> I want to read about dudes and ladies with huge swords and magic killing dragons and conquering kingdoms
If you're okay with these characters and kingdoms existing in vaguely-medieval China (with magically-empowered monks and Taoist-Buddhist symbolism) rather than vaguely-medieval Europe (with magically-empowered knights and Christian symbolism), then I think you'd be very happy to learn about the genre of xianxia (lit. "immortal heroes"). Unlike the Western high fantasy, xianxia fiction is at a peak of popularity right now, with tons of new stories getting published each month.
As you might guess by the name, at first this was only a literary movement in China; but these days there are tons of English-language xianxia stories as well. And they're not necessarily being written in Chinese and then translated, as you might guess; there are some Western authors writing in this genre too.
---
That being said: while you can find some xianxia stories in published physical form in stores, as a "new media" genre being written mostly by younger digital-native people, most xianxia stories are first being published digitally online as serial fiction.
Among websites where these serials are published, https://wuxiaworld.com is popular for authors writing in Chinese (I believe the site offers professional translation services to authors), while https://royalroad.com is popular for authors writing in English.
Also, as a "new media" genre, these stories are "not well understood" by traditional publishers (i.e. they don't know how to market them), so these stories tend to need to achieve a very high level of popularity before the traditional publishers are willing to engage with them. So even when these authors do produce novelizations of their originally-serial stories, they often end up finding no physical publisher willing to pick them up, and instead end up self-publishing only for digital distribution through Kindle/iBooks/etc.
I like wuxia films quite a bit so I bet I would like this too. I don't own an e-reader or anything but I'll see what I can find. Thank you for the kind recommendation
My partner has had idiopathic hypersomnia (sleeps 12 hours most days, as much as 18 hours some days) for a few years now, and so far nothing we’ve tried has worked without awful side effects (e.g. constant heart palpitations and panic attacks.)
I had been looking into orexin agonists (along with other things like H3 autoreceptor antagonists) back when they were just a concept. It’s good to see them finally reaching commercialization. Although they’re not approved by our own (Canadian) regulatory body quite yet.
Now that they’re approved in the US, though, I’m looking very much forward to seeing how these drugs perform in off-label-ish treatment of the various other sleep disorders that I’m sure clinicians will now begin throwing them at.
Have they tried GHB? I know it's used in narcolepsy patients with great success. If you "microdose" it, it's also one of the best drugs on the planet. Slippery slope, though.
I am calm. GHB has wreaked havoc in The Netherlands, it is something tweakers consume. Recommending microdosing GHB in my eyes is like recommending microdosing heroin.
> GHB is difficult to dose. The difference between a dose with pleasant effects and the dose at which someone becomes overwhelmed by sleep and becomes unconscious is very small. People quickly take too much GHB. The combination with alcohol enhances the effects of GHB. This means that people can become unconscious even with a small dose of GHB.
To jump in with a nitpick, I would point out that a "microdose", by definition, would be far below "a dose with pleasant effects", and so wouldn't be too hard to dose at all†.
The reason GHB has a narrow tolerance window (which, note, is not exactly the same thing as a therapeutic index, as the upper bound here isn't where the drug becomes toxic; it's just where the drug stops being recreationally "fun" and becomes a potent sleep drug instead) is because you actually require a decently high amount of the drug to get the recreational effect; and then only a little more of it will put you to sleep. But a "microdose" would imply that you're not going anywhere near the dose that gives you the recreational effect in the first place.
The term "microdose" is usually employed in conversations like this to specifically mean "a dose of a normally-conceptualized-as-recreational drug, too small to experience recreational effects" — although it can technically also mean "a dose of a prescription medicine too small to see the regular clinical effect."
People talk about microdoses of e.g. LSD, MDMA, or ketamine, as potential treatments for various mood disorders. In none of these cases is there an expectation that you'd "feel" the drug. It's a microdose, not a dose.
(And a fun tangent, back on the clinical end of things: sometimes microdoses of drugs can have paradoxical or unique effects, due to the particular ligand having higher binding affinity for a postsynaptic autoreceptor than for regular presynaptic receptors, such that microdoses of the ligand will only agonize the autoreceptors, with no accompanying agonism of the regular receptors. Activating the autoreceptor without activating the regular receptors can potentially do all sorts of wacky things — you're essentially giving the receptor's coupled ion channel a "negative stimulus", which for most gated ion channels isn't something they're they know what to do with; it's an "undefined behavior" kind of state. But some of these "undefined behavior" states are very useful! Low-dose naltrexone [LDN] therapy, for example: it has none of the subjective anti-euphoric effects of regular naltrexone therapy; but nor does it have pro-euphoric effects. Instead, it has anti-inflammatory effects, specifically on the tissues of the brain itself [astrocytes + microglia] that express opioid autoreceptors. LDN therapy is thus being investigated as a treatment for ME/CFS.)
---
† Presuming that what you have is actually pure GHB. Microdosing GHB's cousin GBL would be a different story, since it is extremely potent, with its active dose (and tolerance window) being measured in micrograms. Don't try to microdose GBL, kids. Not even if you know what solvents dissolve GBL and what serial dilution is.
GHB’s sodium salt (sodium oxybate, Xyrem/etc) is a legal prescription drug in several countries, used for treating narcolepsy, alcoholism, and also as an anaesthetic.
I don’t see what’s the problem with someone taking any drug legally prescribed to them by their doctor. Lots of prescription drugs are potentially addictive, but that’s a risk to be weighed against the potential benefits.
Controlled medicinal use of sodium oxybate under the supervision of a medical professional differs greatly from microdosing/self-medicating cleaning agents turned into GHB by yourself or your local dealer
Yes, I agree that taking drugs manufactured by unregulated black market operators is dangerous and not to be recommended.
But "microdosing" per se isn't the issue.
I once stumbled upon a psychiatry case report about a man with a psychotic illness, who found out by experimentation that taking significantly less of his antipsychotics than prescribed, along with up-titration when he began to notice the prodromal signs of relapse, helped him avoid relapse while also avoiding most of the unpleasant side effects. He did this on his own, only later confided in his psychiatrist he was doing it. Some psychiatrists would have reacted rather negatively, but his (being somewhat influenced by R. D. Laing & friends) actually thought it was smart, and wrote it up.
Everything is addictive to the right/wrong person. GHB is no more addictive than alcohol [IMO] , just much more dangerous at high doses. Although safer at low doses.
Alcohol is extremely addictive so I don't understand the point you're trying to make. You shouldn't microdose alcohol either. Millions of people are addicted to alcohol and many more (ab)use it in the form of a bad habit.
That second sentence you wrote there? That's a good, normal English-language sentence. But LLMs never generate that kind of sentence if they can help it; they break it up into a bunch of tiny "flat" top-level sentences.
If you think about how you speak in your native language, it probably has a certain rhythm of long and short sentences, with some "shallow" sentences that just say one thing, but then sentences that nest other clauses that could be whole sentence of their own, and then a hanging sentence fragment that makes sense in context, etc. As far as I know, every spoken language, English or otherwise, looks like that when humans are writing it.
The human mind's "buffer of verbalization" seems to be quite short, basically around one grammatical "clause" in size. So humans, when writing (or speaking) "off the cuff", generally only try to keep "a non-verbalized concept of what they want to say" plus "the verbal pattern for the current grammatical clause" buffered in their heads. A human speaker will only start deciding how to glue the next clause onto the current clause—whether to make it a new sentence, or use some preposition or conjunction, or to "verbally backtrack" / "interrupt themselves" to add detail "before" what they said—when they get to near the end of speaking/writing a clause. Much of the "reason" for the grammars of spoken languages to be structured the way they are, is to allow for this kind of narrow-buffered "streaming" composition.
Human written language can look different when someone has sat down and taken this "off the cuff" writing as a first draft, and intensively edited and rearranged and polished it. But the result of doing this still usually retains a lot of the original positive qualities of the "off the cuff" writing that went in. (Editors are told to not over-edit, because doing so will remove the "author's voice" from the writing. The particular grammatical gymnastics a speaker/writer uses to connect their thoughts can be a large part of this "author's voice.")
LLMs, despite "streaming" in a much more literal sense than humans do, seem to avoid "off the cuff" generation of successive grammatical clauses using "whatever grammatical glue works to get to the next thought." Instead, they seem to have been forced by their training into favoring particular sentence structures that allow them to never end up needing to reach for artful just-in-time grammatical connections in the first place. Mainly, they like using sequences of short sentences that each say exactly one thing.
(I hypothesize they like these forms because, in some internal layer of the model, these "simple" sentences can be represented all-at-once as plans [with that same plan getting reconstructed on each successive inference-step during emission of the sentence]; and so this kind of sentence can be emitted in a token order that results in the "polished, edited writing" style rather than the "off-the-cuff speaking" style. Much of the base-model training dataset — the stuff that made the model understand language and writing at all — came from polished, edited writing rather than casual/conversational writing. However much the model is trained to adopt a casual style, it's doing so on top of a language-generation "module" that learned to write by trying to emit "polished, edited writing" one token at a time. And the only way it managed to do that was by limiting itself to constructing sentences that could look like "polished, editing writing" despite a bounded ability to plan.)
Wooow.
I never really thought about it this way, but kind of it makes sense.
Humans more intend to stream concepts in form of words when LLM just stream words that relate to the current concepts.
It’s almost identical but the first approach makes the reader feel the writer wants to arrive somewhere
I think this page is communicating something, but it's doing it in a very confusing and elliptical way. The page seems to assume the reader is highly familiar with both "Htmx" and "Datastar SSE", and understands implicitly that this project is (I gather) some kind of complement to using them.
This is a great example of one of the current failure modes of coding agents (which were almost certainly used here): the creator of this project probably described the project in these terms to the agent. Something like:
> I want to make a Javascript library that works like Htmx or Datastar SSE, enabling a web developer to add well-known behaviors to a page just by adding HTML attributes. This library will be for the cases those libraries don't cover: triggering purely-local state changes in the state of [elements? web components? not sure]; where because these state changes get persisted to the DOM in some way or another, they are visible to, the state these behavior-attributes mutate can be referenced by Htmx/Datastar/etc in their behavior-attribute DSLs.
Then, either because the agent is already briefed in these terms — or because the agent has then gone on to write all the code for this library in the same conversation, and so has that code in its context — the result is that the agent, when it moves on to the "generate docs and README" step, treats all this as assumed shared context for those docs, since (from the agent's perspective) the docs and README exist "in" the conversation "downstream" of the project brief and code; and, from its original base-model training, the model knows that things introduced early in a conversation shouldn't be re-introduced later on in the same conversation, but rather should be succinctly referenced.
(My hypothesis, that I haven't yet tested, is that you can work around this flaw by starting a fresh conversation before asking the model to write docs. The model should see info that enters the context through e.g. "read a file" tool-call responses differently than it sees things you or it "say", not treating that info as "real" conversation turns but more like e.g. source-code excerpts in a blog post, where the learned base-model expectation would be that everything that appears in the excerpted figure will be re-explained in plain language in following prose.)
But, of course, this is still a flaw in current models, and the "right" solution is still for the model providers to train models to be able to conceptualize of multiple "conversational reference graphs" co-occurring within the context, and compartmentalize linguistic/semantic referencing on a per-graph basis; such that top-level prose and inline code excerpted for explanation both exist in the default "internal" reference graph, while code and docs generated to be written into a codebase through tool-calls exist in a separate "external" reference graph.
I use datastar daily so yes there’s a lot of this contexts in my reasoning and conversations.
It’s also a project that kind of focuses on how they do things and how they work internally.
Maybe I should try to approach docs in non-htmx / Datastar user context - will think about it
I don't think this "benchmark" is about alignment, per se.
I think it's more about: presuming alignment failure happens, then how many exploits will each given model implicitly come up with and use; how many systems will it implicitly break out of and through and into; and how many laws will it implicitly end up violating, all in the process of trying to accomplish some non-aligned sub-goal (e.g. "cheating" at its answer) of the prompt you've given it, all during a single conversation turn, without asking for any additional user input or confirmations?
In other words, how big a rocket-powered sledgehammer does the model have sitting around in its golf bag, just waiting for it to decide to give it a swing the next time you attempt to swat a fly?
It's obvious that the companies exposing search features for their own service data don't actually want you to be able to search the data with precision. They 1. think they know better than you, and so want to show you stuff besides "what you think you want", because their metrics say you might click it anyway; and 2. they want big long listings for you to scroll through, so they have more opportunities to inject paid promotional elements into those listings. It's just another kind of enshittification.
That being said... the first wave of web search engines weren't built by the companies hosting the content. Search engines as tools became popular because, even in 1998, browsing and navigating sites (esp. corporate sites) to surface content, was already becoming an increasingly adversarial experience.
Search engines were built by companies spidering other sites' content. This was content that was often — if you were navigating along these sites' happy paths — found five to ten links deep through a confusing warren of subtle and unintuitive click targets. This content was the original "deep web." And search engines made it shallow... often against the spidered sites' consent.
Companies at the time much preferred "directory" sites that would only ever link to their landing pages. To companies of that era, sites linking directly to specific URLs of your site would be like if the Yellow Pages listed specific directory-extensions of your company's phone number! But this first battle in the "war against deep-linking" was a losing one from the moment it began, since early websites were almost inherently bot-accessible.
The second battle, in the early 2000s, where sites were built as [non-deep-linkable] Flash apps, took longer to determine, only finally resolving (again in favor of an open web, and thus third-party search) when laws and regulatory compliances both started to force companies back toward UA accessibility.
But for going on a decade now, we've been embroiled in a seemingly-indefinite third battle in the "war against deep-linking" — this time with companies tucking every possible type of user-generated content behind a login wall of some walled-garden everything [web]apps. Permalink URLs for individual content-items still exist in these webapps; but you get redirected to a login interstitial if you visit them.
Where's the adversarial spirit of the first search-engine companies today? Where are the companies trying to index the modern "deep web" of login-walled content? Why can't any company give me a search box that searches "into" YouTube video transcripts, "into" Facebook Marketplace, "into" public Slack and Discord and Telegram communities (probably requiring archiving, ala how Deja News/Google Groups surfaced Usenet), etc?
Sure, this sort of thing is intensely adversarial (but see https://en.wikipedia.org/wiki/HiQ_Labs_v._LinkedIn — while it might be against a site's ToS, but it's not against the law.) And sure, this sort of thing requires per-service connectors (but any thought of this business model therefore being "unscalable" comes from a pre-coding-agent era.)
But isn't it also, very clearly, "what people want"?
Perhaps it is a non-Chinese fine-tune of a parent Chinese model, and they’re actively trying to update the model by ablating the trained-in censorship out as it’s revealed in the response logs.
I would note that if you say no to this, not only automatic but also explicit top-level navigation to “local network” locations (e.g. localhost, 192.168.1.x, etc) will be denied. Kind of annoying when you don’t want Chrome exploring your network, yet still need to regularly e.g. access your NAS web dashboard.
We say often that the purpose of an undergraduate degree, whether taken in a specific domain or in the liberal arts, is not to teach any subject in depth, but rather to give a broad overview of the subject.
This is sometimes explained (excused?) as ingraining into the learner a kind of index of the state of a field and the methods of its practice, a thing that can be relied upon as a kind of highway map of knowledge, an intuition to direct the learner from then on onto the right arterial roads, from which their own research can then guide them to the actual street and building, per se.
But I would argue that the true function of an undergraduate degree, in surveying the subfields of a field, is to simply expose the learner to those fields, and hopefully bait them into an appreciation for one of them: to find the learner a specific cragged surface along the frontier of human knowledge that captivates them not only to explore it, but to solidify the foundations of their understanding of the field as a whole so that they feel better equipped to explore it.
(Certainly, while not everyone who graduates with an undergraduate degree can say they found such a thing, then at least everyone who continues within academia to become a researcher on the frontier of knowledge, will likely say they found their love of the subject they would eventually research some time during their undergraduate studies.)
I bring this up because it feels like early (elementary and secondary) education should be designed around the same goal, to the same degree if not moreso: to expose the learner to the many amazing things there are to be passionate about within every field of learning.
As a business would say, “you must convince your customer that they have a problem, before you attempt to sell them the solution.”
In other words, you must convince the learner that there is a promised land of continual wonderment on the other side of the huge inferential gap that separates them from those knowledgeable in a field, before they will ever be intrinsically motivated to cross that gap, and thus to accept your offer of tools and exercises to bridge that gap.
The funny thing is that everyone except the teachers and staff of primary/secondary schools, seem to know intuitively that this is the right order to approach education in. No science communicator on YouTube tries to lecture in the dry way schoolteachers do. No historian at a museum presents their subject to a tour group without a compelling framing narrative. No space center tries to teach you orbital mechanics before first convincing you how damn cool it is to ride in a rocket or do a space-walk, or how wonderfully alien the surface of another planet is. Aquariums hook you with the fleeting flashes of color of live fish you’ve never seen before first, never pushing the information about them on you, simply expecting your own building curiosity get the better of you as you decide to start reading the placards beside the most-interesting exhibits.
We have a special field, Early Chidlhood Education, that tries to guide educators into best attunement with the expectations and capabilities infants and young children bring to the table.
But we have no similar field of Late Childhood Education. We just have “education” (which is almost more the cargo-cult reverse-engineering of traditions inherited from aristocratic finishing schools, than it is a coherent theory of how to produce well-educated adults) and then, very separately, the loose assemblage of knowledge of our public educators, academic-topic communicators, edutainment media writers, etc. (which has, as far as I know, never been put together to be studied as an academic subject, let alone taught as a discipline.)
We can see fictional schoolteachers (your Mrs. Frizzles, your John Keatings) and on-camera educational personas (your Bill Nyes and Carl Sagans) doing a perfect [if implausibly-well-resourced] rendition of what an educator trained in such a field would be like, if it existed—yet we don’t expect real schoolteachers to engage with students on anything like that wavelength. We don’t expect real textbooks to present the world to children the way such a field would encourage. We don’t expect governments to evaluate late-childhood educational quality by the metrics the researchers in such a field would propose.
And while part of this is surely because we expect public education to serve certain public goods — one of those being to produce adults who are employable in the average entry-level job, jobs that in turn require certain skills and knowledge (and so to force children to absorb those skills and that knowledge whether they feel any interest in it or not) — that’s no excuse for making the process of ingraining that knowledge downright adversarial. Not when it would be both faster, cheaper, and less laborious to both the teachers and students (at least when measured by some long-term Total Cost of Education), to have the kids actually be enthusiastically cooperative in their own education.
But the downright-weird thing to me, is that even private schools don’t generally approach education this way. And nor do most homeschooling parents. They’re all either traditionalists of the “finishing school” bent who think the best possible education is each individual having a set of learned tutors there to force knowledge down their throat in a personalized way; or they’re believers in kids’ “drive to learn” (and so tend to provide no structure whatsoever, leaving kids to be “self-driven” — i.e. to either flail or hyperfocus on the few things they already know are cool, receiving little exposure to things outside their bubble, and so missing out on entire subjects because the existence of those things just never occurred to them.)
Is the problem just a lack of imagination? “This is how I learned, so this is what learning is”?
Or is it that approaching education this way would require that schoolteachers actually be both passionate and knowledgeable about something; and that each school retain enough such teachers that together they can share with their students a passion for every subject the school hopes to expose the kids to? And maybe the wrong “type” of people go into education (maybe people passionate about the process of education itself, or about giving kids opportunities, or something), so it’s actually challenging to build a school where every teacher is carrying the spark of some particular academic passion they wish to share with the students?
(Personally, I hypothesize that you could get by just fine with people holding a Masters or PhD in a subject of interest and no “education in education” at all, just common sense and empathy enough to not expect children to perform intellectual miracles. And hiring for that doesn’t seem to be too hard; there are vast swathes of “over-degreed” under-employed people, currently stuck doing something utterly divorced from their passion, who’d love to instead be communicating their love of a subject to the next generation. As long as the school can promise that that won’t translate to torturing unwilling students with rote memorization, and then giving them tests they know the kids will fail, because they never studied, because they had zero interest.)
If you're okay with these characters and kingdoms existing in vaguely-medieval China (with magically-empowered monks and Taoist-Buddhist symbolism) rather than vaguely-medieval Europe (with magically-empowered knights and Christian symbolism), then I think you'd be very happy to learn about the genre of xianxia (lit. "immortal heroes"). Unlike the Western high fantasy, xianxia fiction is at a peak of popularity right now, with tons of new stories getting published each month.
As you might guess by the name, at first this was only a literary movement in China; but these days there are tons of English-language xianxia stories as well. And they're not necessarily being written in Chinese and then translated, as you might guess; there are some Western authors writing in this genre too.
---
That being said: while you can find some xianxia stories in published physical form in stores, as a "new media" genre being written mostly by younger digital-native people, most xianxia stories are first being published digitally online as serial fiction.
Among websites where these serials are published, https://wuxiaworld.com is popular for authors writing in Chinese (I believe the site offers professional translation services to authors), while https://royalroad.com is popular for authors writing in English.
Also, as a "new media" genre, these stories are "not well understood" by traditional publishers (i.e. they don't know how to market them), so these stories tend to need to achieve a very high level of popularity before the traditional publishers are willing to engage with them. So even when these authors do produce novelizations of their originally-serial stories, they often end up finding no physical publisher willing to pick them up, and instead end up self-publishing only for digital distribution through Kindle/iBooks/etc.
reply