Emulator dev rejects AI slop code
One of the most exciting aspects of the open-source retro development scene is the sheer creativity unleashed by projects like the Analogue Pocket and its support for openFPGA, which allows a wide array of homebrew emulators to thrive. However, even in this space of passionate innovation, developers are increasingly facing a thorny problem: managing low-effort, often incoherent code contributions that seem to prioritize trend over technical substance.
The frustration often stems from contributions that feel less like helpful code and more like carelessly generated output. For example, the developers behind the SNES emulator core for the Analogue Pocket have found themselves dealing with submissions that cause significant headaches.
Recently, a contributor named nasonliu offered an “experimental MSU-1 streaming” feature for the Pocket SNES core. While the idea itself might seem novel, the implementation came packaged as nearly 50,000 lines of code generated entirely by a large language model.
The core developer, agg23, who ported the SNES FPGA core to the hardware, was left with an overwhelming task: reviewing this massive block of machine-generated code. The central dilemma was clear: the contributor expected the reviewer to simply accept the code, but the reality was a deluge of unrelated material that required manual sorting and cleanup.
“You sent me a nearly 50k line diff, unasked for, completely LLM built, and you somehow expect me to review it,” agg23 noted, expressing clear exasperation. He pointed out that there was “SO MUCH unrelated garbage” within the submission.
This experience highlights a wider community tension. While many coders embrace artificial intelligence as a powerful tool to boost efficiency, there is a growing divide between those who see AI as an accelerator and those who view it as a shortcut that bypasses essential understanding. In fan communities creating emulators and novel ways to play classic games, experienced developers feel bombarded by code from those who do not fully grasp the underlying mechanics of the system they are attempting to modify.
This sentiment is not unique. Developers working on other major projects, such as the PS3 emulator RPCS3, have also voiced similar concerns. They have urged contributors to adopt a more rigorous approach, demanding that developers understand the code they are modifying from top to bottom before submitting it. The plea has been to “stop submitting AI slop code” and prioritize contributing genuinely useful, well-understood work to the community.
The takeaway is clear: while AI can generate code, it cannot replace the essential human skill of debugging, understanding context, and respecting the time and brainpower of maintainers. For the open-source community to continue innovating and expanding classic gaming experiences, the focus must shift from sheer volume to genuine quality and thoughtful contribution.
Buy on Amazon
Credit
