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

Show HN: I AI-coded a tower defense game and documented the whole process

github.com

Post reply on HN
Print view View original

131–140 of 163 posts

Re: Show HN: I AI-coded a tower defense game and documented the whole process

#131

Post by ukuina »

MattRix wrote: post #108
davidmurdoch wrote: post #104

> AIs like to write a lot of code I vibe coded a greenfield side project last weekend for the first time and I was not prepared for this. It wrote probably 5x more functions than it needed or used, and it absolutely did not trust the type definitions. It added runtime guards for so many random property accesses. I enjoyed watching it go from taking credit for writing new files and changes, and then slowly forgetting…

In my experience Claude Sonnet is much more verbose than Claude Opus, and writes worse code as a result. The difference is pretty striking once you try using them both for the same task.

It generally feels like Opus gives you the 5th or 10th iteration, but Sonnet gives you the first possible solution.

Re: Show HN: I AI-coded a tower defense game and documented the whole process

#133

Post by kenjackson »

unclebucknasty wrote: post #128
twalkz wrote: post #30

I'm really enjoying reading over the prompts used for development: ( https://github.com/maciej-trebacz/tower-of-time-game/blob/ma... ) A lot of posts about "vibe coding success stories" would have you believe that with the right mix of MCPs, some complex claude code orchestration flow that uses 20 agents in parallel, and a bunch of LLM-generated rules files you can one-shot a game like this with the prompt "create a…

> a strong and thorough idea of what you want, broken up into hundreds of smaller problems, with specific architectural steers on the really critical pieces. Serious question: at what point is it easier to just write the code?

Depends. If you have written other Tower Defense games then it’s probably really close to that line. If you just took a CS class in high school then this vibe approach is probably 20x faster.

My aunt would always tell me that making fresh pasta or grounding your own meat was basically just as fast as buying it. And while it may have have been true for her it definitely wasn’t for me.

Re: Show HN: I AI-coded a tower defense game and documented the whole process

#134

Post by unclebucknasty »

skydhash wrote: post #129
unclebucknasty wrote: post #128

Earlier quoted context omitted.

> a strong and thorough idea of what you want, broken up into hundreds of smaller problems, with specific architectural steers on the really critical pieces. Serious question: at what point is it easier to just write the code?

And if it's a work project, you're going to spend a few years working on the same tech. So by the time you're done, there's going to be templates, snippets,... that you can quickly reuse for any prototyping with the tech. You would be faster by the fact that you know that it's correct and you don't have to review it. Helps greatly with mental load. I remember initializing a project in React by lifting whole modules o…

All of this, and highlighting this part:

>You would be faster by the fact that you know that it's correct and you don't have to review it. Helps greatly with mental load.

I keep thinking maybe it's me who's just not getting the vibe coding hype. Or maybe my writing vs reading code efficiency is skewed towards writing more than most people's. Because the idea of validating and fixing code vs just writing it doesn't feel efficient or quality-oriented.

Then, there's the idea that it will suddenly break code that previously worked.

Overall, I keep hearing people advocating for providing the AI more details, new approaches/processes/etc. to try to get the right output. It makes me wonder if things might be coming full circle. I mean, there has to be some point where it's better to just write the code and be done with it.

Re: Show HN: I AI-coded a tower defense game and documented the whole process

#135

Post by Brajeshwar »

mordechai9000 wrote: post #92
Brajeshwar wrote: post #24

Yours is beautiful; the code is too. I'm sure you had a lot more hand than just using AI. I stopped coding a long time ago. Recently, after a few friends insisted on trying out AI-Assistance codes and I tinkered. And all I came up was a Bubble Wrap popper, and a silencer. :-) https://bubble-pop.oinam.com https://void.oinam.com

Running chrome on android and the bubbles aren't popping for me. Count is staying at zero. Are you open to PRs? :-)

Yes Please. Absolutely. Feel free to send in any edits. https://github.com/oinam/bubble-pop

Re: Show HN: I AI-coded a tower defense game and documented the whole process

#136

Post by freehorse »

M4v3R wrote: post #25
freehorse wrote: post #22

A bug in the intro: in the first round in the first playthrough my turret destroyed one of the critter and the other reached the tower. There was no other prompt or anything happening in the game after that and had to restart. The next time, the turret did not destroy any critter, the prompt to use backspace appeared and the game progressed normally.

Interesting, I actually had the turret destroy one of the enemies several time but it didn't prevent the tutorial message from showing up. I'll look into it though, thanks for the report!

Yeah I cannot reproduce it in subsequent runs even if this happens, no idea what happened in that.

Re: Show HN: I AI-coded a tower defense game and documented the whole process

#137

Post by aprilthird2021 »

AnotherGoodName wrote: post #86
aprilthird2021 wrote: post #79

Earlier quoted context omitted.

Would that have taken you weeks though? I imagine reading through a few articles and examples could have gotten you there. I never heard of Manim before but found these pretty quickly: https://docs.manim.community/en/stable/examples.html https://manimclass.com/plot-a-function-in-manim/ I am not trying to pick at you, but it feels like what I am currently able to do with AI, shave off a few hours, but not weeks. I agr…

A single graph might save hours. A full feature series where each graph type has yet new syntax to learn is indeed much more. Especially when there's followups, "let's make the graph move over the left half of the screen and then the next animation shows in the right half?" which again were one shot done in minutes with AI. For me just to gain the context of how to move the animation into the left half smoothly and t…

> I stand strongly by my assertion that it saved me a week (at least!) all up.

Fair enough. More power to you then. I'll keep looking for some other examples. Thanks for sharing!

Re: Show HN: I AI-coded a tower defense game and documented the whole process

#138

Post by marviel »

twalkz wrote: post #30

I'm really enjoying reading over the prompts used for development: ( https://github.com/maciej-trebacz/tower-of-time-game/blob/ma... ) A lot of posts about "vibe coding success stories" would have you believe that with the right mix of MCPs, some complex claude code orchestration flow that uses 20 agents in parallel, and a bunch of LLM-generated rules files you can one-shot a game like this with the prompt "create a…

I totally agree!

this is the idea behind my recent post actually[1] where I recommend people use AI to write specs before they code. If all you have to do is a human is edit the spec, not write it from scratch, you're more likely to actually make one.

[1] https://lukebechtel.com/blog/vibe-speccing

Re: Show HN: I AI-coded a tower defense game and documented the whole process

#139

Post by blks »

AndrewKemendo wrote: post #94
johnrob wrote: post #76

Earlier quoted context omitted.

I see 100x used quite a bit related to LLM productivity. It seems extreme because it implies one could generate a year’s worth of value in a few days. I would think delivering features involves too much non coding work for this to be possible.

But that’s precisely what I’m saying is that what I can do today by myself in a couple of days would have taken me a year with a team of three people The key limiting factor to any project as somebody else in this thread said was “people alignment are the number one hindrance in project speed” So 10 years ago if I wanted to make a web application that does complex shit I’d have to go and hire a handful of experts hav…

Sorry, I call bs, unless you were very poor developer without any skills to manage people.

Re: Show HN: I AI-coded a tower defense game and documented the whole process

#140

Post by blks »

AnotherGoodName wrote: post #14
mgdev wrote: post #6

This is awesome. I've been in software for 20+ years now as well. One thing I've noticed is many (most?) people in our cohort are very skeptical of AI coding (or simply aren't paying attention). I recently developed a large-ish app (~34k SLOC) primarily using AI. My impression is the leverage you get out of it is exponentially proportional to the quality of your instructions, the structure of your interactions, and t…

I've come to this same conclusion pretty strongly in the past few months in particular. I actually had negative comments on my experience with AI previously. For all the talk of AI hitting a ceiling the latest tools have improved greatly. I'm literally doing things in hours that'd previously take weeks with little issue. I do of course have to think about the prompts and break it down to a fine grained level and i al…

For me the LLMs are failing laughably to be able to generate some examples of any actively developing library, even for simple stuff. They invent non-existing APIs, use years old functions that are long deprecated, et cetera.
Post reply on HN