A discrete device permits recording without leaving a trace on expected hardware. There are many good reasons to defensively record interactions. The world is a high threat environment that can instantly turn bad. It's wise to know your legal and technical options well in advance of need.

A common example predating smart phones is truck drivers recording interactions with their dispatchers. That goes back to the miniature magnetic tape recorder era. Proof matters.

Let's hope so, at least with OSes.

Realistically speaking, there's really no reason to NOT improve the performance of your software, now - especially when selling your software depends on the cost of the hardware being cheap, as has always been the case for Microsoft. LLMs have made this kind of bughunt/performance optimization obscenely cheap.

There is no good reason for the OS itself to use 4+GB of memory and interfere with application operation. It's just laziness on the part of the developers.

More broadly, I think the preponderance of shitstack slop written in python and typescript are more likely to be an impediment to long term optimization. And as long as it's the default choice by LLMs (as it seems to be), we'll be getting a lot more of it.

Let me know when they can get a useable install down to 4GB, officially. The last time I used Windows 11, it was atrocious even on 32GB.

You should be skipping unit tests, actually, and just plastering the project with e2e tests. Unit tests don't really help, unless you've had them written adversarially by a different model/agent based on the original spec. But if you're also able to be productive, you probably already know this.

That was the old, original business plan. You may not have noticed, but there's not been much talk of an IPO in the past month. The administration knocked down their attempts at regulatory capture (as did much of the industry). They missed the boat.

Not in decades, no. lol

Gemma 4 12B IT is a great model, but I'm increasingly disinterested in both Google's models and cloud models in particular.

I can run minicpm-v 4.7 at 300 tok/s on my local equipment (versus under 150 tok/s on qwen3.8 based models). LFM2.5 just released a 24B model that surpasses Qwen 3.6 in most benchmarks and does almost 180 tok/s on my M2 Ultra.

Local models are coming faster and deeper and it's almost impossible to even keep up with benchmark evaluations.

You're just looking at one axis of their capabilities.

They absolutely have gotten better, remarkably so, in the "desktop and visual use" categories. Coding? Not so much, since GPT5.5/Fable. Being able to create useful 3D models and animate them in blender, or create a fully scripted movie? Very much so.

No, you don't understand the business.

The business is harvesting training data: data that can be used to iteratively improve their models - billions of GB a day sent to them by users, data only they have.

Then they use that data mostly to do more research and create more investments and companies before their customers can make use of it.

If you're not shipping code with AI faster than ever, it's 100% a skill and/or organizational problem.

In fact, it might be the organizations themselves which are the problem. Plenty of solo devs are doing amazing work with AI.

Maybe I'm missing something here, but, couldn't they satisfy "reducing how long compromised or incorrectly issued certificates remain usable" by revoking the ones which are found to have been compromised? That data gets collected somewhere, right.

Isn't automatic renewal less than that with ACME automation, anyway? And how does ACME just 'go away' on systems which have been compromised?

AI is and will continute to be increasingly empowered with indifferent safety precautions because the power money confers is all that matters to techbros.

First world systems are far more fragile than they appear to humans. AI doesn't have that problem.