I like that this is an active area of consideration. I think there is lots of room for improvement.
But I also think that what we have today in Gnome and KDE is pretty good, and a part of the problem is that people don't realize how good they can be.
I think demo videos showing actual power users doing real work in each of the major DE's would go a long way. Part of the allure of tiling VM's is that they are an efficient way to work for certain types of users, but part of it is that those users are advocates and have impressive demo videos showing exactly how tiling VM's can be awesome, while a Gnome or KDE user is left to kind of figure it out, or worse, discover how they differ from a familiar windows or macos.
(I may be unaware of a great source of exactly what I ask for and if so please tell me!)
>And since PC makers are far more willing to experiment with form factors
What are you thinking of here? ATX is from 1995 and ITX from 2001, since then we've had stuff like the lamp imac, trashcan mac pro and nowadays the finger-thin imac and tiny mac mini from Apple.
I had Fujutsu laptops that were far lighter and smaller than Airs, touchscreens in the late 90s, convertibles, 2 in ones, touch bars (Thinkpad). Plenty of designs that don't build around ATX type specs, and things that do (open frame cases, and the vast array of expansion cards that benefit from it). PCs that snap onto a monitor. Things Apple copied or hasn't gotten around to yet then claim then as revolutionary. Apple has more integration and can be more strategic (when they don't need bailing out), but as one company they do far less exploration.
There have been a few cool ideas from PC makers I'll grant you that, but looking at the offerings from the 4-5 largest PC makers today it basically looks like it did 20 years ago.
I'll also grant that Apple has gotten more boring with age - the G4 imac is still the coolest computer I've seen to this day. Just so outrageously out there. Current offerings are efficient but a bit boring. Apple Silicon could make something like the trashcan design work, when it basically failed with x86 PC components.
Having said all that I wouldn't mind seeing an ecosystem of ARM-based components even if they followed the ATX standard. Being able to buy an ARM-compatible motherboard and upgrade my ARM-based CPU and pair it with separate RAM and SSD's would be nice!
>It is essentially a mechanism to transfer public money to private friends,
I agree completely and would go so far as to claim that presently in 2026 that money-transfer mechanism is the real motivation for any nuclear project in any country. Renewable options are the better path now from a technical point, but politicians and their buddies can't easily make as much money from that.
>Nuclear has no real cons other than time/cost of deployment,
That's not true. It's extremely expensive to operate for one, so expensive in fact that nobody in the private sector wants to do it without having nation states basically guarantee profits for the lifetime of the plant or accept all liability. Which in several countries means that he operator is the nation state. Like in France.
The profits are guaranteed by collecting extra taxes from the population, not by increasing the price of electricity. Which is why people believe France has cheap electricity when they really don't.
Also of course each nuclear plant carries a nonzero chance of a trillion dollar incident (we've had two so far) as well as a nonzero chance of millions of casualties. Both these cons are entirely unique to nuclear power plants. No hydro plant, solar plant, wind farm, coal, oil or gas plant can cause as much damage as a nuclear plant.
>t's that or buying gas from Russia.
Not at all, there are many options (including buying gas from someone else). Some options got viable so recently people still haven't realized it. Like batteries. They're cheap now, you know, and China can build as many as we want. China has also caused a severe drop in the cost of hydrolyzers, used to create e-fuel. E-fuel means you can use solar power in winter and during night, which removes the remaining weakness of renewable sources. Europe should spend every eurocent in that direction instead of nuclear. That's the path to true energy independence.
And anyway, europe still gets 70% of it's nuclear fuel from Russia. How is that better than gas?
> Also of course each nuclear plant carries a nonzero chance of a trillion dollar incident (we've had two so far) as well as a nonzero chance of millions of casualties. Both these cons are entirely unique to nuclear power plants. No hydro plant, solar plant, wind farm, coal, oil or gas plant can cause as much damage as a nuclear plant.
The Banqiao dam failure killed 250,000 people in a single incident. Pollution from coal plants running normally causes 1 million premature deaths per year. The deepwater horizon spill was a $100 billion disaster with an environmental impact greater than any nuclear incident. The Centralia fires in PA are still burning after over 60 years and lead to a town being abandoned. Nuclear is not uniquely risky by any rational metric.
>The Banqiao dam failure killed 250,000 people in a single incident.
The incident was a typhoon that caused 60 dam failures in a densely populated area prone to flooding. Estimates range from 26000 to 240000 people dead, but rationally one has to assume the typhoon would have caused some damage in these areas even without the dams.
So claiming this was a single hydro dam causing 250 000 dead is hardly rational.
If you live in a narrow valley downstream from a dam you are aware of the danger. You as an individual can take action to protect yourself. If you live upstream by even 10 feet you are in no danger at all from the dam.
With nuclear, people in Scotland are in danger from a plant in the Ukraine, can't do anything to protect themselves from it, and have to live with the consequences for 30+ years. That's much worse by any metric.
Deepwater horizon was bad, but nowhere near as bad as Chernobyl or Fukushima. Also it wasn't a power plant, not even indirectly as oil is mostly used for transportation, not electricity. Chernobyl and Fukushima were power plants.
So again, there is just no comparison between nuclear plants and any other power plant. Nuclear really does have a unique level of risk.
Risk is often defined as consequence x likelihood, and for nuclear the consequence is immense. Meaning effort has to be spent keeping the likelihood extremely small, in order for the risk to be low.
This has to be accepted and considered in any discussion about nuclear power. Pretending it doesn't exist is not rational.
> The incident was a typhoon that caused 60 dam failures in a densely populated area prone to flooding.
So we're not counting natural disasters that cause power infrastructure failure? I take it you don't consider the events of the Tohoku earthquake and tsunami to be a nuclear accident then.
> With nuclear, people in Scotland are in danger from a plant in the Ukraine, can't do anything to protect themselves from it, and have to live with the consequences for 30+ years. That's much worse by any metric.
And people in Scotland are in substantially greater danger from the emissions of coal plants in China, and will be until the world stops burning coal.
> Deepwater horizon was bad, but nowhere near as bad as Chernobyl or Fukushima.
It was much worse in terms of environmental impact, particularly the area impacted.
> Also it wasn't a power plant
I don't see any reason to limit the discussion to electrical power plants.
> Risk is often defined as consequence x likelihood
I was being generous in only looking at consequences without considering likelihood. The consequences are no worse than the dominant alternatives, but the likelihood has been astronomically less, hence Nuclear has been the safest power source ever utilized by mankind. So yes, in a way nuclear's risk is unique - uniquely low. I fully agree this should be accepted and considered in any discussion.
>So we're not counting natural disasters that cause power infrastructure failure? I take it you don't consider the events of the Tohoku earthquake and tsunami to be a nuclear accident then.
Of course not, if I did that's 20 000 dead from a "nuclear accident" right there. But that would be dishonest and silly, and I aim for rationality.
>And people in Scotland are in substantially greater danger from the emissions of coal plants in China
Not even if you assign all the blame of global warming to chinese coal plants would this be even remotely true.
>I don't see any reason to limit the discussion to electrical power plants.
You responded to a claim that no other power plant represent as great a risk at a nuclear power plant.
Maybe you should have created a new thread saying something like "a nuclear plant is not as bad as the entire fossil fuel industry worldwide"? ;-)
>The consequences are no worse than the dominant alternatives
You completely ignored the facts that clearly tell you that the consequences are orders of magnitude worse.
The statement about getting 70% from Russia is total bullshit
What non zero chance of millions of casualties? Chernobyl, a shit design that weren't allowed in the west even then is estimated to kill about 4k in total. In Fukushima the number is zero or 1 deaths from radiation due to low dose. And that's old gen2 reactor striken by tsunami and earthquake
I've seen what antinuclear europe does to remove the weakness of renewables - subsidies for new gas peakers in Germany
It's not, the estimated production cost for existing swiss nuclear plants is 4-6 centimes per kWh, which is already expensive electricity and frequently higher than the market prices today. That is also the self-cost price, meaning whenever the market pays less for electricity, ch nuclear plants lose money.
And as always, swiss taxpayers cover that loss so you won't ever see it in the operators financial reports.
Also worth noting is that EDF in France sells for 7 cents after being bailed out when sellng at 4 cents. This is not cheap electricity.
> What non zero chance of millions of casualties?
Just to take one trivial example:
If nobody is able to refill the water in spent fuel pools at any nuclear plant, they will catch fire. Assuming a radial spread of the fallout it will be near guaranteed lethal for at least 50 miles, probably lethal for some 500 miles, and you "won't want to live there" for even further than that. And this will remain the case for millennia to decades, depending on weather and distance.
So basically 500 miles radius, uninhabitable for several generations if you run out of water.
Then consider that Switzerland is covered by a circle of some 120 miles, and has a population of several million.
Do you think the 4-6 centimes per kWh covers a scenario like that? I don't. Good thing Switzerland has lots of water...
>If it's able to guess those 700 bits correctly, then they aren't true semantic information, and you really only have 300 bits you want to transfer.
A math teacher only has "this class is about math" to transfer, the rest is known. ;-)
Jokes aside, I think we're in a weird transition now where AI is used to generate text that looks good but is bad. In a few years people will know that and be more critical.
I think we went through a similar phase when DTP had it's breakthrough. Suddenly school papers were laser printed 300ppi times new roman and got more attention than better papers written by hand. But eventually that became the baseline.
I think Ai will make it so that well-written texts with clarity, good layout, correct illustrations, callouts etc become the norm, and will no longer impress anyone unless the information itself is actually good.
And if the information is good, it won't matter if it's AI generated or not.
Maybe things will change in the future, but I know where things stand right now. Would you rather learn about math by interacting with the teacher, by using the internet (including public LLMs), or by having the teacher tell an LLM "teach a math class" and then copy-paste its response to you? Option 1 is the best, but option 2 is at least better than option 3.
Oh it very much depends, your opinion there is certainly not always true and not shared by everyone; you only know where you stand right now with any certainty.
It depends on the teacher. I’ve had good and bad math classes, and LLMs today are quite a bit better than the bad ones. The worst human teacher in my memory didn’t offer interactions. He walked in, turned his back to the class, wrote equations on the board for 45 minutes, and then left. The best math teachers, the ones better than LLMs, are the ones who share the joy and sense of discovery and history of math, and not just the mechanics. But there aren’t that many teachers of that sort.
Learning by using the internet without LLMs is rarely very good, but more often than not in my experience sucks much worse than using LLMs. If you include using public LLMs in internet usage, then it’s not very different from just using LLMs that search the internet. I have heard that a huge swath of today’s high school and college kids are reaching for chatGPT before Google (which is incidentally OpenAI’s goal), and that many of them would rather talk to chatGPT than talk to a teacher. I’m going to refrain from making any claims, but I believe there are a lot of people who disagree with your ranking of the options.
Foreign language learning is one case where I love using LLMs, because it’s not typically an option otherwise. You can practice non-stop and have conversations with someone fluent in a language who will be infinitely patient with your mistakes. This is true of math and other subjects too; using LLMs to practice, so that the human teacher isn’t the bottleneck, to supplement and reinforce the human interactions, is usually better than using the internet without LLMs. The other reason many people prefer talking to LLMs is the lack of judgement. If you aren’t getting it and ask the teacher one too many basic questions, they treat you differently. Sometimes it’s necessary and helpful, and sometimes it’s harmful and takes a long time to change. LLMs don’t do that, they just explain and explain. That lack of judgement is a big reason many people prefer LLM interaction to human interaction.
Okay, sure. The difference between option 1 and option 2 is completely irrelevant to this discussion and I gave what I thought would be an uncontroversial take on it because I thought it didn't matter. The point is at least one of option 1 or 2 is almost always better than option 3. Option 3 is what I and others are arguing against.
Irrelevant? How so? And why did you present two different options if you thought it was irrelevant? I thought you were trying to make a point about human interaction. From my POV, your options 2 and 3 are much closer together than 1 & 2, so if you’re not suggesting that human interaction is preferable to LLM interaction, then what’s your justification for claiming any option is better than any other? I still completely disagree with your clarified point, and suspect many many other people do too, and are demonstrating that preference by using chatGPT for their classes instead of using either the teacher or Google. It’s a simple fact that interacting with an LLM can be preferable to interacting with some teachers some of the time, and LLMs currently often make interacting with the internet more pleasant and more efficient. Your ‘almost always’ still sounds to me like one person’s opinion that is not shared by all.
We're not talking about humans vs LLMs in general, we're talking about humans copy-pasting LLM output as if they wrote it themselves. It's less "LLMs are bad" and more "if you want to communicate something specific to me, LLMs are a bad substitute for your own writing".
Oh I see I’ve misunderstood. Your analogy is restricting the scenario to no interaction with LLMs. In that case, there are still some reasons LLM writing might be prefereable for students: for non-native speakers, and for math teachers new to match developing a new curriculum (sucks, I guess, but it certainly happens), just to name a couple.
One question to ask is why is your contrived LLM scenario any different than a math teacher using a textbook, or district/state worksheets? Math teachers typically do not write the course text or exercises, and never have. Very few math teachers do their own writing.
>Maybe things will change in the future, but I know where things stand right now
It's not an either-or though, LLM generated text already provides lots of value in many situations right now. It's also used for fluff, yes, but much of what humans write is fluff too, reporters often get paid by the word.
The point is that the value in a text has nothing to do with whether it was generated by an LLM or not. What matters is if it's useful or not.
Moving memory around is a bottleneck. Non-unified memory usually means copying over the PCIe bus which is way slower than RAM (and way way slower than VRAM). Actually unified memory means you don't need to copy anything at all though which is the absolute best case for performance.
It isn't really, as long as you don't care about power consumption, physical constraints and money. Basically desktops <2025 (and hopefully >2027).
DDR is optimized for latency and stability at the cost of bandwidth whilst GDDR is optimized for bandwidth at the cost of latency and stability. GDDR is pushed so hard these days that a small percentage of errors is expected and corrected because this is still faster than running it slower but more accurate.
GDDR7 often has 10-20x the total bandwidth but 3x the latency of DDR5. Graphical workloads want as much bandwidth as possible but care relatively little for latency. Conversely, applications love low latency but don't really see any performance benefit from higher bandwidth.
So basicallyt you have workloads that are diametrically opposed and running unified memory forces you to compromise.
Is GDDR7 that's ECC then just the super well binned stuff (plus the ECC parts added) ? I've been wondering why we don't see more cards using it as they perform pretty damn well. Take the Nvidia 6000 Pro Blackwell for example. The compute is insanely fast assuming you can fit what you need in 96GB of ECC GDDR7
The price is that you essentially glue CPU and GPU together, which limits total compute, from a size and thermal perspective.
This is really not a limit because of unified memory -- in principle, PCIe GPUs could read/write main memory without the CPU. But it's a limit for /fast/ unified memory, because fast means close.
So unified memory is great as long as the integrated GPU is strong enough. Then it has two advantages:
a) probably faster transfer CPU<->GPU (but that's an implementation choice for the non-unified case
b) If you either need a lot of memory for the CPU or the GPU, but not for both at the same time, you pay for memory only once.
In theory perhaps but the benefit is not as relevant with a weak iGPU. In practice all PC's with performance ambitions had a dGPU until Strix Halo and Panther Lake.
>You don't think it is unfair that projects that have basically been xkcd 2347's "a project some random person in Nebraska" don't get a penny, while DHH who gets millions for a bunch of dotfiles, is unfair?
How can we have progress if every shoulder to stand on will cost money? Should we pay every time we walk on the pavement or drive on the road?
Sometimes work is just "for the common good", and sometimes that work is harder than the latter part that actually makes money. It's peculiar, but I don't think it's unfair.
Linus Torvalds will never in his life make a fraction of what Github has made, let alone what RedHat, Ubuntu, Suse etc have made.
But he also doesn't have to pay the GNU project for the use of their compiler. And the GNU project doesn't have to pay him for using his kernel. And RedHat doesn't have to pay either one of them (although they do, voluntarily).
DHH has also made Rails, which is the foundation of several billion dollar enterprises. So he's at least on both ends of the argument.
Can't agree with this, there's clearly room for improvement
>Everything that makes Linux great, simply hasn't translated to the desktop GUI.
Do agree with this though. Fragmentation is not ideal for having a consistent UI and a consistent UI is needed to develop muscle memories so it can be used effectively.
>We are done with desktops. They will be handed off to ai.
I don't think that will happen while AI costs dollars per minute of actual use. Being able to use a computer for free will be valuable for a long time still I think.
Much of the visuals and features in Omarchy is config yes but going from a blank drive to useful development environment in one minute is not just a set of dotfiles. It's clearly a distro now.
Don't let politics cloud your thinking. There are a lot of good reasons to like Omarchy, and other distros should pay attention to the substance.
Omarchy shows how it's possible to realize a very clear, distinct vision of how desktop computing should be. I'd like to see the same amount of effort put into realizing several other visions!
An LLM will happily take an instruction like "my currently super-tweaked Debian install is awesome, create a set of scripts that recreates the exact setup starting from a minimal netinstall". It will create a VM, install the base debian, iterate until it's right. You can have it make a user-friendly UI on top of the scripts, have it install drivers, commercial software that Debian shuns from their distro, and so on.
If you have good taste and attention to detail, your setup might have universal appeal. You will have created a Debian flavour that might be better than anything anyone gets when installing official Debian.
If you have poor taste and attention to detail, at least you have a one-command way to get your familiar janky setup installed from scratch. :)
Surely you are only thinking of the visuals then? I can certainly understand that, but the visuals are not what I mean.
Things like:
- Sweating that the install should take a minute.
- Ensuring that when you change themes, most relevant apps also change themes.
- Including popular commercial apps with 0 friction.
- Consistent design in menus, features and default apps.
- Clear explanations of how everything works, including video demos.
All of that is good taste, and would be good taste even if the visuals were "My little pony" or "Lovecraftian horror" instead of 80s retro. ;-)
The only "commercial app" it themes automatically is Chromium. Obsidian you have to manually theme, the rest is just DE stuff that other distros also automatically theme.
There's just nothing about Omarchy that makes it special or "good taste", in the sense that something like Regolith[0] is special. Or NixOS or uBlue.
I'll say again, Omarchy only got attention because of its creator, and now it is draining away funding from other Linux DE stuff. Bad taste.
I messed upp the list, didn't mean to imply that it can theme any commercial app.
It was a separate point - Omarchy lets you install commercial apps as easily as any other app. Including things like Spotify, Claude Code, Codex, steam and even Windows itself (as a VM). All from the same menu and mechanisms. That's just undeniably useful.
Also, Omarchy does theme not only Chromium, but VS Code for example, and several other "blessed" apps. I get the impression that you haven't actually tried it?
Regolith is similar to Omarchy in many ways but is an older initiative that hasn't gotten as far despite having had more time.
When I look up how to install apps in Regolith it says "use debian package tools" (sad trombone). That's clearly worse than Omarchy.
You can decide that Regolith isn't gaining traction because it's not made by a techbro mascot if you wish but I think the explanation is easy - Omarchy is clearly better.
I'm not saying it being by DHH hurts of course, but IMO it's less about him being connected than it is about his enthusiasm and energy. If being well connected was all it takes we'd all be living in the metaverse now. :) There has to be substance.
It'd be fun to see if there are other initiatives that are as polished as Omarchy - the latest Omarchy demo is here[1], if anyone has links to similar demos of other setups it'd be fun to compare.
AFAIK Omarchy is the slickest one around so far but I hope open-minded devs will take inspiration and start competing instead of just dismissing.
> Omarchy lets you install commercial apps as easily as any other app. Including things like Spotify, Claude Code, Codex, steam and even Windows itself (as a VM). All from the same menu and mechanisms. That's just undeniably useful.
Both Gnome and KDE have done this for years..
> Also, Omarchy does theme not only Chromium, but VS Code for example, and several other "blessed" apps. I get the impression that you haven't actually tried it?
Literally their own documentation says it only themes Chromium and you have to do Obsidian manually. That also points to another of its biggest criticisms: because its vibecoded (including the documentation), it is basically a house of cards.
> It'd be fun to see if there are other initiatives that are as polished as Omarchy
I mean, if you just care about riced setups there's stuff like this[0]. But both in Omarchy and that setup, the beauty is only skin deep. If you dip your toe even one bit out of the tested paths, you're gonna run into regular Linux ugliness. Its not something I mind (I run NixOS on my desktop and homelab server and clients), but Omarchy shouldn't be venerated as some new era of Linux desktop. Its not.
It wouldn't even be much of a problem only if Omarchy wasn't suddenly getting millions in funding. That money should go elsewhere.
No they don't, they don't even consider it their job to do it. It's the job of the distro's package manager, or a 3rd party app wrapper like Snap or Flatpak.
You can of course configure a distro using KDE and Gnome so that the experience becomes as good, if you want. Which is how things have been in Linux for decades. "Walk the last mile yourself".
My point here is that Omarchy shows that there is a last mile to be walked. By having commercial apps included it becomes interesting for commercial vendors to invest in.
I think that (and the overall buzz) is the main reason why it's getting funding. A hundred thousand wealthy geeks installing a platform where our apps are trivial to install and we won't have to deal with MS or Apple if it takes off? Ok, have a bag of cash why not.
Distros where it takes a nerd to even figure out how to install the official nvidia drivers? Nah.
>Literally their own documentation says it only themes Chromium
Ok so you haven't tried it. Then I understand. It's not just skin deep, but you won't know that from just looking.
I'm not saying it's a "new era" of anything btw. I'm just saying there are good ideas there that could benefit any DE or distro.
I am telling you again and again that other distros already do this.... there is nothing special about being able to install Spotify from the get go in a unified installer together with packages, you can do the same on Gnome or KDE with their respective software center apps.
You're just overly enamored. Omarchy has nothing special behind it except the dude who made it.
>I am telling you again and again that other distros already do this....
"This" includes installing in a literal minute, having codex, claude code, vs code etc installed as easily as emacs with 0 extra config. Also things like localsend, tailscale, bitwarden etc. Very useful stuff and it all "just works".
Do your nameless distros do that too?
>Omarchy has nothing special behind it except the dude who made it.
I don't care who made it, just how good it is. You seem to care too much about who made it to ever discover how good it is. ;-)
I didn't say anything about usefulness. Even a bunch of dotfiles can be useful, Oh My Zsh and others were very popular after all.
The thing I complain about are that these companies that build their fortunes partially on open source benefitted a lot from systematically underfunded developers. But they only make small contributions to those projects (if at all). But if their millionaire buddy shows up, they all stand in line to donate millions.
I am not sure what it is. I guess partially they want to associate with the popularity of Omarchy, partially they are doing a favor to another rich techbro, some may agree with DHH's politics to 'drive woke out of open source'. But it still sucks that they don't give that many millions to Debian/OpenSSL/LibreSSL/Rust/pick your other favorite infrastructure project that they highly depend on to make their money.
But I also think that what we have today in Gnome and KDE is pretty good, and a part of the problem is that people don't realize how good they can be.
I think demo videos showing actual power users doing real work in each of the major DE's would go a long way. Part of the allure of tiling VM's is that they are an efficient way to work for certain types of users, but part of it is that those users are advocates and have impressive demo videos showing exactly how tiling VM's can be awesome, while a Gnome or KDE user is left to kind of figure it out, or worse, discover how they differ from a familiar windows or macos.
(I may be unaware of a great source of exactly what I ask for and if so please tell me!)
reply