Off-and-on trying out an account over at @tal@oleo.cafe due to scraping bots bogging down lemmy.today to the point of near-unusability.
how i dream of buy a few 12 tb drives.
Rotational hard drive prices are up, but not that far, not like RAM, especially if you’re not getting the largest drives possible.
That’s maybe a doubling on 12TB drives from the trough. I mean, sure, it’s significant, but if you were already talking about getting several, you can certainly get one.
You didn’t really have a 3D printer setup until its various lights were being driven by a DMX512 rig.
A criminal case has been opened against the man for intentional destruction or damage to property.
Unless Russia has some sort of legal technicality that upgrades a crime to being intentional (like the murder felony rule in the US, where normally murder requires intent as an element, but there are cases where a killing can become upgraded to murder without that intent), I’d think that it’d be tough to prove. I mean, it sounds like the guy said that he severely dicked up and then just ran off. That sounds more plausible to me than that he planned to set his van on fire and then drove it, flaming, to somewhere that he wanted to do damage.
You could maybe get him with some crime that involved reckless disregard for whatever he destroyed, but unless Russian law works differently from the US, that’s a different bar from deliberate intent.
And a lot of that can absolutely be used for intelligence work.
I mean, probably, but by the same token, so can binoculars, a smartphone with a camera, a camera with a telephoto lens, a radio, and a ton of other stuff that lots of people will have.
And if you’re trying to spy, (a) you don’t need a large number of drones — that’s important if you’re using them as disposable weapons — and (b) my guess is that most of those tiny, short-range ones would be of pretty limited value in most cases.
EDIT: I’d also add that my guess is that one thing that Ukraine is probably not terribly short of is abovehead intelligence. As Mike Kofman pointed out repeatedly, the military value of the intelligence that the US provided to Ukraine is probably greater than the weapons. They’ve got satellites and a bunch of other sensor platforms. And the US isn’t the only one providing intelligence. And Ukraine can already fly drones — albeit perhaps more-expensive, longer-range ones — over Crimea out of unoccupied territory. That’s not to say that there isn’t still utility to drones — I mean, maybe flying one into a warehouse with an open door is safer than going in person. But by-and-large, my bet is that the most useful intelligence Ukraine gets from people behind the lines isn’t aerial photography.
I am not particularly sympathetic, but there is a group that takes this quite seriously:
https://en.wikipedia.org/wiki/Voluntary_Human_Extinction_Movement
The Voluntary Human Extinction Movement (VHEMT[A]) is an environmental movement that calls for all people to abstain from reproduction in order to cause the gradual voluntary extinction of humankind. VHEMT supports human extinction primarily because it would prevent environmental degradation.
The governors of Osaka, Fukuoka and Aichi have expressed interest. “Aichi Prefecture and Nagoya possess all the functions necessary to serve as a second capital. I’m the first in the nation to make a bid,” Aichi Governor Hideaki Omura said. Fukuoka Governor Seitaro Hattori added Friday: “We will aim for the designation as ‘all Fukuoka’ by working with municipalities and the local business community.”
For purely historical value, I gotta say Kyoto.
https://en.wikipedia.org/wiki/Kyoto
Kyoto is one of the oldest municipalities in Japan, having been chosen in 794 as the new seat of Japan’s imperial court by Emperor Kanmu. …The emperors of Japan ruled from Kyoto in the following eleven centuries until 1869…The capital was relocated from Kyoto to Tokyo after the Meiji Restoration.
searches
Aw yeah.
https://en.wikipedia.org/wiki/Capital_of_Japan
As of 2007, there is a movement to transfer the government functions of the capital from Tokyo while retaining Tokyo as the de facto capital, with the Gifu-Aichi region, the Mie-Kio region and other regions submitting bids for a de jure capital. Officially, the relocation is referred to as “capital functions relocation” instead of “capital relocation”, or as “relocation of the Diet and other organizations”.[19][20]
In 2023, the Government of Japan moved the Agency for Cultural Affairs to Kyoto. This was the first time that a central government office has been relocated outside Tokyo since Tokyo was designated as the capital.[21][22]
Because AI has become so lucrative, the three major RAM manufacturers have shifted from mostly producing conventional DRAM to focusing more of their capacity on HBM, which is what AI data centers need.
The author probably wants “DIMMs” rather than “DRAM”. HBM memory is DRAM too (well, I guess theoretically you could make non-DRAM memory with an HBM interface, but…)
That’s pre-atomic-weapons-testing steel right there.
https://en.wiktionary.org/wiki/close_enough_for_government_work
Adjective
close enough for government work (not comparable)
- (US, idiomatic, humorous, derogatory) Adequate; acceptable as government work, imperfect but not worth the time nor effort of improving.
Synonyms: good enough for government work
Ah, fair enough — didn’t know what the prize was.
I’m not sure that this is a good idea. Futures markets on things like assassinations have always run the risk of incentivizing people to do things that they shouldn’t, and this is similar.
Ukraine probably has a military plan in place to hit specific targets. You want people to target the targets judged to be optimal. If someone in the Ukrainian military can make money by making such a bet and then targeting the target that they have bet on, there is now an incentive separating their interests from that of the state.
Additionally, depending upon whether the betting information is made public (and whether that information, even if not made public, is properly secured), if someone bets using insider information, it may leak target information ahead of time, which may be detrimental to the Ukrainian state. Russia will have nation-state intelligence agency levels of capability in going after things in the war like strategic air targets.
To put even more context on that…when I first used a diffusion model, I was using the Automatic1111 frontend (well, okay, actually, I think it was a variant, but the first one I put much time into). I was using Stable Diffusion 1.5 or a derived model, can’t recall.
Automatic1111 looks more-or-less like what I think most of the cloud service systems look like. At its core, you plug in a text prompt, you click a button, you get an image.
In 2026, I pretty much only use ComfyUI, which has a directed-acyclic-graph-based system akin to what some image processing software or visual-programming systems use. IIRC, GIMP uses a DAG internally these days, though doesn’t directly expose it to users. searches Yeah, GEGL. And AFAICT, that’s generally where the momentum is — it’s not just me. That’s a lot more powerful for making changes early in a workflow and re-running it, though more difficult to learn. And from a user standpoint, it looks wildly different.
I’m not saying that there was no transferable knowledge there. You’re still working with a prompt somewhere in there. And the parameters that go to models are similar. But…you’re talking about what is effectively a new syntax for prompts. A wildly different set of limitations to work around with those prompts. Shifting from a UI that is basically an “input a string and click go” to a visual programming language.
And…I haven’t even listed some of the other of the changes. Uh, let’s see. Stable Diffusion 1.5 supports negative prompts; one can list prompt terms that should not be presented in an image; Flux does not. Flux has some sort of natural-language-processing black magic that frequently lets one use English-language descriptions to do things like “Divide an image into five separated panels. The first panel should contain…”. With SD 1.5, one would have to use some sort of plugin to subdivide an image and manually apply different prompts to different regions of an image.
Like, virtually the only thing in common between the two are sampler settings (and I use different sampler settings with each model) and the fact that both have a prompt, and there’s virtually no overlap in prompt syntax. And that was over a timespan of three years. Well, less than three years.
I can believe that you can teach someone to directly use a model. I can believe that you can teach someone to write software that uses a model. I can believe that they can produce useful output using that model. But where I have some very real problems is buying into the idea that in 2026, we’ve reached the point where the user skillset is going to be long-lasting enough that this is where one is going to get the best return. It’s like…I don’t know, training people to fly the Wright Flyer en masse:
https://en.wikipedia.org/wiki/Wright_Flyer
The Flyer design depended on wing-warping controlled by a hip cradle under the pilot, and a foreplane or “canard” for pitch control, features which would not scale and produced a hard-to-control aircraft.
You’d be right if you predicted in 1905 that heavier-than-air aircraft were going to be a very important thing in the future, but it was extremely early days then, and the control interface and the challenges that the pilot faces are probably going to be very different in short order.
Secondary school pupils in England will be able to study subjects like manufacturing and AI in addition to core academic learning under reforms outlined by Prime Minister Andy Burnham.
Huh.
considers
I’m still a little dubious that we’re at the point that it makes a lot of sense to dump a lot of people into AI applications. Like, yes, please put as many people as possible on the core research, the kind of thing that people at, say, Nvidia or Google are doing. Figure out how to train a model more-efficiently — that’d be great. Fundamental work towards AGI? Please. New data structures for machine learning other than multilayer neural nets? That’d be great, all important research areas. But “writing programs that make use of existing AI systems and utilizing existing AI systems day-to-day”…the problem I have is that this is early, and the thing is in a state of rapid change. We’re still knocking out low-hanging fruit, solving problems. You build a workflow that uses some model that exists today, and even if the model continues to be available, the limitations and workarounds for a model twelve months later may be quite different. I’d want that to stabilize before shifting a lot of people into using models, so they aren’t having to re-learn a lot of stuff.
Like, Stable Diffusion 1.5 was released in late 2022. The problems that you’d see with that model and the prompting techniques to deal with it and work around limitations are very much not the same problems and limitations that exist for Flux 2 in 2026. Stable Diffusion supports token weighting; Flux does not. Stable Diffusion 1.5 performed poorly with English sentences, wanted a list of tokens; Flux does not. Stable Diffusion 1.5 had problems with fingers; Flux 2 does not. If you spend time getting someone to learn how to operate a diffusion model of 2026, is that skillset going to be similarly relatively-obsolete in another three years? Because if so, the student has to be providing an enormously high return-on-educational-investment over each of those three years to make it sensible to invest lots of time in making people expert in using that model.
I’m not saying that nobody should learn to use them, but I don’t know if this is a “mobilize government and shift things ASAP across society” situation. And that’s not because I don’t think that AI is important. It’s just because I think that from a usage standpoint, things are changing very quickly right now.
The best skills to teach someone, all else held equal, are the skills that they can use for their entire working life. I strongly suspect that the great bulk of LLM or diffusion model user skillset is going to be quite different in a relatively-short period of time, and I can very readily believe that we won’t even be using the relatively-primitive “talk to a raw LLM” within a decade or two, have some considerably-more-sophisticated system moving towards AGI.
Yes, pretty much any skill has value. But lets say that you have the time and funds to teach a student, oh…statistics or how to use an LLM. Which is going to provide the best return on investment for the educational time? Because that’s the real metric that matters.
The deep ties between Iran and Iraq, the two biggest Shi’ite-majority countries
I mean…
https://en.wikipedia.org/wiki/Irreligion_in_Iran
Non-religious Iranians are officially unrecognized by the government, and one must declare oneself as a member of one of the four recognized faiths in order to avail oneself of many of the rights of citizenship.[9][10]
Citizens of the Islamic Republic of Iran are officially divided into four categories: Muslims, Zoroastrians, Jews and Christians.
I feel like that kind of limits how meaningful it is to say that a given percentage of the population belongs to some religion in a state.
People who don’t like codeberg can just go somewhere else (even back to github) and/or found/support an alternative that better suits their needs.
searches
https://gitgud.io/ provides free GitLab hosting for open source projects.
https://gitlab.com/ has a free tier for personal/open source stuff, though looks like it has a 5 user limit on said free tier.
Gitea can be self-hosted. It doesn’t look like they provide free hosting themselves, though, just paid. Apparently it’s considerably less resource-intensive to host than GitLab.
GNU Savannah is available if one is writing projects under a specific set of licenses (the GPL family).
The Power of Siberia pipeline reached its designed capacity, allowing Gazprom to supply more natural gas to China … – 38.8 billion cubic meters. The compensation, however, is far from equivalent. Prices for gas sold to China are linked to a basket of petroleum products and are subject to a substantial discount, while the lack of alternative export routes leaves Moscow with little bargaining power.
The lack of alternative export routes leaves Moscow with little bargaining power in selling oil to China.
Yeah, from the various articles that have been coming up in the news on the long-term oil and gas deals that China has been cutting with Russia, I’m actually surprised that China is using said leverage as hard as it is. Whatever negotiator they have on it is extracting as much wealth from Russia as he can.
Negotiations between Russia and China over the proposed “Power of Siberia-2” natural gas pipeline have hit a complete deadlock following Beijing’s demands for steep price discounts, according to a report by The Wall Street Journal.
Sources familiar with the talks said the Russian negotiating team “hit a wall” after Beijing demanded that Moscow price the gas at its heavily subsidized domestic rate of about $50 per thousand cubic meters.
The pricing demand comes despite China already receiving substantial concessions. This year, Beijing pays $258.8 per thousand cubic meters—a 39% discount compared to the $420.2 average paid by Gazprom’s other international clients.
Internal Russian economic forecasts show China’s price is slated to drop further to $223.9 next year. Still, Beijing is pushing for a rate eight times cheaper than what Russia’s remaining global customers pay.
The natural resources may be inside Russian territory, but Russia’s just going to be doing the work of extracting them and then handing most of the benefits off to China.
If Russia is forced to take said deals, China is going to be doing very well off it until Russia manages to climb out of the international relations hole that it’s in. But the concern that I’d have from China’s standpoint is that at those prices, Russia is going to do whatever it possibly can to figure out a way to cut the deal off as soon as it’s able.
In effect, Russia’s wealthiest state linked companies are being allowed to finance, equip, and direct their own armed formations to defend private industrial assets. This fundamentally shifts the balance of power between the Kremlin and Russia’s oligarchs by giving major corporations armed personnel under their effective control.
I mean, it’s been a good three years since the last time a wealthy, politically-connected Russian tried doing an armored push on Moscow.
If it comes to a contest between production capacity for weaponized drones between Taiwan and China, I think that Taiwan is going to be in trouble. China overwhelmingly dominates global drone production capacity.
If it came to a contest of that sort between the US and China, the US isn’t, in 2026, going to be able to compete in mass either.
searches
The drone market is unlike any other advanced technology sector. Companies based in China and subsidized by the Chinese government control 90% of the consumer drone market, 70% or more of the enterprise market, and 92% of the state and local first responder market. Chinese company DJI holds the vast majority of that market share.
That did not happen by accident. In 2015, China launched “Made in China 2025,” a ten-year, whole-of-society effort to invest in key industries, including drones, to ensure China’s world leadership and market dominance.
The US does have various projects to aim to counter that, but does not yet, as of 2026, have that sort of capability. If China started maximum production of weaponized drones right now, using existing civilian and military infrastructure, I don’t think that we’d presently have a defensive-type counter. We’d probably have to go on offense, try to destroy their industrial capacity.
One project I recall some discussion about is Replicator.
https://en.wikipedia.org/wiki/Replicator_(United_States_military)
Replicator is the name of a United States Department of Defense program intended to pioneer ways to cheaply produce large amounts of weapons or systems for the U.S. military.
It was announced on Aug. 23, 2023, by Deputy Defense Secretary Kathleen Hicks as a means of quickly producing weapons to deter and counter China. In her announcement, Hicks said the first type of weapons to be produced under Replicator were to be autonomous systems; she did not specify whether these would be, for example, aerial drones or unmanned vessels.[1]
The Replicator program announced its first batch of contracts on May 6, 2024; they included purchases of uncrewed watercraft, aerial drones, and anti-drone defenses “of various sizes and payloads from several traditional and non-traditional vendors.” Some other contracts “remain classified, including others in the maritime domain and some in the counter-UAS portfolio”, a Defense Department release said.[2]
On September 30, 2024, Defense Secretary Lloyd Austin III announced a new phase of the program, dubbed Replicator 2, that would focus on anti-drone defensive systems.[3]
every IP address
every IPv4 address
Hmm.
A lot of these are going to be just request-driven. That is, they aren’t really doing anything while idling. Thus, they don’t really need to be actually running.
If the idle memory usage is an actual concern — and honestly, the OS is probably just gonna page things out if the software is truly idle and it needs the memory for something else and it has paging space…I kind of assume that someone has built some sort of system that works like inetd, which avoided leaving non-containerized servers running. You register a port, and then inetd listens on it. When someone connects, it starts up the daemon in question (say, Apache or whatever). Just need to start a container on demand here. Maybe close it down after a long-enough period of inactivity.
EDIT:
https://old.reddit.com/r/docker/comments/72sdyf/startrun_a_container_when_incoming_traffic/