adornKey

Sounds like a decent challenge. Needs AI Translation built into the forum infrastructure... But I'd give it a try. Before there was AI, I chatted a few times successfully in Chinese only using some translation tools. Now with AI it should be even easier.

In case there is a decent forum about tech somewhere in the Alibaba cloud, I'll be interested. HN has built a certain unhealthy bias in a lot of subjects. But so far other places haven't gained enough traction, yet.

Nice. I used Q4_K_M to have some headroom. But yours seems to fit nicely.

Why? Unlike loads involving a real physical process there is absolutely no need for AI-training to be constant.

Ozone is an interesting topic. CFCs seem to be very potent climate gases. But I haven't checked any calculations about them, yet. I'd love to see a good analysis of the absorption-spectrum. Adding something new to the atmosphere has a lot of warming potential - but the question always is how fast it reaches a level of saturation. For ozone and CFCs years of media coverage haven't brought any insight. Having 3 different updated versions of Dupont-products in the atmosphere could be good or bad - most likely people haven't bothered to check, yet... But they're all full of furious knowledge. People "know" that banning CFCs "cured" the ozone hole - but they don't ask why it shrunk too early, and why the situation hasn't changed at all for decades now...

I think most likely the banning was good - but the reasons don't really make sense.

Hm. Nowadays a lot of games look alike. They use some 3D-Engine and most of the work is in the 3D-modelling and writing some interesting scripts.

For modelling and scripting I think we're not far away. A lot of games just reused old historical stories or fiction and a lot of stories feel like cheap soap operas. As soon as an AI can separate the good from the bad scrips it'll be mostly done.

Today even tiny CPUs are really fast. Locally you have to mess up badly to run into trouble. But of course people will do exactly that...

Most real world problems still can be solved with 32-bit software, so the last ~20 years running out of RAM always counted as "using defective hardware". AI workloads now make things interesting again, but it's not that easy to hit the ceiling with real world workload.

Cache is indeed very important. Optimisations like that are gone when you go for distributed computing. Sometimes adding a single nop can do wonders. I wonder how many percent of developers have something in their toolbox to profile for that.

I once found that my old Microsoft Ergonomic Keyboard had an option to disable the Caps Lock Key in the settings. I immediately used that. Then after a reboot Caps Lock was ON and the Key was disabled - Shift now also didn't disable Caps Lock...

That taught me that sometimes you need the Caps lock Key badly...

Software can control the state of the key - and you need a way to get rid of that. Back in the day - to enforce only Upper-Case Text in some Text-Field - some software even messed with global Casp Lock State when entering the Text-Field...

But remember that markets can stay irrational longer than anyone can hold his breath. If they get more funding there's a good chance they'll invest more in the destruction of the remaining production capacity. Admitting that with normal pricing anyone could have a decent AI-machine for 2K is hard - prices for acceptable AI-machines most likely will go >10K first.

Maybe sycophantic nature is a good fit for the legal system. A successful lawyer once told me that the most important thing is to know your judge. Objectivity isn't a big thing in court. They'll cite random newspaper articles as evidence and throw out expert opinions - if they like. There might be a way to appeal - but that road often is not functional.

The relation of consciousness and being in control is interesting. I'm not so sure about the decision-making experiments - maybe they just measured latency in the communication layers?

I think being in control is a rare thing. It's hard to change even the smallest habits - the default result is failure - but sometimes it's possible. The only thing that seems to work reliably is subconscious manipulation of people using propaganda and repetition. That's for me the main reason to believe that at least 90% of people don't have active functional consciousness in their loop.

Decisions of Plants work in different time scales, so they're hard to perceive, but I don't think they work that different - it all boils down to maximising some gain-function using some chemistry for memory.

Your theory about going automaton is interesting. I've seen intelligent persons that communicated very well going "crazy". (I think women do that a lot) - like a looping LLM. And later they came back to normal - but since then I have some doubts about their internal state...

One step further is to ask how conscious your mind actually is. There is a lot happening on autopilot - and everybody usually checks out for a few hours at night. Maybe consciousness is a rare temporary thing.

I think evidence suggests that humans aren't conscious most of the time. So it wouldn't surprise me if 95% of the time people are just stochastic parrots. But maybe that number is even close or equal to 100%.

Intellectually a lot of humans perform worse than LLMs and a lot of people (most of them) are completely unable to process abstract concepts and basic logic at all. Can those people truly be called conscious? Is consciousness worth something without the ability to reason?

Great! Now add a liquid ocean and add scattering in that liquid. Stretch goals: absorption and reflections...

I'm running a server in the 5K-league. And the results are very good. I get about 150 Tokens/s from Qwen3 for coding. And about 50 Tokens/s from the newer non-MoE Qwens.

I wouldn't bother with less than 32GB of VRAM. With 16GB you can already run something usable, but 32GB gives you much more power. 9B and 14B are only interesting if you want to tune models yourself. The sweet spot now seem to be around 27B-35B.

Real Linux on phones is a thing. They're usable, but most hardware is getting old. E.g. PinePhone still works fine, but they recently announced that it's unlikely that we see a new version. They mention that it's hard to be competitive with hardware when people can install PostmarketOS and SailfishOS on cheap old Android Devices for a similar experience.

https://pine64.org/2026/03/24/march_2026_fosdem/#where-is-th...

In the long run - without PinePhone - people will lose more and more control over hardware and drivers.

Good point. Also, most humans seem to have no problems believing even stories that are self-contradictory. Philosophers from all periods have often stated that the situation with human mind and reasoning is almost hopeless.

The news here is that AI has too much trust in the internet. The first time I allowed tool-calling, it started googling up some nonsense instead of thinking... But I think at least it's possible for the AI to evaluate the quality of the source - you just have to ask for an analysis, and you'll get a reasonable evaluation. With humans, something like that just doesn't work - they'll get aggressive or might even start throwing bananas...

I think using violet as a name for the entire color-range around (~128, 0, 255) is also common. So in a sense purple is an element of the violet color-range. But as points they are distinct. I think purple is more specific - as a color-range it'd cover less area.

Purple has a lot more red. (157, 0, 255) vs. (128, 0, 255). Good to have learned something today...

And don't forget open source games. Before going for the indies, I'd suggest downloading and winning all the available major open source roguelikes. And after that, start creating mods/patches for those. Once you're done with that - and not too old of age - maybe think about spending some money on games again.

So, to follow up I'd say in the context of EML I think ln(-1) = pi*i is the natural fit. Pi and Tau are both more advanced constructs.

The construction so far uses ln(-1) to get to pi - so far no easy way to tau.

Although x + y is surprisingly more complicated than you'd expect at first. The construction first goes for exp(x) and ln(x) then to x - y and finally uses -y to get to x + y.

Ätsch!

Having the focal point up close for a long time isn't that good for the eyes, so sitting closer than an arms length to a desk monitor isn't an idea that lasts well.

100 dpi with subpixel rendering already maxes out angular resolution (horizontal). It doesn't max out everything (retinal), so you still see some artifacts, but practically this is not that relevant. The price in energy/bandwidth rises quadratic for very little gain.

To get the equivalent of 4K at 100 ppi - with 200 ppi you have to put the burden of 8K onto the GPU... For now this is absolutely not good - High ppi is ok for small monitors and handheld devices, but for a decent desk with several good monitors GPUs just aren't ready yet.

But the angular resolution of the eye doesn't rise. For a desktop monitor 100 ppi practically already reached the limits. Anything beyond that is just additional burden for the GPU and a waste of bandwidth. Surely you can increase resolution just to make font rendering easier, but you also have to pay the price in energy consumption or speed - without any visible improvement.

Microsoft did a lot of great work on Fonts in the past. Recently it looked like they abandoned per monitor subpixel-rendering?! In which direction are they heading?

The explanation seems to be that it looked good in some old fonts. But I think it was always some kind of abuse. On old Typewriters the accents were usually used for accents (é è). They didn't move the cursor, so using them for apostrophes wasn't that comfortable and interrupted writing flow. Accent + space looks a bit like a quotation mark, but the right place of an accent is usually on top of a letter.

A more general approach are Encyclopedias of integer series. I think that works better than just focusing on single numbers. Hm. How many numbers are there, that are interesting, but not part of a series?

Maybe that game exists, but it's an old word and you can find references for it that are hundreds of years old. Its meaning fits to the browser. Your statement that it's just some kind of reference to some special game is not correct.

A few years ago I liked Ars Technica, but then somehow I think quality went down the drain. Did something happen to them a few years ago? If they get rid of the crazy reporters and go AI only - maybe the quality will improve again to a readable level.