It's a great cover.

I don't have any illusions about whether the media should be unbiased. If someone is a child rapist then the media should call them a child rapist, even if they are the president.

Those two things are synergistic, though, not separate.

Well, you could also put it in a metal[lized] container.

There has never been and possibly never will be a source of only facts.

Even if there were, which facts were covered and which were not would itself be a form of bias.

I am very in favor of facts, and outlets with just facts existing, that's a great idea. But how are they going to pay the bills? And what do you think most people will be able to do with raw, contextless facts? Or, how much factual context do you think they should provide? And down the rabbit hole we go!

I think there's probably a lot of potential primary reasons to record something. Sometimes audio is ephemeral and your goal is text output, in which case the primary goal is recognizability.

However, ogg is a container, and it can contain FLAC.

...ok, and just got done building and running the C version. Got some errors like

(garagejam:23589): Gtk-WARNING **: 12:51:54.197: Broken accounting of active state for widget 0x5579c0ad17c0(GtkBox)

but it seems to be functioning.

I hit record and it opened up a track and found I think my webcam. Or, he said up his sleeve, my LG TV. It's quite opaque, though. There's no apparent audio device selection, though I have several inputs to choose from. To assuage your concerns about audio fidelity, it actually tells me it recorded the track as a .wav. At the bottom of the screen it says it's 48 kHz. It puts the files into Music, where your music library normally goes, I'd put that in Documents but anyhoo.

track-3-take1.wav: RIFF (little-endian) data, WAVE audio, Microsoft PCM, 16 bit, stereo 48000 Hz

Thanks to it being based on a framework (GTK4) I can right-click on text-fields and insert emoji, but I still can't find anywhere to click to select audio devices. I'm going to say you can stop trying, but not for the reason you were thinking.

Restarting nuclear power plants should also take a long time, if not quite as long as building them, to be safe. It will be interesting to see how that goes.

On one hand, the bulk of the vulnerabilities are a fart in a windstorm. On the other, the AI has the patience and speed and parallelism to check many things at once, and pluck fruit humans wouldn't have. Therefore more of these difficult-to-exploit vulnerabilities are going to be exploitable in practice than have been in the past, and it's no comfort if it's difficult and rare if it happens to you and it's the beachhead for an invasion of your network.

So I do think that it is accurate to say that it is "finding vulnerabilities at a rate we humans couldn't", and while yes it's accurate that "competent software engineering would have avoided in the first place" it's also true that there's a lot of incompetent software engineering occurring. It's reasonable to believe that there's going to be a period of pain as a result, and we are already seeing some of these effects.

While it would be nice if only high quality software ever came into being in the future, since that's not going to happen, maybe it's beneficial if "AI" helps some people make their software more secure. It does seem like there's lots of other ground to cover before doing that, though, and maybe the best use of the AI is to tell people to fuck off until they've learned about the no doubt highly bullet-pointed and italicized X, Y, and Z.

You don't get anything for free.

The obvious thing to do with all AI content is to have AI check it. But every time you do that, you're spending more computing resources running more queries. That's why humans write tests, so they don't have to test functions themselves. If you are hoping to check your work as well as a human would, you're going to have to re-check at least the important (tricky) lines in the context of some number of the preceding and following lines both in review and as you write the code, because that's a thing humans do. So far you can increase your chances of getting it right "the first time" only by making "the first time" more recursive and expensive, and it already costs too much.

TL;DR: No.

I picked one thing at psuedorandom to check up on:

Another analysis earlier this year by researchers at the Massachusetts Institute of Technology found that, between 2010 and 2024, data centers increased average electricity prices for residential customers by just 2.1%.

This isn't random because when I see "average" I also see red.

What does the study actually conclude?

Data center entry increases average retail electricity prices by 2.7 percent. Residential, commercial, and industrial rates all increase, with larger percentage effects for commercial and industrial customers. These results indicate that the costs of accommodating data center growth extend beyond the facilities themselves and are passed through to incumbent ratepayers, including households and existing firms.

2.7%, not 2.1%, and we pay for it whether it's up front, or on the back end. I also liked this part:

Regulatory environment further shapes the magnitude and mechanisms of pass-through. Price effects are generally larger in deregulated states[...]

Enjoy, Texans! Though perhaps that's something they can address now?

Rather than go on to check on other "citations" used in this story, I'll let someone else do that (or maybe do it in a separate comment, if inspiration strikes) and instead close this comment with this paragraph from the fine citation:

Finally, the distributional consequences of data center growth extend beyond differences across customer classes and utility institutions. Price effects are larger in some places that already faced higher electricity prices or lower levels of economic prosperity. These results are suggestive rather than definitive evidence on household-level energy burdens, but they underscore that the costs of infrastructure-intensive growth may fall unevenly across communities. As data center demand continues to expand, especially with the newest wave of hyperscale facilities, the design of tariffs, cost-causation mechanisms, and regulatory cost recovery will become increasingly important for determining whether the costs of new infrastructure are paid by large-load customers, incumbent ratepayers, utility shareholders, or taxpayers.

Can you name a news outlet which has only facts?

Same ground clearance as a Megane, less than a BMW 1 series. Not high up. Has an upright seating position but not particularly tall.

Most actual hatchbacks even in Europe are plenty low, I've read the specs. But you've also got a whole class in Europe of "people movers" which are taller which doesn't even exist as a class in the USA, though some of our CUVs are definitely representative of that idea — mostly Hondas. Here, if we want that we generally use a minivan, because we pay less for fuel (not so much less right now, but still generally less, especially outside of California) and because we don't pay such an insurance premium penalty for vehicle mass.

My theory is that this is basically the leveling lift taken to a ridiculous extreme.

Right. For those who don't know, pickups are normally built like that and then people put "leveling lift" kits on them which lift the front to match the rear. If you never put anything in the bed then this is a cheap way to get more ground clearance, because there's less in the front. (Except for the diff, yeah, but you can either get a shield for that or just steer it around obstacles.) But then if they put stuff in the back, the front of the truck is sticking WAY up in the air, and their headlights are shining in your face. They never, ever re-aim the headlights.

Automakers make what will a) sell and b) make the most profit, and most individuals are buying vehicles based on style, while fleet buyers are buying them based on estimated TCO over an approximately ten year period. So a genuinely good pickup which is very reliable and is also very useful to an individual tradesman and also very maintainable for a long time is a very hard sell. Fleet buyers don't want to pay for vehicles designed to last 20+ years, and they buy the bulk of the base model "tradesman" type pickups. That's why those vehicles are scarce on or even literally absent from showroom floors — often those configurations are only available to fleet buyers, because there's not enough per-unit profit in them to make it worth it to sell them to you or me. Consider the tale of the Nissan NV200.

They always had different frames and axles, OBS trucks only had the same bodies, which wasn't true before that body style and then wasn't true again after the OBS. OBS super duty models also only used a frame stiffener rail instead of a unique frame, but they did use a unique rear axle (Dana 80 instead of Sterling.)

That seems ultra weird. Why would it be hard to make a RHD model of a vehicle with no steering linkage?