There are a number of games — particularly those since SSD storage became pretty much the norm — that do not load a whole level, but load while playing. Why not just load the game (or the level)? Because it lets devices with less RAM store more stuff than they otherwise would. You have more relaxed caps on things lile texture size.

I’d assume that the engine is still going to try to predict what content is going to be needed prior to it being needed, that it’s got a strategy more sophisticated than just “load resource first time it’s requested”. But SSD storage is going to be more forgiving of a miss than rotational storage.

Bethesda uses the term “cell”. I know that their engine uses a square grid as the loading unit, loads a sort of circle of cells around the cell where the player is.

searches

It sounds like the Unreal 5 terminology is “sublevel”, for the units loaded.

https://old.reddit.com/r/unrealengine/comments/1tchk8k/is_level_streaming_even_worth_it_if_your_game/

We might have a habit of being curious and seeing things and patterns that aren’t really there. And it’s frustrating to watch it misfire.

But…maybe it’s worth it to come up with silly ideas on a fair number of occasions if that same tendency sometimes leads us to make valid discoveries that we wouldn’t otherwise have made. Sometimes even people who have wrong ideas run over important things.

There was this guy once who disregarded well-established (and correct) geophysical science with some cockamamie theory that he thought would make him rich. He tried to sell a bunch of people on funding his ideas, who mostly turned him down. Eventually, he managed to get funds and set out to prove his ideas. That experiment would have been suicidal for him — because his ideas were totally wrong, making the experiment lethal — if not for the great good fortune of, during the course of this experiment, making a (very important) real discovery that saved his life. He then tried — not very successfully — to claim that his new discovery actually showed that his ideas were correct.

He was a dreamer, and a bit of a nut, and he was wrong. His name was Christopher Columbus, and he had convinced himself that the circumference of the world was much smaller than was actually the case, and that if there was nothing but ocean to the west of Europe, one could safely reach Asia in a single trip of the sort that ships of the time could make. Instead, he discovered the New World.

There were a lot of other people who went looking down other weird rabbit holes with other weird ideas. They usually didn’t come up with anything worthwhile.

But…that discovery was worth a whole lot of poking around blind alleys.

So, hey, maybe a human tendency towards a bit of being a crank and running down crazy ideas is worthwhile, if it’s the price we pay to occasionally come up with something really amazing down a weird rabbit hole.

That comment is quoting text from an article.

So, from past reading, I believe that a lot of jobs in brick-and-mortar banks — which have historically had a major brick-and-mortar presence — have been on a long decline. That’s a phenomenon that well predates popular LLM use, brought about by Internet, ATMs, and similar, which tended to displace brick-and-mortar stuff.

I physically went by my local bank the other month (for the first time in a long time) and while the building was still there, it was like a tomb. All of the service desks that had once held tellers were empty. There was one person in an office handling the few people coming by in person.

searches

https://www.alliantcreditunion.org/money-mentor/the-rise-of-online-banking-in-the-us

Every year, more and more Americans are going online to do their banking.

According to Pew Research Center, 46 percent of adults in the United States banked online in 2010. That number grew to 51 percent in 2013, and in 2015, a full 81 percent of adults who manage their household’s finances banked online.

https://www.branchspot.com/bank-branches-closed/

U.S. banks have been closing branches for years — thinning out “banking deserts” where the nearest branch is now miles away. Branchspot tracks 939 closed or relocated branches nationwide, sourced from the FDIC.

I agree that “open source” is often incorrectly used to refer to open-weight models, things that are more analogous to closed-source software where you get the binaries.

That being said, the article does correctly use “open weights”. The submitted title here lacks it, but the current title of the article does appear to have it, so either the article author updated the title or the submitter truncated it.

Anxiety was on display on Tuesday, when the Nasdaq 100 fell 1.3% amid growing uncertainty about the eventual payoffs from massive artificial-intelligence investments. The S&P 500 declined 0.2%.

“IPOs are inherently more volatile,” Amy Wu Silverman, head of derivatives strategy at RBC Capital Markets, wrote in a note to clients. Given the size and scope of SpaceX, “it is likely that the vol spread between Nasdaq and S&P remains wide until we close in on the inclusion of SpaceX into the S&P as well,” she added.

The bank’s equity sales team recommends investors buy puts on the Invesco QQQ Trust Series 1 ETF and the VanEck Semiconductor ETF to express their bearish views on the tech sector.

It’s certainly good for index fund investors (e.g. S&P 500 index funds lile SPY or VOO) whose indices didn’t recently get their index’s rules rewritten for rapid inclusion of some of these IPOs relative to those who did (NASDAQ 100 index funds like QQQ or QQQM, or Russell).

Replacing the stolen license plate readers will cost Winona roughly $24,000 total.

That’s one possible response to them being removed, I suppose.

Chief among them was Seewa AI, which offered its users the “perfect AI girlfriend”, drew them in with intimate conversations and promised that all chats were strictly confidential. (They weren’t.)

End of the day, if it’s in the cloud instead of local — and this applies for pretty much any service, not just LLM models — you’re relying on trust in the service provider (as well as anyone who might buy them, and trust that they are able to secure their network against people who might break into it).

Only real exception to that is situations where you use client software from another source and the service provider only gets to see any data in encrypted form — think of someone providing storage for a restic backup, say.

In 2026, we don’t have enough memory to do local LLM computation everywhere, so there’s a real limitation versus most other forms of SaaS — we can’t choose to run LLMs locally for more than a fraction of society. Cloud-based LLMs require less memory to provision one machine in the cloud and share that hardware, since any one user is only using it a fraction of the time. But as the memory shortage ends, I expect you’ll have more people running LLMs locally. Still be more-expensive from a hardware standpoint than cloud computation, though.

Right — that’s why I’m talking about a PMU and serial device server linked to a serial console and BIOS as an alternative. That’s another way to accomplish that.

My guess was that IPMI might provide a more-standardized interface, and notabot said that this was the case.

You can mount an ISO image through the BMC and boot the server from there.

I mean, you can typically do a TFTP boot on the systems that I’ve seen that have serial-capable BIOS setups.

considers

Heck, I think that all of my modern personal non-serial-port-BIOS systems can do that, even.

Baseboard management controllers are miniature computers that are embedded into the motherboards of virtually every enterprise server. The microcontrollers, typically abbreviated as BMCs, run with their own operating system firmware, network stack, and IP address. Administrators rely on them to monitor the physical status of large fleets of servers and to perform a variety of tasks, including rebooting machines, installing updates, and even reinstalling operating systems. BMCs provide what’s known as “lights out” and “out-of-band” management because they work even when servers they’re attached to are turned off or are unresponsive.

Hmm. I don’t really understand what the major benefit is to having an integrated management system over just providing a BIOS that can expose its interface on a serial port and having a serial device server attached to the thing, which then means that you aren’t tied to a particular management system. I’d also kinda think that it’d be cheaper to use a serial device server if you have a ton of servers in a rack that need to be managed, since one serial port server can handle all of them. Use a smart PMU for power control.

Maybe interfaces are more standardized or something, for automated tasks?

Eh, this one’s fixable, because you only have a few sources of legit imagery, and it’s all contemporary and digital. Just have Google Earth render the image server-side and add a cryptographic signature to any legit satellite images that someone saves.

You can have AI-generated satellite imagery and image-is-proof satellite images coexisting. Can’t do that for most images, but for this particular niche, you can.

Honestly, GNU/Linux isn’t better today from a user standpoint. I mean, Linux has the technical subsystems to sandbox software, yes — like, the move to Wayland makes isolating graphical software possible, where X11 imposed major limitations on one’s ability to do that. Various containerizing systems exist. But the typical software package distributed via Steam, say, in 2026, can see the filesystem as a whole.

Like, I can hack something up with something like firejail using that functionality that may-or-may-not work for a given game on a given system, but that’s not a real, universal solution for users. My mother is not going to do that.

Mobile OSes at least isolate most apps from most of the filesystem and some sensors, and provide some level of reasonably-accessible transparency into what apps can do. They still have their own issues that annoy me, like software normally being able to phone home with what information it does get. But that’s still way more isolation than a user who is playing a video game on GNU/Linux is getting.

Global, always-generally-available connectivity brought us a lot of nice things, but it’s also come with a lot of not-so-great things, and some of the not-so-great things is a loss of privacy and a loss of control over the things that we own. :-(

I mean, that’s really true of products in general.

I think that there are some real consumer law issues to answer where someone buys a product and then a manufacturer can control the product subsequent to the purchase. Like, say I buy a BMW with the expectation that I am not going to be seeing ads, based on the way it acts at the time that I bought it. At that point, I’m locked into owning that BMW. Then the manufacturer degrades my experience, maybe to the point where I no longer consider the vehicle to be the one I wanted.

Like, what is my remedy here? If the vehicle has OTA updates, I can never know what it’s going to be doing down the line, so I can’t make an informed decision. Having informed consumers is one of the basic requirements for markets to be efficient.

Maybe I could say “Well, that’s just a case of let the buyer beware. We’ll rely on reputation…some auto manufacturers will piss off consumers and those will be avoided”. But…if someone doesn’t buy cars that frequently, that’s a long iteration cycle.

I’m actually more concerned about video games. Like, if I’m a publisher, once I sell a video game, I don’t normally have more revenue coming in from it. But if I do have the ability to update that video game down the line. And I can update it to go harvesting user data, which turns it into a revenue stream. A lot of video game publishers go defunct on a regular basis or sell IP or whatnot, so there, even the “rely on reputation” route is more questionable.

It’s absolutely worth requesting your data to be deleted regardless of where you live by claiming to live in California…

I do think it’s possible that some of them won’t delete your data upon request, failing to comply with the law

That’s not failure to comply with the law if you don’t actually live in California.

People have a legal location of residence in the US.

If you want to establish legal residency in a state, you’re typically looking at doing something like spending the majority of the year there.

EDIT: I’d also be pretty confident, though the article doesn’t talk about it, that California won’t normally have jurisdiction for California state law unless a data broker is doing business in California, and that that will also be a constraining factor. Like, if a data broker company is only doing business in, say, Russia, the US will not enforce rulings against them regarding California state law in general.

Ehh. It’d kill browser “search text” functionality, as well as Reader Mode on Firefox.

They can’t make competitive dedicated memory in a near-term timeframe — the kind of timeframe that would matter, if they are to get something in during the shortage — and Cook has explicitly said that Apple will not try.

However, they do make their own CPUs, and they could maybe make a large-onboard-cache version, kind of akin to what AMD has been doing. That’d make the CPU more amenable to use of slower memory for most applications, which might open up more memory options.

“Technology, for a long time, has been promising people that they’re going to work less and they’re going to have all this leisure," Altman said, adding that things have gone in this direction.

He explained that people now have more time for leisure and a higher quality of life than in previous points in history.

"But somehow we never get the promise of the 4-hour workweek on a mass scale in society. And I don’t expect AI to change that,” Altman said.

OpenAI CEO added that people’s expectations always go up, and they want more.

“We think of new things to do, to create for each other, to want for ourselves. People are very focused on how they’re doing relative to other people,” he said.

Altman then spoke about the competition that drives the economy and people’s desire to be useful to others and to create.

Altman’s predictions are in stark contrast to the vision of Elon Musk, CEO of SpaceX, the company behind the chatbot Grok.

In a recent interview with the Economist, Musk envisioned a future in which jobs are optional, and money doesn’t matter, but he didn’t provide much detail on the transition to this level of abundance.

Yeah, I think that Altman is probably closer to being right than Musk is, at least for some time to come. If you’re talking about the kind of systems that OpenAI and xAI are working on, then certainly.

A lot of people — on the Threadiverse I see plenty posting — are really worried about permanent expanding structural unemployment. Something like “there will be no jobs” or something like that. And, okay, it’s not immediately obvious why that wouldn’t happen. And there have been some very prestigious people who have predicted otherwise.

https://www.npr.org/2015/08/13/432122637/keynes-predicted-we-would-be-working-15-hour-weeks-why-was-he-so-wrong

The economist John Maynard Keynes once wrote an essay titled “Economic Possibilities For Our Grandchildren.” It was 1930. And in the essay, he made a startling prediction. Keynes figured that by the time his children had grown up, basically now, people might be working just 15 hours a week. David Kestenbaum with our PLANET MONEY team tried to sort out where Keynes went wrong.

To date, what has consistently always happened over time is that as our productive capacity increases, someone finds some way to create a higher standard of living using that productive capacity. That higher standard of living has always been more-appealing than the leisure time. If we wanted the standard of living that was present in 1900, say, we could work a lot less and just accept that standard of living. What happened instead of increasing leisure was increasing standard of living. Larger houses, air conditioning, cars, video games, elaborate movies, out-of-season fruits and vegetables, all that.

It’s possible that that phenomenon will end, but I’d bet against it in the near term. We’re really good at finding new things that people would like.

Now, I do expect that there will come a point where you can produce an AGI that is truly superhuman in the sense that it’s more-capable in every significant way than a human. I do not think that that is going to happen in a year or ten years, but it might not be that far down the road. And then, unless you see demand specifically for something made by humans — think of the way that there is demand for, say, expensive handmade fountain pens, something like Veblen goods — we’re going to have to rejigger things for a post-scarcity society.

https://en.wikipedia.org/wiki/Post-scarcity

Post-scarcity is a theoretical economic situation in which most goods can be produced in great abundance with minimal human labor, so that they become available to all very cheaply or even freely.[1][2]

Post-scarcity does not mean that scarcity has been eliminated for all goods and services. Instead it means that all people can easily have their basic survival needs met along with some significant proportion of their desires for goods and services.[3] Writers on the topic often emphasize that some commodities will remain scarce in a post-scarcity society.[4][5][6][7]

But I would be pretty confident that ChatGPT isn’t getting humanity to a post-scarcity society in the near future.

And I think that there are a whole host of other things that would change in human society in such an environment that I’m not sure that structural unemployment would be the first thing I’d think about. Like…what does the military balance of power look like in a world where an individual can hypothetically manufacture robots that manufacture factories that manufacture military drones, say? I mean, it’s a awfully different world in a lot of ways.

I hope it means that AMD is going to be more competitive on the parallel compute front. That’d put pressure on Nvidia to bring their prices down.

Righr now, Nvidia is able to charge an awful lot, because there’s little by way of alternatives.