Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

I'm helping my dog vibe code games

calebleak.com

Post reply on HN
Print view View original

371–380 of 404 posts

Re: I'm helping my dog vibe code games

#371

Post by advael »

krupan wrote: post #370
advael wrote: post #368

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 th…

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

Re: I'm helping my dog vibe code games

#372

Post by mikkupikku »

DrewADesign wrote: post #362
mikkupikku wrote: post #349

Earlier quoted context omitted.

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 i…

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 f…

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.

Re: I'm helping my dog vibe code games

#373

Post by krupan »

wazHFsRy wrote: post #235
with wrote: post #37

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.

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

Re: I'm helping my dog vibe code games

#374

Post by sillyfluke »

cleak wrote: post #179
cs702 wrote: post #81

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 comme…

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.

Re: I'm helping my dog vibe code games

#375

Post by brunooliv »

notatoad wrote: post #173

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

Re: I'm helping my dog vibe code games

#376

Post by cleak »

MrGreenTea wrote: post #247
singron wrote: post #222

Earlier quoted context omitted.

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…

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.

Re: I'm helping my dog vibe code games

#377

Post by Forgeties79 »

groestl wrote: post #346
jraph wrote: post #253

Earlier quoted context omitted.

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.

This whole discussion is wild to me. Comparing people and machines like this is not productive. It is not actually answering a serious question.

Re: I'm helping my dog vibe code games

#378

Post by cleak »

singron wrote: post #222
cyber_kinetist wrote: post #212

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 e…

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…

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.

Re: I'm helping my dog vibe code games

#379

Post by cleak »

bdashdash wrote: post #285
cyber_kinetist wrote: post #212

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 e…

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.

Re: I'm helping my dog vibe code games

#380

Post by cleak »

cezart wrote: post #127
ajspig1 wrote: post #54

Earlier quoted context omitted.

+ 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.
Post reply on HN