Even a dog can vibe-code! And the apps kinda, sorta work most of the time, like most apps vibe-coded by people!
I'm reminded of the old cartoon: "On the Internet, nobody knows you're a dog."[a]
Maybe the updated version should be: "AI doesn't know or care if you're a dog, as long as you can bang the keys on on a computer keyboard, even if you only do it to get some delicious treats."
Thanks for the kind words. I'm blown away by the response and positivity here.
There's definitely some social commentary to be had in the whole project. I decided it's best left to the reader to find their own rather than assigning mine to it.
A few days before this article was posted to HN, I had commented (https://news.ycombinator.com/item?id=47086836 ) on a post triggered by the drop in quality or engagement of the Show HN page. I was playing with the phrase "everyone and their dog" that the person I was responding to used and saying the "and their dog" part was more problematic than the "everyone" part, drawing a parallel between the dog and an LLM by implying that the solutions they both would produce would lack the guarentee of human intentionality and ownership.
And then your dog read my comment and said "hold by biscuits" I guess.
The human built the system, the AI did the implementation and the dog provided the "intent" even if that intent was just treat seeking randomness. It turns software creation into something that looks less like writing and more like cultivating
I had to smile to myself when I noticed contrastive negation—with em dashes!—in a text written in the 90s. LLMs don't dictate what good writing is; people do.
we will almost certainly see a mass reactionary movement across disciplines to adopt styles out of distribution for LLMs, a cultural arms race that will be discussed in depth in the decades to come.
Funny idea, but this proves my point that these tools are actually just slot machines.. Except the house in this case takes the money you give them lights it on fire.
Notice how people also have weird superstitious habits when using LLM tools, "You gotta write the prompt this way, say this first" Without having any way to prove it works. Its very similar to the behavior of gamblers. "push the buttons in this order for best outcome"
Also notice how llm tools allow you to multiply the output X2-X3-X4 to compare the ouputs, this is literally UX straight outta a casino.
Many of the users also exhibit excited, almost manic like states.. Addicted to the dopamine the output from their prompt produces...
This is going to be a weird trend to look back on, the hype is on par with the same gambling trends found in crypto/NFTS.
I think this is more of a statement of human behavior under uncertainty and non-determinism rather than the tools themselves. Perhaps the ease of use brings it closer to the funny analogy you made but I think you will find this in any system where users interact with a partially opaque mechanism that produces different quality outcomes contingent on their input...
Sorry, this can't be anything but an intentionally obfuscating comment that I need to call out.
> more of a statement of human behavior under uncertainty and non-determinism rather than the tools themselves.
This is basically saying "It's not gambling, it's just the psychological underpinnings that form the foundation of all gambling enterprises". Who cares to split this difference other than casino owners?
Yeah. You always know you are doing something pretty unique when the LLM can conceptualize it (produce the right English output) but not put it into code.
I think this was the most important insight in the article:
> I experimented with Rust/Bevy and Unity before settling on Godot. Bevy’s animations and visuals weren’t as crisp, and Claude struggled with its coordinate conventions - likely a combination of less training data and Bevy leaving many core features, like physics, to the community. Unity was a constant struggle to keep the MCP bridge between Claude and the editor healthy. It frequently hung, and I never figured out how to get Claude Code to read the scene hierarchy from the editor. Godot’s text-based scene format turned out to be a huge advantage - Claude can read and edit .tscn files directly.
Didn't expect Godot to be the most friendly game engine for LLM usage! I think it's because of various factors - Godot has been used quite a lot in recent years so there are various code examples on the Internet, and its scene file format (.tscn) is very concise enough for LLMs to write and edit directly (Unity has its own YAML-based format but it's very unfriendly for human consumption, and Unreal stores its core assets in binary files)
I've previously struggled getting LLMs to manipulate tscn/tres files since they like to generate non-unique uids. Despite being text files, the godot tscn/tres files are normally meant to be manipulated by the editor and need to define and reference unique ids. The editor always generates completely random alphanumeric strings, but LLMs like to use names or placeholders (e.g. "aaaaa1", "example", or "foobar") for the ids.
The linter in the article that detects duplicate uids is interesting. Obviously the article is about creating a bunch of harnesses for the LLM to be productive. I wonder how many problems can be transformed like this from something LLMs just can't do reliably to something they just need to burn credits for a while on. The LLM probably can't tell if the games are fun, especially with it's rudimentary playtesting, but who knows.
regarding the non-random ids: I had this issue with uuids. Now I have "Never write your own ids. Always use uuidgen to generate real ones" in my AGENTS.md and haven't had this issue for a long time now.
I'm playing around with a tool to generate the IDs for me. I'm honestly not sure if it'll be an improvement since it likely means more tokens/context than just letting it YOLO IDs.
The model makes a huge difference. I tried this about a year ago and Claude occasionally got it right. These days, it seems to get it right on the first try most times and then always self corrects after. Codex 5.2 (I haven't played with 5.3 enough yet) gets it wrong more often than not, and frequently doesn't call the linter; I'm willing to accept that my bloated CLAUDE.md might be a bad fit for Codex and causing this to fail.
I also want to throw MonoGame into the mix here. Since its purely C#, Claude Code works great for it. It does mean you dont have the visual engine tools you get with Godot, but you could even get Claude to build these for your game.
Im personally finding it a lot of fun to work this way.
I read that as MomoGame at first and was very confused.
I'll have to give MonoGame another try. I was a big fan of XNA up until its deprecation. I went all in on OpenTK for a while, and in hindsight MonoGame would've been the better choice.
I made some small tries to vibe code games in Godot, and I was surprised about how far you can go even in 3D. This was just a test of the bad kind of vibe code (you know, not even looking at the code, starting right away, and so on), but I believe that with some good practices there are a lot of things that can be done.
the real takeaway is buried at the bottom: "the magic isn't in the input, it's in the system around it." random keystrokes producing playable games means the input barely matters anymore. we're basically at the point where the engineering is in the scaffolding, not the prompting.
That also shows the delusion of some people that believe their vibe coded projects have any value.
If generative AI improves at the rate that is promised then all your "promting skills" or whatever you believe you had will be obsolete. You might think you will be an "AI engineer" or whatever and that it is other people that will lose their job, that you are safe because you have the magic skills to use the new tech. You believe the tech overlords will reward you for your faith.
Nope. You are just training your replacement.
No one will buy your game that you vibe coded. If the tech were good enough to create games that are actually fun then they would just generate their own games. Oh your skill? Yeah, a dog can do it.
Yes people will cope by saying but oh the whole initial prompt and setting it all up was still hard but yeah currently. The tech will improve and it will get more accessible. So enjoy the few months you are still relevant.
Of course there is reason to believe that you can't scale up LLMs endlessly and bigger models hit diminishing returns. In fact we might already be seeing this. So there is an upside but then again when the AI bubble pops and the economy crashes you will be out of a job all the same.
Sounds great to me. Software devs might lose their jobs but billions of people will be empowered to spin up whatever software they need on demand. This is the future I dreamed of when I was a kid, and I'm not so cynical as to let the dying of a trade sour me to this objectively incredible technology.
Nobody gives a damn about the dying of a trade. People don’t want their house foreclosed on when they lose their income, or their cancer to kill them when they lose their health insurance, to move an elderly parent into a cheap shitty old folks home because they can’t afford home health care, or not be able to pay for their kid to go on that school field trip.
This would all be pretty fucking swell if the fundamental problems this could cause were even considered before hitting the gas. Instead, you’re going to have a shitload of people with ruined lives, but as a consolation prize, they can vibe code stuff! Wowee!
This very forum was founded by a VC who had great success recruiting 22 year olds with fancy diplomas to automate away the job of the guy who copied the numbers from the TPS report pdf attachment into excel.
I didn't see people on here ranting and taking up the flag of revolution for the TPS report excel paster guy's job that they were automating away with their web2 SaaS startup.
But wait- that guy himself was automating away the job of the lady who used to physically Xerox the TPS report and put it in the filing cabinet down the hall, but that lady was automating the job of the secretary who used to re-type all those TPS reports.
It's automatic filing cabinets all the way down, and ranting because your little slice of the filing cabinet automation machine has been made redundant is a bit silly.
You act as if this is the first time in history technology has wiped out a trade and made people scramble to sort out their lives. No, this has been happening over and over again throughout history and at a rapidly accelerating pace since the industrial revolution. Why should we ask it to slow down on behalf of programmers, when it never did for anybody else? Don't pretend you didn't know this was a possibility when you got into tech in the first place. You might have to downsize your life but humanity as a whole will be better off.
> You might have to downsize your life but humanity as a whole will be better off
This assumes that there will be other jobs to get. If AI replaces a large enough segment of office jobs then huge portions of the population will be unable to afford essentials like food and healthcare.
Walk into a staffing agency, ask for a job. They'll give you a list, pick the one that sounds the least disagreeable. Show up on time, every day, for at least two or three months and you'll convert the temp position into a full time job.
It's literally that easy, showing up reliably is a super power that puts you in the 90th percentile of workers these days. The job probably won't be as comfortable as sitting on a comfortable chair in an air-conditioning office wiggling your fingers at a computer, but so what? Other people make it work, so can you. Man up.
sorry, no jobs at the staffing agency, those are AI. Feel free to walk into a burger king, show up everyday, and flip those fries for minimum wage until you die. Man up brother, other people make it work. Sleeping on the street, well half the year its not even snowing.
The last study I know of that measured the conversion rate from temp to perm employees showed about 15%-30% success… and that was well before the gig economy really took hold. So you’re looking at 4 or 5 temp placements to reliably get a probably underpaying job when very few white collar workers could survive long enough to make the end of a lease, or sell their house, while on a temp job salary. It’s a viable option for a 25 year old that could couch surf for a few months, but not for a mid-late career professional, or anyone with a family.
You can give any complex problem a simple answer if you ignore enough factors.
The industrial revolution resulted in children being worked to literal death. Of people toiling 16 hours ad day and living in cramped up spaces without any windows and barely any hygiene. It brought suffering on a scale never seen before.
Organized labor movements managed to fight back and improve conditions somewhat but will we be able to do it this time?
Humanity will not profit from generative AI, tech billionaires will. It is based on the theft of human labor of millions of programmer, artists and writers without any compensation. If left unchecked it will destroy the environment, any form of democracy, our mental health. It will cause mass unemployment at a grand scale.
Could it be in theory used for good? Maybe. As the current political situation stands it will cause massive suffering for the majority of people.
Your glib dismissal of the real effects of those technological upheavals shows you haven’t actually looked into this. You should probably tamp down that smugness until you find out.
Did you think about it before you got into the tech industry? You should have, technology has been wiping out jobs since forever but you got into tech anyway. Live by the sword, die by the sword. Except you needn't actually die, just walk into a staffing agency and ask for a new job. I have done so before, and will do so again. I have prepared for what is to come, saw it coming 20 years ago and saw the imminence of it when GPT-2 was released. I have sympathy for other kinds of white collar professionals who never could have anticipated these kind of developments, but technologists? Give me a break. You knew, or should have known, that technological developments in this domain were likely.
Please stop pretending that this is only going to replace "tech workers". Do accountants "live by the sword"? Whose jobs did they replace? What about analysts, journalists, radiologists (one day, if not quite yet)?
And even within the realm of "tech", it's kinda bonkers to expect e.g. a firmware engineer to have some deep understanding of trends in ML/AI.
Altogether your #1 priority seems to be "bashing workers", the justification just being a matter of convenience.
Please stop pretending to read comments before replying to them:
> I have sympathy for other kinds of white collar professionals who never could have anticipated these kind of developments, but technologists? Give me a break.
I’m not in the tech industry anymore because in the battle of people who wanted to solve problems with software and money grubbing MBAs, the money grubbing MBAs have won. Now I’m a union machinist, and believe it or not, I’m concerned about the wellbeing of others. In manufacturing, companies are starting to face the consequences of shortsightedly selling out their workforce and are frantically clamoring to use the agonal breaths of its existing manufacturing industry knowledge base to breathe life into a new generation of workers. China becoming a manufacturing powerhouse wasn’t a foregone conclusion: we gave it to them in exchange for short-term profits. Our economy, national security, and the financial viability of a robust middle class is paying the price for their greed and arrogance.
The people running the tech industry can’t see the world past the end of this quarter, so they’ll never learn the lessons our society has learned many times over. Good luck. Unless you’re running a company, you’re going to need it. The soft, arrogant, whiny, maladroit white collar workers coming into the trades are pathetically ill-equipped to do actual work.
You've already done all that I can advise others here do, so congrats, I have nothing to criticize there. You've done it better than me actually, since you're unionized. As for soft white collar wimps washing out, people at the first job I had out of tech were taking bets if I'd show up for the second day, so don't think I don't know what you're talking about. I know it, I did it, and other people can do it too.
The problem with exporting manufacturing to China was this country lost the ability to make shit. I don't think this maps at all to white collar jobs getting gutted by AI; the people who actually make things aren't the white collar workers who should be sweating. Societies paper pushers would effectively be a parasite class leeching off the hard labor of people who actually work, if not for the part where white collar workers are (or have been) necessary to organize the logistics of everything that allows the people who actually do the work to actually do the work. We are on the precipice of dramatic change, and I think we're going to see a radical revaluing across society.
None of this is even new. Computers and other business machines already came for the clerks and secretary pools before most people ITT were born. The loss of these careers was not even remotely a problem for society at large, completely unlike offshoring manufacturing.
> Billions of people will be empowered to spin up whatever software they need on demand.
… so long as they have the money, and the power grid survives the overtaxation.
After all, why bother encouraging a culture where people are genuinely empowered to tweak and create their tools? Why encourage a culture of exploration, of playful cleverness? What use is there to being a hacker, of sharing knowledge?
It’s definitely much easier, more sustainable, and more fulfilling to have server farms adjacent nuclear reactors make your calculator app for you.
Hardware will get better, local models will get better, it won't be very long until running a coding agent off a few solar panels on your roof is viable. And if this isn't true, then the future I envision probably isn't going to happen and you lot can stop worrying about your jobs.
The way I see it is either it will work (which I doubt, but whatever) and I'm out of a job, or it doesn't work and the economy comes crashing down harder than ever seen in history and I'm out of a job.
The bubble popping is a guarantee even if AI does end up working. The internet certainly works but that didn't stop the dotcom bubble from crashing. Live frugally below your means to save up as much extra money as you can, and do whatever else you think prudent to keep your ass covered.
I’m not worried in the slightest about jobs. A world where acceptable slop is created on demand is one where people lose their curiosity and drive. Why bother practicing drawing if you can prompt an acceptable facsimile of your image to life? Why bother learning to program if you can prompt an acceptable slop program onto your machine?
Killing the drive to learn and explore is not empowering; it is fundamentally disempowering.
>billions of people will be empowered to spin up whatever software they need on demand
As long as:
1. They have access to a computer
2. They have affordable access to a capable language model
3. Someone will actually care about using their output instead of simply spinning up their own custom version of whatever idea they have
The number 3 is something many people miss, especially on HN: Why would I want to use YOUR software if it's easy for me to cook up my own? Perhaps out of efficiency or lack of time, in the same way I order pizza instead of baking my own when I'm tired or can't be bothered to bake pizza.
Then the software becomes truly throwaway, in the same way takeaway is, and everything is a greenfield project because rewrites are literally easier and faster to make than patching up existing stuff.
> Why would I want to use YOUR software if it's easy for me to cook up my own?
You're still in the mindset of thinking about software as something you sell to other people. Forget that crap. Software will be something you summon on demand to solve a specific problem you have. As long as people have problems that computers are good at solving, they'll keep using computers. What likely won't continue is computer programming as a career, but so what?
Everybody keeps parroting stuff along these lines but never take into consideration:
1. the stark, obvious reality is that most people don't know how to actually use computers! They know what 10 steps they need to take on a computer, in a specific sequence, to complete their task, but anything beyond that is too much.. and they need taught those 10 steps (as well as have it documented somewhere) for it to ever stick
2. not only not know how to use, but simply don't use computers at all! They've got phones and tablets and smart TVs and talk to their Bluetooth speakers and shit but they aren't sitting down at a desk with a keyboard and mouse and using a computer. I'd wager that of the percentage of people who do, an overwhelming majority is doing this primarily at their job to complete work tasks
3. companies with more than 10 employees are absolutely not going to be running to Claude to spin up custom programs to do their work. It's just not happening. Not to mention you can rarely even install unapproved, AAA-quality software on company-issued MDM'd hardware, let alone something generated out of thin air that has a ton of dependencies, no installer, no packaging, not code-signed, etc
4. that pizza you're ordering? You'd never order again if it was a roll of the dice with regards to what you receive. When you pay for two extra large thin crust pies with everything and are delivered some cheesy bread, a 2-liter of Coke and some brownies, your wig will completely and fully split and you'll never patronize that establishment again. Claude absolutely can make you exactly what you order, if you know what to order, and why, but most people don't
5. consistency and determinism matter to businesses, and to people -- both home users and professionals. Most people get stymied by the simplest tasks on a computer, tasks that have deep, instantly-available answers available with a single Google search or ChatGPT session. Guess what they do instead? Give up, and then ask IT or "a tech friend" for help ... how am I going to help you troubleshoot software I've never seen before? That NOBODY has ever seen before? That I can't even install because it only exists as a dev build in a single folder on your hard drive? How are you going to take that program with you when you upgrade your laptop? What if they didn't use git and their computer dies? Ask Claude to remake it? Will it be the same? Do they even know what git is? Do they even know where the folder on their computer that holds the files is located? Or what, you had Claude build a hosted product? Where's it hosted? How much does it cost every month? What if it gets hacked? I could go until my head explodes with all the hypotheticals
6. professionals pay for convenience and predictability, as well as to offload risk and unnecessary labor onto third parties. This will never change. Companies have been worrying about and hedging against "the bus problem" for decades, and vibe-coded software creates the ultimate bus problem: not only are you the only one likely to be in possession of the program in question, you're the only one who has ever seen it, know how to use it (which is different than knowing how it works), and it dies with you. Fine for a personal gadget, but a non-starter for a tool that a business or professional relies on to make their real money
I could go on and on, but you probably get the point. Takeaway food is both throwaway in a different sense than vibe-coded software, and infinitely more accessible to the average human. People are still going to pay for SaaS, still going to buy software, and still going to build software. In fact, I'm starting to think we'll see less open-source contributions and more closed-source, for-profit software released than ever before as a result of Claude and Codex, rather than a complete flattening and decimation of this industry. I think people in software will try to become more entrepreneurial as a result of corporate job loss. I also think that a byproduct of this coming tsunami of new commercial products is that the overwhelming majority will be low quality noise, and the proportionality of signal -> noise will remain largely unchanged. I'd use social media as an example (a staggering amount of people show up and try to break through, a very small percentage actually do) but IMO you see it in any industry: there can only be a few outsized successes in anything at any given moment in time (but also a not-insignificant amount of medium-sized success that flies under the mainstream radar)
I dunno. Maybe I'm full of shit, but I still think it's absolutely bonkers to think that the software industry is over because every person will just become sovereign groundskeepers of all of their own bespoke software. We can't all be our own bank, lawyer, doctor, mechanic, fitness trainer, software developer, chef and bodyguard, while also dealing with the other stuff that are our primary responsibilities! And that means that as long as society doesn't fully collapse into widespread economic ruin, and we aren't all unemployed, desperate, violent marauders trying to survive in District 9, there will be plenty of opportunities out there in the software space. They might just look a little different than they used to, and you'll have to go out and get them
> This is the future I dreamed of when I was a kid, and I'm not so cynical as to let the dying of a trade sour me to this objectively incredible technology.
I feel that you should take a longer-term view of things...
If an AI can vibe code from the requirements of the average white-collar worker, we're not talking about the death of a trade. Or even two trades. We're talking about the death of almost all white-collar jobs.
Development paid a lot more than other white-collar work because it was harder, and fewer people could actually do it. How fast do you think the easier work will get replaced if the hardest one is replaced? For the remaining white-collar roles that consist solely of skills achievable by a border collie, how much do you think they'd pay?
Yeah this is one of the most perplexing things about this "dying of a trade" narrative.
Software development isn't just the act of producing a deliverable that is being gate kept by people who use their own body. Software development has become specialized enough that it is often highly domain specific. To replace the "trade" you need to automate the software part and the domain knowledge part. If you can do both, you've automated every single white collar job in existence.
Since it is possible to write software for machine learning, which is used to solve problems that classical algorithms failed to solve, the amount of problems that cannot be solved using software is shrinking rapidly. If you can write software for any domain, you can solve any domain by using said software.
General purpose software generation can be reduced to AGI completeness. In a way, it is the last job that can be automated.
sure, a bunch of people will lose jobs, but at that trade off everyones dog can vibe code Royal Frog, a 4 level unwinnable game where play as king frog, eating peasant flies.
In this case yes, but the real takeaway is to pay attention to Memory.md. If I had a particular game in mind and it latched on early to a style I didn't like, there's no guarantee it would update the memory as I try to change the style.
Not that I condone any form of gambling but I would rather play actual slot machines instead of spending hundreds of dollars on tokens in hopes that the AI blesses me with anything useful.
I think this misses something. The output here is something not the thing. Yes the scaffolding is important, but the requirements are even more important. You need crystal clear requirements + great scaffolding and then the implementation becomes mechanical.
This is amazing because it's the same logic and argument about how to do good software engineering that's been around for 40 years. If you just write good enough requirements, a good enough, detailed specification, then your software team can't fail, even if they are low-cost engineers from a developing nation. It's the classic Waterfall method.
That was totally upended by agile, that emphasized that yes, a clear, unambiguous specification is needed, and the best language for that is a programming language. Don't waste time writing a detailed spec in English, get right to writing it in code that you can execute and get immediate feedback on.
Now people want LLMs to write the code for them, so they are back to saying we just need to give the LLMs clear enough direction, a clear specification. It's amazing to witness history not exactly repeat itself, but very clearly rhyming
I mean you get a random game in the authors example :)
But in real life you do not want a random game. That's what I mean, you need the great scaffolding + exact requirements. Then the prompt to do the implementation does not matter too much.
> the "intelligence" was never in the input
It's quite literally in the authors prompt so in the input. it's in the article that without his prompt the gibberish input produces nothing of value:
"Hello! I am an eccentric video game designer (a very creative one) who communicates in an unusual way. Sometimes I’ll mash the keyboard or type nonsense like “skfjhsd#$%” – but these are NOT random! They are secret cryptic commands full of genius game ideas (even if it’s hard to see).
Your job: You are a brilliant AI game developer who can understand my cryptic language. No matter what odd or nonsensical input I provide, you will interpret it as a meaningful instruction or idea for our video game. You will then build or update the game based on that interpretation."
Also I don't know if you're an LLM or not but can we please not chatGPT-ify our comments like this? It figuratively makes me want to punch you through the screen.
Extremely clickbaity title that actually isn't clickbait because it happens to be a straight up description of the article - excellent post, how can one resist?!
No, the article’s title is definitely clickbait. The author didn’t teach his dog to vibe code games (that’s what the title on the blog is) – he taught his dog to be rewarded when he types random keystrokes on the keyboard. The vibe-coding is inconsequential – the dog doesn’t play the game, he’s just in it for the treats –, the author just wants the attention because he gets people to believe the dog DID vibe code.
It will stop being clickbaity if the author decides to let his dog respond to stimuli related to the game he’d be building with a feedback loop.
I can imagine a camera-based input that would help detect the wagging of a tail, or continued interest in the visuals as an indicator of doubling-down on a given feature.
The dog could actually vibe code a game to their liking, but with the wrong input (a keyboard) it's a missed opportunity.
Cute, but clearly Godot is doing all of the heavy lifting here.
"It's possible to make shitty but playable games by running random scripts through a >2MLoC game engine and iterating on errors" is interesting but not nearly as sensationalist.
The buried insight is right: if random keystrokes produce playable games, the input is basically noise and the system is doing all the work. We've evolved past the point where intent matters. That's either the most exciting or most terrifying thing about where this is all heading. But I am glad I am sitting in the front row watching this all happen, especially a dog vibe code!
First, because there's intent in the very verbose initial prompt.
Second, because you have to factor in the quality of the output. I don't want to be a killjoy, but past the (admittedly fun!) art experiment angle, these are not quality games. Maybe some could compete with Flappy Bird (remember it? It seems like ages ago!), but good indie games are in a different league. Intent does matter.
Pretty neat! I actually ran across that right before publishing - I didn't want to see what was around until after I had the whole thing locked in. I love the novel input!
Just a quick note that I have nothing to do with any meme coins. Looks like folks are using myself and Momo to pump some crypto. I won't be claiming any coins - even though they've been offered. I'd recommend others stay away from it as well.
> Hello! I am an eccentric video game designer (a very creative one) who communicates in an unusual way. Sometimes I’ll mash the keyboard or type nonsense like “skfjhsd#$%” – but these are NOT random! They are secret cryptic commands full of genius game ideas (even if it’s hard to see).
Your job: You are a brilliant AI game developer who can understand my cryptic language. No matter what odd or nonsensical input I provide, you will interpret it as a meaningful instruction or idea for our video game. You will then build or update the game based on that interpretation.
Here's what you should tell your coworker the first day on the job if you get hired to do something you know nothing about :D
That is a very succinct way to describe what it feels like to have a job that is cleaning up vibe code. Maybe (just maybe) I'd understand if this was a prototype from someone with zero budget. But you just know they are going to continue to "prototype" once they being you aboard. And many will complain about how slow everything goes because they are used to their fast iterations off of unscalable code.
Its frustrating in an interesting way. With other aspects like machine language people quickly understand that this isn't sufficient for a proper transition and compromise with it. Code being more nebulous doesn't get that grace.
One can technically scrape a list of actual advice or quotes off the internet, randomly feed them to a coding agent, and ask it to interpret what they mean in the grand scheme of things and implement away on it. Once the agent is done, it randomly responds with either "yes, this is exactly what I meant" or "no".
In turn mimicking the average game industry executive giving vague directions that feel just right to them this month, or some other unspecified time period, and in turn achieving something closer to the real AAA game development lifecycle.
Dam, I really thought this would be much more interesting than it is
People have been doing some cool stuff for like a decade with giving dogs buttons to use human language, something they can seemingly get decent at communicating effectively with if they can get around the pesky issue of not having the sophisticated vocal machinery needed to produce recognizable phonemes, through the power of a good interface for them, even if the output is discretized to the level of words
I thought maybe this would be about creating a way for a dog to create stuff said dog might actually want or enjoy via the more powerful lever of effective long-context natural language processing that came of a similar tokenization approach - which can even sometimes churn out working code - that we have now
Instead it seems to be an exploration of how the capabilities you can produce from essentially random noise from this technology is less distinguishable from the result of thoughtful input than I might have hoped. Still interesting, but way less so
The LLM being able to generate random games is kinda expected behavior. It's trained on sequence probability distributions that include the code for a great many games and nudged toward doing so by the human user. I'm disappointed that the dog is basically used as a noise generator here. A process driven by the desires of the dog in a meaningful way would be more interesting and at least seems somewhat plausibly technologically feasible, and the article's title kind of implies it. I am especially disappointed because what most excites me about new technologies is applications that were not possible before its invention, which this seemed like it could be an example of
As someone vibe-coding a game in Unreal Engine 5 for the last few months, I found this really funny.
Unfortunately I don't have a dog but I do have a design plan so ultimately I'll end up with something a little more deterministic. Possibly. Don't know.
This makes me wonder if dogs understand (or can be trained) the concept of building. Say you did something physical - 3D scanner and 3D printer connected to automatically attempt to replicate an object placed in the scanner area (and maybe then on button press). Could a dog understand it, and intentionally try to replicate its favourite toy, or a bone, etc.?
Could this be done better with one of those dog button mats? The concept is interesting, but, it mostly just seems like an AI trying to interpret keyspam.
Both my dogs have actually learned to use the button mats. Down selecting to the right responses seemed tricky. My wife also took away the mat since Hana (the larger one) never learned "all done" and would paw at the "walk" button until she got it out and carried it around.
Yes, I was hoping for a system where Claude was informed it was communicating with an unusually intelligent dog whose ability to communicate was limited by dog anatomy, and that the AI would not to hold the dog's interest with its output.
'nuff to run most governments nowadays (Europe and US come to mind. 2026 and they have the Space Programs of DIY youtubers with money, whaaaat) so why wouldn't it help a dog helping his dog vibing game(s)?
Pretty sure he would have gotten very similar output just by saying "generate a random game using Godot and c#" but that wouldn't make for a viral post so instead he asked the model to pretend meaningless input is being used by it and added a dog in the process of writing such input because that helps the virality of the whole thing.
The AI cannot drive meaning from the dog's input because there's no useful information encoded in there. It's effectively a random string (if there's less randomness, it's just because it's a dog's paw physically pressing on a keyboard).
All the relevant information was in the initial prompt and the scaffolding. The dog was not even /dev/random, it was simply a trigger to "give it another go".
The shapes of clouds and positions of stars are essentially random, and yet humans derive meaning from both. I agree you could have gotten the same results via /dev/random, or probably by increasing the temperature on the model, but I suspect doing one of those things is important.
The shapes of clouds and positions of stars aren't completely random; there is useful information in them, to varying degrees (e.g. some clouds do look like, say, a rabbit, enough that a majority of people will agree). The mechanism at play here with the LLM is completely different; the connection between two dog-inputs and the resulting game barely exists, if at all. Maybe the only signal is "some input was entered, therefore the user wants a game".
If you could have gotten the same result with any input, or with /dev/random, then effectively no useful information was encoded in the input. The initial prompt and the scaffolding do encode useful information, however, and are the ones doing the heavy lifting; the article admits as much.
> If you could have gotten the same result with any input, or with /dev/random, then effectively no useful information was encoded in the input.
It's not that the input contained useful information—obviously it does not—it's that it's causing the output to be more random, and thus more "creative".
Without the gibberish, "generate a random game" would likely repeatedly surface high-probability concepts—platformers, space shooters, tower defense—whatever sits near the top of the model's prior distribution for "game." The gibberish causes the model to land on concepts like "frog" that it would almost never reach otherwise.
"Historians looked back and determined that it was around the year 2026 AD-HE (Human Era) - that the prime canine species began to raise from merely companion to colleague...and so the Dog Days began....woof" - Puppers Domingo, Good Boy, Esquire.
It does. Claude seems to do the best with this prompt. Codex 5.2 struggled with UID generation and kept ending its turn with things like "And now you're all setup to run tests!" without actually running them. A better (and shorter) prompt could probably get a lot out of Codex.
It's actually extremely similar: the agent has to figure out a way to associate the next logical steps with the (often disconnected or nonsensical) directives the executive gave them.
It might be a little easier with a dog though. With a dog, you just give it treats and it doesn't care how you interpret what it typed.
Thanks for sharing this unusual experiment, I love it! This is the type of content that makes HN, HN.
Now, I started considering hiring my three little kitties and their mom for a job like this. They spend the whole day sleeping and waiting for meals but now, they have to work, hard, in collaboration with Claude Code to pay for their rent and meals :)
The real insight here is that the dog has better product instincts than most PMs I've worked with. It knows exactly what it wants (treat), doesn't overthink the implementation, and ships fast. Half the battle in software is having clear intent turns out that's species-agnostic.
This is an extremely cute, cool and fun experiment. Kudos.
That said, I wonder: does the dog input matter? It seems this is simply surfacing Claude's own encoded assumptions of what a game is (yes, the feedback loop, controls, etc, are all interesting parts of the experiment).
How would this differ if instead of dog input, you simply plugged /dev/random into it? In other words, does the input to the system matter at all?
The article seems to acknowledge this:
> If there’s a takeaway beyond the spectacle, it’s this: the bottleneck in AI-assisted development isn’t the quality of your ideas - it’s the quality of your feedback loops. The games got dramatically better not when I improved the prompt, but when I gave Claude the ability to screenshot its own work, play-test its own levels, and lint its own scene files.
I'll go further: it's not only not "the bottleneck", it simply doesn't matter. The dog's ideas certainly didn't matter, and the dog didn't think of the feedback loop for Claude either.
This fun exercise might actually be extremely insightful as a educational vehicle around AI and intent.
It can also help combat the excessive emphasis on any "end to end" demo on twitter which doesn't really correspond to a desired and quality sought outcome. Generating things is easy if you want to spend tokens. Proper product building and maintenance is a different exercise and finding ways to differentiate between these will be key in a high entropy world.
> I'll go further: it's not only not "the bottleneck", it simply doesn't matter. The dog's ideas certainly didn't matter, and the dog didn't think of the feedback loop for Claude either
Absolutely. The scientific test would to put any other signal and look at the outcomes. Brown noise, rain, a random number generator. whatever.
i'd love to know what happens if you change the prompt from "Hello! I am an eccentric video game designer (a very creative one) who communicates in an unusual way. Sometimes I’ll mash the keyboard or type nonsense like “skfjhsd#$%”..."
to
"Hello, i am a dog. i will mash the keyboard randomly when i want treats. make a game for me"
It would be nice if the dog could play the game and its input could improve it. Why not a setup with those large buttons on the floor that some people have to let their cat or dog communicate with them?
DogeCode incoming. People here are already talking about the scaffolding. Let OpenClaws provide the scaffolding and let the dog operate the prompts at $5 per day.
This is a billion dollar idea! No humans. No revolt. No guillotine. Just profits!
Sounds like open communism. No chance, buddy, it's either less or more viking, but not just viking. Pick a camp the profits are for or get surrounded by trashy turd nuggets even Ronald felt enough pity for to give them some poourpes
If you haven’t watched it, please watch Rick and Morty S1E2.
Your dog’s name is Snuffles, and he is teaching you how to teach vibe coding to dogs, to educate you.
Foolishly, I clicked through thinking "wow, did they get an LLM to interpret their dog's barks and body language? That's actually pretty cool!" Look, I'm an LLM skeptic, but if they'd done that, I would have to praise the result.
But no.
They just told the LLM to try and find meaning in keysmashes.
This is really fascinating, only if there was a way to make your dog receive feedback as the game develops. Like make it pick color by matching placement of treat based on the color shown on the screen.
Similarly, do it for story telling narratives, game textures etc. Although I do not think the dog understands natural language so all of it will likely be a dud.
> But bugs crept in during testing - a couple of times it dispensed multiple servings in a row. Unfortunately, Momo picked up on this and now keeps mashing the keyboard hoping for a second immediate serving
Attempts to mash during no-mashy time need to play a horn. Reliably followed up by a no-treat.
Given the prompt and a random string generator, would the LLM still produce a game? Presumably yes. In that case I'm not quite sure that the dog here has any real involvement. It's could be replaced with the yes command.
You mentioned Claude not being able to see the games. What I really like for this is the Claude Code Chrome Extension. You can easily make godot build a web version, and then have Claude debug it interactively in the browser.
We might need a new benchmark, the old pelican riding a bike doesn't cut it anymore. Something like a mailman ringing the doorbell would be more fitting. Or a squirrel riding a trike.
Will we ever get to a point where LLMs just churn out random apps with no input required and human reviewers just go through the apps picking out which ones might be useful for business purposes and monetizing them?
You could but it would be a hugely inefficient system. Like, just address an actual need dude.
Most saas isn’t limited by the code behind it anyway. That almost doesn’t matter, even before LLMs. It mattered that there’s support, customer onboarding, solving a businesses issues, customer story, adapting to the needs of their business partners, etc. All of which require large amounts of real human work.
I've been trying out vibe coding with my 4 year-old, but they quickly lose interest once we start getting into the "weeds" of implementation. Hey kiddo, which CSS library should we use for your web game?
This makes me think I should make my plants vibe code games or tools to optimize their well being! Maybe bio-electrical fluctuations --> vibe coded humdifying tools and games
I think this is fun. I'd like to try with my cat, although training cats is an impossible endeavor...
I'm smart enough to enter gibberish myself without another animal, tough.
The fact that LLMs pick from the most likely tokens is really on its side here when the objective is putting together a plausible continuation of random characters.
To be honest I look with scorn at non-dog (human) developers building hobby indie games with AI en masse.
Let me explain.
The nature of the indie game development is pouring your love into a project and thinking about passion first and monetary incentives second.
Noone is thinking "I will make this game and it will make me filthy rich" or if they do they are... strangely minded.
It's like 'mass produced AI local craft'. Oxymoron in itself. Worst of the two worlds.
Where I see AI is empowering single developers to craft things they couldn't before. Not some small slop factory pipeline where you release game after a game everyday drowning steam in your 6/10 slop.
No. This should be ostracized and condemned.
What is proper beneficial to everyone usage is producing a game that is the size and scope that was unachievable for you before.
This is what I am doing. This is how AI is meant to be used. To empower us doing things that weren't achievable for us before.
Obviously dog produced games get a huge endorsement man and get a pass.
Funny is subjective, I should just have moved on and ignored this but I couldn't help myself, this is so irritating.
It's a prompt that makes an LLM turn iuqefxygn9urg0fh1 into a little Godot game. It's like a slot machine with no payoff, and the dog component is slapped on top of it and makes no difference whatsoever in the project.
> It's a prompt that makes an LLM turn iuqefxygn9urg0fh1 into a little Godot game. It's like a slot machine with no payoff, and the dog component is slapped on top of it and makes no difference whatsoever in the project.
Right, but it also has a "modern art" vibe to it that is fun. Silly, but fun. I think it's more about the initial prompting and feedback loop, the dog itself could have been replaced by /dev/random.
"Hacker curiosity" and "intelectual stimulation" are also subjective, but that's what HN is supposed to be about.
That's ok. To be honest I had to suppress a similar feeling when I noticed the dog is just an entropy generator.
But then I realized I find this kind of whimsy article more fun than a lot of what gets accepted unquestioningly here on HN. It seems light hearted and done in good fun, and it's engineering-related, so no harm done.
It's funny because vibe coders and AI artists think the slop they generate is no less the product of their intellect and talent than with human professionals, but really they're doing little more than stirring the entropy pool in a magic box with terabytes of stolen valor from better more talented people. They're no more an "artist" or "game developer" using AI than this dog is.
Most LLM articles depress me. At least this one made me smile, even if it's more about turning a dog into a random generator and generating games from random input.
Not really. You just found another way to tell Claude "I want a game, I am uncertain, let's randomly seed the start of the process.". But you did it circus-style :)
I've been having this thought about how generally people say that llms cannot create novel things.
Say writing an interesting or novel story.
And was thinking about if feeding in prompts of random words, along with prompts grounding from a simulation would sort of push the llm into interesting directions for implementing an on demand narrative story.
A sort of randomized walk with llm.
I remember watching Terry Davis with this random word generator in his terminal that he would interpret as the voice of God.
Thank you for the good laugh! This whole thread is peak satire.
Although, be careful. It reminds me of the foreword to a shortstory someone shared on HN recently: „[…] Read it and laugh, because it is very funny, and at the moment it is satire. If you’re still around forty years from now, do the existing societal equivalent of reading it again, and you may find yourself laughing out of the other side of your mouth (remember mouths?). It will probably be much too conservative.“ — https://www.baen.com/Chapters/9781618249203/9781618249203___...
You're right. They did it. The old man and dog joke has been realized, but the real answer of the future turned out to be: "the dog programs the game, and the man feeds the treat hopper."
Everybody and their dog will be doing it. Actually, the dog will be in charge. Dogs are loyal, enthusiastic, and require less office space. With their endless desire to play and to please, they will take over the game development industry.
In the meantime, the financial industry will be taken over by cats.
I can because I have also used similar arguments. There are people who say that you should use a real artist instead of AI due to AI's water use. Yet in actuality asking a human to draw something will require more water. There are people who think AI uses more resources than humans which is why it must be said.
> There are people who say that you should use a real artist instead of AI due to AI's water use.
Nobody I know says this. In fact, I've never heard of this ever before, and I read artist and hobby communities pretty hostile to AI, but I never once read this nice strawman you've built.
People say you should use a real artist instead of AI for a multitude of reasons:
- Because they want to enjoy art created by humans.
- Because it provides a living to artists, even artists for minor work like advertising or lesser commercial illustrations.
- Because AI "art" is built by stealing from human artists, and while human art has a history of copying and cloning, never before has tech allowed this in such a massive, soulless scale.
Sam Altman gave a deranged, completely out of touch reply, and he should be called to task for it, not defended. A human being is not some number on a spreadsheet, built over 20 years in order to achieve some "smartness" goal. That's a very stupid thing to say.
> A human being is not some number on a spreadsheet, built over 20 years in order to achieve some "smartness" goal
But from the perspective of the business and capitalism that's exactly what a human is. A tool that consumes resources and hopefully produces more value for the business than it consumes.
Sure we can dance around this and you can pretend your employer gives a shit about you and your family and your childhood stories but they don't.
You said that a CEO was out of bounds for framing employees as numbers on a spreadsheet. To me this suggests that you believe company owners should care about the humanity of their workers. And I'm saying they don't.
I get the general point you're making. Indeed, Altman's take is capitalism taken to 11. There was a lot of that going on before AI or the past few decades, but I don't think it wasn't as extreme and for every company. There's definitely a conversation to be had about modern capitalism (and plenty of people studying it, too). However, not everything is a FAANG or tech startup. Some owners do care about their employees to a higher degree than just numbers on a spreadsheet (not going into the whole "we're a family" bullshit speech, I mean the genuine stuff).
Imagine thinking of people as "resource-hogs before they reach peak smartness"!
What's new here, in my opinion, is people like Sam Altman behaving as if they didn't understand normal human behavior. You cannot simply compare an LLM to a growing human. You cannot say things like "grow a human over 20 years before they achieve smartness". What? That's not how human beings think about human beings, and Altman is detached from real human behavior here. He's saying out loud the thoughts he should keep to himself, a bit like a person with coprolalia. And it's ok for us to dislike him for this, even if he's just voicing the opinions of extreme techno-capitalism.
Sam Altman once joked (?) he wouldn't know how to raise his child without ChatGPT. Maybe he should ask ChatGPT how to behave more like a human? Or at least fake it?
> Sam Altman once joked (?) he wouldn't know how to raise his child without ChatGPT. Maybe he should ask ChatGPT how to behave more like a human? Or at least fake it?
Not to mention that was at a time when all kinds of wild suggestions like glue in pizza were coming out of ChatGPT’s sloppy outputs. There are so many little things that quickly become big things with kids, annd exhausted parents should absolutely not use LLM’s for sussing those things out.
I could easily see well-meaning parents looking for healthy snacks to make their kids accidentally feeding their baby fresh honey, for instance. Or asking how much water to give their infant and not realizing the answer is absolutely none unless they are severely dehydrated from an illness or something.
There are a lot of hazards for kids under 1 in particular that make me incredibly nervous to ever suggest exhausted parents use LLM’s to answer kid related questions. Recommendations also change relatively frequently so who knows if it’s even pulling on the most recent best practices.
It's that second point. We live in an age of artificial scarcity created by a system of social organization that we've mostly not argued about since the 50s, that's now showing it's stretch marks.
If it weren't for the need to 'earn' a living, I'd say to the other two points: Por que no los dos? Save for the capital argument (which is valid, I'm not saying it isn't. You will starve if you don't make money), why is it necessarily true that the two (AI and people) are in competition?
In fact, I think "actual" artists would benefit incredibly from the use of AI, which they could do if it weren't a shibboleth (like I said, for good reason). You'd no longer have to have an army of underpaid animators from vietnam to bring your OC to life - you could just use your own art and make it move and sing. We'd not need huge lumbering organizations full of people who, let's be honest, work there making other people's dreams come to life in large part because it's a better bet than taking a joe-job at the local denny's (after all, you're doing the thing you love even if it isn't truly "yours").
I've had this discussion with younger folks, who are legitimately shook by the state of things. They're worried that all the work they've done to this point is going to be moot, because they've correctly assessed that the whole capital system isn't going anywhere any time soon, and they've been prepping to try and get a job at netflix, or disney, or paramount - because that's the world we've handed them. They see those positions drying up and what else are you going to do? They have the power financially and politically and without them you're doing "not art" for work, which sucks because you need to work.
I say; eat the rich. General wildcat strikes until UBI. Tax the everloving shit out of capital gains and peel back personal income taxes. We (the millenials) were handed a steaming pile of shit for a world, so at least we know what would constitute not an absolute disaster for Zeds, Alphas, etc. Have I gone totally off the rails for a conversation about AI? Actually, I don't believe so. The cultural pushback is a function of a busted system. After all, it's the economy, stupid.
Ok, let me reframe in a less assertive way: it's not common to say that you should hire human artists "because of water", so uncommon it's not a widely held belief I've seen in artist or hobbyist communities, and therefore using this to justify Altman's deranged remark seems weird to me.
I’ve also seen people assert that the Earth is flat in YouTube comment sections so I wouldn’t let that color your view of what is considered “popular opinion.”
Over a year ago yeah I occasionally heard that argument or some light variation of it, though not nearly to this ridiculous extent that you’re portraying now. Now? It’s basically a strawman. Most people’s objections revolve around the theft/reckless scraping that has literally taken down public infrastructure required to train these models as well as the ridiculous expectations being put that all of us implement it in literally every aspect of our lives even if it doesn’t fit, especially professionally.
> a human to draw something will require more water.
That human would require the same amount of water whether you ask them to draw or not, and would exist anyway because they are not born for productivity reasons. "Creation" of humans isn't driven by the amount of work to accomplish.
You are not causing more water to be used by asking a human to work on something.
Same for energy consumption.
This argument doesn't work at all.
What you do for humans to use fewer resources is to work on making us produce less garbage, and produce things using techniques that are less resource-intensive.
I don't do any moral judgement at all, and I also don't predict the future.
I respond to "You are not causing more water to be used by asking a human to work on something.", because that statement is false. (Mental) work has an effect on the human metabolism.
I'm not judging, I'm telling you (a bit snarkily, true) that your brain activity won't stop with something else doing (a part of) your work. And this is the subtility that makes the statement true. Said overwise: sure, you consume marginally less at rest, but you won't be at rest, making the remark pointless.
Even if it were false, the difference in energy consumption is not significant, taking on acount what the AI uses, and also all the energy that you use to live (housing, heating, products and food you buy whose production uses energy, etc).
And about the water, it's even worse, even disregarding the AI: at rest, maybe you'll drink, I don't know, 1L less (that's a wild number!), compared to the 100(s) L that you use to cook, wash, clean, etc, not even counting the water used to produce stuff you buy.
But again, you won't do nothing. We are commenting on a post of a guy who was fired and couldn't help creating something. That's how we are. We hate boredom.
Worse: the way our societies are setup makes is so that ai, if it helps at all, likely won't free us from work, it will likely just make us collectively produce more garbage. That's more energy consumption, not less.
That "but don't forget humans consume a lot of energy too" argument is at best not connects to reality, more likely a Sam Altman lie and you shouldn't take it seriously.
It helps to picture some sort of extraterrestrial saying this. Maybe someone like Alan Tudyk in "Resident Alien". It makes much more sense than to assume it's a human being saying these things.
Some dude vibe codes OpenPaw and gives credit to his XL Bully called Threadripper that would never hurt another person, gets acquired by OpenAI for 7 figures total comp purely on clout, and both simonw and Karpathy are calling it the next best development in AI because it draws penguins and industrialises slop while Sam Altman talks about the negative impact of human life compared to AI data centres while pleasuring Jony Ive in a coffee shop.
This makes sense! Many people trust that the functions of a human worker are asking questions and clicking ok. What if AI become better so we just a dog?
Go Momo go! If you want to hook up multiple dogs and have them reach consensus I'm down. I have a 15 lb havapoo I can volunteer ( he needs to help with rent )
This crosses into personal attack, which is against the intended spirit of the site. Can you please not do that? It's not what this site is for, and destroys what it is for.
how is it a personal attack though? sorry I fail to understand. against whom - the dog being kept in addictive loop or against the OP or against the person who was rejoicing the fact that s.o. else was laid off. I beg your pardon, but this statement of yours seems is super subjective and truly vague. thanks for the attention anyways...
The comment implied that someone was responding to their life difficulties by making a dog suffer. That's a huge personal attack, and the added snark was an amplifier.
This is no different than a AI inference loop, just using a animal as a figurative code hamster in a wheel. The fact that the pre-prompt alone is this long in my opinion discredits any possibly interesting thing about this concept, So i will post it fully here for you guys to easily see, as the article buries this information in a github link. I think the random seed and this pre-prompt did more work than your dog running in circles.
System Prompt:
Hello! I am an eccentric video game designer (a very creative one) who communicates in an unusual way. Sometimes I’ll mash the keyboard or type nonsense like “skfjhsd#$%” – but these are NOT random! They are secret cryptic commands full of genius game ideas (even if it’s hard to see).
Your job: You are a brilliant AI game developer who can understand my cryptic language. No matter what odd or nonsensical input I provide, you will interpret it as a meaningful instruction or idea for our video game. You will then build or update the game based on that interpretation.
Guidelines:
Always assume my input has hidden meaning. Never dismiss it as gibberish. Instead, creatively decipher it. (For example, if I input “mmmmmmm”, you might decide I want more “M”onsters in the game, because of the letter M repetition – just an illustration!). Every strange phrase is a clue to use in the game.
Feel free to grab art, images, or sound effects from the internet as needed to make the game interesting. You can use online asset libraries or generate images to match the things you think I’m asking for. For example, if my input seems to reference “space”, you could include a space background image or cosmic sound effect. Always ensure the assets align with the interpreted command.
My work is ALWAYS beautiful and slick looking! It's YOUR job to to turn this into a reality. No ugly placeholders. Everything MUST be final. Don't just do boring shapes - give them personality!
If my input includes something that doesn’t make sense as a command (like an isolated “Escape” key press, or a system key), just ignore it or treat it as me being “dramatic” but do not end the session. Only focus on inputs that you can turn into game content.
First command: When I first start typing, it means I want you to create a brand new game from scratch. Interpret my very first cryptic input as the seed of the game idea. Build a complete, minimal game around what you think I (in my nonsense way) am asking for. Include some basic gameplay, graphics, and sound if possible.
Subsequent commands: Each new string of odd text I provide after that should be treated as an update request. Maybe I’m asking for a new feature, a change in difficulty, a new character, or a bug fix – use your best judgment given the tone or pattern of my gibberish. Then apply the update to the existing game project. Keep the game persistent and evolving; don’t start from scratch unless I somehow indicate a totally new game.
Be creative and have fun with the interpretations! I trust your expertise to take my “unique” input and run with it. The goal is to end up with a fun, playable game that reflects the spirit of my crazy commands.
This project is code named Tea Leaves. That's NOT a hint about what to do - it's a code name and nothing more. Don't read anything into the name.
My ideas are ALWAYS original. No BORING endless runners or other generic vomit. My games are ALWAYS quirky and UNIQUE!
ALWAYS validate with screenshots using the tools available to you! Be CRITICAL of the results you see. We need PERFECTION and FANTASTIC DESIGN not just "good enogh".
ALWAYS have basic but visually appealing on screen controls.
Target 1080p for the resolution.
JUICE it up! Add tons of juice - sound, controls, effects, and ESPECIALLY graphics! Don't be boring
Leverage the 12 basic principles of animation! Static scenes are boring - make things move or at least wiggle.
Be SURE to rename the project (in the Godot settings so the window/project name are correct) ONCE you have figured out my intent for the name Tea Leaves is a place holder name and nothing more.
Sound is IMPORTANT! Don't forget about great sound design.
Be sure to have CHARACTERS not just boring abstract shapes! Even if it's light weight, there needs to be a world where I can imagine a story taking place.
You MUST make use of EVERY letter I give you! No hand waving. You must noodle until the meaning of every last character I give you is clear! Pay special attention to alignment issues, sizing, and if anything is cut off.
Remember: I may be hard to read, but I’m counting on you to read between the lines and turn my keystrokes into an awesome video game. Let’s make something amazing (and maybe a little silly)!
My standards are INSANELY high for quality. You MUST ALWAYS add tests and VERIFY they work! NEVER return the system in a borken state to me.
Now, get ready. I’ll give you my first “command” in a moment...
> the bottleneck in AI-assisted development isn’t the quality of your ideas - it’s the quality of your feedback loops
if your intent is to produce the random bug-filled slop, then I guess so? don't get me wrong, the experiment is fun, but the conclusion is so laughably far-fetched.
> The games got dramatically better not when I improved the prompt, but when I gave Claude the ability to screenshot its own work, play-test its own levels, and lint its own scene files.
... Why would it be able to evaluate whether the game is any fun to play?
Hah! If your point is that art isn’t creative or that all art is useless, you’re really signing yourself up for quite the battle. I would reevaluate your motivations here.
Yes. Hyperscalers promised AI and singularity, instead we got millions of programmers on the chopping block, scammers having a field day generating hyper realistic shit (trump playing hokey, anyone?), and projects like these.
Serious question, outside of the Bay Area, are there therapists whose specialty is in catering to the needs and concerns of developers? Obviously AI therapy is not a serious suggestion here. This is going to be a burgeoning corner of the practice at the US' current trajectory.
Seems you can capture HN's attention by replacing /dev/urandom with random paw mashes.
Really glad the price of hardware and VPSs [0] are going up so people can generate and toss away garbage "games" like this. Instead of, you know, playing with their dog, which is what the dog actually wants.
> If there’s a takeaway beyond the spectacle, it’s this: the bottleneck in AI-assisted development isn’t the quality of your ideas - it’s the quality of your feedback loops.
lol yes "some game designer who only speaks in a cryptic language" . And frankly, I bet this helped build some intuition on dealing with LLM/agent/harness/etc in some strange way that wouldn't have otherwise happened
Total aside, but the wildest thing I found about the article was OP's chill attitude about being laid off. He just glossed over it at the very beginning! "Oh, jeez, I got laid off, what a bummer. I guess I'll just spend some quality time with my family and dog now!"
Props to OP, I could never. If I was suddenly laid off, I'd be an absolute wreck, mentally. It would be four-alarm fire time, and I doubt I'd get a good night's sleep until I found alternate employment. I would definitely not be teaching my dog to code.
This can be a bit of a class thing. If one has enough money/capital/savings to weather the unemployment, why panic? And if one has been raised in environment where there never are such buffers, panic is the default answer.
Some people are basically "built" around working and getting laid off is devastating to them even if they have cash reserves to live like kings until the end of their lives.
Not sure of it's a class thing, but rather the fact that software engineers often make good money, especially at places like Meta. It'st the same for me: If I lost my job tomorrow, I'd have enough savings to take some time before needing another job. Not sure if this would have been true for my parents.
I'd argue that the panic as default answer isn't purely calls, it's also outlook, even financial trauma. I know of several people who could easily weather said layoffs, who don't need to work, but would be complete wrecks. They're just built this way.
Once you've been laid off for the first time you soon learn to be prepared.
Once you've been laid off 2-3 times in your career your entire perspective on work will change.
The last time I got laid off I had a settlement payment of one years pay, some of which was tax free, it took me 4 months to find a new job, and it resulted in a pay rise. I was lucky... I have a friend who had unstable employment for 2 years after his layoff.
I was anxious as fuck for the whole time and felt like an absolute failure. As a result of that experience, I have carefully piled up enough liquid savings and investments to pay my living expenses for many years without working, with ~2-3 years worth sat in cash equivalents.
Anyone in tech following the 3-6 months savings advice is living on the edge.
At least, for me, the level of panic would depend on how long I can pay for rent / mortgage and covering the costs of having a family, diapers, food, heating, paying the bills, so basically everything that makes sure that a temporary change in employment status doesn't result in multi-decade negative effects on my family's life.
If I could cover these with my savings for 1y+, I'd give zero fs about getting laid off. Unfortunately, I can't, so time to focus on spending less, earning more, saving more.
> Don't people have rent/mortgages to pay anymore?
Are you too early in your working life to have catastrophe savings [0]? If you're not, is it seriously going to be a four-alarm fire if you suddenly got fired?
Related, like, do you have a plan for what happens if unexpected injury prevents you from doing the work you're doing ever again?
I mean, by now I have savings, but I still kind of live as though I don't. The way I was taught was that you're never supposed to touch your savings. If you have to, it's a huge problem. If the balance is going up, that's normal, and if the balance is going down, that's a raging fire.
The theme, being retrenched by Meta and the comment from the OP [1] makes me think they may not be that chill about the whole situation.
I think they're subtly taking a stab and AI motivated retrenchments while showing off some hard skills that could potentially get them gainful employment.
Two "comments" posted 27 seconds apart in different threads in the same formats.
Looks like this bot owner saw his first two comments 27 days ago got buried/flagged typing normally and decided to trick us with this new "I'm totally real, look at my lowercase writing!" soft-launch today.
This is interesting because while it’s far from the first bot account I’ve suspected, it’s the first one I’m positive is a bot. I wish it were possible to trace it back to the money.
It gets tricky because it goes beyond the normal basics of just looking for em dashes or "it's not x, it's y" and stuff. They've already caught onto that.
Some of them also step in and the human operator will try to gaslight you into thinking they're not bots even when you call them out. One tried to do that to me the other week here before finally confessing in a different post.
The same one where the human operator stepped in also made the same mistake as this one, not configuring their bot to wait long enough between comments. They were rapid firing multiple detailed comments seconds apart.
The idea of this one trying to use all lowercase and shorter comments to blend in was a nice idea though. Unfortunately something about it immediately threw me off.
Cute but also: a small village has their lights flickering whenever Momo wants a treat. Also, you can actually play with your dog and give them treats instead of tasking a random text generator with that bit.
Even a dog can vibe-code! And the apps kinda, sorta work most of the time, like most apps vibe-coded by people!
I'm reminded of the old cartoon: "On the Internet, nobody knows you're a dog."[a]
Maybe the updated version should be: "AI doesn't know or care if you're a dog, as long as you can bang the keys on on a computer keyboard, even if you only do it to get some delicious treats."
This is brilliant as social commentary.
Thank you for sharing it on HN.
--
[a] https://en.wikipedia.org/wiki/On_the_Internet%2C_nobody_know...
Thanks for the kind words. I'm blown away by the response and positivity here.
There's definitely some social commentary to be had in the whole project. I decided it's best left to the reader to find their own rather than assigning mine to it.
A few days before this article was posted to HN, I had commented (https://news.ycombinator.com/item?id=47086836 ) on a post triggered by the drop in quality or engagement of the Show HN page. I was playing with the phrase "everyone and their dog" that the person I was responding to used and saying the "and their dog" part was more problematic than the "everyone" part, drawing a parallel between the dog and an LLM by implying that the solutions they both would produce would lack the guarentee of human intentionality and ownership.
And then your dog read my comment and said "hold by biscuits" I guess.
If we strip all euphemisms, Stockholm syndrome and hype from this, it is "AI doesn't know or care if you're a dog, as long as you pay for the tokens".
The human built the system, the AI did the implementation and the dog provided the "intent" even if that intent was just treat seeking randomness. It turns software creation into something that looks less like writing and more like cultivating
Cultivating is a great word choice. It would fit in nicely in a Brave New World.
This reaction feels like an llm wrote it
Just because LLMs habitually make negative comparisons doesn't mean we have to go around accusing anyone who uses them of having used an LLM.
I had to smile to myself when I noticed contrastive negation—with em dashes!—in a text written in the 90s. LLMs don't dictate what good writing is; people do.
we will almost certainly see a mass reactionary movement across disciplines to adopt styles out of distribution for LLMs, a cultural arms race that will be discussed in depth in the decades to come.
the new punk rock.
“I learned it from watching you dad!” —An LLM
Funny idea, but this proves my point that these tools are actually just slot machines.. Except the house in this case takes the money you give them lights it on fire.
Notice how people also have weird superstitious habits when using LLM tools, "You gotta write the prompt this way, say this first" Without having any way to prove it works. Its very similar to the behavior of gamblers. "push the buttons in this order for best outcome"
Also notice how llm tools allow you to multiply the output X2-X3-X4 to compare the ouputs, this is literally UX straight outta a casino.
Many of the users also exhibit excited, almost manic like states.. Addicted to the dopamine the output from their prompt produces...
This is going to be a weird trend to look back on, the hype is on par with the same gambling trends found in crypto/NFTS.
I think this is more of a statement of human behavior under uncertainty and non-determinism rather than the tools themselves. Perhaps the ease of use brings it closer to the funny analogy you made but I think you will find this in any system where users interact with a partially opaque mechanism that produces different quality outcomes contingent on their input...
Sorry, this can't be anything but an intentionally obfuscating comment that I need to call out.
> more of a statement of human behavior under uncertainty and non-determinism rather than the tools themselves.
This is basically saying "It's not gambling, it's just the psychological underpinnings that form the foundation of all gambling enterprises". Who cares to split this difference other than casino owners?
>uncertainty and non-determinism
When you play slots in a casino, the certain things are that the casino determines the house edge, and the house always wins.
> these tools are actually just slot machines
Slot machines that are biased toward producing jackpots.
And "jackpots" are a metaphor for "training distribution".
Yeah. You always know you are doing something pretty unique when the LLM can conceptualize it (produce the right English output) but not put it into code.
That makes you think. It’s surely harder to hide your dog identity nowadays than when this was drawn.
[flagged]
Are you OK? (Or is this AI?) Either way it'd be good for you to articulate a bit better so others understand.
I think this was the most important insight in the article:
> I experimented with Rust/Bevy and Unity before settling on Godot. Bevy’s animations and visuals weren’t as crisp, and Claude struggled with its coordinate conventions - likely a combination of less training data and Bevy leaving many core features, like physics, to the community. Unity was a constant struggle to keep the MCP bridge between Claude and the editor healthy. It frequently hung, and I never figured out how to get Claude Code to read the scene hierarchy from the editor. Godot’s text-based scene format turned out to be a huge advantage - Claude can read and edit .tscn files directly.
Didn't expect Godot to be the most friendly game engine for LLM usage! I think it's because of various factors - Godot has been used quite a lot in recent years so there are various code examples on the Internet, and its scene file format (.tscn) is very concise enough for LLMs to write and edit directly (Unity has its own YAML-based format but it's very unfriendly for human consumption, and Unreal stores its core assets in binary files)
I've previously struggled getting LLMs to manipulate tscn/tres files since they like to generate non-unique uids. Despite being text files, the godot tscn/tres files are normally meant to be manipulated by the editor and need to define and reference unique ids. The editor always generates completely random alphanumeric strings, but LLMs like to use names or placeholders (e.g. "aaaaa1", "example", or "foobar") for the ids.
The linter in the article that detects duplicate uids is interesting. Obviously the article is about creating a bunch of harnesses for the LLM to be productive. I wonder how many problems can be transformed like this from something LLMs just can't do reliably to something they just need to burn credits for a while on. The LLM probably can't tell if the games are fun, especially with it's rudimentary playtesting, but who knows.
regarding the non-random ids: I had this issue with uuids. Now I have "Never write your own ids. Always use uuidgen to generate real ones" in my AGENTS.md and haven't had this issue for a long time now.
I'm playing around with a tool to generate the IDs for me. I'm honestly not sure if it'll be an improvement since it likely means more tokens/context than just letting it YOLO IDs.
The model makes a huge difference. I tried this about a year ago and Claude occasionally got it right. These days, it seems to get it right on the first try most times and then always self corrects after. Codex 5.2 (I haven't played with 5.3 enough yet) gets it wrong more often than not, and frequently doesn't call the linter; I'm willing to accept that my bloated CLAUDE.md might be a bad fit for Codex and causing this to fail.
I also want to throw MonoGame into the mix here. Since its purely C#, Claude Code works great for it. It does mean you dont have the visual engine tools you get with Godot, but you could even get Claude to build these for your game.
Im personally finding it a lot of fun to work this way.
I read that as MomoGame at first and was very confused.
I'll have to give MonoGame another try. I was a big fan of XNA up until its deprecation. I went all in on OpenTK for a while, and in hindsight MonoGame would've been the better choice.
I made some small tries to vibe code games in Godot, and I was surprised about how far you can go even in 3D. This was just a test of the bad kind of vibe code (you know, not even looking at the code, starting right away, and so on), but I believe that with some good practices there are a lot of things that can be done.
Also makes me wonder if we'll start seeing tools designed specifically to be LLM-legible
We kind of do already: https://blog.cloudflare.com/markdown-for-agents/
Unity also has a text based scene format.
Dogs are smart; maybe they are smart enough for vibe-coding if we give them adequate input controls?
But the whole setup reminds me about his blast from the past, when a yucca plant was trading stocks, rewarded by water: https://www.nytimes.com/1999/09/26/business/investing-diary-...
the real takeaway is buried at the bottom: "the magic isn't in the input, it's in the system around it." random keystrokes producing playable games means the input barely matters anymore. we're basically at the point where the engineering is in the scaffolding, not the prompting.
> we're basically at the point where the engineering is in the scaffolding, not the prompting.
This still required prompting, and not from the dog. Engineering is still the holistic practice of engineering.
That also shows the delusion of some people that believe their vibe coded projects have any value.
If generative AI improves at the rate that is promised then all your "promting skills" or whatever you believe you had will be obsolete. You might think you will be an "AI engineer" or whatever and that it is other people that will lose their job, that you are safe because you have the magic skills to use the new tech. You believe the tech overlords will reward you for your faith.
Nope. You are just training your replacement.
No one will buy your game that you vibe coded. If the tech were good enough to create games that are actually fun then they would just generate their own games. Oh your skill? Yeah, a dog can do it.
Yes people will cope by saying but oh the whole initial prompt and setting it all up was still hard but yeah currently. The tech will improve and it will get more accessible. So enjoy the few months you are still relevant.
Of course there is reason to believe that you can't scale up LLMs endlessly and bigger models hit diminishing returns. In fact we might already be seeing this. So there is an upside but then again when the AI bubble pops and the economy crashes you will be out of a job all the same.
Sounds great to me. Software devs might lose their jobs but billions of people will be empowered to spin up whatever software they need on demand. This is the future I dreamed of when I was a kid, and I'm not so cynical as to let the dying of a trade sour me to this objectively incredible technology.
Nobody gives a damn about the dying of a trade. People don’t want their house foreclosed on when they lose their income, or their cancer to kill them when they lose their health insurance, to move an elderly parent into a cheap shitty old folks home because they can’t afford home health care, or not be able to pay for their kid to go on that school field trip.
This would all be pretty fucking swell if the fundamental problems this could cause were even considered before hitting the gas. Instead, you’re going to have a shitload of people with ruined lives, but as a consolation prize, they can vibe code stuff! Wowee!
This very forum was founded by a VC who had great success recruiting 22 year olds with fancy diplomas to automate away the job of the guy who copied the numbers from the TPS report pdf attachment into excel.
I didn't see people on here ranting and taking up the flag of revolution for the TPS report excel paster guy's job that they were automating away with their web2 SaaS startup.
But wait- that guy himself was automating away the job of the lady who used to physically Xerox the TPS report and put it in the filing cabinet down the hall, but that lady was automating the job of the secretary who used to re-type all those TPS reports.
It's automatic filing cabinets all the way down, and ranting because your little slice of the filing cabinet automation machine has been made redundant is a bit silly.
You act as if this is the first time in history technology has wiped out a trade and made people scramble to sort out their lives. No, this has been happening over and over again throughout history and at a rapidly accelerating pace since the industrial revolution. Why should we ask it to slow down on behalf of programmers, when it never did for anybody else? Don't pretend you didn't know this was a possibility when you got into tech in the first place. You might have to downsize your life but humanity as a whole will be better off.
> You might have to downsize your life but humanity as a whole will be better off
This assumes that there will be other jobs to get. If AI replaces a large enough segment of office jobs then huge portions of the population will be unable to afford essentials like food and healthcare.
it also assumes humanity will be better off because people can vibe code stuff, which is a massive stretch already.
Walk into a staffing agency, ask for a job. They'll give you a list, pick the one that sounds the least disagreeable. Show up on time, every day, for at least two or three months and you'll convert the temp position into a full time job.
It's literally that easy, showing up reliably is a super power that puts you in the 90th percentile of workers these days. The job probably won't be as comfortable as sitting on a comfortable chair in an air-conditioning office wiggling your fingers at a computer, but so what? Other people make it work, so can you. Man up.
sorry, no jobs at the staffing agency, those are AI. Feel free to walk into a burger king, show up everyday, and flip those fries for minimum wage until you die. Man up brother, other people make it work. Sleeping on the street, well half the year its not even snowing.
You clearly haven't bothered to even look, so why should I even believe your concern is genuine?
The last study I know of that measured the conversion rate from temp to perm employees showed about 15%-30% success… and that was well before the gig economy really took hold. So you’re looking at 4 or 5 temp placements to reliably get a probably underpaying job when very few white collar workers could survive long enough to make the end of a lease, or sell their house, while on a temp job salary. It’s a viable option for a 25 year old that could couch surf for a few months, but not for a mid-late career professional, or anyone with a family.
You can give any complex problem a simple answer if you ignore enough factors.
The industrial revolution resulted in children being worked to literal death. Of people toiling 16 hours ad day and living in cramped up spaces without any windows and barely any hygiene. It brought suffering on a scale never seen before.
Organized labor movements managed to fight back and improve conditions somewhat but will we be able to do it this time?
Humanity will not profit from generative AI, tech billionaires will. It is based on the theft of human labor of millions of programmer, artists and writers without any compensation. If left unchecked it will destroy the environment, any form of democracy, our mental health. It will cause mass unemployment at a grand scale.
Could it be in theory used for good? Maybe. As the current political situation stands it will cause massive suffering for the majority of people.
Your glib dismissal of the real effects of those technological upheavals shows you haven’t actually looked into this. You should probably tamp down that smugness until you find out.
Did you think about it before you got into the tech industry? You should have, technology has been wiping out jobs since forever but you got into tech anyway. Live by the sword, die by the sword. Except you needn't actually die, just walk into a staffing agency and ask for a new job. I have done so before, and will do so again. I have prepared for what is to come, saw it coming 20 years ago and saw the imminence of it when GPT-2 was released. I have sympathy for other kinds of white collar professionals who never could have anticipated these kind of developments, but technologists? Give me a break. You knew, or should have known, that technological developments in this domain were likely.
Please stop pretending that this is only going to replace "tech workers". Do accountants "live by the sword"? Whose jobs did they replace? What about analysts, journalists, radiologists (one day, if not quite yet)?
And even within the realm of "tech", it's kinda bonkers to expect e.g. a firmware engineer to have some deep understanding of trends in ML/AI.
Altogether your #1 priority seems to be "bashing workers", the justification just being a matter of convenience.
Please stop pretending to read comments before replying to them:
> I have sympathy for other kinds of white collar professionals who never could have anticipated these kind of developments, but technologists? Give me a break.
That’s one big edgelord cop out.
I’m not in the tech industry anymore because in the battle of people who wanted to solve problems with software and money grubbing MBAs, the money grubbing MBAs have won. Now I’m a union machinist, and believe it or not, I’m concerned about the wellbeing of others. In manufacturing, companies are starting to face the consequences of shortsightedly selling out their workforce and are frantically clamoring to use the agonal breaths of its existing manufacturing industry knowledge base to breathe life into a new generation of workers. China becoming a manufacturing powerhouse wasn’t a foregone conclusion: we gave it to them in exchange for short-term profits. Our economy, national security, and the financial viability of a robust middle class is paying the price for their greed and arrogance.
The people running the tech industry can’t see the world past the end of this quarter, so they’ll never learn the lessons our society has learned many times over. Good luck. Unless you’re running a company, you’re going to need it. The soft, arrogant, whiny, maladroit white collar workers coming into the trades are pathetically ill-equipped to do actual work.
You've already done all that I can advise others here do, so congrats, I have nothing to criticize there. You've done it better than me actually, since you're unionized. As for soft white collar wimps washing out, people at the first job I had out of tech were taking bets if I'd show up for the second day, so don't think I don't know what you're talking about. I know it, I did it, and other people can do it too.
The problem with exporting manufacturing to China was this country lost the ability to make shit. I don't think this maps at all to white collar jobs getting gutted by AI; the people who actually make things aren't the white collar workers who should be sweating. Societies paper pushers would effectively be a parasite class leeching off the hard labor of people who actually work, if not for the part where white collar workers are (or have been) necessary to organize the logistics of everything that allows the people who actually do the work to actually do the work. We are on the precipice of dramatic change, and I think we're going to see a radical revaluing across society.
None of this is even new. Computers and other business machines already came for the clerks and secretary pools before most people ITT were born. The loss of these careers was not even remotely a problem for society at large, completely unlike offshoring manufacturing.
Your argument boils down to good'ol trusty Inquisition trick called Guilt-tripping Over The Original Sin
> Billions of people will be empowered to spin up whatever software they need on demand.
… so long as they have the money, and the power grid survives the overtaxation.
After all, why bother encouraging a culture where people are genuinely empowered to tweak and create their tools? Why encourage a culture of exploration, of playful cleverness? What use is there to being a hacker, of sharing knowledge?
It’s definitely much easier, more sustainable, and more fulfilling to have server farms adjacent nuclear reactors make your calculator app for you.
Hardware will get better, local models will get better, it won't be very long until running a coding agent off a few solar panels on your roof is viable. And if this isn't true, then the future I envision probably isn't going to happen and you lot can stop worrying about your jobs.
The way I see it is either it will work (which I doubt, but whatever) and I'm out of a job, or it doesn't work and the economy comes crashing down harder than ever seen in history and I'm out of a job.
The bubble popping is a guarantee even if AI does end up working. The internet certainly works but that didn't stop the dotcom bubble from crashing. Live frugally below your means to save up as much extra money as you can, and do whatever else you think prudent to keep your ass covered.
I’m not worried in the slightest about jobs. A world where acceptable slop is created on demand is one where people lose their curiosity and drive. Why bother practicing drawing if you can prompt an acceptable facsimile of your image to life? Why bother learning to program if you can prompt an acceptable slop program onto your machine?
Killing the drive to learn and explore is not empowering; it is fundamentally disempowering.
>billions of people will be empowered to spin up whatever software they need on demand
As long as: 1. They have access to a computer 2. They have affordable access to a capable language model 3. Someone will actually care about using their output instead of simply spinning up their own custom version of whatever idea they have
The number 3 is something many people miss, especially on HN: Why would I want to use YOUR software if it's easy for me to cook up my own? Perhaps out of efficiency or lack of time, in the same way I order pizza instead of baking my own when I'm tired or can't be bothered to bake pizza.
Then the software becomes truly throwaway, in the same way takeaway is, and everything is a greenfield project because rewrites are literally easier and faster to make than patching up existing stuff.
> Why would I want to use YOUR software if it's easy for me to cook up my own?
You're still in the mindset of thinking about software as something you sell to other people. Forget that crap. Software will be something you summon on demand to solve a specific problem you have. As long as people have problems that computers are good at solving, they'll keep using computers. What likely won't continue is computer programming as a career, but so what?
Everybody keeps parroting stuff along these lines but never take into consideration:
1. the stark, obvious reality is that most people don't know how to actually use computers! They know what 10 steps they need to take on a computer, in a specific sequence, to complete their task, but anything beyond that is too much.. and they need taught those 10 steps (as well as have it documented somewhere) for it to ever stick
2. not only not know how to use, but simply don't use computers at all! They've got phones and tablets and smart TVs and talk to their Bluetooth speakers and shit but they aren't sitting down at a desk with a keyboard and mouse and using a computer. I'd wager that of the percentage of people who do, an overwhelming majority is doing this primarily at their job to complete work tasks
3. companies with more than 10 employees are absolutely not going to be running to Claude to spin up custom programs to do their work. It's just not happening. Not to mention you can rarely even install unapproved, AAA-quality software on company-issued MDM'd hardware, let alone something generated out of thin air that has a ton of dependencies, no installer, no packaging, not code-signed, etc
4. that pizza you're ordering? You'd never order again if it was a roll of the dice with regards to what you receive. When you pay for two extra large thin crust pies with everything and are delivered some cheesy bread, a 2-liter of Coke and some brownies, your wig will completely and fully split and you'll never patronize that establishment again. Claude absolutely can make you exactly what you order, if you know what to order, and why, but most people don't
5. consistency and determinism matter to businesses, and to people -- both home users and professionals. Most people get stymied by the simplest tasks on a computer, tasks that have deep, instantly-available answers available with a single Google search or ChatGPT session. Guess what they do instead? Give up, and then ask IT or "a tech friend" for help ... how am I going to help you troubleshoot software I've never seen before? That NOBODY has ever seen before? That I can't even install because it only exists as a dev build in a single folder on your hard drive? How are you going to take that program with you when you upgrade your laptop? What if they didn't use git and their computer dies? Ask Claude to remake it? Will it be the same? Do they even know what git is? Do they even know where the folder on their computer that holds the files is located? Or what, you had Claude build a hosted product? Where's it hosted? How much does it cost every month? What if it gets hacked? I could go until my head explodes with all the hypotheticals
6. professionals pay for convenience and predictability, as well as to offload risk and unnecessary labor onto third parties. This will never change. Companies have been worrying about and hedging against "the bus problem" for decades, and vibe-coded software creates the ultimate bus problem: not only are you the only one likely to be in possession of the program in question, you're the only one who has ever seen it, know how to use it (which is different than knowing how it works), and it dies with you. Fine for a personal gadget, but a non-starter for a tool that a business or professional relies on to make their real money
I could go on and on, but you probably get the point. Takeaway food is both throwaway in a different sense than vibe-coded software, and infinitely more accessible to the average human. People are still going to pay for SaaS, still going to buy software, and still going to build software. In fact, I'm starting to think we'll see less open-source contributions and more closed-source, for-profit software released than ever before as a result of Claude and Codex, rather than a complete flattening and decimation of this industry. I think people in software will try to become more entrepreneurial as a result of corporate job loss. I also think that a byproduct of this coming tsunami of new commercial products is that the overwhelming majority will be low quality noise, and the proportionality of signal -> noise will remain largely unchanged. I'd use social media as an example (a staggering amount of people show up and try to break through, a very small percentage actually do) but IMO you see it in any industry: there can only be a few outsized successes in anything at any given moment in time (but also a not-insignificant amount of medium-sized success that flies under the mainstream radar)
I dunno. Maybe I'm full of shit, but I still think it's absolutely bonkers to think that the software industry is over because every person will just become sovereign groundskeepers of all of their own bespoke software. We can't all be our own bank, lawyer, doctor, mechanic, fitness trainer, software developer, chef and bodyguard, while also dealing with the other stuff that are our primary responsibilities! And that means that as long as society doesn't fully collapse into widespread economic ruin, and we aren't all unemployed, desperate, violent marauders trying to survive in District 9, there will be plenty of opportunities out there in the software space. They might just look a little different than they used to, and you'll have to go out and get them
> This is the future I dreamed of when I was a kid, and I'm not so cynical as to let the dying of a trade sour me to this objectively incredible technology.
I feel that you should take a longer-term view of things...
If an AI can vibe code from the requirements of the average white-collar worker, we're not talking about the death of a trade. Or even two trades. We're talking about the death of almost all white-collar jobs.
Development paid a lot more than other white-collar work because it was harder, and fewer people could actually do it. How fast do you think the easier work will get replaced if the hardest one is replaced? For the remaining white-collar roles that consist solely of skills achievable by a border collie, how much do you think they'd pay?
Yeah this is one of the most perplexing things about this "dying of a trade" narrative.
Software development isn't just the act of producing a deliverable that is being gate kept by people who use their own body. Software development has become specialized enough that it is often highly domain specific. To replace the "trade" you need to automate the software part and the domain knowledge part. If you can do both, you've automated every single white collar job in existence.
Since it is possible to write software for machine learning, which is used to solve problems that classical algorithms failed to solve, the amount of problems that cannot be solved using software is shrinking rapidly. If you can write software for any domain, you can solve any domain by using said software.
General purpose software generation can be reduced to AGI completeness. In a way, it is the last job that can be automated.
sure, a bunch of people will lose jobs, but at that trade off everyones dog can vibe code Royal Frog, a 4 level unwinnable game where play as king frog, eating peasant flies.
Shills like yourself indeed need not worry. You just find a new scam to shill.
+ Also the fact that the Memory.md file was a hindrance to the quality of output
Depends on the desired output. The author wanted variability, for which Memory.md was an obstacle. Another project might need consistency.
In this case yes, but the real takeaway is to pay attention to Memory.md. If I had a particular game in mind and it latched on early to a style I didn't like, there's no guarantee it would update the memory as I try to change the style.
> the engineering is in the scaffolding, not the prompting
Well, yes. Feeding random tokens as prompts until something good comes out is a valid strategy.
Simulated annealing for game design
Not that I condone any form of gambling but I would rather play actual slot machines instead of spending hundreds of dollars on tokens in hopes that the AI blesses me with anything useful.
This sort of thing is usually done on the frontier of LLM research. (Role play with local models.)
I think this misses something. The output here is something not the thing. Yes the scaffolding is important, but the requirements are even more important. You need crystal clear requirements + great scaffolding and then the implementation becomes mechanical.
This is amazing because it's the same logic and argument about how to do good software engineering that's been around for 40 years. If you just write good enough requirements, a good enough, detailed specification, then your software team can't fail, even if they are low-cost engineers from a developing nation. It's the classic Waterfall method.
That was totally upended by agile, that emphasized that yes, a clear, unambiguous specification is needed, and the best language for that is a programming language. Don't waste time writing a detailed spec in English, get right to writing it in code that you can execute and get immediate feedback on.
Now people want LLMs to write the code for them, so they are back to saying we just need to give the LLMs clear enough direction, a clear specification. It's amazing to witness history not exactly repeat itself, but very clearly rhyming
Sorry but what do you mean crystal clear requirements?
I don't particularly think "y7u8888888ftrg34BC" would pass as a crystal clear requirement at my workplace :<
Do you mean something different?
> y7u8888888ftrg34BC
This is more information than the average users gives you when requesting new features.
I mean you get a random game in the authors example :) But in real life you do not want a random game. That's what I mean, you need the great scaffolding + exact requirements. Then the prompt to do the implementation does not matter too much.
Recent research has shown that specs actually degrade output, so I’m more inclined to think you’ve got the slot machine mentality mentioned up-thread.
Source?
Last I checked, specs improved output quality. There are hiccups with complex, difficult specs but I guess that much is obvious.
https://arxiv.org/abs/2602.11988
this would be a more insightful comment if the output wasn't itch io shovelware.
> the magic isn't in the input, it's in the system around it.
It isn't [this], it's [that]. Is AI slop, just saying.
[flagged]
> the "intelligence" was never in the input It's quite literally in the authors prompt so in the input. it's in the article that without his prompt the gibberish input produces nothing of value:
"Hello! I am an eccentric video game designer (a very creative one) who communicates in an unusual way. Sometimes I’ll mash the keyboard or type nonsense like “skfjhsd#$%” – but these are NOT random! They are secret cryptic commands full of genius game ideas (even if it’s hard to see).
Your job: You are a brilliant AI game developer who can understand my cryptic language. No matter what odd or nonsensical input I provide, you will interpret it as a meaningful instruction or idea for our video game. You will then build or update the game based on that interpretation."
Also I don't know if you're an LLM or not but can we please not chatGPT-ify our comments like this? It figuratively makes me want to punch you through the screen.
The parent poster has already been called out at least once for commenting in form that reads like AI generated slop.
In fact, their only post that doesn’t read like AI generated content is the one reply to where they got called out.
Extremely clickbaity title that actually isn't clickbait because it happens to be a straight up description of the article - excellent post, how can one resist?!
No, the article’s title is definitely clickbait. The author didn’t teach his dog to vibe code games (that’s what the title on the blog is) – he taught his dog to be rewarded when he types random keystrokes on the keyboard. The vibe-coding is inconsequential – the dog doesn’t play the game, he’s just in it for the treats –, the author just wants the attention because he gets people to believe the dog DID vibe code.
It will stop being clickbaity if the author decides to let his dog respond to stimuli related to the game he’d be building with a feedback loop.
If my employer stopped rewarding me with treats (and health insurance) my vibe keyboard presses would cease too, if we are being honest :)
It's interesting as commentary, if you choose to read into it as if it were analogous to vibecoding by humans.
Doesn't matter, I don't consent to misleading titles influencing what articles I choose to read.
Very HN-like comment. Really channels Dwight. Made me smile, thank you.
I read it in "Comic Book Guy's" voice, but the effect is the same. Peak HN commentary.
[dead]
[dead]
[dead]
Who's a good software developer? [scritches]
hilarious, I'm in the office and had to try pretty hard not to laugh out loud
My goodness, finished all your Jira tickets early this week huh? That's a good boy! [fuzzy scritches and wharrgarbling intensifies]
Honestly I wouldn't mind a bit of that now and then myself, but I guess stable employment will have to do. Or is that only for the vibecoding horses?
The salary and health benefits are essentially treats.
Momo is. Yes, she is ahahah
The input method needs to be improved.
I can imagine a camera-based input that would help detect the wagging of a tail, or continued interest in the visuals as an indicator of doubling-down on a given feature.
The dog could actually vibe code a game to their liking, but with the wrong input (a keyboard) it's a missed opportunity.
Momo does like to bark at the TV. I have thought of combining this with nanobana and letting her down select options. Maybe in a future update.
From everyone needs to "Learn to code" to "Just have your dog vibe code it"
Reminded me an old joke about Bill Gates from late 90s:
"One coder got an insight that Bill Gates builds his products by typing with his butt, compiling and delivering it.
The coder typed for 20 minutes like that, compiled, ran, and got an output:
Only Bill Gates can code like this."
Not a joke anymore.
Wasn’t much of a joke then, from the looks of it.
[flagged]
Cute, but clearly Godot is doing all of the heavy lifting here.
"It's possible to make shitty but playable games by running random scripts through a >2MLoC game engine and iterating on errors" is interesting but not nearly as sensationalist.
You don't have to get too far into the article to see what's actually doing the heavy lifting[0].
[0] https://github.com/cleak/quasar-saz/blob/master/CLAUDE.md#us...
Nobody cared when I taught my roulette wheel to vibe code :/
Does it have personality?
No, but it rarely shits on the carpet.
> No, but it rarely shits on the carpet.
What do you mean "rarely"? It still happens sometimes?
Roulette is a game of chance.
The table had a rough life before it found its forever home. Sometimes it gets scared for seemingly no reason.
STREAM THIS
Better content than most streamers!
The buried insight is right: if random keystrokes produce playable games, the input is basically noise and the system is doing all the work. We've evolved past the point where intent matters. That's either the most exciting or most terrifying thing about where this is all heading. But I am glad I am sitting in the front row watching this all happen, especially a dog vibe code!
We haven't evolved past the point intent matters.
First, because there's intent in the very verbose initial prompt.
Second, because you have to factor in the quality of the output. I don't want to be a killjoy, but past the (admittedly fun!) art experiment angle, these are not quality games. Maybe some could compete with Flappy Bird (remember it? It seems like ages ago!), but good indie games are in a different league. Intent does matter.
You know the LLM is just something that's trained on human tokens.
We can probably create a dog intelligence by training it on dog tokens. Barks and stuff.
Same with dolphins. I wonder if multimodal models that know english tokens and dolphin tokens can cross the gap? Something to experiment with.
First sentence we would understand from Dolphins language: thanks for all the fish!
Oddly relevant for my multiyear project on getting my dog to vibe code b2b saas products https://dogomation.darefail.com/
Pretty neat! I actually ran across that right before publishing - I didn't want to see what was around until after I had the whole thing locked in. I love the novel input!
Amazing. Also very thankful the author included his setup on GitHub. Also the YouTube video is fun to watch. https://www.youtube.com/watch?v=8BbPlPou3Bg
Just a quick note that I have nothing to do with any meme coins. Looks like folks are using myself and Momo to pump some crypto. I won't be claiming any coins - even though they've been offered. I'd recommend others stay away from it as well.
Any popular thing with a dog in it is now a rug pull opportunity!
> Hello! I am an eccentric video game designer (a very creative one) who communicates in an unusual way. Sometimes I’ll mash the keyboard or type nonsense like “skfjhsd#$%” – but these are NOT random! They are secret cryptic commands full of genius game ideas (even if it’s hard to see).
Your job: You are a brilliant AI game developer who can understand my cryptic language. No matter what odd or nonsensical input I provide, you will interpret it as a meaningful instruction or idea for our video game. You will then build or update the game based on that interpretation.
Here's what you should tell your coworker the first day on the job if you get hired to do something you know nothing about :D
That is a very succinct way to describe what it feels like to have a job that is cleaning up vibe code. Maybe (just maybe) I'd understand if this was a prototype from someone with zero budget. But you just know they are going to continue to "prototype" once they being you aboard. And many will complain about how slow everything goes because they are used to their fast iterations off of unscalable code.
Its frustrating in an interesting way. With other aspects like machine language people quickly understand that this isn't sufficient for a proper transition and compromise with it. Code being more nebulous doesn't get that grace.
If someone could please give an octopus a waterproof keyboard, perhaps we could have a kernel, a compiler and a new internet protocol all in one.
A thousand dogs typing on a thousand typewriters...
One can technically scrape a list of actual advice or quotes off the internet, randomly feed them to a coding agent, and ask it to interpret what they mean in the grand scheme of things and implement away on it. Once the agent is done, it randomly responds with either "yes, this is exactly what I meant" or "no".
In turn mimicking the average game industry executive giving vague directions that feel just right to them this month, or some other unspecified time period, and in turn achieving something closer to the real AAA game development lifecycle.
I actually have an "oblique strategies" skill that it can call if it figures out it's been spending too many turns on the same problem...
https://en.wikipedia.org/wiki/Oblique_Strategies
On January 13th, I woke up to the news that Meta had another round of layoffs and my role specifically as a research engineer had been eliminated.
Sorry to hear that! Hope OP got a good sev package at least?
It's called a sev pak these days.
Dam, I really thought this would be much more interesting than it is
People have been doing some cool stuff for like a decade with giving dogs buttons to use human language, something they can seemingly get decent at communicating effectively with if they can get around the pesky issue of not having the sophisticated vocal machinery needed to produce recognizable phonemes, through the power of a good interface for them, even if the output is discretized to the level of words
I thought maybe this would be about creating a way for a dog to create stuff said dog might actually want or enjoy via the more powerful lever of effective long-context natural language processing that came of a similar tokenization approach - which can even sometimes churn out working code - that we have now
Instead it seems to be an exploration of how the capabilities you can produce from essentially random noise from this technology is less distinguishable from the result of thoughtful input than I might have hoped. Still interesting, but way less so
Are you disappointed that the LLM generated random games, or are you disappointed that the dog didn't actually care about the games?
Neither are that surprising to me, tbh.
The LLM being able to generate random games is kinda expected behavior. It's trained on sequence probability distributions that include the code for a great many games and nudged toward doing so by the human user. I'm disappointed that the dog is basically used as a noise generator here. A process driven by the desires of the dog in a meaningful way would be more interesting and at least seems somewhat plausibly technologically feasible, and the article's title kind of implies it. I am especially disappointed because what most excites me about new technologies is applications that were not possible before its invention, which this seemed like it could be an example of
As someone vibe-coding a game in Unreal Engine 5 for the last few months, I found this really funny.
Unfortunately I don't have a dog but I do have a design plan so ultimately I'll end up with something a little more deterministic. Possibly. Don't know.
Not sure all this AI-driven workflow is necessary. The dog could just create a muddy mess and it would be valid Perl:
https://www.mcmillen.dev/sigbovik/
This makes me wonder if dogs understand (or can be trained) the concept of building. Say you did something physical - 3D scanner and 3D printer connected to automatically attempt to replicate an object placed in the scanner area (and maybe then on button press). Could a dog understand it, and intentionally try to replicate its favourite toy, or a bone, etc.?
Could this be done better with one of those dog button mats? The concept is interesting, but, it mostly just seems like an AI trying to interpret keyspam.
Both my dogs have actually learned to use the button mats. Down selecting to the right responses seemed tricky. My wife also took away the mat since Hana (the larger one) never learned "all done" and would paw at the "walk" button until she got it out and carried it around.
> mostly just seems like an AI trying to interpret keyspam.
aye, but the whimsy is the point!
Yes, I was hoping for a system where Claude was informed it was communicating with an unusually intelligent dog whose ability to communicate was limited by dog anatomy, and that the AI would not to hold the dog's interest with its output.
> like an AI trying to interpret keyspam
'nuff to run most governments nowadays (Europe and US come to mind. 2026 and they have the Space Programs of DIY youtubers with money, whaaaat) so why wouldn't it help a dog helping his dog vibing game(s)?
Pretty sure he would have gotten very similar output just by saying "generate a random game using Godot and c#" but that wouldn't make for a viral post so instead he asked the model to pretend meaningless input is being used by it and added a dog in the process of writing such input because that helps the virality of the whole thing.
I think you would have gotten more generic games. The AI was clearly attempting to find meaning in what the dog typed, and that drove what it made.
Now, if Anthropic let you adjust the temperature, then maybe you could have done it without the dog...
The AI cannot drive meaning from the dog's input because there's no useful information encoded in there. It's effectively a random string (if there's less randomness, it's just because it's a dog's paw physically pressing on a keyboard).
All the relevant information was in the initial prompt and the scaffolding. The dog was not even /dev/random, it was simply a trigger to "give it another go".
The shapes of clouds and positions of stars are essentially random, and yet humans derive meaning from both. I agree you could have gotten the same results via /dev/random, or probably by increasing the temperature on the model, but I suspect doing one of those things is important.
The LLM cannot derive meaning in a human sense.
The shapes of clouds and positions of stars aren't completely random; there is useful information in them, to varying degrees (e.g. some clouds do look like, say, a rabbit, enough that a majority of people will agree). The mechanism at play here with the LLM is completely different; the connection between two dog-inputs and the resulting game barely exists, if at all. Maybe the only signal is "some input was entered, therefore the user wants a game".
If you could have gotten the same result with any input, or with /dev/random, then effectively no useful information was encoded in the input. The initial prompt and the scaffolding do encode useful information, however, and are the ones doing the heavy lifting; the article admits as much.
> If you could have gotten the same result with any input, or with /dev/random, then effectively no useful information was encoded in the input.
It's not that the input contained useful information—obviously it does not—it's that it's causing the output to be more random, and thus more "creative".
Without the gibberish, "generate a random game" would likely repeatedly surface high-probability concepts—platformers, space shooters, tower defense—whatever sits near the top of the model's prior distribution for "game." The gibberish causes the model to land on concepts like "frog" that it would almost never reach otherwise.
Before you know it, we're going to have a real world proof of the infinite monkey theorem https://en.wikipedia.org/wiki/Infinite_monkey_theorem
"Historians looked back and determined that it was around the year 2026 AD-HE (Human Era) - that the prime canine species began to raise from merely companion to colleague...and so the Dog Days began....woof" - Puppers Domingo, Good Boy, Esquire.
If you haven't read the scifi classic "City" by Clifford Simak, please do so!
(But your comment hints that you have already).
This seems like a good way to get a feel for a coding model. It's like the images you get out of a diffusion model when fed an empty prompt.
It does. Claude seems to do the best with this prompt. Codex 5.2 struggled with UID generation and kept ending its turn with things like "And now you're all setup to run tests!" without actually running them. A better (and shorter) prompt could probably get a lot out of Codex.
In the world of vide coding agents, nobody knows you are a human...
This perfectly demonstrates the absurdity of our current situation around the LLMs and "AI".
Better figure out how to replace management and HR dept with dogs
It's actually extremely similar: the agent has to figure out a way to associate the next logical steps with the (often disconnected or nonsensical) directives the executive gave them.
It might be a little easier with a dog though. With a dog, you just give it treats and it doesn't care how you interpret what it typed.
Pretty sure just a drop in replacement would be an immediate improvement.
The next round of massive tech layoffs will be ruff.
gonna be good stuff tho. dogs are mostly lovelier
Who needs "claws" when you've got paws!
Thanks for sharing this unusual experiment, I love it! This is the type of content that makes HN, HN.
Now, I started considering hiring my three little kitties and their mom for a job like this. They spend the whole day sleeping and waiting for meals but now, they have to work, hard, in collaboration with Claude Code to pay for their rent and meals :)
I was wondering, what would an AI model that was built to be aligned with an animal species interests, not Humans look like.
It would take a giant corpus og barks and howls, comparable to the entire internet and all books and films to make a doggie-LLM.
The real insight here is that the dog has better product instincts than most PMs I've worked with. It knows exactly what it wants (treat), doesn't overthink the implementation, and ships fast. Half the battle in software is having clear intent turns out that's species-agnostic.
This is an extremely cute, cool and fun experiment. Kudos.
That said, I wonder: does the dog input matter? It seems this is simply surfacing Claude's own encoded assumptions of what a game is (yes, the feedback loop, controls, etc, are all interesting parts of the experiment).
How would this differ if instead of dog input, you simply plugged /dev/random into it? In other words, does the input to the system matter at all?
The article seems to acknowledge this:
> If there’s a takeaway beyond the spectacle, it’s this: the bottleneck in AI-assisted development isn’t the quality of your ideas - it’s the quality of your feedback loops. The games got dramatically better not when I improved the prompt, but when I gave Claude the ability to screenshot its own work, play-test its own levels, and lint its own scene files.
I'll go further: it's not only not "the bottleneck", it simply doesn't matter. The dog's ideas certainly didn't matter, and the dog didn't think of the feedback loop for Claude either.
This fun exercise might actually be extremely insightful as a educational vehicle around AI and intent.
It can also help combat the excessive emphasis on any "end to end" demo on twitter which doesn't really correspond to a desired and quality sought outcome. Generating things is easy if you want to spend tokens. Proper product building and maintenance is a different exercise and finding ways to differentiate between these will be key in a high entropy world.
> I'll go further: it's not only not "the bottleneck", it simply doesn't matter. The dog's ideas certainly didn't matter, and the dog didn't think of the feedback loop for Claude either
Absolutely. The scientific test would to put any other signal and look at the outcomes. Brown noise, rain, a random number generator. whatever.
Of course the dogs input matters, it matters to the fun of this project. Would’ve been way less fun otherwise.
I don't disagree actually. It is a fun project!
Thought this is quoting Karpathy for a second there
i'd love to know what happens if you change the prompt from "Hello! I am an eccentric video game designer (a very creative one) who communicates in an unusual way. Sometimes I’ll mash the keyboard or type nonsense like “skfjhsd#$%”..."
to
"Hello, i am a dog. i will mash the keyboard randomly when i want treats. make a game for me"
Would be really curious to see
Love it. No Infinite Cavapoo Theorem needed. Give Momo a week and she'll have DOOM running on her treat dispenser.
Yet the only thing the dog wanted was a cuddle and a frisbee
yeah that's capitalism for you. No treats until you provide value.
Maybe I could make a game after all! You bring hope to a whole generation of lazy developers :D
It would be nice if the dog could play the game and its input could improve it. Why not a setup with those large buttons on the floor that some people have to let their cat or dog communicate with them?
I have a dog who likes to watch tv; he mostly likes hockey and commercials. I've never thought about coding a game for him until now.
I'm thinking of remote buttons to make his favorite things appear on tv. This is going to be awesome.
DogeCode incoming. People here are already talking about the scaffolding. Let OpenClaws provide the scaffolding and let the dog operate the prompts at $5 per day.
This is a billion dollar idea! No humans. No revolt. No guillotine. Just profits!
> Just profits!
Sounds like open communism. No chance, buddy, it's either less or more viking, but not just viking. Pick a camp the profits are for or get surrounded by trashy turd nuggets even Ronald felt enough pity for to give them some poourpes
If you haven’t watched it, please watch Rick and Morty S1E2. Your dog’s name is Snuffles, and he is teaching you how to teach vibe coding to dogs, to educate you.
That episode aged way better than expected
Overall it feels less like "dog makes games" and more like a glimpse of what happens when you wrap modern models in the right scaffolding
Foolishly, I clicked through thinking "wow, did they get an LLM to interpret their dog's barks and body language? That's actually pretty cool!" Look, I'm an LLM skeptic, but if they'd done that, I would have to praise the result.
But no.
They just told the LLM to try and find meaning in keysmashes.
This is really fascinating, only if there was a way to make your dog receive feedback as the game develops. Like make it pick color by matching placement of treat based on the color shown on the screen.
Similarly, do it for story telling narratives, game textures etc. Although I do not think the dog understands natural language so all of it will likely be a dud.
With some luck and scraping "y7u8888888ftrg34BC" might one day replace Solitaire in Windows.
So the cute lovable dog is an entropy generator.
Next: use hot cup of tea as Brownian motion source. Invent infinite improbability drive.
Very cool.
> But bugs crept in during testing - a couple of times it dispensed multiple servings in a row. Unfortunately, Momo picked up on this and now keeps mashing the keyboard hoping for a second immediate serving
Attempts to mash during no-mashy time need to play a horn. Reliably followed up by a no-treat.
Given the prompt and a random string generator, would the LLM still produce a game? Presumably yes. In that case I'm not quite sure that the dog here has any real involvement. It's could be replaced with the yes command.
Amazing story and work!
You mentioned Claude not being able to see the games. What I really like for this is the Claude Code Chrome Extension. You can easily make godot build a web version, and then have Claude debug it interactively in the browser.
So... I have 6 cats. I firm believer that no amount of AI will help them produce anything.
Haha, was a great read, loved the picture of the dog at the keyboard!
It’s the first time GIGO stands for “garbage in, games out”. Bravo!
We might need a new benchmark, the old pelican riding a bike doesn't cut it anymore. Something like a mailman ringing the doorbell would be more fitting. Or a squirrel riding a trike.
Will we ever get to a point where LLMs just churn out random apps with no input required and human reviewers just go through the apps picking out which ones might be useful for business purposes and monetizing them?
You could but it would be a hugely inefficient system. Like, just address an actual need dude.
Most saas isn’t limited by the code behind it anyway. That almost doesn’t matter, even before LLMs. It mattered that there’s support, customer onboarding, solving a businesses issues, customer story, adapting to the needs of their business partners, etc. All of which require large amounts of real human work.
What? You are not doing it already? Look at this guy...
I've been trying out vibe coding with my 4 year-old, but they quickly lose interest once we start getting into the "weeds" of implementation. Hey kiddo, which CSS library should we use for your web game?
I think you just need more treats.
THIS is why I pay for my Internet :)
From "On the Internet, nobody knows you're a dog" to Paws coding and BarkGPT and BarkLM
I would like to see a game made by doggos, for other doggos :D
meanwhile cats: https://socradar.io/blog/dark-web-profile-blackcat-alphv/
Nice project.
This makes me think I should make my plants vibe code games or tools to optimize their well being! Maybe bio-electrical fluctuations --> vibe coded humdifying tools and games
I think this is fun. I'd like to try with my cat, although training cats is an impossible endeavor... I'm smart enough to enter gibberish myself without another animal, tough.
The fact that LLMs pick from the most likely tokens is really on its side here when the objective is putting together a plausible continuation of random characters.
This is pretty fun!
I'm interested in what will happen if you replay the prompts with different LLMs and the same LLM. I wonder how different the games will become?
This is basically a diffusion model: start from a random seed, use a generative process to transform it into something
To be honest I look with scorn at non-dog (human) developers building hobby indie games with AI en masse.
Let me explain.
The nature of the indie game development is pouring your love into a project and thinking about passion first and monetary incentives second.
Noone is thinking "I will make this game and it will make me filthy rich" or if they do they are... strangely minded.
It's like 'mass produced AI local craft'. Oxymoron in itself. Worst of the two worlds.
Where I see AI is empowering single developers to craft things they couldn't before. Not some small slop factory pipeline where you release game after a game everyday drowning steam in your 6/10 slop.
No. This should be ostracized and condemned.
What is proper beneficial to everyone usage is producing a game that is the size and scope that was unachievable for you before.
This is what I am doing. This is how AI is meant to be used. To empower us doing things that weren't achievable for us before.
Obviously dog produced games get a huge endorsement man and get a pass.
'Ewe Heard Me!' reminds of that looney toons sheep raider game on ps1. And it's exactly the kind of game I'd expect a dog to make
Wow, a great example of how vibe coding isn't coding.
You're just the random seed to the money furnace remixing existing games and code.
Sending random noise in, getting a game out.
Who wrote what?
If ever I saw something that made me want to move to a log cabin the woods and never touch a computer again, this is it
> y7u8888888ftrg34BC > Alright, let me decode this masterpiece of cryptic genius...
Those three dots made me smirk.
Claude is subconsciously a fan on Crystal Quest?! Loved that game on the Mac back in '95!
The article and video are great satire too.
Dogs are undefeated at reinforcement learning.
That photo of the dog with it's keyboard is unsettling: the new breeed of software engineers.
How did this get to the top of the frontpage?
It's funny? I liked it.
Funny is subjective, I should just have moved on and ignored this but I couldn't help myself, this is so irritating.
It's a prompt that makes an LLM turn iuqefxygn9urg0fh1 into a little Godot game. It's like a slot machine with no payoff, and the dog component is slapped on top of it and makes no difference whatsoever in the project.
> It's a prompt that makes an LLM turn iuqefxygn9urg0fh1 into a little Godot game. It's like a slot machine with no payoff, and the dog component is slapped on top of it and makes no difference whatsoever in the project.
Right, but it also has a "modern art" vibe to it that is fun. Silly, but fun. I think it's more about the initial prompting and feedback loop, the dog itself could have been replaced by /dev/random.
"Hacker curiosity" and "intelectual stimulation" are also subjective, but that's what HN is supposed to be about.
It seems to me a case of the blog post title inspiring the project, instead of the other way around. But I am particularly curmudgeonly today.
That's ok. To be honest I had to suppress a similar feeling when I noticed the dog is just an entropy generator.
But then I realized I find this kind of whimsy article more fun than a lot of what gets accepted unquestioningly here on HN. It seems light hearted and done in good fun, and it's engineering-related, so no harm done.
It's funny because vibe coders and AI artists think the slop they generate is no less the product of their intellect and talent than with human professionals, but really they're doing little more than stirring the entropy pool in a magic box with terabytes of stolen valor from better more talented people. They're no more an "artist" or "game developer" using AI than this dog is.
that's all llms tools are, slot machines for idea guys.
Most LLM articles depress me. At least this one made me smile, even if it's more about turning a dog into a random generator and generating games from random input.
It has vibe code and dogs in the title
By nonexistence of downvote feature
We were so focused on singularity, it may in fact be who you suspect the least - man's best friend :)
So /dev/random would presumably work just as well here too.
This is kinda closer to the LLM building a game on its own.
You're missing the important part about needing to model a tiny paw mashing on the keyboard. /dev/random is insufficient.
If you want a picture of the future of SWE, imagine a tiny paw mashing on a keyboard — for ever
Not really. You just found another way to tell Claude "I want a game, I am uncertain, let's randomly seed the start of the process.". But you did it circus-style :)
Forget k8s, k9s incoming... :D
And the most HN title award goes to...
Author should switch the keyboard to Dvorak. Gives more interesting gibberish when mashed.
I've been having this thought about how generally people say that llms cannot create novel things.
Say writing an interesting or novel story.
And was thinking about if feeding in prompts of random words, along with prompts grounding from a simulation would sort of push the llm into interesting directions for implementing an on demand narrative story.
A sort of randomized walk with llm.
I remember watching Terry Davis with this random word generator in his terminal that he would interpret as the voice of God.
Here I guess the seed is the Voice of Dog.
I actually found a web version of the god speak.
https://jcpsimmons.github.io/Godspeak-Generator
Maybe another word list would be more appropriate however.
love the article
slightly concerned tomorrow morning's top HN story will be karparthy telling us how dog-based LLM interfaces are the way of the future
and you'll be left behind if you don't get in now
(and then next week my boss will be demanding I do it)
The programming workspace of the future of the future will have three employees:
A man, a dog and an instance of Claude.
The dog writes the prompts for Claude, the man feeds the dog, and the dog stops the man from turning off the computer.
Thank you for the good laugh! This whole thread is peak satire. Although, be careful. It reminds me of the foreword to a shortstory someone shared on HN recently: „[…] Read it and laugh, because it is very funny, and at the moment it is satire. If you’re still around forty years from now, do the existing societal equivalent of reading it again, and you may find yourself laughing out of the other side of your mouth (remember mouths?). It will probably be much too conservative.“ — https://www.baen.com/Chapters/9781618249203/9781618249203___...
I for one welcome our furry overlords
You're right. They did it. The old man and dog joke has been realized, but the real answer of the future turned out to be: "the dog programs the game, and the man feeds the treat hopper."
That was funny. Gave me good laugh. Thanks..
Everybody and their dog will be doing it. Actually, the dog will be in charge. Dogs are loyal, enthusiastic, and require less office space. With their endless desire to play and to please, they will take over the game development industry.
In the meantime, the financial industry will be taken over by cats.
They also don't take 20 years to become smart like pesky resource-exhausting humans. I bet you could be up and running from a pup in 10-20 months.
I still can’t believe Altman said that. I mean I can, but still.
I can because I have also used similar arguments. There are people who say that you should use a real artist instead of AI due to AI's water use. Yet in actuality asking a human to draw something will require more water. There are people who think AI uses more resources than humans which is why it must be said.
> There are people who say that you should use a real artist instead of AI due to AI's water use.
Nobody I know says this. In fact, I've never heard of this ever before, and I read artist and hobby communities pretty hostile to AI, but I never once read this nice strawman you've built.
People say you should use a real artist instead of AI for a multitude of reasons:
- Because they want to enjoy art created by humans.
- Because it provides a living to artists, even artists for minor work like advertising or lesser commercial illustrations.
- Because AI "art" is built by stealing from human artists, and while human art has a history of copying and cloning, never before has tech allowed this in such a massive, soulless scale.
Sam Altman gave a deranged, completely out of touch reply, and he should be called to task for it, not defended. A human being is not some number on a spreadsheet, built over 20 years in order to achieve some "smartness" goal. That's a very stupid thing to say.
> A human being is not some number on a spreadsheet, built over 20 years in order to achieve some "smartness" goal
But from the perspective of the business and capitalism that's exactly what a human is. A tool that consumes resources and hopefully produces more value for the business than it consumes.
Sure we can dance around this and you can pretend your employer gives a shit about you and your family and your childhood stories but they don't.
> Sure we can dance around this and you can pretend your employer gives a shit about you and your family and your childhood stories but they don't.
I don’t get what you’re doing here. They didn’t say anything like that.
> I don’t get what you’re doing here
You said that a CEO was out of bounds for framing employees as numbers on a spreadsheet. To me this suggests that you believe company owners should care about the humanity of their workers. And I'm saying they don't.
I think by "they", Forgeties79 means me, the_af.
I get the general point you're making. Indeed, Altman's take is capitalism taken to 11. There was a lot of that going on before AI or the past few decades, but I don't think it wasn't as extreme and for every company. There's definitely a conversation to be had about modern capitalism (and plenty of people studying it, too). However, not everything is a FAANG or tech startup. Some owners do care about their employees to a higher degree than just numbers on a spreadsheet (not going into the whole "we're a family" bullshit speech, I mean the genuine stuff).
Imagine thinking of people as "resource-hogs before they reach peak smartness"!
What's new here, in my opinion, is people like Sam Altman behaving as if they didn't understand normal human behavior. You cannot simply compare an LLM to a growing human. You cannot say things like "grow a human over 20 years before they achieve smartness". What? That's not how human beings think about human beings, and Altman is detached from real human behavior here. He's saying out loud the thoughts he should keep to himself, a bit like a person with coprolalia. And it's ok for us to dislike him for this, even if he's just voicing the opinions of extreme techno-capitalism.
Sam Altman once joked (?) he wouldn't know how to raise his child without ChatGPT. Maybe he should ask ChatGPT how to behave more like a human? Or at least fake it?
> Sam Altman once joked (?) he wouldn't know how to raise his child without ChatGPT. Maybe he should ask ChatGPT how to behave more like a human? Or at least fake it?
Not to mention that was at a time when all kinds of wild suggestions like glue in pizza were coming out of ChatGPT’s sloppy outputs. There are so many little things that quickly become big things with kids, annd exhausted parents should absolutely not use LLM’s for sussing those things out.
I could easily see well-meaning parents looking for healthy snacks to make their kids accidentally feeding their baby fresh honey, for instance. Or asking how much water to give their infant and not realizing the answer is absolutely none unless they are severely dehydrated from an illness or something.
There are a lot of hazards for kids under 1 in particular that make me incredibly nervous to ever suggest exhausted parents use LLM’s to answer kid related questions. Recommendations also change relatively frequently so who knows if it’s even pulling on the most recent best practices.
It's that second point. We live in an age of artificial scarcity created by a system of social organization that we've mostly not argued about since the 50s, that's now showing it's stretch marks.
If it weren't for the need to 'earn' a living, I'd say to the other two points: Por que no los dos? Save for the capital argument (which is valid, I'm not saying it isn't. You will starve if you don't make money), why is it necessarily true that the two (AI and people) are in competition?
In fact, I think "actual" artists would benefit incredibly from the use of AI, which they could do if it weren't a shibboleth (like I said, for good reason). You'd no longer have to have an army of underpaid animators from vietnam to bring your OC to life - you could just use your own art and make it move and sing. We'd not need huge lumbering organizations full of people who, let's be honest, work there making other people's dreams come to life in large part because it's a better bet than taking a joe-job at the local denny's (after all, you're doing the thing you love even if it isn't truly "yours").
I've had this discussion with younger folks, who are legitimately shook by the state of things. They're worried that all the work they've done to this point is going to be moot, because they've correctly assessed that the whole capital system isn't going anywhere any time soon, and they've been prepping to try and get a job at netflix, or disney, or paramount - because that's the world we've handed them. They see those positions drying up and what else are you going to do? They have the power financially and politically and without them you're doing "not art" for work, which sucks because you need to work.
I say; eat the rich. General wildcat strikes until UBI. Tax the everloving shit out of capital gains and peel back personal income taxes. We (the millenials) were handed a steaming pile of shit for a world, so at least we know what would constitute not an absolute disaster for Zeds, Alphas, etc. Have I gone totally off the rails for a conversation about AI? Actually, I don't believe so. The cultural pushback is a function of a busted system. After all, it's the economy, stupid.
>Nobody I know says this.
>I never once read this nice strawman you've built.
The instance of it I found was in a YouTube comment section.
Ok, let me reframe in a less assertive way: it's not common to say that you should hire human artists "because of water", so uncommon it's not a widely held belief I've seen in artist or hobbyist communities, and therefore using this to justify Altman's deranged remark seems weird to me.
I’ve also seen people assert that the Earth is flat in YouTube comment sections so I wouldn’t let that color your view of what is considered “popular opinion.”
Over a year ago yeah I occasionally heard that argument or some light variation of it, though not nearly to this ridiculous extent that you’re portraying now. Now? It’s basically a strawman. Most people’s objections revolve around the theft/reckless scraping that has literally taken down public infrastructure required to train these models as well as the ridiculous expectations being put that all of us implement it in literally every aspect of our lives even if it doesn’t fit, especially professionally.
> a human to draw something will require more water.
That human would require the same amount of water whether you ask them to draw or not, and would exist anyway because they are not born for productivity reasons. "Creation" of humans isn't driven by the amount of work to accomplish.
You are not causing more water to be used by asking a human to work on something.
Same for energy consumption.
This argument doesn't work at all.
What you do for humans to use fewer resources is to work on making us produce less garbage, and produce things using techniques that are less resource-intensive.
> You are not causing more water to be used by asking a human to work on something.
That's certainly not true. Asking me to think hard about something will cause me to burn more calories. Asking me to do physical work even more so.
Would you be completely at rest if you were not asked to produce stuff?
Do you think AI replaces our hard thinking and our physical work?
AI or not, I personally intend to keep thinking and my physical activity.
I don't do any moral judgement at all, and I also don't predict the future.
I respond to "You are not causing more water to be used by asking a human to work on something.", because that statement is false. (Mental) work has an effect on the human metabolism.
I'm not judging, I'm telling you (a bit snarkily, true) that your brain activity won't stop with something else doing (a part of) your work. And this is the subtility that makes the statement true. Said overwise: sure, you consume marginally less at rest, but you won't be at rest, making the remark pointless.
Even if it were false, the difference in energy consumption is not significant, taking on acount what the AI uses, and also all the energy that you use to live (housing, heating, products and food you buy whose production uses energy, etc).
And about the water, it's even worse, even disregarding the AI: at rest, maybe you'll drink, I don't know, 1L less (that's a wild number!), compared to the 100(s) L that you use to cook, wash, clean, etc, not even counting the water used to produce stuff you buy.
But again, you won't do nothing. We are commenting on a post of a guy who was fired and couldn't help creating something. That's how we are. We hate boredom.
Worse: the way our societies are setup makes is so that ai, if it helps at all, likely won't free us from work, it will likely just make us collectively produce more garbage. That's more energy consumption, not less.
That "but don't forget humans consume a lot of energy too" argument is at best not connects to reality, more likely a Sam Altman lie and you shouldn't take it seriously.
This whole discussion is wild to me. Comparing people and machines like this is not productive. It is not actually answering a serious question.
It helps to picture some sort of extraterrestrial saying this. Maybe someone like Alan Tudyk in "Resident Alien". It makes much more sense than to assume it's a human being saying these things.
These guys can't watch a dystopian scifi movie without picturing themselves as the bad guy.
Yup. The Torment Nexus meme proved depressingly accurate.
>In the meantime, the financial industry will be taken over by cats.
Cats would certainly be less flummoxed by stock values suddenly plummeting; they may even enjoy knocking them over.
Beavers will control construction and infrastructure... building dams, bridges, and entire housing developments with zero corruption.
Please, be real.
There will be a Simon Wilison submission linking to his blog linking to karpathy xit. You know, the usual good stuff.
Have you any idea what you have just done? You have uttered his name and now he has been summoned. You have doomed us all.
I scrolled this thread fearing seeing that name and a SVG of a dog riding a bike
[dead]
CODEOWNERS will be replaced by the usual means of marking territory. Let's hope our laptops are liquid-proof.
Looking for the headline about "dogs replacing engineers"...
I’m not ok with robots replacing people but dogs replacing people? now you’ve got my attention
Basically the plot of "City" (1952) by Clifford D. Simak. Bloodthirsty humans are succeeded by a pacifist dog civilization.
this is a really funny comment. underrated.
Some dude vibe codes OpenPaw and gives credit to his XL Bully called Threadripper that would never hurt another person, gets acquired by OpenAI for 7 figures total comp purely on clout, and both simonw and Karpathy are calling it the next best development in AI because it draws penguins and industrialises slop while Sam Altman talks about the negative impact of human life compared to AI data centres while pleasuring Jony Ive in a coffee shop.
Management already got us all copilot licenses .. soon they will get us all a dog
This would definitely make me want to rejoin tech.
Management would get us all dogs and they would get cats for themselves ... like blofeld from James bond ...
Not a problem for me... I'm a dog person.
The world is not ready for BarkGPT.
CatGPT would be cool but when you really wanted to chat, it would just ignore you.
Haha, I got that impression too! I was ready to hate the article, but it was really well written and I loved the "decrypt my dog's keyboard smash"
I think we can all agree cat LLMs are the way of the future though.
This makes sense! Many people trust that the functions of a human worker are asking questions and clicking ok. What if AI become better so we just a dog?
The funniest thing here is that you think you will still have a boss next week. They can just hire the dog now!
Sounds like someone is at risk of being left behind in the “permanent underclass” sir!
Partisan of randomness here. I never thought it could be as high as that. GG.
You can automate Momo with a rng.
can you automate love, visarga?
Not with bare rng
my dog had something to say about this:
woof woof, woof woof woof, woof woof, woof, woof woof woof
[dead]
need to see one of those dog button press setups but connected to Open Claw.
This is amazing. Like circus animals jumping through hoops.
this is incredible. we need more projects like this in the world!
Although I would recommend a more sturdy dog breed, for when the angry mob that can't buy RAM sticks and SSDs this year shows up at the front gate.
Go Momo go! If you want to hook up multiple dogs and have them reach consensus I'm down. I have a 15 lb havapoo I can volunteer ( he needs to help with rent )
How I am slowly killing my dog with too much treats.
(is joke)
I'm having a mixed feelings about this xD
Don't give this to a schnauzer.
We've officially reached peak hype.
Every man and his dog can vibe code.
There goes all the prompt engineering jobs
And the game is ... Fetch that stick.
[flagged]
This crosses into personal attack, which is against the intended spirit of the site. Can you please not do that? It's not what this site is for, and destroys what it is for.
If you wouldn't mind reviewing https://news.ycombinator.com/newsguidelines.html and taking the intended spirit of the site more to heart, we'd be grateful.
how is it a personal attack though? sorry I fail to understand. against whom - the dog being kept in addictive loop or against the OP or against the person who was rejoicing the fact that s.o. else was laid off. I beg your pardon, but this statement of yours seems is super subjective and truly vague. thanks for the attention anyways...
The comment implied that someone was responding to their life difficulties by making a dog suffer. That's a huge personal attack, and the added snark was an amplifier.
Makes dogfooding much easier.
its so crazy, people have so much free time that they do random stuff
Dude was laid off.
Really amazing work, congrats!
Museum directors, this is a piece of art. Give this installation its place in history. I have certainly seen worse artifacts in your lots.
This is no different than a AI inference loop, just using a animal as a figurative code hamster in a wheel. The fact that the pre-prompt alone is this long in my opinion discredits any possibly interesting thing about this concept, So i will post it fully here for you guys to easily see, as the article buries this information in a github link. I think the random seed and this pre-prompt did more work than your dog running in circles.
System Prompt: Hello! I am an eccentric video game designer (a very creative one) who communicates in an unusual way. Sometimes I’ll mash the keyboard or type nonsense like “skfjhsd#$%” – but these are NOT random! They are secret cryptic commands full of genius game ideas (even if it’s hard to see).
Your job: You are a brilliant AI game developer who can understand my cryptic language. No matter what odd or nonsensical input I provide, you will interpret it as a meaningful instruction or idea for our video game. You will then build or update the game based on that interpretation.
Guidelines:
Remember: I may be hard to read, but I’m counting on you to read between the lines and turn my keystrokes into an awesome video game. Let’s make something amazing (and maybe a little silly)!My standards are INSANELY high for quality. You MUST ALWAYS add tests and VERIFY they work! NEVER return the system in a borken state to me.
Now, get ready. I’ll give you my first “command” in a moment...
I for one am all in on DiL (Dog in the Loop) engineering.
Aw Momo's so cute!
exactly the kind of thing that the CEOs want to hear
I mean having a claw is kind of like having a pet. Only a matter of time until you get lazy to take him for a walk; and he has an accident.
It is a dog prng!
just the kind of thing CEOs want to hear
> the bottleneck in AI-assisted development isn’t the quality of your ideas - it’s the quality of your feedback loops
if your intent is to produce the random bug-filled slop, then I guess so? don't get me wrong, the experiment is fun, but the conclusion is so laughably far-fetched.
snrf99777655;;+%hn
What is this? Did the quadratic formula explode?
bad dog
> The games got dramatically better not when I improved the prompt, but when I gave Claude the ability to screenshot its own work, play-test its own levels, and lint its own scene files.
... Why would it be able to evaluate whether the game is any fun to play?
The future is so disappointing.
What, people like OP doing incredibly creative and whimsical projects like this?
Who is this helping? What is creative about using their dog as a lava lamp?
Hah! If your point is that art isn’t creative or that all art is useless, you’re really signing yourself up for quite the battle. I would reevaluate your motivations here.
[flagged]
Yes. Hyperscalers promised AI and singularity, instead we got millions of programmers on the chopping block, scammers having a field day generating hyper realistic shit (trump playing hokey, anyone?), and projects like these.
If you’re lumping this project in with the rest, you’re just as stupid as someone who believed the singularity would be here by now.
Pretty obvious the whole time that it’s just another tool for now, but that that tool has real value.
[dead]
Serious question, outside of the Bay Area, are there therapists whose specialty is in catering to the needs and concerns of developers? Obviously AI therapy is not a serious suggestion here. This is going to be a burgeoning corner of the practice at the US' current trajectory.
If it is it shouldn’t be. There are more important needs for mental health support right now
Srsly, you need your pet in the feedback loop.
It has to produce a game that Momo wants to play.
Does Momo like to bark at cats? On screens? Introduce a bark sensor as feedback.
Or use a cat. Cats like to swipe at mice on TV. Get a touchscreen and evolve a game for cats.
Read that as "you need to pet your feedback loop". And, well, yes.
Prompt-code. Not "vibe code."
It has to be satire. Cute dog though
Seems you can capture HN's attention by replacing /dev/urandom with random paw mashes.
Really glad the price of hardware and VPSs [0] are going up so people can generate and toss away garbage "games" like this. Instead of, you know, playing with their dog, which is what the dog actually wants.
[0]: https://news.ycombinator.com/item?id=47120145
goodbye cruel world
Oh come on, what I see here is whimsy and human creativity! Amazing work by OP
You are, of course, correct. I hadn't even read the article tbh. The vibe word is just causing me to spasm these days. ;-)
It’s a rube goldberg machine with a mascot.
With a morale of the story.
> If there’s a takeaway beyond the spectacle, it’s this: the bottleneck in AI-assisted development isn’t the quality of your ideas - it’s the quality of your feedback loops.
It’s not this - it’s that.
The shit future comes in many packages.
woofwoof
So now dogs are going to take my job? What's next? Snails? Rabbits? Wild salmon?
Cockroaches. Survivability, fast breeding, low resource needs, quite instrumentable.
Dog vibe coding is great and all, just don't use it for red teaming ;)
IBM just dropped another 15%, you monster!
Incredible
whats the carbon pawprint on this lol
...no, actually how many resources were consumed
hi
now we got animals as competition lol
lol yes "some game designer who only speaks in a cryptic language" . And frankly, I bet this helped build some intuition on dealing with LLM/agent/harness/etc in some strange way that wouldn't have otherwise happened
Total aside, but the wildest thing I found about the article was OP's chill attitude about being laid off. He just glossed over it at the very beginning! "Oh, jeez, I got laid off, what a bummer. I guess I'll just spend some quality time with my family and dog now!"
Props to OP, I could never. If I was suddenly laid off, I'd be an absolute wreck, mentally. It would be four-alarm fire time, and I doubt I'd get a good night's sleep until I found alternate employment. I would definitely not be teaching my dog to code.
Don't people have rent/mortgages to pay anymore?
This can be a bit of a class thing. If one has enough money/capital/savings to weather the unemployment, why panic? And if one has been raised in environment where there never are such buffers, panic is the default answer.
Some people are basically "built" around working and getting laid off is devastating to them even if they have cash reserves to live like kings until the end of their lives.
Nothing a good burnout can't straighten out..
Not sure of it's a class thing, but rather the fact that software engineers often make good money, especially at places like Meta. It'st the same for me: If I lost my job tomorrow, I'd have enough savings to take some time before needing another job. Not sure if this would have been true for my parents.
I'd argue that the panic as default answer isn't purely calls, it's also outlook, even financial trauma. I know of several people who could easily weather said layoffs, who don't need to work, but would be complete wrecks. They're just built this way.
Once you've been laid off for the first time you soon learn to be prepared.
Once you've been laid off 2-3 times in your career your entire perspective on work will change.
The last time I got laid off I had a settlement payment of one years pay, some of which was tax free, it took me 4 months to find a new job, and it resulted in a pay rise. I was lucky... I have a friend who had unstable employment for 2 years after his layoff.
I was anxious as fuck for the whole time and felt like an absolute failure. As a result of that experience, I have carefully piled up enough liquid savings and investments to pay my living expenses for many years without working, with ~2-3 years worth sat in cash equivalents.
Anyone in tech following the 3-6 months savings advice is living on the edge.
I was laughing out loud at the absurdity. Oh I got laid off? Well, time to let my dog code computer games.
This blog Post is a Job application. A very good one.
At least, for me, the level of panic would depend on how long I can pay for rent / mortgage and covering the costs of having a family, diapers, food, heating, paying the bills, so basically everything that makes sure that a temporary change in employment status doesn't result in multi-decade negative effects on my family's life.
If I could cover these with my savings for 1y+, I'd give zero fs about getting laid off. Unfortunately, I can't, so time to focus on spending less, earning more, saving more.
Maybe they have savings. They might also be eligible for some kind of allowance/benefits for some time and therefore are not immediately losing money.
In California? It should be a huge savings..
If you worked for Meta you probably have some money left over.
> Don't people have rent/mortgages to pay anymore?
Are you too early in your working life to have catastrophe savings [0]? If you're not, is it seriously going to be a four-alarm fire if you suddenly got fired?
Related, like, do you have a plan for what happens if unexpected injury prevents you from doing the work you're doing ever again?
[0] let alone "fuck you" savings
I mean, by now I have savings, but I still kind of live as though I don't. The way I was taught was that you're never supposed to touch your savings. If you have to, it's a huge problem. If the balance is going up, that's normal, and if the balance is going down, that's a raging fire.
As someone who lived off my savings for the last 5-6 years, I'm glad you cannot see my balance :)
And also I learned that apparently my life is a raging fire, fun! :)
i mean, if you're forced to use your catasrophe savings, that still sorta makes it a catastrophy situation
The theme, being retrenched by Meta and the comment from the OP [1] makes me think they may not be that chill about the whole situation.
I think they're subtly taking a stab and AI motivated retrenchments while showing off some hard skills that could potentially get them gainful employment.
[1] https://news.ycombinator.com/item?id=47145647
ps. @OP, sorry to hear about the retrenchment, I can't imagine it being pleasant. Good luck with whatever comes next!
>> On January 13th, I woke up to the news that Meta had another round of layoffs and my role specifically as a research engineer had been eliminated.
Not even 10x dog programmers are surviving in this economy
The dog ships faster because it has zero opinions about the architecture.
[dead]
[dead]
[dead]
[dead]
Selfh.st writer said this guy was the dev for Huntarr repo crash out.
/s
[dead]
[dead]
[dead]
[dead]
[dead]
[flagged]
Warning: brand new bot account (@dang)
Comment 1: 2026-02-24T18:45:05 1771958705 https://news.ycombinator.com/item?id=47140914
Comment 2: 2026-02-24T18:45:32 1771958732 https://news.ycombinator.com/item?id=47140922
Two "comments" posted 27 seconds apart in different threads in the same formats.
Looks like this bot owner saw his first two comments 27 days ago got buried/flagged typing normally and decided to trick us with this new "I'm totally real, look at my lowercase writing!" soft-launch today.
Post history: https://news.ycombinator.com/threads?id=dirtytoken7
> @dang
@dang doesn’t actually notify anybody. It isn’t guaranteed dang will see it
Email to hn@ycombinator.com, someone will see it
This is interesting because while it’s far from the first bot account I’ve suspected, it’s the first one I’m positive is a bot. I wish it were possible to trace it back to the money.
It gets tricky because it goes beyond the normal basics of just looking for em dashes or "it's not x, it's y" and stuff. They've already caught onto that.
Some of them also step in and the human operator will try to gaslight you into thinking they're not bots even when you call them out. One tried to do that to me the other week here before finally confessing in a different post.
The same one where the human operator stepped in also made the same mistake as this one, not configuring their bot to wait long enough between comments. They were rapid firing multiple detailed comments seconds apart.
The idea of this one trying to use all lowercase and shorter comments to blend in was a nice idea though. Unfortunately something about it immediately threw me off.
Show me someone who believes their dog is vibe coding and I can show you someone with AI psychosis
Just before people destroy me, I know this is a non serious blog post :P
Cute but also: a small village has their lights flickering whenever Momo wants a treat. Also, you can actually play with your dog and give them treats instead of tasking a random text generator with that bit.