asveikau

> It's also fighting their very successful advertising campaigns

If you want to talk about successful advertising campaigns, things like age verification laws are actually being lobbied by Meta. In other words, Meta wants to do this and is spending money to make it happen. That should make you suspicious.

> Being a parent

I am a parent. I really hate it when people pull out this "parents say this ..." crap as if it's a shortcut to winning an argument. They don't want to have a discussion on the merits. A lot of the people pulling out that line seem to be childless, if it matters.

I'm the primary caretaker to two kids. So yes.

I don't understand the legislative pushes for some of these things.

Schools can already say "no phones during instructional hours". Many of them do. I guess legislating it can mandate the policy across a school district, state, country etc., so perhaps that makes sense.

But others call for outright bans for kids, all the time. Parents can also restrict usage or set rules for appropriate use. Many do.

> My opinion on local LLMs [...] :

>You avoid the uncanny valley where the model appears to be intelligent before doing something stupid; the stupidity is self-evident and this helps calibrate your interactions.

This seems to be capturing a feeling I am finding messing with local LLMs: popular software around this seems to be slop coded by no talent hacks. It's a common experience to see a python process pegged at 100% CPU with a huge amount of memory allocated, seemingly doing something that could be a simple shell script. That is before stuff reaches the LLM.

If that runs on somebody else's machine, you don't notice. When you see it in top it's frustrating.

It's probably just that I've seen too many weird comments on here lately.

Looks like the decline started before the ai boom.

I can relate. I was active there from 2009 until about 2014, which looks rather like a plateau in the graph. It still showed up in Google searches but I mostly just lost interest in participating.

That hasn't been my experience at all. The quality of these things is absolutely terrible. Maybe if you have no taste and no actual skill, you can't tell the difference.

The fact that you have chosen to say this about generated amd64 assembly is telling. This is a terribly pointless exercise. Further, the llms are bad at more niche languages especially. Even experimenting with having them write C, which is quite a bit more common than writing straight assembly, I have seen them fall over quite severely.

The linux kernel also has its own headers (IIRC it was something like <asm/unistd.h> and/or <sys/syscall.h>, but might depend on architecture and version) where it will provide the stub asm statements.

In my memory the syscall ABI has changed a few times (i386 had int $0x80, then sysenter, then abstracting it in the vdso, then amd64 has 'syscall'), so it may be easier to let the kernel header provide the mechanism.

I can have a machine generated X server in assembly too. It's called cc -s.

Or:

   $ objdump --disassemble /usr/lib/xorg/Xorg

Maybe I'm getting too old for this, but I really don't see the point in having AI generate assembly code for this.

Phones are skimping on RAM lately because of the price spikes due to the AI boom. Phones and laptops are sometimes debuting with less RAM than their predecessors of the same brand.

    replacements = {
        "seam": "whatchamacallit",

That seems like it would work whatchamacallitlessly.

I had an edit in my comment about military spending being an exception but I decided to leave it out to not distract from the core idea.

Though there is nuance there too, as some wasteful military spending seems to be more of a jobs program for specific congressional districts.

Because Bill Clinton pushed Democrats rightwards with his electoral policy of "triangulation". That was the start of establishment Dems embracing Reaganomics.

Any time people talk about cutting government spending, they are exploiting naivety of the audience. When you actually cut, nobody likes it. There isn't really much waste in federal spending. Most programs are important to somebody for good reasons. From this perspective, DOGE was always an obvious con.

I haven't really seen it as an acronym "UAF", but I can't recall the first time I heard "use after free". It was probably in the previous century.

The idea that Claude came up with it is ridiculous.

> Are you theorizing that the churn causes bugs?

Seems to be the case.

How many times do you see a bug investigation and it's determined when the bug was introduced?

Do you ever look at the diff that introduced it to understand what was going on in the project at the time? Often, it's in service to a new feature. Sometimes the original change is questionable when you consider you traded it for a severe bug.

I am fluent in both languages and I can't succinctly translate it.

If I changed the meaning of the sentence, I could say, "can I study civil rights if I'm a lefty?"

If it were text, I might give you the translation I already gave yesterday and add a footnote explaining the pun.

Neither of these are ideal. One changes the meaning, the other is hard to understand.

Coming late to this thread, but this is reminding me that many remote execution vulnerabilities have come from font rendering code paths.

If there is a turing complete VM inside the font renderer that makes perfect sense.

I'm not sure if it's the case but at one point on Windows, font rendering happened in the kernel.

Yeah that's weird. The instructions are not even hard to read. I don't understand what an LLM would add to this.

This is how well structured teams have worked for a long time.

It sounds pretty similar, for example, to team structure ideas proposed by Fred Brooks in The Mythical Man Month from 1975.

Came here to complain about this workflow too.

If you don't want to resort to piracy, there are many vendors that sell epubs with weak DRM and presumably give money to publishers. Ebooks.com is one. If you have not already, I would recommend looking into calibre for managing such titles.

(I did this even when I used to buy from Kindle, first thing I would do is break the DRM and put it into calibre even if I was only reading on Amazon devices, because I never trusted Amazon in the first place. But supposedly the DRM breaking flow is broken with new kindle releases.)

My point is that a succinct and correct translation for the pun does not exist in the target language.

Puns generally do not translate well.

Where they do, there needs to be a coincidence where words with multiple definitions happen to have the same meanings in both languages, or there happens to be a similar saying in the target language. Often this is not the case. Translators often just make up entirely new material to substitute in for that.

It's funny that TFA seems to use the comparison to honey as disparagement, rather than interpret the same information as an endorsement of the helpfulness of honey.

I'll give you a very simple example I saw recently. I saw a dad joke in Spanish on social media and it gave me a chuckle. It was a comment to a lawyer's video and it translates to "can I study law if I'm a leftie?" It relies on the fact that law is also a word for "right".

AI won't be able to succinctly translate that joke and have it hit the same way. As an experiment I just fed that into ChatGPT, and it did explain the pun in 6 paragraphs with quotations and a bulleted list, but that kills the simplicity of the humor.

I don't know your history, but this sounds like a monolingual person not understanding and being arrogant about it. It's not a niche hobby if that's literally your life. Many people are not multilingual as a hobby, or as a choice.

I think any multilingual person will be skeptical of your analysis. Translation is not a substitute for understanding the original. It's good that we have translations, but it isn't the same. As you get into consuming art, literature, poetry, commentary, this is relevant.

I got a Quadra 650 (if memory serves) super cheap on eBay circa 2003 and put A/UX on it successfully. I still have that machine but it hasn't been powered on in quite some time.

I think I wrote this in an HN comment before, but knowing what I do about linguistics and people, I think some of the AI writing styles will be re-enforced and organically more popular even for human authored writing over the next few years. We're all being unconsciously exposed to those language structures and they will feel comfortable, normalized, and we'll be inclined to repeat them.

I don't like it necessarily, but that's the way it works. Repeat exposure creates that.

I haven't lived in Seattle since 2011. I remember Belltown having one or two blocks that felt a little spotty or like you wanted your guard up, but the rest of it was thoroughly gentrified by the time I left, and it was a common choice to live for new college grads from the suburbs arriving to work in tech. I have to assume the process has continued.