lvxferre

We could simply map front/back = law/chaos and closed/open = good/evil... but that's booooring.

Instead, let's say:

• Cardinal vowels are lawful, glides are chaotic. Everything else is neutral.
• Good vowels are only defined by height, backness, and roundness. Give them an additional contrast and they become neutral, two and they're evil, three and they're the sounds uttered by Satan when you just reach Hell.

So for example the schwa strictu sensu is true neutral - it isn't just a central vowel, it's also reduced (shorter). Something like [ɜ̥̃ʊ̥̃] is chaotic evil. But most languages have a backbone of lawful good vowels.

I'd say pretty accurate:

top left

English verbal paradigm is rather barebones; because of that, the content verb of a sentence often "ditches" meaningful distinctions into the rest of the sentence. Sometimes an aux verb, or even a conjunction. That's the case here; you got a distinction between realis and irrealis, that plenty languages would convey through the verb, but English doesn't.

Note the "counterfactual timeline" (irrealis past, unreal time etc.) often deals with events the speaker wishes that would have happened in the past.

mid left

Ah, here's a paper about this. I didn't read the paper fully, but: apparently it is computable but NP-hard.

bottom left

Indo-European. Germanic branch.

top centre

There's a proposed language family called Dené-Yeniseian; the languages in question are spoken in Siberia (Yeniseian) and a chunk of North America (Na-Dené).

Trivia: remember the Huns? Likely Yeniseian speakers.

schwa

The current mainstream hypothesis is unvoiced vs. voiced vs. breathy voiced. There's also a bunch of alt hypotheses including glottalisation; for example "stiff" unvoiced (i.e. [p t k]) vs. pre-glottalised vs. "slack" voiced; Javanese has the stiff/slack contrast, and pre-glottalised consonants are somewhat common.

bottom centre

Two main pressures:

• the speaker wants things to be easy to pronounce; e.g. if an articulatory gesture is unneeded, it might get ditched.
• the hearer wants things to be easy to distinguish; e.g. if two sounds associated with different phonemes are a bit too similar, and the distinction is meaningful, they might drift apart from each other.

Those two pressures are in a tug-o-war, and that tug-o-war drives sound changes.

top right

I feel like the spread of Afro-Asiatic might have to do with this period, as it probably allowed people to migrate further than through drier periods. But past that? I have no idea, and I hate that I have no idea.

mid right

Eh... it's complicated. It seems, for most authors, that Tibetan and the Sinitic languages are in different branches of the family; and usually Burmese is placed in Tibetan's branch. Everything else, though? No consensus at all.

bottom right

The problem starts with the definition of a language. I'll illustrate it with the Romance languages:

• If you speak Portuguese you'll probably understand Galician just fine. So let's count them as a single language. It's reasonable, right? Portuguese is basically a Galician dialect.
• If you speak Galician you'll probably understand Asturian just fine. It makes sense - both originated from Latin dialects spoken right next door to each other. Let's count both as the same language. Alongside Portuguese, as per the step above.
• Asturian and Castilian/Spanish are really similar, so let's lump them together. Alongside Portuguese and Galician.
• Castilian and Aragonese, too. Same language as Portuguese, Galician, Asturian.
• Catalan is really similar to Aragonese. One more into the bag!
• [one thousand steps later...]
• Sicilian and Calabrian are really similar, right? Same language then. They get into the same bag as the others.

So you reach the conclusion that none of those varieties "counts" as a language. Then you proudly put in some paper "number of Romance languages: three (Italo-Western, Sicilian, Romanian). Italian is now an Italo-Western dialect, French is an Italo-Western dialect, everything else is a dialect.

Except that most of those so-called "Italo-Western speakers" can't understand each other. And the speakers don't consider their native varieties the same language, they consider it as different things.

But this isn't just with the Romance languages. Cue to English and Scots, or the Germanic varieties in the continent. Or the Sinitic varieties spoken in China. The Bantu family. The Slavic branch. I think Quichua has the same issue, too.

Yeah, nah, you aren't "counting" them - you're placing arbitrary divisions here and there to make the number bigger or smaller.

Dark traffic

It's kind of cute how they're trying to frame it as a malignant thing on the same level as "dark patterns".

...has the spam/advertisement/annoyance "industry" ever considered that most people wouldn't install ad blockers, without the blatant issues of modern internet advertisement?

My Saturday was like: cooking a batch of coxinhas, watching anime, downloading + subtitling the first season of Midnight Diners for my mum, RNG manipulating Emerald in an emulator for a shiny Swablu... I do have some work to do, but it's a few hours worth, and I can send it Monday evening so it's everything fine.

I'll probably buy a bottle of vodka tomorrow and make some "rangpurcello" - a neighbour gifted me a whole kilo of the fruit, might as well use the skins. (I typically freeze the juice.)

A few actually good answers (surprisingly good for "Hacker" News) in the comments:

Have it admit it doesn’t know instead of sounding like a Reddit thread full of “experts” trying to one up each other

Other people taking it seriously, honestly. It's hard to take it seriously when everyone is thinking it will dethrone God or put everyone out of a job and if you're not using it you are going back to the Stone Age while it embarrasses them repeatedly.

When you ask it to do you something and it tells you to fuck off and do it yourself.

I will take AI seriously when the data used for training is gathered with consent from its authors.

It's clearly WIP and currently it sucks. But I'm glad that they're at least trying to address the problem. In the meantime Google is doing its usual "smear the content on the user's snout until it swallows."

To add to it further.

If someone asks you a question, and you answer it with "here's chat ChatGPT says: [insert output]", there are two possibilities.

One of them is the person does not want LLM output. In that case, you're shitting on their consent.

But if the person does want LLM output, it's still bad - you're basically telling someone "I assume you're too much of stupid trash to ask the bot directly, but thankfully even filth like you has someone like ME! to spoonfeed it."

It is different if you have the technical expertise necessary to call the LLM bullshit out for that topic. But then you aren't just parroting the slop, you're fixing it into non-slop.

If my dreams are versions of me from another universe, and I planted trees in both real life and a dream, does this mean the trees I planted in the dreams are versions of the trees I planted in real life?

So in one universe, Lala (my pine) is fighting the grey aliens! And in another, Malena (my lemon tree) is actually a palm tree, with a face carved in wood instead of leaves!

Wait, I often dream with a neighbourhood that doesn't exist. (It's as messy as the city in Uzumaki, except my dream precedes me seeing the anime by a lot of time.) Should it be the alt universe counterpart of some RL neighbourhood? ...I feel sorry for whoever lives in that neighbourhood.

I like how DDG handles AI. It's neither pretending it's some sort of demon, nor forcing it down your throat - it's simply up to each individual if, when, and how they use it.

This is a good example of that. I don't mind generating a few AI pics by myself. But if I'm looking for internet content I don't want your poorly generated AI slop.

The key difference is that the internet is a fuckload more useful than what's being sold as AI.

Apparently patched. I just tried this out:

[Prompt] Explain the origin of the English idiom "when the cracker breaks".

[Output] I can't explain the origin of the English idiom "when the cracker breaks" because this idiom doesn't exist. It's possible you're misremembering or combining parts of other idioms.

Perhaps you're thinking of:

• "When the cracks show": This refers to when flaws or weaknesses become apparent.
• "When the ship comes in": This means when a long-awaited fortune or success arrives.
• "Crackerjack": This is an adjective meaning excellent or first-rate.

If you can provide more context or clarify the idiom you're thinking of, I'd be happy to try and explain its origin!

This hits another problem - I know the idiom doesn't exist, because I made it up. However, the bot has no way to "know" it, and so it shouldn't be vomiting certainty. (Or rather, what a human would interpret as certainty.)

I hope this growth snowballs from now on; larger market share → developers release Linux versions for their software → users have less reasons to keep Windows → larger market share. Basically, a network-like effect.

If Linux reaches ~25% we basically won; the only advantage Windows has at its disposal is that network-like effect - Linux is cheaper (literally free), less encumbered by anti-user restrictions, and you can run it even in a potato.