soldeace

> Readers can detect LLM words in the parts per trillion.

Oh, we do. We can also detect hackeneyed, LLM-made sentences where just one word had been replaced. "That's where it's going to f*** you".

I'm definitely going to use a variation of it for learning new programming languages.

Not everyone feels comfortable running third-party opaque code in their computers.

Honest question, as I've just recently started fiddling with Meshtastic: could it be that the mesh is not set up correctly for a dense environment? (e.g. using LongFast rather than MediumFast, or not having more nodes configured as client_mute?) I know the conditions may be wildly different, but just as an example, the guy in this video says he saw no big issues on a hamvention with 300+ nodes: https://www.youtube.com/watch?v=aBfHAPpjtk4

You forgot to mention https://smolweb.org/

My wife and I tried to use Briar to communicate after we had been reallocated two seat rows apart in a flight. It didn't work at all. Messages arrived hours later, when they arrived.

In a recent project I was asked to create a user story classifier to identify whether stories were "new development" or "maintenance of existing features". I tried both approaches, embeddings + cosine distance vs. directly asking a language model to classify the user story. The embeddings approach was, despite being fueled by the most powerful SOTA embedding model available, surprisingly worse than simply asking GPT 4.1 to give me the correct label.

Despite being theoretically possible, much of the signal directed at the Moon would be absorbed in the upper atmosphere at this wavelength. On the other hand, the 10-40m bands are fantastic for long-range "earthly" communication (when the conditions are proper).

I used that website everyday when I was prepping for my ham license upgrade and got reasonably good after a while, being 25 WPM my most comfortable speed. But then I learned that the CW exam in Brazil was carried out at 5 WPM. When I tried that speed, much to my surprise I couldn't understand a single word. I had to relearn slow Morse on lcwo.net from scratch weeks before the test. My takeaway was that our brains seem to get super specialized, so if you're studying for a CW exam yourself, I do recommend immersing yourself in CW at roughly the same speed as the exam.

At any rate, really cool website!

Only after I've got my ham radio license that I learned how these USB switches are annoying sources of RFI, even the more expensive ones. KVM switches are fine though.

For the best part of my life I use a controlled set of tags[1] rather than hierarchical categories. This is mostly due to the fact that stuff can be a lot of things at the same time.

That said, one of the best use cases for Johnny's system I've found is when you have to share an online drive with hundreds of people, where you can't use tags, and even if you could, there would be no consensus. Strangely, nowadays I can find my way around a huge project's online files quite easily just by the prefix numbers of each categorical level.

[1] https://karl-voit.at/2022/01/29/How-to-Use-Tags/

> Anyone who has ever used IRC knows that there is nothing even remotely complicated about using it, but the terminology and the steps required to use one are ostensibly terrifying enough to reliably keep the technically illiterate at bay.

This remark, topped with the author's piece on "normiefication", is the kind of intellectual elitism that reliably keeps me away from IRC whenever I think of coming back to it.

At least for this particular case it was not a matter of teaching journalists statistics or anything STEM-related. If it weren't for the leaked messages, we'd never hear about it. Epistemic sincerity and a good notion of statistics are important for sure, but giving whistleblowers legal cover and a means of releasing this kind of material is just as important.

I understand that learning English may have not been that hard for you, and given that you come from a Scandinavian country, most likely you speak near-perfect English (I've been to Norway once, it's amazing how even the old lady in the yarn store speaks English so fluently!) Also, no one is contesting the benefits of learning something new.

But keep in mind that different cultures face significantly different difficulties when it comes to pick up a foreign language. I'm based in Brazil and I can tell you with reasonable confidence: the average person in Latin America struggles A LOT to get past Harry Potter books and achieve full work proficiency. This can be due to significantly different grammatical structures, word roots, phonemes, and whatnot, or---most importantly---due to the fact that most of us can't afford quality language courses at affordable prices. In reality, good and affordable schools in our own native language is considered a privilege to many.

So either due to structural differences between languages (especially those that don't share the same Germanic roots as English), or due to economical and social issues, some non-English speakers have to spend hours of deliberate practice to be on the same ground as people from a few other countries, in academia or in a multi-cultural IT team. I myself can't count the number of hours of pronunciation practice I amassed throughout the years for being too afraid of sounding dumb in my daily scrums. This is something I'm pretty sure a native speaker doesn't have to mind with when starting at a new job.

That said, I don't see how the "UK vs. rest of Europe in numbers of papers" could support the claim that a head start doesn't exist. There are a number of infinitely more relevant variables that could explain scientific throughput by country.

I really like the typeface, but for some reason it looks "blurred" in comparison to all other mono fonts I have installed. Anyone noticed the same?

One of the most useful features on Twitter for me was the option to mute retweets. After doing this, my feed's signal-to-noise ratio skyrocketed.

Unsurprised to see Paul Graham's home page among the top 10 most similar websites to news.ycombinator.com, but a bit surprised about xkcd: https://explore2.marginalia.nu/search?domain=news.ycombinato...

For the record, I'm now interested in your note-taking system and would love to read more about it.

David Chapman, hands down. The kind of insights I get reading his pieces can't be found anywhere else on the internet: https://meaningness.com

What a pleasant piece by the way. I really enjoyed his style.

As others said, the fact is that scumbags exist everywhere, and Brazil is certainly not exempted from this. Another fact is that Brazil is obscenely huge. The country is almost the size of Europe, and the number of online Brazilians nowadays is probably larger than some European countries' entire population. So, given that the ratio of scumbags is probably and roughly the same for every country in the world, the chance of anyone running into a scumbag Brazilian online is naturally larger than average, which tricks our biased brains into believing that "whoa, Brazil must be filled with scumbags".

And of course, the notoriously pervasive "stray dog syndrome", a sentiment that every Brazilian thing is inferior than its imported counterpart (including manufactured goods, culture and whatnot), and which is hammered into the unconscious mind of 11 out of 10 Brazilians from a very young age (myself included), only helps strength this confirmation bias.

Once I had an argument with a PhD biologist at my university because she was doing a Fisher's exact test in the most insane way I had ever seen: taking floating point ratios instead of integers, and because the stats program she used didn't take floating points as valid inputs for the test---obviously---, her workaround was to multiply these numbers by 100 to get integers. She refused to back down on this madness because "her supervisor did it like this, and his supervisor before him". In the end, her paper was peer reviewed and published and it was the first in a series of let downs that made me decide not to pursue a career in academia. Sometimes it's not like researchers want to tamper with their results. It's just that statistical ignorance is widespread and affect even the reviewers that were supposed to be the first line of defense. Which is a huge problem for an environment that relies too much on p-values and not so much on explaining them.

I have a blast being a senior data scientist, and I've recently turned down a manager role precisely because of the points you raised, despite the better pay. Your comment made me glad of my decision.

The investigative work in that piece reminds me of this old case: https://www.youtube.com/watch?v=OAI8S2houW4

Which is not always feasible. Some companies enforce a no-WebRTC policy, so that installing the app winds up being the only possible solution.

Are those mods? Cool, didn't know that! I'll definitely have to check it out. I remember that Death Rally had very cool mod tracks too (by Purple Motion if I'm not mistaken).

The first digital music files I'd had contact with were mods downloaded from a BBS. (Or did they come in those computer magazines? Can't remember now.) Space Debris was among them. Listening to it now immediately brings me flashbacks of 13-year-old me with the same hundred mods on repeat on MOD4WIN, creating levels for Duke Nukem 3D and feeling quite da h4x0r on IRC. These were good times, and mods were just the perfect soundtrack for that. Thank you Markus!

By the way, it's still possible to listen to plenty of mods online on https://modarchive.org/

This project is AMAZING. Earlier this week I had the exact same idea of a project while disentangling a bowl of spagetthi code. I'm glad I've seen it here today so I didn't have to reinvent the wheel. Also, I'm impressed by the quality of the documentation. It's pure eyecandy. Kudos!