mholm

People in business class can buy this laptop. That doesn’t mean the people that buy this laptop fly business class

I'll note that many of the glass blocks they use are available all over, for a few dollars. I bought a few Doric glass blocks, stuck some diffused LED strips below them, and printed a stand/enclosure, all for under $15. Good gift for friends.

The calculus here is more grey than it seems. 'Plant based leather' is largely plastic, while the leather industry (especially around cheap accessories) is made from meat industry byproducts. The hides would be destroyed/trashed otherwise. There's also the durability and biodegradable aspect of leather, over plant-based (let's not even talk about 'vegan leather', which is just vinyl).

Not to say that the leather industry doesn't still contribute money to the meat industry, just that this change reads more as greenwashing than an upgrade.

Switching models is _very_ expensive in compute (you have to rerun everything from the beginning), and highly variable in cost. Cursor tried doing this for awhile, but inconsistent performance/usage means most users turned it off and pick models specifically.

The article mentions this is about a legal requirement, not Anthropic considering adding one. They state they and many others already have one.

Their power needs must be absurdly strict, considering they didn’t want the 12v -> 220v -> 12v loss for an M1 chip.

As models train up the intelligence ladder, many common tasks will hit fully diminished returns, and instead it'll just get progressively cheaper to do that task. But the tasks that AI is capable of doing are also expanding. I'm not sure 'Some tasks don't require the peak of the frontier' is worth worrying about, from an AI finance perspective.

OpenAI has paused training for multiple weeks, and is still working on releasing a full postmortem. This is not getting swept under the rug. A lot of the engineers internally are very worried.

There's been a lot of effort into the writing space, and the models genuinely prefer this style. You can let them iterate on the same idea 100 times, rewrite sentences, determine what works best — and they'll still verb the noun, do rule of 3, and keep the same monotonous structure.

This was true in the ChatGPT era. Now we're in a world with reasoning tokens, where a model can thoroughly plan out the response it wants to make. If anything, it makes the style worse.

Black for Python https://github.com/psf/black

> I guarantee you somewhere in their backlog is an item that says "require app store for app installations on macOS"

This has been said for over a decade and it still hasn't happened, and Apple hasn't even tried to do so.

I think a better way to model their approach is to push non-technical users further and further away from the typical .dmg app installation approach, and put hoops up to prevent them from following malicious instructions, like the privacy/security and unverified developer hoops. Mildly annoying for technical users, but near impassable for the tech illiterate. We've been seeing this transition across windows and android for a long time, and it seems to be a net positive, if my family is anything to go by. I spend much less time on tech support now.

> He reported returns of 439% for the first half of 2026

> which posted a loss of 67% in July

So it's still up 44% this year? The article notes this, but seems unnecessarily adversarial against an investor who is still wildly successful.

98.99999%

Yes, they host it on their own infrastructure.

I don't believe it's published anywhere, but it's common knowledge to Apple engineers. I can second the poster's assertion that they run Claude internally.

The excessive spending was part of the point. The market is watching AI spend very closely. AI spend is a proxy for AI absorption. Expanding your MS contract to include Copilot for your rank and file doesn't move the meter much. Letting your engineers run up 100k/month shows extreme absorption. The share price increase potentially pays for the entire token bill once Wall Street goes through your earnings.

The study itself from late 2025 is also written to be _highly_ favorable to non-ai usage. Developers on a codebase they're already experienced with, largely using these tools for the first time. Of course they aren't faster! They've already done the hard part! They're also open-source maintainers, which, ime, means they're already a cut above the average dev.

I'd be much more curious on a study that used regular developers on brand new-to-them codebases for significant features or migrations. I'm personally encountering what I'd describe as a 200%+ productivity boost.

These are not competing with Kobo. These are micro-ereaders, often much smaller than a phone. They're characterized by extremely weak processors and minimal hardware. 'Apps' are not a concept here yet.

Yeah. I also noticed a fairly significant speed increase when I started using Safari again. Unfortunate. There are just too many features that Safari supports perfectly, that Orion either doesn't support at all, or has flakey support for. Especially the Apple ecosystem ones, which I'd think Apple doesn't expose, but Chrome seems to support some of them just fine, like 2fa codes.

I'm in the same boat. Nothing else is as rock solid or has the color accuracy of Hue. You can immediately tell when it's some random brand

Once you start doing frequent business through PayPal, they begin 'holding' some of your money to form a buffer against chargebacks and fraud. This doesn't happen often at the consumer level, but dominates within the small business world.

> It is unlikely this kind of agentic workflow will ever get cheaper

This seems like a very '500kB is all you'll ever need' thing to say. This exact scenario has seen a precipitous decline in cost and increase in reliability over the past year.

What questions? When I go into auto mode, it doesn't come back until it accidentally/intentionally tries to slip the guardrails, or completes the task. My prompt will generally include information on what it's allowed to do to accomplish the task, where to test, etc. Simple, but effective.

As someone who loved Supreme Commander, I really wanted to like BAR, but I found it felt like death by a thousand cuts. I last played more than a year ago, so things might be different, but the QoL felt really lacking.

If a teenager on your street said he was going to spend $1,000 to customize his Honda Civic for his needs, you'd believe him. If he says he's going to build a brand new car, better than a Honda civic, for $10,000, you'd laugh and say good luck.

OpenAI and Anthropic don't make small models. Google happens to already have a billion devices that would benefit from small models, so they made one. Google basically gets 1 billion per year for free*.

Predictable for a year or two, yes. I know how much I'll be paying a decade from now, exempting property tax and insurance fluctuations.

I was in middle school in 2007

Beyond the financials, the psychological impact of both being able to make greater-than-superficial changes, and having extremely predictable payments for years without worrying about substantial rent increases, is substantial.

I redid/improved the bathroom to exactly what I wanted. I renovated the kitchen. I added paneling to the walls. I added a few outlets to rooms that needed more. I wouldn't do these things in an apartment, because rent could go up any year and exploit me for liking my home. Property value has gone up by 50% in the years since I bought.