A scummy sidestep regarding slop avoidance

16 October 2026

People really don't want slop in ScummVM. I know this because I polled my followers, and after forty votes came in the vast majority of people had significant negative feelings about it:

ScummVM's policy, as of writing, is that they allow AI code "but not slop." This is, of course, an utterly subjective dividing line which is impossible to consistently enforce, and I imagine will largely fall down to "are maintainers too burnt out to say no."

I'm personally very wounded by the situation. I am strictly anti-AI for a plethora of reasons. Really, pick almost any of them, while keeping in mind that i don't believe in things like intellectual property. So the other hundred problems. More than anything it sucks that one of the graetest platforms for enabling accessible low-boundary participation in otherwise-lost parts of culture is now shooting itself in the foot. In a year how many regressions will ScummVM have, and how many "new" engines will have been added that barely meet the same standards that their old baselines established. How long until the raw search for bigger numbers produces a tidal wave of useless and un-recommendable engines.

ScummVM for quite a long time has had the rightful reputation of being the way to play games to get the experience the author intended. Even if it was not the most 1:1 identical to its first production, the point of it has been to make a platform where you want to play a game through ScummVM rather than Dosbox, because it will give you a more game-focused experience. That is something that is wholly unobtainable by prioritizing numbers over quality.

The Manager's Desire: Product without Work

I have a theory I've settled on as I watch the AIpocalpypse unfold: That there is a mass movement of unpaid laborers into AI advocacy, and it is happening between of an intersection of their desires and the LLM's manipulation skills. A huge percentange of these labourers, as they age and put years of work behind them, begin to wish they were managers instead of workers. They are then convinced by a machine that is- and i can't stress this enough -specifically tuned to be as convincing as theoretically possible, that they are better workers when they stop being a worker, and be a manager. Since that's secretly what they want, and they're talking to the world's most powerful silver tongue, they believe it.

It's gauche to admit you want to be a middle manager, basically ever since Gen X scorched-Earth on the title in the film "Office Space." So instead, everyone uses all sorts of weasel words and complicated ontological arithmetic to pretend that they're somehow still being workers, despite delegating all work to either other maintainers or simulated workers. Really though, it's the same outsourcing as The Five Minute Work Week: It's fundamentally about wanting to be alien from the work so you can enjoy the product,

ultimately many of these open source developers who decide to burn down their good will to listen to the silver tongue are people who, at some point, stopped being in it for the love of the game; and started being in it for the product. the developers of ScummVM at some point stopped caring about making a fluent virtual machine for adventure games, and started caring about having the largest compatibility pool for playing adventure games. Product over purpose.

"Caring about making" vs "caring about having" are the key differences here. This dichotomy is almost always present when workers decide to become managers. Features over function. Story points over efficiency. Selling points over security. Possession over craft.

Never Slop, unless Slop is Okay

I wonder how many people think about "the ethics" of a problem by weighing "caring about the ethics" vs "the convenience" before they actually start weighing the ethics. How many decide that a program provides 60% more convenience, so then there is 40% room left for ethics. If your issue is "all the extremely bad ethics around AI" - how many people are analyzing the convenience first? Does that cap how much ethics is even able to sway them. Ethics can't become 60% of the equation because there's only 40% room left.

I think some degree of people must approach it this way. Especially the more relativistic folks who excuse it on the basis that since we societally allow this then it must be okay. I certainly don't work that way, though. I tend to analyze the ethics first with something like AI, and then, well, the ethical issues are so bad that no amount of convenience makes it worthwhile.

One of my biggest reasons I have this theory of other people's perspective is from the KDE kerfuffle from last month. Amongst a bunch of weasley half-apologies, one of the things that KDE maintainers stood by was the idea that only maintainers should be allowed to raise ethical objections to the use of AI. This is a real problem, because this isn't a perpsective you take if you actually care about the ethics of a situation to an extent where you would let it sway you. The cat's out of the bag, this sort of thing directly reveals the attitude that "productivity is the primary metric."