Oh how I wish...
Steam would turn their Proton engine-eyes to Microsoft Office. That is the one single suite that stops me moving my daily driver.
And no, LibreOffice (or the hideous online apps) doesn't cut it, before you say.
The latest edition of Valve's monthly Steam Hardware & Software Survey is out, showing a rise in Steam usage on Linux. Penguin likes to play! In isolation, the numbers aren't all that impressive. Linux usage is at 3.05 percent, up 0.37 percentage points from last month. However, it's a significant uptick compared to the …
I agree, there are things about Microsoft Office that Libre won't touch... and some that it shouldn't IMHO. But there will be a time when you quit hanging there, let go of the rope and trust that there is something below that will catch your fall. I had to say goodbye to a lot of things, like Halo (emulation works to a point but the mouse is an arse) and other games that I'll never be able to play again (... there is a discussion to be had on whether I could play them before... but hey...)
Being honest, if you're waiting for Office to run on Linux, you'll never move. There will always be regrets... but there will also be benefits... so make the jump.
I'm 3 years into switching to Linux and couldn't agree more. There's no way I would ever go back to Windows but it would make life a hell of a lot easier if Office would run locally. None of the open source office suites can compete, they're just not as polished, and online Office is so stripped of advanced features that it's not usable for my use cases.
I struggle on with OnlyOffice as I've found it the best of a bad bunch but unfortunately occasionally I've got to launch a W11 VM to get something done and I wish I didn't.
Well, with MS Office, the Wine/Proton devs face an uphill battle, since MS can (and apparently already has) bork(ed) it deliberately under Wine.
But personally, I prefer Libre even if MS Office was available - nothing pestering me to save to OneDrive (with several clicks on semi-hidden UI elements needed to avoid it), no telemetry, no AI crap, and I always hated that Ribbon anyway.
Sorry I can only give you one upvote!
I've used Open/LibreOffice for over 20 years now and had almost no problems with it - less issues than with MS Office 2007 onwards.
Most of my customers don't even realise that I'm using LO as the reports get issued as PDF!
I have 3 remaining customers who use MS Office where I need to issue .docx or .xlsx files and hence I retain a copy of Office 2010 for those times when I need to use MS Office.
LibreOffice can render, edit and save .docx and .xlsx files (unlike office, to spec, but you'll want MOX mode rather than to spec) and I suspect the output files would render more reliably than the output of office 2010 in later versions of office.
For all the times I've edited .xlsx files - nobody has ever noticed that I'm not using office.
That rings a bell from the early days when Lotus 1-2-3 was still the front runner in spreadsheets. Transferring data between 1-2-3 and Excel was a nightmare as the file produced through the export/import function kept breaking. Every time Lotus would supply a fix, MS would supply another break
If Microsoft Office is the only thing keeping you on Linux... Why haven't you changed your workflow to something that doesn't use MS Office?
There are very (VERY) few industries where "I have to use MS Office or fail" is true, and these days, I'm not at all sure what those industries are. Perhaps someone in your industry might enlighten me?.
> Steam would turn their Proton engine-eyes to Microsoft Office.
Have you tried it? I have read some reports that some parts work.
I don't use it much but I keep MS Word around on my Linux laptops. I like Word 97, which is blisteringly fast on even a 15-year-old laptop, but with slightly more work, Word 2003 works fine as well. I don't want any later version as I loathe the ribbon interface, but in Word 2003, you can install all the service releases, and then install the Office 2007 file converters, and then it happily loads and saves .DOCX files just fine.
I have no need for Excel. I like Excel but LibreOffice Calc does all I need. Ditto PowerPoint, except that I don't like it -- but LibreOffice Impress is fine for me. I have no use for Outlook, OneDrive, Sharepoint, any of that junk. I have successfully connected to OneDrive/Sharepoint volumes with `rclone` and `onedrive` though. In a prior role I ran Linux but had a corporate OneDrive account with 2TB of storage, so I wanted to configure my machine to automatically backup to my OneDrive. Never got that working, but manual file copies were fine.
So, yes, some bits, the standalone apps, can and do work... but you might have to settle for older versions, as I am confident that despite is protestations, MS is engaged in a continuous sabotage campaign to keep Office 365 a moving target that WINE can't hit.
Since about WINE 8 I have not needed it, but before that, CrossOver Office made Word 2003 run when bare unaided WINE couldn't.
Failing that, at some point soon, I plan to test and write up WinApps.
https://github.com/Fmstrat/winapps
WinApps is great but has issues. Things like (and this is a few months since I last tested it) that you only get the icon of the last application opened and file type association is a mixed bag.
WinBoat works quite well, but also has the same kind of issues (I believe it's a fork of WinApps, so unsurprising).
I actually like OnlyOffice - it has superb compatibility but the very obscure Russian ties are a bit objectionable.
I must admit I have not tried Office under Proton - I couldn't see any way of making it work via it. I will revisit.
I have almost completed transition from Kubuntu to Debian KDE. SNAP issues history. KVM/QEMU/Libvirt gives us Debian Sid VM for the more exciting stuff
Debian delivers stability and the one of the edgiest of rolling distros on one desktop. Result.
Old school linux distributions still exist. Slackware 15.0 is very stable and can be configured with Pipewire or just alsa as well as the default pulseaudio. Some find applications a tad dated and run Slackware current (the in-development tree). Either way you will have to take charge of configuration (I tend to use mostly out of the box defaults with some slackbuilds).
PClinuxOS has a rolling release model but curated (and, it has to be said, opinionated ), so more recent packages and probably won't break on you.
And for something completely different, there is always OpenBSD.
Slackware isn't for everyone tho. For example, it's kernel is old so it's support for newer hardware made within the last year is non-existant (and the version of MESA is equally old that you'll have a hell of a time getting the newest GPUs supported). The older version of Libreoffice is almost certainly not going to include the recent improvements that improves Word document formatting tremendously. And god help you if you want Wine to be a version that was made this year.
True, people with recent hardware might have to look at Slackware Current (Kernel 6.12.57 mesa 25.2.6). For some of us with museum quality hardware, the older mesa in Slackware 15 is actually an advantage.
I'm a bit confused about your LibreOffice concern. Alien Bob's package is at 25.8.2 for both Slackware 15 and Slackware Current. The slackbuild.org build scripts point to 25.8.2, both the actual compile script (think 12 hours on fast hardware) and the script that repackages the LibreOffice rpms.
As well as a host of new gaming oriented gaming distros (CatchOS, Bazzite etc) that have really caught on the the fragging set. Valve's continued investment in Proton is a thing of beauty and a number of CPU intensive games run better than under Windows. Impressive.
So much so that people are installing Bazzite on the Asus "Xbox" Ally to get some performance out of the handheld.
Say it with me: Windows is the problem with Windows handhelds
Bazzite fixes the Xbox Ally and shames Windows yet again.
To be fair - "just because a game launches doesn't mean it runs well enough to play" is equally applicable on a Windows box! I've spent many many hours trying to get rid of glitches in Windows games on running natively on Windows, if I ever have to spend 20 minutes googling and tweaking a Windows game to run on Linux, I'll give it a pass.
I've exclusively played games on Linux since 2020 and my experience is that nine times out of ten I actually have more of a fight with the bloatware launchers that big companies bundle that I do with the actual game.
I'd also say I have more luck with games than productivity applications. Because they want their own look and feel, they have their own in game menu systems which are a direct translation. If I have a glitchy application in Wine, it's usually the Linux desktop window dressing not quite matching up with where it would be in Windows, leading to blank or unclickable buttons or Windows bouncing off into the ether (Adobe, I'm looking at you).
My experience matches yours. If something has a launcher, there's a better than even chance I have to do some stupid tweaking to get it to work. If it's got a dumb implementation of 'anti-cheat' (which is almost all of them) then it detects "this isn't Windows, fail on purpose" which is the current most likely reason for any game to fail to launch on Linux.
...is that a lot of YouTube tech people are trying Linux for gaming with generally OK results. The issue is mainly anti-cheat which cannot be used under Linux, which is stopping the AAA games running.
However as more and more smaller games are getting played, along with old titles that refuse to die, it could slowly increase.
Im at some point going to switch fully to Linux as the one thing holding me back, the Affinity suite, the new Afffitiy seems to be able to run with some tweaking.
I think I read somewhere that after the Crowd Strike fail Microsoft is going to work to block these things from running at the Kernel level they do.
That means these anti-cheats may have to retool, maybe they can then work on Linux.
EA's Javelin essentially stopped me from buying Battlefield 6 as I'm on Bazzite these days.
So, developers will have a choice of safe, secure user space APIs, or dangerous insecure but probably slightly more powerful kernel APIs. Even if they do have the same power, convincing existing software makers to switch from an API that works, to one they havent tried won't be easy.
I can't see them moving away from what they know.
As far as I'm aware, the only anti-cheat system that don't work on Linux are the kernel level ones, and to be honest, not sure I'd wan't to be installing those on Windows anyway! (Not that I use Windows for home use anymore anyway).
Granted this does mean quite a few of the bigger multiplayer games won't work on Linux, unless the devs change how this works, something that they might have to do if Microsoft go through with their supposed intention to remove kernel level access.
The big problem with this is that multiplayer games suck when people cheat. Impossible shots, impossible to shoot at. Takes the fun out of any game when cheaters join in.
I don't like software running on kernel level but it is better than letting the few rotten apples ruin games for the rest.
What is your solution to multiplayer games if anticheat is removed?
I never said remove anticheat, the thing is you don't need kernel level access for anticheat, this is a choice the devs for specific games have made. There are plenty of multiplayer games out there with anticheat that don't touch the kernel.
These devs might not have much choice in the future anyway, as Microsoft have stated they plan to remove kernel level access, thanks to issues like CrowdStrike.
The likely best long term way to identify cheaters is to monitor behaviour from the server side, rather than trying to detect cheat software on the client side.
Except that monitoring from the server side doesn't work either. EA actually tried that and their anticheat banned great players more often than cheaters.
https://www.youtube.com/watch?v=GBKRdsHBef0&t=216s
I'm not speaking for KLAC, just saying that server-side isn't perfect either.
I think the only thing that works is good old fashioned game logging and community reports. But then that will take EA hundreds of man hours and cost them a few thousand dollars a month to hire someone to go through replays of reported players. So of course they'd rather have some tentacle-laden monster fondle your kernel instead of making their shareholder see that they're not making as much money as they'd like...
You're assuming that anti-cheat software that blocks Linux is doing anything useful. Valve's in-house anti-cheat is designed to run on Linux, with or without Wine, and nothing really stops companies from using that instead of rolling their own, or going to another third-party that doesn't understand how to write good Linux anti-cheat software.
The other is, "don't make the game free to play, and ban the cheaters when detected so they have to buy another copy of the game to grief again." The biggest problem with cheaters is in free games that cost cheaters nothing other than "create new email address and/or game account" to cheat in.
Cheaters get a lot of press, but the actual percentage isn't that high in "pay for account" games.
is an increase in coverage of gaming on Linux (excluding Steam Deck) on popular YouTube (and other sites) tech and gaming channels, that previously only ever covered gaming on Windows desktops (or laptops).
No big shift over yet, but the fact that it's being talked about at all on these quite large channels (millions of subscribers) is a big change from just a year ago, plus the comment sections are full of people who've either already switched to Linux, or people curious about giving it a go. Some channels like Gamers Nexus (2.5m followers) are talking about adding in Linux benchmarking.
A common theme in discussions seems to be many people just hating the direction Windows is going, and looking for a alternate path.
Also likely helped that it hit mainstream tech news sites (such as Toms Hardware) recently that someone on YouTube got hold of a ROG Xbox Ally (an official XBox, aka Microsoft, handheld gaming system that runs a slimmed down Windows 11), and stuck Linux (Bazzite) on it, and got Kingdom Come: Deliverance 2 to run at 32% better FPS than Windows. Most other games didn't get the same boost, but had apparently a more consistent FPS (smoother game play) in Linux than under Windows. Also sleep mode (enter and exit) was basically instant under Linux, but took around 40 seconds to sleep and 15 seconds to resume in Windows.
> If this keeps up it will be 10% in seven years.
Only if there is no change in rate.
The EoL of Win10 pissed off a lot of people. If they find that Linux is actually fine now, most of their favourite games work, and their computer runs faster, then news will spread.
It's tricky, predicting the future, but the point here is that a critical point may have been reached: a shite new version of Windows, an old one intentionally killed after an aggressive campaign of free upgrades and promises it was the last ever release that would be updated forever... and now, the compatibility is there.
We don't know for sure yet but this could be the tipping point. We know that multiple vendors are selling Linux boxes primarily intended to run Windows apps -- AAA games. That's a big deal.
Chuckle. Who knows, I suspect it's difficult to predict currently until we start to see some longer term trends.
The gamingonlinux site has a tracker for Linux market share (based on the Steam hardware survey), so shows the trend over time. : https://www.gamingonlinux.com/steam-tracker/
Basically it hit 1% around mid 2021
2% around the start of 2024,
Now 3% in Sep 2025
So:
~2.5 years to go from 1% to 2%
~1.8 years to go from 2% to 3%
There seems to have been a marked uptick in just the last 6 months, with around 2 thirds of that increase to 3% being in those last 6 months.
If the last 6 months trend continues, we could see 10% in less than 5 years, but this could just be a short term trend due to Win 10 EOL, and things might settle down again.
Oddly, if you look at English only Steam users (last chart on the above page), the % for the Linux share jumps to 6.61%, seems Linux is way more popular with English speaking users for some reason!
Looking at some absolute numbers, based on Valve reporting they had 132 million active monthly users back in 2022 (very likely higher now), that means an estimated 4 million plus active Linux users each month (based on the 3% number).
> Linux is way more popular with English speaking users
I think this is one of the things where Linux and FOSS are more than one thing to more than one group of users.
For one group, FOSS is easier to translate. So you can get it in your language, and it's useful to people who aren't very well-educated and well-travelled and who don't speak multiple languages.
(I am not being dismissive here. I am absolutely horrible in 6 languages other than English, and I am shameless enough to torture natives in their own language, even if they speak mine. This amazes lots of people but I'm terrible -- I am just spread thin.) I deeply and profoundly respect proper true polyglots who are decently fluent. I can just about hold an arbitrary conversation or read comic books in French, but nothing else really.)
FOSS wins here because all the code is right there and you can grep it for plaintext, and even a non-programmer can contribute by translating text... and then you get the famed "stone soup" effect, where it's useful to a monoglot even when only _some_ is translated, then each time someone comes along and does a bit more, it gets better and better and better. There's no single "enough" or "ready" point. It's incremental, in a way few companies would be willing to release, put their name on and ship.
But the other position is that you need a bit of a tech clue to find Linux and FOSS and use it at all... and people with that sort of initiative are more likely to speak enough English to be able to use a computer in English for the much greater range of software and docs and news and how-tos you can get.
So it appeals to people at both ends of the bell curve, while the commercial stuff with pro translations sits in the middle as the tool of the oppressed masses, enslaved by commercial software. ;-)
It doesn't get talked about much but there is another place this is true: China and Chinese. Tons of software that is not localised in anything but Chinese and probably never will be.
Much as was true in Japan 30-40 years ago -- entire computer hardware families never seen in the West, entire OS families, from FM Towns and Sharp X68 tower boxes, to the several weird x86 families -- and the amazing TRON family of OSes, including BTRON and iTRON.
Once mainstream Western OSes could handle kanji and the kana, the Japanese-only hardware faded away. Japan could afford Western kit.
My impression is that China ran on old pirated versions of Windows until quite recently, but the government of the PROC is leaning heavily on people and organisations to switch to FOSS... led, of course, by Android.
When it comes to translations, that is generally enabled by a library called GNU gettext, which is free software - not "FOSS".
If the programmer goes and includes the gettext init and header and does _("string to translate") for every string that is beneficial to translate, that does indeed allow even non-programmers to translate the software.
The .po format is pretty good in that all you need is a text editor to finish off a translation - you can see; msgid "Apply" msgstr "Apply" and go translate it by changing msgstr to;
msgstr "適用"
For many programs - all that needs to be translated is the help output and the manual, as that allows the user to understand what say cp and --dereference does and use it, despite not knowing any English.
There is nothing wrong with commercial software - what is wrong and what does the enslaving is proprietary software.
Several years ago, microsoft got the Chinese by making many of such unauthorized copies of windows cease to function (Billy let the unauthorized copies happen, as he wanted the Chinese to use and become dependent on his - rather than having factories use say LiGNUx instead), but as far as I can tell, there hasn't really been that much action taken to fix the problem permanently.
Android is just as proprietary as windows, but I guess how you can make custom Android builds that partially does what you want, despite the extreme difficulty, is practically much better than windows (which does not allow custom builds at all).
There was so much misleading info and disinformation in this I felt I had to make some corrections.
> which is free software - not "FOSS".
FOSS: Free and Open Source Software.
FS is a subset of FOSS. It is generally not productive or helpful to quibble about that. Saying "FS != OS" all right, but that's not what you did.
> msgid "Apply" msgstr "Apply" and go translate it by changing msgstr to;
> msgstr "適用"
Which requires Unicode support and appropriate fonts for a start. Then it additional requires a way to _type_ 適用.
You misrepresent as simple something which is not.
> Android is just as proprietary as windows
This is not true.
Android is Linux. It has a different libc but Bionic too is FOSS. The runtimes are FOSS. AOSP is FOSS. There are multiple rebuilds of it.
Here are the Git repos:
https://android.googlesource.com/
Here's Bionic for example:
https://android.googlesource.com/platform/bionic/
Now, a lot of _apps_ don't work without Google Play and Google Apps, but that is NOT THE SAME THING and trying to represent one as the other is much MUCH more misleading than saying "X is Free Software not FOSS."
Behave. That includes not telling lies. If you don't know the difference, if you don't know if something is true, then go check first.
>Behave. That includes not telling lies.
It's quite sad that people accuse me of telling lies when do I extensive research and check before making any claim and try to tell the truth to the best of my ability.
Let us compare Android and Windows;
- Windows doesn't really restrict your ability to install and run software on windows (if it doesn't have a signature, it will pop up a nasty warning, but the software will still run (windows tablet did, but that is discontinued)), while Android refuses to install software unless it has a signature (it's a unreasonable task to re-sign an apk - for example after merely editing the apk config flag to add the external storage flag - you need to work out how to sign it and of course the official signing program is proprietary software) and it is scheduled for future Android versions to disallow the installation of software without google's signature (it is unclear as to how long workarounds like `adb install` will work).
- Windows is fractionally source-available to approved people via the "shared source initiative", while Android is partially source-available to approved parties are not forbidden from receiving the source code (there are far more proprietary things than google play services and google apps) - with google providing non-corresponding source code for many parts under a free license.
- The windows "sdk" is under a proprietary license, while the Android "sdk" is under a proprietary license.
- Windows uses a proprietary kernel (the NT kernel), while Android uses a proprietary kernel (Linux).
- There are unauthorized custom modified versions of windows, while there are also unauthorized custom builds of Android ("AOSP" lacks everything needed to actually get a device working - the developer needs to include the missing proprietary software in an unauthorized manner).
As you can see, Windows and Android are uniquely proprietary in different ways - which makes Android just as proprietary as Windows in practice, even though far more non-corresponding source code of Android is available.
The explanation below is only partially sourced (I didn't want to build a wall of text too big) - if anyone needs any further details, just ask.
>The runtimes are FOSS.
To develop anything for the "runtimes" of Android, you'll need at least the Android SDK, NDK and platform-tools, which are proprietary software under a proprietary license (the same license applies whether bundled with studio or not); https://developer.android.com/studio/terms https://developer.android.com/ndk/downloads https://developer.android.com/tools/releases/platform-tools
That license contains many proprietary restrictions, but restrictions of exceptionable notability are;
"You may not use this SDK to develop applications for other platforms (including non-compatible implementations of Android) or to develop another SDK."
"You may not use the SDK for any purpose not expressly permitted by the License Agreement. Except to the extent required by applicable third party licenses, you may not copy (except for backup purposes), modify, adapt, redistribute, decompile, reverse engineer, disassemble, or create derivative works of the SDK or any part of the SDK." (applicable third party licenses is software not from google).
google claims to provide source code of the SDK separately (I have not checked if that is in fact complete source code - I have read a claim that it contains binaries not in source form) and there is a rebuild project; https://codeberg.org/Starfish/SDK-Rebuilds - but what google provides is not corresponding source code, as a lot of Android software that is developed with google's SDK (that does not use google play services) will not compile against the rebuild SDK.
fdroid for that reason does not use the rebuild SDK to build software - instead they use the proprietary google SDK; https://f-droid.org/docs/Installing_the_Server_and_Repo_Tools/#proprietary-non-free-libraries
>AOSP is FOSS.
"AOSP" completely lacks the most important software required to get any device working.
In most cases, or most devices, to get a working device, proprietary software like (but not limited to) proprietary drivers need to be extracted out of vendors proprietary versions of Android and then unauthorizely distributed as part of the distribution.
Google in the past provided downloads of the recent required proprietary software for their devices (starting from Android 16 they don't), but in fact distribution of such software was and is forbidden; "These files are for use only on your personal devices and may not be redistributed"; https://developers.google.com/android/drivers).
>FOSS: Free and Open Source Software.
Most people who read that assume it means gratis, source-available software.
The only way to resolve that confusion would be to describe what free software is (takes <30 seconds - "By free I mean freedom, not gratis - free software is software that respects the users freedom to run, study, modify and/or share unmodified or modified versions of the software. Everyone has these freedoms with food recipes and should have them with software too. Free software can be either gratis, or cost money".) and then describe what is "open source" (10+ minutes - as you need to go over the 10 requirements of the osd, the development model and open source intelligence - 90% of people are going to stop listening and all they'll remember is "open source" == source-available) and then you need to come up with something as to why free software and "open source" should be grouped together - which is a disaster.
>Which requires Unicode support and appropriate fonts for a start. Then it additional requires a way to _type_ 適用.
To edit a .po file, all you need is Emacs and unifont (although you may want to install mozc or anthy, rather than having to type C-x 8 RET 9069 C-x 8 RET 7528).
>Android is Linux. >It has a different libc
Android is far more than just the kernel, Linux and binonic and an OS is far more than a kernel and a libc.
Linux isn't fully source-available, despite being the poster child for "open source".
It contains completely undocumented init sequences without a single comment, object code without source code disguised as arrays of numbers and the project also develops against and maintains proprietary software without source code (it's stored in another git repo for convenience reasons (massive binary blobs mess up the git history), but the name is "linux-firmware.git" - not "firmware.git" (i.e. for any kernel) for some reason - which means despite the "open source" development model, Linux is proprietary software
Android does not clean up the proprietary software out of Linux when it uses it and proprietary Linux drivers are extremely common for Android.
>There are multiple rebuilds of it.
CyanogenMod, GrapheneOS and LineageOS are all proprietary software.
There is Replicant - but questionably it refuses to install an apk without a signature and you cannot permanently disable system services (if you extract the boot partition, disable an unwanted service that will not break the device and repack it with the exact settings the device will fail to boot).
I suspect many non-English Linux users work in IT or have an interest in it and who install the OS in English because they can speak it, English is still the language of IT, and they'd rather not put up with a patchwork of software which is in English only or half or poorly translated into their native language, in spite of there also being software which has been properly translated into their native language (Firefox, LibreOffice...).
It would have been better to break it down by country, if that data is available.
> English is still the language of IT
In the West, and I suspect _only_ in the West.
English is not the world's only _lingua franca_. The Roman alphabet isn't either.
I formerly lived in a part of the world where the Cyrillic alphabet is also common. Cyrillic is not "the Russian alphabet" you know. Bulgarian, Serbian, Ukrainian, etc. are not Russian but they're written in forms of Cyrillic and -- news flash -- _different_ forms of Cyrillic. Just as English doesn't have a letter š or a letter č or ć or even a way to make the difference between č (church -> čurč) and ć (where I don't think an example is possible).
400 million odd people understand a bit of Russian and live in countries where it's a common 2nd language. One hated in recent decades but even so.
There's a dialect of Chinese written in Cyrillic.
https://en.wikipedia.org/wiki/Dungan_language
Japanese, Korean, Vietnamese and lots of other used to be written in Chinese characters.
About 200 million speak Mandarin Chinese as a second language.
Remember that the view from here may not reach everywhere and there are other places where the view is very different.
If we're talking stats, I don't see Linux desktop use by country disproportionately high in Canada, US, UK, Oz, NZ compared to the worldwide value but I do see it high in India at 6% which is double the worldwide average.
If anecdotal evidence counts, I've heard there a lot of sites which are full of English IT book PDFs on websites which aren't English speaking countries...
I haven't seen it mentioned in the comments so just letting people know..
If you're thinking of gaming on Linux take a look at the ProtonDB website.
If the game(s) you play are ranked as platinum or gold , you'll generally be fine out the box.
When I switched to Linux 2 years ago I used ProtonDB to see which of my games would come across with me too Linux.
Amazingly, 38 out of 41 in my Steam library!
Obviously your mileage may vary depending on your game choices. As others have already mentioned, quite a few of the triple AAAs have anticheat that makes life more tricky.
It's also surprising how fast games get ranked on ProtonDB nowadays. It used to be a couple of months following release before you got an idea of whether the game would be ok on Linux, now it's usually a day or two.
Anno 117 has already got Platinum status, and that only gets released on 13 Nov.
I just checked my own library, I've been on Steam since it launched (over 2 decades ago!) and have 100s of games (some AAA games, I also buy lots of smaller indie titles, and often wait for sales, especially for the AAA titles, let them fix bugs and performance issues before I buy!).
79% are Platinum or Gold, so run out-of-the-box (Platinum just works, Gold usually needs a small tweak [*] and then runs perfect afterwards)
13% are Silver, so might have minor glitches, but otherwise still works fine.
2% are Bronze, so typically have issues, such as crashing, and might need more tweaking to get to run stable.
1% Borked.
The remaining ~5% are games with no reports (typically very old games no one plays anymore).
Note, I do not play many competitive multiplayer games (War Thunder being one exception, and that's native Linux). So nothing in my library has the kernel level anti-cheat.
Of the borked games, none of them were current, or even vaguely recent. 2 of them were very early VR experiences (e.g. NVIDIA VR Funhouse). 2 were multiplayer games from years back, where the companies have shut down and the servers are long since gone (Marvel Heroes Omega for example), another 2 are de-listed titles (e.g. Age of Empires Online) and 1 was an open beta which ended back in 2015! The only Borked slightly more recent title is a VR video player called 'GizmoVR Video Player', which is free, and seems to have been abandoned about 5 years ago, although seems to still work in Windows.
Of the no report ones, most of those seem to be things like discontinued expansions/DLC, such as 'Sword of the Stars: Argos Naval Yard Expansion', as this was rolled into a Complete Collection version years ago.
* Note, any needed tweaks are documented in ProtonDB for that game, so generally no searching forums etc needed, and typically require either some command line parameters adding (in Steam for that specific game), or might need a specific run time installing, which is typically managed via protontricks.
Hmmm my brother was a die hard Windows user but Win 11 pushed him over the edge. I recommended Linux (Febian) so he tried it but unfortunately it did not go well.
He had TRIM NCQ problems with a Samsung 870 SSD drive. He researched the issues, and concluded from all the articles that Linux and SSD don't play well. Most solutions seem to be disabling NCQ or blacklisting at the kernel level with reduced performance.
He gave up and concluded that Linux is not ready yet and went back to Windows.
I've been running Mint off an SSD for years without trouble (and gaming with Steam too!), and I've never run into any problems. Based on your comment, I did some googling, and as far as I can see, CNQ TRIM problems with Samsung drives were fixed in kernel 4.2 - from 2016. I haven't heard of the "Febian" distro, but I'd expect any modern one to have a 6.x kernel, and work fine.
I expect they meant Debian.
I haven't heard of any "CNQ TRIM" problems. Glass houses, etc.
Looking at the latest kernel source libata-core.c
"solves" the problem by blacklisting devices. That is turning off NCQ, which dramatically reduces performance of the SSD. I guess this is not something most end users notice if they are unlucky enough to have one of the blacklisted drives (and combination of SATA controller).
I don't believe your brother's experience is typical.
I have three devices in my house running different Linux flavours.
All three run either M.2 or 2.5" SSD or a combo of both.
One of mine is actually running a Samsung 870 Evo.
None of them have ever had any issues with the drives.
A lot of my team I work with in my org (15 of us) have also been bitten by the W11-is-the-last-straw bug and hopped over to Linux.
We have a Linux teams chat dedicated to helping one another answer questions etc.
SSD issues have not cropped up once since we started the thread 16 months ago.
Sounds to me like your brother was unlucky.
The issue seems to be lack of interest in delivering a non-defective product on Samsung's part, as Linux does implement SATA and NCQ and TRIM to the specification; https://bugzilla.kernel.org/show_bug.cgi?id=203475
I wonder what Debian's doing, as looking at GNU Linux-libre 5.4.260 as far back as November 2023, the issue is fixed by disabling NCQ TRIM in drivers/ata/libata-core.c (NCQ itself is not disabled unless you have an old ATI SATA controller which requires disabling NCQ entirely).
With no NCQ TRIM, but NCQ enabled, as long as you schedule a periodic trim with cron or whatever (usually already done via a systemd service on systemd distros), there will be no meaningful performance decrease.
Window's kernel seems to handle defective SSD's far worse; https://wccftech.com/windows-11-latest-update-is-reportedly-causing-widespread-ssd-failures/
I've had no issues with SSDs on LiGNUx - even when I forget to setup TRIM, there isn't a slowdown until months later and you just run fstrim.
It takes some skill compile things cross platform between windows and GNU if the used IDE doesn't have a button that automatically does it - which most game devs seem to lack.
Considering that such games are almost always proprietary software, it's arguably a feature for the games not to run.