678: AI Hype, Browser Journey, and Content Creation Incentives  ## Metadata - Author: [[ShopTalk]] - Full Title: 678: AI Hype, Browser Journey, and Content Creation Incentives - Category: #podcasts - URL: https://share.snipd.com/episode/1b56c339-1175-4a0b-a63a-0d31c6e2c09c ## Document Notes ## Highlights - **AI Hype Is Emotional, Not Always Transformative** - New AI releases create a hype cycle where improvements feel incremental but emotional reactions are large. - GPT-5 feels like better contextual understanding rather than a magical leap, per the hosts. Transcript: Dave Rupert I'm changing fire. Fire ship makes a funny Simon Willison says something smart. So it's currently at time of recording, we are in the hype cycle, you know, like new thing came out, you know, iPhone with better camera takes better pictures. You know, that's, that's where we're at. You know, for me, it's like, does it better understand how I'm talking to it? Does that make sense? Like, like in my, does when I type, does it understand what I'm getting at? And does it feed back information that is a more accurate be in my wheelhouse, you know? So, yeah. Chris Coyier Cause gosh, it's also complicated, you know, like it's very emotional. When you just type a paragraph, you're not giving it any context. But we can use AI in all these other circumstances where context is abundant and it automatically happens and it's just gunning everywhere. Dave Rupert Yeah. Chris Coyier Have you heard of this idea of context rot that's kind of out there? Dave Rupert It's this idea where if you give it more context, the models actually perform worse. So like if you have a large code base with millions and billions of lines of code, it actually performs worse on tasks as the context gets bigger, which is weird or interesting because The opposite, I feel like I've heard the opposite for like five years. Oh, you just got to give it more context, write more paragraphs, do more things, you know. But apparently at a certain point, if you give it way too much context, like my entire code base, it starts tanking. And so anyway, that's an interesting problem, I guess, problem space. That's an interesting problem space. One thing that GPT-5 or OpenAI's thing has is, you know, ask it to code up a website, and it'll use React and Tailwind because that's what they all do. And then you hit preview, and it'll spin up a working web app. I don't want that going ([Time 0:02:00](https://share.snipd.com/snip/8d76a48b-24da-4d44-b892-a6256ec308ea)) - **Generative Code Creates Reviewer Burden** - Chris tried Claude Code on a company mono-repo and it generated a lot of files and config quickly. - He discarded most of it because he couldn't confidently own code he didn't fully understand or want to pass to teammates. Transcript: Chris Coyier Yeah. But so I, I spin it up and I was like, go look, you know, take a look around, you know, a, right when you do it, the first thing it wants to do is make a clod that MD file, which is like, I don't Know, must use some different model. If it figures out all kinds of crap about your app. And the first time I ever saw this was the sidebar augment that you put in VS code. It's like a copilot competitor. It did this too. The first time you use it, it like goes into explore mode and gives itself a chunk load of, of context about your app way more than your paragraphs. You're typing in it. It knows a ton about your app and you know what else it's got. This is crucial. I think that has your dot get folder, which is every commit you ever made in your life. Oh, whoops. Think of the, think it, well, sure. But think of the context it has there and how you talk and work and what's important to you and what commits look like in your company and stuff. I think that's crazy. Okay. Dave Rupert Well, it might have some API keys in there. Okay. Like, whoops. Chris Coyier Like, okay. Yeah. Sue me. Yeah. I don't know. I I'm talking on a turn almost. I know it's right on disc, so it's sitting there. I assume it's using it. I have seen it run little Git logs and stuff to figure stuff out, so I just assume that it's doing that stuff. But anyway, it figures it out for me, but it touches like a zillion files. And the way we generate Nginx config is special and it figured it out. And I thought it did a pretty good job, but then it didn't quite work. And then we had to adjust some go server code to, and it gets done. And I'm 30 minutes into this freaking doing it. And I think it's done, but then I'm like, I'm throwing it away. I'm throwing it in the garbage I cannot ask Alex to review this code that I don't even fully get, and I'm not really responsible for. I won't do it. I can't do it. It's like it's turning my 30 minutes into his two hours. ([Time 0:07:21](https://share.snipd.com/snip/3ff5a541-50ba-4f96-bc94-99313bf66374)) - **Require Human Review For AI Output** - Validate AI-generated documentation and make human review mandatory before merging. - Treat PR reviewers as quality checkers, not sole QA, and require author validation first. Transcript: Dave Rupert Just fully documented. And somebody figured out, hey, I can just say, can you generate a readme? Like a proper full readme? And it was like a – and that's super helpful just for like we just need more readmes or like usages how to use it you know not just beyond storybook and so uh in kind of the context of the storybook Stories and the you know the the ts doc like it could actually do a good job um but if my my request was can we make sure these are human reviewed before pr because or like, you know, validated, Make sure that it's all correct so that the PR reviewer isn't in charge of quality assurance. Like they should be from like a, you know, just things going in the repo are correct sense, but like, you know, like slopping and making it somebody else's problem is not the, the culture We want to build internally. So that's kind of, so I think there's sort of like this ethics of like, you know, like I can generate fast and, you know, fix my server problem in 30 minutes, but who, who am I passing that Problem to somebody or, or what? So I think that's a good concern. Chris Coyier Right. It's almost like a, I don't know, you want to say, even if you used AI, do you have as many hours into this as you expect it's going to take to review? Like, isn't there, that was almost a Simon Williason thing too, was that like, I don't want to read something that takes me longer to read than it took you time to create. Yeah. Which is some kind of, you know, like, yes. Oh, it almost feels like, of course, now to me. Dave Rupert Yeah, yeah. For sure. Well, yeah. In this situation, it was just something we didn't do. ([Time 0:10:06](https://share.snipd.com/snip/f0604da8-65b1-4328-be26-00c697ac9eee)) - **Contextual Tests Beat One-Liners For Real Tasks** - Short, context-free prompts (the 'blueberry' tests) are useful for model comparison but miss real-world complexity. - Real value emerges when models ingest full repos, histories, and context to reason across tasks. Transcript: Chris Coyier Of what I think about is when a thing drops like this, there are people that do the blueberry thing. And you're like, okay, you got it. You tricked it. Good for you. You gave it one sentence. I'm like so less interested in take down yeah just just isn't it way more interesting to think what does this model do if i give it my entire repo my entire git history i'm asking it to do Complex tasks for me what how does it reason about them? That type of stuff feels like a thousand miles different from how many blueberries. But it's not without some merit. Because I watched a Simon Williason video where he was like, what happened in the last year in AI? And he's like the best guy in the world to explain it. And he does have a one line prompt, which he uses to compare models, which does seem kind of fair, actually. And his is like, have you seen this video? It's like he types in like, do a, not a penguin, a big pink bird. Pelican riding a bicycle. Yeah. Yeah. It gets better and better over time. And I think that's going to become more and more of a compelling prompt as you watch these models get better and better. It does end up being a decent proxy for it. So it's like, sure, you know, sometimes the contextless one line prompt is a little interesting. But for me, I'm like, I want to know what does it do in these like 30 minute agentic blast throughs of asking it to do complex is like very different. And I also don't mind that it's just incrementally better, even if that's what it turns out to be is like GTP five is a little better and Claude 4.4 comes out or something and it's a little Bit better and they just get a little bit better it seems okay to me i mean it took and i hate to every time we mention this i feel like it does still bother me that it has to just chew up the internet With no permission whatsoever to to to get these gains so i'll always caveat that still don't love it but ([Time 0:12:10](https://share.snipd.com/snip/98ecbb29-4625-427b-82a8-09571f3a422c)) - **Browser Swap Misplaced Blame** - Chris left Orion for a while because of a login issue he believed was the browser. - He later discovered the bug was in CodePen, and praised Orion for interesting features like built-in VisBug. Transcript: Dave Rupert That's cool. My son dyed his hair. I'm not going to join him on that one. It's a little too midlife crisis-y. So I'm not going to do that. Chris Coyier Well, I'm still on my browser journey a little bit. Tell me about it. Yeah, well, I said I was using Orion by Kagi or whatever, because I kind of liked that vibe, even though I'm not like the biggest WebKit fan. I like the existence of WebKit. WebKit's fine or whatever. It's just that I don't like the dev tools very much and yada, yada, yada. But at one point I was like, I got to leave Orion for this one thing that I couldn't get logged into CodePen. It had some weird handling of cookies or something. Like something about our system in Orion just was broken, and I could not figure it out. And then, sure enough, like the week after I have given up on Orion, it was totally our bug. Dave Rupert Whoopsie doopsie. Chris Coyier Like a webkit bug? Like a Safari bug, basically? No, no, no. It was our code pens handling of of what was going on was wrong and it was a little niche but it just it's like if you you know start your session on this very specific page that you know me As a heavy user almost always start on it made a cookie in a weird format and then so it didn't let you log in but i hadn't really it would have happened in any browser. It wasn't just Orion anyway. So I might go back to that too. I just didn't, I think I called out that it had, there was some Orion, Orion was too buggy to use and it was not Orion. It was me. So sorry, Orion, you're still a pretty cool browser. ([Time 0:17:10](https://share.snipd.com/snip/4d1aeddf-765a-4af2-a846-78ed131db2b1)) - **Use Stable Browsers For Daily Workflows** - Try alternative browsers like Edge when experimental features in Canary are cumbersome. - Prefer stable builds for compatibility with tools like Riverside and fewer disruptive updates. Transcript: Chris Coyier So I'm basically on, I went to Chrome Canary and I kind of regret it now. I went there just because they have a cool, I've mentioned this too many times, probably, but there's like a command flag to do split tabs. And I kind of liked how they did it. But you have to open it with the flag. So if you just like click it in your dock and open it, it's not, it doesn't have it, you have to go to the command line and open it that way and pass the flag at the command line. And then it opens up with that ability in it. It's not even just a flag that you can turn on in the UI. Every time you open the app, you have to do the flag. So if you close it and then just click the icon again to open it up, you don't have the feature anymore. And it's just too, like they did that on purpose. They're like trying to make it hard to use because they want you to like, you know. Dave Rupert Can I interest you in a Edge browser? It has this feature by default. Does it really? Chris Coyier Yeah. Dave Rupert Yeah. Okay. Chris Coyier Don't know what this shortcut is that could also impact the uh yeah i mean of course that's what i liked in r because you option click anything and it does the thing all right i'm downloading Edge right now we'll see how it goes yeah try it out try it out anyway there's um that is a feature it has uh by default I believe. Dave Rupert So unless I'm leaking. Chris Coyier Yeah, I never gave Brave a big shot because I just don't like the Brave leadership all that much. And then the yeah, anyway, but I kind of regret Canary because it was about three times a day. You get this big thing and it's like discord, even stable discord. They like ship code every day. ([Time 0:21:15](https://share.snipd.com/snip/eda48d9c-7ab7-4495-acf5-0103b0b019e3)) - **Human Voice Still Wins Over Raw Answers** - AI may replicate factual answers from sites, but original content offers personality, context, and community value. - Being known and producing unique perspective still provides career and audience benefits that AI can't fully replace. Transcript: Chris Coyier Oh, have we not? Do we not do that? That's cool. That seemed to ring a bell. It kind of gives AI. Dave Rupert We're back on the AI. But you were responding to a syntax episode, right? Yeah. Chris Coyier Well, it's always flattering when you hear your name on a podcast. And it was an episode of them where they were talking. Wes was mentioning, and he specifically is like, I wonder what Chris thinks. I'm like, oh, cracks knuckles. I will tell you what I think. Because Wes made the point, as a content author, the business is getting shaky. Dave Rupert Right. That was sort of his his vibe. Chris Coyier You know, not that it was ever like a rock solid, flourishing, easy way to make money. But sure. Yeah. Dave Rupert Right. Right. His comment was kind of like, you know, you know, you spent a lot of time on CSS tricks, put up all that information and you got out and maybe you got out at the right time, you know, just from Chris Coyier A, just from a, yeah, he literally said that, you know, like, or he kind of walked it back, but then was like, yeah, kind of, kind of the right time. But he was talking about like the right time as far as like AI is concerned. And like, do we even need this information anymore? I mean, whatever, the AI has sucked it up and now they can answer the same questions. So why would you go to CSS Tricks for an answer when you just don't need to anymore? They've already sucked up all that information and can answer it for you. Well, I don't know. That's why I always thought, I said this many times running CSS Tricks, that I wanted people to find information there, and that's great, and answer questions. But I also wanted people to read it like a newspaper. You read it because it's the local rag for the thing that you're interested in. And I wanted to make it interesting and have a little bit of that in there. Not really journalism, really, because I just didn't adhere, but a little bit. Dave Rupert Think it needs like i tweeted about height stretch or sorry i blue sky skated about height stretch and it popped off people didn't know i didn't know people wanted to figure it out miriam Chris Coyier Made a video it's awesome that would have been perfect fodder when i was running css tricks to make the little post about it or whatever. Dave Rupert It's just like a, oh, dang, that's cool. Like, it's a compendium of cool things that are out there and in the evolving language of CSS. Chris Coyier Totally. And I'm still, that's the thing. I hate talking about it too much like the past tense because I'm like, I still do this, you know. Dave Rupert Still do this, yeah like i'm out of my uh what is it the where you got like uh the the no non-compete is over and i am now at a friend of masters blog a lot yeah well yeah i helped that blog out Chris Coyier And right there because it's like i have some muscle memory for wanting to write right right and i write the code pen spark every week. And I have a topic. And I put links and stuff in there. I do this freaking podcast with you. Every week, yeah. I work on a tool to help other people make demos of stuff. That has a podcast as well. Yeah, dude. I have lots of stuff. So anyway, I'm not like done making content. But those things are a little easier to look at and be like oh yeah you do code pen stuff because code pen is a business that like has pro members and stuff and it's basically marketing for That business well that's a part of the story then okay yeah you know west's own the syntax is owned by by century century yeah and it's like well you're producing this content because It makes money because of that. But before that, it made money in other ways. There's always incentive. It's like, follow the money a little bit, as gross as that is to sound. Can you extract money from this even in a world of AI? And it's like, so far, it's like, I don't know, it doesn't seem like it's changed that much. And even if you don't follow the money, follow the incentive. And the incentive could just be, I like hearts. I like it when people click the like button. That does it for me. A lot of people do things just for the validation. I mean, ask a painter, you know? They're probably getting very little money from what they're doing. They're doing it because they like it and they like it when they make other people happy, you know. And every interview with a comedian is like, man, when I make people laugh, that is the good stuff for me. ([Time 0:28:41](https://share.snipd.com/snip/a3a73502-4707-4ea6-ab88-133b54a22f71)) - **Incentives Drive Creation Beyond Money** - Follow incentives, not just money, to predict what content creators will produce. - Validation, community, and learning motivate many creators beyond direct monetization. Transcript: Chris Coyier Can you extract money from this even in a world of AI? And it's like, so far, it's like, I don't know, it doesn't seem like it's changed that much. And even if you don't follow the money, follow the incentive. And the incentive could just be, I like hearts. I like it when people click the like button. That does it for me. A lot of people do things just for the validation. I mean, ask a painter, you know? They're probably getting very little money from what they're doing. They're doing it because they like it and they like it when they make other people happy, you know. And every interview with a comedian is like, man, when I make people laugh, that is the good stuff for me. Yeah, yeah. And they want to make money doing it, but it's not the only reason they do it. Dave Rupert Yeah. You know, I think there's other reasons, too. I mean, just, I don't know. For me, it's like, I have ideas in my head. They have to come out or I think about them too much, like almost like obsessive or like some sort of, I don't know, OCD sort of thing. But, you know, like that can be a value to getting things out, you know, like, you know, getting your name, you know, you're not, maybe the, the, the ability to become famous writing content Is done. I don't know. Like, I'm just hypothesizing. But like, you can be known, you know, like by by a like minded group of of Internet readers or tech enthusiasts. You know, you can be known. And that's good for for the bottom line. You know, that's good for you, for your career, even if it's not career-based. Just knowing, being known is helpful. Chris Coyier It probably will continue to be. Yeah. Dave Rupert And that can backfire, too, because if you're a level 3000 A-hole, now I know that and it's documented so that's good but you know if you're you know level 3000 nice person shoot that's Great if you seem like you know you kind of vibe out that you're an interesting person to uh work with i'm sure more people would read your blog and think i could work with this person so Yeah, yeah. ([Time 0:32:19](https://share.snipd.com/snip/35973b02-5dbb-4195-a770-c65cc29f8cdb)) - **Ship Small, Useful Developer Tools** - Use streaming and small public experiments to iterate and sharpen ideas without chasing massive growth. - Publish small starters (like a CSS starter) as copy-paste tools for your own projects and community feedback. Transcript: Chris Coyier I've been streaming, you know, as you know, because I published, posted in the little sharing channel. Almost every day I get on for minutes to an hour or so and talk about stuff. You know, it's just proving that like, gosh, you can just talk about anything in tech. There's a million things. And this whole week, for one, you know, less than 10 for sure in the stream. Like I'm not, it's not growing. It's, that's just what it is. That's the popularity I have as a streamer. And I'm cool with that. I'm getting better at it. I'm getting better at video generally. I don't know. I just, to me, it's like whatever. Obviously, I could never, I can't even pay my electric bill with what I earn from it. In fact, if anything, it costs me money to do. Dave Rupert Well, yeah. But if you're going to fart on cam, you want that in front of 10 people, not like 1,000 people, you know? Chris Coyier Yeah. Anyway, I wouldn't mean to talk about that so much. But obviously, like, I'm incentive. I like doing it. Yeah, right? Like, if I hated doing it, I wouldn't be doing it. So this week, I was like head on my list of things. And I'm like, yeah, let's just do it. It's to make a CSS, I'll say reset once, but the, I, my goal of this thing is not to make it a CSS reset, but to make my own, I've never done it before. Make my own CSS starter, you know? Okay. Yeah. Maybe I'll publish it. But those seems, those things seem to get a little traction when you publish one, you know, anybody puts out the, like, this is my reset. People are like, Ooh, I'm going to use, know they i don't know what it is it's no catnip for a certain kind of eric meyer and andy bell bought lamborghinis that's like how they fund their I looked at andy's yesterday there's some good stuff in there but that's what had it stuck in my mind is because when andy did it originally maybe we did it on this show but i like i just like Nitpicked it just like for fun not because i think andy's wrong or bad at this or anything but i'm like nope nope nope i would do it differently no i'd do it differently here you know like Just little little css i don't know yeah manners or something so i'm doing my own and i started with andy's but i wanted to look at everyone in existence i started with eric myers you know And be like is there anything in here that you'd like still do which is pretty much no yeah no oh really yeah not really interesting there's one line that's a table as border collapse or Border or that one that like makes the borders on table yeah Table down into one border. Uh, border collapse, collapse, right? Border collapse, or table collapse, collapse or something. Dave Rupert Yeah. Nah, I'm going to find it. Chris Coyier Collapse, collapse. Um, I like that line. That's a good one for tables. I think one. Yeah. That's it though. Well, cause like border collapse, collapse. It is border collapse then okay yeah but you use it on tables it only matters on and you put it on the table element not on the cells for whatever yes yes half of the entire reset is like listing A ton of common elements and then just saying like margin zero padding zero on it you know as a way to like avoid using the star selector or something but i don't even do that on the star selector In this reset i'm like so far i'm like nah like i'll rip it off a few things where i don't like it like the body but i will i'm not going to do it for for headers and stuff like i think the fact Dave Rupert That they have a little not even for border spaces you're going content oh no no no no, no. Chris Coyier That's the top thing. That's in every reset ever. Okay. As soon as that was, like, there's none that do it. Although, mostly you see it as star comma, star colon colon before, star colon colon after border box. So you get the pseudo elements too. But like, there's a little variation of it where you set it just on the HTML element, box sizing border box. And then on that selector I just described, you say box sizing inherit. And it's a little tiny bit better. Dave Rupert It's just a twist. Yeah. Chris Coyier It's a tiny twist on it. And I'll tell you why. Because it's like if there's an area of the site where you, for some reason, want to turn off that box sizing behavior. Now you can just do it on like the div or whatever. That's the parent of all of it. And it turns off that behavior for the whole div and inside of it, which is usually what you want. Whereas if you do the star selector way only, you can't like it, you turned it off for one element and then it's immediate child has the wrong box sizing again you have to like this is just A way of like you can turn off that that box sizing behavior on a whole dom sub tree easier and it's like it's not a big deal but it's just a tiny tiny bit better and it's like if it's going to Be in a style sheet like this that you intend to like the theoretically you're using on lots of stuff then just do the better version yeah anyway i'm not even done with it yet i'm just picking Out all the little stuff that i like kind of going adding stuff that i like not adding stuff that i don't like and it's a starter it's not it's not intended to be like npm installed and just Used as like version 3.8 or whatever where you just, it's not that it's just like, I'll link it up my pens. I'll probably copy and paste it into a new project and monkey with it. It's just a little starter pack, a little reference. Dave Rupert Copy, paste in. Yeah. Yeah. Yeah. That's ([Time 0:34:41](https://share.snipd.com/snip/16864e1e-4b99-44ba-ad34-626c2269894e)) - **AI Could Undercut Future Knowledge Creation** - If incentives to produce original web resources vanish, the training data pool for AI may shrink and knowledge sharing could decline. - The hosts worry this devaluation could reduce future human-driven discovery and documentation. Transcript: Chris Coyier Long-term implication that I'm not sure we spelled out here is like, if the motivation goes away, if there's no money in it anymore or, and, or all the incentive is gone for doing this Kind of thing, then people don't do it. Yeah. Yeah. Maybe like, not to be like the next CSS tricks, but let's, I'll just use that as a useful proxy. What if nobody ever does that again for new technologies or new things, because there's no incentive to do it. Nobody wants to make that kind of website anymore. That is this possibly a fount of knowledge about new stuff and people sharing and whatever. Because that's what makes the AIs smarter to train against. But if there's no AI to – there's no website to slurp up all that crap from because there's no incentive to do it, then we all get dumber instead of smarter. That's the kind of the – I think the extension of that thinking that is worth spelling out. I don't know that, I don't think that's a particularly big risk somehow. I just don't, I don't know why. Yeah. Dave Rupert Yeah. Chris Coyier Hard. I couldn't really comment. Dave Rupert I don't know. Yeah. Chris Coyier I don't know. People just talk, they just type stuff to each other and whatever that'll, there'll be some way to get that slurped up and no but i'm sure the risk of people who were doing this and it was Dave Rupert Part of their livelihoods you know like became a part you know like i do think there is a risk that you know gets devalued so much because i can just ask an ai for the same thing. ([Time 0:41:12](https://share.snipd.com/snip/7fbdde57-9352-4bae-8b7c-0300c1ba000e)) - **Item Flow Seeks Cross-Layout Consistency** - Item flow/masonry proposals aim to unify layout needs across grid and flex, offering cross-layout primitives. - A shared 'item' concept could enable features like balanced wrapping and masonry without duplicating layout models. Transcript: Chris Coyier Well, there's an interesting blog drop on the WebKit blog, which we recommended in our RSS show yesterday. Yes, yes, yes. The second part of that item flow thing, which was, I think, you know, Jen meant to do a second part right when the first one dropped. That item flow was a big bomb drop from them of saying, like, look at how cool this could be. This has to do with the masonry discussion. Yeah, yeah. We had item flow, that could do masonry, but it could possibly do lots of cool stuff all across the board, which is the kind of like cool, interesting thinking that going slow in the web World often leads to. I thought there was some kind of fun drama between part one and part two here is Google's like, we're just going to ship our version behind a flag. Yeah, right. They just did that, which is behind a flag. That's what flags are for. So it's hard to really slap their wrist for this. But there is a little bit of like, well, if lots of people try it, it's going to influence their thinking that this is the one that they like, theoretically. And it did not mention like ah maybe it did mention i can't i shouldn't say that but obviously this item flow is like has some legs you know people are a little excited about it but there I think people are also scratching their heads a little bit like who's who's got the most sway here like if apple says this is absolutely the way do they have it in their soul to just not Do it if it's like somehow not the winning choice or whatever i don't i don't know i've never gotten an answer i think you asked somebody from apple like hey are you prepared to abandon This they're not gonna get answer you. Dave Rupert Yeah. Well, that's, yeah, even though my edge adjacency, the Chrome sort of release there, soft release was sort of shocking, you know, just, I guess, just from a, like, I didn't know where We were at on that, you know? So I haven't really looked into... Chris Coyier Firefox still has their really old school one implemented too, which was grid template columns or grid template rows masonry. Okay. If you just write that, it just works in Firefox. It's not like they unshipped that. Way to go to Firefox. Really weird. Dave Rupert Yeah three browsers have like a slightly different one good that's what we want um uh the the the did they adopt did edge or chrome sorry chrome adopt any of the like item flow stuff at all I don't think they've said anything about it kind of just ignore, there's an item pack property proposal they mentioned in the post. So maybe that's what they're... Chris Coyier I think they acknowledge its existence, but not like if we're willing to start prototyping. That's just publicly, I don't follow the GitHub stuff where I'm sure that stuff is flowing a lot thicker with people from all these companies actually talking about this. I mostly just read the blog posts that come out, you know, as interested as I am. And as I'm a little, not burned out, but I'm a little bit like, yeah, just whatever. I kind of felt like that way before, like either one of these is okay. I could learn it. I hate to be terribly unopinionated, but like they're both pretty good. So whatever it does sort of last time we did a run through of this. We, I think we agreed that item flow looked like the best option. Probably. Dave Rupert I liked it from a, it floats all boats perspective. Like I felt like grid gets better. Flex gets better, you know, gets implemented. ([Time 0:43:07](https://share.snipd.com/snip/34f927cf-d541-4e5d-9230-838f194abfd7)) - **Choose Interop Over Niche New Types** - Prioritize web compatibility and interop when new layout features emerge. - Favor solutions that 'float all boats' to reduce fragmentation across browsers and frameworks. Transcript: Dave Rupert I'm not a browser engineer. Maybe that that would perspective would help. You know, it's like, hey, if I if I'm a browser engineer and you want masonry and grid combined, now I have to put an if statement inside every single like layout, you know, check or every Single layout calculation and it slows it down by X billion percent, like then now I'm listening. Now I'm like, oh, I don't want slow layout. You know what I mean? Like, so that feels like, you know, we should listen to that and pay attention to that, you know? Chris Coyier So yeah, it is funny how we generally ignore that. Right. We're like, I don't know. Obviously it won't go into a browser if it's slow. Dave Rupert Right. So, like, y'all do pick the best one because I don't know that part. But, yeah, I just, yeah, I think, like, yeah, I just kind of wish I knew a bit more about that end of it, you know, sort of. Yeah. You know, if everything takes a huge hit, but, you know, I don't know if WebKit's like, hey, we have this, then, you know, WebKit obviously doesn't think it's a problem. WebKit and Blink probably aren't super different at this point, but maybe they are. Chris Coyier Pretty hardly. People take this stuff seriously. We have it pretty good, honestly. Yeah. Dave Rupert Do you think we're getting it like in 2025 or are we six, seven, eight? Nah. We'll call it six. Nah. Chris Coyier I'm interested in the, you know, the more interested, I'm always just like the biggest super fan of compat, web compat or whatever they call interop. Interop, yeah. Just rip the interop, dude. I'd rather see that stuff go out than masonry go out. Until I need masonry and then. Yeah, well, like this interop thing, man. Dave Rupert Like, yeah, I was thinking about it the other day. It's like, Chris, off the top of your head, what's the latest JavaScript feature? Chris Coyier I have absolutely no idea. Dave Rupert Yep. Nailed it. So like, you know, what's the latest CSS feature? And you're like, too many to choose, you know, like, you know, height, height, stretch, you know, I don't know. Chris Coyier You know, like JavaScript does evolve, but yeah, it's funny. You look at the list of the year and you're like, Ooh, that is some niche stuff. Dave Rupert It just feels like yeah well just yeah yeah it's like array group or object group by or whatever you know but um but i i wonder if like uh i wonder if there's a yeah um i i guess i wonder if the I guess I wonder if Interop works very well. That's what I'm wondering. And could JavaScript benefit from that? Like, hey, what engines are we bringing? I guess they operate totally differently, though. They have a TC39 spec board, and they don't quite have like hey everybody write your own thing because that's interesting levels that they move stuff up through and all that they don't Quite have like the css shotgun and and reconcile strategy that we have ([Time 0:50:10](https://share.snipd.com/snip/bcf55799-fc71-4fd5-94ca-bbcfb02bee22))