Welcome to the standup. Today I have

with me Teach TV Trash Dev and Casey

Miratory, legendary game dev and snack

influencer. And we are going to be

talking about Codeberg today.

Uh anyways, sorry for those that are not

familiar. Codeberg is an online code

hosting platform with a focus on free

and open-source software. It is part of

their mission. A big part of their stuff

is that people who open source their

code should have a extremely permissive

license for everybody to be able to

enjoy and bask in the glories of

open-source software. Uh with that said,

there's been a recent proliferation of

LLM generated code in which is putting

strain on anyone that hosts code, i.e.

GitHub. GitHub has had a lot of issues

and I think we're all familiar with

GitHub's issues. Well, recently Codeberg

came out and said they will no longer

allow mostly vibecoded projects to be a

part of codebergonline.net.

Right? So, you can no longer host your

vibecoded stuff on code. They gave a

whole series of reasons as to why and

that is what we're going to discuss

today. Is co is this a good move by

Codeberg? Are we happy with it? Uh will

this work out for them or is this

terrible? All right. So, there we go. So

this is the addendum that they made. You

must not share projects that mostly

consist of code written by generative AI

tools including services such as claude

open a uh open AAI codecs. Such projects

have unclear copyright status. See the

requirements in section 211 section 213

and furthermore have little safeguards

to ensure that they do not include

harmful code. Those other sections just

uh they're just directly above saying

like hey we have certain standards. We

want certain things to be uh uh like if

you are posting a project, we want it to

be this type of permissive license. You

cannot guarantee that by having an LLM

is what they're effectively attempting

to say. So this is the line that caused

the controversy that everybody is

freaking out about. Okay, we're getting

a 504. So they might actually like they

might be getting hammered right now. Oh

yes, yes, we got it. And so they also uh

posted a blog post and they had a nice

little image right here which it looks

like they've also just recently deleted

which is super unfortunate because it

was right here within the first couple

but it showed the actual voting um it

showed the actual voting. So that has

been somehow removed or changed. But the

voting goes that 2/3 of all Codeberg

people supported the no LLM policies,

whereas about 1/3 said, "Hey, we don't

support the LLM stuff. You should be

able to have LLM projects."

>> I could be misremembering, but I'd add

one caveat to that, which is it wasn't

2/3 of Codeberg people. It was 2/3 of

people who voted, which is 50% of

codeberg people. So it's kind of 2/3 of

50% if you

>> Yes. Yes. Sorry. It was it was of the

yes the voting class of codeberg because

there is two different classes of

codeberg. There's the you know people

that can just host a small project and

of course the people that are members

that are involved that have voting

rights and all that. So that that side

of things

>> well I mean I think it's 50% of the like

it's they were talking about how many

people participated which may just be

people didn't vote like you know you had

rights to vote but didn't vote. I don't

know. Like I'm just saying they they

were sort of it was not clear exactly

how many people out of all of the people

it was. But just a slight caveat.

>> Here we see the elusive programmer. A

simple creature that spends [music] most

of its time working alone, often in

darkness. But what's this? Someone being

wrong on the [music] internet. Our coder

springs into action, reaching top speeds

of 120 words per minute before flash. A

light mode website. The natural enemy of

these code lovers [music] stuns our

friend. The chase is called off. We'll

have to get them next time.

When not on their computers, they can

spend [music] hours drawing crude

symbols on something they call

whiteboards. Researchers have discovered

thousands [music] of dialects, often

with more than a dozen used in a single

office. However, [music] no linguist has

yet deciphered what their purpose is.

Vein creatures, their bodies have

evolved over a millennia to be able to

sit in unusual postures while looking at

themselves online. This will often last

for many hours using the excuse they're

waiting [music] for code review,

depressed to why they're so inactive.

And finally, after a long day of

accomplishing very little, our keyboard

warriors [music] ready for bed. Quick

read and it's lights out.

Good night, little coder.

So, how do I sleep so well at night?

Well, I have Sentry to help me crush

those bugs. And I'm not I'm not talking

about like little teeny tiny South

Dakota bugs that die in the winter. I'm

talking about big mean jungle bugs and

I'm not scared of any of them by the way

just but I can squash those bugs with

seer by century. Their blog post went on

and shared several points about what

exactly uh is their reasoning behind and

a few of the big reasons of course is

that LLM put a natural strain on

hardware costs. So, the replacement of

their hardware and them being fully

self-hosted is being directly impacted.

Going from 700 non-freedom units, euros

to 3500s for a uh SSD is directly

impacting how much uh hardware they can

buy. That's realistic. Second, it

impacts actual infrastructure because

there's a higher rate of commits, more

code being generated, more of the

platform is being used by these uh by

people who can just generate huge

amounts of code and it's largely

untouched. Their big argument was that

you have a single piece of code which is

tens of thousands or hundreds of

thousands of lines long and it's just

one person interacting with it. Whereas

something like say Ghosty or some other

project that's hosted on Codeberg, they

have tons and tons of people interacting

with it. It creates a very healthy

community. And so it's like

disproportionately costing them more for

something that creates no community. And

so they're like, we don't really like

that. And then thirdly, they said, hey,

the copyright status, we don't really

know about that. And then fourth, they

said, hey, we really don't like

single-use software. Meaning like you

shouldn't generate something for your

exact need. Instead, it should be built

upon something that a bunch of people

should be able to have benefits and

that's kind of the hosting or style they

wish to have on their platform.

So I I believe I have given the the

rough outline as to everything. There's

also a very lengthy blog statement um

which is called protecting our floss

commons from LLMs

that was posted um here and you should

have the link to that because that's

pretty helpful because it kind of goes

through all of the

>> issues that they were concerned about

and kind of explained their goals which

is pretty important to understand like

the motivation of a terms of service

change obviously. So that was pretty I I

found that pretty helpful to just like

know why they were doing it, right? Um,

>> yep. I tried to explain most their

reasons. That's where

>> I said I feel like you got most of that

in your in your

>> I missed a few. I missed a few for sure,

but I really Yeah,

>> I missed this part which I don't think I

did very well. Let me say this last part

which was losing trust in each other was

another big point which is when you go

to a as a maintainer when you open up a

project if you have notifications and it

turns out the person on the other side

often well-meaning as they say put in

loweffort poll requests it kind of

builds into you a bias thinking that all

poll requests and all new people are

just going to be a bunch of LLM or

loweffort people and it makes you more

resistant to new people coming into your

community makes it harder for people

that are active in your community to be

able to like really reach you and have

good discussions because now

everything's being passed through this

filter of oh it's an LLM being

bothersome and so it kind of tears down

communities and creates a more toxic

place whereas if you didn't have that

you'd have good conversation much more

meaningful conversation at least that is

their intended argument

>> good summary prime

>> now prime you are the one who wanted to

talk about this originally you you were

like I want to talk about this so maybe

you could start by saying why did you

want to talk about it

>> for me I I I will first start off with

what I thought was their most

interesting point which was the last one

that that they said which is the tear

down of communities. Mhm.

>> Uh I I do think that there's something

very interesting to that because I've

noticed it a lot even just in my basic

communication is that anytime I read

something that's over three sentences

long.

>> I am now like more and more going I

think this person just generated this

response because it's just like way too

it's just like way too uh superolous.

like there's just the language is all

wrong and it just like and I realize

that it makes me stop really reading a

lot of people's responses and it's kind

of tearing down the community whether I

real like it wasn't

>> it was a uh it was my favorite part of

the four-part diagram which is called an

unknown known I didn't know I knew it

but I just knew it inherently and I just

operate under this bias where I just

don't even realize that I'm doing it to

myself whereas like I start looking at a

lot of stuff and going oh this is just

LLM generated without even giving them a

fair shake or really consideration to

their argument and I see why this is

actually tearing down uh tearing down

communities and codeberg being a place

that's probably much more tight-knit

than say GitHub. I bet you a lot of the

maintainers of projects know other

maintainers of project community is very

very important. And so I'm just curious

like how is that something that they

should be concerned about or is it

better just to say hey we we're growing

and adapting to the everanging field of

software engineering and we should just

move away from this even though I can

see why it's beneficial. I find this

part at least very interesting or very

unique.

>> Yeah. I mean I guess one thing I would

say is reading through their blog post I

didn't really hit any part where I was

like well that doesn't make any sense.

like it it kind of was just like very

reasonable. Now, I don't know for a fact

that the reasons that they're giving are

the real reasons like or whether they

just hate LLMs or something and and just

we're going to ban them either way. You

know, I don't know whether it was

rationalization or whether it was

actually like these were the motivating

factors. But to me, it doesn't really

matter. It's like the explanations they

gave were very sound explanations. I

felt like all of them were pretty

reasonable. about the only thing that I

would say as a slight counterpoint was I

was like well you know could you just

turn this into a monetary policy could

it just be that well if you want to do

this like if you want to use LLMs to

generate a lot of code we're going to

charge you a lot of money like that's

going to cost you so like you can't use

this service for free uh you're going to

have to pay us if you submit LLM

generated stuff uh you're already paying

for the LLM so you know whatever or

something like that because at least

that way it would it would be something

where um it could be adjusted. But I

also understand the perspective of if

they just don't want this, like if

they're just like we don't we don't see

this as being something we want to be

involved with at all, then I can

understand the policy because they're

just basically saying like look, we

don't want this at all. So we don't want

to we don't want to pessimize it. We

don't want to reduce its use. We just

want to get rid of it and h have those

people go somewhere else. Totally fine

if that's their decision.

>> I have a quick comment. So the I thought

the same thing as Casey said. I thought

all like this blog post was good. I

thought everything was written very

well. I agreed with everything. What

what I can't get past is like

I don't understand how you can even

enforce such a thing that they're saying

here. So there's like a lot of like

blurry lines where it's like you can't

have a project that's mostly created by

like an AI LLM, which that just blurs

the lines. Now it's like well who's

deciding what's mostly or who's not?

Like it almost feels like you really

can't really have a repo on Coberg

unless you have like some like past

reputation of being like some great open

source person like someone who's coming

in like it almost shuts off anybody new

to like the space. Um because they have

no previous like like credentials or

anything like that. And if you can't

prove that they they didn't create with

an LLM, you're going to almost just

assume just kind of like what you said

earlier, you almost just assume

everyone's using LLM. Like I've never

seen such detailed PR requests or PR

descriptions until like this past

[laughter] year. And every time I see

like a wellthoughtout PR description,

I'm like I'm not reading it because it's

slop. Um, which is funny cuz when I do

it, I don't feel that way. I'm like,

yeah, this is well detailed. But when

someone does it to me, I get like I get

like offended almost. [laughter] I'm

like, wow, you didn't take the time. So

funny. This morning I actually got a on

one of my open source projects I got uh

[clears throat] a pull request and it

was definitely just like autogenerated

and I was like almost pissed off when I

got it. But at the same time like this

whole past week I've been using uh LLMs

to just like knock out a bunch of like

tech debt. But for whatever reason when

I saw someone do it to me and it wasn't

me doing it I could I just got like

super bad for some reason. I'm like I'm

going to close it. But yeah, I don't

know. I can't explain the I can't

explain the psychology there right now.

It's weird.

I I think that's kind of a thing that is

almost to be expected though because uh

everyone

this is common for humans right like

it's just a thing that everyone does

which is you believe that you are

responsible when you use something like

I am a good driver and I drive

responsibly right but these other people

>> that's why I can text and drive case

>> yeah [laughter]

right it's like I can look at the map

but if someone else is doing that

they're just irresponsible Well, they're

going to kill somebody and whatever,

right? And uh you can name just about

anything, gun ownership, drinking,

whatever it is, drugs. It's always like

the people who think that they can do it

responsibly don't just automatically

then think, well, anyone can do it

responsibly. They're like, no, everyone

else could, right? LLMs feel a lot like

that to me. Um, I feel a lot I I feel

sort of ambiently that people generally

are like I think that AI is bad and that

it's generating a lot of slop, but of

course I use it and I use it really well

and I make like good stuff with it. It's

like no, you probably make roughly the

same stuff as everyone else and you're

just like downgrading what they're

making and upgrading what you're making

kind of unfairly. uh because probably

their stuff isn't as bad as you're

making it out to be and your stuff

probably isn't as good as you're making

it out to be, right? Uh and that's just

how it goes, unfortunately. So, I think

you're touching on something very real.

I don't think it's specific to AI, but I

do think it probably is happening a lot

right now.

>> I would say

>> I think that summarizes my feelings

exactly. No sign, but yeah,

>> I was going to say the other thing for

like Oh, go ahead, Trash. No, sorry. No,

no, no. Go for Well, my last comment is

after I read this, I almost felt like it

was like a Masttodon blue sky blog post

and I was just like, "Wow, this is this

is like very pretentious. Like, come

on." But then I get it. Like after

reading this blog article, I was like,

"Okay, okay, this all makes sense." Like

financially, too, it also makes sense.

Um, but I still feel like it's going to

shut out a lot of like future like open

source creators that do use LLM but also

use them responsibly. But to your point,

Casey, it's like how do we know they're

using it responsibly? because I use it

responsibly, you know. So, it all it all

just seems very handwavy, not very

enforcable, and pretty much like you got

to know somebody almost kind of thing,

which I don't really like. I don't know.

>> Tee, you were going to say something.

>> Yeah, the thing I was going to say was

it's uh I I think similar to the thing

you're saying, Casey, but maybe slightly

different, too, is like for my code

base, I don't care if I fill it up with

trash, right? Like if I do something in

my house and I scuff up the wall, I'm

kind of like, "Well, that stinks. I wish

I didn't scuff up the wall." But like if

I was at my friend's house and I scuffed

up their wall, I would feel like really

bad cuz it's like their thing, right? So

I do think there's like a little bit of

an asymmetry

there where like if I do something to

someone else's repo, I do feel like I

should do better than I would do on my

own repo, right? So, I do think like at

least for I feel like a responsibility

that like, oh, I'm going into your

house. What's your rules? How do you do

stuff? I'm going to do it your way. You

leave shoes on. That's disgusting, but

I'll do it at your house. [laughter] My

house, you're taking your shoes off.

Okay. Everybody at the door. Okay.

>> Reasonable.

>> Gotcha. [laughter]

>> Um, okay. So, I wanted to jump back on

uh trash. What were you saying? What

What What did you just say? What was the

Cuz I had a great Oh, yeah. Okay, I got

it. Ghosty. How does Ghosty work in this

world? For those that don't know, Ghosty

was a terminal uh started off being

written by Mitchell Hashimoto in his

retirement years and then gain gained

popularity and over the last

>> He's unretired as of today, by the way.

>> Yeah, he's unretired.

>> He's back in the

>> And over uh over the last like year and

a half, he's been talking about how he

uses AI all the time to generate uh

patches and all that. Like, how do you

even enforce the idea of mostly? Like

why would Ghosty deserve a spot on

Codeberg because we know at least one

person committing to Codeberg uses AI

regularly? Like what is the what is the

the mark of too much slop? Like I don't

I don't know what that is.

>> I I kind of have something tangential to

say. I don't think there is a way to

know, but I have something sort of to

add to that or to sort of it's not

really a pylon, but it's like a thing

that goes along with it, I think, which

is I think there's I based on their blog

post, they have a lot of goals like

there were a lot of things outlined

there, right? And I think there's a

simpler version that has less goals. So,

if your goal is just that you want the

projects to be good and you want them to

work well and you want the system as a

whole to work well, then really I think

the only policy you probably need in

that scenario is no one should be able

to tell this was done by an LLM. That's

the goal, right? The goal should be I

get the pull request and it's been

looked over correctly. There's no

obvious mistakes. The thing isn't too

verbose. like somebody prepared it so

that it looks like a nice human

generated pull request and maybe it was

generated largely by the LLM, but a

human was there to make sure that you

didn't get the slop, right? And so

that's the real thing that like a

project that just doesn't want to drown

in lowquality stuff should be concerned

with, right? But then there's a separate

thing where they've piled on all these

other things which I don't necessarily

disagree with. So I'm not saying pile on

in a bad sense. I'm just saying they've

added a lot more past that baseline of

just don't disrupt the service. They're

basically saying like they don't really

want to invest time in the LLM sort of

economy. They're saying they want to

invest in people, right? I've heard this

from multiple places. I believe this was

the stated reason why Zigg stopped uh

accepting LLM generated political

>> contributions. And then exactly that

>> they basically said I think Andrew Kelly

literally said in an interview he said

something like we want to invest in

people who we believe will become future

contributors to Zigg and we want those

people to be spending a lot of time in

the code. Again don't know if that's the

real reason or not. I can't like intuit

it or or I can't like see into people's

souls and know like what they're

actually saying. But as a statement, it

is very reasonable, right?

>> Yes. If that's your goal is to have is

to cultivate people who are going to

contribute, then not wanting them to be

using LMS as the way that they learn

about the codebase could be a very

reasonable thing to say. Like I don't

know, right? I'm not in a position to

really uh judge that. So I do think

there's kind of this there's kind of

different stages, right? when you get to

a certain point, the goals that you're

trying to sort of go for kind of have to

be like, "Yeah, we don't want you using

LLMs much because we think we we think

there's something to preserve in the

human layer and we're going for that."

>> So, I guess I guess the follow-up

question is that this obviously caused a

huge amount of uh kind of stir on the

internet and I I like you said, I like

their if you take what they're saying at

face value. Yes. and you believe they're

being like honest in all of their

thoughts that it's purely a very

practical educational type of pursuit,

then you would say, "Okay, well this I

think it's totally fine." Like I think

that there's nothing wrong with it. But

obviously the internet kind of freaked

out and there was people just saying how

they're going to get left behind and all

these different kind of things. Uh is

this actually a real concern or is this

just like a Twitterism that people throw

out there that actually has no real

impact into the future of Codeberg?

Does anyone else want to go first

because I have something to say about

the Left Behind thing? That's kind of

its own direction.

>> Oh, with Kirk Cameron, famous movie

>> and book series.

>> Really famous series.

>> And book series.

>> Yeah.

>> What do you mean left behind? In what

sense?

>> Come back, idiot.

>> Yeah. [laughter] Yeah. And then there's

a bunch of people here get left behind

here. Okay.

>> Okay.

>> No, I mean like uh

>> like they're good. I'm talking about

like Cobs, but like I don't understand

like why they'd be left behind if they

took out LLMs.

Well, I think the I the implication is

that if everyone's going to be

programming with LLMs in the future

because that's the new way that people

are

>> well that people are starting to program

now like most people are using these

things or at least a significant portion

of developers are using these things

then if you ban them you are effectively

not going to have enough people right

doing work to to contribute to codeberg

repositories I think I think right I

mean that's just the basic implication

is there won't be enough programmers is

doing programming without these LLMs to

to constitute a you know a relevant set

of of projects on Codeberg. Is that I

mean am I just out

>> they're not banning LLM straight out.

They're just banning like strictly like

only LLM stuff, right? That's what I'm

getting.

>> Well, yeah,

>> but I guess that's like that's where the

gets weird.

>> Yeah. I mean the I I mean I don't know.

I'm just saying like I think that's what

they mean by left behind is just like

because people will be mostly mostly

they will be generating code with LLMs

and if you're doing that you know that's

that's not allowed

>> right

>> so Casey you may not know this but uh

sorry T you can go next sorry I just got

to throw this in there trash knows this

very very well which is a very common

Silicon Valleyism so I think this is

actually a holdover from just another

time which is that if you were to build

a product in 2021

>> and you didn't use React

Mhm.

>> You were effectively making it so that

no one can work on your project and it's

going to be super hard to hire and all

these kind of things like you saw that

kind of phrasing and phrasiology

constantly. So I do think that that that

mentality whatever it is yet again shows

up in these type of arguments which is

well if you're not doing the current

thing then you're done. You're cooked.

You're you know whatever the kid you're

chopped the kids say. I think that's

what they say. All right. Sorry teach

go. Yeah, I was going to say I think

there's

like if you think of Stack Overflow and

it like used to be a company and

basically now it's not is my

understanding, right? Like it used to be

popular and now I think they get like

sometimes days go by with no questions

asked. Am I crazy? Is it that small now?

>> It is. No way.

>> Yeah, it's like it's like 10 questions

on average a day was the last time I

looked.

>> That makes sense.

>> And it used to be like over 100,000 or

something insane, right? Sometimes it

peaked something crazy. whatever. Um,

so I think like from a if you're saying

like from a business perspective, I

think like some of what they're, you

know, what people are getting at is like

Codeberg, you won't be able to like pay

people to work on your stuff or like to

host it or do all these other things

because there's literally just not going

to be enough people like doing whatever

version it is that Codeberg says is like

acceptable coding policy. I think is

like the the conjecture. The thing that

I always feel like is kind of crazy for

that where I just, you know, want to

take a step back and make sure people

are considering like a lot, you know,

spent a lot of time working on Neoim.

Did a lot of fun stuff on my setup,

whatever. I had a lot of fun doing that.

Some of it was productive, some of it

was just fun. I always uh said as much.

I know that a bunch of people absolutely

demolished me using Notepad or Notepad++

or like Ed still and made a whatever

metric you would pick like crushed me in

term number of users, amount of money,

amount of money raised, cultural impact

like you could pick any vertical and

they crushed my entire career, right?

So, I always feel like the like you'll

get left behind because you're not using

whatever the current thing is is like

usually very overstated in just in terms

of like there are people who are just

like either much more persistent or much

more lucky or work harder or whatever

it's going to be and they like still win

and carve out their thing and

unironically uh and without malice the

successful by their own standards a

phrase that I know Prime is familiar

with at least

>> my favorite phrase by the way

>> that's a really good

>> is actually is actually true though if

codeberg doesn't want to be acquired by

Microsoft and be the default place where

people host code instead they want to be

a place that's like a sustainable

business where people can share code

with other humans then they can like

still meet that goal without getting

every new programmer so it's like you

can't you might get left behind by the

like VC hype cycle thing [clears throat]

like you'll never get the growth that

like your venture capital firm would

have wanted you to get, but that doesn't

mean that the project is like a failure,

which is a thing that I just feel like

is so just like everyone's talking way

past each other about that online in the

conversation.

>> I I feel like that's 100% correct. It's

like the idea of getting left behind

doesn't really make very much sense

because it suggests that everyone's goal

is to be at the forefront of something,

right? But that's not right. Like there

are, you know, like you've got people

like the Amish who are just like, you

know what, I'm just not going to use

electricity, right, in 2026. Like people

have to have the freedom to define their

own ways of living. And one of the

things that I try to tell people,

although people rarely seem to want to

listen, is that that is a feature, not a

bug, of humanity. Because you never

really know when humanity in large

numbers is going to go do something

fantastically stupid. Like we have no

idea if AI is going to be a huge

mistake. It might be for reasons we just

can't foresee, right? And so having some

percentage of the population who doesn't

like it reflexively and chooses to go do

something else, that is not bad. Even if

they're wrong, it's good. It's

insurance. It's basically like

humanity's protection

about everyone doing something dumb,

right? So you always want a diversity of

sort of directions forward where some

amount of your populace is not doing the

new thing. Even if the new thing turns

out to be correct, it was not a good

idea to try to pressure them into not

going their own way. Because having that

sort of different set of people doing

different things for different reasons

is why we survive. And the instant you

try to homogenize everybody and say

everyone should do it this one way, you

are effectively like eliminating

humanity's greatest strength, which is

not all doing the same thing all the

time.

>> They did. This was the problem in

Interstellar. They all had the exact

same corn. That's why they had to go

inside of a black hole.

>> I mean, I don't know.

>> Trash. That's literally a bookshelf.

>> No way. What?

>> Then Michael Matthew McConey had to go

behind a bookshelf and like and like

punch it to like get some stuff happen.

It's It's hard to recover from corn.

[laughter]

>> If you all are doing the same corn, it's

GG's. Okay. Timothy Shalomé might

disappear from your timeline and never

get referenced in the last 45 minutes of

your movie. Okay.

>> You get hit by this huge wave, like the

biggest wave you've ever seen, and you

have very little warning trash.

>> Okay.

>> Very little warning. Noted. Noted on

corn. Got it.

>> Yeah. And then then Matt Damon lies to

you about the planet. It's there's a ton

of bad stuff that happens in this.

>> He's a good guy when he's on Earth, but

every time you see him not on Earth,

he's a bad guy. Think about it.

>> That's right.

>> Uh something I I you know, I feel like

people don't talk about is Matthew

McConn spent like years behind that

bookshelf.

>> I mean, it's probably a pretty cool

little space back there.

>> I'm just throwing it out there. It was a

long time. He was he was he was out

there for a long time. patient man.

>> Very patient.

>> Very patient.

>> He got binary.

>> He He doesn't have

>> trash. That is what happens though, by

the way. They all have the same corn and

then they gets afflicted by the same

disease and they can't stop it because

they ran out of corn. That's what they

can't live anymore.

>> I had no idea.

>> You're not paying enough attention.

Trash.

>> I'm not paying attention. I'm not.

>> Yeah.

>> Trash. But the the serious point to echo

Casey's point is

>> if you love AI, you should be thankful

that there exists people putting out new

untainted fresh training and keeping the

light alive of uh trades that we may

forget how to do. We don't actually want

to have the meme where it's like we lost

the technology, whatever the meme is.

Like my dad showed me this TV and asked

how did we put this on the moon? He said

son we lost the technology. You don't

actually want that. So, you should be

thankful that people exist that are

like, I hate AI. I'm going to keep

writing code by hand.

>> That's a good

>> This is This is the exact same thing as

back in the old old days where people

were like, why are you working on like

game engine stuff? Why don't you just

use Unity? And the answer is don't. Why

would you ever say that to somebody? The

people who you're saying that to are the

people who built Unity. Like, that's why

you have a Unity is because of these.

You should just say, "Great, have fun

with that. I'm going to leave you alone

because I want my unity later." Right?

Like that's it's people are arguing

against their own self-interest. Like

people who are LLM afficionados are

arguing against their own self-interest

when they tell other people not to

pursue other p. It doesn't make any

sense. It's like the Codeberg people's

blog makes perfect sense. The people who

are saying they shouldn't do this or

they're going to be left behind. I don't

really take that seriously because

that's not what you would say if you

actually thought about the system as a

whole. And again, so when you look at

those people's motivations, it kind of

feels more like they're just

self-conscious or something like they

they they don't have enough self-esteem

to just go, I'm using LLM and I'm fine

with it and I don't mind if other people

aren't in the band. It's like somebody

else not doing it is like makes them

afraid maybe there's something wrong

with them or something and they have to

get everyone involved in this thing and

everyone has to agree it's great to do

this and that there's nothing wrong with

using them. It's like just have some

self-confidence. You use LLMs, it's

fine. You're fine with it. Go ahead and

let the people who don't don't it it

just what's the problem, right? I don't

understand.

>> Casey, can I read you a QUOTE FROM

GEORGE HOT?

>> ALSO, I called it in the chat. As soon

as Casey said this, I typed in the cat,

"Ge hots reference incoming." Prime is

so excited. [laughter]

>> Tee, I actually had a completely

different point, but when I saw that, I

was like, "Oh, that's a good reference.

I'm going to use that."

>> Your face [laughter] lit up. You did a

little bit of shaping, future shaping by

accident.

>> What is What is it?

>> All right. Well, let me let me show you.

So, this is from his famous blog post

called I love LLMs. I hate the I hate

the hype. Where he also casually drops

that he runs GLM2 just to set up his

[laughter] Linux machine. you know,

whatever. A multi-million dollar machine

just to set up Linux. Uh, you're the

Linux, by the way. Uh, so anyways, he

said this, "What I don't like, uh, is

two things. One, the constant

about the window closing or perpetual

underclass or falling hopelessly behind.

This is a negative valance hype. Not

only is it not true, it's mostly

designed to make you feel bad about

yourself and move to shitty San

Francisco where everything really does

suck like how these people claim."

[laughter] Oh, it gets better, Casey. It

gets better.

>> All right. Again, just in case anyone's

wondering, hots, open invite whenever

you want to be on here.

>> Hot stakes. I love these hots takes

>> dude. He's my favorite writer uh of the

modern age. And two, this straw man

jumped from, oh hey, it's a fancy

autocomplete, smart compiler, better

search engine to it's like going to own

the whole Lite Cone, bro. If you aren't

in San Francisco and at the right

parties, there's going to be a flash of

light in the sky one day and you're not

even going to know what happened. But

everything just changed. I'll bet you

everything I I have that this doesn't

happen. The people per perpetuating this

are terrible people, but the justice is

that this is how they feel inside all

the time themselves.

>> I I guess, you know, all all I'm doing

is is unknowingly paring geoh here. So,

yes, I mean, that's kind of what I was

trying to say, but put more eloquently,

I suppose.

>> I I love that phrase, but uh I mean, I I

I do want to throw this out there that

there's this there's this one big

problem when it comes to AI. Well,

really it comes to anyone talking about

technology, and I do this myself all the

time because Neo Vim's still the

greatest editor if you want to actually

be a real programmer. But, uh,

[laughter] the thing is,

>> is that when you approach a problem,

especially like you'll definitely see

this from people recommending LLM is

that they can identify and say, "Oh,

this is a part that LLMs are not that

good at, but don't worry, it's really

great at backends." Oh, no, don't worry.

It's really great at this other thing.

And and so I I hearken it to like

something like the gentleman amnesia,

which is you open up a magazine, you

read the first thing about horses. You

say, "Wow, they're so smart about

horses. Look at all these cool facts."

You go to JavaScript and you read it and

they're like, "Oh, they know nothing

about JavaScript. They're so they're

such idiots." You turn the page and

you're like, "Oh man, this be article is

so amazing. Look, I know so much about

bees now." It's like you completely

forgot the fact that when you knew

something, then it was not that good.

But when you don't know something, it's

like the most fantastic thing in the

universe. So it has some strong

parallels to that. And I think that

people often forget that what they work

on and what they're doing may be very

good with LLMs. That doesn't mean

everybody has a problem that's really

aligned with LLM. Like if I was just

building uh some SAS tools that is

taking stuff from the back end and

putting it on the front end, like you

bet your sweet ass I'd be using LLM like

wild. Like I'd be setting up every last

thing cuz it's like this is such a

mechanical obvious, you know, put the

data in the box, bro. Like there's not

like a lot you got to do there. And so

it's like this is a perfect example of

you saying, "Hey, you could probably get

to the point of even automating PR

review through a lot of this stuff with

barely like any oversight." Whereas like

that you can't do that for everything.

But people see this one thing and they

go, "Dude, why aren't you automating PR

review? It's obvious. You should do it

this way. This is the future, idiot."

But like their their circle is so much

different than other people's circle,

but it's the you know, I see it this

therefore everybody has this exact same

thing. And also they feel really crappy

on the inside based on geohods. And so,

uh, that's kind of like how I look at

this and why people are angry at

Codeberg is because they see their

future and what they're doing and they

see codeberg going against it and they

cannot imagine a world in which is

different than their own world. Well, I

would also just point out the difference

in goals at least because from my

perspective I I've never really been

super concerned I mean I guess I am

concerned sort of because I have to use

software about how good LLMs are

compared to human programming, right?

I'm slightly concerned because like if

they were if they don't get any better

maybe that's a problem because people

are using them. So I I you know I kind

of want them to improve in the sense

that if they don't improve we're kind of

in trouble. But you know that sort of

that's a separate thing. So assuming

that they get good enough that it's just

fine all the time and they can write

really good code and we don't really

have to worry about it too much. to me

that's really just not part of the

equation at least for me because it's

like I like to program like I want to do

it right and and so when I look at

things other things in life uh I think

te used chess as an example before I

find that very compelling it's like if I

want to play a game of chess I'm not

thinking well I'm not going to play that

game of chess because I know that I

could get a computer that can play it

better than me [laughter] I still just

want to do it right

>> you know

>> it's good

>> yeah so

>> I Like TJ, I I actually even prefer

that's one of TJ's really great

analogies. A second one of his is the

Amish chair one, which you kind of

alluded to, which is Amish people still

make chairs that I can go buy.

>> Yeah. Well, and the other thing I was

going to say is

>> why not just massproduce it from IKEA,

idiot? Just use the massproduced one,

dummy.

>> Drum drum machines versus live music,

right? It's like we have things that can

do stuff on a type. So, it's like I may

just want to learn to play the drums

because I enjoy playing the drums. It's

not that I think I will be more precise

than a drum machine or whatever.

[laughter] It's just that I wanted to

play the drums, right? And so to me,

those things should be acceptable. Like

if you say that, people should not be

likeounding you for some reason. It's a

perfectly rational thing that exists in

all other things. Like, hey, we could

probably design a robot that hits a

baseball really well. That doesn't mean

that we shouldn't have people playing

baseball as something we watch. Like,

it's just it might just be something we

want to do and that should be fine. and

you should not care because if again if

you're not really self-conscious or

something and you and you actually have

some self-confidence, then if you want

to go build the future with LLMs, those

people are not in your way, right?

They're like not they're not creating a

problem by going and doing something a

different way. So, what's the what's the

issue? Like, why do you have to convince

them not to? I I don't I don't see it.

Hey guys, if you like this episode, you

can watch the rest of it on the Spotify.

And don't forget to LIKE AND SUBSCRIBE.

WOO! WOO. See you later.

>> Boot up the day.

Voting [music]

errors [singing] on my screen.

Terminal [music]

coffee

and

living the dream.