Compiler output is highly predictable, but it doesn't make the inverse true. LLMs are not themselves compilers.
Now the games industry will try it again for real and find a way to rebrand cloud gaming and trick everyone to believe its better than running games on your own machine, only for them to stream the game back to you without you ever owning it.
Would it have gotten traction in the future? Don't know, but with Google it was certainly never happening.
"X license disallows this current use without express permission"
"Z gave me permission , here's the pdf, they gave us permission regardless of what licensing says going forward."
<insert pdf generated from a text file that says "I'm Z and I gave permission.">
"Okay, we will continue forward while noting that Z gave us permission regardless of what licensing says going forward."
What's more interesting is the new ability to easily create ports of old games that let you play old games legally: Bringing your copy of the original's data files and playing on a modernised engine with high-resolution, unlocked FPS, bugfixes, etc. In the past couple of months, I've seen a bunch of vibe-coded ports of DOS games for which the source code was lost.
If you've noticed the obscene amount of ports, remakes, and remasters (most of which are done terribly) in recent years, it's obvious why certain people are pushing this as IP infringement. Companies were loving the fact that they could pull from a back-catalog of their PS2 games, get it compiling again, do some minor fixes to support modern widescreen/controllers/etc, maybe do some minor graphical enhancements, then resell a 20-year-old game for $60. Far less work than a new title, and far less risk since it has a built-in fanbase.
It's exploitative and shitty, to be clear, but it's understandable why some people are very upset about this.
All of this power and still there's nothing new I want to play.
I'm tired of these terribly made AI projects in this space. It's just noise and low quality spam. I miss when these were done legitimately.
Some script kiddie ran this through a chat bot once and probably doesn't even understand the code.
For those who don't understand, watch this segment of DF Retro Super Show - https://www.youtube.com/watch?v=A6gJTdk_BkI&t=165s
The point of decomps/recomps was never for some altruistic reason related to skill or art.
The point was people wanting to play games. This just adds to that. Yes - the people who did it before were skilled, smart and deserve recognition. But these ports are exactly that - ports so that people can play their old games or relive them.
I don't know where my quake cds are. Even if I found them I'd have to figure out what mods to run to have it feel good. Then I'd have to find communities to play with.
This is super cool to me.
Probably will consume gazillion compute but now seems like sky is the limit, on a good compute day at least
By the time we get there, AI can probably be able to recreate the game based on the video alone. I think we are already there, the fidelity of it might vary.
What will probably happen is that frontier models will lobotomize models to refuse reverse engineering. That will buy some time, however with open weight models, some will find the way to fine tune the models to support RE if magically everybody agrees on the consensus that RE should be blocked.
moritzwarhier•1h ago
One turn involved ChatGPT explicitly fixing C++ source code to fix beat detection for the available audio sources after I complained that it was broken.
Hands down, sure, the most respectable work, apart from the original AVS code, was done by the Emscripten contributors. Anyway, it readjusted my expectations, even after my day job switched to agentic coding as the primary development tool.