Stop Slopware

In practice we've been doing setting the threshold of what we care about for a long time already, with firmware, operating systems, libraries, etc. You can always go one level deeper if you want more. Did you want to know how readline works? What about terminal control characters? What about pty? What about the input? The keyboard interface? HID? Device drivers? USB? Packet transmission? Serial interfaces? Electrical connections? Signal integrity?

LLM code provides just another type of available abstraction where we can stop in learning, but not really something entirely new.

I've been using Gemini / Antigravity to make a virtual pet for my kids using Love2D / lua. I never coded in my life and have no intention of learning (but I've been learning a lot about game systems and logic, which has been fun). The game is coming along very well and it looks very pretty (I'm a professional illustrator) and if I decide to publish it, no one will care the code is made in AI. It's very high effort, to be honest. You'd had to look into the code to know AI made it.

I now get why so many people are making AI art. I see their "work" as an illustrator and it is absolute slop, but I can see now how it might be fun and even liberating for people who don't make a living with it. So I now think twice before calling AI art "slop". Sure, it may be slop, but it's making a lot of people happy and probably opening up new carreer paths for people.

And yes, I've been affected financially because of this... but I get it.

I used to be intrigued by _software craftsmanship_, but I mainly saw people using it to practice heavy gatekeeping. Be it on the job, in code reviews, in job interviews...

Furthermore, given how they behave in a cult-like way, it feels like they are straight-up delusional.

People working as consultants for big retail chains, talking all day long about "the craft". Nobody cares. They sell trash. They don't put marble in their store. They don't want fancy software. Furthermore, if, by trying to force "the craft" to their peers, all they do is making the life of others miserable... Just stop. Please.

Now my approach is as follow:

if stakeholders are only interested in two things (how-much it cost, when it's ready), which is 99% of the case at $JOB, then make something that does the job and that won't make you hate yourself if you have to maintain it in a year

if I'm the stakeholder, like creating internal tooling that nobody asked for but that will solve issues, then yes, I do things as good as I want them to be

same for working on FOSS on my personal time

Your point also touches on the idea that new things are being created that might well never have. Like your virtual pet. You might have been commissioned to illustrate such a thing but most likely not, and it wouldn’t have been “yours.” It reminds me of when desktop publishing, MIDI sequencers, or PowerPoint took off and people produced all sorts of things that were largely not of high artisanal quality but it was new stuff, people got personal value from it (as it was harder to spread stuff around pre-Internet) and the tools eventually matured into what all the pros now use anyway.

That said, I concede the critics have many valid points and concerns and it’s going to be interesting to see how we navigate this flood of “stuff” at a scale never seen before. (I suspect it’ll end up like YouTube and video. Ultra long tail. Most stuff never seeing more than a few eyeballs and a smaller group getting the lion’s share of attention, as with most things. Did YouTube change TV and video production more generally? Yes! But it also didn’t destroy it..)

I don't think we should shame beginners for using AI. Instead, we should encourage them to try working manually and/or learn fundamentals.

Also, public communities have been overrun with low-quality beginner posts for longer than AI. Moreover, you can't convince all beginners to stop using AI, because some are malicious and/or socially inept. Moreso than shame or encouragement, we need a filter for low-quality projects (perhaps a combination of AI flagging and manual review). That would benefit both the beginners and the people who shame them because they're bothered.