my role in the podcasting community or but I haven't I haven't upgraded yet. I actually plan to do the whole RC thing this year, but I just sort of didn't. Looks like they managed without me.
Jonathan Hall:I guess so. I guess so.
Shay Nehmad:Next time. What's when's next time? August?
Jonathan Hall:Yeah. In August. So I've been using it. I've upgraded two projects already. One of them, basically, a big credit application.
Jonathan Hall:I already I was so excited to get rid of that hacky pointer to nonsense where you can't create a pointer to a constant.
Shay Nehmad:Yep. Yep.
Jonathan Hall:One of those, I had over a thousand lines changed just just with that change alone. Oh my god. Felt
Shay Nehmad:so good.
Jonathan Hall:It felt so good to just put new in in place. Yeah.
Shay Nehmad:Alright. Well, we obviously highly recommend you update, if you haven't already. But I understand if, like, you're working in a like, we I I would recommend updating if you're, like, a library maintainer or if you have side projects like me. You're working in a big company, maybe wait for the first minor patch. It shouldn't take too long anyway.
Shay Nehmad:Although, again, this is not really the first minor patch. Right? There's been, like, three or four RCs before this. This is pretty well tested, but I would wait with me one more. Alright.
Shay Nehmad:So this is the the fallout of, 01/26. We're gonna do a quick lighting round, and then we're gonna talk about a big community discussion about what to do with change lists generated by AI.
Jonathan Hall:Oh, yeah. That's gonna be fun.
Shay Nehmad:But let's jump through a few small updates first. Lightning round. Wanted to highlight blog post from a listener that updated on the channel, on our cupago channel, which you are welcome to join. It's, like, getting pretty crowded in here. It's almost 700 people, which is crazy.
Shay Nehmad:But there's a new gopher on the scene. I really like seeing these, blog posts where someone from a different sort of discipline is moving into Go. But El Gofer, which, you you gotta admit is a pretty strong name, it's Augustine Oliveros, is moving from front end to Go using learning Go, an idiomatic approach to real world programming, the John Bodner book, whom we also had on the show. Right? John Bodner?
Jonathan Hall:Indeed. Yeah. Yeah. Sure.
Jon:So software engineer, my current title is staff engineer. I'm at Datadog. I work in making it easier to onboard customers to our APM tools. I've been doing software engineering professionally since '97. That's too many years for me to count the moment.
Jon:Twenty six years. So,
Shay Nehmad:yeah, if you wanna go follow lgopher, lgopher.fly.dev, and see how his journey goes. And, Augustine, now we're all listening, so you have to become really good.
Jonathan Hall:Is it El Gopher or El Gopher?
Shay Nehmad:You're the you have the that accent down. When we get a Russian, front end developer moving to Go, it could be like That's right. Oh, it's Gopher. But you can do the this Spanish accent. Alright.
Shay Nehmad:And again, it's a person from all around the world. I I think it's Argentinian, actually.
Jonathan Hall:Yeah. I think you're right. I think you said he was from Argentina.
Shay Nehmad:Welcome. We are a welcoming community. What do you have for the lighting round?
Jonathan Hall:So we I'm tired of 01/26 already. I wanna talk about one twenty seven. There's, something that has already been merged to the one twenty seven branch to tip proposal to drop PPC 64 big Indian support from elf v one. No. I'm sorry.
Jonathan Hall:Transition from elf v one to elf v two because the elves just weren't good enough.
Shay Nehmad:So this is like Can I can I admit something? I never remember which one is little Indian and which one is vague Indian. I don't
Jonathan Hall:I don't either. And, you know, honestly, unless you're doing low level stuff, you shouldn't need to care about that detail.
Shay Nehmad:So even when I when I was doing low level stuff, I had to I had a little cheat sheet made out of, like, notebook that I posted on the wall with, like, a few shortcuts, also examples of begin the end of the night because I didn't remember. What does this mean for me? Nothing. But the basic idea here,
Jonathan Hall:there are no actively supported PowerPC 64 ELF v one Linux distributions. So why keep supporting it? So my understanding, and I'm not really an expert in this, but my understanding is that we're gonna still support p p c 64 hardware, but not the old elf v one binary format. Now I don't know when elf v two came around, and I don't even know the differences between elf v one and elf v two. I didn't know there were two of them.
Jonathan Hall:I I know there's a difference between ADA out and elf from, like, twenty years ago or something. So I don't know the details here, but my understanding is you can still use that hardware. You just have to use the new binary format now, which you're probably already doing since there are no Linux distributions that still support the old binary format. That's my understanding. If you use PPC 64 and I'm wrong, please just send me a message and tell me how wrong I am.
Jonathan Hall:I'd love to know for my own edification, and I'll do a correction on the show.
Shay Nehmad:Cool. If you can educate John on the public channel where I can see him being wrong as well, I would appreciate it.
Jonathan Hall:Or do it on LinkedIn or Reddit. That's where I'm told I'm wrong on that.
Shay Nehmad:LinkedIn is perfect. Yeah. Okay. Let's move to a quick ad break and then talk about AI.
Jonathan Hall:Oh, let's do it. Thanks for listening to the show. We'd love to have your community support. The best way you can support the show is by sharing it with friends and colleagues or leaving a comment wherever you listen to your podcasts. Leave a review, leave a rating.
Jonathan Hall:That just helps to get the publicity out and get more listeners. You can also support us financially. This is an expensive hobby. If you want to become a Patreon, you can contribute a little bit of your hard earned cash to help pay for editing and hosting fees. And we also have some swag you could buy.
Jonathan Hall:If you'd like to buy a mug, if you don't have one, or a t shirt or anything like that, we have cute little Brewster gopher swag available. All of that's at cupago.dev. That's our website. You can also find links there to past episodes and our contact details. If you want to send us an email, you want to tell me how wrong I am about VPC64 support, etcetera.
Jonathan Hall:If you'd like to come on the show and talk about what you're doing in GO, we'd love to hear from you too. We are always looking for interesting gophers to talk to, and we are looking for information about community events. If you are hosting a meetup or attending a conference or anything along those lines that we don't know about, please let us know so we can include it in the show and let our other fellow gophers know what's happening. Let's end this little break, and let's talk with Shay about some AI.
Shay Nehmad:Okay. We kept this part for after the break because it's a bit more of a discussion, but there is a very interesting Google Group's sort of, you know, email chain going on. What is it about?
Jonathan Hall:What should we do with CLs generated by AI? Lance Taylor, who we've had on the show also we say this so many times because everybody's been on the show. He asked, do we have a policy about accepting CLs that were written by AI? And in particular, he's talking about, you know, Claude or maybe other tools will say coauthored by Claude Opus blah blah blah as a tagline on the git commit. And the question is like, should we reject these?
Jonathan Hall:Should we embrace them? Should we edit them? What? And, yeah, there's been quite a bit of discussion about this. What do you like to what what's what's been your highlight or your your takeaway from this?
Shay Nehmad:This has been a big I don't know. I've seen it all across open source. Mhmm. One approach by the HashiCorp former CTO who developed Gozi, and I'm blanking on their name right now, was we're just not gonna accept pull requests from people we don't know. And if someone puts in, like, a Slop AI pull request, they're gonna get banned forever.
Shay Nehmad:So, like, the focus was on trust. We don't want, like, low level AI Slop, contributions. Mhmm. Some people are, like, on the other end, and, you know, yesterday I was in an AI meetup in San Francisco with the most, like they called it, and, you know, I'm cringing here in the video, Jonathan can see it, but I hope it carries through voice as well. Like, a guy a person called Den from Anthropic just sitting on stage in front of, like, a 100 people and, like, of course, I'm AGI pilled.
Shay Nehmad:I'm AGI pilled to the max. Like, AI can do everything for me. I don't even look at the code anymore. And you have, like, these two extremes. Right?
Shay Nehmad:One one extreme is, like, I will only accept contributions from people I I know and trust, and the other one is like, oh, code doesn't matter anymore. We're operating on, like, a higher abstraction level. It doesn't even matter. I don't think anybody is saying don't use AI tools, period, anymore.
Jonathan Hall:Some people are, but, yeah, that's not what this thread is about.
Shay Nehmad:Yeah. And there are two sides to this debate, I I think. There's software engineering concerns and legal concerns, both of which Ross Cox in his, like, really, really great answer, which, like, could have been a seminal blog post. It's really funny that it's, like, on Google Groups, which is, like, the most non Twitter y or Redditor y space, I feel, but it it really reads like a great blog post breaking them down. I have nothing to say about the legal concerns.
Shay Nehmad:I have my moral, opinions on it, but that train has left. Like, the AI labs managed to steal all content and get away with it. We're done. That we're after that. That door has been opened.
Jonathan Hall:Yeah.
Shay Nehmad:I don't think any legal concern is is even relevant or interesting at this point, just because it's too late. We could have done it in 2023, but it's too late now. I think the interesting concerns are software engineering concerns. Like, what if someone put puts in an AI code review, doesn't read it, and submit it? Is that okay?
Jonathan Hall:Are you asking my opinion now?
Shay Nehmad:Yes. What do you think?
Jonathan Hall:I think that it is asking the wrong question.
Shay Nehmad:What's the right question then?
Jonathan Hall:I don't think it matters if they read it per se. What matters is, is it good code? We should judge the code How do you
Shay Nehmad:determine if it's good code, though?
Jonathan Hall:Oh, that's what the code review is for.
Shay Nehmad:So I don't I don't agree with you at all. What Russ writes here, and I really like, is that from a social standpoint, when I sent a pull request at work, my obligation and my code reviewers expectation was that I did the best work I can do. Sure. And I don't expect them to, for example, spell check my code. Right?
Jonathan Hall:Right.
Shay Nehmad:So I don't know how much of a problem this was for you, but for me, you know, working with mostly most of my life in Israel, most of the people I worked with, English was their second language. So, you know, typing out retrieval correctly, if you're working, like, on an AI system right now, I bet even you might trip up and write it, you know, I before e or whatever. Yeah. You don't expect the code reviewer to find stuff like that. Right?
Shay Nehmad:You wouldn't wanna waste their time with something like that. And at least right now with generative systems, they might get like a wrong assumption or a wrong direction really early on. And then like a huge pull request is built on like some wrong prediction. From a social standpoint, I would expect to get to get the pull request where someone reviewed it first and and checked that it was good. Someone, some person.
Jonathan Hall:I I haven't heard you disagree with me yet.
Shay Nehmad:No. But you're saying it's the code reviewer's responsibility and
Jonathan Hall:No. This time I'm I'm not saying that. I'm saying that the code should stand on its own merits.
Shay Nehmad:Okay. But do but you do expect the person who submitted. If I
Jonathan Hall:have an AI, whether it's one that exists today or a theoretical one that exists in ten years, that can produce high quality code, I don't think it is necessarily the responsibility of the submitter to review the code. If the submitter has full faith in that AI, right, we need to judge the work on its own merits. Now if AIs generate slop, and most of them do right now, at least much of the time, then, yes, the the the submitter so here here's what I'm saying. The person who submits the pull request is saying, I believe this is quality, and I'm staking my reputation on it. And I think that that's what Russ is talking about here, and I agree with that.
Jonathan Hall:If I, as the submitter, believe my AI is foolproof enough or whatever, you know, it's generating good enough code that I don't need to review it first, That's that's me saying I have faith in this code.
Shay Nehmad:Basically, your responsibilities are not lessened when you use AI tools.
Jonathan Hall:Exactly. Right. Right.
Shay Nehmad:But it's very easy when you use AI tools, sort of, like, let go and produce lower quality work. I've dev I don't know. Personally, I've felt it. I've felt, like, the lure of opening a background agent. You know, I'm working on a b to b SaaS cybersecurity application, and then I open the UI and I see, like, a tiny bug.
Shay Nehmad:And then I, like, tag a cursor on Slack and I, hey, cursor, screenshot, go fix it. Then it opens a pull request. I've been tempted to just click merge without even looking at the code. It's so easy. Or when reviewing, you know, not like actually opening the code and thinking about the structure and the context of where it added this specific function or whatever, but rather just look at it.
Shay Nehmad:Does it seem reasonable? Will it work? Click merge. And I don't know. I'm the more one of the more detail oriented developers that I I know personally.
Shay Nehmad:I I used to be, like, a very pedantic person, and and it's very easy to to drop it, especially when the incentive of becoming, like, an a Go contributor is this is something very lucrative for a lot of people. You don't understand this about open source normally, but having Go standard library contributor on your resume, especially if you're, like, getting started out, a lot of people are that's their recommendation for getting your first job. Right? Oh, just find some open source project and contribute to it. Mhmm.
Shay Nehmad:And then you end up with a lot of people just, like, changing stuff in a read me uselessly just to get that on their resume because they're looking for a job. On the other side, the maintainers have to deal with huge mountains of, like, sloppy hours. So I think the bar is higher. I think you can't just say, I trust that this AI generated code is good enough. I think you must self review it.
Jonathan Hall:I agree with that in practical terms, but in in on principle, I think we still need to judge the code on its own merit. I mean, what whether the slop comes from AI or from Stack Overflow copy and paste or whatever, slop is slop, and we need to judge it based on on its merits, not based on the fact that per se that it will just because it comes from Stack Overflow doesn't mean it's actually bad either. Right? Maybe you copied some great stuff from Stack Overflow.
Shay Nehmad:But but I think there's a longer game here. Like, if someone comes to contribute to the Go library, every contribution on its own needs to be measured on its merit. But because Stack Overflow is actually a really good example, in my opinion. Because I used to, contribute to Stack Overflow, like, go while on the bus hunting for answers and trying to answer them. Not because every single question on its merit was interesting to me, but because there was, like, a longer game of, oh, you start contributing to Stack Overflow and then you gain more reputation and then you can get more tags and you get like, the community recognizes your contribution in a more long term way.
Shay Nehmad:I think of, I don't know, a Cup O'Go episodes, think of them as the same. Like, some of them are are seminal and very important, but the actual important thing is that you and I are doing this in a continuous way. Mhmm. So people listening might trust our our voice more. You know what I mean?
Shay Nehmad:Just because we do it on a longer term. When you see someone contributing and they put in a lot, a lot of effort and you see it, they typed out things manually or maybe they post a livestream of them of themselves working on it, then you see that they worked really hard on it and the result is subpar, I would be inclined to help them and get the result to be good. You know what I mean? Because I appreciate their effort. The the process is interesting because there's a longer term play here of, okay, they'll come back next year and contribute more and more, and maybe eventually they will become a core contributor of the compiler team five years from now.
Shay Nehmad:Sure. Whereas if you just judge on merit every time, you're like, oh, okay. I'll just reject this person's contributions because they're subpar even though they worked really hard on it because AI lowered the bar for everybody can submit subpar pull requests.
Jonathan Hall:So I I don't I don't think we should let AI lower the bar. That's not the point at all. I I I don't Yeah. I don't think so. I do wanna call out there's a comment here.
Jonathan Hall:I'm gonna I'm gonna disagree publicly with Rob Pike probably for the first time. Oh. Rob Pike, one of his comments here is just exactly what was written by the AI. That matters. Was it the code?
Jonathan Hall:Was it the CL description? And who has responsibility for maintenance? This is a slippery slope. Be careful on your first step. I recommend simply saying no.
Jonathan Hall:So I don't disagree with the whole thing, but I do disagree with his opening statement. What was written by AI? That matters. I I don't believe that it does matter. I do think the question who has responsibility for maintenance matters, but I don't think what did the AI write is what matters.
Jonathan Hall:And and to to elaborate your or or to to call out again what Russ says later on in his response, we use all sorts of tools, whether they're code generation tools or not. You know, code generation tools are not new. They're usually deterministic up until the AI era. But just because you have a generated code because just yet you generated a code from an OpenAI spec, for example an OpenAPI spec, for example, or from gRPC or GraphQL spec, that doesn't make sloppy generated code acceptable, does it? You still want that generated code to be of high quality.
Jonathan Hall:And who's responsible for maintenance is still a valid concern in those cases. So, yeah, I don't think it matters what was written by AI. What I think matters is, is it good code, and is it maintainable code and all of those things. And, yes, there is a different social responsibility or burden or opportunity when it's written by, say, a junior developer that we want to mentor and bring up to speed versus written by a sloppy AI. You know, if it's a sloppy AI, just reject it.
Jonathan Hall:If it's a junior developer, we want to nurture them. So there is a difference in that regard. But as far as should we accept the code or not as it is, I don't think that matters whether it's generated by a human or by AI or any other means.
Shay Nehmad:It's written by a human and generated by AI. But, yeah, I also think it's again, maybe I'm just a San Francisco brained, I think it's too late. Like, would be surprised if less than if more than 50% of the code in the recent change of this is not AI generated. And finally, there's the one tiny issue of co authored by, which I really like. So I don't know if you noticed this, but if you use Clot or whatever, it adds a little co authored by ClotCode or co authored by Cursor.
Shay Nehmad:I don't like these in my in my private projects, but I didn't super mind. But I really like I I Russ really crystallized why I don't like them, which is I'm gonna quote, as far as I can tell, the only purpose of this line is to provide free advertising to the companies making coding tools. Let's not. And that I can totally agree with. Just removing these lines is is pretty smart.
Jonathan Hall:So I'm gonna say that I I the exact same thing resonated with me. So I took the time to figure out how to disable that in Clubcode, and it's no longer doing that for me as of this morning.
Shay Nehmad:Alright. Well, I think it's people should just read it. It's a very interesting discussion. Maybe we can get Russ or someone else from this discussion to dive more deeply into it. But the link to this, like, discussion is on the show notes, and it's a really nuanced take.
Shay Nehmad:I think just reading it is really, really good. Again, Russ has a way to write that just cuts cuts through the noise and writes really, really clearly. I love it.
Jonathan Hall:I agree.
Shay Nehmad:Well done.
Jonathan Hall:Thanks, Russ.
Shay Nehmad:Alright. So AI in changelist? Probably yes. AI advertisement in changelist? Probably no.
Shay Nehmad:That's it for this week, folks.
Jonathan Hall:Yeah. AI is the solved problem now. We can get on with our lives.
Shay Nehmad:Program exited. Program exit up. Goodbye.
Creators and Guests
