It ain’t over until members of DOGE and Musk have been put to sleep. They will just do this again, when the Regime needs vandals to do the dirty deeds.

Windows itself is quite good. The big problems come from who is managing it. If Microsoft wasn’t invasive, I would gladly throw a couple hundred bucks at them and get on with my computing life.

My latest version of Thunderbird comes with the far right side panel being used to read email, the center panel a list, and the left is my boxes.

If I were god king of (USA/Europe/Asia…the Universe), I would subsidize repairable laptops. It would save resources in the long run, for both humanity and the planet.

Here’s hoping that the bubble pops. I want to fill my machine with RDIMMs, and swap all of my SATA SSDs with 8tb 2280s.

In theory, completely rewriting legacy software with AI could be a great thing - doing so every 3 or 4 years, allowing newer and better AI to tackle the task, optimizing for every development made in technology or standards.

However, this still requires human minds to define, guide, and verify these projects. AI opens up possibilities for better development, but executives seem determined to fritter away such potential by axing the human element.

It is like a certain story…

I think AI is cool and useful, but it doesn’t obsolete people. AI doesn’t have hands, nor does it have social circles of experts to rely upon, or lived experience.

Closest thing to that is using Windows Enterprise. Here is previous PSA of mine…

00000

PSA, for people sticking to Windows:

You can get a reasonable level of privacy by installing Windows Enterprise via RUFUS, which also has options for removing restrictions during installation. Massgravel is used to activate your copy of Windows, the Github also having .ISOs for you to use with RUFUS.

ShutUp10 is a piece of software that goes a step further, allowing you to toggle off many bad things, uninstall Microsoft’s AI, and gives a description of what you are tweaking does. The premium version also automatically applies your settings at all times, reverting Microsoft’s constant tweaking of your settings.

RUFUS

Massgravel

ShutUp10

I am guessing this is to amp up the stock price of the GrokSpaceLink IPO.

GrokAI is a poison pill that has been fed into a healthy rocket-satellite company. Sucks for the folks who value practical infrastructure that contributes to society.

A law should be passed, where people who blind a Palantir cannot be prosecuted.

I am figuring on selling off my previous machine once the teething issues with the new one has been resolved. Given these specs, what would be a fair price, Lemmings? I was guessing around $2,250 without the drives, but I am not really on the pulse of the market.

00000

• CPU = Ryzen 5950x + Noctua heatsink with Kryosheet.

• PSU = SeaSonic Vertex GX-1200

• MOBO = Crosshair Hero VIII + WiFi

• CASE = Rosewill Thor V2

• RAM = DDR4 3600, 128gb total, came as a kit of 4 sticks.

• STORAGE (OPTIONAL) = x2 to x5 4tb SSD SATA drives, non-NVME.

00000

The biggest issue here is doing an RMA on the CPU. It is unstable when PBO is enabled. Since I suck at techno surgery, I wouldn’t be able to install the replacement without forking over money to a technician. Maybe I can make that contingent on buying at least 2 SSD SATA drives?

My new machine has only 4 SATA ports, and the SATA card to support all five SSDs and the optical disc drive is taking up a GPU slot. Being able to buy some 2280 M2 drives would solve that problem for me. That would gel nicely with a technician handling the transfer of the SATA SSDs back to the old machine for the new owner, since the tech can also get the M2 and a third GPU into my new PC.

Courtesy of Kagi’s search AI:

000000000

It appears there is a slight misunderstanding of the acronym: Yann LeCun’s architecture is called JEPA (Joint Embedding Predictive Architecture), not JASP.

JEPA differs from Large Language Models (LLMs) primarily in how they learn, what they predict, and how they represent the world. While LLMs are “word models” designed to generate sequences of tokens, JEPA is intended to create “world models” that understand the underlying physics and logic of reality 3 4 .

The key differences are summarized below:

1. Generative vs. Predictive (Non-Generative)

   LLMs are Generative: They operate by predicting the next token in a sequence (generative AI) 2 . This approach often leads to hallucinations because the model focuses on statistical probability rather than factual ground truth 6 . JEPA is Predictive: Instead of generating every single pixel or word, JEPA predicts latent representations (embeddings) in a hidden space 5 . It tries to learn what is “plausible” rather than attempting to reconstruct every single detail of the input.

2. Word Models vs. World Models

   LLMs are “Word Models”: They learn from text and treat intelligence as a language manipulation task 4 . LeCun argues that language captures only a small subset of human thinking and cannot represent high-dimensional physical spaces 2 7 . JEPA aims for “World Models”: It is designed to understand cause and effect, physics, and the physical environment 1 . This allows the system to reason from first principles and plan sequences of actions, which is a prerequisite for autonomous AI 1 .

3. System 1 vs. System 2 Thinking

   LLMs (System 1): LeCun describes LLMs as “System 1” processes—they are reactive and perform a fixed amount of computation to produce each token 2 . JEPA (Path to System 2): By incorporating world models, JEPA is intended to enable “System 2” thinking—the ability to plan, reason, and deliberate before acting 1 .

Summary Comparison Table Feature Large Language Models (LLMs) JEPA / World Models Core Goal Predict next token (Text/Code) Predict latent state (Reality/Physics) Method Generative (Pixel/Token by pixel/token) Joint Embedding (Non-generative) Domain Linguistic/Statistical patterns Physical/Causal understanding Weakness Hallucinations, lacks physical grounding Limited fluency in natural language Cognition Reactive (System 1) Planning/Reasoning (System 2)

Around 4 years. The creation of any microchip requires extremely stable geology, pristine rooms, high levels of training, Scrooge McDuck moneybins, and so on. It isn’t an industry that can turn on a dime.

Wizard Pope didn’t specify which head. The rope and the pruning shears are not mutually exclusive.

The internet survived the Dotcom bubble. Many businesses died, but some survived. The glut of hardware from the dead will end up in the hands of individuals, upstart companies, and the corporations that outlived their peers.

In any case, I believe the definition of frontier AI will change, as would the hardware. My build is based on what local AI is available today, with some wiggle room left over. I believe that it is around 2030, when DDR6 and other major shifts are likely to happen, that we will see the definitions change.

In any case, a Q4 of GLM 5.2 is about 460ish gigs of VRAM+RAM. While expensive, that ain’t out of reach for an AI hobbyist.

White Hats can just set a script kiddiez AI to the task of jiggling doorknobs and seeing what happens. The blade cuts both ways. In the long run, I think automated defense testing will make cybersecurity much better.

There is a key difference, that makes the genie impossible to bottle: People can run local AI on their own machines. Fans of nuclear power can’t easily build nuclear plants in their backyard. A pity, the world could use more nuclear energy. 😔

Anyhow, I am looking forward to someday using frontier-grade AI on my PC. Just need the AI bubble to pop, so that I can afford the terrabytes of memory that would be needed to comfortably run it.

I bought some DDR5 RDIMMs, 2x16gb 5600mhz. Cost me $606 at Ebay, but I haven’t yet got a chance to install and test them. Word of warning: stay on top of your shipping order, and if the seller is slow to ship or the shipment is late, message them and open a case with Ebay.

Hopefully my sticks are the real thing, but the biggest problem is the belabored shipping. I bought a Threadripper 7965WX on Ebay for $1500, and the return window is closing. It sucks when a seller holds up an entire project by being dumb about getting stuff mailed.

I tried buying some rather cheap DDR5 RDIMMs through Alibaba. The process to actually buy is pretty vague, and I found that the suppliers I contacted really suck at communication. The two I spoke to, didn’t acknowledge the details I was actually asking about, and spoke broken English at best. The shipper I spoke with, Cooperate Logistics, had fairly clear communication, but was very slow at conversation - it took about two or three days for our dialogues to conclude. They tried to buy the goods on my behalf, but they couldn’t - the supplier had to send the goods over to the shipper’s place, THEN it would go to me.

There is potential in buying from Alibaba, but many of the suppliers feel sketchy, and the bad communications make it hard to feel trust. I ultimately gave up. 😔