380 karma · joined March 5, 2025
Secure Boot being "a choice" on PC is an exception, not the norm. On just about every other device, the vendor is going to take a boot, shove it up your ass, and say "it's there to make your ass more secure" if you complain.
Payment processors have ways of passing some of the chargeback risks onto the stores, and it's not like Steam itself is chargeback central. If you just want free games, pirating them is extremely easy, and trying to abuse chargebacks gets you banned.
The reason why Apple or Nintendo go out of their way to make this impossible isn't user security. It's the "security" of their 30% App Store cut.
Out in the wild, Secure Boot exists to "secure" vendor revenue streams - and PCs are the only devices where it's even possible for the user to disable it. Most of the time.
What's happening in smartphone space is enough of a reason to treat Secure Boot on PC like an ongoing attack. The only reason why there are still legitimate ways to disable or adjust it is that most PC manufacturers don't have their own app store.
"My computer was compromised with an early boot stage hypervisor backdoor" happens basically never. It's an attack vector that exists almost entirely in the minds of infosec fucktards.
"My brand new device ships with vendor-selected boot certificates that can't be changed, can't be overridden, and control what software I can install onto my own device" happens with every other smartphone, gaming console, car, and even some PCs.
"Trusted Computing" is, and always was, about making sure that the user doesn't actually own his device. This is the real, tangible attack vector - and the target of this attack is user freedom and choice.
And then they just... gave up?
I don't know what happened to them. When AI breakthrough happened, I expected them to put up a fight. They never did.
We've been in a continuous AI breakthrough since 2022. In 2000, this kind of thing was nonexistent.
What you're missing is: today, we're nearing the point where actual general purpose robots become viable.
Which means: the purpose of a robot is no longer to sit at a factory line and precisely execute the same exact motions on repeat 24/7. The purpose of the next generation of robots is to learn generalized behaviors, adapt to circumstances, and carry out circumstance-specific actions with active sensor feedback. Which means completely different requirements for effectors.
Which means: repeatability can go get fucked, for one.
We also keep training LLMs to work with greater autonomy, on longer timescales, and tackle more complex goals.
Whether LLMs are "actually thinking" or have "volition" is pointless pseudo-philosophical bickering. What's real and measurable is that they are extremely complex and extremely capable - and both metrics are expected to increase.
If you expect an advanced AI to pose the same risks as a faulty thermostat, you're delusional.
My point stands though: no Artemis mission has plans to launch a full Apollo style capsule + lander stack. Artemis HLS, SpaceX and Blue Origin versions both, flies entirely on its own. So what's required of both SLS and any would-be SLS replacement is a far less demanding mission than what Saturn V has done in the past.
Which calls into question: what SLS is even for? Does it add value to the mission, or is the mission subtracted from to justify adding SLS to it?
>Simply put these systems are NOT dynamic, they can not modify based on experience, they lack reflection.
We already have many, many, many attempts to put LLMs towards the task of self-modification - and some of them can be used to extract meaningful capability improvements. I expect more advances to come - online learning is extremely desirable, and a lot of people are working on it.
I wish I could hammer one thing through the skull of every "AI SAFETY ISNT REAL" moron: if you only start thinking about AI safety after AI becomes capable of causing an extinction level safety incident, it's going to be a little too late.
A hitherto unknown limit to scaling laws? GPU compute being discounted - as opposed to in a massive fuckoff "people would kill for an H200" shortage at all times?
What are your sources on that? A fat fucking blunt?
That deadline is never going to hold - too many things are just nowhere near ready. By now, I expect NET 2030.
Now, it seems like even that might end up not happening. What a shitshow.
Saturn V was less powerful than SLS - but it could send an entire mission in a single launch. Capsule, lander and all.
A lot of what NASA has been doing with SLS is just trying to... rationalize its existence. This is what gave us NRHO, Gateway and others.
Reportedly, some of the people at NASA just believe that having an inefficient, wasteful and corrupt space program is better than not having it.
If he claims that giving LLMs a Python interpreter is a huge win for his paradigm, then major AI companies have been "winning" since 2022.
"See, LLMs that are allowed to use Python perform better than ones that aren't, and Python is symbolic, so I was right all along!"
Looks like a surrender to me.
Most top-of-their-field researchers are on top of their field because they really love it, and are willing to sink insane amount of hours into doing things they love.
It starts off being wrong ("Opus 4 has maxed out LLM coding performance"), then keeps being wrong ("LLM inference is sold at a loss"), and tries to mask just how wrong it at any point in time is by pivoting from one flavor of bullshit to another on a dime, running laps a manic headless chicken.
Can you stop it?
You still need the Starlink coverage to be enabled on system level, and the local government to care enough to track down Starlink users at the same time. I can't imagine that there are many countries that meet both criteria.
If we're using relative addresses, this would, of course, result in 69 different addresses to compress - each relative address being the difference between 0x1337 and the position of the code that calls it.
If we're using absolute addresses, we get the same exact address 0x1337 repeated 69 times - which is way more compressor friendly.
Probably the single oldest trick in the code compression book.
"Good company" is subjective, but to argue that the company that built the backbone of modern web didn't make anything novel or monetizable is a bit short-sighted, don't you find?
There are still massive gains to be had from scaling up - but frontier training runs have converged on "about the largest model that we can fit into our existing hardware for training and inference". Going bigger than that comes with non-linear cost increases. The next generations of AI hardware are expected to push that envelope.
The reason why major AI companies prioritize things like reasoning modes and RLVR over scaling the base models up is that reasoning and RLVR give real world performance gains cheaper and faster. Once scaling up becomes cheaper, or once the gains you can squeeze out of RLVR deplete, they'll get back to scaling up once again.
They can sit on their asses, but going against people who actually try to do something to address climate change is a step too far.