Hacker News Reader

Comment by InvisibleUp | Hacker News Reader

You can draw a line from Xenowhirl's disassembly of Sonic 1 and Simon Wai's research of the Sonic 2 prototype to the creation of Sonic Mania and the blossoming fangame / ROM hacking community, and from there the careers of a lot of indie devs. It's not that these people aren't capable of talking to each other, it's that they get to know each other in a personal and professional context by bonding over a project they have a shared passion over. If these disassemblies just popped up on GitHub one day, and if the fan games and the ROM hacks were created by solo devs with an AI model... well, we end up with quite a few neat projects. But we don't end up with the rich, vibrant communities that get created in the inherent process of collaborating on a project.

Of course, in a post-AI world people will still want to work together on things. But these sorts of tools do disincentivize it. That's the point I'm getting at here.

View on HN

I think people who are doing these decomps aren't necessarily passionate about them to that extent. If they were a huge fan of the history of Sonic, they could use the resources now available to see how the game works, but they would also be internally driven to go beyond just a dump of the game's code on a repo. AI cannot decompile the social history of these games and the stories of the people who made them. It's just a matter if the person actually cares about that. Who knows what drives such an unusual amount of passion.

I feel a lot of these decomps were created by people not as passionate as your examples, because they are only showing up now, not in the pre-AI era. There is a flood of AI decomps only now because AI has lowered the barrier enough to where a project a person could not be passionate enough about to spend years of their limited time on Earth toiling on it is now easily accomplishable. It's a huge ask to put years of your life towards one thing. People might have obligations, work, family, children, more important hobbies, yet still wish one day they could get around to that one project, but couldn't find a way of prioritizing it over other things. I can't fault them for that. Now they can get it done in a month and still have time for other things that matter to them.

The claim is "we lose the rich communities that only exist from manual work." I still think the people who want those communities, like the people who still create art by hand in spite of the existence of diffusion, will keep doing their own thing. They were going to from the start, AI or not, but now that AI is a thing the separation is clearer to people on the outside, since they are seen to not rely on AI.

The communities that will appear because we can now organize AI decomp projects wouldn't have appeared at all with the barrier so high, had it not been easy enough to start them. The blocker was other obligations and not having the time to devote that was intrinsically required to finish them by hand, with the vast majority of people still being a member of outer society.

Dare I say it, maybe not all of the people into reverse engineering really enjoyed the process. It's just they put up with learning it because if you wanted to modernize an abandoned game in the pre-AI era, you had no choice but to learn those things. Only the people with the interest and patience remained, and those are the examples we are able to point to, while the people with only enough passion to get it done with AI assistance left a while ago and are only now reappearing.

I think you may be missing a key point here. These communities spring up, at least in part, because a group of people all want X to exist and building X alone isn't really feasible. This necessity of collaboration can be a major driver of community-building, especially for people who are very technically competent but not particularly outgoing.

If an AI-generated version of X already exists, at least some of the people who might have joined the "let's build X together" community aren't going to sign up anymore. Some of them may feel discouraged because AI got there first, while some may simply feel like the AI version is good enough for their purposes.

You may be right that new and different communities will spring up around AI-generated projects. I'd argue something is still lost, even if something new is gained in the process. I think there's something special about communities of people who are passionate enough to do truly hard things that require dedicating large amounts of time and effort. While AI can help people accomplish far more than they otherwise would've, reducing the amount of human effort and attention that goes into any given task is almost inherent in any AI use.

It is sad that these communities will be lost. Unfortunately, to a lot of people who just wanted the code and some mods, they now are an artifact of their time. Machine code was meant to be read by machines, not humans. The challenge of turning machine code into code that humans can read was a hard problem requiring months or years of effort that was only an unintentional byproduct of the compilation process, not a challenge deliberately laid out by every compiler for humans to gain a sense of accomplishment each time they emitted an object file. That we are creative enough to derive a sense of satisfaction by creating those challenges for ourselves was a happy coincidence always vulnerable to this kind of disruption.

I am not saying that this way is better, but people might start asking themselves when they have a tool sitting right in front of them that they can use to increase their productivity in this domain on the order of 1,000 times or more, if the value of holding together their community overrides the amount of time and maintenance they will have to devote to maintaining it and the project by rejecting AI. Accomplishing the goal they set out for to begin with, having the source code of the game delivered, can be accomplished by any single person now, and that leaves the technically competent communities fragile to breaking up by any one person splitting off, dumping all the code somewhere, and the less passionate leaving one community for another. How many of those people are there for the community itself, and how many are there to just have the code? It will become clearer with anyone being able to split off and use AI to finish things, but I am confident the most passionate people will still remain on the other side. It's just that the motivations of those involved were not as obvious without the distinction between hobby-as-passion and the primarily goal-driven, until now.

Reply on news.ycombinator.com