Must be nice to have a rich parent.
post
Rich kid uses parents' money to hire a company to port the game
hire a company to port the game
That traditionally implies actual humans.
A new way to look at vibe coding to point out that it isn't coding.
It's hiring a company. In the year 2012, you could email someone with a request for something you want made, and if you have money they'll reply back over network the code you need.
AI companies just tightened that loop. You hired a company to write software for you. That doesn't make you a developer or a programmer.
It's the difference between: wow this 14 year old developed a pc port for ocarina of time
Vs.
Wow this 14 year old spend money to develop a pc port for ocarina of time
My team has been trying for years to get apps for internal use and we have waited for IT to either buy or make something. One day I just said, "fuck it" and started building the tools we need. I'm not anywhere near done, but the team has started using them, and it's increasing productivity.
So someone thanked me for coding the tools, and I said, "I'm not really coding. I'm more acting as a product manager. I gathered the requirements from the users, brought it to the developer team, and act as the go between." I think that's the best analogy. And like a product manager, I have to take on faith that the devs are doing what I ask without being stupid.
People say shit like this all the time, and it always irritates me.
"We built a house down south." No you didn't, you hired a contractor to build a house for you.
AI has certainly made it worse, though. I keep having conversations with people where "I'm writing a novel" somehow turns into "I'm having ChatGPT write me a novel."
I have an entirely free LLM I regularly run locally on my SteamDeck. I didnt sign a contract for anything, I just downloaded it.
I could power the Deck running the LLM indefinately with maybe $300 worth of solar panels and a camping grade power station.
Now it ain't nearly as fast as hooking in to an LLM thats being run on a data center, but that's tool use, not contracting an outside firm.
We can democratize this technology if we want to, we do not have to rely on environmentally destructive, oligarch enriching cloud computing.
Could I ask what model you use locally? I’ve got a bunch of computers and vijia cards and would love to experiment with a local model. I’ve never used the subscription models, but I’d love to play around with something secure and private!
Bazzite -> LM Studio (installed via BoldBrew)
Uh what... model am i on now lol... hold on...
Qwen 3.5 9B DeepSeek v4 Flash by 'Jackrong'
But yeah generally, Qwen 3 / Qwen 3.5 type models, will work reasonably well and reasonably fast, at 4B or 2B. 9B works if you close basically everything other than LM Studio lol, or, this latest DeepSeek v4 Flash variant is considerably more performant than just standard Qwen 3.5 9B.
But yeah the way Bazzite does LM Studio via BoldBrew, it is at least theoretically reasonably containerized and isolated... not even gonna pretend I am competent enough to give any guarantees there, but also, I don't think anyone is capable of saying anything LLM related is 100% certainly correctly understood.
Oh uh, yeah, you're gonna wanna be docked/plugged in to a power source, if you want more than ...idk, 0.5 to 1.5 hrs of battery life while the LLM is actively working on something.
Thank you so much for your detailed response! My user tag I don’t remember setting for you proves correct again!
I never ended up picking up a steam deck cuz I don’t leave the house, so I’ll be using a desktop with maybe a 1070 or 2070s maybe? I’ve never tried Bazzite but I’ve got a bunch of different distros on various machines at home. I’ll throw Bazzite on my Medicat drive. Thank you so much again!
YMMV with other hardware, I genuinely have no clue rofl!
Maybe start with 2B... try to work up from there?
Maybe Nvidia cards are just... better generally, due to generally better CUDA support for LLMs than ROCm?
I have no idea!
The Deck uses an AMD APU, RDNA 2, shared LPDDR5 RAM... so... yeah i have no idea how to translate that to a more standard PC set up, in terms of performance expectations. Tooooo many variables.
That sounds like a cool setup, what kinda stuff do you use it for?
But it's also used by actual programmers with strong credentials who seem to actually use them intelligently (e.g Torvalds) which muddies the water a lot for the general public, and allows these people who are basically the programming version of a script kiddie (leveraging powerful tools they barely understand to accomplish things that look impressive on a surface level, and then misrepresenting the results).
We added a child-sized crank to the orphan crushing machine so they can help participate younger than ever!
Of course there are now whales in Open Source, lol
I wonder how far $9000 would have gone if it was donated to real people working on a port instead.
In either case I'll never lower myself to voluntarily consume AI slop.
Answer: something that would've run much better on much lower end hardware and worked for more than one game, but would've taken much longer to produce (since actual human thought went into it) which is the only metric people seem to pay attention to.
Immediate convenience is the metric we've been trained to exclusively focus on. Bread and circuses and all that. It's why I've been purposefully adding inconveniencea to my life to try to build up some of my patience and preservation again.
Also more money. $9000 gets you one dev for what, 2, maybe 3 months?
$9k won't get you far in the US or most of Europe.
Given that he is almost certainly using a $100 or $200/month plan, he could probably only afford a 30-60 minute consultation.
🤷♀️
Lets say you get a dev working for 320 hours over the course of two month on the project for that money. I hate AI as the next person, but I honestly don't think he would have achieved anything noteworthy to the Normies in those two month.
$9000?
At $45 per hour ... call this a contract job with no taxes or health benefits or anything ... thats 200 hours of dev time or 25 8-hour workdays.
You'd almost certainly need at least 3 months (probably more) to lay a solid foundation and get a base level, functional proof of concept, if you had only person, one pretty darn highly skilled low level expert working on this.
So $9000 would produce nothing useable, just another call for fund raising.
And I'm sure you could find many programmers qualified and capable of this task who would call $45 an hour for doing this solo, lowballing them.
Doing a project like this with only humans is certainly possible, and likely would yield a higher quality result, but it would take substantially more time and money, orders of magnitude more.
Sorry, but just being entirely, blanket, always mad at AI/LLMs is stupid.
These things do actually have very sensible use cases, and reverse engineering a giant undocumented black box program is one of the most clear cut of these.
Its also possible to run these things locally... its slower than a cloud LLM service, but its still very usefully fast, and isn't funding some corpo datacenter.
99% of the problems that are being caused by LLMs right now are due to idiot humans using them in very obviously stupid ways.
LLMs are tools, like hammers, guns, chainsaws, forklifts, etc. If you don't know what you're doing, ita very easy to cause a lot of damage. Also like other human tools, you cannot uninvent them.
This kind of use of an LLM, as a tool assisting in a very complex reverse engineering project, directed and overseen by actual human experts, not just trying to naievely one shot it with a novice in charge... is almost certainly the optimal or most reasonable way to balance the [fast, cheap, good] factors in approaching a project like this.
I'm not saying this random 14 yo is doing this in the best possible way, not even close, but I am saying that if you think just $9000 thrown at a or multiple humans could produce anything even comparable, then you do not understand anything about the actual realities of working in the computer software development field.
Its also possible to run these things locally… its slower than a cloud LLM service, but its still very usefully fast, and isn’t funding some corpo datacenter.
But it wasn't done locally, which begs the question: can local models reverse engineer on any kind of reasonable time scale?
That is a genuinely good question!
Which I am currently investigating in my spare time, lol.
I don’t know, by my little ollama GPT OSS is pretty great at merging mods together for games and updating old mods to work with new game versions. under my guidance it’s been pretty effective, I also have been a Software Engineer since 2019.
But again this is under the category of not super difficult and basically tedious stuff I don’t feel like doing anymore (been modding games since Oblivion 2006) and I just want to say this is far more convienent.
Sure there are many bugs still, but I’d say the most useful feature isn’t making something new or reverse engineering it to be implemented different or something else. It’s just altering it enough to better fit my needs, but at the drop of text like during gaming. Usually via merging mods into mod packs and working through the bugs myself.
But I have never let it touch anything real or used for work.
Tedious stuff
This annoys me a lot with complaints about "vibe coding". Like, I know how to do it, I just don't feel like spending 5 hours building and debugging a script I am probably going to run once to convert a bunch of data. I just want the task done so I can move on and do actually enjoyable things.
I get all the arguments against AI "art" and "writing" and "music" etc, actual creative areas. But coding is just tedious boring nonsense 99% of the time. It's not the same. Get the computer, to translate human words to computer words, and be done.
Thank you for a nuanced take in the middle of this generated outrage.
Meanwhile, the persone above you with 24 upvotes advancing estimates with no other source than "trust my gut instinct bro."
14 year olds waste money on all kinds of stupid shit. If he wasn't spending on this it still wouldn't have been donated to a homeless shelter or something like that.
I never blew $9000 when I was 14.
I never even saw that much money at one time until I was in my 20s.
I've gotten in the habit of looking at all software for LLM commits on their Github page.
Check out the SlopScan extension
I will, thanks.
Edit:
Brave gets an F with 47 Claude or Copilot commits.
Cake Wallet gets and A+ with 48 Claude commits.
Notesnook is in S tier with 8 commits from Copilot.
Not a fan of how they score.
Yep, and if there's no .md or co-author, I'll check the first commit and that will usually expose them.
Anyone, who actually already has some experience in developing software and takes on a serious project on their own, doesn't just start using git at already 50k lines in. All slop projects, I've seen thus far, have like 20k - 50k lines in their first commit, or, in case of this super hyped (totally natural hype, btw) KytyPS5 - 200,000. Yes, TWO HUNDRED THOUSAND, in their first commit. Like how do you only think about version control at already 200,000 lines in. Utterly incompetent. And remember, if they had just moved the remote (gitlab to gh, f.e.), the full git history would obviously be there.
Now, ofc LLMs are gonna be way more competent than these people would, buuuuut, since these people have no fucking idea what code they are pushing, I am not trusting them, ever.
So the developer is Anthropic 🤷♂️ If I order some specific pizza, am I the cheff who made it?
It took piracy to make AI. It took piracy to make AI profitable.
i just called out another "developer" for saying they made a program, and then said they had claude code it for them and all they did was "test" it. not a fucking developer if you didn't write the fucking code. it's like saying i just came back from a trip to Hawaii and all you did was look at pictures of Hawaii from google image search on your phone
I am very confused. $9,000 in API costs, but also Claude Max subscription? Shouldn't it be $100-200/mo?
I had thought its either/or, either you pay per token, or a fixed subscription.
I think the number is just what it would have cost in pure AI tokens if they weren't using the subscription.
They said in the article that this kid was making new accounts whenever he would hit the "monthly max" on his old accounts, implying that the subscription still only gets you a certain amount of tokens. I dunno personally because I've never used this stuff.
The article says something about making new accounts when they ran out of tokens or whatever.
I have no idea how much usage $200/month gets but it sounds like they burned through it, the paid another $200.
For those of you who feel sorry for Nintendo with this situation, I blamed Nintendo. It's come to the point where they have to realize that PC is a viable platform that they need to utilize instead of avoiding it. When they avoid it and don't do an official release, this is what they get. Your games on PC are inevitable now with or without you. It's a reality. Adopt.
what did they get? 5 million units in sales in the first week. The vast majority of the Nintendo audience doesn't buy gaming PCs. The real problem for Nintendo is something like an R4 cartridge but for the Switch 2
That's not even an issue for them now with checking unique IDs. They're just outright banning entire switch 2 consoles that use duplicate unique cartridge IDs which appear after dumping the game data onto a Migswitch cart. They see the same ID in two places at once? That ID is poisoned now. Any time that ID comes online, it bans a console. The fact that they're doing this and still remaining profitable is a testament to customers blind love for this company's products.
It has also killed the second hand game market for me regarding switch and switch 2 games.
I don’t feel sorry for Nintendo, I think it will be funny when they inevitably take legal action against this ignorant arrogant child. As is tradition.
Can't wait for the shoe to drop and horror stories of the "hEaD dEv" have their computer wiped because the generative AI agent did something it wasn't tasked to do. That's always the hope so that way we can laugh at their punishment.
That, or in the case of generative AI, I am willing to side with the capitalist cthulu cult leaders since I hate the generattive AI cthulu beast "we" are creating even more. In other words, for once would absolutely side with sintendo if they sued the kid into taking down "their" "work" all because of the usage of generative AI.
Hell, maybe the parents, or more likely some 3rd party paid by the parents if the kid is constantly going through accounts and can afford to do it, would then step in and teach that kid a lesson. I'd hope it's a lesson on being a decent human and not using generative AI for anything, but if they're a rich enough family, you know there won't be any serious life lesson.
Makes more sense to sue Anthropic for producing the code rather than using the kid. I'm all for consequences but I can't imagine a 14 year old has the money to fight such a lawsuit and even if the parents do, it's very likely that Nintendo wouldn't get much out of them if they won. And it's not going to disencentivize others from trying the same thing while the medium to do it still exists.
I legit started to cheer for Sony and Nintendo recently lol. I hate how people act like this is a good thing.
Some random 14yo kid has 9k dollars to spend on Claude code. Fuck my life.
The issue lies with knowing what the code does that AI produces, if you don't have a clue and still use it that is when you are a vibe coder in my book.
If you just use it to increase productivity and improve code then I don't see any issue with that and it does not make you any less of a developer. Times have changed, Stack Overflow is from the past.
I program longer than that kid lives.
If I slop out code, I still see it as a piece of crap. It doesn't matter whether I understand it or proof-read it. If I ever do that, I stop being a programmer.
You don't see any advantage to having code auto written to spec and then being able to tweak the result? Just simply as a time saving/efficiency hack?
Honestly curious, not baiting here.
The thing is that it's an extra slog because the auto written code 'looks' right much better than it is right. And reviewing that sort of code is exhausting because you have to be on your toes at all times.
Also, it suggests you have a comprehensive and suitable spec that describes everything, just in prose. At least in my experience, the spec is a token starting point with the design being fleshed out as you go in practice. Trying to make the spec comprehensive up front can be much more painful than guiding the implementation piece by piece.
It all depends on what the scope is, and how vague your specification can be and still drive results. For example, "take this existing running code and assets and make it run exactly the same way it should today, but on a different runtime" is roughly a scenario that LLM can be pretty free with and is more likely to get closer to what you want than a new sort of thing.
I'm surprised at the negative response to my comment, the statistics do not lie about Stack Overflow usage dropping rapidly. AI clearly objectively helps people. It is not all slop, I think people think about AI emotionally and see everything it touches as bad.
It actually has been insane though. I remember as a kid losing my n64 in a flood and waiting months at a time for single bug fixes to pj64 and some other emulator, plus the graphics, audio, and control plugins. Then in recent years while I was too busy with work recomps like ship of harkinian dropped and gave native+ performance on literally any hardware.
Now on top of the existing iffy at best legality we have llm code, but its moving freakishly fast. Literally 2 days before ocarina released I tried it out on one of the myriad forks of the eden switch 1 emulator, and based on the before times I would have said its months away from 80% playable.
Its been 6 days? It started at playable if you are okay with relaunching to clear full screen graphical bugs and just tolerating things like leaves disappearing when you get close to them, or inconsistent fps that capped at maybe 46fps even for people with bleeding edge gpus. Now barring specific gpu issues you can expect fairly consistent 60 frames with some new scene hangups from shaders compiling, and maybe the occasional model flashing white or glow from behind geometry seams for like 1 frame. On a still decent am4 cpu and a 5700xt gpu.
It's most definitely not just one kid's multiple ai subscriptions though. Its like 30+ with some of them also having multiple. There are so many forks of forks of the emulators just cherrypicking single bug fixes amongst many that are just ai bullshit that only had the appearance of working under very specific conditions. Currently as the gamer finding which fork gives functional performance on your specific hardware is the hard part, but eventually the working fixes will consolidate on the main forks.
I hate to be "that guy" but who/what do you think is writing Windows updates
I mean I assume that’s what’s happening and why it all fucking sucks.



all 95 comments