>unified memory

This is just marketing speak. Stop repeating marketing. It isnt a walled garden, its a walled prison.

Unified memory is just regular memory. There is nothing special about integrated GPUs.

Isn’t that is how it’s called though? PS4/PS5, Xbox consoles all referred to it like that on the spec sheets.

Ugh, more "AI" hype. How useful are the cited hardware features for NON-"AI" processing?

Does it mean I can buy chips that are on the boards and solder them if they go bad?

It sounds like repairability means dividing device into smaller not repairable parts and make extra money off of it.

For instance, can I get those replaceable ports on Mouser?

Repairwashing.

Shame the keyboards have a copilot key. That doesn't sound so bad until you see that the thing emits a key chord, not a scancode, making it annoying to remap. But you can.

The most annoying part is that the key matrix isn't set up to 3-key rollover with the copilot key like it would be for a real modifier key. (I'd assumed they'd just keep the matrix they used when there was a modifier in that spot. Nope.) Consequently, some key combinations, e.g. ralt-rcontrol-spacebar, don't work. Press them, nothing happens. Infuriating.

I love this. T14 gen 7 was the first NB I a actually bought for myself, and it's great to know that USB-C ports can just be replaced that easilly without soldering and that it was designed from the start with repairability in mind. Non-A USB ports is something that always ends up failing.

lobster: Like rust, python and ruby all mixed together

Very cool

The game currently has a Mixed (65%) rating on Steam. Granted, some negative reviews are shallow, but some mention important issues. Regardless, a Minecraft clone is not exactly groundbreaking in terms of gameplay.

This is to say that technical merits are rarely good indicators of a good game. As a gamer, I don't really care about the game engine, and even less about the language it's written in. Good programmers often obsess about these details, but it's easy to miss the forest for the trees, which is what I think happened here. Game design is a separate skill from game development, and not many people excel at both.

Still, it's great seeing this here, as the technical achievements are quite remarkable.

What GUI text editor widget does it use, or is it home-grown?

I don’t see it mentioned.

? Seems to support GTK, Qt and ncurses?

looks like it uses scintilla

>why can't i find love in san francisco

Because that´s the last thing going on your mind in San Francisco. You have long ago before going there manifest to get funding and make money, The rest is blank.

No need to ask AI for that LOL

> We heard feedback that GPT‑5.2 Instant would sometimes refuse questions it should be able to answer safely, or respond in ways that feel overly cautious or preachy, particularly around sensitive topics.

Lol it won't solve the issue when ChatGPT treats me like a teenager and tells me to ask my parents about everything (I just don't want to provide my ID to OpenAI to verify my age). Btw that's why I stopped using ChatGPT in my everyday life

surprisingly accurate! Is Gary the AI equivalent of the “nothing ever happens” guy?

Now do this for every single person with actual power

Now do the Pentagon. Gary Marcus is uninteresting cause he has no Power over anything.

> Don Knuth

Please put this in the title! I know it's editorializing, but this is such an important detail.

This is such a gem. I almost didn't read it.

Edit: I also feel like ~knuth in the URL should render on HN like social media handles.

Previous discussion: https://news.ycombinator.com/item?id=47230710

Enjoyed the article! As someone who's worked on bits of collaborative software (and is currently helping build one at work), I'd caution people from using CRDTs in general. A central server is the right tool for the job in most cases; there are certain things that are difficult to handle with CRDTs, like permissions and acquiring locks on objects.

Edit: I had an excerpt here which I completely misread. Sorry.

Heh. Find how to grant permissions/ acquire lock in git. You can not. That is fundamental to distributed systems.

Downside: harder to think about it all.

Upside: a rocket may hit the datacenter.

From what I remember about Figma, it can be proclaimed CRDT. Google Docs got their sync algorithm before CRDT was even known (yep, I remember those days!).

Here's a 2019 Figma article on this, not sure if its representative of their current system

> OTs power most collaborative text-based apps such as Google Docs. They’re the most well-known technique but in researching them, we quickly realized they were overkill for what we wanted to achieve ...

> Figma's tech is instead inspired by something called CRDTs, which stands for conflict-free replicated data types. ... Figma isn't using true CRDTs though. CRDTs are designed for decentralized systems where there is no single central authority to decide what the final state should be

https://www.figma.com/blog/how-figmas-multiplayer-technology...

the weird physics are mildly infuriating. still funny though

That is the joke, I think. The game is to touch anything and try to not make the rest fall down.

Simply clicking on the empty background already makes things fall down.

In his latest essay, Leonardo de Moura makes a compelling case that if AI is going to write a significant portion of the world’s software, then verification must scale alongside generation. Testing and code review were never sufficient guarantees, even for human-written systems; with AI accelerating output, they become fundamentally inadequate. Leo argues that the only sustainable path forward is machine-checked formal verification — shifting effort from debugging to precise specification, and from informal reasoning to mathematical proof checked by a small, auditable kernel. This is precisely the vision behind Lean: a platform where programs and proofs coexist, enabling AI not just to generate code, but to generate code with correctness guarantees. Rather than slowing development, Lean-style verification enables trustworthy automation at scale.

no one wants to believe this but there will be a point soon when an ai code review meets your compliance requirements to go to production. is that 2026? no. but it will come

We already have specifications though, so that’s not different. What happens when the AI is wrong and wont let anyone deploy to production?

Interesting, I have built https://github.com/michaellee8/voice-agent-devkit-mcp exactly for this, launch a chromium instance with virtual devices powered by Pulsewire and then hook it up with tts and stt so that playwright can finally have mouth and ears. Any chance we can talk?

That's actually interesting. Is it a dependancy on user to create the HTTP endpoints for the /speak and /transcript?

One of our learnings has been to allow plugging into existing frameworks easily. Example - livekit, pipecat etc.

Happy to talk if you can reach out to me on linkedin - https://www.linkedin.com/in/tarush-agarwal/

> he'd been offered a promotion, to an Engineering Manager role

Funny how this lateral move to another function is seen as a promotion.

I've done both for significant amounts of time, and rather than a blanket, utilitarian "dont become a manager", I'd go with the antithesis to that blog buried at the very end:

> So why am I still an EM [...] the main reason is that I enjoy my job

EM positions come in all sorts of shapes and sizes, and it's an entirely different function from that of a developer. I had tremendous fun being a manager in a couple startups, where left with lots of autonomy I could learn about, then experiment with better ways to deliver than "let's do 2w sprints" and ship shit. The human management was interesting, especially the continuous improvement side of things: it's especially exhilarating when you find something someone can do better and have a durable impact on their career ; it's especially tiring when you have to become something at the convergence of a psychiatrist, a referee and a nanny.

In large companies, the job isn't the same. You're stripped from autonomy and forced into a bureaucratic aspect of things. Dates are the main control dial that VPs have, so your main goal is to provide random dates, track random dates, make sure it's gonna be delivered at random dates, and make up excuses for why that date was not met.

After alternating a couple of times between the two functions, I figured development is what brings me the most joy, so I staid with it. But to each their own, and you might want to be a manager:

- if you have a true interest in the function, go fo it. There's a lot of learning to be done (the main problem with bad managers, I believe, is that they're thrown there because they were good devs, and they just make shit up rather than learn) and you'll discover things

- at the opposite side of the article's thesis, AI is a chance for you to innovate as a manager. The bureaucratic aspect I mentioned can be smoothed by it, and new tools mean a new way of working, so good times to experiment!

- don't just do it for the utilitarian side of things. Developing your career is important, but you also need to do it a sustainable way. Something I keep telling: it sucks to be good at something you hate. So do something you like.

- it is not my experience that pay is lower, Amazon paid SDMs more than SDEs, Microsoft pays them the same.

- titles mean very little. VP at MyFavoritePet who employs 12 people is not the same job as VP at Amazon. Principal (not principle - makes my eyes bleed every time) is harder to achieve at Amazon than at Facebook. Not because the job is more complex, but just because they define things differently.

//Funny how this lateral move to another function is seen as a promotion.

Not at all. IC salaries outside of the absolute top-tier companies are capped, and were traditionally always capped lower than any degree of Senior Management prior to the 2000s.

More to the point, they were capped illegally and in collusion with the main players in the game, completely separate from market forces.

This was ably demonstrated by the class action taken when five former software engineers sued Apple, Google, Adobe Systems, and Intel in a Federal District Court in California for colluding in an “overarching conspiracy” to keep wages low by promising not to poach each other’s employees.

https://equitablegrowth.org/aftermath-wage-collusion-silicon...

65,000 software engineers eventually claimed they were unable to jump companies for higher pay because of a series of non-solicitation agreements by the likes of Sergey Brin, Eric Schmidt, and Apple's Steve Jobs.

Outside of VC/PE funded American tech hotspots, this depression of salaries for IC roles still tends to be the case - particularly in Europe - for whatever reason.

Simply put, the promotion is in the remuneration; the lateral move in functionality is simply a required re-alignment of role and responsibility to meet the expectations of the 'Leadership' tier - something always distinct from original job function, be it in Sales, HR, or Engineering.

I took an EM role. About a year later they eliminated all EMs in the US and replaced them with people in Poland. So I guess take the EM role if you're in Poland.

They sent a clarifying email today

> Our email yesterday was imprecise relative to our actual new Terms. To be specific:

> You must be at least 18 years old to use the Service (Zed’s AI-enabled software-as-a-service offering, including features like account creation/sign in, Zed Free and Zed Pro, and collaboration). See https://zed.dev/terms#21-eligibility. We set the threshold at 18 due to children's data privacy obligations under COPPA, equivalent international frameworks, and an increasing number of state and regional laws that extend protections to anyone under 18. Those regulations require parental consent verification, age-gated data handling, and separate retention policies for minors. Building and maintaining that infrastructure is a real cost for a small team, and getting it wrong carries regulatory risk. Setting the line at 18 lets us maintain a single privacy framework for all account holders without carve-outs.

> Zed's Software (open source code editing software) is governed by our open source licenses. In cases where the open source license can govern, it will over the Terms. See https://zed.dev/terms#24-restrictions.

I use Zed, but not the AI bits of it. Works really well as a plain code editor. I hope they remember there are folks like me who just want a better code editor than the miserable shite that is VSCode, without all the LLM stuff in it.

Not that I have a problem with the LLM stuff, I just use the LLM in a shell and then use Zed to fix the problems in the output.

You might like this then https://gram.liten.app/

"The shortest poem is a name". The longest memory i own is also a name.

There's been a lot of "the world doesn't work the way I want it to" on HN recently. I suspect this is a function of an aging readership more than anything particularly groundbreaking about hot takes on the up and coming tech.

"Anything invented after you're thirty-five is against the natural order of things." Douglas Adams

I'm reminded of the Air Canada customer service chatbot. It completely made up a refund policy (and there are still people on HN who insist LLMs don't hallucinate) and a court ruled the company had to honor it [1].

The only way to deal with this is to make the implentation not worth it by constantly bypassing it to speak to a human and/or making it cost money by getting it to give you things you're not otherwise entitled to.

I really wonder how these things will handle prompt injection and similar things. I have no confidence any of this is secure.

Wait until this comes to healthcare and it'll be chatbots handling appeals to prior authorization denials, wasting even more physician time.

[1]: https://www.wired.com/story/air-canada-chatbot-refund-policy...

Do you enjoy reading slop? I fail to see how this is a controversial take.

Their problem is that it looks great and all, but there's just zero reason to upgrade my M2 MBA -- until I'm forced to install Tahoe. When it's time to buy, I'll grab one of these or whatever's latest, I'm sure, and I'll get another Mac without thinking twice about it, but I'm not even sure I'd notice the giant speed difference?

I did get tricked into putting Tahoe (or whatever the iOS version is called) on my iPhone 12 Pro though, and my phone is now sluggish and sad, so I am going to have to upgrade it, which I'm carrying quite a lot of resentment about. Hoping I can hold off until the fold phone.

  The difference is in the number of GPU cores. One chip has an 8-core GPU, while the other has a 10-core GPU.