Indeed, you are not obligated to use the Microsoft time servers in Windows. You really should use more local ones, as those have a much better idea of when DST events happen.

The maintainers of the Microsoft time servers have been spectacularly bad in my case, here in the South-Americas. Always timing issues when DST was applied. But after configuring 1.south-america.pool.ntp.org all transitions went practically always perfectly. So no, no more Microsoft time severs for me. However, time-zones do remain a problem, I am in the -03:00 zone, but Microsoft puts my actual location in the -04:00 zone. As this country stopped with DST in 2025, this has become a permanent issue, it looks like.

On a somewhat related sidenote: Getting rid of DST might sound like a good idea in theory. After almost 2 years of no DST, I would like DST to be re-instated. It really does mees up your sleep schedule/circadian rhythm.

Are those gas turbine not very inefficient power generation sources? As in consume a lot of natural gas and not deliver that much electricity...and all that making a noise requires ear protection when you are in the vicinity of 1?

And as these are apparently so inefficient, you'll know that datacenters will deploy many of those in 1 location...because they will have to. Anyone here ever lived around a small(ish) airport or military airport? Then you'll know how much and how far the racket of a single jet-engine can carry. Imagine 30 of these stationary jet-engines running 24/7.

Jet-engines also tend to generate quite a lot heat. So in an area about 30 miles around a jet-powered datacenter the ambient temperature increases about 5 degrees. Meaning that in summer you'll get and noise-pollution and a much higher electricity-bill as you will need to run the airco(s) in your home longer and harder. And not only the cooling system in your house needs to work a lot harder for a lot longer, the cooling systems for the datacenter itself also need to work a lot harder 24/7.

Jet-engines are a dumb way to power any type of datacenter. Relatively quick to build and quite reliable in operation, sure. But very costly, because you'll spend a lot of serious money on natural gas to fuel it 24/7 and not getting that much in return for all that spend, while polluting the datacenter and large surrounding area with much more heat and ear-destroying noise.

WiFi - in essence radio signals. Subject to all the benefits and disadvantages of radio signals. So, no Wifi in my home at all. Cabling is worked out of sight and I tend to work/play in the same locations, so I don't bother with laptops either.

An extra set of keys are in a lockbox in a safe and secure location outside my home. Furthermore., During my obligatory military service I have worked in locations that were deemed essential for national security and in control of top secret information distribution (either electronically or more analog ways). And that during war-time, so rest assured that I'm well aware of security protocols.

My post clearly didn't convey my intent that TPM and Secure boot are security measurements, just not as useful as you believe them to be. And yes, taking away hardware through military means is and always remains a thing for getting around security. Maybe you are lulled into a false sense of safety, just by thinking/accepting that attacks on security are only done remotely nowadays. They are not.

Again too many words to say: "if I can put my paws on hardware, the data contained on it is mine." The only thing you are correct about is the level of motivation. Remote security attacks require much less motivation than "local" security attacks. But once the motivation is there, TPM and Secure Boot won't help much securing your data.

Still, given the tone in your response, it will be safe to assume that we will only agree to disagree from this point on.

Sure, TPM and secure boot are being sold to you as security measures. And these are...secure-ish.

TPM 2.0 could already be brute-forced within 45 minutes with hardware from 10+ years ago. Now, with much faster hardware + AI...should be a whole lot less time. Because TPM 2.0 is already a very old security method and there is no TPM 2.0 or even TPM 2.0.1 being planned either. This update cadence is way worse than OpenSSL's cadence. Or TLS's cadence for that matter.

Secure boot is in a nutsshell a sandbox that verifies if your computer's hardware is allowed to start up it's operating system. Sounds great in theory, I'll grant it that. However, it is in full control by Microsoft. So a single layer of security. By definition: "a layer of one == a layer of none".

In theory one could say that TPM and secure boot are security measurements. In practice, each is much less effective as the narrative wants us to believe. When combined, things don't fare that much better.

The "bonus" for Microsoft is that with those security measures, they have far more control over your computer than anyone should want. Microsoft hasn't had a great track-record in this field. Add to that the severe lack of progress regarding TPM...then I won't describe this as security, only as secure-ish. Because it isn't deserving of more.

But that is mostly caused by NVidia's lackluster (yes, I know they are improving somewhat, but still) driver support in Linux. I have 10 Proxmox nodes deployed. 2 of them use NVidia GPUs. Not even good ones or modern GPUs, but GPUs you would expect the kernel to support for several years already.

The nodes with NVidia GPUs barely manage an uptime of 3 days. The other nodes use either Intel or AMD GPUs. And those run solidly for months. I have spent time migrating Proxmox from v8.x to v9.x as instructed, but it doesn't matter. The nodes with NVidia GPUs remain as unstable with Proxmox v9.2.x as they were with v8.4.x

Sure, there are "gaming"-distros and the main distros (+ derivatives) might work ok-ish with NVidia GPUs, I'm already aware. Still, I imagine NVidia to be the proverbial red-headed step-child under the care of a spiteful red-headed stepmother. And no, not in the way some 'hub of corn'-website wants you to believe, but more in the style of fairytales from the Grimm-brothers.

DDG is using Bing as underlying search engine, as far as I know. So no wonder that DDG isn't doing searches all that well anymore. Here in South-America, Bing always has shown disappointing results. Since their inception, really. Maybe that it was different if one hailed from 'freedom country' and that it has become as useless in the US as it has been everywhere else.

Most of the income OpenAI and Antheropic do generate seems to come from other AI startups, like for example: Perplexity. Seem the income from the frontier labs is divided as 90% coming from other AI startups, about 7% from 'pay-for-use'-customers and the remainder is coming from customers with a subscription.

If you pay for tokens from any cloud-AI provider or pay for power to run your own LLM(s) with this new software...I don't see that much difference to be frank.
In both cases I lose money and in both cases I have to hope that financial loss produced results that makes me more money than I spent.

Local is more Private and I happen to have a large roof...well, more than one, completely filled with solar panels. And I live in a place with a lot of sun-hours, so on sunny days I really don't feel that much of a financial burden running my own LLM cluster. And there is still room left on anther location on my terrain, where I could place quite some more panels. Those cloud-tokens may be faster and perhaps even of better quality, but me spending on those is relentless. And I will always have questions about privacy with them.

Nah, I like LLM solution where I am more in control of. And to be honest, the Qwen 3.8 LLM (27b) I can run locally, generates decent quality tokens. Having a few of those instances managed by a piece of software NVidia is now offering...in a location where I have much more control regarding energy-safety and privacy, sacrificing speed and a little quality, it really doesn't sound all that bad to me.

I recently saw a company's bill for API access to cloud-AI (routed over Claude, Codex, Gemini and Grok) with some 10 very active AI users for a single month that ended up to be almost 18.000 Euro, close to 20.000 USD in freedom money. Who knows how much that company spends in a single year on AI. Even that monthly bill would already give me enough for all the panels I would need. Another similar month would pay for quite some GPU upgrades, even with the excessively inflated hardware prices of today. A 3rd month like that and I would have enough money to pay my yearly energy bill while running my local AI cluster full tilt for 24/7. The rest of the year...gravy/profit?!?

Cloud-AI, especially from OpenAI and Anthropic, is expensive. But with their totally unsustainable token pricing from today and in the past, it is worth it. However, if either OpenAI or Anthropic has to become financially sustainable, let alone profitable, their current pricing is simply far too low. A tenfold in token-price would do it, but at that point a local LLM cluster becomes financially far more interesting and not without too much loss in token quality. Especially if you start with such a cluster now and give it opportunity to apply reinforcement learning on itself for your specific use-cases and workloads.

A far more sustainable AI/LLM future, if you would ask me. And yes, I am aware you didn't.

Your statement is absolutely correct.

But that doesn't mean China is above fueling that particular fire. Or that they aren't doing that out of spite.

All those free and near free models coming out of China, you think they do that to "stimulate openness"? Nothing more than PR-speak for spite. And as a bonus, every investing-fool it putting far too much money into AI, which is money that could have been used to protect countries against the Chinese onslaught. Either against their ideology or against their financial reach or against their ever increasing military might. Probably a combination of all 3 will cement their place in global history for the next 100 to 150 years.

The Chinese never plan ahead for the next 2 quarters, like most of the world is cajoled into doing. They plan years ahead. Which makes strategies like they show in this day and age possible...and way worse, sustainable.

The decline of the US empire already started at the end of Obama's second stint, Trump 1 threw that decline onto a kayak through the Grand Rapid, Biden navigated back to a less wild river, to make it easy for Trump 2 to dump the decay into a barrel and throwing that barrel in Niagra Falls...

To my understanding, Oracle didn't just bet everything on AI, they bet on OpenAI specifically. Every other hyperscaler placed their chips on both OpenAI and Anthropic, Only Oracle didn't.

And yes, when OpenAI falls, it is bound to take Oracle down with it. As a whole, not just their cloud division, perhaps even reduce Ellison to a mere millionaire. Especially as the Paramount-Warner Bros merger deal has hit some snags and Ellison appears to have put up his own wealth as collateral with that deal.

It comes in the form of a chip? then it could be housed in a cartridge-type form and then exchanged just as easy as those were with the Nintendo consoles of yesteryear, the C-64 etc. At least with those 2 types of computers I have personal experience running cartridges and it was already fine back then.

Any line carrying 50.000 Volt or more, those hang in the air on pylon masts all over the globe, including the EU. Those are the lines that do not tend to break that often. And air-cooling is a must. The transmission lines that carry less than 10.000 Volt, those do tend to break for many, many reasons when hanging in the air.

The sections of Belgium, France, Germany and the Netherlands that I frequented, all those "small" lines are all underground. I'm sure you could find sections of the electrical grid hanging in the air when traveling to backwaters of these countries. But in larger villages, cities, industry-parks etc. you'll find that the grid is buried. Because people need energy-security. Most countries in the EU have actual laws that make sure of that.

Say what you want, the western parts of the EU decides to bury most of their electricity grid, making weather far less of a problem than it is for any country that is not inclined to do this.

Investing in home batteries seems like a wise idea. Even more beneficial if you can aid the charging of this home battery with solar panels. Because it becomes clearer that the centralized generation and distribution game is too brittle to maintain, Weather, but also greed, are very powerful foes that can cripple any grid.

Unless you do it properly and bury your electrical grid as much as you are able, to eliminate at least the weather as a foe.

Isn't New York not a state located quite close to the Canadian border? Would that not be a good location for energy-availability and -security? Building is something you do once. Having access to power from US and Canada, will make for a continuous functioning datacenter. Which is a quite important factor in profitable exploitation of such any datacenter.

As I understand it:
The Chinese AI labs employ only (very) smart Chinese people, so 100%. In all of the AI labs from the US, 35% of employees are Chinese as well.

So, if the West still wants a part in the future of the next generation and beyond, the West should look exactly how the Chinese teach their children and follow that example. As a matter of fact, we (the West) were already doing that, but let the teaching standards slip dramatically and introduce computers/tablets in their classes and now also AI on those devices...to add insult to injury, so to say.

In my opinion, an AI ban on digital devices in schools that teach to kids between 4 and 13 would be be the best thing to happen to those kids. Dropping of the "new math" and/or computers/tablets in such schools also great. Even learning how to write in cursive on paper with an utensil should be encouraged, as this makes students understand their (native) language so much netter and as a bonus improve their reading skills leaps and bounds. All skills that form the best base to tackle computers/tablets and AI. At 13 kid's minds are still sponges anyway, so they will pick that up fast.

For those that can, do work with computers/AI etc. For those that can't (or simply don't care), there are still many jobs in society that will be great for them to build productive lives.

What should be made easier, is learning later in life. Not everyone reaches their peak thinking/reasoning performance early in school. There shouldn't be a stigma on those persons that want (and possibly could) get "up to speed" later in life. And AI could actually be a great help/tutor with that goal. But first, schools should train young minds to accomplish as much as possible, under their own power...just as education used to be once in the West and still is in China.

With this ban Norway seems to be heading towards a path that keeps that nation still relevant on the global scale 50 years from now, with the people that live there and then.