Retreat to attack

So we were supposed to be in San Francisco, but we’re not in San Francisco.

No, we’re not. What a sadness, you know? Well, happy I’m not stuck in an airport or trying to get to someplace, but… For those who don’t know, our government is shut down, and therefore the FAA, and TSA, and all the acronyms A are not showing up A… And so we would have a bad flight A.

Those TLAs, man… A! Show up A, you know?

Yeah, you know what I’m trying to say? Let’s get to San Francisco, A.

Therefore, we canceled our friends, you know? We told them “Hey, don’t come record with us this week. We’re going to be in San Francisco.” However, we are not in San Francisco. So we are here, in our home studios, ready to Friends… And we’re going to do kind of a, you know…

A grab bag.

A director’s cut. Changelog News director’s cut.

Deep behind the scene… I would say extended cut.

Yeah…

You know, extended director’s cut, you know?

Take the 10 minutes, make it 90.

Criterion edition.

60, 90…

Steelbook edition. You get the leaflet –

Do you watch those director’s cuts when you’re going to go deep, and he’s whispering “Here’s where I should have done this”, or “I was trying to capture why…”

I would say on certain movies. Certain movies I do want to go deep… But generally no. It’s just not my thing.

I used to do it back when I had odious free time, and was much more of a movie fan than I am now… And I liked them, although they do get pretty boring.

Yeah. I mean, it’s not as good as the movie, right?

Right. It’s like “Well, I’m looking at the movie, but I’m listening to a guy talk about the movie, so…”

Yeah. Sometimes you get nice extras, I would say, about the plot, or the twist, or the storyline, that you don’t get in the movie… So in those cases, I do enjoy it. But that’s not always the case.

Okay. So after this show, let us know - do you like the director’s cut of Changelog News, or would you prefer just the 10-minute “Give me the quotes and get out of here, man. Get out of here”?

We shall see.

So there’s some good stuff in this week’s news, I thought… I looked at the list and I was like “Dang, dude.” Not that I did well, but like there’s lots of good stuff out there… My favorite of which I put right in the top, which is the Alive Internet Theory.” As you know I’ve had – I don’t know if you’d call it an obsession, or… I’ve had frequent references to the dead internet theory. I feel like I’ve been experiencing the dead internet of late… Do you feel it, Adam? Do you feel the dead internet, or is it just a theory to you?

You know, dead internet is essentially like bots commenting, and like bot content everywhere…

Bots commenting, bots creating, bot blog posts… AI slop, just everywhere.

I think I would experience that more, sadly, in the TikTok world. I feel like it’s on repeat, I feel like it’s all about product placement, I feel it’s all about like selling stuff in the TikTok shop… I feel like there’s creators who just literally go into the shop and get incentivized to shill things in there… They may like them, but I feel like it’s just like the only reason you’re making content is because you can get some version of that product sent to you for free, and then you can schlep it and make some money. There’s not a real relationship there, so… That’s not a bot, really.

It’s more like zombie internet. It’s like a person, but they’re driven so much by the algorithm that they’re like the undead.

That’s right. Yeah, that’s true. I would say not really… Not really.

Okay. Now, I do think that the AI slop is coming to the world of video and audio. I think it’s behind where it is in prose, because of course, the text models were the first ones to get good enough to trick us a little bit… But with Sora 2, and that whole deal - I’m not sure if you were… I didn’t join the social network for Sora, but I saw all of the cross posts elsewhere of what people are making with it… And that one’s actually okay. I’m not the hugest fan of Open AI as an org, but that was at least an attempt to be transparent about like “This is a social network for AI slop”, and that’s what it is. And so you want to get sloppy? Let’s go get sloppy together. But let’s not act as if we are real. Let’s just go ahead and embrace the fact…

And there’s some – I mean, there’s funny, interesting, cool stuff that you can do, and so I’m not against the content necessarily… Especially because you know what you’re getting. You’re like “Oh, here it is.” But when they’re on the YouTubes and on the TikToks and the Instagrams trying to act as if they’re humans, that’s where I get a little bit offended.

So I definitely have been feeling it of late, but I was uplifted by Spencer Chang. And I did reach out to Spencer, asked him to come on the show. I haven’t heard back yet; this was just yesterday. So hopefully he’ll come on and talk to us, because he’s doing all kinds of cool stuff… And he created the Alive Internet Theory; of course, an answer, a counter narrative, and a very cool website to the Dead Internet Theory, which basically states his premise - which I did quote in News - that the Internet will always be filled with real people looking for each other, answering calls for help, and sharing laughs, even in the midst of the arguing. Now, there’s lots of arguing, for sure.

[00:07:52.27] And the cool thing about this website is it basically went down through the ages, it scrolls the timeline from 2004, and updates kind of the style of websites based on where you are in that timeline, all the way up through present. And you can click a time period and then start the experience, and it’s going to now load up with audio, which probably doesn’t translate, but it’s gonna little load up – oh, it’s too loud, actually… I’m blasting myself. I had to close it. [laughs]

I did that. I had to close it, too. I tried to hit Refresh, but it wouldn’t work, so…

Unfortunately, it loads audio very loud, of like a digital scrapbook of randomness about humans, and all of our analog, jagged, weird ways have put onto the internet during that time frame. And they’re all pulled from Internet Archive, and of course, the Wayback Machine… And it’s just super-rad, and it made me happy. I was like “You know what…?” It kind of reminded me of the nostalgic feeling we had going back to the Winamp skins…

Yeah.

Remember that show we did last year?

So much fun.

Yeah.

Winamp is the best.

So thanks, Spencer, for restoring a little bit of my hope in the internet, and the fact that - yes, over 50% of new articles, according to a recent study, or maybe it’s right at 50 percent, of all blog posts and articles are now written by AI… And that’s always going to increase from here. But the humans are there, man. We’re there, and we can find each other, and we can talk to each other, and we can care.

Yeah, I don’t know what to make of this, because “Written by AI” is somewhat potentially a misnomer, in the fact that… Sure, written by AI, but thought up and desired by a human. Orchestrated, directed by. And I’m not really sure where I sit with that still yet, because it’s really a conundrum, that I use a lot of AI to do things, and it writes a lot of documentation for like little things I’m doing… And if it wasn’t for that - geez, I would never even show up and do it because it’s just too much time. I don’t have the time for that. So I’m a little torn with that. Where do you sit with that? And the fact that just the blanket statement of “written by AI”, but maybe not the subcontext of “directed by and influenced by and ideated by a human”? The only reason it’s there is because a human was like “Hey, this is cool. AI, write it.” Boom. There it is.

I am totally okay with what I could just call perhaps utilitarian uses of AI generated words such as documentation, such as explanation. And especially when directed by humans who desire that content in order to better understand something. I’m also okay with it in the context of information, such as Wikipedia, and now the competing Grokopedia, which seems to be so far not as good as Wikipedia, but I don’t disagree with the premise of Grokopedia, which is like “Let’s remove as much as we can human bias.” Now, also the models are biased, so that’s why I say “As much as we can.” But let’s let information be written by an unbiased, as much as we can, source… Because we want just the information sometimes, and that’s fine. And so I have no problem with it in the context of those things. I don’t even necessarily call that slop, unless it’s actually sloppy. Similar to the way that I instruct a coding agent to write code for me, and I direct it to write code. As long as I’m involved in that process, and I approve of the code, it provides a use and a value.

Where it really feels sloppy to me and bad is when it’s creative to entertain, to persuade, these things, and it’s – or to SEO… I mean, I try to think of the reasons why they do that; when it’s just content farms powered by AI…

[00:12:18.13] I see that actually a lot on YouTube, where you’ll search for something, in particular around a product… I do some product exploration, like an appliance or something like that… An example is if we’re buying a new air fryer, let’s say for Christmas… If you’re doing this - whatever, listen up. You go on YouTube, you find maybe the Ninja version of it, and you’ve got this other version of it, maybe from Cuisinart, or whomever, and you start versus-ing them… Cuisinart versus Ninja. Well, you’re going to find some good people out there. And I have no idea how they make this their living, to compare air fryers. And then you will eventually find – well, probably pretty quickly find some AI-generated, or certainly not a human. It doesn’t even sound very human when it’s speaking, too; it sounds like a computer voice. And it’s clearly just an infomercial for the views. And it ranks high… And maybe it does get some views, and some of those views perpetuate more views in the algorithm, I don’t know… But those things drive me crazy. Or just straight up – like, you can tell no human was involved in this. Or there was, and it was just just for the views.

Yeah.

Not for the engagement, or the true enjoyment.

And I’m honestly only really offended when I can tell. Right? Because it’s kind of like that comparison, like a toupee, or a boob job. That’s the two things.

Gosh, Jerod…

You only know that a person has one when it’s a bad one. You know?

Yeah. I love your comparisons.

Because if it’s good, then you can’t tell anyways. You don’t even know that guy has a toupee. You think it’s his hair. It’s when it’s bad that you’re like “Dude’s wearing a toupee.” It’s kind of offensive. Silly.

Sometimes you never know, but…

When you don’t know, you’re fine with it. When you do know, now it’s bad. That’s the way I feel about, I guess, these things. It’s like, if you fooled me, then good on you, I guess… It was good enough that I thought a person did it. But when I find out it’s acting like it’s a person and then I realize “Ah, it’s not a person”, that’s when I get mad.

How about this? Maybe a Civic, a Honda Civic wrapped in a Ferrari shell.

Okay.

That would upset me. Would that upset you?

Yeah. You open the hood on that sucker…

Yeah. Like, “What is this? Four-cylinder DH-whatever thing”, you know? I don’t even know what the engines are in Civics anymore, but it was something like that. Overhead cams, you know… OHC…

Or the real tall person in the trench coat, and then you realize it’s like three short people standing on their shoulders…

Yeah…

Yeah, that’s going on a lot… But you know, when it tricks you, you’re none the wiser, so… I think once we get beyond the uncanny valley, where we can no longer detect - and I’m not sure if and when that takes place - then perhaps the dead internet theory will be just an undead thing that we don’t even know.

Yeah. To pull an exact pull quote from, I guess, them, the person who wrote this - what was his name again?

Spencer.

Spencer. Thank you, Spencer. And then I guess by proxy, you quoted his quote, so I’m quoting you quoting him…

Okay, this is a Michael Scott moment.

That’s right. We’re going deep here, a couple layers. You said “The internet will always be filled with real people looking for each other, answering calls for help, and sharing laughs, even in the midst of arguing.” And I think that’s when the internet will always be what we consider, since the ’90s, or whenevers, when you and I began to do dial-up, or it was cool to email a friend kind of thing, you know?

Totally.

[00:16:03.07] Because that’s when the internet is still the internet. Because I want to show up to a place that is globally connected and informational shareable forever, so long as it’s shareable with other human beings that have empathy, have care, have grace, mercy… All those things that the human beings embody.

Sure.

When that’s true, while that’s true, I will enjoy the internet. When that becomes less true or not as true, I’ll probably go on a walkabout, or move to the cabin in the woods that I got reserved, you know?

Break: [00:16:44.03]

Well, let me share this with you then, which is a bit of a curveball, because we didn’t discuss discussing it… But it is in News, it was in the lower three, which get in the newsletter, and they get mentions, but they don’t get actually coverage, which is Mesh-tastic. And it’s in the same vein now of like what if there wasn’t necessarily – or there still was this global connected internet that maybe only robots use to talk to each other, maybe at our behest, maybe not… And there’s no people there. I mean, because honestly, if it continues to ramp up and the uncanny valley stays, meaning it’s just AI slop and I know it, I’m not ignorant of it, I might check out. I might be like “Yeah, no longer surfing the web. Just doing other stuff.” Okay, so that might happen. And there’s some social networks that they say has already happened.

A lot of people claim that X is now mostly robots… Which if it’s true - and I don’t know - that’s hilarious, because one of the reasons Elon Musk bought it was to get rid of the robots. Anyways… Mesh-tastic. So this is cool…

It is cool.

An open source, off grid, decentralized mesh network that’s built to run on affordable, low-power devices. So this uses inexpensive LoRa radios, which - I don’t know what LoRa means. Do you know what LoRa means?

I’m just hearing it for the first time, I think…

L-O-R-A. So let’s find out in real time. Long range. So LoRa just means long range. I already knew it was long range.

Oh yeah, that makes sense.

I just didn’t know that LoRa means long range.

Low range off grid… Is that what that stands for?

No, LoRa just stands – it’s from long range, a physical proprietary radio… Oh, it’s an actual – it’s a proper noun. Okay, so that is different. A proprietary radio communication technique based on spread spectrum modulation. So it’s not just an acronym or a shortening, it’s actually a technology that these things use.

What I’m seeing says “LoRa, L-O-R-A, is a software technique that adapts large language models by training only a small set of new weights.” Is that what you’re tracking?

I think these are probably namespace conflicts, because this is all about communications.

Oh, it does say two different technologies. Here I am way off, okay? A wireless communication protocol, or the thing I just injected, which is the wrong thing.

[laughs] That’s alright, [unintelligible 00:22:10.21] was definitely on topic for most of our conversations these days.

My bad, y’all. I’m catching up.

I’m just glad that this isn’t that. So this is long range radios. And this Meshtastic software, I think you flash it onto specific devices. They have a list of devices that you can put it on… And it sets up basically a mesh network that is not just at your house. I mean, most mesh networks that I think of is like Bluetooth, AirDrop, AirPlay kind of things… But this is like, you connect all of these inexpensive radios, and it can go long range. And they’re not kidding. When they say long range - 331 kilometers long. That’s the record, at least.

Wow.

Using Meshtastic.

This is meant to be used by humans to send messages?

Yeah. You know, it’s like the old carrier pigeons, man… If you’re going to send a message, you’re going to send it over Meshtastic. So this would be like if you and I – I could probably do this with like Nick Nisi, for instance. He lives across town from me, probably 30 miles, which is going to fit inside this 331 kilometers. And he could get one of these radios set up at his house, running Meshtastic, and I could get one at my house… In fact, maybe we should do this. It’d be fun…

Yeah. Let’s try this out.

…as a test, as an experiment. And he would get one set up at his house, and I’d have one at my house, and then we connect our devices to the radios as if they’re Wi-Fi networks, basically. And so we are now meshed. We have basically a LAN, by way of Meshtastic. And so it’s an off-grid, meaning no one else can connect to that, it’s decentralized… We could probably set up radios all around, and they all would connect to each other… And it creates a mesh network that you can run your apps, run your comms, run whatever you want. All encrypted, excellent battery life, optional GPS features… I mean, this is a cool project.

[00:24:22.23] This is meant to be radio-based, not a device. I mean, because they’re mentioning Raspberry Pis. Are they thinking the Pico, I think, in particular? Is it only meant to be radio-based, not like literally IP-based through the internet?

I think that these things can talk to each other via this LoRa technology, which I assume is a protocol, and then they act as simply like routers that your local network then connects to. So these are like bridging local networks via this hardware and software. So I’m sure it can run on all kinds of software, but the hardware is a set list of specific supported devices, such as the Seed Card Tracker, the Heltech Mesh Node, the Nano G2 Ultra, etc. A bunch of specific devices. They have antennas, so you can go for long range… And I assume the 331 kilometers has some pretty stinking big antennas on them to get that far.

And the question is - I see stuff like this and I think “That’s cool, I’m really impressed.” And “Would I go through all that?” If I was sick of the internet, but I still wanted to communicate with people - and I assume that you could probably get like a regional network set up. It’s not just like me and Nick Nisi’s houses, but you could probably have an entire mesh network that’s like a city. Maybe even multiple cities. 331 kilometers is a long way. And then you could just have your own little network that’s just like “Yeah, I can run my iMessage against– you know, I can talk to those people over there in Lincoln, on their iMessage.”

Yeah.

Now, iMessage might have to get out beyond your network, and talk to Apple’s iCloud servers… I don’t know.

That’s what I’m thinking now - like, okay, the network is one thing, but then what can you do on the network? Like, what kind of protocols does it support? Is it only messaging? Is it kind of an API layer in a way, where the network gets created and then there’s an API that you can send messages? Is there a length of message? Is there a certain amount of bytes you can send, or that kind of thing? I think that’s what would describe to me its utility. Because I’m thinking there’s unique businesses that might want a reliable communication pattern whenever the internet is not always available. You might be in spotty areas. I was just on a retreat, and I’ve got T-Mobile, and while there, I basically had SOS. I came in and out of like one bar, or half a bar… And so I wasn’t really trying to use my phone anyways, because it was a retreat; that’s the reason you retreat, to attack… But nonetheless, I still kind of wanted to have my tether to the world, you know? Like, don’t take my phone…

Did you just say you retreat to attack?

Yeah, retreat to attack.

What’s that mean in this context?

Well, I believe… Okay, great. Let’s go there.

Okay.

I believe to truly attack like we have to, as men - I’m a man - I’m called to for my family, and my life, and who I am…

Sure.

I believe to attack, in the most kindest way, the world…

Yeah, [unintelligible 00:27:37.10]

…my problems, whatever. Yeah.

Sure. Yeah, your life.

If I’m going to have the energy to do what I’ve got to do, then I cannot just do it by just constantly attacking. I’ve got to retreat, to recharge, to collect, to reaffirm, to examine, so that I can better attack. So I believe you have to retreat to properly attack the world.

Gotcha.

And I mean that in a positive way, of course.

Yeah, I get you. So when you’re retreating, you don’t necessarily want your phone, but you still want to be connected, just in case.

[00:28:07.18] That’s right. Yeah, I mean, something could happen, you know?

Right.

I need my… I need my fix, okay? My addiction.

What’s the longest you’ve gone without your phone recently?

Honestly, while I was at the retreat, I left my phone in the cabin all the important times. I literally left it there…

Hours at a time. Days.

Yeah, like the whole day.

Nice.

And I did it as a force function, but it was a little strange at first. I kept tapping my pocket…

Right?

…because one of the habits I have as an individual who doesn’t want to lose this expensive device in my pocket is to make sure it’s there.

Yeah.

So I will occasionally, let’s say once every five and a half minutes… I’m just kidding. [laughs] Once at some sort of intermittent measure, I’m tapping my leg, or I’m touching my right pocket, which is where I keep my phone… So pick-pocketers, now you know.

Now you know.

Really easy to get me.

I think most people – you know, most people are right-handed. It’s going to be in your right pocket. Me? You don’t know where it’s going to be. So…

You just don’t know. Back pocket, front pocket, no pocket… In the shirt… Who knows? It could be in your armpit… Sneaky people put it in their armpit, man.

Yeah… Like with a little case that wraps over your shoulder?

I left it back there though, man, and I felt good. I felt good. I was proud of myself, honestly, for doing that, because I think… It’s so sad to say this, but I think in today’s world, for a normal first world person who has access to the things we have access to, the privilege we have access to the things we have access to, to leave this connected device that we pretty much rely upon to navigate… Like, if I got in my truck, I know where most things are at, but if I’m going into Austin to a particular building, I’m probably going to map it.

Sure. Because why not?

Because I [unintelligible 00:29:52.28] remember that.

Yeah, because why not?

Yeah. That’s what I was there for, man.

Especially if you’ve got Waze. You’ve got Waze?

I’ve got Waze, man. I’ve got all the Waze.

I know you do. You taught me about Waze. Yeah, you taught me about Waze.

Waze all day.

The thing that’s nice about Omaha - small town living; we’re a small, big town - is you don’t really need Waze, because the traffic’s just fine, pretty much everywhere you go. There’s times, but for the most part, you’re not going to get a major reroute because of traffic. That being said, they’re so useful, and yet disconnecting from them is so freeing. I had the same experience with my watch, my Apple watch, which finally died… I haven’t bought one since, which kind of tells the story. Because I was addicted to the notifications, and the rings, and the things… On your wrist. And when it died, I was like “I don’t know. 300 bucks… I’m going to go a week or so without it and just see.” And it was like – you get a little bit of withdrawals… It’s not like quitting tobacco, which I’ve also done. That was way harder. But you definitely get withdrawals, like “Oh… Where is it?” You ever look at your wrist and there’s nothing on it? It’s just weird. Like, why do I keep looking at it? There’s nothing there.

And then beyond that, it was – it’s just been pure freedom ever since. I just feel like that was a little prison that I erected for myself and I lived in it. And then I just happened to escape, and I don’t need it. And it doesn’t actually make my life enough better to be worth all of the baggage. But the phone itself is way more useful than the watches… And so that one I’m much more connected to and addicted to. But I have been leaving it on the charger. That’s my recent move. I call it a return to analog, which is like, let’s just focus on the analog, let’s get away from the digital. When we’re away from the digital, let’s actually get away from the digital, and leave the phone on the charger.

This elongates this conversation a bit, but this is a resurgence for a lot of folks to be against the norms. Like, drinking has been a norm for a lot of folks. There’s a lot of people like – I’m now sober. I haven’t drank all year. I quit drinking in January. January 1, I quit drinking. Haven’t had a drink since, and so I’m free of alcohol in my life. I never plan to go back. And I’ve never felt better my whole entire life. And I’ve got so many friends that are either sober-curious, or going there. And it’s become a trend for young folks, younger folks, who are just like basically anti anything that is not health conscious.

[00:32:21.17] Yeah.

It’s like, my routine is to get up at 5:45 every day, journal for 30 minutes, work out for an hour and a half, and then go slay work, and then spend time with friends. Like, that’s the dream of a lot of young folks these days. And that was not the case for me when I was young. It was like “You better get your butt out there and work. Okay? You better find your career and make some money. You better build a family and have those things…” But a lot of things are – like this return to analog, this return to soberness, this return to just putting pure things into your body, and less toxins, less poisons that really shape a lot of Americans’ lives.

Yeah. Yeah, I saw a trend recently amongst teens, where they will sit for 15 minutes, or whatever the time period is. And they’ll set their phone up and they’ll start recording, which is to keep themselves to task, basically. Because they’re going to publish it. So there’s still like a digital component to this, and I’m sure it’s a trend, so they want to get all the kudos from their friends and stuff by putting it up on the internet… But they just sit there. So they put their phone… It’s almost – it’s not meditation, but it’s just like “I’m not going to listen to music, I’m not going to have my phone, I’m not going to do anything. I’m just going to sit here for 15 minutes, and think.” And they find out it’s really hard, but it’s like a challenge, and they’re treating it like a challenge, like “Can you do this?” And they all started doing it. So yeah, these things are – anytime there’s a move in a certain direction, eventually there’s a counter move in the other direction.

Yeah.

And we’re starting to see some of that. But back to Meshtastic, you mentioned you might use it out there in no man’s land…

That’s right.

…and that’s actually the primary use case, I think, for this.

I think so too, yeah.

Which is like, in the boondocks. It says right there it’s for areas without existing or reliable communications infrastructure. So I’m sure there’s lots of really good uses for this in areas where they don’t have fiber runs going to their house. So a really cool project. I wonder if we should dig deeper and maybe get some Meshtastic folks on the show and hear about some of their cool use cases, because I guarantee you people are using this.

I would love to explore how to – I don’t have a big enough brain, I guess, to explore how to potentially use this unique way, besides the ones we’ve talked about. Like, what are some really, really interesting uses of it? That’s what I would love to explore.

Yeah. Really, really interesting uses. Probably out in the boondocks where there’s needs that we just don’t even know about, you know?

Well, even - I wonder if you can connect… Imagine you’re a 10-location boutique retreat or hotel. Most of your places are in the boons, but you’re literally all over the nation. Is there a way to strategically place nodes to enable this mesh network so that no matter what the connected case is for everyone else, you’ve got some version of your own private network built on top of this Meshtastic?

Mm-hm.

I mean, that’s what you’ve just said, but it’s a bit more expanded, you know?

Right. In the case of multiple states maybe, you’d probably have some sort of a VPN connecting the two networks, and then the mesh would be like an extender into the boondocks. And so you could probably accomplish that, but I bet you would use the actual internet in certain ways to actually bridge anything beyond. I mean, sure, that 331 kilometers record is probably not the best packet delivery, and everything…

Yeah, you’re gonna get latency, dropping packets.

[00:36:02.21] Yeah, that’s the record, but you want to be closer than that… So I assume if you’re like – we want one big network across multiple states, and then at each location we reach out two miles to this cabin, or whatever it is. Then your Meshtastic would be meshing from your cabin back to whatever you have at the headquarters there in that place, which is just on your local network, and then that would bridge via VPN or some other solution to your other headquarters, which happens to be in Tennessee, or whatever.

Yeah. You know what I’ve found myself exploring more of? Specifications. I never thought I’d be such a dang specification nerd, either creating them or exploring them.

You want to create specifications?

Heck yeah, man… All day.

Why…? [laughs]

I mean, not myself, but my little AI buddies, of course…

Gotcha. So you’re slopping us with some specs.

I mean, you can call it that… You can call it that… But I don’t think – I don’t call it that…

[laughs]

I call it exploring.

One man’s slop is another man’s treasure.

That’s true.

So what kind of specs are you writing, man? The most recent one was a way to version a Go CLI. Because I’d gotten a couple now, and I’m like “They’re all versioned weird”, and I’m not really writing all the code, I’m reviewing it. I see the patterns… But I was like “Can we just kind of create one unified way to version Go CLIs that are in this realm of what I’m doing?” And so I created a specification for how I want to utilize versioning inside of a Go CLI. And so now I just point the next thing I’m making to that specification, and say “Adhere to it”, and it does, and life goes on and it’s amazing.

The other one most recently was I swapped out 7Z, because it’s actually not that good. It’s good, but it has some flaws that I was not personally aware of. And so now I’ve become exposed to ZStandard, which was created by Meta. It’s a compression algorithm. And so now my new thing is I’m combining TAR, which has stood the test of time, the TAR archive protocol, with ZStandard, which is actually ZSTD. And so now I’m looking at these specifications thinking “Okay, great. If I’m going to implement an archiving tool that bridges the gap”, what I’m calling ZStandard for mere mortals, is what Zarch… is about…

ZStandard for mere mortals.

That’s right, ZStandard, which is the compression algorithm that Meta made. Now, it’s just a compression algorithm. It’s not an implementation, it’s a compression algorithm. And so to archive with TAR and use ZStandard - well, you can use that pretty easy with the command line, but there’s no API, there’s no sparse behind it, there’s no CLI… And so I’m building a CLI, because I archive a lot of stuff, as you may know… And I want to do it in the best possible way. So my most recent thing was going – I’ve been really enjoying this as a nerd, too. I’m going deep; I’m not that deep yet, but I’m trying to go deep into the TAR format. When did it begin? Where does it get utilized? Because what happened was is I had this really great tool, but I was using it on Arch Linux, by the way, and it just wasn’t working. I kept having CRC errors when I was trying to compress something and test it… I’m like “What in the world is going on here? Is it Dropbox? Is it the way I’ve used Rclone? Is it because I never really tested the memory? Because I’ve got 128 gigs of memory; is it because I never really tested the memory properly? What is the root cause of 7z not working properly on Arch?” Well, it turns out that 7Z has some flaws that I was not aware of, and I didn’t experience those flaws on macOS. They’re non-existent there. So on macOS, the tool works great. On Linux, it does not. And so I’m like “Well, I can’t spend my time building this, because it’s… That’s lame.” I want the good stuff, man. If I’m going to pour myself a glass of compression, I want the whole glass. You know what I’m saying?

[00:40:12.10] And so I went back to the board and I’m like “What’s going on here?” And so obviously, TAR, and I learned about ZStandard… And so to round about back to the specification, I’m working on - and I think it’s kind of there; it’s at version 1.1 or, 1.0.1 or something like that. I’m semvering, Jerod, okay? I’m botching this a little bit. But I’ve got the specification for how to implement TAR in an archiving tool.

And so there’s a lot of stuff about the way you’d manifest a JSON file first, and in what order you would actually create the archive to properly support PACS, or POSIX, and things like that in a TAR format, so that you don’t have jacked up archives. And then you compress them on top of that with ZStandard.

So I’ve become like this little specification nerd. I just love creating specs, because I can point at them when I’m working with agents and say “That’s the best version. I like that version. That’s a version that represents the software I want to make.” And so I feel like with these specs - one, I’m learning, and two, I’m defining what I think is the right way, so that when I repeat myself, I can repeat it against that standard.

Gotcha.

Long story short.

So you’re doing spec-driven development.

Oh, absolutely. All day long. Yeah.

[laughs]

I mean, not literally all day long, but like literally, when I’m doing it - yes, that’s the way. It’s a lot of planning, a lot of thinking, which I find is the weirdest state of play I never thought I would find myself in. When I first heard about vibe coding, I was like “Ah, those guys are stupid. What are they doing? Gosh, can’t you just learn to code already?” And now I’m like “Why?” Why would you go through such depths when you can have your Rust, and your Go, and you can eat it, too? Let’s do it, right? And I can just work with the AI and within these realms. I’m reading a ton of code, I’m loving it, enjoying it… I’m not personally writing it, but I’m the one who’s visioning different things. And so this move from 7zarch to Zarch has been fun.

Mm-hm. Well, your code cup runneth over. All this time I thought ZSTD was what German folks get when they’re too promiscuous, you know?

That’s right. Me too.

“I have ZSTD!”

[laughs] But instead it’s Zstandard. And thank you, Meta, for making that.

How is it better? You say it’s better. It compresses better? It’s faster? In what ways do you measure better?

I don’t have the facts fully in front of me, but – so let me go to their website and see if I can get it. Okay, so compressor name. You’ve got ZSTD… No, it’s not the German version of it… You’ve got Zlib, you got Quicklz, you’ve got Lz4, you’ve got Snappy and various other ones that they compare against. And so when Zstd - which is short for ZStandard - compresses, it can compress at a ratio of 2.896 (I’m reading this chart here), versus let’s say, Lz4, which is 2.101. So those are very similar. It’s actually compressing at a better ratio. Lz4 will compress at 675 megabits per second, whereas the compression speed of ZStandard is 510 megabits per second. So it’s pretty fast, and it decompresses just as fast. So it has a similar compression ratio, but it’s faster on the way in and way out.

[00:43:49.11] And one of the biggest issues I’ve had with the thing I built was like “My gosh, it takes so long to compress. And it takes so long to uncompress.” Unbearably too long. So long that I was like “Why am I even doing this?” But I like that, because I’ve now learned so much more making all those mistakes… And I’m like “Man–” I was kind of upset going down the wrong road, and really happily going down the wrong road. Like “This is awesome. It’s so cool. It’s useful”, and it is useful. And so all I’ve done now is basically swap out the tool I made with a different compression algorithm, and then now I’m re-implementing against this tar spec how to best implement a tar archive.

And so I didn’t lose a lot of runway in terms of what I built in time. I just swapped out the format and re-implemented the way it does TAR, and we’re none the wiser. So now we’re back to square one. But I’ve learned a lot going down the wrong roads, and I now have a brand new version of empathy I’d never had for what I would call developers, right? Software developers.

[laughs] [unintelligible 00:44:54.21]

[unintelligible 00:44:55.04] consider myself a pseudo-software developer. You know this, I’ve been an imposter for many years. If you didn’t know this, now you know. And I’m like “Wow, this is what it feels like to go down the wrong road”, and not be pissed, and still be kind of happy, because you learned so much along the way.

Yeah.

So that’s kind of cool.

And then later on, people will come by and say “Well, why didn’t you just use ZStandard?” And you’re like “Well, because I didn’t know it existed.”

That’s right. I didn’t know it existed. I had to make those mistakes, I had to try it on Arch Linux, and be like “Why is this thing not working?” …only to go back to root cause, which is a great principle - go back to root cause analysis. Where is this problem at? What’s causing it? And then you know what helped me, believe it or not? Stack Overflow.

Hm… Shout-out.

Yeah. It was comments on Stack Overflow, comments on Reddit about issues with 7Z on Arch. And if I didn’t find those, I’d have been like “7Z works everywhere. It’s amazing. It’s awesome”, which is what I thought. And it is kind of awesome, but it has some flaws that I just can’t live with.

Well, that’s how progress works.

Yeah.

So in the real world now - so we’re looking at a compression speed versus ratio image that compares… Is Zlib - is that behind 7Z, or is that a different thing?

That is a different thing. I think 7Z uses – I think it’s LZMA4. No, what is it…? Gosh, you’re testing my memory here, man… Hold on a second, I’ll get you some information. Yeah, I was close. LZMA – I was right, LZMA2. And so there’s LZMA the original compression, and then the fellow who wrote this software, his name is all over the actual stuff. Let me see what his name is, if I can get to it quickly… 7Z, the archive format, was created by – so it’s both a format and an archive tool in one. So it’s not only the compression algorithm, the format, it’s also the tool that does the job. And so it’s made by a fellow named Igor Pavlov. And so I think he rewrote LZMA, if I understand the history correctly, and LZMA2 is his compression version of it that 7Z uses. It’s great, it does a great job, it’s just notoriously slow. Do you like slow, Jerod?

Slow music, slow jams…? Slow dancing…?

Slow compression algorithms…

No, I hate it.

Slow internet?

No, I hate it.

It’s the worst, right?

Slow’s the worst.

Yeah, so…

Unless it’s a slow jam. Those are good.

Several things were getting me. I was like “Why does it take so long, this compression stuff?”

So what’s the real-world win for you? So let’s say you’re compressing one of our episodes, and it’s like 10 gigabytes. Do you know the time win for you? Like, it used to be 20 minutes, now it’s seven minutes? Or…

Yeah… So it used to seriously tap my CPU. Like, I use this application called – I think it’s CleanMyMac Z, or X… CleanMyMac X.

Yeah, I use the same one.

[00:48:05.14] And it hangs out in your system tray, it’s up there, and if you go and tap it, you will see CPU. And it will tell you the temperature of your CPU. And so whenever I was compressing with 7Z or 7Zarch, it was always as high as possible. And it’s like “Hey, listen, if you don’t stop doing what you’re doing, your computer is going to blow up, basically.” I mean, it was being very kind about it… And I was like “Ah, it’s an Apple. It’s an M1, it’s got it.” So I just never worried about it.

So it would really, really hurt the CPU for quite a bit, and it would take – I mean, now that we have like 20-gig, 30-gig projects, easily 20 minutes, if not more… To the point where I’m like – I’d forget about it, you know? And it just became this, as you know – you know about our Dropbox issues. It became like “I can’t keep up because this tool’s jacked up.”

And I’m trying to compress and archive in a way that gives me a little compression, but gives me an assurance of a single file that I can get back later. So I want a good in and a good out. And I thought 7Z was that. And it just turned out that it was great for Mac. So it’s great if you’re only on Mac, at least in my experience… And I could be wrong. If you know more about 7Z and I’m way off, please call me out… But that’s been my experience.

Well, one thing I know about 7Z is it’s been around forever. I think when I very first started using Linux, it was there. And so it’s old technology, and there are reasons why progress makes steady, sometimes slow, sometimes a little backstepping, but eventually, forward motion and improvement. So what was that same 30 gigabytes? Are you on the ZStandard now? Like, you’ve completely replaced and you’re running your new thing? Or are you still working on it?

This is super, super-early days, so I can only give you my latest/greatest, but… Time-wise - so much faster. Like, I was like “What, you’re done already? There’s no way.” And I would check it –

Like what, like from 20 minutes to 5?

Yeah, it’s so fast. Like three minutes, five minutes, maybe ten minutes on like a… So we had a couple archives that were like in the 50-gig range, and that was about eight minutes. So I’ll be in like the three minute range for most of our stuff… And if I do it right, if I do this TAR specification well enough, then we’ll have a really good manifest.json at the very top of the archive, inside the archive container, the thing that makes up the TAR archive. And it will have all the cool stuff. It will have – one of the other things about archives I’ve learned is they’re attack vectors. And I’m not a pentester, I’m not trying to like hack things, but – I obviously pay attention, like you do, but I never considered that “Wow, I should probably do a better job with this, because if I don’t do it right, then…” I mean, I don’t think our archives are in jeopardy, but if I open source this, when I open source this, and others use it, and they start getting jacked up because they’re attack vectors - well, I’ve got to do a better job of security.

And so when I started to craft that specification, it really exposed me to like a couple of different points around what was important. And one of the more important points was being secure. And I’d never thought about that. So there’s certain things you could do with following file paths and when to cap it, when not to do it… And again, I’m not super-steeped in it, but it’s quite tantalizing once you have this – I would say AI is like this guide that can learn or knows a lot about most of the things you want to know about. And if it doesn’t know it, it can find a way to learn with you, and help you learn, too. And so I find that that thing, that ability in today’s age is just so tantalizing. The fact that if I have a curiosity - sure, it may hallucinate, it may go down the wrong road for a little bit, but for the most part, I can explore some territories I almost would never have before, not because I didn’t have the ability, but because the guide wasn’t resilient to my path.

[00:52:18.20] Like, if I hit a blocker, AI finds a way around it through new information, or a different path, or an alternative. Whereas a human might be like “Okay, I’ve hit the end of this book, the appendix isn’t helping me, and Amazon’s down for the day, because the latest AWS outage happened.” I don’t know, whatever it might be. I can no longer explore over this hurdle. But AI finds a way to help you through all these things, and I just think that’s such a wild thing to have, to be able to talk to our codebase in ways we have never been able to, literally ever, in humanity. That to me is what gets me excited about software today.

Yeah, it’s an entirely new experience, and one that is quite interesting at the lowest, and exciting at the highest. It depends on the day, and the concept, and how well it’s performing, oftentimes…

Oh, yeah. I mean, I wanted to punch the screen today. My AI was not supporting me. I’m like “Listen. Listen. I gave you clear instructions. I know you’re smart. You’re going way over there. Can you please stay here? Right here. This is where you’re safe at. Don’t go over there. Don’t do that. Don’t touch that. Stop changing that file…!” I’m just – I’m being facetious when I say that. But it’s a version of that. It’s like, “Listen to me, please! We have a mission. Stick to the mission.” Over there, looking at squirrels… Get out of here. Sometimes it listens well, and sometimes it’s like “Listen, I’m just going to do whatever I want to do, and you can’t stop me, because you don’t have any control over me. I do what I want.”

Yeah. Yeah, one thing that I’ve been doing recently, which has given me some success - it requires a little bit of patience in one direction, but it helps me avoid a lot of the tailspins… It’s just like basically rebooting the AI, and just – keeping my conversations short and the context windows fresh, and all the backlog of things it already knows about the project… Yeah, just make it go read the files again. So I will do a couple of features, and the first time I see it, you can almost notice, “Oh, it’s building up a hallucination path, or something. I don’t know what it is”, but you can kind of just feel it. And I just Control+C that sucker and just start fresh, and I just tell it what I want it to do, without any of the… Sometimes you’re like “Dang, it already knows this part, and it already knows we’re working on this…” and other times I’m like “Nah, who cares? It can go figure it out again.” That’s the patience part. It’s like “I’ll check your files and see what you’ve been up to”, and you’re like “Well, you just knew that until I Control+C’ed you, and now you don’t know what I’m up to.” But I feel like it’s – this is Claude Code specifically, which is the tool I’ve probably been using solely for coding for the last two weeks. I haven’t been using anything else. It’s so much better on its first feature than it is on its sixth. It’s just better. And I’m not sure why. It may be because it just doesn’t have any baggage… That’s the way I kind of mentalize that.

Yeah.

And so I just Control+C that thing and I’m like “Let’s just have–” It’s like a brain wipe. It’s like “I’m going to get a new Claude Code that doesn’t know my codebase anymore.” And it’ll figure it out and it’ll do the feature and it’s going to be better than if I have the one that knows the codebase, but it also has - I don’t know, it needs to be defragged, or something. Have you ever tried that, or have you had experience with starting fresh? I know you [unintelligible 00:55:33.06]

That’s exactly what I do.

Oh, you do do that. Okay.

[00:55:38.05] Yeah. I have a process where I keep a context file, like a context.md file, and I keep a context-guide.md, which is how I want this context file to be structured if it needs to be structured, if it needs to be deep; so it’s like multiple agents, multiple features, just kind of keeping a lot of things in check… And so if I’m like “Hey, you’re misbehaving. Just dump your context to this file, and I’m going to clear the context.” It’s like a threat. It’s a very kind – I’m being kind; I’m just joking around when I say that, but it’s like, “You’re going too far, you’re not operating well… Just dump whatever context you have and what you think is next to this file, and I’m going to clear the context.” And I do that happily, and I have like an unfurl command that says “Bring yourself back online, read this file, and measure it against this guide. So this is how you should read it. And report back when you’re ready”, kind of thing, or “Confirm when you’re ready.”

Okay.

And that’s what it does. It goes and reads that, it dumps this context, I clear it, it comes back, I say “Bring yourself online…” It’s like I’m in Westworld or something like that, which is where I got that line at… It’s like “Bring yourself back online, Dolores.” It’s from Westworld. And so they do that, and it checks it out and it’s like “Okay, great. I see you’re working on this, I’m looking at that feature… What do you want to do next?” I’m like “Sweet, now you’re smart again. Let’s go.”

Gotcha. So we’re doing slightly different things. I don’t let it remember anything. I just tell it – like, I don’t want to have any of that context at all. I just want it to be like a new human. And so it’s more like the Men in Black Thing, you know…

“Where are we at here?”

Yeah, exactly.

“What are we doing here?”

I’m just wiping memory. I’m just like “No context, no nothing. You’re done.” I’m going to open up a new Claude Code on the same directory, and I’m going to tell it what I want it to do. It could be copy-paste what I told the last Claude Code, but this is a brand new Claude Code. It doesn’t know my project, it doesn’t know anything. It has a claude.md, so it has some rules that it can follow there… And that’s about it. It’s going to figure it out from square one.

And so that to me is like a full reboot, versus - maybe you’re just doing like a go to sleep and wake back up thing. Neuralizer, that’s the word I couldn’t think of. Men in Black neuralizer.

Yeah. Neuralizer. I will review that context file, so I don’t like to blindly just let it do that. It’s a way for me to know what it’s thinking, too.

Sure.

Or what it thinks is next. And so if I look at that, whatever it dumps and it’s way off, then I won’t tell it to bring itself back online by reviewing the context… I’ll just either delete it or skip it. But it’s a way for me also to have an awareness of what it thought was going on, and how far off it was from the truth… And there’s times I’m like “Wow, that was not true. That’s a hallucination there. We’re not working on that feature.” Or “That’s not what’s happening here”, and I won’t do that. But it’s a way to save myself from having to type as many keystrokes, or do something to like retell its context, or even give it indirection… I can just say “Review this.” If I agree with it, of course… And go from there.

Break: [00:58:38.01]

Well, let’s turn now to a new AI role that’s exploding. We’ve been discussing our AI roles…

That’s right.

…but we are not officially employed by anybody. Well, I guess we’re officially employed by our own company, but we aren’t certainly passing our resume around. But if we were, we would want to have the acronym FDE on our resume.

Yes.

Because that’s the new hotness according to interviewquery.com. Reporting on a report, as the internet does, gladly - and I’m actually reporting on the report of the report… So if you’re listening to this, and then you go tell your friends about FDEs, you are four layers of inception. A Financial Times report talks about monthly job postings, and this new role, FDE, Forward Deployed Engineers, are looking to get hired. There’s been new job postings to the tune of 800%, which of course tells you it was really low before, and now it’s not as low…

Yeah. It could be a bubble.

Yeah. I mean, it could also be from nothing to some, and the 800% is much easier when you have small numbers… But nonetheless, it’s moving, and it’s a new position led by the forerunners, as they call it, in the AI race: Anthropic, OpenAI etc. People who are out there building the models, but also the software around the models, and trying to nab up as many customers as they possibly can, before their money runs dry, to have a new kind of engineer… Which is somewhat novel, I would say.

Mm-hm.

I call it AI-led, because – well, you’re kind of leading the AI, but the AI is leading you…? I don’t know, who does wag the dog, or the tail wag the dog? How do you say that?

Who’s got the leash? Who’s holding the leash? Is it the human or is it the dog?

That’s right. Does the dog wag the tail, or does the tail wag the dog? There, I got it. The question is that, and the answer is we’re not sure, but if you want to be an FDE, you’re going to have to wag the AI… Because their whole point is to be deployed with customers. This is kind of how they’re different than a typical software engineer. They’re one-on-one with customers, and they’re tailoring AI models, whether that means RAG, whether that means fine-tuning, whether that means literally just setting up the right model to use, and providing it context… I’m not sure if that matters so much. That’s probably up to the FDE, to decide what is tailoring in the context of the customer… And working directly with the end users in order to give them what they need. Your thoughts on this new thing, that is new?

Well, I have news, Jerod…

Okay.

I have met, talked to, and experienced an FDE before, recently.

Oh, yeah?

We’re working with Redis on an engagement, so they’re one of our sponsors… I love Redis. I’m actually quite excited that Redis will want to sponsor us, because they’ve never…

That’s cool.

It’s so cool, right?

Yeah, for sure.

And they obviously have a lot of stuff around their vector search… You know, Redis is really this in-memory cache that’s utilized in so many places… But they’re really solving some cool problems around the AI vector search and embeddings. I don’t even get it. I don’t even play in that world very well, so I don’t really understand the full spectrum of it, but I sat down and I talked to Tyler Hutcherson from Redis, and he is a forward-deployed engineer.

Oh, he is?

[01:05:48.16] Yeah. And his entire job is to obviously talk to people like me, in some context, but the other part of his job, the 99.9 other percent of his job, which is not talking to me, is being at a customer. Like, literally being there for sometimes multiple weeks, or a month. And so he’ll kind of come in and come out… And his whole entire job at Redis is to not really at all sell Redis, by any means. That’s not his job. His job is super-about what are they doing, how are they leveraging Redis, and how is it working for them, and how can they make Redis better for them. So kind of like that – they’ll kind of like either do integrations, they’ll do migrations, they’ll do… Like they work there, like they work for that company, but all they are to Redis is a customer. They’re not paying Tyler any extra, to my knowledge, and Tyler actually works with a team of these folks.

So that’s the point of the forward-deployed, is like, you are not just with the customer, you’re like kind of working for them.

Yeah. And - I mean, you’ve got to be fully embedded to have the empathy factor. You’re deep in the water with them, you’re wading with them to kind of figure out “Where’s the next drop off?” or “Where’s the land?” I don’t know. What’s the point of wading in the water? I guess you’re swimming, maybe? Who knows. You’re knee-deep in something, right? And sometimes you’re not knee-deep in what you wanna be.

That’s right. You’ve got to figure out what it is that you’re knee-deep in.

Sometimes it’s poo, sometimes it’s water.

That’s right.

Whole job is about figuring out how Redis is being utilized, implementing features, taking that stuff back to product, product iterating it out, now it becomes in the next version of it… And so it’s just like the rinse and repeat cycle. I think that’s – it’s smart. It really is smart, because you’ve got that boots on the ground mentality, you’ve got an engineer doing an engineer’s job, and their whole entire job is – the success metric around it is just make the customer enjoy using it more, and solve their problems.

Right.

And then hopefully, if that is a product feature we should support, bring that back to product, let’s make it a true feature… And you’ve got, as best as you can, truth, which is customer usage. How do they fall down with it? How do they triumph with it? How did it work out in different deployments, or on different architectures? And you take that back to product and you make it real, and now they’re better, and other customers who had that problem’s better… And obviously, if it fits into the Redis – I mean, this is not a Redis ad, but this is the only context I have. I’m sorry.

That’s alright.

Good for you, Redis. You got this one for free.

You contested him.

Yeah… What was I saying now…? If it fits into the Redis way of product, that’s great. I mean, that’s really a – that’d be a fun job, I think.

Yeah. That does sound cool. Let me bring this back to a company like Anthropic here, out of Redis… Because that makes total sense in the context of Redis. Where I think it’s going with these AI model situations is you have a company like Anthropic, who wants other companies to hire Claude, to use Claude as much as they possibly can, and to get the most value out of Claude that they can. Well, as we’ve seen lately, there has been a difficult path towards successful adoption of AI in the enterprise, whether you believe the 95% numbers or not. Our recent show with Sean Goedecke, he does not believe those numbers. You can go read his blog post why. But certainly, there’s plenty of projects that have failed to adopt AI inside the enterprise. The promise is there, you get excited, you go out and you’re just like “Well, let’s just slap it on something and see if it’s just all unicorns and rainbows”, and it’s not.

And so these FDEs, these forward-deployed engineers that work for Anthropic - they would perhaps go in and be deployed into a Fortune 500, or whoever it is, or maybe it’s a small Claude customer, but could be using it 10x, 100x more if they were having success… And so they are there to help them wield that model in a way they get a huge value out of it. And I think that does sound like a fun job. That would be fun.

[01:10:10.21] Yeah, I think so, too. I mean, yeah, wow. What a time to be an FDE, right?

That’s right.

This burgeoning brand new title. What a time to be that.

Right. If you’re not an FDE and you want to become an FDE, how do you get there? Well, according to this article - that’s a summary of another article that I’m summarizing for you… They say this is a clear signal of a shift from research to real-world results, and there’s a clear message that AI fluency isn’t enough. So if you’re fluent with AI tools - not enough. Customer-facing AI skills are the real edge. It says “The most in-demand tech workers in 2026 are not just coders, but also communicators, problem solvers, and translators between AI systems and human needs.”

So there’s a short list of skills to be acquired. If you have those skills, throw FDE on your resume. If you don’t, acquire away… And I think that while this might not be the end-all be-all title - because you know, every couple years we need a new title. This is like the new data science, according to this. Maybe it’s a flash in the pan. But if it does have staying power, then those are the skills you’re going to need. It’s not just the ability to direct your agents in a successful way, you have to actually help other people direct their agents in a successful way.

Yeah. That’s interesting, even the way it phrases this… It says – I’m going to re-say it again, because you said it good, but I’m going to say it good, too…

[laughs] Say it better.

Maybe I’ll suck at it, who knows? Let me try.

Okay.

The most in-demand – I’m gonna…

You’re doing bad.

I’m just laughing for the fun of it. I don’t even want to try. I just want to read it.

Okay, read it.

The most in-demand… This is profound. That’s why I want to read it again. The most in-demand tech workers in 2026 - that’s not this year; this is 2025 - are not just coders, developers, but also communicators, problem solvers, and translators between AI systems and human needs. And that is such a wild thing, because there are so many folks out there like that. So when we talk about disruption and job replacement, how many folks were in product teams, or the solution makers and problem solver spaces where they were sort of moving around specs and they would say “Well, what do you do here?” “Well, I go get the spec from the customers and I bring it to the engineers.” That’s what they do around here, right? And they’re like “Well, gosh. Everybody’s taking my job. All these AIs are taking my job now.” Well, not anymore, because you can now go into this role, because so many of those people - they may be displaced because of change or disruption. I think that’s like a snapshot of a lot of those folks. Go do that. Go do that.

Mm-hm.

And if it’s going to be in demand next year, and this episode is true - this is November 11th, 2025; when it ships it’s plus four days. Do the math. I can’t, I’m stupid… Go and do this job. Be an FDE for Anthropic or the Anthropic likes.

Yeah. Alright, let’s do one more. Let’s do one more deep-dive, and then we’ll call it a day.

One more.

Dead Framework Theory. You know I like these theories, you know?

So many theories today, man…

Gosh…

Don’t confuse it with the Alive Internet Theory, okay?

That’s right. This is different.

We’re back to the dead.

This is different.

Kill the frameworks.

[01:13:41.07] This is the opposite of Alive. This is dead. This was written by Paul Kinlan, and he actually wrote a piece last year which I included in Changelog News. It was called “Will Developers Care About Frameworks in the Future?” And he is thinking he was kind of wrong about that, but he was wrong in kind of a weird way, or a timing way… What he’s found, and what was news to me - which is why I definitely had to include this - is basically now that more and more people are building with AI agents, it has pushed React into this foundational layer of web development, which he thinks is unassailable at this point. Not only do the models know a lot about React because they’ve been trained on vast amounts of React docs, and examples, and open source code, but it’s also in the best interest of the tool creators if they want their customers, like Cursor, for instance, or V0, or Bolt, these tools, these vibe coding tools or whatever you want to call them, they want their customer, the developer/end user to have the best chance at success. And so they’re just picking React for that reason. So it’s like, it’s not the winner of the game, it is the game now. And so much so that they’re hard-coding it into their system prompts. So if you’re like “Hey, make me a web app”, the system prompt’s like “Make it a React app”, basically.

That’s right.

And so that’s why the dead framework theory is like maybe frameworks are over, not for the reasons that Paul was thinking last year, but now he’s thinking because basically React is its substrate.

Fully seated, yeah. You know, what I find interesting about this is just that, is that if you’re going to code something else – let me rewind and think about what I’m trying to say here… Maybe a layer above that is that what if future engineers, 2026 engineers maybe even, right? Because that’s so close. That is the future, obviously, but it’s so close that’s not truly the future, but it is the future. But what if in 2026 engineers’ jobs shift from innovating and handing it to humans, but they’re innovating it to hand it to an LLM, so that the LLM can then be using a well-defined standard.

Now, in this case, React has had 20 – well, not 20 years. How many years? 14-ish years? 12 years?

What year did it come out? I can’t remember. 2014, maybe?

Yeah, so… A lot, you know? 11 years.

Call it a decade at least…

Yeah, a decade. So it’s been seated for a decade.

Hold on… Let’s get the number. Yeah, 12 years ago. I drilled it, 12 years.

Okay, smart man over there. Good for you.

Good guess. I guessed good.

You get a cookie today.

I guessed good.

So 12 years - alright, that’s there. But what if future engineers – I almost want this as a developer. I would rather have a super-good engineer who’s way better than me, who really knows the things, teach the LLM that really good thing; make the framework, make the standard, define the specs even, so that whenever I utilize the AI and those kind of things, it’s just kind of there. It is the substrate, like React has become in this case, in your example here. Like, what if the LLM already knows how to write the language, so I don’t have to learn how to write the language? I can learn through using it, but it already knows, because it was taught by an engineer. It’s almost the same concept. It was taught to use and deploy really good, well-formed – some may not like React, but really good, well-formed React applications, because it is a substrate. It is the winner, and so why debate it? …and just hardcode it into everything, because that’s the way. It is the standard format, it’s got the most battle-tested tests against it, it’s been utilized in all these places… And sure, it’s got its blemishes and its flaws, but it’s the standard, so why fight it?

[01:18:04.22] Kind of the same thing, but the opposite way, which is not the React way where it was born 12 years ago and it’s become that through usage, but the opposite way where it kind of comes in from the top. It’s like, it’s teaching the LLM at the top, and it’s coming down to the LLM through my desire to build out new things.

In this case, something like Go and Rust are pretty staples. I’m trying my best to never touch Rust again, because I’m just not that good. I’m just not that good. Go, I can hang out there. Nothing against Go, it’s just easier. It’s easier to read, it’s easier to work with.

That’s a good thing.

It’s kind of more fun, for me at least… And I feel like Rust is so cool, but I’m just not that cool to touch it, okay? It is no doubt super-cool. I’m just not worthy.

Yeah, I’m tracking what you’re saying. I’m wondering… There’s like a gap between here and there, obviously, the world that you’re painting and the world that we live in… And ironically perhaps, the reason why at least Paul says that these tool creators, as in model creators, are front-loading React is because developers need to maintain the apps that are being created, and developers know React. So you’re giving them the best chance of actually maintaining the produced code, because of this deep-seated knowledge across the industry of the last 12 years, of how to build React-based websites. And if you’re going to top-load something else that we don’t have to look at, then I couldn’t imagine React being the technology of choice. There would be no reason to choose React, because their browser already is a platform, and the agents could speak directly to the browser via the DOM, and wouldn’t need something that’s maintainable, because they’re the ones maintaining it, and they already speak the browser. You know what I’m saying? So because the humans are still in the loop, we are getting React apps, but if we took them out of the loop, we wouldn’t need to have React apps anymore.

Ooph…

But we aren’t out of the loop, and then will we be anytime soon…?

Yeah, I don’t know if we will be either. Let me clarify… As you were describing that, it came to me more points of clarity. So what if really good Go developers were really interested in, or even employed to make sure that Claude Code knew exactly how to structure a Go CLI? So whenever I spin up a new CLI, I’m not having to start from scratch with context lost, or what is the problem I’m solving, or “Oh, because it has these specifics.” Like, what if it came from top down? I don’t have to think about structure, I don’t have to think about where the commands go, where the internals go, where the security goes, all these different things. It just is, because someone who’s really good with all the Go idioms is teaching the LLM so that I can start off with all these sort of like – I wouldn’t even call them training wheels, right? This is a dang tank. I don’t have to even get a bike, I can go right to the fighter jet… Which is totally different, when you’ve got two wheels and two wings, and massive engines, and really high speeds… Tom Cruise, all that good stuff.

Now you know how he retreats to attack right there, man. The fighter jet. Let’s go.

And even so far as like versioning. I had fun doing that, versioning a Go CLI. What if somebody already knew that and taught the LLM, so that it was always –

That that’s an implementation detail.

Yeah, it was just an implementation detail. I didn’t have to do that.

Right.

[01:21:42.28] I don’t know how that flexes out over time, but that to me is cool. Like, why should I have to learn all these things when the LLM can be taught it from a human, and even learn from the AI? …and this weird cycle goes on and on. But it kind of comes down from the top. Why do I have to sort of start out from scratch, telling my LLM like “Hey, no, you totally botched this structure, and now we have to spend a day or two hours refactoring.” A day in AI time is like one hour, right? [unintelligible 01:22:10.22] is like “We’ve got two weeks to do this”, I’m like “Dude, you’re going to be done with this in like a minute. I’m not even worried about these weeks”, right?

Seven minutes…

Yeah. So that to me is kind of a weird thing, and it’s kind of a wild thing to think about.

Yeah, I see what you’re saying. I think a slight trepidation from any sort of top-down design, and the reason why it can be bad is when it doesn’t fit your circumstance, but you’re stuck with it, because that’s what the agent knows… And so you now have to use – you have to build a Go CLI the way the agent says to build it, because that’s how it builds it…

You’re not actually part of that conversation at all. And if you tell it to change it, it’s like “Sorry, I can’t do that, Adam.”

Yeah. “My ways are not those ways, and so you have to go that way yourself.”

Right, everybody has their own opinions and their own taste. And when your taste clashes with the top-down - well, the top’s at the top, and you’re at the bottom, so you know…

Or even things in Rust like memory safety. Like “Hey, don’t ever do it that way.” That’s the wrong way. That’s always the wrong way. I don’t care if you’re writing a Rust CLI, or a Rust backend, or low-level tooling that taps into eBPF, whatever. It’s never right. Never do it that way.

There aren’t so many absolutes in software development, though.

Are there not?

Well, why do you think “It depends” is one of our most common phrases?

Gosh, that’s true. Okay, fine. It depends…

[laughs]

I concede. I concede. But I still have once. I still have once.

It was a fun ride. Yeah, well… Now you’re retreating. I’m waiting for the attack. I’m waiting for him to attack. He just retreated a little bit… Okay, so… Good thoughts. Interesting –

Well, I can appreciate the “It depends.” I’ve heard it about one million times in my life, on this podcast and throughout my career. So I can empathize with that. I think my dream is that help me think a little less, so I can think about vision, product, and user experience.

Oh, I think that we could possibly get there without the top-down approach. I think that the more that we codify and continuously improve on best practices, and then feed them back into the systems, then the systems will get better at working at increasingly higher levels. I don’t think if we take us out of the nitty-gritty that we want our agents creating React apps any longer. I just don’t see the purpose of that. But in the meantime, I get it… I also think it’s kind of sad, because you just have this accident of history. Not that React is an accident of history, but you have this timing, and this – talk about codification… A freezing in time, all of a sudden. Look at this chart. [unintelligible 01:24:54.27] It’s crazy. So this is according to builtwith.com. He says there were 13 million plus sites outside of the top 1 million deployed with React in the last 12 months. So there’s a straight up chart, if we look at React usage over time. Starting in – I mean, it started gaining in 2018, up through 2022, but in 2023, a.k.a. coding agents starting… They’re coding LLMs telling you how to write code, and now they’re agenting for us. Agentaking? I don’t know what the verb there is, but… It goes straight up. I mean, it’s just insane how many new React sites are coming out today.

You could do the same comparison to WordPress in a different era, right? At one point, by and large, anybody developing a blog or building out a basic website for SMBs, medium-sized companies, even some larger ones…

Oh, for sure.

WordPress, right? Same curve, no LLM.

[01:25:54.29] Yeah.

I mean, when you are the de facto, that’s what happens.

Yeah.

So nothing against that, but I think that’s…

No, I’m not here to blame React, or anything. I think it’s different because there was human choice involved, and there is human choice… It’s just like two degrees of separation away.

Yeah.

It’s like this self-redirecting feedback loop, where just React’s popular because React’s popular because React’s popular because React’s popular… And when do we step in and say “Yeah, but…”? We had a chance to say “Yeah, WordPress is cool, but what about…?” What else did people build websites with? [laughs] What about Ghost? Here you go.

Rails?

What about Ghost?

Yeah, I suppose Ghost, yeah. Sure. But I think they sort of have more choice, like Ruby on Rails, and different things like that…

There’s choice. Yeah.

Build your own.

That’s right.

Build your own WordPress, kind of thing.

Yeah, we can still build our own things… I think that what we need to do in the future is just build everything ourselves, you know? Just like, who needs libraries? That’s where I’m headed.

Is that right?

If the agents get us there. If they get us there, yeah.

Explore that a bit more. How do you mean – I mean, because I think testing frameworks are still important… Protocols and known processes that work, that are battle-tested, fast, readable, digestible…

I’m not against protocol. I’m just talking about libraries and frameworks.

Right. Okay…

Certainly protocols and common bits that glue everything together. Here’s a quote… So I’ve been working on a little blog post that I’m not ready to say much about, but I will say…

Okay…

I’ll say this much. Our friend Luis Villa was writing on his newsletter that I read, and he’s been agentic coding now for about a month… Like in earnest. Like, having fun building stuff, just like you are. And he wrote out some of his observations after a month of doing this.

Yeah.

And he says – here’s one of his observations. “Each of my new tools relies heavily on FOSS libraries”, free and open source…

Yeah.

“…but I suspect many fewer than would have been the case in pre-LLM code. Hard to say if this makes existing core libraries even more important and valuable, or what.” So he’s finding what I’m finding, which is like there – you’ve still got your React, but all of the other stuff that you used to pull in, all those other ancillary libraries, those utilities, those helpers, especially the smaller ones… It’s like, it’s a burden.

Yeah.

So just tell your thing to write a thing that does exactly what you need. It doesn’t need to be generic, it doesn’t need to be abstract. It can be exactly what you need and nothing more. And you have way less dependencies over time. And I think that we’re at a point now where we can start to trim our dependencies around the edges, and re-implement their limited functionality just for our own use case, and remove that, and then over time build with less and less dependencies. And I think that that in certain ways is unfortunate, but in other ways it’s kind of a good thing.

Like Leftpad, for example. No one’s –

For sure Leftpad.

In this future, Leftpad will be less – it’ll probably never be a dependency again.

It would never be.

Not so much specifically, but the idea of Leftpad being a dependency. In this new world, an LLM can easily replace that thing, and it would. It would never pull it in as a dependency. It’s like “Oh, I need to move–” I forget, what did Leftpad even do? Like padding around it, or something like that?

It padded a string on the left side. It added blank spaces to a string, according to the way you want to use it.

Right. So the LLM would be like “Yo, I just did that. No Leftpad required.”

Almost every developer in human history would also have done that one. That’s why that one’s hilarious. Only in the JavaScript community would we just pull in a package for that. But yes, exactly that. Let me go slightly more complicated. What about a Markdown parser, for instance?

Ooh, okay… Now you’re getting me excited.

[01:30:14.29] Well, because, I mean, Markdown parsing is a known quantity… This is not rocket science. There’s a Markdown parser in every language, people have all written them etc. Well, what if I need a feature where it’s I want to be able to add Markdown links to this text blob? So you can put your square brackets around your text, and then you can put your parentheses, and put your link in there… And we’re going to take that and we’re going to know that that’s a thing, and we’re going to either translate it into HTML, or maybe into an iCal thing… Wherever we can output it, okay? Well, in the past, perhaps, you would pull in a Markdown parser. I’ve got to go get the Markdown library of the day. Or I write a really gnarly regex, right? In the future, why don’t you just write your own little Markdown parser? Because it’s going to take seven minutes for Claude to do that for you. Or 40 seconds. And it might not be a full Markdown parser, that supports – it doesn’t have to support all of that. You don’t need all of that surface area.

And you’re not worried about bold, and italics, and stuff like that. All you need is a link.

Right. And there’s a place where it’s like “Yeah, a regular expression could do that.” But also, you’re like “Well, now I want the links, but also maybe I do want the bold and italics… Okay, my Markdown parser just does bolds, italics, and links. And it’s 500 lines of code”, versus the official Markdown parser for Ruby, which is like… I’m just making this up.

Massive.

I have no idea what it is. It’s 17,000 lines… You have to compile a C extension, blah, blah, blah, blah, blah…

Oh yeah, dang. It’s true.

You know? So like stuff like that.

Yeah.

I think those things – you start to eat in from the fringe, and you start to like eat your way into your dependencies until they’re all gone.

Yeah, I think that’s the kind of open source we won’t truly miss. I think the open source we would miss… Like, if you no longer need a Ruby on Rails. That level of replacement, that might be a different world to live in.

I think those are the last things to go. I think they stick around the longest. Because they’re the biggest, most valuable. I mean, React sticks around the longest.

Yeah. And I think as long as humans desire and need to read the code, or desire to have visibility into the code and it’s never – because, I mean, there could be a world… I mean, I’m living in a world now where I still tell people I’m a technologist and a software developer and a podcaster and all the things I do… And then I tell them “I’m doing things with AI I never thought I would do. Literally, it’s science fiction to me still yet. I still can’t believe what we can do.” It may be a couple more years where that may be true. Like, you just invent a framework, and it’s so good at doing what it needs to do that we stop checking it. Because it’s just so good. It’s always been good. And that might be five years or a few years down the road, but it might get to a point where enough of the human race who’s doing this kind of work and pushing this kind of boundary is like “Yeah, every time the thing does the thing, it’s good, so I don’t really care about the code really ever anymore. I used to, but now I don’t.” I’m not saying I want that world necessarily, but what if that’s the next evolution? It’s just become so good that it could talk to the DOM… And in fact, it has reinvented the browser. DOM’s gone, too. You know?

Bye, DOM…

The things you thought were never going to be replaced are replaced, you know?

Bye, Felicia…

Who knows…?

Yeah, we’ll see. I’m still of the opinion, even though I’ve gone away from my AI winter stances, my plateauing, because I’ve been wrong about the plateau on a couple occasions… I don’t like to be wrong, I’d rather be right. I’m not going to be calling plateaus anytime soon, because I’ve been wrong enough. But I don’t think we get from where we are today, with transformers, to that. I think we need some other – and some other thing most likely will come around. So I’m not saying it’s not going to happen. I just think that we need another step change to get to that level. I think we can get very far, where we are.

Let’s try to hypothesize what that might be then. Can we do that?

[01:34:13.06] Oh my goodness, man… I’m not a computer scientist, I’m a podcaster. [laughs] Technically I’m both, but…

Well, I have some ideas, and I’m a podcaster, too. I don’t know if this is true, but I wonder. I mean, because we are moving at such a pace now where the hard things got easier… And I would love – maybe this is where we should go either this year if we have one more opening for it, which I think we don’t… Maybe it’s a first year kind of thing, because I tried to do it last year… Is dive into quantum computing. I think that that’s becoming the new fertile ground, because we’ve got such advancements in artificial intelligence, LLMs supporting large swaths of generating code, that we’re moving at a faster pace on the things that were hard… So it’s obvious something like that will be next.

We’ve talked a little bit about biocomputing recently, which I thought was kind of wild… And even that was in the research phase. It was cool to have that conversation, and no harm, no foul to the person that we interviewed… She was amazing. But it was still research, so it still wasn’t quite here.

For sure.

It was in the works, they’ve had some issues, they’ve had some stability issues… They’ve got some clear windows of where they can go… But I wonder if that next invention layer, like you say, is not here with transformers, is solved… Because you’ve got Microsoft doing some cool stuff that I’ve heard about where they’ve got a computer in negative 500 degree temperatures doing something. And I’m not in that world, I don’t know about all that stuff, but I saw a video about it…

“Doing something…”

Doing something, you know? Like, this quantum stuff.

Computing… Yeah.

So I wonder if this quantum computer, this thing that I’m not even that familiar with, is the next big thing that comes sooner, because we’re here now so quickly in building software. It’s such a clip that we never had before.

Yeah, I can’t answer that question. If you made me stake my life one way or the other, I would say I don’t think quantum computing takes us to AGI, for instance. I think it’s like a different kind of thing, but I could totally be wrong about that. And I think there is other things that are coming beyond transformers. I just don’t know what they are and when they’re coming, and how far transformers can take us. But even guys like Yann LeCun, which - he is a Computer Scientist, capital C, capital S, and one of the smartest ones that I can fathom in the industry… He says that we’re not getting to AGI with transformers. And so I’ll just take his word for that. I don’t know what the next thing would be, and I’m not the guy to invent it. Somebody else will invent it, and then I’ll complain about it.

You think AGI is required to reinvent the DOM, or the browser?

Yeah, it’d be pretty stinking smart.

…or to make Rails obsolete?

I think it’d be pretty good. AGI-AGI probably not, but something with way more… Conscientiousness? I don’t know.

It has to have actual smarts, not just instant recall of the world’s information, you know?

Yeah, yeah.

That’s my take. But I’d love to hear other people’s take on that, because I’ve been wrong before. I think it was 2017… Back in 2017 I was wrong, and… Yeah, that was the last time.

What were you wrong about? That was the last time? [laughter] I was following you until I realized [unintelligible 01:37:37.06]

I’m just messing around, man…

I’m like “That was a long time ago, Jerod. I’ve been wrong a lot since then. Are you pulling – yes, you’re pulling my leg. Stop doing that.”

Yes, I got you.

Yeah, you did get me for a bit there.

Alright, great conversation. This was fun.

Yeah. Oh, did you push Record?

Oh, shoot…

You did. I got you. I got you.

Oh, you got me back.

That’s just a throwback, y’all. Our last Friends episode we did together, we were sad… If the first few minutes of that episode - go back and check it out - we seem a little gloom… Because we are. We forgot to push Record. We talked for like 45 minutes, total gold, and you missed it, because it was going to /dev/null, and that’s not cool.

[01:38:19.09] That’s right.

Let’s leave the folks with this. Go write an agent. Right?

Yes.

Go write an agent. Listen to Changelog News, go write an agent… Changelog.news. If you’re going there this week - well, that’s going to be there for you. If it’s next week - well, then go back a week… But either way. Thomas - how do you say his last name?

Well, I think it’s pronounced Tacek, because it’s Polish, I believe. But it’s spelled Ptacek.

Yeah, Thomas Ptacek.

There you go.

Co-founder of Fly. We love Fly. We’re hosted on Fly. Thank you, Fly.

Did he co-found?

He’s one of the co-founders, yeah.

Okay, cool.

I think he’s – was he CTO?

I don’t know.

I don’t know what he is. I don’t know, Thomas. Where are you?

I know he’s been around a long time…

I know he’s a co-founder though, over at Fly. We love Fly. We use Fly. Amazing. But he wrote this really awesome read, it’s 13 minutes long, “You should write an agent.” And I read it. I agree. I don’t have time to write an agent, but I think I’d like to. It’d be a fun exercise, because then you get the full empathy factor of what it takes to make one of these things. And maybe you can make your own little Archie, or Alf, or pick a name that’s fun to you… Maybe it’s Susie. I don’t know. Pick a – make yourself a little agent. Who knows?

There you go. And then tell us what you named it.

That’s right.

We love naming things around here.

Man, I tried to name something Archie recently, man… That’s why I said that. And I was bummed, because I think it’s –

Was it your archiving tools?

Yeah… I wanted to call it Archie. And so I think it’s – yeah, Archie Core…

Just spell it wrong on purpose. That’s what everybody else does.

Maybe so… But I think it’s Archie – what was it? Something that was like that. There’s an Archie out there, and I found it, and I was like “Okay, I can’t do that, because this is like AI-native cloud infrastructure.” Is that the most buzzword ever? But that’s what they do, and that’s what they are… Archie Labs, Archie Core… Don’t know who they are really, but I found them and I’m like “I can’t call this thing Archie”, so I called it Zarch.

Nowhere near as good.

I mean, I think Zarch is cool.

Why? Because it’s an arch way, that’s somewhere over there in Germany?

It’s archiving, it’s totally focused on ZStandard, not other formats… This is ZStandard for mere morals, man. Zarch. Not Archie. I think Archie would have been cooler. I’m just trying to make myself feel better, okay?

Right. This is like when you don’t hit Record, and then you still record a show after that, and you try to make yourself feel better…

Yeah. I mean, Archie would have been cooler.

Yeah. Well, you know, just spell it wrong on purpose, man. Just spell it wrong on purpose. It’s all yours.

I guess you could say A-R-C-H-Y. I could try that.

Bam.

I mean, it’s not done. I didn’t release it. I mean, this is a Skunkworks project at this point. It’s my fun name. Zarch sounds cool… We’ll see.

Alright.

Bye, y’all. Bye, friends.

Bye, friends.

Go code an agent, have fun. Listen to News. Change.news. And we’ll see you when we see you.

Bye…

Bye, Friends.