This was a nice game. It would be cool if you added some more purpose to landing the plane, like if the fuel and throttle were limited, making it actually useful to land it. Overall great game and enjoyable. Also would be nice to get some extra points for doing loop-de-loops or tricks with the plane. A shield power-up is also a great idea.
Loved it. That's exactly what AI should be used for. Something that would take us one month at least, full of bugs, AI now delivers in less than a day, ready to ship
This is the essence, AI is a super power, the world is ours
Well I dunno, man, this looks like the sort of amusing prototype I'd expect to take a couple of days - though not to understate the value of getting it in 4 hours (and probably with less than 4 hours' continuous effort from the human). Iteration time is an important factor!
(Regarding the parting shot, I find myself once again, for the numerousth time, reminded of this curious lobsters comment that I'm having difficulty forgetting: https://lobste.rs/s/omq8rt/vibecoding_gets_emacs_patch_rejec... - perhaps not quite entirely directly germane here, I don't think, but, still, somehow, not quite entirely directly not.)
>This is the essence, AI is a super power, the world is ours
Don't me overdramatic. The guy who made Flappy Bird did it in a weekend without AI. A simple game like this wouldn't be difficult in any modern framework. That it can be vibe-coded in less than a day isn't really much value.
Nice job - it kind of reminds me of the rocket barrel levels in Donkey Kong Country: Tropical Freeze.
Feedback: Consider playing with the physics of the plane depending on your pitch, so that some parts of some levels might require you to deliberately pitch down for a few seconds to build up some "stored speed".
It basically does this. If you are in level flight or a dive, you increase speed up to a certain max (level flight max is less than dive max). If you climb, you bleed off speed. To reach high balloons, you often need to build up more speed before you can climb high enough.
There's probably room for improvement, but it's already enough fun that I can play it for several minutes at a time.
Now I'm trying to decide if there's anything else of value to add - without making it too much of anything. I think I'm about finished now.
The fake scanlines and position of the UI elements make them difficult to read. The plane is too slow and its turn radius is too tight, the physics (if there are any) need to be improved to give a sense of weight and velocity to the plane IMO, and that probably necessitates a bigger playfield. Also it feels like it really needs a gun (edit: OK it does.)
Overall it feels like an OK prototype but I agree it needs more polish.
That said, HN is traditionally for sharing code with the community. Maybe for AI submissions people should share their prompts? I don't know. I'd ask to see the code but since you didn't write it you can't really discuss it.
It took me less than 10 seconds into gameplay to pick up bullets for the (already existing) gun. Your comment reads as if you didn't try to play the game for more than a few seconds.
It's a fun little game; no need to be so smugly dismissive.
>Your comment reads as if you didn't try to play the game for more than a few seconds.
I made it a few minutes, I admit it was boring. That isn't my fault. In a better designed games these things would be more obvious.
edit: OK I tried it again, got the gun, it autofires which still isn't fun. If I actually got to control it and had less ammo it would be more challenging. If that's a possible mechanic I couldn't see how to activate it. Still boring.
Yeah, this feels like a vibe-coded game that people who played it for 1 minute are upvoting, but it just really lacks the feel of how even a mildly solid and entertaining game feels. I got bored very quickly.
I flagged it because of HN's AI content policy. I know it technically doesn't apply to submissions, but I think game submissions (or any others) should show more humanity, instead of being like "look at this momentarily pleasing thing Gemini coded for me".
I think it might be interesting to know age ranges for posters, because this made me smile as it reminded me of hours spent trying to perfect Barnstorming by Activision on my Atari, right down to the scan lines and flight dynamics. Not sure it’s supposed to be an actual flight simulator, so your point (if there was one) just seems like being a prick for no particular reason. Did skin come off your nose while playing?
I'm giving my honest impressions from playing the game and suggesting ways that I think it could be improved. If I ever posted anything here I'd expect the same treatment.
Literature is written by human beings, and if the author was in the class I expect they would be able to discuss their process and methodologies.
But the entire point of using an AI to generate code is not to pay attention to any of that. The author of AI generated code did not write said code and cannot therefore provide any insight into the process or methodology beyond "here is the prompt and model I used." Nothing that would be of value to the few of us left on this site who actually care about such things. Nothing, in other words, to gratify intellectual curiosity.
Also regarding scanlines, fake scanlines shouldn't obscure UI elements or make them difficult to read. Actual scanlines on actual CRTs didn't do that.
I agree with you the game is mediorce and feels like a single prompt AI generated game (whether or not it actually is)
But, reguarding AI code ... a movie director directs others (actors, camera people, editors, set builders, ...) to make a movie. They don't do those things themselves. Yet we still ask the director about the movie they directed. I see AI as similar. I direct it. To me there is still value in knowing why a director made certain decisions and their process for getting the result they wanted.
Because getting slop is easy and slightly novel ATM, people post it and it's crap, but I believe some people will use AI as the tool it is and put in enough effort *directing it* to get good results and be worth respect.
>Most literature classes discuss works where the author is unavailable.
This is not actually a literature class, and the author, as such, is available. And the purpose of Show HN is to have a conversation with the author.
>The point of AI is not to generate code and not pay attention…
Yes it is. That's what "vibe coding" is. That's what all AI workflows converge towards. That's how most people use it. And that's what makes discussing AI generated posts so limiting.
Nobody's generating this in four hours and understanding anything, but if I'm wrong I welcome the author's replies. It would make more interesting conversation than usual for these threads, which I guess are just supposed to be for kudos now.
It is fun. Now make it multiplayer over websockets/etc with some dev controlled signalling server (who peers register with/etc) and STUN check to get peer IPs locally.
Great game and this reminds me of it. Thanks for sharing.
This is the essence, AI is a super power, the world is ours
(Regarding the parting shot, I find myself once again, for the numerousth time, reminded of this curious lobsters comment that I'm having difficulty forgetting: https://lobste.rs/s/omq8rt/vibecoding_gets_emacs_patch_rejec... - perhaps not quite entirely directly germane here, I don't think, but, still, somehow, not quite entirely directly not.)
Every step in this argument (particularly this one) seems tenuous at best, and in need of citations.
Don't me overdramatic. The guy who made Flappy Bird did it in a weekend without AI. A simple game like this wouldn't be difficult in any modern framework. That it can be vibe-coded in less than a day isn't really much value.
Feedback: Consider playing with the physics of the plane depending on your pitch, so that some parts of some levels might require you to deliberately pitch down for a few seconds to build up some "stored speed".
There's probably room for improvement, but it's already enough fun that I can play it for several minutes at a time.
Now I'm trying to decide if there's anything else of value to add - without making it too much of anything. I think I'm about finished now.
Overall it feels like an OK prototype but I agree it needs more polish.
That said, HN is traditionally for sharing code with the community. Maybe for AI submissions people should share their prompts? I don't know. I'd ask to see the code but since you didn't write it you can't really discuss it.
It took me less than 10 seconds into gameplay to pick up bullets for the (already existing) gun. Your comment reads as if you didn't try to play the game for more than a few seconds.
It's a fun little game; no need to be so smugly dismissive.
I made it a few minutes, I admit it was boring. That isn't my fault. In a better designed games these things would be more obvious.
edit: OK I tried it again, got the gun, it autofires which still isn't fun. If I actually got to control it and had less ammo it would be more challenging. If that's a possible mechanic I couldn't see how to activate it. Still boring.
I flagged it because of HN's AI content policy. I know it technically doesn't apply to submissions, but I think game submissions (or any others) should show more humanity, instead of being like "look at this momentarily pleasing thing Gemini coded for me".
Brusque but not inaccurate, Welcome to our future I guess.
Huh? Why do you believe this? ever take a literature class?
Regarding scanlines, there is a setting to turn them off.
But the entire point of using an AI to generate code is not to pay attention to any of that. The author of AI generated code did not write said code and cannot therefore provide any insight into the process or methodology beyond "here is the prompt and model I used." Nothing that would be of value to the few of us left on this site who actually care about such things. Nothing, in other words, to gratify intellectual curiosity.
Also regarding scanlines, fake scanlines shouldn't obscure UI elements or make them difficult to read. Actual scanlines on actual CRTs didn't do that.
But, reguarding AI code ... a movie director directs others (actors, camera people, editors, set builders, ...) to make a movie. They don't do those things themselves. Yet we still ask the director about the movie they directed. I see AI as similar. I direct it. To me there is still value in knowing why a director made certain decisions and their process for getting the result they wanted.
Because getting slop is easy and slightly novel ATM, people post it and it's crap, but I believe some people will use AI as the tool it is and put in enough effort *directing it* to get good results and be worth respect.
The point of AI is not to generate code and not pay attention…
This is not actually a literature class, and the author, as such, is available. And the purpose of Show HN is to have a conversation with the author.
>The point of AI is not to generate code and not pay attention…
Yes it is. That's what "vibe coding" is. That's what all AI workflows converge towards. That's how most people use it. And that's what makes discussing AI generated posts so limiting.
> My concept, my guidance, and Gemini Flash doing the work. Took about 4 hours and 50+ conversations with Gemini.
You are right though. It is exhausting since some people just can’t get it through their head that you can use AI and still understand the output.