The AI does not hate you, nor does it love you, but you are made out of atoms which it can use for something else .

https://www.lesswrong.com/posts/eehsGtoQTncuJs3Fd/the-paperclip-maximiser-s-perspective

Happy ending: she is now working in nanotechnology, hoping to design self-replicating assemblers that will clog the world with molecular-scale paperclips, wipe out all life on Earth and continue to sustainably manufacture paperclips for millions of years.

Not quite the same thing, but I really don’t like the ISO (International, what a lot of European use) QWERTY layout compared to the US one. It’s not unusable or anything, but…

I wish that ISO would make some new layout that starts from the layout from US ANSI and then stuffs the European-specific symbols somewhere on the keyboard.

And while I’m dreaming, I’d like that layout to physically swap left control and Caps Lock, so that I don’t have to go swapping it in software everywhere.

And to get rid of Menu and Right Windows and replace it with Compose which is, I think, by far the most-preferable way to get access to a substantial additional number of characters. AltGr or Option permits for a small number of additional characters and is harder to remember for occasional use. The Windows Alt-numpad scheme is also much harder to remember, as is the GTK Control-shift-u <Unicode hex codepoint> convention.

I also don’t use right Control, but I can believe that somewhere out there, someone gets actual use out of it and needs it somewhere comfortable, so I won’t complain about that.

Actually, what I really want, which would solve the above in an even better fashion, is for laptops to use modular, standardized, replaceable keyboards so that I can just buy whatever keyboard I want and slap it on the thing. With external keyboards, as on desktops, the selection is much better.

EDIT: I’d also add that I’ve seen numerous European users saying that they also prefer the US ANSI layout over the ISO layout, so it’s not just me being US-centric, and OP has a comment even saying so themselves in this thread. But if you just use stock US ANSI, then you don’t directly get access to the extended Latin set, which you want in Europe. Though Compose can do that, and OP is, like me, also wanting Compose on his keyboard…

@MrShankels@reddthat.com:

That and DAU (daily active users) are common metrics used by social media companies to measure the size of their userbase, so it permits for comparisons among different entities.

I commented on this myself earlier. Doing the extra shipping and bare legal minimum to count as a transformative step to create a new product in some other country will increase the cost of the good relative to where it is today, so there’s still an impact, but for virtually all products, it’s going to be far less than the 145% increase that tariffs would impose on a directly-shipped-from-China product.

This is waaaay too much effort.

If you are willing and able to do a bit of scripting, it’s not that hard to generate and send a bunch of messages on the Threadiverse.

And there are people who will go to pretty extreme lengths to harass people who they are really upset with. An ugly breakup or something and…

What makes people think it’s harassment?

It’d make a lot of sense to me.

The image quality was poor, and there are AI models that permit one to create absolutely stunningly attractive people, moreso than real photos. Hell, I’ve written scripts myself to automatically drive Stable Diffusion to produce bulk procedural images. Anyone capable of scripting up a bot to send the message in the first place is more than capable of scripting up better generation.

For catfishing, sending multiple duplicate messages to a user, which happened in this case, seems unlikely to be a goal.

I assumed that it couldn’t reasonably be a scam attempt, so was guessing at it being a deanonymization effort, but harassment would make even more sense. If you’re trying to drive lots of angry people to make the victim miserable, it doesn’t matter if the images are annoying — in fact, it only makes them more effective, since hopefully you get more irate users sending material to the victim.

For anti-spam efforts, I think that there are a variety of potential partial solutions. No complete fixes, but some:

• Rate-limiting the comment frequency on new accounts. IIRC, Reddit used this tactic. It does create some issues for (legitimate) use of throwaway accounts in anonymous posts, but there’s no legitimate reason for a new account to blast hundreds of messages an hour, I think. This might already be present, but if not, it’d be a good start. This can be defeated by generating new accounts for each new message or batch of.

• Rate-limiting new account creation from a given IP address, if not already present. An attacker could defeat this via use of a commercial VPN, and if too low, it could create issues for some commercial VPNs.

• Hashing of messages to red-flag identical messages being posted en masse. As best I could tell, the spammer here was posting many identical messages. This can be defeated by a spammer having software slightly modify each message.

• Fuzzy-hashing of messages to red-flag almost identical messages being posted en masse. This can be defeated via text generation methods that are carefully tailored to the fuzzy hashing mechanism to modify messages such that each fuzzy-hashes to a different value.

• A mechanism to permit an account to share blacklists of IP or message hashes and trigger removal of messages on other instances, preferably associated with a specific identifier or account. This permits any other instances to leverage antispam work by one instance; if I want to trust a given antispam admin or bot on lemmy.world, I can. Let an instance admin review and override such removals, maybe. It creates abuse potential for malicious use or inadvertent false positives spanning instances, but I think that it’s necessary to avoid having each instance fight its own lonely antispam battles. Otherwise, new and personal instances risk being buried by a deluge of direct message spam. The same mechanism, if exposed to users and not just instance admins, would also permit for subscribable content filters for people who don’t want to see content of a given sort (e.g. profanity or pornographic content of a particular sort or whatever, not just spam), which is another issue.

Fortunately, as far as I see as a user, we’re not yet at the point that there is much spam on here yet, so this isn’t yet a serious problem. Maybe it’ll never happen, if the userbase never grows much. But if the userbase gets considerably bigger, increasingly-problematic spam will inevitably follow.

not allow images or embedding at all in PMs

I’d add — as someone who was concerned about and posted on the possibility that the aim of the spammer was exposing the IP address associated with the receivers’s username — that even if this wasn’t the aim from this event, it could be in some future event.

I don’t think that disallowing inline images in direct messages will eliminate spam problems, even efforts of this sort, as it’d still be possible for a spammer to spam messages with indirect links to images hosted elsewhere. But it would help avoid leaking IP addresses of the receiving user.

Or at least disallowing inline images in direct images by default. I can imagine maybe someone enabling them on some kind of a private, decoupled-from-the-wider-Fediverse instance on an intranet or whatnot, but I really don’t think that this is something that nearly any instance should actually permit.

pushes rewind button

https://www.theguardian.com/world/2016/nov/15/germany-trump-ttip-trade-deal

Tue 15 Nov 2016

Transatlantic trade deal ‘not realistic’ under Trump, German official says

pushes fast forward button

https://www.politico.eu/article/eu-offers-trump-removal-of-all-tariffs/

April 7, 2025

EU offers Trump removal of all industrial tariffs

“Europe is always ready for a good deal,” says Commission President Ursula von der Leyen.

The Onion staff was fighting a valiant rear-guard action against reality, but it was increasingly clear that the odds were stacked against them.

If you use keys or strong passwords, it really shouldn’t be practical for someone to brute-force.

You can make it more-obnoxious via all sorts of security-through-obscurity routes like portknocking or fail2ban or whatever, or disable direct root login via PermitRootLogin, but those aren’t very effective compared to just using strong credentials.

That’s frame scaling in real-time, rather than offline texture scaling.

The tech demo is part of Microsoft’s Copilot for Gaming push, and features an AI-generated replica of Quake II that is playable in a browser. The Quake II level is very basic and includes blurry enemies and interactions, and Microsoft is limiting the amount of time you can even play this tech demo.

Microsoft is still positioning Muse as an AI model that can help game developers prototype games. When Muse was unveiled in February, Microsoft also mentioned it was exploring how this AI model could help improve classic games, just like Quake II, and bring them to modern hardware.

Okay, here’s a much-less ambitious use of existing AI technology that I think would be vastly-more-useful than whatever they’re off doing: how about just going out and using existing AI upscaling techniques and limited human interaction to statically-upscale the textures by maybe 2x to 4x, take advantage of more VRAM on newer hardware?

The elderly have much higher per-capita healthcare consumption than do people during other points in their life.

One element of the ACA was capping insurance premiums for seniors at an 3:1 ratio, where seniors couldn’t be charged more than 3 times the premiums of people at other ages in life.

If you mean just rely on state-level taxation, it’d create a incentive to work in (low tax) states that didn’t provide state-subsidized health care, then retire in a state that does.

You want any kind of intergenerational wealth transfer to happen at the federal level, else you will tend to get those misincentives.

I mean, that splitter device cannot be USB compliant, if it’s giving more than 5V to a device that never negotiated more than 5V.

I’d also guess that if it’s just silently feeding the second device from what the phone has negotiated, it’s probably not compliant in that it’s probably drawing more from its power source than the phone has negotiated – USB devices are responsible for indicating what they’ll draw.

You could make that multiport device USB-compliant, but it’d require having the splitter be a DC-DC power source and having it negotiate some PD draw sufficient to power both devices.

I use my computer for so many things and I have about 200 applications on my computer. I don’t know why, but it bothers me that everything happens on this one machine as well as seeing so many app icons (even grouped into folders).

If what you want is organization from a workflow standpoint, I think that you’d have an easier time just using some form of launching system that doesn’t show a single monolithic menu of all your installed executables. Either have a launcher that permits breaking up stuff by task and lets you customize those groups, or just use a non-menu-based launching system.

I mean, /usr/bin on my system has 2694 entries. I don’t see them, though, since I’m launching software via bash or tofi, so…shrugs

VMs can have uses, but I’d mostly either use them for software compatibility, or to isolate things for security reasons. They wouldn’t be high on my list of tools to organize workflow.

Ooh, I didn’t know that someone had developed a mechanism to move issues and PRs.

I remember commenting on the fact that while it’s easy to move the source repo itself from location to location, as git makes that easy and self-contained, issues and PRs didn’t enjoy that.

Well, that gives you infinite energy, since you can produce energy-containing stuff.

Hmm.

On one hand, a lot of competition for resources go away.

On the other hand, that’s also pretty disruptive.

I think that that world is going to have a lot of sudden challenges. You don’t have scarcity of any material or existing item that you can break down to less than a 1m cube unless you need it in great bulk, but you also have no ability to control production of things like firearms, explosives, drugs, physical proofs of identity, missiles, weaponized drones, etc.

I can imagine countries or organizations trying to seize the supply of replicators.

If you can disassemble them, this is probably a good way to eliminate bounds on throughput, but honestly, even a little coordination permits for pretty enormous throughput from the get-go. You’ve got a lot of people out there.

It was the only search result relevant to my query.

Paste it into archive.org’s Wayback Machine. Good odds that they’ve stored a copy. I’d do it for you and just link to the page, but you don’t list the URL…

https://web.archive.org/