You don't know if this is a Sherlock Homes mystery. Signs of suicide obvious on the scene might be foul play. My news usually goes "Person X was found dead at home, was suffering from depression, investigation on-going", not calling it "suicide" until after investigation is complete. There was a recent case with a politician in my place. First day "found dead", second day "found hanged", days later "suicide" (as per Prosecutor's office conclusions).

TFA says the videos were some sort of test before dive; during dive the camera was configured for external acquisition, not for dump to SD.

You're joking but latest cat litter machines have cameras. Here from Petkit https://petkit.com/products/pu...

The camera is useful if you have several cats, so you can then check on them while travelling. Cats might get stuck in a cardboard box or push a door and be stuck in a room. A cat litter machine without camera distinguishes them by weight, but that might not be enough to make sure they are all ok if two or more have similar stature.

Their cat litter includes pH colour indicator so they also help checking on the cat's kidney health (which is critical in cats).

Alternative is camera on the feeder, but you can have both, a camera is cheap and those machines are already cloud-connected (which is important; when travelling I want to check the feeder and water pump are operational and the cats have been showing up for food and poo).

As a language models, GPT-5 finding references that solved Erdos problems that the professional mathematician and maintainer of the Erdos problem website was not aware of, shows an actual useful skill for the LLM.

Microsoft has been maintaining a linux distro for the past 5 years, CBL-Mariner recently renamed Azure Linux https://en.wikipedia.org/wiki/...

Don't be so sure. Most people use it because it's in their fibre gateway, router, smart fridge, the wired SIP phone at work that has a single line display, whatever, it's everywhere.

Your proposal is more difficult, less effective, and legally riskier:
* More difficult, because you need to stay awake, wait for the car to enter, then put the cone behind; then repeat the procedure with more Waymos; then herd your angry circling reverse screaming Waymos so they stay stuck behind a single cone. (Maybe someone will vibe code that game.) In the other option you put a cone and you're done.

* Less effective, since the whole purpose was to sleep in peace, and you're not sleeping. Plus the reverse noise still happens so your family and neighbours aren't sleeping either. In the other option everybody sleeps.

* Riskier, because you're then messing with traffic and freedom of movement, Waymo and the possible passenger might sue you. In the other option, you're objectively helping Waymo and the passenger since they likely never intended to drive into your dead end, so it's safe.

Check the World Risk Report, so you then can factor natural disaster risk when choosing a place to live. in https://en.wikipedia.org/wiki/... (2021) or latest edition https://www.welthungerhilfe.or... (2025).

That was last century. Right now all new chipsets (in particular for communication) go in the linux FIRST (before Windows); often before the dongle product is even released. Chip manufacturers do that so phone makers can use them in their next designs.

Last time I compiled drivers... January 2005, I emailed HighPoint Technology as my Asus motherboard had a HPT371 RAID controller, and they kindly replied with the code for the 2.6 kernel. But then I did not need it, linux software RAID (portable to any distro) is a better idea than tying oneself to a particular hardware controller.

Teams-for-linux https://github.com/IsmaelMarti...

The problem is the big distributions persist at promoting GNOME as the default option despite a long history of questionable UI decisions. The constant support enables this sort of decision making at GNOME.

(Though I don't understand much,) I would imagine the deconvolution would be normally done through inverting the point spread function at each pixel. Most likely they need an iterative solver and it can be cumbersome to compute on large images. So instead they let the machine-learning model fill its weights on known cases. In the end it's still matrix algebra, just a linearized method that less expensive computationally.

My concern is my own privacy, and traceability of the police action (even assuming blameless law enforcement and justice. Related to privacy, I want to know when I'm being filmed, to adapt my behaviour (avoid being in an embarrassing situation). Related to police action, accountability is only possible with our ability to check the legitimacy of their action.

I would agree if the police asked me to share a livestream of smartphone or dashcam so they can identify bad guys in the street. I wouldn't want Google/Apple to sign an agreement with law enforcement to remotely enable smartphone cameras as they please.

You probably would want to know and validate the request, even if you always say yes. Same as the difference between regularly donating small amounts to a homeless man, and that man pickpocketing you everytime you pass by. Even though I'm going to say yes without a second thought, you have to ask every single time before taking, using or accessing my belongings.

It might not be new but it's a timely repeat. Next year the European Commission (EC) will review its current objectives. The legacy manufacturers are doing a lot of lobbying right now (such lobbying efforts are in the news right now in France and Germany) and are pushing plug-in hybrids as an alternative solution to the full ban of ICE scheduled for 2035. You and me know hybrids would defeat the purposes of the ICE ban, which is reducing greenhouses gas emissions and reducing dependence on foreign countries (e.g.OPEC). Studies like this one, even when not new, are important to keep scientifically correct information in the public spotlight, and on the table of government officials and lawmakers, in view of upcoming political debates.