And the default DE is a JS app that runs in a webview. You know, the same tech stack we make fun of the Win11 start menu for, but for the whole DE.

Gnome is Javascript that runs in a webview. It’s the same technology stack that we make fun of with the Win11 start menu.

It’s shit technology. No wonder it requires so much RAM.

It’s true. Using floppies is too easy for decent nostalgia usage. To prove the point: how many people do you know who use floppies for nostalgia reasons?

In terms of nostalgia-buying, we millenials are now the older generation. I doubt it’s all the 15-20yo who are buying CDs.

There’s a huge difference between “Creates intelligible single-use text that’s good enough that I can understand what the text is roughly about” and “Creates text at a quality high enough to work as a quotable source”.

For the first use case, infrequent hallucinations are no problem. I read it, if I understand a bit about the topic I might catch it, if not it probably doesn’t matter too much either. Especially if it’s about non-critical topics.

For the second use case, infrequent hallucinations are a massive problem. Most people who use Wikipedia use it like a primary source. Even though sources are linked, they don’t go hunting for sources but instead rely that the information in the article is accurate. Every article is read not only once by one person, but thousands or hundreds of thousands of times. That means every single line is read and believed. You can bet that if there’s a hallucination in there, someone will read it and believe it. That’s requires a completely different level of accuracy, and doing that kind of crap translation work on such a large scale as OKA is a massive disservice.

“Just tell it to not make mistakes.”

Yeah, right.

The only mistake here is that the author switched the term “tools” for “extruders”. They did list four tools (FDM extruder, Pellet Extruder, Ink extruder, Heater).

This sounds to me much more like a human error than an LLM one, because the source material calls them “tools”.

In this work, three out of the four original filament extruders were swapped for a pellet extruder, an ink extruder, and a heater. The tools that make up the final configuration of the machine are:

Filament extruder (Figure 1a): one of the original E3D Hemera direct drive filament extruders of the E3D Motion System and ToolChanger was kept in place. It features an E3D 24 V 30 W heater cartridge, an E3D thermistor cartridge, and a 0.4 mm nozzle.

Pellet extruder (Figure 1b): a Mahor v4 70 W Pellet Extruder (Mahor.xyz, Spain) was incorporated to the system to enable 3D printing from pellets. A custom case was designed and 3D printed to adapt the pellet extruder to the E3D ToolChanger. The case wraps around the extruder and provides anchor points to the E3D toolhead plate and docking port, necessary to allow the pick-up and drop-off of the tool by the robotic arm.

Ink extruder (Figure 1c): a syringe pump was custom-built from scratch, combining an E3D Hemera XS stepper motor, a lead screw, a linear rail, and custom-designed, 3D-printed parts, to enable 3D printing with inks. The syringe pump is designed to be compatible with the docking system of the E3D ToolChanger and accommodates a three-milliliter syringe that can be easily swapped, enabling seamless material exchange.

Heater (Figure 1d): an E3D Hemera extruder with its nozzle and silicon insulation sock removed was installed to enable the curing of inks on the printer bed. During operation, the ink extruder and the heater can be used sequentially: first, the ink extruder prints a pattern; immediately afterward, the heater reproduces the printing trajectory of the syringe, drying the deposited ink as it hovers over it. This strategy enables the drying of ink while printing, facilitating the deposition of subsequent layers on top of the dried ink.

https://www.tandfonline.com/doi/full/10.1080/17452759.2026.2613185#d1e378

They don’t use metal-infused filament as a conductor but conductive ink.

I agree, but to be fair, this is not a new problem, nor is it one limited to the US.

I’m from Austria, and during the London riots (IIRC, that was in 2010 or 2011) I lived in the UK.

My parents frequently sent me news articles and snippets from TV news about things happening in the UK, and it was constant horror stories, almost apocalyptic. They claimed that all of UK was in riot and specifically also mentioned the area where I lived in.

In fact, all that I noticed of the riots was one peaceful demonstration on one afternoon and that was it.

Yeah, that’s not a job I’d be comfortable doing.

I think now is a good time to get into malware AI plugins.

Infinite scroll amplifies the “I’m never going to find that again” problem. That’s the thing I hate most about it.

There’s not really anything on that account, tbh. I only ever used it a few times to talk to friends when playing minecraft.

2D printers are way more difficult than 3D printers. The only reason we didn’t have 3D printers in the 90s is Stratasys and their stranglehold patents. Hobby-level 3D printers only became a thing because the Stratasys patents expired.

Before that they were just able to ask for €70k for what’s essentially a cheap ABS FDM printer.

Teamspeak is proprietary software, but it’s self-hosted and there’s an opensource TS3 client.

Since it’s self-hosted, enshittification is rather limited. If the new version sucks, anyone could just run an older version instead, so their are in constant competition with their own older versions.

Don’t worry, he fucks everyone. He fucks his employees, he fucks his customers, he fucks government and he fucks the victims of his software.

Teamspeak. It’s old and it’s still good. Uses no resources at all compared to Discord and there are thousands of free servers if you don’t want to host one yourself.

I uninstalled Discord yesterday.

I would totally go for that! After being together for a decade, it does become quite difficult to find something novel to enjoy together.

And if they are doing their job right, they also won’t lose that many customers.

Nah, they used oxygen-free bananas though.

The problem is with hardware requirements scaling exponentially with AI performance. Just look at RAM and computation consumption increasing compared to the performance of the models.

Anthropic recently announced that since the performance of one agent isn’t good enough it will just run teams of agents in parallel on single queries, thus just multiplying the hardware consumption.

Exponential growth can only continue for so long.