moritzwarhier

It also doesn't care about what humans feel to be desirable.

Sure, humans like being healthy and strong and sociable.

But natural selection doesn't care about that either, per se. That's just a small slice of what natural selection can result in.

Didn't they do sort of a 270-degree turn on this when they abandoned erotica chatbots?

If not now, when?

I only know about Issac Schlueter, but this page is not scrollable on iOS Safari.

After attempting a second and third time, I saw two intrusive overlay ads (newsletter etc).

Dismissing them didn't help, yes, I have a content blocker, no, I doubt I should disable it.

Kubernetes

Serious answer:

For almost a decade, many of you have relied on our Multi-Account Containers extension to keep work, personal, and privacy-sensitive browsing separate without needing multiple browsers or profiles.

We’ve heard your feedback and understood the value you find in that separation. Now, we’re bringing the power of the Multi-Account Containers extension directly into the heart of Firefox for all to benefit from.

I'm confused too, maybe it's preinstalled, but disabled?

I remember seeing this Fb container symbol in a tab from time to time in my private machine, but I think I did enable something in advance.

Or it's that I just see it extremely rarely due to uBlock origin.

Thanks!

Matches my superficial experiments with trying to tweak Ollama's "modelfile" using some LLaMa- or gpt-oss-based instruction-tuned model as "base".

I need to experiment more with base models. The time from the end of 2019 onwards, when I first came across talktotransformer, it felt so magical.

Getting meaningful things out of these things can feel so... restraining.

And on the other hand: I'm tbh freshly stuck in the stage of being amazed at what current frontier coding models and apps can do.

It's just the same about the definition of "perfect".

Fixing critical bugs is not perfectionism. Fixing low-impact bugs by making high-risk, well-intentioned changes might be.

However, I believe it also shows that "general-purpose" models are not really general purpose and can't really step out of their assistant role. A common and often promoted prompting technique is to prompt a model to "behave like ..." or "you are a ..." which means these instructions do not really work.

I think you're still confusing model and "LLM app" there.

I'm not that versed myself in these things, but you can, for example, look at the conversation templates, stop markers etc. in open weight models on HuggingFace, or play around with these things by yourself and modify them using llama.cpp or ollama (the things I mention in this paragraph are, AFAIK, not part of the model). These, and parameters like temperature, sampling etc. are just the things that can be controlled without touching the model.

Of course, frontier models and their uses have supposably a lot more machinery built around them to orchestrate their usage, apart from even chatbots defaulting to "agentic" behavior for many use cases.

And models still are specialized, and fine-tuned for instruction usage, so things like the conversation template, system prompts won't be enough to bend the characteristics of such a model in all desired directions. But "general-purpose model" has become a very fuzzy term by now.

It does sound and look like it (also the HN submission text...).

I hope I'm not insulting s/1bpp/arbesman here as a person with genuine passion. But to me, the promises "LLMs can prototype your ideas quickly", "LLMs can build from solely a natural language description" increasingly sound like proverbial monkeys-paw-wishes.

Not wrong, but also the opposite of helpful.

I want to be clear here, I'm not talking about coding AIs/LLMs in general, just the particular flow

simple idea ->

working app

All the great apps I've ever used, from the measliest web clients (think of some important form CRUD thing), to the most complex desktop applications (IDEs, DAWs, games):

the moat was and is always attention to detail, and being a good "host" and guide for your users.

This app is so blown up with bombastic text that the opposite applies, I first have to consider what mindset is required to find it interesting.

Seems common.

No, I hinted that families in which phenomena of wellbeing loss are caused by access to resources can be suspect of being dysfunctional, and I suggested that active control over bodily functions is basic and its defect abnormal. I also said that resources must be treated as such, and potential side effects managed

I still don't understand what you wanted to say with the analogy.

That people falling prey to social media in a manner similar to addiction would be comparable to not being able to control bodily functions?

If you wanted to say that broken families and a bad upbringing nurture all kinds of problems like this, and a good upbringing can help prevent them, I'd agree.

That doesn't mean there isn't a lot if grey area though. But mh, it seems that I just don't understand your point.

Also, I found using breathing as an example particularly strange, since it's an involuntary reflex and not learned in a family? Correct me there if this isn't true.

That is another matter, but: the silence of Intellectuals has killed Democratic discourse. The vilest, lowest voices squeal, and nobody calls them with their own deserved names.

Yes and this happens not only but increasingly on Social Media.

I'm not even sure if I think such a law helps anyone, I just understand neither fear nor outrage over it.

In my view, it has no effect at worst, or positive effects in the best case.

Re: free speech, I brought that up because I wanted to understand where you are coming from with your comment and viewing such a law as a form of censorship was my best guess.

You think people only breathe because their "functional families" remind them of it?

Strange, shallow hyperbole.

Social media is causing widespread mental health crises, has killed democratic discourse and regarding the "functional families", I doubt it being helpful there either.

I respect being against censorship and/or surveillance, but portraying Social Media bans for children as plainly stupid in this way is deeply unwarranted IMO.

Free speech does not depend on Meta, and other platforms that intransparently and intentionally promote personalized manipulation of public opinion and monetization of suffering.

Using social media (or better, offering it) is not necessarily a free-speech issue.

In my opinion, free speech is not hindered by banning minors from using social media platforms / banning platforms from monetizing their lives.

The German Physics Society and the German Meteorological Society

are not the German government.

Wishful thinking, at best

What .editorconfig has in common with CLAUDE.md and similar files, is that they should be shared among the team.

Individual guidelines can live in a dot-prefixed subdirectory or the users $HOME.

This is what they already provide.

So it's possible (but not necessary) to add personal pre-prompts, but the shared best-practises and, more importantly, learnings, can be comitted and pushed.

What's exposed on a web server is a different topic, and when that is your repo's root dir, you've done something wrong either way.

I'd suggest "Caveat".

The problem

While an article lends a headline more weight, in incomplete phrases consisting solely of a substantive, "The" is a superfluous rhetorical device.

"The Exorcist" could just as well be named

"Exorcist".

But it was not the style at the time.

We already know it's important. If The Caveat doesn't stand out enough without The, maybe one should consider interleaving it with the preceding text, or increasing the heading level.

Do you want me to increase the heading level of Caveat by using only a single #?

But hear me out: there comes

# The Markdown Trap

In fact, this is not always possible, because heading levels decrease when adding # characters, which limits our headroom.

## The solution

I've implemented a Markdown transpiler that assigns inverted heading levels based on the number of #s.

With # beinh regular body font size, mapped to ######.

Higher heading levels are compiled to style attributes, providing an almost limitless signifikance scale and infinite nesting levels.

So from now on, you can use

  # Heading 

Work your way up to

  ###### The Caveat

And more hash signs make it stand out even more.

(green checkmark)

markdown-transpiler.sh

I know this is about addictive social media feeds as a whole, but I'd like to think a law is being passed to "endanger", no, better outright ban infinite scrolling.

Call it the "pagination law" and combine it with banning personalized feeds as a whole, apart from fully transparent choices made by the user (e.g. follows).

One can dream...

Treating medical care like this, similar to a racketeering scheme, should not be a thing anywhere.

Instead of punishing people for seeking medical care (or plainly requiring it), it would be preferable to have a robust protocol for rejecting patients that do not require care, whether at the hospital or before the ambulance ride.

For this, people would need to want medical care be a humanitarian right and basic pillar of a functioning society, not a business or a bureaucratic system to perform a selection process decreasing the life expectancy of less-affluent people.

Ironically, when there is free medical care and universal insurance, there are also perverse incentives. For example unneeded expensive procedures, prescribing patented, newer drugs where cheaper ones would work, providing ineffective or even detrimental services, etc.

But humanitarian values are out of fashion, because they were never achieved globally.

So the only right people care about increasingly is their right to own property.

Not the same issue, but makes me nostalgic for these simpler times:

https://www.wired.com/2011/04/amazon-flies-24-million/

There was an HN comment pointing out this exact fact in some Anthropic thread recently, that agents are not suitable for indeterminate questions that would better fit a chatbot.

I couldn't bother to find it yesterday, but I think this article was just generated by another user who read that comment and thought it's suitable to bloat into a blogspam submission.

Probably finding suitable comments or other content that can be bloated and used for monetization or engagement farming is already feasible to automate, too.

Disclaimer: didn't click because the headline already tells this and follows an overused generic phrasing.

As a Western German myself (who doesn't vote AfD), I'm inclined to agree.

Division and looking down on people, especially using attributes like "braindead", is arrogant and also just adds to the political capital of the far right.

Things that are not direct calls to violence or tearing down democracy must be acceptable parts of the discussion. And calling people who have opinions you consider hostile or wrong "stupid" has never helped anyone. Even when these opinions already cross the line of what's acceptable or are really "stupid" in your best-faith interpretation.

The Ossi/Wessi thing is especially problematic to bring up I feel. Also, the German East has been attracting industry and science facilities for quite a long time. And parts of Western Germany still seem to have issues with losing their perceived cultural hegemony.

But this process should be welcome to anyone who honestly wants a united Germany state.

Sowing division by looking down on the "former DDR" is a poison to democracy, just like yearning for autocracy.

The term "former DDR" is only recently going out of style in Western Germany, and that already says a lot.

Looking down on the AFD is understandable when you detest right-wing ideology, but it is no more helpful than laughing about people who were nostalgic about parts of the former DDR.

Adding arrogant arguments about wealth, the economy or cosmopolitanism will only increase this division and also the success of the AFD in the West.

For example, rejecting to celebrate and awe at cosmopolitanism and multiculturalism is not "Nazi".

That being said, yes, there are violent Nazis in Germany, many but not only in Eastern Germany.

Pointing fingers at only this problem just further sows division though, and it should also be clear that the far right that is active in the East is not a seperate entity from the far right in Western Germany.

Some regions in Western Germany also have strong far right hegemony, or are shifting to it, especially the poorer regions.

Sowing division about wealth and relating it to xenophobia is not helpful in increasing tolerance or decreasing frustration of working people.

And there is a shrinking part of the German left that ostentatiously is about human values, but has an ambivalent relationship towards capitalism, while rejecting any and all remotely critical sentiments regarding immigration. Often, even rejecting the German state: considering themselves leftists, but at the same time being deeply intertwined with the dominant cultural and political currents, and often enjoying a lot of wealth.

The arrogance of this "wealthy liberal establishment" as an issue is not much different from what helped Trump get elected, I feel.

The contradictory combination of being anti-right, pro-immigration, demanding of material wealth and at the same time claiming the moral arrogance of being on the morally "good" side will not slow down the decline of the left party of the political spectrum.

The whole premise of a "giant secure open source C codebase" seems questionable

Because code review is sometimes not much different from an idealized version of the halting problem, where you would have access to a formalized version of a specification.

In other words, there is no strict definition of what is a security issue.

Or using it correctly... depends on what your system is, or more accurately, your policy on reviewing or quickly merging PRs with "Draft" status.

As long as it's taken as an indicator for WIP, it works. It just doesn't work when acting illiterate of this distinction; and I have often have had PRs switched to "ready", reviewed + merged in a couple of hours.

But when the change list grows, and the PR ages, while still being intentionally maintained, the Draft signal is strong and helpful IMO. Switching an old Draft PR to "ready" after reviving it with changes seems like a useful signal to me.

undefined is not less absurd because there are parts of the standard that allow you to differentiate between

  undefined 

Of course you shouldn't do that, but I once encountered a library that behaved differently depending on whether an option in an options bag object was not present or explicitly set to undefined.

You can run into similar ugliness with function parameters, if you do evil things like using

  arguments

All not things you should do, but taints the "purity" argument, doesn't it? :D

https://tc39.es/ecma262/multipage/indexed-collections.html#s...

Arrays are exotic objects that give special treatment to a certain class of property names. See 10.4.2 for a definition of this special treatment.

I just meant these special properties. The behavior, apart from the square-bracket syntax for construction, can be emulated using Object property descriptors, Symbol.iterator etc, but AFAIK, much of this is retro-fitted.

Not disagreeing with the fact that arrays are almost just regular objects in JS, but the "just" in "just objects" does have nuances, AFAIK.

JIT Optimizations for non-sparse arrays might just be part of a larger hot-path optimization system, but I think there are still differences.

Is it possible to create an object for which

  Array.isArray

E.g.

  Array.from({"0": 1, "1: 2})

  Array.from({length: 2}, (i) => i + 1)

  [1, 2]

  [...Object.values{"a": 1, "b": 2})]

all implicitly use the built-in Array prototype.

I'm not sure if it's possible to build an array using only primitives and functions from the

  Object.

If this is true, it really would be impressive.

I intentionally corrected it, because vanilla JS arrays used to behave like this in some contexts, but I'm not even sure about which ones still.

Since the "Array" is a reference to the prototype, I think it might be outdated due to changes in undefined behavior of runtimes when serializing array objects, or logging them.

I'm pretty sure that [object Array] used to be the result of logging an array at some point.

  Object.prototype.toString

  Array.prototype.join

Thanks for the factual clarification. This is so important when everyone already has their trigger finger on politics. Not meaning that politics are irrelevant here, see sister comment by jobim.

But it's impossible to form a nuanced opinion when political association has a higher priority than the facts; which, again, don't look flattering for the implementers.

I do it this way, but this led to people asking me in reviews why I use NULL^^

My explanation was that it signals intent to me, and is different from some property not being part of the expected object shape or not having been initialized because of some accident or logic failure.

Since then, I've sticked to it, and am "allowed" to use NULL ^^

It can lead to some annoying checks in TS for primitively-typed properties, so for these, I still allow explicit usage of undefined when it's simpler given the surrounding code.

But I agree with you in principle. Using "undefined" as a "second nullish value" and explicitly checking for it is a programming error.

When there's object/areay vs null/undefined, thankfully, the truthiness narrowing often allows me to interface with code relying on "undefined" without explicitly handling this "value" in my own parts of the code base :)

Thanks a lot! I'll give this a try next time I feel like coding for a hobby.

I've tried exactly this "project idea" in a half-assed way some time ago with next.js, Postgres and Prisma before I was out of free-time-in-front-of-computer.

Then abandoned it again it when I started coding for a living again. It was exactly the "modest RT capabilities" thing when I abandoned it. Because I started using Firebase (2023, Supabase was not known to me then) and then I felt like "OK, I managed to get full next.js running properly on shared hosting with limited server-side JS, but now I again need an external cloud provider for a hobby project" (Firebase).

I read a bit about their pricing models and lost interest.

I think next time I'm gonna try see what this stack can do.

In 2024, I heard a very interesting conference presentation about Erlang and its concurrency model, that was after the hobby experiment, and made me want to revisit losely functional languages with strong pattern matching (OCaml too, but that's a different topic).

Previously submitted here a couple of times, last one is 2 years ago.

So I thought, why not re-submit it. Relevance is probably timeless, although it's a neologism that doesn't present a particularly new idea.