als0

Everything aside from apps with SafetyNet dependencies. Banking apps come to mind.

Why? Do you think frontier AI will never generate enough profit to justify its costs? The market seems to move on vibes rather than logic. I can imagine one of those company failing and the other seizing the market.

I've always thought CSP as a robust design pattern where you have no shared state between components and they must communicate with each other using message passing. It also requires synchronous communication (rendezvous-style). If you follow those rules you can have a pretty robust system. Aside from these abstract rules, CSP has more formal research (algebra) but I'm not sure if there are any decent tools available.

TLA gives you a full toolbox and in theory can model whatever you can express. That's very different from a design pattern.

Why decades?

How does battery life compare to Sequoia? I heard there is a huge drop

Google is the new IBM. All the signs are there.

I thought supermarket loyalty cards were just a QR code of a simple serial number. Is it different where you are?

The story about that AMD ARM core is that they were making it for Amazon. Performance was not good enough so it was canned. Amazon then went on to make their own chips called Graviton.

> your average diabetic who has fairly good control over their blood sugars

I think it's huge for diabetics. It's a mental toll for them. They hate having to constantly prick themselves to take readings. It's awkward, inconvenient and you have to do it multiple times to be sure you got a good reading. And worse of all, it's possible to forget to take a reading when you need to.

Tahoe slows it down, but Seqouia is still strong.

Yep, even 600GB/s is barely usable and not fun.

> Requires macOS 26 Tahoe or later

Why? Tahoe hasn’t even reached its first birthday and plenty of us are still on Seqouia not wanting to put up with apple’s disaster release. There’s nothing a terminal needs from Tahoe specifically.

7. Do you want the most contributions?

I hate to admit it, but GitHub has a massive community. I’ve personally seen a project leave it and then contributions dropped significantly.

The larger 3.6 35B is actually a mixture of experts (MOE). This means a small proportion of those B's are actually active. It's fast and suitable for agentic tasks but nowhere near good as the dense 27B model, which has all of its parameters loaded.

This. Despite my personal dislike of Python I would never discourage anyone from using it. It has an extraordinary ecosystem.

Seems to be in the same category as Lua. Can a PL nerd explain what's different about it?

> Can somebodhy explain to me why this isn't implemented yet?

Because right now it's a privacy and security nightmare.

Says it requires a 2TB disk? Must it be internal NVMe?

Have you heard about the cloud?

Reminds me of https://sharedrop.io

So many of these agent sandboxing solutions now. Don’t people search for existing solutions anymore?

What is sketchy about it? I think you know something the rest of us don't.

Leaving is not the hard decision. The hard decision is where to go. Do you have any suggestions?

I don’t think that would have worked. Parents buy phones for their kids since the kids can’t buy them. What would the choices be? Give them a phone or not give them a phone. I don’t think society is ready for that kind of choice anymore.

> Can I run something comparable to Opus 4.6 locally yet?

Sadly, no. The best comparable thing you can get is about Sonnet 3.7

If you show up to a protest then you automatically get put on a police database via facial recognition.

> It was still better than the competition

Plenty of cases where Surface isn't. Microsoft like to think they can make hardware but they're no better than other OEM and it's clearly not a focus for them

For the uninitiated, Service Processor (SP) is just another kind of Baseboard Management Controller.