sundarurfriend

And yet, even for programmers, that output is full of info that is useless 99.9% of the time. That is just bad design. It is basically a debug log that is shown by default all the time, instead of written to a log somewhere for the <0.1% time it's useful.

And I point out the attitude because I've seen it for decades and seen the harm this kind of "it's just harmless extra info" type thinking has brought. I don't know what tone you're reading into it, but I'm quite literally and explicitly criticizing the attitude, not you personally.

Because git is an open-source project written by nerds, it shows you all of this information. Feel free to ignore it!

This is the type of attitude that kept most Unix tools quite user-unfriendly for several decades. What information to show the user, and when, are important design decisions to make. Just dumping it all on the user and making them wade through it is not doing the user a favour. Thankfully newer tools seem to be better about this, which has brought the Unix shell forward by leaps, even if there's still ways to go. (You can still make a conscious design decision that the info needs to be all there, and git is one of tools where that's at least somewhat justifiable, but a lot of the time the attitude is more like in the quoted text, that dumping out more info is always better.)

Agreed with all the rest of it, but

to have the notebooks play nicely with the existing package environment system instead of being self-contained

isn't that just a `Pkg.activate` away? https://plutojl.org/en/docs/packages-advanced/

Because the patterns of consumption and harm are very different. It's the constancy and immediacy of social media that makes up the major part of its harm. It's not something where a 15 minute smoke break or a weekend binge is the main mode of consumption and harm. So the parents have a lot more opportunity and power to handle it if they give it reasonable time and attention.

The PR is here: https://github.com/thesysdev/openui/pull/517

Pretty hilarious, but also not that much of an exaggeration from other more "serious" PRs like this.

I don't use Obsidian, and my assumption when I saw the title was I guess they're gonna be limiting it to a small set of corporate-blessed plugins.

I've come to expect that "The Future Of XYZ" titles from software companies means severely limiting XYZ or preparing XYZ for a shut down!

"whitespace, not brackets" from a sibling comment touches on it, but a lot of people, beginners especially (but not uniquely), are put off by symbols when reading code. Python is less symbol-heavy than most languages, by using whitespace and syntax and words (eg. `and` not `&&`, explicit `lambda x:` rather than `x =>`) in their place. It doesn't go so far as COBOL as to be cumbersome, but far enough to make a difference to a lot of people.

The tools the language gives you to create those abstractions make a lot of difference, however.

He isn't specifically related to the 3D printing or maker space (in this sense) niches. He got popular at first for repairing Apple devices that Apple Stores claimed were unrepairable or quoted enormous repair fees for. He made videos about those cases, and then transitioned into right-to-repair efforts in general. He also started the Consumer Rights wiki: https://consumerrights.wiki/w/Main_Page

It's hard to decouple them as primary vs secondary because Julia is pretty central to what they're doing here. To my understanding, all the actual calculations that this is based on are in Julia, Dyad is basically a layer above it that gives a declarative interface, AI integration that understands that interface, and a GUI that makes it even easier (than the declarative language) to input the model. So funding for Dyad has pretty heavy incentives to go towards improving the Julia ecosystem because that's where its foundations are.

The paywalled (or subscription-walled) portion isn't too long, it's a pretty small article, but here it is: https://removepaywalls.com/https://www.axios.com/2026/04/30/...

Does anyone have a general idea of if $65m is typical, or larger or smaller than the usual funding amounts for these kinds of industry targeted "boring" software?

Despite the framing, I think Dyad's role is more to fill in the areas where Simulink is a pain to use and has been wrangled into shape for lack of better options, than to replace it. The agentic part can be a big pull though, if they can get it to reliably produce what the user, eg. the engineer, asked for, without having to spend more time correcting it than they'd have spent writing or laying it out. Seems plausible because this is a specialized niche-purpose AI, but still not 100% certain it can get there IMO.

I don't know the actual Christian theology, but at least in modern popular interpretations, Lucifer is the Angel of Independence, so that would suggest no!

Scarcity: Why Having Too Little Means So Much is a pretty good book about the psychology of this. The stronger your the necessity for saving (whether from poverty or external influence like here), the deeper it gets embedded in your psyche as well, and can start to feel like "this is just who I am" as in habits around this becoming something you see as intrinsic to your personality.

This is the most basic level of eval, of whether they can produce output that will be considered by someone somewhere (usually a young urban US American) as informal toned. Real human communication is far more nuanced than this, different groups have different linguistic registers they're used to and things outside it sound odd even if they can't articulate why. You could also want to be informal but not over-familiar with the other person (for eg. in a discord chat to a new acquaintance) - actually looking at the outputs here, the Claude output seems best fitting for that (in my subjective view anyway) than to the one you gave it - or want many other little variations.

What makes one cringe and another recognize as familiar and comfortable is also pretty subtle and hard to define. These things need nuanced descriptions and examples to actually get right, and it's in understanding those nuances and figuring out the register of the examples that Grok outshines the others.

As an English-as-second-language speaker and writer, one thing Grok really shines at is capturing the tone and level of "formality" of a piece of text and the replicating it correctly. It seems to understand the little human subtleties of language in a way the other major providers don't. Chatgpt goes overly stiff and formal sounding, or ends up in a weird "aye guvnor" type informal language (Claude is sometimes better but not always).

Grok seems in general better at being "human" in ways that are hard to define: for eg. if I ask it "does this message roughly convey things correctly, to the level it can given this length", it will likely answer like a human would (either a yes or a change suggestion that sticks to the tone and length), while Chatgpt would write a dissertation on the message that still doesn't clear anything up.

Recently I've noticed that Grok seems to have gotten really good at dictation too (that feature where you click the mic to ask it something). Chatgpt has like 90-95% accuracy with my accent, the speech input on Android's Gboard something like 75%, Grok surprisingly gets something like 98% of my words correct.

I've been feeling more optimistic about Mozilla recently than I had in years, since their language in communication seems to have shifted from a Stepford-ish tone of corporate speak to something that feels more authentic and closer to their roots. I don't know if it's the new CEO, or a general cultural shift. (Or just me projecting from little intangible bits of evidence to something I hope for!)

Hearing about positive personnel shift like this now gives me a bunch more optimism on this. I really hope I can go back to the days of unambiguously being in support of Mozilla and their many awesome efforts, without always having to be a bit dubious about their next (mis)step.

That is a useful guide in terms of the personal psychology of how to go about doing it, which is an important side of it, thank you.

I'm also interested in the mechanics of how you actually do it: for eg. your mention of paper maps for travel makes me think if a lot of that becomes workable because you're in planned cities with reliable maps. I'm a mid sized town in India where maps are vague guides for the general layout, but are missing the many many alleys and connecting roads that people actually live on (or have shops at). Roads, road names, traffic restrictions - pretty much every part of it is chaotic and incredibly hard to put together without a GPS on a digital map.

On the family aspect too, do you have a Matrix or similar for the larger family to connect through and share news on (their own travel for eg., or difficulties they might be having, or news like child birth), or do you only use phone calls or texts to connect?

In any case, I can definitely relate to:

even worse, you are mentally always ready to be contacted, for a new dopamine hit of information or a new decision to make.

and feel the negative effects of that, so I'll be moving actively towards what you're suggesting. Maybe to a different point on the line and with different workarounds, but it sounds at least 90% workable and with significant benefits too.

I assume the word is in there for the sake of people who don't know what a honeypot is. It gets them curious that law enforcement set up something fake, even if they don't immediately know what it is for.

Do you have a blog post or similar that describes how you do this?

tree-sitter's design has potential, but my impression is that even after all these years, it is yet to be realized. The speed claims turned out to be largely overstated in practice, for the general variety of usage (rather than single task benchmarks or special cases). And the claim with the grammar system was that, given such a coherent system rather than the much-hated regex parsing, people would be able to write better grammars that are less prone to edge case problems and be less buggy. And maybe that's true in cases like this where someone gets paid to write the grammar and maintain it, but in most common cases, the actual quality of the grammars turn out to be much the same, but with more possibility of regression or breakage. It's possible that in ten years' time, tree-sitter will clearly be the way to go, with more polish all around, but at this point it doesn't feel like an easy strong recommend over the traditional parsing systems.

Does it appear only if you visit the website on mobile? I don't see any street view option when I visit that site on a PC.

Not extensively. The few interactions I've tried on it have been disappointing though. The Voice input is really bad, like significantly worse than any other major AI in the market. And I assumed search would be its strong suit and ran a search-and-compile type prompt (that I usually run on ChatGPT) on Gemini, and it was underwhelming at it. Not as bad as Grok (which was pretty much unusable for this), but noticeably worse than ChatGPT. Maybe Gemini has other strengths that I haven't come across yet, but on that one at least, it was

    ChatGPT 5 ~= Claude > ChatGPT 5.2 > Gemini >> Grok

ChatGPT 5.2 has been a good motivator for me to try out other LLMs because of how bad it is. Both 5.1 and 5.2 have been downgrades in terms of instruction following and accuracy, but 5.2 especially so. The upside is that that's had me using Claude much more, and I like a lot of things about it, both in terms of UI and the answers. It's also gotten me more serious about running local models. So, thank you OpenAI, for forcing me to broaden my horizons!

Because anything that even slightly differs from the standard American phrasing of something must be "LLM generated" these days.

We don't have TikTok here (India), but I find YouTube shorts pretty useful. My feed is a mix of Action Labs shorts, Omar Agamy news, ZTT PC stuff, psychology, videos about animals (pet, domestic, wild), etc. I don't know what your standard for meaningful is, but the shorts are at least as useful as long form videos to me, if not more.

An important bit of context is that I prefer to get detailed information through text, research, sometimes podcasts, but rarely ever videos. The shorts serve as one low effort way (among others) for me to surface new potentially interesting things, to follow up on to the degree I find useful or interesting.

The original comment was about one language that borrowed cipher from Arabic (i.e. English) where the word no longer means zero. So my comment was about a different language that also borrowed the word cipher (i.e. Tamil) where it still retains that meaning.

In Tamil, it still means a zero. It's usually pronounced like 'cyber' though, because Tamil doesn't have the 'f'/'ph' sound natively.

In what sense?

Pretty fun overall. The roasts were really good, the stats were pretty neat, the predictions made me smile. The xkcdupe fell flat for me because it's pretty generic and I've seen many variants of it before, but still pretty impressive for an AI to make. The "Your HN Front Page in 2035" was definitely the weakest part for me. Maybe because every gag in it had already been done in other sections.

One thing I noticed is that it seems to give more weight to submissions (i.e. things I've submitted as posts) than to comments (or at least, doesn't let submissions get drowned in a sea of comments), which turns out as a good thing. And it doesn't seem to care about karma, which is also IMO a really good thing, makes it feel like a deep dive and keeps it interesting.

This made me compare the figures, and: did they accidentally switch those around, or are the Post-training Reasoning and Factuality scores actually significantly lower than the Pre-training ones?

Edit: Just noticed

Also note pre-training and post-training benchmarks are different, so scores are not comparable across plots.

The paper gives more details about the specific benchmarks and the scores obtained in them: https://arxiv.org/html/2512.14856v1#S4