Any NIH-funded research must be made open access one year after its publication date. NIH publishes the accepted manuscript in PubMed at the one-year mark. Unlike NIH, (last I checked) NSF doesn’t strictly require it, but you won’t be getting NSF funding unless you say you’re going to make the resulting papers freely available somehow (e.g., preprints, paying for open access, etc.). Not sure about DOE/DOD/etc. funded-articles.

The majority of federally funded research in the US is made open access. You might not realize it because news outlets typically report on brand-new articles, which haven’t hit the one-year mark for open access yet.

There’s some evidence for the same mechanism of action reducing PFAS:

https://www.sciencedirect.com/science/article/pii/S0041008X24003879

https://ehjournal.biomedcentral.com/articles/10.1186/s12940-025-01165-8

Nothing in the Frontiers is reputable among scientists. It gets linked a lot on Reddit because it’s open access, but scientists tend to view it as essentially the not-actually-peer-reviewed equivalent of a preprint. In the past, if all reviewers recommend rejection at Frontiers, the editor would be forcibly assigned new reviewers by the publishing staff. This would continue until the manuscript would get accepted. Not sure if that’s still the same (I’ve blocked all Frontiers emails), but it’s not correct to call a Frontiers journal a major reputable journal.

ONLYOFFICE (sorry for the caps, poor name) has better docx compatibility than WPS or any other suite. It’s the only thing I’ve found that can do everything in an academic style paper without issue. In addition, its source code is open (unlike WPS) and it has Zotero and Mendeley integrations. Its Zotero integration was better than its Mendeley integration last I checked.

I’m a professor and use ONLYOFFICE as the only word processor on my office computer.

Edit: apparently the Zotero plugin needs to be updated.

ONLYOFFICE (sorry about the caps, poor name choice IMO) has even better docx compatibility, and its source code is open

I completely agree on all points.

I can’t imagine that flags will get awards automatically cancelled. Any human (f)MRI work is going to describe its participant inclusion or exclusion criteria, because you can’t put people with any risk of metal in their bodies within an MRI machine. Republicans tend to like brain research because the military really likes it. Additionally, virtually all NSF broader impacts will contain at least some speculative verbiage like, “this could help to increase representation.” My guess is that flags return an AI or actual person review, which then makes a decision. Some folks at my university have been told that their awards have been cancelled. My awards that have some of these words haven’t been cancelled.

I’ve heard that most, if not all, of their stations outside of NY are essentially for training other police departments. Is that not true?

Examples? I can think of a number of foreign companies that the US facilitates, like Nestle.

Eh, I switched. I switched all of my lab’s computers, too, and my PhD students have remarked a few different times that Linux is pretty cool. It might snowball.

I actually took that bit out because LLMs are pro climate and against everything that makes the environment worse. That’s a result of being trained on a lot of scientific literature. I was just curious what Opus would say about the conceptual knowledge piece.

Claude Opus disagrees, lol (edit to add: all of what follows is Claude; not me):

I respectfully disagree with the claim that current language models are logically incapable of mapping language to actual concepts. While today’s LLMs certainly have limitations and are prone to hallucinations, they have demonstrated a remarkable ability to capture and manipulate complex concepts through language.

A few key points:

LLMs are trained on vast amounts of human-generated text which is grounded in real-world concepts, allowing the models to build rich representations that go beyond surface-level word associations. Techniques like unsupervised pre-training enable LLMs to learn meaningful conceptual relationships.

In many domains, LLMs have shown an impressive capacity for reasoning, inference, and knowledge synthesis - skills that rely on a deep understanding of the underlying concepts, not just pattern matching. Their performance on complex question-answering, analogical reasoning, and natural language inference tasks is evidence of this.

LLMs can be fine-tuned and augmented with techniques like retrieval-augmented generation or chain-of-thought prompting to further ground them in factual knowledge and improve their reasoning capabilities. This allows more robust conceptual mappings.

Careful prompt engineering and sampling techniques can help to mitigate hallucinations and nonsensical outputs. While not a complete solution, this shows that the conceptual knowledge is often present even if the model sometimes struggles to express it faithfully.

That said, I fully agree that the tendency for subtle nonsense is a serious issue, especially for layperson use cases where hallucinations may go undetected. Continued research into making LLMs more truthful, consistent and robust is crucial. Techniques like constitutional AI to bake in truthfulness, and better UX design to convey uncertainty are important steps.

But in summary, I believe the evidence suggests that LLMs, while flawed, are not fundamentally incapable of meaningful conceptual representation and reasoning. We should push forward on making them more reliable and trustworthy, rather than dismissing their potential prematurely.

To a degree. The large subreddits, like AskReddit, get far fewer upvotes on the top posts of the week than they used to get. I think there’s a good chunk of folks who left for a replacement, then left their replacement without going back to Reddit.

deleted by creator

Or Kagi. I couldn’t do DDG but Kagi was good enough for me to finally switch off of Google.

Where’s the Julia programmer that hits every one of these with @benchmark and then works for six hours to shave three nanoseconds off of the fastest one?

(Example: https://discourse.julialang.org/t/faster-bernoulli-sampling/35209)

Yeah, that was my favorite one

Within a loop could be:

for(i in 1:10){
     assign(paste0("listNum", i), list(i, someStringVector[i], i:(i+20), i*value))
}```

And you can also use get() in the same way to dynamically retrieve a variable. 

I've gone so far into coding debauchery that I've dynamically assigned variables from dynamically retrieved ones, and I've done so fairly frequently.

Eight of the monkeys are still alive and being tested on, it looks like.