The goal is banishing any and all questions of what to download, what desktop, what apps, packaging formats, software stores, user accounts, backing up, all of it: run the browser locally, sync the data – and only the data – to the cloud.
That is not really popular in the open source world. Who finances the cloud, is it open, who can read or sell your data?
I think the idea pairing would be with with a very open cloud-app based solution. Something like Sandstorm where you can install your own cloud apps and have them available via the web on all your devices.
I really with Mozilla would get in this game. For most apps for most people, hosted PWAs make the most sense. But right now we're either stuck with local native software or only web apps from closed providers.
It would probably make sense to build an end-to-end encrypted sync library using S3 for storage. Then users can choose which cloud to use, or self-host.
I would argue it goes against the desires of the open source world.
ChromeOS is a product designed to suck people further in to the Google ecosystem. All its limitations, like basically being a web browser with access to the Android Play store, are built around that. The simplicity of not having to worry about anything else is how Google sells that as a benefit of their platform to the masses.
I don't know any of my fellow nerds who want a ChromeOS kind of experience, so why would any of us dedicate our free time to making one?
"I don't know any of my fellow nerds who want a ChromeOS kind of experience, so why would any of us dedicate our free time to making one?"
Interesting perspective. Per the article, this is a method of circumventing the W10 issue coming up.. I would guess that you have family members or friends that might be caught up in that? Even ones that are not nerds?
I think project bluefin has a similarish philosophy of being a "cloud first os", would definitely recommend checking it out: https://projectbluefin.io/
In all honesty though, there's a tonne of easy to recommend distros which don't need set up if you are a "normie" just wanting to get online:
- Elementary
- Mint
- Ubuntu
I think the real "secret sauce" for ChomeOS' success is basically big coorporate backing and being from-a-store-buyable. I think the truth is that nobody other tham techies is likely to want to install an OS themselves.
I recently (two weeks ago) decided to finally switch from Windows to plain Ubuntu as a daily desktop. I'm well experienced with it from managing servers and have used it occasionally at a desktop in the past.
I've had - and continue to have, a ton of minor and a few large problems.
First, wifi. Apparently there's often problems with Intel killer 6e laptop cards, which required me to enter loads of console commands that I didn't understand to fix. I did finally fix it, but I'm not sure exactly what did it. A couple of hours work over a few days to fix this.
I work from an external monitor plugged into my laptops HDMI port with the lid closed, and it seems Ubuntu is massively confused by this. I'm constantly having to open the laptop to get a login screen to show, then logging in and closing it, then opening it again because now the desktop doesn't show.
My laptop screen resets to minimum brightness every restart.
I did a driver update and suddenly my external monitor wasn't recognized at all so I had to roll back.
And at least several hours more on other issues. Oh yes, if I suspend then try to wake, I get a black screen or occasionally the Ubuntu startup screen but can never get back to the login screen until I do a hard reset.
The Bluetooth audio was set to headset quality (mono) and the only way to fix this, following 20 minutes research, was to install a different audio settings app.
All of these things, I can fix. I don't want to, but as a technically savvy user I can do it, with enough hours put in. But there's no way I'd suggest anyone else in my family or friends to use it unless they clearly tell me that they would enjoy tinkering with their operating system.
It is unfortunate that Ubuntu still has the mind share it does. I abandoned it a while ago as too unreliable. It is effectively Debian testing (they call it testing for a reason) with proprietary addons that have even less testing than Debian testing.
Originally, Debian making it difficult to get non open source firmware during installation was a reason for using Ubuntu, but now even that justification gone. Meanwhile Ubuntu anti-features like snap accumulate.
Ooof! That sounds rough and also makes me realise a huge oversight in my original comment.
I've been using linux on desktop for about 12 years and will always pick a laptop that's known to have good compatibility otherwise the endless hardware issues are a big timesuck.
Some distros are better than others for hardware support but I think it'd be disingenious to make out like any solve the problem.
If you have a laptop with good support, or even better, a linux specific one, it really is an experience that can work out of the box.
Maybe that's a big part the appeal of being buy-in-a-shoppable. Somebody has figured out for you what works, and you don't have to be your own guinea pig.
Hardware problems are a frequently understated barrier to switching, particularly for laptop users. Long time Linux users are generally selecting their machines with Linux compatibility in mind and so don’t see much of this, but the corpus of potential switchers haven’t, and it creates a lot of friction. It’s difficult to convince people to replace their perfectly good laptop with something else just for the sake of trying out an alternative OS.
It's good to know that "Linux is only free if your time is worthless" still holds true after all these years. From the way many talk, I honestly thought it might have changed.
Fedora might support your hardware better, (newer kernel) and not have snap.
Windows also sometimes has these kinds of problems. Plus I usually have to dig up many commands to turn off user-hostile defaults or free up a quarter terabyte of disk space wasted. Was an obscure diskadm cli I’ve never heard of.
The perfect OS doesn’t exist, maybe not even a good one.
This is ultimately a business model problem. There are FOSS projects out there; some of them are even listed in the article. The problem is that making it "Just Work(tm)" is hard; it's a lot of work, requiring a large number of engineers. And funny thing, engineers prefer food with their meals.
Saying "FOSS world" we need XXX is pretty useless. As a FOSS mainatinaer, the answer I always give people demanding their favorite pet feature is, "clean, maintainable patches are appreciated", ChromiumOS is free software; someone could take it use it as the basis for something like ChromeOS. But the author of the article has basically admitted that the derivitives of Chromium would quickly fail if Chromium stoped being something that they could free ride off of. Shouldn't that tell you everything about what the problem is with this picture?
Programmers do stuff for free, all the time. People also donate to foundations so some programmers can be paid to make free stuff. It's how we got the Linux desktop as we know it.
Programmers often stop worrying about the details once they're satisfied, though, and there are a lot of details. Yes we could add thumbnails to the file picker, but we also need to support Nvidia hardware and the archive application is missing a few much more interesting compression formats, plus there are a bazillion logged bugs that actually crash software rather than being a minor inconvenience.
In theory someone could set up a collective of programmers, designers, and translators to create a usable Linux desktop for the Windows 10 refugees, but it'll require a lot of effort from a lot of volunteers and a lot of people who happen to share the same principles and preferences.
ChromeOS works because there's a single entity that can hire a small group of designers to tell an army of programmers what they want everything to look and work like.
There are actually projects and companies that fork ChromeOS, but they're pretty small and once Google stops maintaining the OS they're going to collapse pretty quickly.
> ChromeOS works because there's a single entity that can hire a small group of designers to tell an army of programmers what they want everything to look and work like.
ChromeOS works because it's pre-installed on selected hardware devices where it's been configured and tested before any user even get it into their hands which would solve the majority of the problems no matter what OS you're using. Then they also take all power away from the user do make any changes to those systems which helps solve the rest. Ultimately though, most people who know anything at all about computers want the power to make changes to what's installed on them..
Everyone I know who got a chromebook did it first and foremost because they were inexpensive and every one of them at one point or another found themselves frustrated by the limitations. A privacy respecting linux system won't be able to use massive amounts of data collection to lower the price of the hardware like Google can.
People will do a small amount of work for free. But if something requires dozens or hundreds of programmers working full time, the chances of this happening is much less.
And if you want something which just works, it means you need a huge number of people doing grunt work, and that's the kind of thing that people are less likely to want to do in their free time.
SteamOS is the closest thing to making this work. Valve has somehow worked their way into being a software store that not everybody hates.
Companies should be happy that there’s a FOSS community that lets them build up on top of it for free. It is a great deal! I wonder why there isn’t an EpicOS yet.
I don't think this would work. A big part of why ChromeOS is so simple is the Google services backing it. Without the unified login, automated backups, verified app store, and hardware certification ChromeOS wouldn't be the easy to use system that it is.
FWIW, back in 2013 I was working at Deutsche Telekom in a project that was attempting to do "ChromeOS, but for Enterprise", built on top of Ubuntu and a heavily modified Firefox. We had a list of target devices to test against, but given the Ubuntu base it was pretty easy to get it running everywhere.
All the office were based on open source systems. DT wanted to make money mostly on their data plans + device management services. The project failed for one reason: Enterprise IT is the opposite of price sensitive. I had talks with a couple of CIOs who were interested in the idea, but realized that commiting to a solution that could slash their budgets by 80% would be career suicide.
> Many of our technologies are distributed together within Endless OS. This is not your typical Linux distribution. We don’t use rpm, apt, or any other packaging system. In our quest for simplicity and robustness, we use a read-only root file system managed by OSTree with application bundles overlaid on top. Most desktop Linux distributions are oriented towards tech-savvy users and developers, but we have a different target user.
Linux desktop in general suffers from not having good "default" options. I can't really recommend any single distro, can't recommend even a single desktop environment as good for most people. Mint is very good, but it looks painfully dated in many places. Fedora is very good, but Gnome Shell is not for everyone in the slightest and debian base is just too wide-spread to ignore. Ubuntu is very good, but actually it's not that good in 2025 because canonical spent 15 years reinventing the wheel and not finishing any.
KDE has its own problems, though they’ve been improving in recent years. Out of the box defaults still need some work for example, and settings can still be confusing and difficult to navigate for less technical individuals. It also still has some behaviors that aren’t found on other desktops that can put off switchers.
The best DE for switchers would be one that brings as few surprises as possible and has defaults good enough that the overwhelming majority feel no need to change them. Basically, there needs to be near-perfect clones of the two big commercial desktops that don’t try to be clever or shake things up in any way so the switch is close to seamless.
Debian Testing packages may also still be too old, depending on the user. Anybody having to work cross platform or collaborate with people using other platforms is going to wind up dealing with version mismatch issues, since commercial OS users have most of their apps auto-updated to latest without Testing’a lag.
Debian Testing packages are not too old. They are mostly very nearly bleeding edge, if not they’re a month or so behind. Ridiculous you would blindly say Debian Testing is outdated.
Probably fine for most people, but not a good choice for low-hassle setup with "recent" nVidia cards (last ~10 years). That's almost all nVidia's fault, and I can't say any distro handles their cards flawlessly, but Debian is one of the worst: the drivers are (accurately) classified as "nonfree" and so not available by default, setup may require config file tweaking, and KDE Plasma prefers to support Wayland while nVidia prefers to support X11, leading to weird issues (more than usual, anyway).
Yeah, it's kind of niche to find someone who has an nVidia card, uses it for gaming/compute, has some reason to run Linux, and is afraid to dig into config files etc. But Microsoft's behavior of late is creating some demand here, and I too get tired of having to fix things after random kernel/driver updates.
My experience is that distros with traditional package management can't be counted on to work over the long term without manual administrator intervention.
Debian in particular would not be my first choice, or my second, for people who can't do their own system administration.
(And this is a bit of the problem: nobody can agree on this stuff)
Nobody agreeing on this stuff is freedom and beauty. If you want a managed OS there are already many options.
You look at Linux as an issue because someone you know can’t do system administration, something Debian almost requires none of. I look at Linux as an opportunity to rip people off their safe OS and introduce them to something that works but allows deep customization when they’re ready.
You’re operating out of fear, I’m operating out of faith. I really don’t care if the user “gets it”. They barely get their mobile device.
This touches on a larger problem that I think we tech people don't address as much as we should. That the modern big tech OS (Windows, MacOS, Android, iOS) has a huge advantage, which is that it's managed. Actively supported by the vendor, well after the purchase. That, and the heavy restriction of the user makes the computer orders of magnitude more stable.
With Linux this is not the way. The closest I can get is unattended-upgrades.
I think Fedora Silverblue, Bazzite, et. al. is the right direction for linux on PCs. It still feels a little immature yet, but also I'm way more confident in the system's durability than I've ever been in the past.
Based. I run the same setup myself, but "testing" does mean testing, and while it usually works fine, you have to examine apt output and make sure it's sane, before you update your system.
I don’t have problems unless I run apt-dist-upgrade before a release cut (which I never do). Otherwise things have upgraded without pain or issue packages are held back until they’re ready for testing using apt-update/upgrade
I don't think it particularly suffers from not having "good defaults", because it doesn't get adopted by word of mouth. And people are not shopping around The Best OS that they find, and sadly, they evaluate that Linux just doesn't have the juice.
Two things here.
One, OSs get adopted because of the package they arrive in. Laptops have Windows bundled, schools get computers with Windows on the cheap. Phones come with whatever they are integrated - often, not even a major upgrade is possible on them, let alone an alternative OS.
In my opinion, products like the Steam Deck are what get Linux adopted. They provide excellent functionality out of the box.
Two, PCs in this age thrive on interaction. Overwhelmingly, they are bought to interact with other PCs, either directly by network or through formats of specific software, like docx. Therefore, the functionality depends on how well they can participate in a network. This is where Microsoft really made it big with how ruthlessly they pushed Windows as an app platform and Office as the productivity software baseline.
Hence, in the end, what makes Linux really competitive today is the proliferation of web apps, because Linux has first class browsers, and the existence of Wine and its related software. Take these two away and the OS is not even talked about as an alternative to anything, just a specialty, or a curiosity.
Really, an alternative OS can do two things. It can either set the trend, which is not very likely, given that we are talking about alternatives. Or it can follow the trend, or "embrace" it, in MS lingo.
Some do come with Linux preinstalled, but there is a lot of mess in this regard. "Windows tax" was one of such things, the Windows bundling went so hard[0] that it was more expensive to buy a computer without Windows than with[1]. Absolute nonsense, that more or less stopped now. But of course we are well past the Windows platform establishing itself, so now the network effect is a larger contributor I'd say.
We all know about the Windows tax and Microsoft making Windows XP free for Netbooks.
OEMs making custom distros didn't help.
If anything that was the precursor of the Android distros mess that we got afterwards.
They also did the same with CP/M, MS-DOS and Windows since forever, including UNIX clones, why wouldn't they do otherwise with GNU/Linux.
As former owner of an Asus Netbook 1215B bought with Linux pre-installed, even that way wasn't without issues.
OpenGL support was always lagging behind what hardware was capable, GL 3.3 vs GL 4.1, hardware video decoding only worked during Flash heyday, and wlan drivers always had issues with hardware when doing large downloads.
OEMs were smaller than Microsoft, that was the problem I think. Valve seems to have success with Linux, which is maybe because of their niche, and size. Smaller players just don't get to be trendsetters.
Valve has failed to convince studios already targeting POSIX like environments like Android NDK and Orbis OS (PlayStation), to port their games to GNU/Linux.
Had to switch translating Windows games, as plan B kind of solution.
It remains to be seen for how long Microsoft will tolerate Proton, or just like it happened with netbooks, will drive other handheld OEMs to use Windows and thus in a decade from now people will foundly remember Steam Deck.
I went from being a UNIX/Linux zealot during the mid 90's, writing M$ on email signatures, back to Windows as daily driver, because I got tired of all fragmentation, hunting for hardware support, that in theory would work but the actual firmware on the bought device was already another incomplete version, and many other stories that I could share since starting with Linux Slackware 2.0 back in 1995 (first UNIX was Xenix in 1993, followed by DG/UX).
As someone who appreciates Cinnamon’s approach, it does have a number of rough edges that make it feel less smooth and polished than it could be, compared to even e.g. Windows 7 or OS X 10.9. The practical impact is minimal, but for someone coming from outside of the Linux sphere I could see it leaving a less than stellar impression.
I've been running Mint for ... longer than I care to remember, many years, and it's been flawless for me. I see windows users around me raging when it decides to just freeze, or has weird twitchy behavior, and I just keep on going.
And no, I don't spend hours upon hours tweaking and troubleshooting until it's "just right"; Mint convinced me because it was 99% "just right" out of the box.
Compared to when I used W10 and earlier, it's been sheer joy. It just stays out of the way and lets me do what I need to do.
There were a bunch of these during the netbook boom and they were grandparent-proof. Sadly, all of my favorites (Moblin/Meego, Joli OS and xpud) are long dead.
Netbook boom happened and then people actually started trying to use them.
That is why the netbook boom died. Because while they promised to make cheap Linux devices that anybody could use none of them actually delivered on it.
In fact the whole thing was a fiasco.
The classic Linux problem can be described as "9 clicks to shit". Meaning that a Linux desktop looks, on the initial viewing, as something that is actually usable and modern, but once you start clicking around things start going bad. Going bad quickly.
A major problem cause of the problem is "linux is about choice" mentality.
Another problem is that programmers specialize at being good at programming, but things like documentation and UI testing are their own disciplines that are separate and distinct in a lot of meaningful ways. So, in attempt to make up for this, they leave configuration and details up to user choice. It is effectively pawning off the last stages of development onto the end user, who are generally most ill-equipped to make technical decisions on software they are not familiar with.
Instead of presenting something that "just works" users are presented with a myriad of choices and options that they have to make decisions about before ever actually being able to use the software.
Having lots of choices in applications, widget libraries, desktops, ways to install software, init systems, etc etc... results in very significant complexity as it all has to work together.
And complexity breeds bugs.
So each and every user experience ends up being its own unique things. There is no "standard configuration" no "supported configuration" and no "documented configuration".
The end result is that each user is forced to find ways to make the desktop work for their specific use case. Essentially finding a magic combination of things that isn't broken for what they specifically want. The desktop will still be full of buggy behavior, but just not in the way that that particular user cares about.
This creates a extraordinary of frustration and friction when time comes to doing the things that the desktop is actually intended for... which is getting actual work done.
This isn't a problem for a certain class of highly technical users that love technical minutia and configuring things. Knowing how to carve a working system out of a OS riddled with historical artifacts and highly technical choices is a badge of honor for many people.
But this isn't how most people want to do things.
Netbook boom tried to fix this in certain ways, but all that ended up happening is that each little corporation tried to do their own thing and then abandoned it when they realized there was no money to be made and they really lacked the resources and expertise to make it work.
Can't you say most of these things about Windows or even any other useful desktop OSes?
If you want a computer that does useful things and isn't just a locked down social media consumption machine, you end up with a lot of that baggage you've mentioned. Sure MacOS and Windows fare better (insert trillion dollar companies backing it spiel here...) but they are no smooth ride if you are doing anything more than consuming stuff or using a web browser.
I agree, it feels like linux is on a perpetual "under construction" flow, except that lately development has been picking up speed (which is a good thing), but results in an even "bumpier" ride.
And you can't really retreat to a more stable distro (Ubuntu LTS, numbered debian release etc) because you often get stuck with "outdated" software, missing hardware support and missing features (might be essential to your use case). Chasing the bleeding edge ends up being pretty essential in desktop linux but as the name entails it ends in lots of pain for the user. Nerds, hackers, developers and maybe even some power users can definitely have fun and thrive in that environment but the casual consumer.. yikes :(
> Another problem is that programmers specialize at being good at programming, but things like documentation and UI testing are their own disciplines that are separate and distinct in a lot of meaningful ways.
Even having decent UIs is a problem for many Linux apps and many are often deeply unpleasant to use day-to-day. This is one thing that has started to slowly get better in the last decade or so, but I can always tell what applications were designed by programmers and which ones have had at least some UI work done on them.
Also, accepting feedback from users on UI improvements often gets either ignored or de-prioritized in favor of adding new features. It's very frustrating to see an otherwise really fantastic application with a lot of neat functionality hamstrung by a bad UI.
The only desktop to have real meaningful large-ish/professional usability testing is Gnome. And that is only a couple times.
The first time was financed by Sun Microsystems in 2001 for Gnome 1.x and the result was Gnome 2.
The second time was financed by Novell around 2005 or so for their attempt to compete with Microsoft with Novell Linux Desktop. Unfortunately for them the real beneficiary of the improvements that results from their work was Canonical's Ubuntu.
Since then there have been numerous smaller/informal/ad-hoc attempts for both KDE and Gnome.
The results of all of this, of course, is that many "Linux users" believe that Gnome is the result of a conspiracy between IBM, Redhat, and maybe even Microsoft to "destroy the Linux desktop".
So that is fun.
The real success story, from what I can tell, is Blender.
They successfully revamped their user interface without a huge budget. Although it was still financed somewhat by some EU initiative, IIRC. They accomplished this by getting developers sat down next in a big room to actual 3D artists working together to produce a animated feature.
By physically placing users next to devs and having them work together, likely with a great deal of humility and openness, they managed to transform their UI into something that was actually decent.
This is probably a model that can be duplicated by other open source developers, although finding the right type of technical users not experienced in using said software willing to participate is going to always be a major challenge.
I like what EndlessOS is trying to do but it prioritizes the ability to be a complete OS that doesn't need an internet connection, whereas the article wants a browser-first OS (and assumes an internet connection). So aren't those kind of different use cases?
I think it's also the immutable nature of EndlessOS, its not something you can break as easily as a regular Linux distribution, and acts much more like ChromeOS.
I get the sentiment, but honestly anytime I try to use a modern smartphone it almost takes a short flight. Hidden UI gestures, active sides, functionality limited to 'popular' cases, etc. Totally unusable and frustrating, IMHO.
Well there was that novel CloudReady solution, which would have allowed people to repurpose old PCs/laptops and replace them with ChromiumOS, but then Google acquired them, and it became "ChromeOS Flex". [0] Not sure where it went from there. Presumably once ChromeOS gets axed (or merged into Android), Flex would go away, too.
Once the Android+ChromeOS amalgamation becomes usable on laptops, it's a matter of time before the next CloudReady pops up its head. Androidx86 already does a job running Android on PC hardware, it just lacks proper desktop behaviour. If ChromeOS' desktop environment makes it into Android, that problem is pretty much solved.
> Where Google's team put innovative effort into ChromeOS was in making it robust enough to be sold to the masses in the hundreds of millions of units, with no tech support. It's immutable, with image-based updates. It has two root partitions, one of which updates the other, so there's always a known good one to fall back to if an update should fail.
This is probably the only part of the article that is compelling. I guess having a simple, immutable Linux distro that prioritises being hard to break over many other things could indeed be a good thing for getting less tech-savvy users over to Linux.
But even then, who do they expect to maintain it? I would guess that many distros are maintained by people who are passionate about using them, and I suspect the overlap of FOSS maintainers and people who want to use such a locked-down, stripped-back system is small. And if the distro really didn't have many native apps, then users would either need to rely on Google for all their services anyway (somewhat defeating the point), or the "FOSS world" would also need to produce (and maintain, and host) a full suite of browser-based apps to rival Google's. Which is very far from easy.
The idea of "friction" comes up again and again whenever we talk about open products (whether it's FOSS or open platforms like the ones in the Fediverse, etc). People want an experience that is completely smooth and frictionless, while remaining free and open. But IMO freedom is friction. Fundamentally it means being able to choose, and not being completely reliant on megacorps, and being able to tinker and explore, and all of these things are sources of friction. Like, the author complains that there are so many distros to choose from, but how would adding another distro to the mix address that?
FOSS world needs to realize that software and user freedom are political goals and can be only achieved through political action.
Rest is accidental, yet most effort is being spent on that. That's why even though we have more FOSS code than ever, users have least amount of freedom and agency.
The fact that the customer just needs a vehicle to get to and from work, but they're offered tractor wheels and 50k RPM flywheels is exactly the problem.
But there are customers who are looking for tractor wheels and flywheels. Maybe not you or people you know, but there's definitely large industries for both. Just like there's a market for "vehicle to get to work". Same thing exists in software. Linux caters to all the uses, from low power embedded systems to huge high performance computing systems and everything in between.
I'm talking about the wheels that are analogous to the software needed for a decent consumer desktop computer.
How many window managers are there? How many different package managers? Ways to prepare code for said package managers? Scripting shells? File systems? Hell, how many distros are there based on Debian and/or Ubuntu?
If the goal is "free the world from proprietary software", then the talent spent on scratching an itch for a few thousand users is talent wasted because that isn't moving the needle fast enough. The three things driving consumer-grade (read: not enterprise or hobbyist) adoption of Linux right now are ChromeOS, gaming and the fact that Windows 10 support is ending soon. It has nothing to do with the fact that you can find just the right set of utilities for your Arch install.
But there's no money in making a consumer Linux OS. The market has shown this time and again. ChromeOS is effectively a loss leader, getting consumers and students into the Google ecosystem. MacOS is what you get if you want Apple hardware. That's about it for consumer oriented desktop/laptop *nix operating systems today.
So if there's no money in something, then the only people who will show up to work on it are going to want to scratch their own itches because it's a hobby and fun/interesting to them.
Android shouldn't be a loss leader at this point, though maybe it is in certain companies, and it's not exactly what most people think of when they think GNU/Linux.
Steam Deck is only being successful by translating Windows on top of Linux.
Valve failed to convince any major number of studios to bother about GNU/Linux, even when it would be relatively easy to port their games from Android NDK.
Providing drivers for all the hardware and peripherals that all work together in all the permutations is properly hard. I’d argue it’s the root issue.
Google can test each ChromeOS release against their Chromebooks to make sure everything works. If something fails, someone in Google can root cause on a matching hardware configuration easily.
This is not feasible for a general desktop Linux distribution because there are too many permutations. If the distribution is targeted at non-technical users the matter of root causing an issue on a hardware configuration the maintenance team can’t replicate becomes intractable.
A solution might be to literally only support Google Chromebooks, to constrain the possibility space, but even that is a big undertaking if the goal is to match ChromeOS for quality.
The basic problem of this line of thinking has always been (and always will be) that FOSS, in the end, is just a loose-knit community of people who do eachother the favor or sharing out the programs they made. There’s absolutely no reason for that community to go and try to attract a bunch of non-technical people who won’t be able to contribute back.
Who wants to deal with a bunch of users who only switched because Windows 10 stopped updating? I’m going to have to switch my stupid gaming PC away from Windows because it stops updating, and I don’t even want to deal with myself in that context, because he’s a grumpy guy that just wants to relax and play videogames after working (in Linux) all day.
Valve might be able to do it, but it won’t really be a FOSS community project, it will be a proprietary layer on top of FOSS (maybe an Open Source layer on top of FOSS, but it will still be proprietary in the sense that they’ll mostly do the development in-house and in the service of their business model). Which is totally fine, but subservient to their business needs.
I mean… the Register reports on Free Software stuff a lot. Maybe they are the entity that should do it.
A fedora atomic fork with nothing but Chrome/Firefox + Gnome or KDE would do this.
...But who would use that? If you know how to install an operating system on a laptop, you're already not the target audience for this. OEMs aren't going to ship this Linux any more than they would've shipped any others.
Chrome OS worked because it had the marketing budget of a megacorp behind it, one motivated to push people onto their cloud services to cement their monopoly.
I wonder if there were significant numbers of people who actually went out of their way to get a Chromebook over a regular laptop? I'd imagine those that did just did so due to the lower price without knowing the difference, or they had it forced on them by their school/employer.
I've been touting the chromeOS banner for years. It makes a huge amount of sense to have an OS optimized for the place most people are going to be spending 90% of their time - the browser - without any extra bloat. Then if you need more, just add those features. Gone are the days when you couldn't do certain things on a chromebook as well - assuming you have the hardware to support it, you can do almost anything you can do on a "proper" linux distro that you can on a chromebook, because you can run a full containerized distro and all the apps that come from it. I've got VSCode in my dock and can do web development easier than on a Windows desktop.
I think the problem is that the way ChromeOS was originally sold it was extremely user hostile and also made it completely impossible to "just add those features" back so in most people's mind they hear "ChromeOS" and think "oh it's even more useless than Android."
> The goal is banishing any and all questions of what to download, what desktop, what apps, packaging formats, software stores, user accounts, backing up, all of it: run the browser locally, sync the data – and only the data – to the cloud.
ChromeOS is dead simple. So much so that my otherwise computer-illiterate wife can (and does) use it regularly. There's no way that I'd get her to use debian, though, unless I set everything up for her in advance. Choosing the desktop, setting up PPAs / flatpack / whatever (if you don't want the stock firefox browser), tweaking power settings, there's all sorts of faffing about in comparison.
I'd have to set up syncing in Firefox for her (or whatever browser) but to be fair, I helped her set up her account in the chromebook too.
Your computer illiterate wife is using a browser 90% of the time. The next machine you’ll be installing ChromeOS-clone Linux for your wife, she won’t be. It’ll be just as complex to install for her.
She's using a browser all the time. Using debian out of the box is still more complicated. The setup process for debian to get it back down to the bare bones simplicity of ChromeOS is much more time consuming for me.
Hence, as my original post was meant to convey, a ChromeOS clone is a better option to set up a laptop for her than just grabbing debian.
The author's point is to use an immutable or atomic system. Debian uses a traditional package manager and over time those packages can get in a weird state. Battery runs out during package update? Third party repo disappears? GPG key changes? Distribution eliminates 32-bit library support? Nvidia drivers change? Distribution deprecates Xorg? Your everyday user doesn't know how to deal with that.
An A/B partition scheme is very simple and the system as a whole is integration tested in CI. An OSTree based atomic system is pretty close, although there is some added complexity.
Not to mention Debian out-of-the-box is a pretty miserable experience for everyday users. Their Nvidia drivers won't work, they can't figure out how to install Chrome, videos won't play due to missing codecs, etc.
> An OSTree based atomic system is pretty close, although there is some added complexity.
The author mentions this as well but there's no details there. We've been shipping chromeos-style images with universal blue for over 4 years and the ostree parts are invisible to end users. What do you feel takes away from the user experience?
Thanks for your work by the way, I'm a happy uBlue (Bazzite) user.
I think my main concern is something like a known OSTree design issue around UID/GID drift, there was a bug that was partly fixed in 2023 but the issue comes from OSTree assigning known UIDs from the deployment once created, but this may not map to the proper UID on the system the deployment is being blasted to. You can get improper ownership out of this.
Not something I've ever encountered myself, but if I were developing an embedded device it seems like one less thing to worry about.
I have 0 problems installing Debian. Half the features listed in the article and yoyr reply aren’t even used by the average user. I don’t know how you can sit there and claim Debian doesn’t work but Ubuntu does. Please share what you’re smoking.
It's all about security, especially for non-technical user.
The Chrome OS is running trusted boot and immutable partition, and completely closed host system, with no user-installable apt [0]. This means it's impossible to completely break the system - nothing that user does will escape browser sandbox. And, for defense-in-depth, even if browser sandbox got compromised, there is no sudo nor ~/.bashrc nor any other executable per-user configs, so malicious software will disappear after reboot. And if you compromise kernel and modify disk directly, there is verified boot...
And this is not just theoretical benefits, it's stops practical attacks. Are you sure that "bash | curl" command you typed was safe? Are you sure that "pip install" or "npm install" you did didn't actually install malware on your PC?
It is possible to build systems like this based on open-source Linux, but this will be quite a lot of work, and may even require separated trusted builder machine. Definitely out of reach of non-technical user.
aren't there kiosk modes for various linux distros? I wouldn't be suprised if a kiosk mode can be configured to only allow a browser and wifi config, set it up once and then image it. There you go, now you can just write the image to whatever hardware is supported by that particular flavor and it's installed.
edit: i'll add, link your image to a github repo with a readme describing how to get it on disk. there's the project.
I think steamos would be a better example, also mentioned in the article. Focus on a single application by default, immutable A/B OS filesystems, but can be opened if desired.
The goal is banishing any and all questions of what to download, what desktop, what apps, packaging formats, software stores, user accounts, backing up, all of it: run the browser locally, sync the data – and only the data – to the cloud.
That is not really popular in the open source world. Who finances the cloud, is it open, who can read or sell your data?
I think the idea pairing would be with with a very open cloud-app based solution. Something like Sandstorm where you can install your own cloud apps and have them available via the web on all your devices.
I really with Mozilla would get in this game. For most apps for most people, hosted PWAs make the most sense. But right now we're either stuck with local native software or only web apps from closed providers.
It would probably make sense to build an end-to-end encrypted sync library using S3 for storage. Then users can choose which cloud to use, or self-host.
https://yjs.dev/ seems like a useful building block.
Hopefully there would be an option to turn off cloud sync.
I would argue it goes against the desires of the open source world.
ChromeOS is a product designed to suck people further in to the Google ecosystem. All its limitations, like basically being a web browser with access to the Android Play store, are built around that. The simplicity of not having to worry about anything else is how Google sells that as a benefit of their platform to the masses.
I don't know any of my fellow nerds who want a ChromeOS kind of experience, so why would any of us dedicate our free time to making one?
"I don't know any of my fellow nerds who want a ChromeOS kind of experience, so why would any of us dedicate our free time to making one?"
Interesting perspective. Per the article, this is a method of circumventing the W10 issue coming up.. I would guess that you have family members or friends that might be caught up in that? Even ones that are not nerds?
I think project bluefin has a similarish philosophy of being a "cloud first os", would definitely recommend checking it out: https://projectbluefin.io/
In all honesty though, there's a tonne of easy to recommend distros which don't need set up if you are a "normie" just wanting to get online:
- Elementary - Mint - Ubuntu
I think the real "secret sauce" for ChomeOS' success is basically big coorporate backing and being from-a-store-buyable. I think the truth is that nobody other tham techies is likely to want to install an OS themselves.
I recently (two weeks ago) decided to finally switch from Windows to plain Ubuntu as a daily desktop. I'm well experienced with it from managing servers and have used it occasionally at a desktop in the past.
I've had - and continue to have, a ton of minor and a few large problems.
First, wifi. Apparently there's often problems with Intel killer 6e laptop cards, which required me to enter loads of console commands that I didn't understand to fix. I did finally fix it, but I'm not sure exactly what did it. A couple of hours work over a few days to fix this.
I work from an external monitor plugged into my laptops HDMI port with the lid closed, and it seems Ubuntu is massively confused by this. I'm constantly having to open the laptop to get a login screen to show, then logging in and closing it, then opening it again because now the desktop doesn't show.
My laptop screen resets to minimum brightness every restart.
I did a driver update and suddenly my external monitor wasn't recognized at all so I had to roll back.
And at least several hours more on other issues. Oh yes, if I suspend then try to wake, I get a black screen or occasionally the Ubuntu startup screen but can never get back to the login screen until I do a hard reset.
The Bluetooth audio was set to headset quality (mono) and the only way to fix this, following 20 minutes research, was to install a different audio settings app.
All of these things, I can fix. I don't want to, but as a technically savvy user I can do it, with enough hours put in. But there's no way I'd suggest anyone else in my family or friends to use it unless they clearly tell me that they would enjoy tinkering with their operating system.
> from Windows to plain Ubuntu
It is unfortunate that Ubuntu still has the mind share it does. I abandoned it a while ago as too unreliable. It is effectively Debian testing (they call it testing for a reason) with proprietary addons that have even less testing than Debian testing.
Originally, Debian making it difficult to get non open source firmware during installation was a reason for using Ubuntu, but now even that justification gone. Meanwhile Ubuntu anti-features like snap accumulate.
Ooof! That sounds rough and also makes me realise a huge oversight in my original comment.
I've been using linux on desktop for about 12 years and will always pick a laptop that's known to have good compatibility otherwise the endless hardware issues are a big timesuck.
Some distros are better than others for hardware support but I think it'd be disingenious to make out like any solve the problem.
If you have a laptop with good support, or even better, a linux specific one, it really is an experience that can work out of the box.
Maybe that's a big part the appeal of being buy-in-a-shoppable. Somebody has figured out for you what works, and you don't have to be your own guinea pig.
Hardware problems are a frequently understated barrier to switching, particularly for laptop users. Long time Linux users are generally selecting their machines with Linux compatibility in mind and so don’t see much of this, but the corpus of potential switchers haven’t, and it creates a lot of friction. It’s difficult to convince people to replace their perfectly good laptop with something else just for the sake of trying out an alternative OS.
Also, it's pretty common to install linux to revive old laptops that might have outdated hardware.
I guess that's a different baseline?
If I try out linux and it breaks my laptop, I'm mad at linux.
If windows stops supporting my laptop, or updates make it unusable, and linux makes it usable, I love linux.
It's good to know that "Linux is only free if your time is worthless" still holds true after all these years. From the way many talk, I honestly thought it might have changed.
Fedora might support your hardware better, (newer kernel) and not have snap.
Windows also sometimes has these kinds of problems. Plus I usually have to dig up many commands to turn off user-hostile defaults or free up a quarter terabyte of disk space wasted. Was an obscure diskadm cli I’ve never heard of.
The perfect OS doesn’t exist, maybe not even a good one.
This is ultimately a business model problem. There are FOSS projects out there; some of them are even listed in the article. The problem is that making it "Just Work(tm)" is hard; it's a lot of work, requiring a large number of engineers. And funny thing, engineers prefer food with their meals.
Saying "FOSS world" we need XXX is pretty useless. As a FOSS mainatinaer, the answer I always give people demanding their favorite pet feature is, "clean, maintainable patches are appreciated", ChromiumOS is free software; someone could take it use it as the basis for something like ChromeOS. But the author of the article has basically admitted that the derivitives of Chromium would quickly fail if Chromium stoped being something that they could free ride off of. Shouldn't that tell you everything about what the problem is with this picture?
Programmers do stuff for free, all the time. People also donate to foundations so some programmers can be paid to make free stuff. It's how we got the Linux desktop as we know it.
Programmers often stop worrying about the details once they're satisfied, though, and there are a lot of details. Yes we could add thumbnails to the file picker, but we also need to support Nvidia hardware and the archive application is missing a few much more interesting compression formats, plus there are a bazillion logged bugs that actually crash software rather than being a minor inconvenience.
In theory someone could set up a collective of programmers, designers, and translators to create a usable Linux desktop for the Windows 10 refugees, but it'll require a lot of effort from a lot of volunteers and a lot of people who happen to share the same principles and preferences.
ChromeOS works because there's a single entity that can hire a small group of designers to tell an army of programmers what they want everything to look and work like.
There are actually projects and companies that fork ChromeOS, but they're pretty small and once Google stops maintaining the OS they're going to collapse pretty quickly.
> ChromeOS works because there's a single entity that can hire a small group of designers to tell an army of programmers what they want everything to look and work like.
ChromeOS works because it's pre-installed on selected hardware devices where it's been configured and tested before any user even get it into their hands which would solve the majority of the problems no matter what OS you're using. Then they also take all power away from the user do make any changes to those systems which helps solve the rest. Ultimately though, most people who know anything at all about computers want the power to make changes to what's installed on them..
Everyone I know who got a chromebook did it first and foremost because they were inexpensive and every one of them at one point or another found themselves frustrated by the limitations. A privacy respecting linux system won't be able to use massive amounts of data collection to lower the price of the hardware like Google can.
People will do a small amount of work for free. But if something requires dozens or hundreds of programmers working full time, the chances of this happening is much less.
And if you want something which just works, it means you need a huge number of people doing grunt work, and that's the kind of thing that people are less likely to want to do in their free time.
SteamOS is the closest thing to making this work. Valve has somehow worked their way into being a software store that not everybody hates.
Companies should be happy that there’s a FOSS community that lets them build up on top of it for free. It is a great deal! I wonder why there isn’t an EpicOS yet.
I don't think this would work. A big part of why ChromeOS is so simple is the Google services backing it. Without the unified login, automated backups, verified app store, and hardware certification ChromeOS wouldn't be the easy to use system that it is.
FWIW, back in 2013 I was working at Deutsche Telekom in a project that was attempting to do "ChromeOS, but for Enterprise", built on top of Ubuntu and a heavily modified Firefox. We had a list of target devices to test against, but given the Ubuntu base it was pretty easy to get it running everywhere.
All the office were based on open source systems. DT wanted to make money mostly on their data plans + device management services. The project failed for one reason: Enterprise IT is the opposite of price sensitive. I had talks with a couple of CIOs who were interested in the idea, but realized that commiting to a solution that could slash their budgets by 80% would be career suicide.
Why not charge more, then some funds could be sent upstream?
Edit: answer below does not explain why price can’t be raised even higher.
Because the problem was any reduction in cost, and the status quo (Microsoft) was familiar and safe.
"Not being price sensitive" does not mean "looking always for the most expensive solution".
TIL about Endless OS: Recommended for Learners, Parents, and Schools
https://www.endlessos.org/
> Many of our technologies are distributed together within Endless OS. This is not your typical Linux distribution. We don’t use rpm, apt, or any other packaging system. In our quest for simplicity and robustness, we use a read-only root file system managed by OSTree with application bundles overlaid on top. Most desktop Linux distributions are oriented towards tech-savvy users and developers, but we have a different target user.
What are application bundles?
looks like it's just pre-selected software and flatpak. They offer a way to add whatever software you want to the list.
Linux desktop in general suffers from not having good "default" options. I can't really recommend any single distro, can't recommend even a single desktop environment as good for most people. Mint is very good, but it looks painfully dated in many places. Fedora is very good, but Gnome Shell is not for everyone in the slightest and debian base is just too wide-spread to ignore. Ubuntu is very good, but actually it's not that good in 2025 because canonical spent 15 years reinventing the wheel and not finishing any.
> I can't really recommend any single distro, can't recommend even a single desktop environment as good for most people.
I can and do all the time: KDE Plasma + Debian Testing
KDE has its own problems, though they’ve been improving in recent years. Out of the box defaults still need some work for example, and settings can still be confusing and difficult to navigate for less technical individuals. It also still has some behaviors that aren’t found on other desktops that can put off switchers.
The best DE for switchers would be one that brings as few surprises as possible and has defaults good enough that the overwhelming majority feel no need to change them. Basically, there needs to be near-perfect clones of the two big commercial desktops that don’t try to be clever or shake things up in any way so the switch is close to seamless.
Debian Testing packages may also still be too old, depending on the user. Anybody having to work cross platform or collaborate with people using other platforms is going to wind up dealing with version mismatch issues, since commercial OS users have most of their apps auto-updated to latest without Testing’a lag.
Debian Testing packages are not too old. They are mostly very nearly bleeding edge, if not they’re a month or so behind. Ridiculous you would blindly say Debian Testing is outdated.
Debian Testing KDE Plasma version: 6.3.5-1 [0]
Arch Stable KDE Plasma version: 6.4.3-1 [1]
[0] https://packages.debian.org/trixie/plasma-desktop
[1] https://archlinux.org/packages/extra/x86_64/plasma-desktop/
Based on what I’ve found it varies greatly between packages. Some are just a few weeks behind, others months or years.
Probably fine for most people, but not a good choice for low-hassle setup with "recent" nVidia cards (last ~10 years). That's almost all nVidia's fault, and I can't say any distro handles their cards flawlessly, but Debian is one of the worst: the drivers are (accurately) classified as "nonfree" and so not available by default, setup may require config file tweaking, and KDE Plasma prefers to support Wayland while nVidia prefers to support X11, leading to weird issues (more than usual, anyway).
Although gaming on Linux has come leaps and bounds, and is actually in a pretty good place with Proton and all that, it is still a hobbyist thing.
For most non-gamers (modulo some workstation users doing number crunching, but that’s pretty niche), the iGPU is the way to go.
Yeah, it's kind of niche to find someone who has an nVidia card, uses it for gaming/compute, has some reason to run Linux, and is afraid to dig into config files etc. But Microsoft's behavior of late is creating some demand here, and I too get tired of having to fix things after random kernel/driver updates.
My experience is that distros with traditional package management can't be counted on to work over the long term without manual administrator intervention.
Debian in particular would not be my first choice, or my second, for people who can't do their own system administration.
(And this is a bit of the problem: nobody can agree on this stuff)
Nobody agreeing on this stuff is freedom and beauty. If you want a managed OS there are already many options.
You look at Linux as an issue because someone you know can’t do system administration, something Debian almost requires none of. I look at Linux as an opportunity to rip people off their safe OS and introduce them to something that works but allows deep customization when they’re ready.
You’re operating out of fear, I’m operating out of faith. I really don’t care if the user “gets it”. They barely get their mobile device.
This touches on a larger problem that I think we tech people don't address as much as we should. That the modern big tech OS (Windows, MacOS, Android, iOS) has a huge advantage, which is that it's managed. Actively supported by the vendor, well after the purchase. That, and the heavy restriction of the user makes the computer orders of magnitude more stable.
With Linux this is not the way. The closest I can get is unattended-upgrades.
I think Fedora Silverblue, Bazzite, et. al. is the right direction for linux on PCs. It still feels a little immature yet, but also I'm way more confident in the system's durability than I've ever been in the past.
Based. I run the same setup myself, but "testing" does mean testing, and while it usually works fine, you have to examine apt output and make sure it's sane, before you update your system.
I don’t have problems unless I run apt-dist-upgrade before a release cut (which I never do). Otherwise things have upgraded without pain or issue packages are held back until they’re ready for testing using apt-update/upgrade
I don't think it particularly suffers from not having "good defaults", because it doesn't get adopted by word of mouth. And people are not shopping around The Best OS that they find, and sadly, they evaluate that Linux just doesn't have the juice.
Two things here.
One, OSs get adopted because of the package they arrive in. Laptops have Windows bundled, schools get computers with Windows on the cheap. Phones come with whatever they are integrated - often, not even a major upgrade is possible on them, let alone an alternative OS.
In my opinion, products like the Steam Deck are what get Linux adopted. They provide excellent functionality out of the box.
Two, PCs in this age thrive on interaction. Overwhelmingly, they are bought to interact with other PCs, either directly by network or through formats of specific software, like docx. Therefore, the functionality depends on how well they can participate in a network. This is where Microsoft really made it big with how ruthlessly they pushed Windows as an app platform and Office as the productivity software baseline.
Hence, in the end, what makes Linux really competitive today is the proliferation of web apps, because Linux has first class browsers, and the existence of Wine and its related software. Take these two away and the OS is not even talked about as an alternative to anything, just a specialty, or a curiosity.
Really, an alternative OS can do two things. It can either set the trend, which is not very likely, given that we are talking about alternatives. Or it can follow the trend, or "embrace" it, in MS lingo.
Netbooks had Linux pre-installed, and they became an OEM distribution mess, because OEMs can't stop to differentiate themselves.
Some do come with Linux preinstalled, but there is a lot of mess in this regard. "Windows tax" was one of such things, the Windows bundling went so hard[0] that it was more expensive to buy a computer without Windows than with[1]. Absolute nonsense, that more or less stopped now. But of course we are well past the Windows platform establishing itself, so now the network effect is a larger contributor I'd say.
[0] https://en.wikipedia.org/wiki/Bundling_of_Microsoft_Windows#...
[1] https://www.reddit.com/r/linuxmasterrace/comments/4ewnoi/if_...
We all know about the Windows tax and Microsoft making Windows XP free for Netbooks.
OEMs making custom distros didn't help.
If anything that was the precursor of the Android distros mess that we got afterwards.
They also did the same with CP/M, MS-DOS and Windows since forever, including UNIX clones, why wouldn't they do otherwise with GNU/Linux.
As former owner of an Asus Netbook 1215B bought with Linux pre-installed, even that way wasn't without issues.
OpenGL support was always lagging behind what hardware was capable, GL 3.3 vs GL 4.1, hardware video decoding only worked during Flash heyday, and wlan drivers always had issues with hardware when doing large downloads.
OEMs were smaller than Microsoft, that was the problem I think. Valve seems to have success with Linux, which is maybe because of their niche, and size. Smaller players just don't get to be trendsetters.
Success in a relative kind of way.
Valve has failed to convince studios already targeting POSIX like environments like Android NDK and Orbis OS (PlayStation), to port their games to GNU/Linux.
Had to switch translating Windows games, as plan B kind of solution.
It remains to be seen for how long Microsoft will tolerate Proton, or just like it happened with netbooks, will drive other handheld OEMs to use Windows and thus in a decade from now people will foundly remember Steam Deck.
I went from being a UNIX/Linux zealot during the mid 90's, writing M$ on email signatures, back to Windows as daily driver, because I got tired of all fragmentation, hunting for hardware support, that in theory would work but the actual firmware on the bought device was already another incomplete version, and many other stories that I could share since starting with Linux Slackware 2.0 back in 1995 (first UNIX was Xenix in 1993, followed by DG/UX).
How is gnome shell bad for the masses?
What does ”looks painfully dated” mean? I hope that doesn’t mean they avoiding bad ideas like hidden scrollbars or menus.
I think mint+cinnamon is fantastic, my only gripe is that packages get old after a release has been out a year or two.
edit: specifics please.
As someone who appreciates Cinnamon’s approach, it does have a number of rough edges that make it feel less smooth and polished than it could be, compared to even e.g. Windows 7 or OS X 10.9. The practical impact is minimal, but for someone coming from outside of the Linux sphere I could see it leaving a less than stellar impression.
I'm still waiting for specifics.
I've been running Mint for ... longer than I care to remember, many years, and it's been flawless for me. I see windows users around me raging when it decides to just freeze, or has weird twitchy behavior, and I just keep on going.
And no, I don't spend hours upon hours tweaking and troubleshooting until it's "just right"; Mint convinced me because it was 99% "just right" out of the box.
Compared to when I used W10 and earlier, it's been sheer joy. It just stays out of the way and lets me do what I need to do.
There were a bunch of these during the netbook boom and they were grandparent-proof. Sadly, all of my favorites (Moblin/Meego, Joli OS and xpud) are long dead.
Netbook boom happened and then people actually started trying to use them.
That is why the netbook boom died. Because while they promised to make cheap Linux devices that anybody could use none of them actually delivered on it.
In fact the whole thing was a fiasco.
The classic Linux problem can be described as "9 clicks to shit". Meaning that a Linux desktop looks, on the initial viewing, as something that is actually usable and modern, but once you start clicking around things start going bad. Going bad quickly.
A major problem cause of the problem is "linux is about choice" mentality.
Another problem is that programmers specialize at being good at programming, but things like documentation and UI testing are their own disciplines that are separate and distinct in a lot of meaningful ways. So, in attempt to make up for this, they leave configuration and details up to user choice. It is effectively pawning off the last stages of development onto the end user, who are generally most ill-equipped to make technical decisions on software they are not familiar with.
Instead of presenting something that "just works" users are presented with a myriad of choices and options that they have to make decisions about before ever actually being able to use the software.
Having lots of choices in applications, widget libraries, desktops, ways to install software, init systems, etc etc... results in very significant complexity as it all has to work together.
And complexity breeds bugs.
So each and every user experience ends up being its own unique things. There is no "standard configuration" no "supported configuration" and no "documented configuration".
The end result is that each user is forced to find ways to make the desktop work for their specific use case. Essentially finding a magic combination of things that isn't broken for what they specifically want. The desktop will still be full of buggy behavior, but just not in the way that that particular user cares about.
This creates a extraordinary of frustration and friction when time comes to doing the things that the desktop is actually intended for... which is getting actual work done.
This isn't a problem for a certain class of highly technical users that love technical minutia and configuring things. Knowing how to carve a working system out of a OS riddled with historical artifacts and highly technical choices is a badge of honor for many people.
But this isn't how most people want to do things.
Netbook boom tried to fix this in certain ways, but all that ended up happening is that each little corporation tried to do their own thing and then abandoned it when they realized there was no money to be made and they really lacked the resources and expertise to make it work.
Can't you say most of these things about Windows or even any other useful desktop OSes?
If you want a computer that does useful things and isn't just a locked down social media consumption machine, you end up with a lot of that baggage you've mentioned. Sure MacOS and Windows fare better (insert trillion dollar companies backing it spiel here...) but they are no smooth ride if you are doing anything more than consuming stuff or using a web browser.
It is especially acute when it comes to Linux desktop.
I agree, it feels like linux is on a perpetual "under construction" flow, except that lately development has been picking up speed (which is a good thing), but results in an even "bumpier" ride.
And you can't really retreat to a more stable distro (Ubuntu LTS, numbered debian release etc) because you often get stuck with "outdated" software, missing hardware support and missing features (might be essential to your use case). Chasing the bleeding edge ends up being pretty essential in desktop linux but as the name entails it ends in lots of pain for the user. Nerds, hackers, developers and maybe even some power users can definitely have fun and thrive in that environment but the casual consumer.. yikes :(
> Another problem is that programmers specialize at being good at programming, but things like documentation and UI testing are their own disciplines that are separate and distinct in a lot of meaningful ways.
Even having decent UIs is a problem for many Linux apps and many are often deeply unpleasant to use day-to-day. This is one thing that has started to slowly get better in the last decade or so, but I can always tell what applications were designed by programmers and which ones have had at least some UI work done on them.
Also, accepting feedback from users on UI improvements often gets either ignored or de-prioritized in favor of adding new features. It's very frustrating to see an otherwise really fantastic application with a lot of neat functionality hamstrung by a bad UI.
The only desktop to have real meaningful large-ish/professional usability testing is Gnome. And that is only a couple times.
The first time was financed by Sun Microsystems in 2001 for Gnome 1.x and the result was Gnome 2.
The second time was financed by Novell around 2005 or so for their attempt to compete with Microsoft with Novell Linux Desktop. Unfortunately for them the real beneficiary of the improvements that results from their work was Canonical's Ubuntu.
Since then there have been numerous smaller/informal/ad-hoc attempts for both KDE and Gnome.
The results of all of this, of course, is that many "Linux users" believe that Gnome is the result of a conspiracy between IBM, Redhat, and maybe even Microsoft to "destroy the Linux desktop".
So that is fun.
The real success story, from what I can tell, is Blender.
They successfully revamped their user interface without a huge budget. Although it was still financed somewhat by some EU initiative, IIRC. They accomplished this by getting developers sat down next in a big room to actual 3D artists working together to produce a animated feature.
By physically placing users next to devs and having them work together, likely with a great deal of humility and openness, they managed to transform their UI into something that was actually decent.
This is probably a model that can be duplicated by other open source developers, although finding the right type of technical users not experienced in using said software willing to participate is going to always be a major challenge.
I think that's just EndlessOS: https://www.endlessglobal.com/foundation/access/operating-sy...
I like what EndlessOS is trying to do but it prioritizes the ability to be a complete OS that doesn't need an internet connection, whereas the article wants a browser-first OS (and assumes an internet connection). So aren't those kind of different use cases?
I think it's also the immutable nature of EndlessOS, its not something you can break as easily as a regular Linux distribution, and acts much more like ChromeOS.
"Endless OS: A PC as Simple as a Phone"
I get the sentiment, but honestly anytime I try to use a modern smartphone it almost takes a short flight. Hidden UI gestures, active sides, functionality limited to 'popular' cases, etc. Totally unusable and frustrating, IMHO.
Well there was that novel CloudReady solution, which would have allowed people to repurpose old PCs/laptops and replace them with ChromiumOS, but then Google acquired them, and it became "ChromeOS Flex". [0] Not sure where it went from there. Presumably once ChromeOS gets axed (or merged into Android), Flex would go away, too.
[0]https://en.wikipedia.org/wiki/ChromeOS_Flex
Once the Android+ChromeOS amalgamation becomes usable on laptops, it's a matter of time before the next CloudReady pops up its head. Androidx86 already does a job running Android on PC hardware, it just lacks proper desktop behaviour. If ChromeOS' desktop environment makes it into Android, that problem is pretty much solved.
> Where Google's team put innovative effort into ChromeOS was in making it robust enough to be sold to the masses in the hundreds of millions of units, with no tech support. It's immutable, with image-based updates. It has two root partitions, one of which updates the other, so there's always a known good one to fall back to if an update should fail.
This is probably the only part of the article that is compelling. I guess having a simple, immutable Linux distro that prioritises being hard to break over many other things could indeed be a good thing for getting less tech-savvy users over to Linux.
But even then, who do they expect to maintain it? I would guess that many distros are maintained by people who are passionate about using them, and I suspect the overlap of FOSS maintainers and people who want to use such a locked-down, stripped-back system is small. And if the distro really didn't have many native apps, then users would either need to rely on Google for all their services anyway (somewhat defeating the point), or the "FOSS world" would also need to produce (and maintain, and host) a full suite of browser-based apps to rival Google's. Which is very far from easy.
The idea of "friction" comes up again and again whenever we talk about open products (whether it's FOSS or open platforms like the ones in the Fediverse, etc). People want an experience that is completely smooth and frictionless, while remaining free and open. But IMO freedom is friction. Fundamentally it means being able to choose, and not being completely reliant on megacorps, and being able to tinker and explore, and all of these things are sources of friction. Like, the author complains that there are so many distros to choose from, but how would adding another distro to the mix address that?
ChromeOS is genius in edging out Microsoft dominance in office software. Most interns only know how to use Google docs thanks to Chrome books.
FOSS world needs to realize that software and user freedom are political goals and can be only achieved through political action.
Rest is accidental, yet most effort is being spent on that. That's why even though we have more FOSS code than ever, users have least amount of freedom and agency.
We just need to stop reinventing the wheel.
The tractor wheel, the race bicycle wheel or the 50000RPM flywheel?
The fact that the customer just needs a vehicle to get to and from work, but they're offered tractor wheels and 50k RPM flywheels is exactly the problem.
Stop inventing wheels, stop offering esoteric wheels and start solving customer problems simply.
But there are customers who are looking for tractor wheels and flywheels. Maybe not you or people you know, but there's definitely large industries for both. Just like there's a market for "vehicle to get to work". Same thing exists in software. Linux caters to all the uses, from low power embedded systems to huge high performance computing systems and everything in between.
I'm talking about the wheels that are analogous to the software needed for a decent consumer desktop computer.
How many window managers are there? How many different package managers? Ways to prepare code for said package managers? Scripting shells? File systems? Hell, how many distros are there based on Debian and/or Ubuntu?
If the goal is "free the world from proprietary software", then the talent spent on scratching an itch for a few thousand users is talent wasted because that isn't moving the needle fast enough. The three things driving consumer-grade (read: not enterprise or hobbyist) adoption of Linux right now are ChromeOS, gaming and the fact that Windows 10 support is ending soon. It has nothing to do with the fact that you can find just the right set of utilities for your Arch install.
But there's no money in making a consumer Linux OS. The market has shown this time and again. ChromeOS is effectively a loss leader, getting consumers and students into the Google ecosystem. MacOS is what you get if you want Apple hardware. That's about it for consumer oriented desktop/laptop *nix operating systems today.
So if there's no money in something, then the only people who will show up to work on it are going to want to scratch their own itches because it's a hobby and fun/interesting to them.
Do we know the money stats on the Steam Deck?
Android shouldn't be a loss leader at this point, though maybe it is in certain companies, and it's not exactly what most people think of when they think GNU/Linux.
Steam Deck is only being successful by translating Windows on top of Linux.
Valve failed to convince any major number of studios to bother about GNU/Linux, even when it would be relatively easy to port their games from Android NDK.
> Nothing in this is really difficult.
Providing drivers for all the hardware and peripherals that all work together in all the permutations is properly hard. I’d argue it’s the root issue.
Google can test each ChromeOS release against their Chromebooks to make sure everything works. If something fails, someone in Google can root cause on a matching hardware configuration easily.
This is not feasible for a general desktop Linux distribution because there are too many permutations. If the distribution is targeted at non-technical users the matter of root causing an issue on a hardware configuration the maintenance team can’t replicate becomes intractable.
A solution might be to literally only support Google Chromebooks, to constrain the possibility space, but even that is a big undertaking if the goal is to match ChromeOS for quality.
There are too many Linux distributions so we need to create a new one?
Relevant: https://xkcd.com/927/
It will never happen due to the religious issues FOSS has with hardware support, that is why Firefox OS died, and only LG cares about WebOS.
This "simplicity" the author strives for is a fallacy. Simplicity means something different for everyone.
The basic problem of this line of thinking has always been (and always will be) that FOSS, in the end, is just a loose-knit community of people who do eachother the favor or sharing out the programs they made. There’s absolutely no reason for that community to go and try to attract a bunch of non-technical people who won’t be able to contribute back.
Who wants to deal with a bunch of users who only switched because Windows 10 stopped updating? I’m going to have to switch my stupid gaming PC away from Windows because it stops updating, and I don’t even want to deal with myself in that context, because he’s a grumpy guy that just wants to relax and play videogames after working (in Linux) all day.
Valve might be able to do it, but it won’t really be a FOSS community project, it will be a proprietary layer on top of FOSS (maybe an Open Source layer on top of FOSS, but it will still be proprietary in the sense that they’ll mostly do the development in-house and in the service of their business model). Which is totally fine, but subservient to their business needs.
I mean… the Register reports on Free Software stuff a lot. Maybe they are the entity that should do it.
A fedora atomic fork with nothing but Chrome/Firefox + Gnome or KDE would do this.
...But who would use that? If you know how to install an operating system on a laptop, you're already not the target audience for this. OEMs aren't going to ship this Linux any more than they would've shipped any others.
Chrome OS worked because it had the marketing budget of a megacorp behind it, one motivated to push people onto their cloud services to cement their monopoly.
I wonder if there were significant numbers of people who actually went out of their way to get a Chromebook over a regular laptop? I'd imagine those that did just did so due to the lower price without knowing the difference, or they had it forced on them by their school/employer.
I've been touting the chromeOS banner for years. It makes a huge amount of sense to have an OS optimized for the place most people are going to be spending 90% of their time - the browser - without any extra bloat. Then if you need more, just add those features. Gone are the days when you couldn't do certain things on a chromebook as well - assuming you have the hardware to support it, you can do almost anything you can do on a "proper" linux distro that you can on a chromebook, because you can run a full containerized distro and all the apps that come from it. I've got VSCode in my dock and can do web development easier than on a Windows desktop.
I think the problem is that the way ChromeOS was originally sold it was extremely user hostile and also made it completely impossible to "just add those features" back so in most people's mind they hear "ChromeOS" and think "oh it's even more useless than Android."
Uhh…doesn’t a basic Debian install already do this? If you can’t be bothered to install basic af distro, then what is a ChromeOS clome going to do?
> The goal is banishing any and all questions of what to download, what desktop, what apps, packaging formats, software stores, user accounts, backing up, all of it: run the browser locally, sync the data – and only the data – to the cloud.
ChromeOS is dead simple. So much so that my otherwise computer-illiterate wife can (and does) use it regularly. There's no way that I'd get her to use debian, though, unless I set everything up for her in advance. Choosing the desktop, setting up PPAs / flatpack / whatever (if you don't want the stock firefox browser), tweaking power settings, there's all sorts of faffing about in comparison.
I'd have to set up syncing in Firefox for her (or whatever browser) but to be fair, I helped her set up her account in the chromebook too.
Your computer illiterate wife is using a browser 90% of the time. The next machine you’ll be installing ChromeOS-clone Linux for your wife, she won’t be. It’ll be just as complex to install for her.
She's using a browser all the time. Using debian out of the box is still more complicated. The setup process for debian to get it back down to the bare bones simplicity of ChromeOS is much more time consuming for me.
Hence, as my original post was meant to convey, a ChromeOS clone is a better option to set up a laptop for her than just grabbing debian.
The author's point is to use an immutable or atomic system. Debian uses a traditional package manager and over time those packages can get in a weird state. Battery runs out during package update? Third party repo disappears? GPG key changes? Distribution eliminates 32-bit library support? Nvidia drivers change? Distribution deprecates Xorg? Your everyday user doesn't know how to deal with that.
An A/B partition scheme is very simple and the system as a whole is integration tested in CI. An OSTree based atomic system is pretty close, although there is some added complexity.
Not to mention Debian out-of-the-box is a pretty miserable experience for everyday users. Their Nvidia drivers won't work, they can't figure out how to install Chrome, videos won't play due to missing codecs, etc.
> An OSTree based atomic system is pretty close, although there is some added complexity.
The author mentions this as well but there's no details there. We've been shipping chromeos-style images with universal blue for over 4 years and the ostree parts are invisible to end users. What do you feel takes away from the user experience?
Thanks for your work by the way, I'm a happy uBlue (Bazzite) user.
I think my main concern is something like a known OSTree design issue around UID/GID drift, there was a bug that was partly fixed in 2023 but the issue comes from OSTree assigning known UIDs from the deployment once created, but this may not map to the proper UID on the system the deployment is being blasted to. You can get improper ownership out of this.
Not something I've ever encountered myself, but if I were developing an embedded device it seems like one less thing to worry about.
Interesting, I don't think I've seen this in the wild (~30 million pulls across ublue) - but I think I've seen this issue mentioned before.
I'll bring it up during the next bootc meeting[1], which are public btw! Thanks for using bazzite!
1: https://github.com/bootc-dev/bootc?tab=readme-ov-file#commun...
I have 0 problems installing Debian. Half the features listed in the article and yoyr reply aren’t even used by the average user. I don’t know how you can sit there and claim Debian doesn’t work but Ubuntu does. Please share what you’re smoking.
It's all about security, especially for non-technical user.
The Chrome OS is running trusted boot and immutable partition, and completely closed host system, with no user-installable apt [0]. This means it's impossible to completely break the system - nothing that user does will escape browser sandbox. And, for defense-in-depth, even if browser sandbox got compromised, there is no sudo nor ~/.bashrc nor any other executable per-user configs, so malicious software will disappear after reboot. And if you compromise kernel and modify disk directly, there is verified boot...
And this is not just theoretical benefits, it's stops practical attacks. Are you sure that "bash | curl" command you typed was safe? Are you sure that "pip install" or "npm install" you did didn't actually install malware on your PC?
It is possible to build systems like this based on open-source Linux, but this will be quite a lot of work, and may even require separated trusted builder machine. Definitely out of reach of non-technical user.
[0] https://www.chromium.org/chromium-os/chromiumos-design-docs/...
aren't there kiosk modes for various linux distros? I wouldn't be suprised if a kiosk mode can be configured to only allow a browser and wifi config, set it up once and then image it. There you go, now you can just write the image to whatever hardware is supported by that particular flavor and it's installed.
edit: i'll add, link your image to a github repo with a readme describing how to get it on disk. there's the project.
I think steamos would be a better example, also mentioned in the article. Focus on a single application by default, immutable A/B OS filesystems, but can be opened if desired.
Not even remotely close.
This is why it was such a revolution when Ubuntu came out and quickly dominated the Linux desktop space. When it was new.
It took Debian and made it something approachable. But even then it was a 1/3rd of what is described as needed in the article.