Yeah, that was a typo in my submission. I should have said 2021, not 2024 given that the REDIT post was '4 years ago'. Sorry.

I guess the logical next step is to capture the heat output as hot water, concentrate the heat somehow (or heat the water a bit more) and use steam to drive a turbine producing electricity. Ye cannae break the laws of physics, but it should be possible for a datacentre to recoup at least part of its electricity costs this way? Essentially a steam-driven power station where the heating element is a bank of GPUs with water running over them.

It's all about theta waves. Those who have become "clear" and can harmonize the frequency of their theta are known as operating thetans.

They took things they knew from the outset they didn't have a right to, like the oft-discussed Books3 database. They knew it was pirated, had an email chain discussing paying for the books, and decided to use it anyway. It was a wilful disregard of copyright law because it was faster and easier to use piracy for profit.

There are a TON of differences. Probably the biggest is that the machine version can read the entirety of all known creations.

Humans can study some a book in a few days, watch a movie in an hour or two, a web page in a few minutes. Machine learning can pull in thousands in the time it has taken you to read this.

Similarly for output, writing a book takes months to years, staging photos takes time and tools, feature films are hundred million dollar multi-year endeavors.

The human cost is a huge part of the economic difference. The AI industry has made fortunes by sweeping in everything ever created, authorized or not. Companies like Meta now have email trails showing they could have moved for authorized access, but like a thief that it was easier to just grab known-unauthorized materials and profit immediately rather than compensate people for the use.

Combined the two are unacceptable. They could pay but they refuse, they claim the only way to operate is mass infringement on the scale of all humanity, that if they don't get unfettered access to everything humans have ever created, without compensation, so they can maximize profit.

Well I know that you can't argue over personal tastes, and many people like modal editors, but I don't think it is about "educating yourself". Perhaps the opposite is true, as this interview with vi's creator, Bill Joy, explains:

REVIEW: What would you do differently?
JOY: I wish we hadn't used all the keys on the keyboard. I think the interesting thing is that vi is really a mode-based editor. I think as mode-based editors go, it's pretty good. One of the good things about EMACS, though, is its programmability and the modelessness. Those are two ideas which never occurred to me.

I never had a Surface Studio. But I always wanted one for its 4500x3000 display. Microsoft did a good job in pushing 3:2 aspect ratio and driving the PC market away from the horrible letterboxing that dominated laptops and monitors for a decade. It's a pity that panel was never sold in a standalone monitor (Huawei talked about it but the product never reached the market).

Did you try asking one chatbot to check the quotations given by the other chatbot? If you ask the AI to find something then it will do its best to please you. But if you ask it "is this quotation real" or "is there any evidence for X" then at least some of the time it can perform the useful service of saying "no, can't find it".

I upgraded my video card recently. I need four DisplayPort outputs so I picked a Nvidia RTX A2000 (old generation, not Ada). The prices on ebay.co.uk looked good value. Then I looked at the seller, and he had about a dozen of these cards for sale. I guess Bitcoin or crypto mining costs have reached some threshold where these cards no longer make money.

(The A2000 is a power-limited card drawing only 70 watts, intended for workstations, but I guess that might also make it suitable for mining.)

They increase the cost to customers and cre revenue for the government, but they do not stop trade.

For small, normal tariffs there is no real difference in trade. It just goes to government coffers as a hidden tax.

The current trade war will certainly increase costs, but still the goods will flow. Nothing is stopped, just a bit more pressure on people who are sensitive to costs. Certainly the rich don't care about a few cents or a few dollars. The billionaires especially don't care, they can pay hundreds to have a special sandwich delivered to them fresh at their vacation location, what's a few bucks at Amazon when they are also getting same day delivery?

If stopping trade was the point, there are trade embargos and import bans and government seizure of goods that could be invoked.

When I search for anything, Gemini pops up despite it being useless.

When I tell my phone to play the news or play some music or tell me the weather, Assistant was disabled and now Gemini tries to do it, but badly.

Features I liked on my phone were removed against my will and against my preference, now instead of something useful it just says "I am a large language model and I can't do that useful task".

When I use work tools that use Gmail, Gemini pops up and I can't turn it off.

When I use Google Docs, because that's what work requires, Gemini pops up repeatedly telling me it wants to be useful, it's worse than Clippy ever was.

Probably 10,000 of those "uses" were just me personally telling Gemini it is a useless pile of garbage that if it caught fire it could at least provide warmth and heat as a dumpster fire, it is less valuable than that. It is a waste of bandwidth, unwanted, being aggressively forced on the victims using Google products as their enshitification converts useful tools into monetization.

Yes, and it it was such a bad merger, why didn't the Justice Department tell them not to do it in the first place? The DOJ is supposed to review mergers of large companies. I'm no fan of Zuck or Meta, but seems kins chickenshit to come after them after being silent on the IG buyout.

Quite a few scenarios can't do it, actually.

If your scenario fits in the box --- and the vast majority of cases fit in the box --- then the ACME protocol works great. Publicly accessible, ability to modify mainstream DNS TXT record, and public access to port 80, and able to get online on the public internet periodically at least once every 3 months, you're great. If you're in the most typical scenario it works just fine. Box in a datacenter, system runs in an always-on network, using any of the major DNS providers, even if your network is mostly private but one at the top level is accessible so you can do a *.example.com wildcard registration, the vast majority of people have no issue with it.

But that's not everybody, that's almost everybody. Some scenarios are excluded. What if the device is not using a mainstream DNS provider? What if the device can't provide the port on the domain's address? What if the device can't get online frequent enough? What if the obscure configuration doesn't allow for DNS challenges? What if you don't own the level you can do a wildcard registration? What if you can't satisfy the round trip timeout? Live in a rural place, where service comes through "internet on a bike" but still want security? Need to serve from a device that can't reach the public internet for many months at a time? There are plenty of obscure situations where the protocol doesn't work.

People in those scenarios still want the security, but they don't fit in the box.

Yup. They are overdue for poisoning bot requests. Block the hosting domains, errors and black holes, feed them the same errors every time about how they can get the copied version of the databases at cost. This is not a new problem, companies have detected and killed bot traffic for decades now.