AI Is Decompiling Every Game
Summary
The Standup hosts discuss the rapid rise of AI-assisted game decompilation and the resulting backlash from the retro preservation community. Casey Muratori argues that while AI can speed up reverse engineering, the resulting code is often "slop" that lacks the value of original machine code for true preservation or learning. The group also analyzes the legal risks of cloning proprietary software like Photoshop, noting that clean-room implementations do not protect against copyright infringement. Finally, they debate whether these mashups represent a genuine threat to the industry or merely …
Topics discussed
Intro: Casey's 'Legendary' moment and moving to Philly
Hockey, Olympics, and the whiteboard marker story
Overview: AI game mashups and modding potential
Technical barriers: Hardware, source code, and modding
Sponsor: Sentry and gaslighting AI agents
Nostalgia vs. novelty: Will mashups impact the industry?
Physics engine mapping vs. full decompilation mods
Adobe suite clones and the 'clean room' concept
GPU command tracing and manual reverse engineering
AI decompilation quality: 'AI slop' and preservation
Preservation value: Machine code vs. readable source
Learning from decompiles and the need for human cleanup
Consumer indifference to code quality and frame rates
Legal risks: Copyright infringement in AI clones
Market dynamics: Adobe revenue vs. game piracy
Industry knowledge sharing and standardized engines
Monopoly concerns and the economics of free software
Court competence and the future of software law
Fan wishlist: Specific game mashup ideas
Anti-cheat systems and AI-assisted bypasses
Linux gaming, kernel anti-cheat, and hardware costs
Final mashups: Pokemon bowling and Harry Potter
Listen ad-free on Castria