For those unfamiliar, CachyOS doesn’t ship a single kernel…

Well, they ship a default kernel, yes, but one is immediately presented with a dedicated GUI for swapping between them.

There are all sorts of variants. There are kernels with different schedulers, a server/workstation kernel, a “deckified” kernel, a LLVM built kernel, and all sorts of niches optimized for specific applications. Making switching frictionless is central to CachyOS, hence they build variants for different CPUs, and drivers paired to each one, so you don’t even need dkms to switch between them.

They are adding an EAC kernel + guide as an option.

It’s yet another niche option.

That is a critical distinction. I don’t judge people for all the weird niche stuff they want to do on their PC; playing Fortnite isn’t even in the top half of that weirdness. Providing that as an option is fine.

…If they were to ship it by default?

Yeah. Then I would be concerned.

But Lemmy/Piefed could be structured to mitigate it. For instance, there could be a “community notes” header like Twitter (one of its few useful UX innovations, IMO). Or perhaps Piefed’s repost consolidation feature could be extended to include “reposts” of an original source, like tabloids running newspaper headlines.

…But yeah. You are not wrong. Ultimately I think it’s a user sentiment issue. I’ve discussed this in other comment sections, and a good chunk of Lemmy outright say “Posts fitting with my views are more important than factual accuracy.” Others questioned why we should even care about primary sources, or sources with a reputable track record.

Hence users like the headline above. And it makes me sad that, even when pointed out, even after everything we’ve escaped from and how we’ve seen the internet deteriorate before our eyes, folks here don’t see that as the priority issue.

Yet it’s just upvoted to the front.

TBH this aspect of Lemmy/Piefed (clickbait constantly topping feeds) has disappointed me. I guess I shouldn’t be surprised; it’s structured like Reddit, so one would expect it to have the same structural issues.

…But coming from Reddit, I thought folks here would prize information hygiene above all.

I mean, if every “AI” company went belly-up tomorrow, if development of any new model was banned globally, no exceptions… it would still not go away.

Deepseek 4.1-tier models would be useful for decades, and cheap to run. Heck, I think the scaffolding around them still needs years of development to not be so half-baked.

You aren’t wrong. That sentiment is everywhere in the comments below, and highly voted.

I mean, this isn’t true from my perspective. All the Chinese LLM labs are still developing full steam ahead. Model/architectures are getting updated so quickly it’s a meme amongst inference framework devs, because they can’t even support one model before the next one is already out.

…They’ve always been thrifty, though. The difference is they don’t need a trillion dollar bubble to fund their modest, more sustainable dev/inference costs.

I mean, I sympathize, disease is miserable, but that’s too far in the opposite direction?

The risk of genetic engineering leading to cascading social inequity isn’t trivial. There are many, many solid sci-fi plots that illustrate exactly this.

And one could argue “just do it equitably!” Yes. That is the goal. It’s a work in progress. It’s kinda like saying “AGI will solve our problems” or “we’ll just geoengineer our way out of climate change.”

Like others said, it looks better next quarter than a bunch of new projects. Short term gain vs long.

But also, their expenses from all that hosting and usage are high. They’ve masked it creatively, but that only works so long.

Maybe they’re trying to make their balance sheet look less catastrophic in the short term (again, at the expense of the long term).

Whatever it is, that will be the underlying theme. Things must look good immediately, at the expense of sustainability.

They needed this before “AI.” As there’s some evidence social media was already ruining attention spans.

TBH I have felt some degradation myself. And I’m more social media skeptic than most because, well… gestures at being on Lemmy.

Most companies won’t do this, though. That would require worrying about the company’s health years out, rather than next quarter.

Indeed. But many pollsters (like Nate) acknowledged this huge miscalculation in hindsight, and have tried to dissect that and predict with that in mind.

In fact, the original source addresses this and discusses the topic extensively. It’s most of the interview:

I think Democrats have the emotional feeling that when there’s a polling error, it is against them. Because Hillary Clinton and Joe Biden and just in general — polls actually have repeatedly underestimated Donald Trump. But is that true? And is it particularly true in midterms? Have polling errors been consistently against Democrats?

Good content, Nate Silver knows what he’s talking about, but why didn’t you link the original source?

https://www.nytimes.com/2026/10/27/opinion/ezra-klein-podcast-nate-silver.html

It’s not paywalled. And mediaite is a questionable source, to the extent that even Wikipedia advises against citing them:

https://en.wikipedia.org/wiki/Wikipedia:Reliable_sources/Perennial_sources

It’s generally best practice, most ethical, and good information hygiene, to link the original source instead of a downstream “repost,” or at least to link it in the post text.

Maybe questionably truthful.

But at this point, I’d settle for a not-Cyberpunk future. The prospect of a total Chrome monopoly horrifies me, hence I’m all for this ad.

Yep. This is why Nvidia is preaching open weights literally every chance they get, too.

Still. It’s a better stance, even if for selfish reasons.

It doesn’t literally mean “90% failure rate vs 100%”

In the HLE benchmark, for example, we may be talking about 45% test score for MiMo Pro or GLM or something, vs 55% for Sonnet and 61% for maxed out Claude.

But further complicating that, those particular open weight models do better than Claude in one particular hallucination benchmark, meaning Claude/ChatGPT might be worse at RAG/extraction type workloads. It’s like this across the board, with different models trading blows in different areas.

By “90%,” he really means “it’s so close the better model depends on the particular application, workload, harness, sampling configuration” and such, and that none of these models are close to “100% accuracy” across the board.

One other thing: Anthropic, OpenAI, and Grok have all been accused of “dumbing down” models under high demand to speed up inference. If your application depends on them, and they are the only provider, that could throw a wrench in the workflow.

But open weights? Pick your provider. You get an API as consistent as you are willing to pay for.

In other words, that “100%” isn’t really representing the potential inconsistency one gets being locked to a single provider.

Yeah. I see this attitude on Lemmy a lot. In Linux circles in particular, but elsewhere too.

No one is perfectly insulated from the rest of the world, as much as they’d like to think they are.