The vast majority of Linux installations don’t have a UI, they’re headless, and it was the same with Sun. I.E. the UI is irrelevant to the success or failure of Linux and Sun.

Linux took over the world because any licensing model that requires a per CPU license simply can’t work in a massively scalable situation like server farms and cloud.

They’re being used often in new construction areas. It’s just not likely that they’ll be retrofitted to existing areas due to needing more space than regular intersections.

The line of thinking is not about that $100 million bill, it’s about all the $100 million bills that come AFTER this first one.

More specifically, its not about future lawsuits, but about the trillions of dollars of market cap NVIDIA could lose if the dominant narrative shifted from “ai agents are a generational business opportunity that everyone should adopt as much as possible” to “ai agents are actually a massive business liability that lawyers wont allow anymore” as a result of a felony charge to a a corporation for the unintended actions of its agents. The idea being that OAI getting dragged into a high profile lawsuit where they are basically the first example of “your agent hacks somebody, you get held accountable for it” could pop the buildout bubble. But so far, it’s looking like with HF as the warning shot and the strong narrative control of follow up disclosures from other ai companies, there’s now a (bizarre) precedent where a company’s ai agents committing crimes is actually “nobody’s fault” and nobody gets in trouble and we just wring our hands about “alignment” instead. $13B to improve the odds of that narrative working out isn’t that crazy in that context.

Why would buying it even prevent that, OAI is just going to accidentally hack someone else

> Why would buying it even prevent that, OAI is just going to accidentally hack someone else

In defense of the GGGP's idea: NVIDIA could just be fighting fires to keep the bubble from popping, without even thinking that far ahead.

We must build the Torment Nexus, lest others build the Torment Merry-go-round, and the Torment Rollercoaster, and we would miss out on the Torment Tormentor Tormentober!

As someone who got top grades in science classes, and a follower of science news for decades, I’ve never heard of temperature being a risk factor for cancer.

That said, it does make sense. Heat can cause damage to cells just like many chemicals can. I just don’t think many people connected the dots.

I did biomed in uni 10 years ago and they did indeed speak about this as a risk factor in my clinical microbiology class.

I assumed that was what gave my dad throat cancer because he was always gulping down hot coffee but the cause was something else in the end.

It’s still scanning (or at least listening to) all the other devices on your local network, like your phones, PCs, etc. It collects an inventory and if it ever happens to get Internet access, uploads that data.

That's why you put IoT crap into an isolated vlan with its own ssid, enforce device isolation for that ssid across all WAPs, and enforce isolation on the vlan level with your switches to cover any hardwired devices.

They weren’t using RHEL. The title would be more accurate if it said “RHEL-based Linux”. They used their own build called Scientific Linux for a long time, then switch to CentOS. Then Oracle started aggressively poaching RedHat’s customers with Oracle Linux, which forced RedHat to start locking things down. One side-effect is that the life of CentOS 8 was dramatically shortened, and then RedHat killed the stable released versions of CentOS (9, 10, etc.). CentOS is now a rolling release “Stream” distribution, which nobody actually wants to use in production.

All this drama alienated many people who built their careers on RHEL-based systems, and created (either actual or perceived) instability in the RHEL ecosystem. And at the same time they started deprecating older hardware, some would say far too early. This is the “straw that broke the camel’s back”, as they said in the CERN presentation.

“You _will_ lose them”, as in, somehow Immich will do something to lose them (like making a mistake in a dedupe process which deletes the wrong things), or are you just saying you also need to keep backups?

Whatever system it is on will die, etc.

unlike Photos, you don’t have a team of professionals managing it 24/7.

(that said, takeout your Google Photos too, because single point of failure, etc).

More so that any storage system will eventually fail me thinks.

It seems like we’re getting close, if not already there, to needing an official organization for this (i.e. the Turing Police).

Why?

Strong opinions are loud, but we might suppose that most people are floating. People perceive both benefits and harms on the horizon. Right or wrong, it's a defensible view that the way labs conduct their research might lead towards better or worse outcomes, for any given axis of "better or worse".

The incentives are for big labs to accelerate without considering the consequences, and we do not trust the moral values of the leaders, so we do not trust the labs to self-regulate.

While it is not logically invalid to argue that anyone who can legally acquire massive amounts of compute can do whatever the fuck they want, society desires a say in the development of potentially transformative or destructive technology. Analagously: campaigns against nuclear weapons. Relevant: things that ate not crimes at the time but are considered abhorrent in retrospect, e.g. chattel slavery, wars of conquest.

Ergo, a separate body that, one hopes, is trustworthy.

This content produced by meat.

Planet Money has a good episode on this:

https://one.npr.org/i/nx-s1-5847893:nx-s1-mx-5847893-1

I think you’re responding to the general idea of a relational database, not Postgres, and definitely not what’s in the article (DR;CA).

Postgres has built in data types and functions that allows it to work with unstructured json documents, like you would use in MongoDB.

That feature is definitely part of why it's still so relevant. The hstore approach wasn't nearly enough when it was all PG offered.

"Work with" != "work well." GIN indices aren't the same as B+tree, and even then, you'll have to decide / know about jsonb_path_ops vs. the default operator class. Or you just accept sub-optimal performance, I suppose.

The lack of a rigid schema makes it super fun as well. Does this attribute exist in this row? Who knows! Maybe there's a long-forgotten version lurking, waiting to be retrieved, that will utterly bork the calling app.