I do not mind when I am coding with Claude and it uses all the typical claudisms. I am much more bothered when I am reading a blog post, email, or other form of prose and I see those same claudisms.
I guess they are not annoying since I know I am talking to an LLM and expect the typical responses. When I am reading prose online that I previously would have expected a human to write, it can be quite jarring to realize its an LLM.
I don't know if this is a good framing. "Too much" is subjective, and every heavy AI user will assert that they're just unlocking their potential, that calculators didn't make us dumber, etc.
But to latch onto the calculator argument: if you outsource adding numbers to a calculator, you're still you. On the flip side, if you use an LLM do most of your thinking, what's left? We have people here who use LLMs to raise their children, to manage relationships, to design products. So what's your unique contribution to this world - is it the prompt you once wrote? You're standing in front of a token-generating machine, pulling a lever, sometimes receiving gifts. Is that your edge, your unique experience, your purpose in life?
Many LLM maximalists say they use the tech to learn new things, but to what effect? Are you going to apply that knowledge of physics or computer science yourself, or will you just prompt the LLM again?
In my mind, it's pretty simple: I'm a human, LLMs are not. If a human writes a novel, it's inherently worth more because it's hard-earned and anchored to experiences we share. I want to support that. And I want to be a human who can write novels, the old-fashioned way. I'm not good at lifting weights or running, so my thinking is the only thing I have.
The article is very, very light with details. The university or research center is not named. No scientist is named. No link. Nothing that tells "look, we're telling you real, solid, serious stuff."
Here is another article with that details : https://www.techspot.com/news/112051-japan-finds-way-recover...
This game was so good, it's as if the developer was given a team of PhD-level experts to work on it.
With a headline like that the graph should be at the beginning. Usually you'd put it up front and then talk about the graph, not put it at the bottom and talk about how people aren't talking about it.
Anyway, I scrolled down to the graph and skipped the text. We are currently 4 std deviations above the mean with respect to El Niño temperature.
But there's also a historical line -4 std deviations from the mean. Was that an eventful year too? I can't tell and the graph is at such low resolution that the source URL isn't visible. If the graph and data is so important, shouldn't we care more about presentation? This is either super sloppy or deliberate obfuscation.
Look, I'm all for good reporting on climate. This just doesn't feel like it.
Yeah it's kinda crazy you can't legally take them down even if they are banned/contract expires. IKE Skelton, a county commissioner took it into his own hands and they were pressing felony charges on him. Not sure what ended up happening. Basically flock wouldn't respond to take them down, he felt it was his duty to remove them, he brought them back to his office, and then the state hunted him down.
Here is a podcast about it. https://internationalflavor.podbean.com/e/the-surveillance-s...
The best part is that flock owns the cameras and the poles so even when the contract expires the cameras keep running and recording data that flock can sell to e.g. CHP, LASD, FBI, Palantir; and LAPD can just call them and access the data
the flock scam was engineered to be resilient to political pressure by giving departments and jursidictions this fake exit ability while the data continues to be harvested, it is a noose that only tightens; the amount of flock cameras recording only ever goes up not down.
Whisper is the wrong model to benchmark against, or rather, there are better models that are state of the art now like Nemotron and Parakeet both by Nvidia, as well as Mistral's Voxtral and Cohere Transcribe.
However, what's funny is, RIP to a lot of the paid apps that simply wrap Whisper, I'm sure Apple will make a native GUI such as a recorder app for macOS that obviates the need for these wrappers, which everyone seems to be vibe coding these days.
DOGE fired a whole bunch of NIH staff that processed high scoring grants to get them ready for Notices of Awards (the official document that starts moving funds, etc). Meanwhile, the administration now requires final approval of any grant by non NIH political staff.
Consequently, Science is slowing down (and that is outside of other shenanigans). What used to take 3 months is taking 9 or more.
For the many medical research Institutions where the dominant system for professors is soft money (no or partial tenure, salary is provided by research grants), there is a real crisis.
To try to make up the shortfall,we are submitting any more grants, doing less actual sciencee are submitting even more grants, and exacerbating the staffing issue at NIH.
DOGE found an actually highly efficient Federal government, doing what was lawfully passed legislation asked, and destroyed it anyway (instead of passing legislation to remove programs, the lawful way).
> There isn't really much waste in federal spending.
There's actually a lot of waste. DOGE just didn't go after it. Check out DOD and all the 9- and 10-figure programs that get canceled without delivering anything, and whose work is often useless for follow-on work. OCX is a recent example, costing around $6 billion and took so long that the program it was supposed to replace ended up just doing the work instead. Essentially nothing of OCX will be retained. This isn't really unusual in the DOD.
I will keep banging this drum until people listen:
Trying to use markdown files to limit access should never be treated as a security guarantee at all.
This is a form of in-band signalling that goes into a machine that, among other things, tries to read between the lines of your requests, extrapolate user desires, and please the user.
The only sane way to address this is using a control plane. A well-built harness can do this; a sandbox can do this; hell, a carefully-chosen `umask` can do this; but both of those are liable to introduce notification fatigue in the user.
It’s a bit alarming how cloudflare is establishing itself as arbiter of all things bots…both on blocking and allowing.
Doesn’t seem healthy for the internet as a whole
So many of the replies are saying that they should've restricted access using .md files and whatnot. Is really any guarantee that they even follow those? It seems like even if you ask pretty please don't touch those files, there's a chance they will. So many people have just willingly installed spyware on their computers and big tech calls this the next big thing.
A drum I've been banging increasingly often recently is that having friction and time to work ideas over in your mind adds huge amounts of value. Vibe coded projects have this very specific, well, vibe to them where you can clearly see that the lack of time to digest has allowed the person to not challenge their own worst impulses. You can see it in the feature bloat, the lack of depth and polish in core features and the wild asides you tend to talk yourself out of still on display.
This website is making heavy use of IP range blocking. Here's an uncensored link: https://web.archive.org/web/20260713092155/https://www.lyreb...
Alternatively, since the link that was posted is just an AI copyright theft site, use the original instead: https://climatecasino.substack.com/p/some-monsters-are-real
Discussion: https://news.ycombinator.com/item?id=48890533
They get abandoned because they get generated on a whim.
Sunk cost fallacy can be a feature: if you have spent a lot of blood, sweat, and tears on a project, you are more likely to push it through adversity and the doldrums that inevitably one will encounter. If all it took was one of those momentarily brilliant ideas and a prompt on Claude to produce something, there is no attachment whatsoever to it.
Speaking as the ‘average programmer’, I have dozens of brilliant ideas per day that don’t stand the test of time or scrutiny, and the very few that pass the filter don’t seem that interesting days later, or worth the effort at all.
Ideas have always been cheap. Now, proof of concepts have become as cheap. I don’t care about your Show HN unless you have spent a month on it.
There's so much good stuff in this post.
Can't help to think of a recent HN post about most AI-generated projects being abandoned within months. Why?
Because value of a project is not in the code produced. It's in the amount of battle-testing that code has seen.
Battle-tested, mature code > fresh rewrite.
Existing Zig codebase has seen X amount of battle-testing. Rust rewrite: 0 (except -I'm assuming- passing test suites). Also:
"this was a port to unsafe Rust, allowing a literal file-by-file migration to minimize risk"
How is that better than the Zig codebase you started with?
Now if that's further migrated to safe Rust, put into production & gathered feedback from lots of users, yes then you have something. As it is, the impressive bit is do such a big rewrite & result seems to work ok. Are Bun users happy with this?
To me it reads like Bun was forked. Will the Zig version survive? Will the Rust one? Both? All options ok.
Edit: and fwiw, I don't think Zig community should get triggered on any of this. It says nothing about how suitable Zig is or isn't for project xyz, and Zig community is big enough to carry their own project & applications besides Bun.
If it should be front-page news, shouldn't it also be at the top of the article, rather than right at the bottom?
> Anthropic is not in the programming language market; their post about rewriting Bun in Rust is full of technical details that led to improving the end product for their users
Anthropic absolutely is in the programming language market. If/since AI makes rewrites to certain languages relatively easy, a success story will tie the given language(s) to the given AI company.
Rust may have a tremendous success in the future, because it's much easier to write it with AI (ignoring for a moment whether that's really a good thing). The implication is that Anthropic has a stake in Rust's success.
Also, to be kept in mind that devs advertising successfull rewrites often hide some aspects that are unfavorable to the narrative; typically, how bad was the code before the rewrite), although there are other (significant) aspects that have been omitted.
> Zig's response is a sour opinion piece full of personal attacks.
I take you haven't read Andrew Kelley's article (here: https://andrewkelley.me/post/my-thoughts-bun-rust-rewrite.ht...).
Summary:
- Jarred has written Bun with very bad engineering standards
- Jarred has managed public relations very poorly (e.g. ghosting the Zig foundation)
- When they rewrote the project to Rust, and described Zig as poor choice, there has been a negative fallout for Zig
- The ZSF is obviously upset because of the poor publicity
This is summarized at the end of the post:
> Zig users who knew next to none of these facts and have only the surface level understanding that an ex-Zig-user is getting trashed by the language creator. Such people might reasonably worry that might happen to them
As a matter of fact, I also believed the same after reading's Bun's post. This is undeserved though, and that's what Kelley explains.
There's definitely a personal attack somewhat, and this is addressed in the last (added later) section.
I stand with Andrew.
As someone who's been following Sumner's work closely for years, Kelley's accusations are very much true even if unkind. While the results are useful and cool, it a wankfluencer op from start to finish. I dare you to refute thus.
And I say all this as someone who does agentic development 8hrs a day and someone who always pestered my team to opt for Rust and Deno instead of Node. Call a spade a spade, the rewrite was poorly justified and one in a long lines of successful psyops Dario and co. cooked and delivered.
Now, would Andrew's message have been better received if it had better "decorum"? Maybe. But I'm glad he stayed honest to himself instead and didn't have a PR team ghostwrite his thoughts. You have to appreciate that.
I think like most people, I don’t have a problem with Andrew “calling a spade a spade,” even if I find his reasoning motivated. The bigger problem with the post is that it talks out of both ends of the mouth: it’s clearly meant as a personal attack, but also insists that it isn’t.
When I read the post, my first thought was that I wouldn’t want to build things in Zig, because any technical decision I make, good or bad, might subject me to this kind of article from their BDFL. I can’t conceive of the leadership of the Python or Rust or any other community I’ve ever worked with doing something like that.
Yeah, exactly. It's weird that Zig even responded to that. Imagining that your studio switched from Unity to Unreal and Unity proceeded to release a hit piece attacking your codebase quality and workplace environment.
Did we read the same Anthropic and Andrew Kelly's posts? Anthropic is not in the programming language market; their post about rewriting Bun in Rust is full of technical details that led to improving the end product for their users. Zig's response is a sour opinion piece full of personal attacks.
For context, I'm using Codex and have no interest in either Zig or Rust, so just observing this drama from the sidelines.
I think classifying it as an allergy or a status thing is a little too glib. I’ve read and reviewed, conservatively, hundreds of AI generated documents for work, and “written”/commissioned a bunch too. My biggest issue is that it’s impossible to engage with and give feedback on an AI written document, because it’s impossible to know whether misconceptions or gaps are because the author doesn’t understand the material deeply enough or the author does but the AI doesn’t and the author’s not proofreading carefully enough. Or if a surprising idea is raised — is it the authors insight, can they elaborate on it, where did it come from, etc?
Hackernews isn’t work, obviously, but “it’s impossible to engage deeper with the material because the author doesn’t really exist” is sort of a problem for a discussion site. If the human coauthor puts in enough work, they can make sure the doc really reflects their views and their understanding, but in my experience that’s much less common.
I wish Count Binface all the best for the Clacton by-election.
Edited to add: Some of my favourite commentary around this by-election is along the lines of:
A fundamentally un-serious candidate with no coherent policies or political experience running against Count Binface.
We don't allow genai text on HN itself - see https://news.ycombinator.com/newsguidelines.html#generated and https://news.ycombinator.com/item?id=47340079. How to enforce it is a separate question, of course, but the rule exists.
We don't have a similar rule yet about article content but my sense is that the community mostly doesn't want to read it—or, to put it more conservatively, discounts it. This is why we see so many "just show me the prompt" responses, along with others like this: https://news.ycombinator.com/genai-pushback. I built that list so I have something to send to users who email about why their genai articles got flagged.
It's a fascinating arms race right now: the AIs are training on the humans but the human hivemind is also training on the AIs. Readers are developing allergic sensitivities to language that sounds like an LLM produced it. The AIs will adapt to this, but the humans will adapt in turn. Where it ends up is anyone's guess. I have an optimistic view, but I've already been wrong about this so many times that I have low confidence in it.
The current picture is that there is an emerging class distinction between writing (and writers) that use genai vs. writing that does not. As soon as the "this sounds like an LLM" allergy kicks in, the writing instantly gets relegated to a low-status bucket in the reader's mind. That doesn't mean it won't still get looked at - but it is now under a stigma.
(I was rather pleased with the originality of this until I remembered pg had come up with "writes and write-nots" in https://paulgraham.com/writes.html. Oh well, it's the point that matters.)
This has the happy flipside that anyone who would like readers to classify their article as high-status rather than low-status can apply the judo move of simply writing it themselves.
Now I need to add the disclaimer that none of this is a dismissal of LLM technology per se. We rely on it heavily, and there's no question that it's useful. The question is how to use it (pg again: https://x.com/paulg/status/2058871512451412457) and whether one should use it on writing that one publishes to other humans.
To turn to OP's questions:
> Should HN add the ability to flag articles as AI-generated? [...] it could just show up as an indicator
Flagging-as-just-an-indicator would be tagging, which we've always resisted adding to HN, but I wouldn't rule it out.
What I do think we'll (finally) add is a "please give a reason why you flagged this post" step, and "because I think it's genai" will be one choice among several (spam, offtopic, mean, etc.)
> Why is the regular voting system not enough?
The regular voting system is never enough. https://hn.algolia.com/?dateRange=all&page=0&prefix=false&so...
> Should HN change in response to the gen AI era?
To this I am tempted to reply with https://news.ycombinator.com/item?id=48887149 in an homage to https://news.ycombinator.com/item?id=3742902.
Not OP but my frustrations come from it being impossible to ignore and outright distracting.
I've found the same thing showing with Claude-coded/designed front ends that overuse the same semi-monospaced fonts, Blue/Yellow/Red palette and rounded corner borders. It isn't that it is bad, but it often isn't fit for purpose.
You're right it wont change anything, but authors shouldn't be surprised when people who care about their time/attention comment on low/no effort pieces.
> One tanh call on the right input is a per-OS signature. Claim macOS, return Linux math bits, and you have contradicted your own User-Agent.
They (or rather the LLM that wrote this) missed that this is possibly fingerprintable to browser version range, which is slightly more interesting. Most users aren't spoofing their user agent headers to be a different operating system. Most fingerprinting solutions aren't trying to infer your operating system, they only care about semi-unique things that show up.
It's an interesting finding. I wish they had taken some time to have a real person write it up. This is too heavily LLM written to ignore.
This line: "this is my main argument against the valuation of frontier labs. It’s not that AI won’t create that much value, it’s that they won’t capture it."
That is a very astute and concise way to explain everything about how the frontier labs are behaving and how they're trying to push more people to pay token rates for the best models. At the current subscription prices ($100 or $200 a month for a generous, though bounded, amount of tokens), frontier models are a no-brainer, most folks and companies will use them. But, at token rates, 10x or 100x the cost of open models or what I was spending on the frontier models a month ago? That is a harder question to answer "yes" to. I certainly wouldn't spend $1000 a month for the best model, much less $10,000; my employer might pay $1000/month, but definitely not $10,000. The frontier labs need everyone to answer "yes" to spending 100x what they currently spend to justify the valuations, and it's just not going to happen as long as everyone knows how to make these models.
Both OpenAI and Anthropic are trying to figure that out now. Anthropic, in particular, has their finger on the trigger...they want to push people to usage-based billing for Fable. But, OpenAI released 5.6 Sol, competitive with Fable (or close enough), and it's available via subscription (even the $20 subscription!), and there's no moat keeping someone from switching. If Anthropic really does end Fable access on the subscription plans in a few days, I predict a large market move back toward OpenAI.
The market isn't going to bear the cost of making the frontiers investment make sense.
> Numbers like that buy a model a real migration effort.
Such a silly choice of words. I wish the human directing the LLM writing the article put some effort into rewriting the worst examples of LLM style.
> But it did extremely well, and the promise was immediate and specific: builds finishing in less than half the wall-clock time, at 27% lower cost, scoring at or above our incumbent on completed work.
The way the LLMs write (Claude perhaps?) With short phrases separated by colons, commas or full stops, is so poor and frustrating.
There some good insights behind this article, so it's worth reading, for example below, but it isn't easy to read.
> Earlier GPT models cached implicitly on partial prefix matches, which gave decent hit rates for free. GPT-5.6 dropped partial-prefix matching:
Social media companies became so obsessed about maximizing ROI on short form video content that they stopped being a platform to share with friends and turned into Temu Youtube. You won’t see your friends stuff on any of them because it’s designed to work that way. Group chats are the only way to have a meaningful conversation if you a casual non-technical Internet dweller.
They don't have Claude write assembly because there is no training corpus on people making CRUD apps in assembly.
I'm as hateful of LLMs hollowing out the job market as the next guy, but the reality is the frontier LLMs are really good at writing anything that's been done and documented on the Internet a million times and unfortunately most of what software devs have been doing the last couple decades is shitting out cookie cutter CRUD apps.
I have my doubts about whether the state of the industry is going to advance as long as we're having LLMs do all the creation, but that's another diatribe.
> where’s all this new magical software that the productivity improvements should imply?
It's running, privately, in my homelab.
I think we are entering what I call the "have it your way" era. If an open source project doesn't do exactly what you want it to do, fork it, or create a new version. It's too easy.
This makes me a bit concerned about the future of open source. Upstreaming used to be worth it, since maintaining a fork is effort too. But now the balance has shifted significantly. Especially with many projects becoming a lot stricter about contributing, and some becoming outright hostile to AI. I can't blame them. But I think the effect will be that improvements are less likely to make it back to the community as AI adoption increases.
What really burns tokens is sub agents. I once gave Claude Code a pretty big task, and it immediately launched 7 sub agents which burned through my budget before even one of them was finished. Tried again 5 hours later: same result.
If I let the main agent do the same task sequentially, it was no problem at all. I don't know if it's really just communication and orchestration that makes sub agents so inefficient, or if Anthropic figured that most people using sub agents pay per token on a big corporate account, so this is an easy way to make more money from tokenmaxxers.
Apparently it's not obvious to everyone, but if you can't write code, you can't review it. I do know people, and companies, that says: "So what, we ask Claude to write the code, Codex will then do the review". The thing that then strikes me as odd is that they still ask for the code in Python, Java, or some other high level language.... Why? Just ask Claude to dump out assembly, or a compiled binary, but no, they don't trust the LLM that much. They still want to be able to read the code. So they need developers that can read, debug and reason about the code, yet they don't want to give them the training that's required to do this?
My opinion is that claude code uses more tokens simply because Anthropic makes more money that way and forces people into their subscriptions. This is supported by the fact that they won't let you use your sub on a different coding agent. I use pi btw.
Building visualizations with LLMs has been a major boost for my CS classes:
https://htmx.org/essays/universities-and-ai/#demos-visualiza...
Many visualizations that I have always wanted but just didn't have the time to build, I now have.
To give an example, I wanted a simplified 8-bit computer to complement the 16-bit teaching computer I use and designed this in a few days with the help of claude:
Terry Tao using coding agents to build apps means we're one step away from a Fields Medalist asking an LLM why his Docker container won't start, just like the rest of us.
I met Vint at the Kech Institute for Space Studies. He arrived to help us look at in-space data centers for planetary science throughout the solar system. He was a big proponent of delay-tolerant networking and other useful networking stacks, so he was the "rep" for that layer of problems.
Just the nicest guy you could imagine. He took the note-takers job during our breakouts, had beers with us after the session, and asked really good questions and never asserted anything the whole time.
What a legend.
The purpose of a college degree is NOT a job... but the purpose of a college LOAN is 100% a job, and it's very important to differentiate between the two.
The whole point of the loan is to buy time; you don't want to wait for when you have savings to purchase the degree, you want to do it now. If you are not doing it for the job, then why the loan, what's the rush?
If knowledge and prestige is all that matters, then don't take the loan, take the scenic route, get your degree slowly as and when you have the time and money, and one day you will have something to look back at.
But if you are doing it so you can start earning as soon as possible, when you are still young and energetic... then you are doing it for the job, and in that case the degree better be financially worth it.
You have the right to a degree in XYZ... you should NOT have the right to a taxpayer backed grant/aid/loan/whatever to gain said degree unless you're on a reasonable path to become a tax payer yourself as soon as you are done with the degree.
> But this new test, known as "do no harm," raises some thorny questions about the purpose of college. Like: Is it just about making more money?
If this is a fair question to ask students, then it is a fair question to ask the schools as well. They are the ones charging enormous amounts of money to students for this.
This doesn’t prevent people from learning to paint or play the clarinet. It prevents students from taking out enormous loans for it.
GitHub Copilot engineer here working on identity, safety, and privacy - no, even Microsoft doesn’t have access to all GitHub repos.
As years have passed since the acquisition “company” delineations have blurred a bit, but Microsoft employees still need to go through a separate onboarding process to access any GitHub company resources (internal repositories, telemetry, documentation, etc.), and then we have an additional layer of entitlements to gate and audit access to any sensitive data, including user data.
Very few employees within GitHub proper even have access to view private repositories, and in the rare cases where that’s done for legal or safety reasons the repository owner is notified.
There are currently no OpenAI employees with access to GitHub systems, so there’s about 4 layers of protection in place to prevent private repositories access. We do genuinely take user data protection and privacy seriously.
"It uploads the whole repository — every tracked file's content plus git history — independent of what the agent reads"
Holy cow!!!! I mean I kinda expected Elon would do something like this to try to catch-up.. but this is extremely concerning.
This is precisely the reason, even though their pricing is competitive and grok-4.5 is actually good enough, I chose not to go with them.
On the spectrum or go gentle vs fight, I'd have to say, now is the time is history where "fight" makes the most sense.
This is not abstract for me. I have not one, but two forms of cancer.
Both were considered incurable when I was diagnosed.
Both have treatments now that, IN SOME PEOPLE, lead to remission.
I still don't know which group I am, but I'd be dead from either one by now, if I hadn't elected to treat.
New treatments, for SOME cancers are literally coming out monthly.
So the fact that you can't be cured today, does mean there won't be a better treatment by next year, if you can hang on.
I should find out soon on my more aggressive one. Either way, I plan on continuing to try.
The subtitle is:
> I already did.
They repeat multiple times in the article that asking Claude was something they already did. So this isn’t an anti-LLM article.
This seems to be a communication problem. The other party either doesn’t know that they’ve put a lot of effort into researching this already, or their trying to give a gentle let-down instead of saying they don’t have time for this.
For the first case, the solution is to explain what you did to reach this point. People are more interested in helping those who have already tried helping themselves.
The second case is more of a social situation with an infinite number of explanations. Some times you have to read the room and realize that someone may not be interested in having those conversations with you. Some times it’s only in the moment (we all have bad days where we want to be left alone) but other times it’s a signal that they’re not interested in discussing this topic with you or maybe even anyone else.
This is no joke. I've done the crossing from Moloka‘i to Oahu (~45 miles) in a canoe several times, and those open ocean waves can get very nasty (largest I've dealt with were around 15m tall). I can't imagine the mental endurance required here, let alone the physical. My longest crossing took 9 hours, and I was completely drained by the time I touched shore. 44 days is absolutely insane.
Such a huge accomplishment.
Why is it a big deal?
Nvidia invested $2b into CoreWeave for 9% equity stake. CoreWeave is spending $35b in CapEx in 2026. Therefore, Nvidia's investment is only 5.7% of CoreWeave's single year CapEx. The other $32b is coming from other sources that isn't Nvidia. This is hardly circular.
Nvidia invests in Neoclouds because it's a hedge against hyperscalers having too much power, ie designing and prioritizing their own chips, and not fully using Nvidia's rack design. Neoclouds give hyperscalers competition. Neoclouds accept Nvidia investments because it allows them to secure Nvidia chips first, which is a competitive advantage since new Nvidia chips have been as much as ~5-20x more efficient than old Nvidia chips.
Nvidia was planning to directly compete against hyperscalers through DGX Cloud. They cancelled public DGX Cloud access when they found that investing in Neoclouds would accomplish the same goals without having to compete against their biggest customers.
If you're Nvidia, it's smart because Neoclouds that you have a large stake in will deploy your full stack from GPUs to networking to storage racks. They will share valuable usage data back to you so you can design a better next generation. Hyperscalers are likely a lot less cooperative, prefer to use their own designs if possible, and will guard their usage data.
It's kind of buried here, but Kelsey is the fastest human to do this. She beat the male record holder's time by 6 days.
I had a discussion regarding this some time ago with my grandchild who has an ADHD diagnosis. She has troubles being in noisy (especially visually) environments, yet she finds my home (relatively large home full of books, music always playing etc) comforting. She explained that all this stuff in my home is interesting for her and speaks with her - "It's you and grandma, it's full of stories". But the very modern and "must be comforting" environment in school full of patterns and pictures drawn on walls etc is just irritating – "There is no stories, just noise".
SpaceX needs to claim there’s a need for 100k more satellites to prop up unreasonable valuations. This is no different than Elon claiming Tesla owners would be renting out their cars as FSD taxis while at work (next year, we swear guys!!!)
In a functioning economy he’d have faced criminal charges for knowingly misleading investors and customers about a dozen times over by now. It’s one thing to set lofty goals internally to keep your workforce motivated and innovative. It’s something else entirely to state things publicly with a targeted date when you know there’s absolutely no chance it will ever happen.
If you've ever been in an home owned for generations, filled with books and knickknacks and heirlooms and family photos, despite the clutter it all feels comforting in a way that modern decor doesn't.
The article doesn't touch much on why modern decor emerged as it did. It's a market response where everyone needs to (or feels the need to) pick up and move at a moment's notice. Companies are either expanding or like to think they'll be expanding soon. People move jobs so often that they have a hard time feeling settled where they are, so they design for that possibility. The modern aesthetic is one of planned impermanence.
I feel that way about street advertising, beautiful European cities with historical buildings all around and suddenly a big screen/panel asking you to buy whatever.
The article headline makes it seem like the factories couldn't make the gloves.
But further down it says that the cost was double and factories couldn't get buyers.
These are very different failure modes, and speak to very different solutions.
It's nuts how well "Superintelligence: The Idea That Eats Smart People"[0] aged. That talk is a decade old by now and still hits just as hard as it did back then, despite the incredible advances made in AI in the meantime.
[0] https://idlewords.com/talks/superintelligence.htm
This is religious fervor folks, as AI 2027 was.
I grew up in evangelical christianity, and to them the end of the world is just around the corner, the same way it has been since I was a small child and likely will be when we are all gone. This isn't science. This isn't hypothesis experiment record results. This is very expensive astrology, shiny rock collecting, ritualistic meaning-making and self-justification.
Yall, with your incredible wealth and resources you could do real good in this world and make society better, healthier, better educated, and the whole world more equal, just, and reduce the desperation and suffering. Reject the false and self-serving narratives that empathy doesn't matter, that altruism isn't "effective". You can change a person's whole life in a moment.
OpenAI is a company built on copyright violation.
That means it’s in the corporate DNA to treat laws as things for little people.
Apple have deep enough pockets that they can actually sue OpenAI but I bet OpenAI are surprised they got caught.
Now ask yourself, would the Codex agents on your machine ever over step legal boundaries? Would OpenAI ever make use of data you, voluntarily, send to their servers?
If they did could your company afford to sue OpenAI and would it still be too late to save the business?
I have started to see what I think are star link satellites at night on walks with my kids. It actually makes me sad to see that on person owns the night sky and is changing the literal stars my kids will grow up with. It feels different when it’s the government that theoretically represents people but when it’s one person that feels truly depressing.
I would not like to be dismissive, but to me this article feels like an exercise in creative writing rather than a report to be taken seriously. The entire experience feels like a choose your own adventure game, seems like their stylistic intent.
I am not sure if alternative reality fiction is the best way to approach real and serious AI risks.
I am also not sure, with the amount of emdashes and the style of prose, that the entire article was not AI generated.
AI is going to be a mature scientific field. There are going to be efficiency improvements in training and inference. New paradigms are going to emerge with better multimodality, real time streaming and real time interfaces. Models are going to converge on the limits of our data available for pre and post training, improvements will be incremental and spiky in domains.
I am not sure who the AI 2040 article is for. I suspect it is intended to be a digestible piece of media for the financial class.
AI is going to be a useful technology and its impacts across the economy and global will be broadly distributed. Because AI represents the distillation of the very best human knowledge and expertise. AI is compression of human capabilities, the very best ones. Maybe the argument is that in verifiable domains, such as model training, AI models can supercede humans. I don't think so. A human's high level thinking, our incredibly more efficient semantic/neural compression, our ability to switch tasks and achieve the creative insight is not replicated through the current paradigm.
I had a client send me an ACH that was legitimately a fat finger extra zero. For me, it was a "lot" of truck payments. For them, it was a rounding error that they were unaware of until I reached out and let them know about their mistake. I couldn't wait to make it right with them because it bothered me so much because suddenly I had a pile of money that was theirs and not mine.
This is basically the end of OpenAI hardware. This is by far worst than the Waymo vs. Uber lawsuit which killed the Uber self driving project.
Also if you are a business using OpenAI models, I would highly suggest you do not because they are most likely looking at your code and IP.
Until the industry addresses the Original Sin of Generative AI (and the ascendance of Thievery Corporations), we should expect more and more of this. So far, theft has been rewarded. As long as you make enough money, people seem to be okay with ignoring long-lasting impacts of intellectual theft. As long as you become King of the Cannibals, it seems many are happy to remember you as King and not as the Cannibal.
An acquaintance of mine was accidentally wired about $100k when it was supposed to be $5k. Before it could be reversed, they moved accounts and immediately bought a one way flight out of country. They then changed all socials and handles. They are now ignoring all court documents and are on track to get a default judgement against them.
Their rationale? “It’s mine, they owed me this”. They are 100% convinced that they are in the right, not just that they can keep it but that they actually intended to send them this to begin with. I get it $100k isn’t nothing but they’re also throwing their life away for less than what they used to make a year in salary.
People do weird things when given sudden access to money or power.
>This is how you behave when you think you're so much smarter than everyone around you that consequences don't apply you.
Spot on perfect. I see this too often and not just in tech.
OpenAI really doesn’t have infinite money. They have a lot of money, sure, but it is being burned like crazy, we know this. It is widely known that they are deeply unprofitable.
Compare that with Apple, a company that throws off billions of cash every quarter. This isn’t a legit comparison.
It gets even worse. The person not only kept the laptop and used an exploit to download confidential Apple documents, they bragged about it to a contact who was still working at Apple who was also feeding him information:
> Liu allegedly kept an Apple-issued laptop after departing the company and exploited a vulnerability to download dozens of confidential Apple documents while he was working at OpenAI. He also maintained a relationship with Yu-Ting "Alyssa" Peng, an Apple employee who continued to give him updates on Apple's projects, vendor decisions, and engineering details. When Liu learned he still had access to Apple's systems, he texted Peng "LOL, I found out I can access the [network storage], so funny."
This is how you behave when you think you're so much smarter than everyone around you that consequences don't apply to you.
Whenever I leave a company I make sure everything that belongs to the company goes back to them and I wipe any access credentials or authenticator codes that might be on any of my devices. I can't imagine being so brazen that you'd keep the company laptop and then start using an exploit to download confidential information for your new employer.
Doing it at a the company that most aggressively enforces secrecy is even crazier.
Unrelated to the accomplishment or proof itself, but it's interesting how much of the prompt, even in this latest-and-greatest model, is spent essentially telling the model to actually solve the problem. Things like "Reject status reports, vague optimism, and claims that an unproved global compatibility statement is 'routine'."
Also a lot prompt spent feeding it strategies, which feel like they should/will eventually be deduced by the model itself, not explicitly stated. That's not to take away from the outcome in any way; rather, it feels sort of like when you would prompt GPT 4, "think through your answer step by step," as a sort of proto-chain of thought.
Part of the problem is that the difficulty curve becomes, like, superexponential if you try to do the actual math. Fairly elementary atoms require the full theory of quantum mechanics to justify rigorously, and anything more complicated than that requires huge bodies of specialist knowledge on approximation schemes (I assume; I haven't studied them, but given that helium already requires approximations I'm assuming the trend continues..)
Of course, they could still do a much better job useful providing pointers into this knowledge, instead of just handwaving over it and insisting on rote memorization.
I had incredible difficulties with Chemistry, more than any other subject, because most everything was hand waved away, requiring mostly rote memorization. I could never get an intuitive understanding, partly because my profs seemingly refusing to think about things from a physics perspective. My physics prof was able to help with some of it. It was very odd.
If I would have stuck with it, would things have improved?
OpenAI is about to get ROCKED on this. From this report, this looks open and shut. Apple has basically infinite money and incredible lawyers. Not sure what OpenAI can counter with unless they have clear, hard evidence this hasn’t been happening.
A company that behaves like this in one area, cannot be trusted in any area. Any enterprise that endorses/allows OpenAI products to be used is taking a big risk.
Some pretty damning stuff:
> OpenAI also instructs new hires on how to avoid scrutiny when they leave Apple. For example, Mr. Tan warns them not to tell Apple that they have taken jobs at OpenAI, so they can stay at Apple as long as they can.
> Apple says it discovered a pattern of OpenAI recruits emailing themselves confidential information when leaving Apple, including Tan.
> OpenAI apparently used confidential Apple hardware information when approaching Apple suppliers, and tricked one company into using a "specific trade secret metal-finishing technique" for an OpenAI device by claiming it had Apple's permission to do so.
> Liu allegedly kept an Apple-issued laptop after departing the company and exploited a vulnerability to download dozens of confidential Apple documents while he was working at OpenAI.
Non-competes and the like are gross but what's described here isn't just "bring your expertise to OpenAI" it's "here is how to steal secrets on your way out" which is even grosser.
QuadRF creator here. Happy to answer questions!
We have a quick demo video as well: https://m.youtube.com/watch?v=QvniJk3uNyA
Along with a deeper dive video: https://m.youtube.com/watch?v=zdJ9Tbm8ALg
We didn't give Jeff great direction on camera alignment calibration or setting the radio gain but he seemed to mostly figure it out. We're improving the UI based on his suggestions (it's open source so you can customize it too)
The RF augmented reality is just one of many applications of this brand new 4x4 MIMO software-defined radio built from the ground up. The AR uses a web app to stream RF points that your phone/laptop browser then live-merges with your local camera in the browser. I've been obsessed with low latency and high frame rate to make it a truly AR experience. More technical details at https://QuadRF.com/
> We saw in a movie how motorcycles can jump over bridges. We used AI to learn how to do this. We gave it information, like what motorcycles we use and the distance we need to jump and so on and it gave us steps on what we have to do. We practiced a lot and kept asking questions. We dug holes and filled them with broken glass and fire to practice. 18 of us died in the process. Eight of us managed to do it. The next time we attacked, we could jump.
Now listen, I'm not saying we need to give these guys more AI, but it clearly isn't yielding bad outcomes for us here.
"You're absolutely correct! For it to be a good practice ground you need to fill the trenches with broken glass and light the whole thing on fire"
When Starlink first became available here in poor-ish Central-EU, I was excited. Then, only months later, but after years of planning: EU funding brought fiber to my farm area, at ~$25/900mbps 10ms.
While my story is just n=1, I don't understand the huge upside for Starlink outside of Africa or India, where they have <.1% the money to spend on such things.
However, I am dumb, and very open to be convinced.
Currently working at an older style defense company and this fits but I think momentum is a better reference. There are no financial incentives to risk on new process. Gatekeepers, siloing, bureaucracy, and risk aversion act to stop and slow.
I have worked startups and early stage companies prior and used that experience to force developmental projects and gotten prototypes and patents through the resistance. My coworkers who lack that experience get shut down often before they even start.
If you are not in the chosen group or have a fully fledged business case with 5 levels of managerial approval it’s dead on arrival. To anyone in this sort of role it’s not blindness where you lose the skill, it’s stagnation. The moment you leave you move again. The blind fish never gets their eyes back.
I'm not trying to be reductive but the article's a lot of words for "We're vibecoding our app now and the glorious (almost almighty) Haskell compiler is too slow for the agent to iterate it's mistakes until it gets it right."
Wow. Not a Haskell user, but a big user of other languages with expressive type systems (mostly Scala; some Rust). My experience is the complete opposite. I can't imagine using a language without a good type system to catch all the junk the LLM produces. In fact I thought people would move away from languages from poor type systems, like Python, given the cost of using languages with expressive type systems has decreased with LLMs.
Having designed a good number of internal tools for teams of developers I couldn't agree more.
Earlier I had the tendency to "leave the guts" open, thinking my users were developers and would want that. All it did was put obstacles in my teammates actually doing their work. My teammates must use the tools I made for them to achieve work the company needs them to do, they don't want, nor should they want to, fiddle with a little tool they won't find anywhere else.
I still leave a lot of escape hatches, but I try to design the internal tools in such way as to make the users fall into a pit of success.
Edit: also, error messages, error messages, error messages and auto suggestions for common errors
Edit 2: also the number of people only addressing the examples in the post rather than the spirit of the post is... disappointing.
I think teaching a child to trust an LLM from a formative age is horrifically irresponsible.
If anything, an app should be made where a child learns to correct an LLM's mistakes and learn that it isn't trustworthy.
Actually, better, don't put an LLM in front of children. At all.
EDIT: If a use case is for children who can't afford good education, then use an LLM to make educational materials for children, review them, and make them available for free. After all, the contents are ripped off from human educators anyway.
This is the absolutely horrific next stage for social media platforms:
- They're already well able to surface the most addictive short video for a specific user out of millions of real videos.
- But these millions of real videos are just darts thrown into the space of "videos that could hook the user", in the end even the best-selected of them is not perfect.
- Now, behold! AI allows to generate the perfect video to surgically hit all the switches in the viewer's brain and turn it into a zombie hooked for days on end.
Let's hope our regulations hit these "social networks" hard enough so that never dare deploy this kind of technology.
The EU is a farce, an undemocratic virtue signaling organization, and this is why:
- The Parliament voted against the first reading of this proposal twice in 2026, the first time they only supported limited cases for it, while the second time they actually defeated it fully.
- The Commission didn't care, and kept the proposal on the table by refusing to withdraw it.
- Once the Commission does that, the proposal goes on second-reading (despite the first-reading having defeated it) and it is established in a very PERVERSE way in EU law that to AVOID passing the proposal in second-reading you need ABSOLUTE majority which is incredibly hard to pursue (you would think that we would need an absolute majority to PASS a proposal that was previously defeated on first-reading, not instead needing absolute majority to DENY a previously defeated proposal that was again forced to the table).
- Furthermore, absences in practicality count as "No" on the rejection. So of course they scheduled the vote in the summer when notoriously there will be many absences.
By never withdrawing a defeated proposal they can effectively and in practicality pursue any agenda they want (it requires a massive mobilization effort to find absolute majority to defeat any proposal, especially when absences for any reason effectively count against rejection).
In PRACTICALITY, the Commission can pursue any agenda whenever and however they want, and throw the votes down the drain.
EU's democracy is lipstick on a pig.
> Society should find a better way to pay for ambulances
Society has this figured out, at least a decent solution that works until we find a perfect one. Only the US society seems to be unable to find a solution.
I don't think the question is "Should ambulances be a thing?" though. It's a question of "Should someone in a situation where they need an ambulance have to balance the potentially life-threatening impact of saying no versus the potentially financially ruinous impact of saying yes?"
The (fairly obvious) answer to that no one should be in that situation. It's horrible. Society should find a better way to pay for ambulances. Most of the world has accepted that some system to spread the cost among everyone is better than putting people in that situation.
That’s not true for Postgres however: due to its usage of a shared memory pool, whenever a subprocess is terminated unexpectedly, Postgres will kill all other processes and enter recovery mode, replaying the WAL, during which time it will not accept connection requests.
It does this because it can’t possibly know whether the dying process did bad things to the shared memory pool.
Takes some extraordinary lack of self awareness to write something like this while burning tens of billions of dollars a year spearheading the "Metaverse", which is as big a digression from the company's core competency as you could possibly get. And soon after publishing this he would go on to lay off a large chunk of the team and more than 20,000 employees total.
The dev is a person from Indonesia (Rizky Nova) who's device has 16GB of ram.
Being able to use the Unreal Engine for free to develop this is awesome. This couldn't have happened 10 years ago.
> I remember watching HN and seeing every time there was something Rust related trending, there was ALWAYS a post made shortly after trying to hype Zig and this went on for like 4 years.
You just got a tiny taste of what Rust enthusiasts have been doing to every C++ related submission here on HN for years.
I think that means those MEPs are not doing their jobs. They are the representatives of their people, but somehow they left for vacation before the last day of the session. They failed in the most important part of their duty.
Almost every time I see a halfway polished “solo developer” game, they did not do all the work themselves. Especially, they usually hire out the music, maybe other sounds, and much of the artwork. Sometimes they also have freelancers doing marketing and such. Sometimes even some paid help writing the software.
I highlight this not to bring those developers down, but because I think it’s important people understand how these things actually come to be, so they aren’t discouraged to try themselves by thinking they ought to actually be doing 100% of the work solo. That’s pretty rare.
I spent a couple years managing a Postgres cluster with a petabyte of data. I wrote a couple blog posts from my work then[0][1]. I also wrote dozens of posts on the Postgres internals[2]. I've also given talks on how to generate fractals with SQL[3] and how to write a lisp interpreter in SQL[4].
[0] https://www.heap.io/blog/testing-database-changes-right-way
[1] https://www.heap.io/blog/analyzing-performance-millions-sql-...
[2] https://malisper.me/table-of-contents/
[3] https://www.youtube.com/watch?v=xKoYIvMFnoQ
[4] https://www.youtube.com/watch?v=MPSMH8w7nfw
This shouldn't be ignored in the discussion here:
The job performed by the humans was broader than what was requested of the model in this benchmark: humans also had to find the relevant invoices (searching through mailboxes, or requesting them from providers) and reason through any circumstances which cannot be inferred from the bank feed and invoices/receipts on their own. In the benchmark these circumstances are presented to the model as “user notes."
And I would be pretty nervous about asking any of the frontier LLMs to retrieve invoices: "cool, Claude logged that it found the May 6th bill from the paper supplier, I am sure it didn't just make something up arbitrary, then compound on the error by agentically iterating over the made-up invoice lurking in its reasoning traces. I checked the first 30 times and there were no problems!"
Not only was Chat Control 1.0 already rejected twice by the European Parliament but:
- This vote took place on last day of the session when many MEPs had already left for Summer vacation - 112 MEPs of 719 didn't vote.
- The vote was called only two days before as an "Rule 170 - Urgent procedure" - 73 MEPs missed the vote making it "urgent". Normally it takes months of procedure to come up for a final vote.
I just installed this. I am very confused. I no longer have a Codex app on my computer. ChatGPT is now Codex.
But what happened to ChatGPT? Where am I supposed to casually chat?
Also, when you toggle btween ChatGPT Work and ChatGPT Codex, nothing changes. This is super confusing. Can someone from the OpenAI team clarify the difference btwn the modes? Does chatgpt work have more business-y related plugins turned on by default?
Edit: So it seems like the only place you can actually chat with chatgpt is in an awkward homeless nested window. idk. The chatgpt interface wasn't great (desperately needed artifacts), but I still used it a lot. I can't see this change going well with a lot of the casual users.
Edit2: In their awkward homeless nested chat mode, you cannot even edit past messages. this is a mess, why was the team so zealous to pull the switch on unification in this state? guessing there was internal pressure to juice codex's growth but, based on what im seeing, they did it by torching chatgpt?
Edit 3: Ok so it seems like ChatGPT is still around, but renamed to "ChatGPT Classic". Seems like it wont be long for this world because there's no place to download ChatGPT Classic should you choose to uninstall it. The dmg at https://chatgpt.com/download/ only contains the new ChatGPT.
GPT-5.6 Sol sets a new SOTA on ARC-AGI-3: 7.8%
Sol is the first verified frontier model to ever beat an ARC-AGI-3 game
https://arcprize.org/results/openai-gpt-5-6
Codex has arguably been better than Claude Code for months now, but it's flown under the radar because it just didn't capture the same viral marketing effect and OpenAI in general has had more optics / PR issues than Anthropic amongst the online developer crowd. I use the word "better" not in the sense that the underlying GPT models are fundamentally smarter or more intelligent, but rather that as a product Codex is just simpler, cheaper, and abundantly reliable and low-drama.
Heavens to Betsy. My server alerted me that traffic was far exceeding a median Thursday. For the record, I am a fairly active user on HN, but I was not responsible for sharing this link here, directly or otherwise. I have little stomach for self-promotion (probably to my detriment).
Funny to see that they did not include Fable 5 in their GeneBench and LifeSciBench comparisons because "it does not answer advanced biology questions and refuses the majority of questions in this eval".
Winner by default!
There is so much less drama involved with the Codex world. You don't realize how oppressive CC is until you've escaped it. Outages, weird restrictions, degradation, accelerated usage, etc etc etc.
The developer's guide (https://developers.openai.com/api/docs/guides/latest-model) has some interesting semantic tips for using the model:
> Intent understanding: GPT-5.6 can better infer the user’s underlying goal and intended level of work without you specifying every step. Continue to state important constraints, approval boundaries, and success criteria explicitly.
> Original image detail: GPT-5.6 preserves the original dimensions of images sent with original or auto detail instead of resizing them to a patch budget or pixel-dimension limit.
> Use shorter prompts: In internal evaluations, replacing long, explicit system prompts with minimal prompts improved scores by roughly 10–15%, while reducing total tokens by 41–66% and cost by 33–67%.
> Avoid generic brevity instructions: GPT-5.6 is more sensitive than GPT-5.5 to instructions such as “Be concise,” “Keep it short,” or “Use minimal text.”
> Control warmth: GPT-5.6 does not become meaningfully better when prompted to be broadly friendlier or more empathetic.
I wrote a post on linkedin last year titled "Do not use AI to write."
Boy, was it controversial. I could not believe how hard some people were pushing back in the comments.
Quoting from myself there:
"When you write your own words, you are forging your own voice. It is distinctive, conveys your unique world view, and connects with others in a way that is specific to you alone.
If you use an AI tool to write for you instead, you lose all of that."
That seems blindingly self-evident to me, but apparently a lot of folks disagree.
Something else I said:
"Writing is hard because thinking is hard. When you write, you forge your thoughts, distinctions, mental models and even feelings into the clarity of precision that the written word demands. When you outsource your writing to an AI tool, you lose more than you know."
I guess a lot of people don't want to bother with all that.
"To authorities responsible for the measurement and distribution of time" is just the best preamble ever.
One of the most interesting innovations in the Ukraine war is their internal market place for drones, letting each drone group decide which drones they want to procure and use in battle.
It is not a top-down decision, production and supply as other armies use for their weapons logistics.
A very insightful, and correct, piece.
I'll quote in full the following, which I think gets to the heart of the matter. If you have no push, you can't apply pressure to the point.
> The notion that amateurs talk tactics and professionals talk logistics is frequently discussed in military academies and war colleges, yet it is rarely reflected in the Army’s budget requests or modernization priorities. The outdated concept of the tooth-to-tail ratio, which implies the logistical tail is a bureaucratic waste that must be minimized to support the combat teeth, must be fundamentally reexamined. In modern warfare, the tail is the primary target. If the tail is severed, the teeth are rendered useless.
Stupid parliamentary trick: Hold the vote on the day before the summer break - ensuring that many people have already returned to their home countries. Then use a sort of "reverse" parliamentary trick: the default is that this legislation is accepted. They needed an absolute majority - not of voting members, but of all members - to reject it.
Result: 314 against, 276 in favor, 17 abstentions, 113 absent
The EU is well on the way to becoming a totalitarian government.
ETA: It is shocking that 276 members of parliament would vote to support this. Are so many so naive? Or being paid off?
I have learned so much reading Andrew’s code and as I said in the original post: Bun would never have happened without Zig.
> The post claims they were fuzzing their Zig code, while during our calls the whole Bun team told us that they were not fuzzing anything. This appears to be an outright fabrication.
Fuzzilli integration: https://github.com/oven-sh/bun/pull/24826
Merged PRs fixing issues Fuzzilli found in Bun’s Zig code:
- https://github.com/oven-sh/bun/pull/28926
- https://github.com/oven-sh/bun/pull/28934
- https://github.com/oven-sh/bun/pull/29255
- https://github.com/oven-sh/bun/pull/29210
- https://github.com/oven-sh/bun/pull/29199
Searching “Fuzzilli” shows more PRs: https://github.com/search?q=repo%3Aoven-sh%2Fbun+is%3Apr+Fuz...
At the beginning of William Gibson’s Neuromancer, the protagonist is trying to sell 3 MB of RAM in underground markets. This is often cited as one of the ways the book has not aged well. But, looking at the direction of the memory market now… maybe we just haven’t gotten there yet.
This is a nice piece of democracy right here:
"a measure it had rejected twice in March. Although a majority of voting Members of the European Parliament (MEPs) actually opposed the regulation (314 against, 276 in favor, 17 abstentions), the motion to reject it failed to secure the required absolute majority of 361 votes. As a result, mass scanning is now permitted again until 2028."
"Oh no we can't get a majority to pass the law!"
"Have you tried getting a majority to not pass the law?"
"Worth a shot!"
"It worked, should we also do this multiple times?"
"Of course not! Pass the law, quickly!"
This is unprofessional and embarrassing for Zig.
I know very little about Jared but his article yesterday, which I read, seemed appreciative of Zig. I now learn he's donated significant chunks of money to them.
This entire article is publicly and personally attacking him for choosing a different product.
It's insane to me that Andrew thinks this post will somehow exonerate Zig when it really just makes them look childish. Or maybe he doesn't care, and just wants to attack Jarred?
> Two, I actually don't have any personal criticisms of Jarred
That’s quite a statement to make at the end of a post that seems to contain little else…all just thinly veiled.
Saying someone has „beginner energy“ but reframing it as a faux positive (this person fails and thus learns)
Or saying the grapevine says someone is a „stinky manager“? Basically I’m not saying this person is bad it’s just that I need to bring up on this blog that everyone agrees this person is bad.
All seems to be in very poor taste even if true…
Roberta Metsola's actions this week jeopardise the legitimacy of the EU project as a whole.
It's clear that member countries use the EU as a blame-laundering mechanism to pass domestically unpopular laws, but the forcing of this vote under the urgency procedure that requires absolute majority to reject, on the last EP session before summer break is so blatant that it might awaken people that might've overlooked the structural failures of the EU and finally radicalise them
EDIT: bad wording, it's not that the urgency procedure causes the voting to require absolute majority, it's that an absolute majority second-reading is forced through an emergency procedure which is designed for first readings of legislation that's the implied meaning above
> I actually don't have any personal criticisms of Jarred
The whole post felt like a personal criticism of Jarred.
Hey author here. Wasn't expecting to see this up.
To concisely give an overview of the project, I've been experimenting with using LLMs to build a better version of Postgres. Postgres is 30 years old and we've learned a lot about databases since hten. A lot of the techniques that work for doing a rewrite are also useful for doing a rearchitecture.
I'm now working on a new, not yet published version of pgrust that incorporates a lot of techniques. Currently the new version:
- Passes 100% of Postgres regression suite
- Implements a thread per connection model instead of the process per connection model Postgres does
- Is 50% faster than Postgres on transaction workloads
- Is ~300x faster than Postgres on analytical workloads. Right now it's 2x slower than Clickhouse on clickbench and I think it's possible to get faster than Clickhouse
I think the distinction with the Chinese models (or with any of the other models) is that they aren't particularly vocal and obviously active about their politics. I don't see how political commentary about Musk's is somehow forbidden when the man constantly reminds everyone about his political position and is simultaneously the face of his companies and obvious beneficiary. Furthermore, he's also very obviously interfered with the model development in ways that are quite ridiculous compared to other labs to insert his political opinion.
I don't think you need to somehow get personally offended by every Tesla on the road but it seems ridiculous to ask people to not be political about a such obviously political figure.
So depressing to read the non-stop political comments here. I’m using GLM at the moment, it’s Chinese and backed by god knows who, and no one cares. I really wanted to see what the experts thought of new Grok tech and how the model compares etc. I wish I could turn off the non-technical comments somehow, could literally just go to reddit if I want to see garbage like this. Am I supposed to get emotional every time I see a Tesla drive by? HN was so much better than this.
Where have the hardcore nerds gone? How is the model, is it good at coding? What does this mean for competition and pricing?
Does nobody read the fineprint?
By submitting, posting, or publishing your content, suggestions, enhancement requests, recommendations, feedback, information, data, or comments (“Content”) to any Website or Online Service, you are granting Cloudflare a perpetual, irrevocable, worldwide, non-exclusive, royalty-free right and license (with the right to sublicense) to use, incorporate, exploit, display, perform, reproduce, distribute, and prepare derivative works of your Content.
If you're ok with that, fine. But I'm not.
I don't think I have a "burnout", but LLMs are really exhausting due to amount of pressure they generate. No one is really pushing me to increase my workload, but at every moment there is always something ready, done by my clankers or clankers of other people that I could be unblocking. In the past (before LLMs) it was already hard to keep up, but now it feels like there's 10x more things waiting at any given time, and there could be 10x more if everyone just "optimized" and streamlined processes fed the AI even more tasks in parallel faster. It just being a bottleneck of everything, all the time is tiring...
I am happy about all the little side-projects, and ideas it help my realize, and I enjoy exploring this new world, but I've noticed LLMs feed my unhealthy "don't want to take a break and waste time being idle" mindset, and I need to correct it.
W.r.t. article's main complain - I think the similar thing happened due to factory manufacturing automation. What used to be a varied skillful craft in a shop became standing in a single place of an assembly line doing the exact same thing whole day. LLM took away the more creative and variable part of the work, and left the repetitive QA rubber-stamping. Probably some of the mitigations used back then could be rediscovered today.
Personally I don't care that they used AI to rewrite Bun to Rust. Even if 1.4 is not good enough it will probably get better over time.
What has pushed me back to Node is seeing how amateurish the transition has been handled.
- No LTS support for the Zig version regarding CVEs etc.
- Huge bugs like the 3MB memory leak mentioned in the blog post abandoned in the Zig version to basically force people into the Rust version to fix their apps in production.
- Zero involvement with the Bun community about such a major decision. One day it was "stop the drama I'm just playing with this" and a couple of days later "yolo merged to main".
Jarred basically keeps operating as if he was a lone hacker working on his personal project.
Shout out to Louis Rossmann for doing a ton of work on Right to repair.
He started a website called Consumer Rights Wiki to document anti-consumer practices.
https://consumerrights.wiki/w/Main_Page
He's also involved with FULU Foundation which has a bounty of 25k to get Ring cameras working without Amazon's servers.
https://bounties.fulu.org/bounties/ring-video-doorbells
Has it occurred to you that _all_ model providers are actively trying to shape their models' replies to fit their preferred political narratives?
> Deere must pay $1 million collectively to the five states for antitrust enforcement costs and will be subject to strict compliance oversight for the next 10 years.
$1 million fine for probably $10 billion in profit. I know what lesson I'd learn if my only personal value was maximizing shareholder value. The compliance part can be dealt with later.
I am amazed at people's willingness to use Grok. The company is so transparently morally bankrupt. They're the only AI company that seems okay with CSAM (or at least don't do as much to stop it)
Why give them money?
It would be one thing if they were the only game in town but thats definitely not the case.
Without commenting on Bun itself as a project, or the nature of the rewrite, it can't be good for Zig that a naive rewrite away from it fixed memory leaks, improved stability, shrunk binary size by 20%, and improved performance by 5%.
I just don't think that I can ever trust an xAI model knowing that they are actively trying to shape its replies to fit a political narrative. How can you trust their models to be reliable in a business setting with the foreknowledge that their models are being nudged around in the backend?
Netlify made this 10 years ago... they even copied the name! https://app.netlify.com/drop
The speed up numbers based on their testing:
Codebase | TypeScript 6 | TypeScript 7 | Speedup
------------|--------------|--------------|--------
vscode | 125.7s | 10.6s | 11.9x
sentry | 139.8s | 15.7s | 8.9x
bluesky | 24.3s | 2.8s | 8.7x
playwright | 12.8s | 1.47s | 8.7x
tldraw | 11.2s | 1.46s | 7.7x
Quick question: How does this affect downstream tools like tsdown and esbuild, which need to build the TypeScript codebase? Can I use TS 7 and current tsdown together?
It seems to be extremely economical - 4x better reasoning efficiency compared to Opus while being priced at $2/$6. For comparison, GPT 5.4 is $2.5/$15, GPT 5.5/5.6 are $5/$30, Opus 4.8 is $5/$25, Fable is $10/$50.
And by benchmarks (unless they gamed them), seems to be at around Opus 4.7 level, which is what Elon mentioned in https://x.com/elonmusk/status/2074911038286295049.
I guess the Cursor data was very useful.
Remember when people would argue about how types weren't worth the effort?
I love TypeScript, if nothing else for how it's been able to popularize types.
I had preview access to this one for a few weeks. It's very good. I had one conversation that lasted a full hour while I was walking the dog, got some good brainstorming done against one of my projects.
The best feature is that it can delegate questions out to GPT-5.5 in the background, so you're no longer restricted to a voice model that's several years behind the frontier.
I did report a fun bug with it though: it was interrupting me and laughing at my (not really intended as) jokes while I was still talking! They seem to have clamped that behavior down thankfully, it felt a bit rude and condescending.
Worked on my torso