As someone who has spent an inordinate time researching prompt engineering technique for producing more interesting images, it's a good thing the base case with an underspecified prompt is boring as it gives the model more malleability for more interesting prompts.
One of the reasons image AI slop has been so prevalent is that the slop is the base case (as derived from tons of RLHF because the median user has no taste) and it is becoming increasingly difficult to steer those models even when the prompt is more specified.
Per-token Inference is likely profitable (the news was before recent price cuts that may make the numbers less valid), but subscriptions if fully utilized are 100x value, and the profit margins aren't that high.
The entire reason you're seeing a lot of more ambitious reverse-engineering/decomp projects as of late is due to Claude Opus 5.5, which is cheaper than models have been over the past year, in addition to being smarter.
The backlash against vibecoded decomps and using a lot of compute for unimpactful projects was met with the counterargument "why can't you vibecode something useful like Photoshop instead?"
No, the backlash was because vibecoded decomps were trash, incomplete, and quickly abandoned once the attention rush wore off. You'll never guess what's happening to these!
Just recently there was a massive kerfuffle in the Minecraft modding community because a mod called Thaumcraft was recreated after it was effectively abandoned for 7 years. The mod's management team threatened copyright legal action.
It's extremely ironic seeing the modding community, which generally play fast and loose with IP rights, all of a sudden care about them when AI made modding trivial.
You can see some of the community responses here, it's hilarious.
It's basically "you can't make a mod inspired by the other mod we're hugging the rights for"
Completely missing the part they didn't ask for permission to mod the base game either..... and that Thaumcraft (the original mod) had assets consisting mostly of random images copied off google images. Yes you can't make that up. "You can't steal what I've rightfully stolen!"
I am actually quite curious if the original Thaumcraft mod author Azanor is on HN. I know a lot of the original generation of Minecraft modders (pre twitch, mostly YouTube days, Minecraft modding changed a lot post twitch) are fairly senior software engineers (most famous one is probably Sethbling, though he didn't really do modding).
There's a "Direwolf20" on hacker news too but that user has nothing to do with the famous Minecraft modder afaik.
I mean, the REAL backlash they got was "So, are you going to abandon this on in 72 hours too?" at which point, once again, without fail, they do.
You'll find drama for decomps of games with active projects. Which is why the DK64 styled all over the AI Slop. You find zero drama for those with not. You may find some with translation projects, but most people already figured out you can just give them 0 attention, there are live translation tools (that you can run on dual screen devices even).
And then there's morons like the PSO on the browser who tried adding merch links and got C&D'd within 12 hours.
I'm sorry but isn't this the whole point of free and open source software? We have been saying this for decades.
Maintainers, you don't owe anyone anything. Put your own life and mental health first. Things like we specifically disclaim all liability and that this software is available free of cost with no assumption of ongoing support etc.
OP isn't just making a straw man argument, they're making the entire straw family.
> Those defending the poor performance of vibe coded apps have parroted the famous lines AI guys used in the past - “It is only days old” and “this is the worst it gets”. They make the mistake of equating human app development time with generative AI development time (hundreds of times faster). A vibe coded app is, in human hours, not only a few days old but closer to two or three years old. If a brush engine or transform tool works badly after the human equivalent of three years of development that is a demonstration of bad planning or an AI model not doing a good job.
I legitimately cannot parse the logic behind this argument.
It’s really badly put but I thiiiink they’re trying to say that due to the pace you can work with AI, you can have something technically complex but badly engineered quickly. Like three years of bad engineering in an afternoon? Or something? It’s still bad as even three months of bad engineering would normally contain far more foresight than a vibe coded weekend nonsense project
I see it like this: because greenfield coding is strong with AI, but improving/large rewriting an existing codebase is getting progressively harder and worse, if you don't get the architecture right in the first try, improve them in later rounds will be very hard,if not futile. Thus the poor performance is not a typical day 1 problem as with human coding, but a pestilence.
The article is strangely supportive of Adobe, far beyond a normal person's sense of ridicule.
Software development is about revenue, but specifically Adobe's revenue:
> Automated or semi-automated cloning to steal revenue away from developers
Pay us, don't innovate:
> They’re spending minimum $2K to vibe code a Photoshop competitor when Affinity Photo is free and Photoshop costs $200 a year. Not very smart economics.
> Commercial apps innovate and lead. Open source apps like Blender follow the lead. Innovation costs a lot of money.
> Instead of hurling hate speech and bile at thousands of software engineers who lead innovation at large companies (Adobe, Autodesk, Maxxon, Celsys, and others), show them some gratitude for their efforts. They are the engineers who innovate and everyone else follows. Just because they work at a big company doesn't mean they are committing human rights abuses. We have a generation brainwashed by envy and hate instead of urging everyone to be greater and bigger.
Why do people hate Adobe more than other companies? What's the difference? Only sensitive insiders would pretend to ignore decades of enshittification.
> Clip Paint Studio, Maya, 3ds Max and Cinema 4D, btw, are all subscription applications. But because those apps aren’t in demand by pirates Autodesk has not experienced the hate and bullying campaigns that Adobe has received.
People noticed the "underage children", but defending Adobe customer support is even more ridiculous:
> And when it comes to customer support things become even sadder looking. One or two guys in a Discord chatroom cannot support a global user base, especially when those two guys are hunting for underage children to speak to on Discord (yes, I have seen what vibe coders do).
That’s not my experience at all, as a European who frequently travels to the US for work. If anything, I’d say that restaurants in the US are slightly more expensive without tips, definitely more expensive with.
> I’d say that restaurants in the US are slightly more expensive without tips
That's entirely possible, but then again people make far more in the U.S., so the cost of living is generally higher. Taking that into account, I don't think the cost of eating out is higher than in Europe.
My experience is that European restaurants are cheaper and better. Especially in 2026 with the Trump inflation hellscape. Everything from McDonalds to local high end restaurants.
Certain commodities are even more shocking. You’ll get the best loaf of bread that you ever had. in France, Spain, or Italy for two euro. In most American cities, unless there happens to be a functional local bakery, you’re looking at a seven dollar defrosted loaf of bread at a supermarket that tastes like cardboard.
Since people keep doing it: no, you don't have to make an allusion to Simon's pelicans every time a Hacker News thread about a new LLM pops up. It's a lower-effort joke than even Reddit memes.
I'd just read it as social friction, just like I'd read your comment as that very same thing.
This is the consensus mechanism doing its job, essentially.
__
Though to be fair, the way I frame it assumes no connections between nodes and independent choices, when in reality, we have groups supporting each other.
So it's not necessarily the best mechanism, as social cohesion and other such dysfunctions might be steering away from the objectively correct solution through not necessarily rational biases.
Or rather not necessarily rational when viewed in just the specific context, but possibly rational when zooming out and considering whole-subsystem health.
Like posting the exact same personal brand-building under each new model you mean?
As said, there is no right or wrong here. Well, technically there is and it is my opinion (obviously), but if we take a step back, it's exactly what I just described and you (unfortunately) discarded through bulldozing.
The """"thought leaders"""" will have to live with the fact that some people just don't think that their work is adding all that much value. It's a bit unpleasant for the ego of course, but that's kinda the trade when making money with fluff.
That was effectively required to match GPT-6.1 Sol (costs and caching prices are now equal). Sonnet 5.5 made zero sense to use over Opus 5.5 under the old cache prices.
One of the reasons image AI slop has been so prevalent is that the slop is the base case (as derived from tons of RLHF because the median user has no taste) and it is becoming increasingly difficult to steer those models even when the prompt is more specified.
reply