First: what does this question mean? What is a good or best Linux? It means an accomplished OS: a Linux operating system (OS) that is good (or even best) as operated by the end-user . This clarifies a question that is often set aside: namely, there is already a standard set by the market and by the users on what a good OS is. Discussing Linux comes down to comparing it to this standard. This standard is set by the operating systems on the market: Windows, Mac and the mobile Linux-based Android. Linux OS-es for desktops have yet to set the trend here; and the fact that the trend is already set for them makes it easier to answer the question above.
If Linux is or not “better” in itself (as a kernel and possible developments) compared to Windows and Mac is a different discussion. Here the meaning of a “best” operating system is: a system that is best as operated by the end-user, a user considered as being the same for all operating systems.
I will not discuss here in depth the shortcomings of Mac or Windows which might make Linux be considered better. The problem is: what do they have that Linux lacks as far as the end-user experience is concerned? What is the standard that these two (and lately the mobile Android) have set and that Linux (for desktops) should follow? It is a quality that is highlighted even by the shortcomings of these rivals: first, they are intended as finished, well-rounded products, ready for use. (1) (An end-user should not have to bother about plugins that don’t properly show and switch keyboard layouts, about the wifi being reset, about difficulties with printer drivers, about missing icons and especially not about compiling program packages.) Second, the present standard concerning the general aesthetic and design quality of an OS (2) is pretty clear, as everybody has seen a Android or iOS phone, a Windows or Mac computer.
These two points should be enough to set the standard for the question about the best Linux OS. (It is beside the point the fact that these qualities are more easily and necessarily imposed by the fact that big companies are behind the rival OS-es. Big companies or open-source, there is a standard of quality for the general end-user that has to be respected by the “best” Linux, and this is what the question is about.)
The question will not be therefore answered by comparing Linuxes between themselves but by comparing end-products to the standard defined above.
On the other hand, it is necessary to see what are the main possible differences between Linux operating systems for desktops and at what levels they are defined in order to understand better the situation:
the kernel: there are different kernel versions
(they are named with numbers: 3.13, 4.2, etc)
the “type” of Linux, or the Linux base:
the compiling, the packaging, the repositories, the maintainers can
be different, ending up as a Linux base for a Linux OS
(these bases are Debian,
Fedora, Arch, and not too many others)
the desktop environment (DE): there
are different graphical user interfaces that a Linux OS may have
(Gnome 3, KDE/Plasma, Cinnamon, Xfce, etc.) and the quality of the
integration between GUI and kernel/base may be different.
The term OS is used at different levels. Debian has its own end-user version, that is, a Debian-packaged and maintained Linux with a GUI/DE on top, which can vary (you can have the Gnome desktop version of Debian or the Xfce desktop version.) While this is called “Debian OS”, there is also the Debian base for other operating systems, like Ubuntu. But Ubuntu also has this kind of multiple-level meaning. Ubuntu is based on Debian, a Debian without a GUI and without its repositories, but with Ubuntu repositories and maintainers. This “new” Linux is also branded Ubuntu as an end-user OS but also as a base for other end-user OS-es, some of which can be called Ubuntu “flavors” (Xubuntu, Kubuntu, Ubuntu Mate, as they stay close to Ubuntu while using a different DE than the Ubuntu flagship, which uses the Unity DE), and some of which become different OS-es (Linux Mint, elementary OS, etc.) within different Linux OS projects.
It all comes down to the different levels of open-sourcing and branding and to the different maintainers and/or companies behind a Linux project. This term is arguably the most useful to understand the Linux world and the possibilities each Linux OS has of becoming “the best”.
A Linux project may be just packaging and maintaining of a Linux base plus various options regarding desktop environments and optional packages, ending up as a finished Linux OS to a lesser degree than others, and therefore not having a real chance of meeting the standard set above. Others may go from the level of packaging, maintaining and setting repositories down to the minutest details of the end-user level. This end-user level is the one that counts in this discussion, and the fact that the end-user OS was developed from scratch inside the same Linux project or not may be more or less important.
As the finished product is all that counts here, the difference between the goals of the various Linux projects are very important. That is, the main goal of some Linux projects is not to provide a finished product, as much as to follow certain guidelines or a certain philosophy. (The ‘Linux philosophy’ may be in some cases what stops a Linux OS from becoming a good OS.)
Therefore, in order to identify the “best” Linux as a “finished Linux OS for the end user”, I need to look for Linux projects that aim at providing a such OS.
Debian and Arch do not have this aim.
Debian is focused on setting a draconian open-source standard that makes the end-product unfit for the end-user. The user for Debian is not as much an end-user as an ‘end-follower’, the follower of the open-source philosophy, which can be also a developer (of open-source products). (Debian open-source limitations will necessarily incite a user to develop its own products or to look for them, if he/she wants to follow Debian philosophy, or, otherwise, to install non-open-source products or to integrate Debian base into a different Linux project, like in Ubuntu or Linux Mint. In all these cases the user is not really an ‘end-user’.)
On the opposite side, Arch gives access to the largest Linux database of software products but has the aim of giving the user total freedom and responsibility of building the operating system. (An arch-user is not an end-user, but rather a ‘starting-user’.)
This being said, let’s take a closer look at some Linux projects that might be providing what I am looking for here.
First, what is my personal experience with various DE/OS:
I have tried Xfce in Xubuntu and Linux Mint; also in Manjaro. Manjaro was an opportunity to try a super-light desktop, Fluxbox. A lot of LXDE in Lubuntu when I had a 1GB RAM laptop. Some Ubuntu-Unity (not as bad as some say, a lot lighter than expected). A bit of Cinnamon (but lately I am trying it again) and Mate in Mint, some Fedora and Bodhi Linux. In the late couple of years I extensively used elementaryOS and also KDE, mostly in Linux Mint KDE 17.2 (14.04 ubuntu-based, still using KDE4). I also tried to keep an eye on Plasma 5.
I have not really used Gnome 3, it is the only desktop that I found too heavy on my resources (not impacting speed and usability, but stressing my fan, which may be a limitation of my system).
I have also tested Opensuse (Xfce and KDE) and other more exotic distros like Deepin and Solus OS.
I have realized that desktop environments were not equally good in all Linux system/projects, as they were more or less tuned and polished; also, that installing different DEs on top of the default one is one of the biggest mistakes one can make in Linux. I became convinced that if one really wants the best Linux one should stick with the default desktop and select a system which has the desired desktop as default, therefore changing the system altogether than just the desktop environment. This is only normal, given that a good product can only be the result of an effort made by providers to integrate as much as possible the GUI and the non-GUI levels of the system but also the different GUI elements.
Therefore it doesn’t make much sense judging the desktop environments in themselves, because they are made better by the different projects, but differences between them may be important, and also the DE is an important element in judging the OS. The effort made within a Linux project of creating a finished (‘perfect’) OS will result mainly in refining and improving the DE and its integration and sometimes in creating a DE to achieve this goal. Different DE-s, whether created for a specific system or not are themselves different as far as the degree of completeness is concerned. KDE and Gnome 3 are more complete/finished (‘perfected’) than Xfce and LXDE, and in-house DE-s are bound to be more so inside their home project (Unity in Ubuntu, Panthon Desktop in Elementary OS, Cinnamon and Mate in Linux Mint, Pidgin in Solus) than outside it.
is the most known and most largely used Linux OS. But it’s a paradox. While is has the best support both financially (by Canonical) and as a community, there is some ambiguity on whether there is one Ubuntu OS that can be considered the most used Linux or whether this status is related more to Ubuntu as a base for other OS-es, especially Linux Mint, which is a separate project.
Although Ubuntu is the clear favorite for the title as it aims closely at the goal defined above, the end results are somewhat ambiguous. The Ubuntu project includes the creation of a new DE (based on Gnome) called Unity, which is rather different in design than other such interfaces. As such, Ubuntu is a finished and specific end-product OS, sometime called ‘Ubuntu Unity’.
At the same time, Ubuntu has accepted as co-related other systems that were created by replacing Unity with other DE-s, within initially separate but much smaller projects, the so-called ‘flavors’. The integration between these and the Ubuntu project may in fact vary, some of them could be easily seen as separate, but defined by a very limited separation from the main Ubuntu project, a kind of potential separate projects or half-developed projects. It is maybe an intentional scale with gradual steps of rapprochement to the Ubuntu project. Small projects were created with the limited goal of adding new desktops into the Ubuntu family (recently Mate) which then were gradually accepted as Ubuntu ‘flavors’.
What happens is that the Ubuntu ‘flavors’ lack the resources and the commitment that are needed by a Linux project (and thus rely on Ubuntu project for data and support, which is enough motivation for them it seems), but on the other hand are not able to get as much support as the Ubuntu flagship, Ubuntu Unity. The result is that it is not always clear if a flavor is really an Ubuntu as a finished product or just a test of a different DE on top of Ubuntu. The ambiguity is enhanced by many talking on askubuntu about how easy is to replace one DE with another or put one DE on top of another. On this background the flavors appear as just a cleaner way of using other DE than Unity on top of Ubuntu.
Considering the end result of these flavoring of Ubuntu, I think that their stability benefits greatly from the fact that they never go into troubled waters far from the Ubuntu shore, which is for them not just a ‘Ubuntu-base’ (like for Linux Mint or elementary OS), but is the ‘Ubuntu OS’ itself (the end-result of he ‘Ubuntu project’) minus Unity + KDE or Xfce or LXDE or Mate. On the other hand, their quality as really finished products is affected and varies according to the quality of the desktop environment: Kubuntu looks much more as a complete and finished OS than Xubuntu and Lubuntu because the KDE desktop itself is more complete and finished than Xfce and LXDE, which as such would require some supplementary polishing (that they don’t necessarily get as Ubuntu flavors; it is as if they were the answer to a question like ‘is Xfce going to work fine on Ubuntu?’ - and the answer is ‘yes’; and that’s all; while the question ‘is Xubuntu a finished OS worthy of the Ubuntu name?’ has to be answered by the negative).
Therefore, Xubuntu, Lubuntu and Kubuntu are good systems that can give a rather clear image of the value of the desktop environment that they use. But that doesn’t mean that they make the best use of these DE-s or of Ubuntu in general. Linux Mint does a far better job, and those who think they can find something more accomplished than Unity will not improve things by just replacing the DE or by going to one of the flavors, instead they should try Linux Mint. Ubuntu Mate is a good example of a bad choice. Why not Linux Mint Mate, giving that Mate is an old product of LM?
Ubuntu Unity, although lighter than expected, and although they finally made it possible in 16.04 LTS to move the panel to the bottom (big deal!!!) it is still refreshingly buggy after so many years of development! It should be the steady rock of Ubuntu and Linux but it's not! It looks good and complete, and feels how a good GTK desktop should feel, but still with unacceptable bugs on a LTS version!
As for the other Ubuntu flavors, as I said: all are better in Mint.
Testing Ubuntu Mate I was able to see how rudimentary the Mate desktop may be without the LM effort. The same is true for Xubuntu vs LM Xfce. I haven’t yet found a project that does the same for LXDE. Lubuntu can be considered fine by some I guess, namely on a low end PC, but is limited. A thing of the past for me. Kubuntu is a step ahead as LM has not yet adopted Plasma 5. Kubuntu was for me the best Linux I used until now and may deserve to enter the race for the best, but that is mainly the merit of the DE.
I don’t mean that a medium user could not rather easily fix most of the shortcomings of the systems just mentioned. But the purpose here is to identify an accomplished OS, a system that doesn’t need all that supplementary tweaking from the part of the user.
Is based on Ubuntu, but provide also Debian-based versions that I have not tested but that should keep the Debian philosophy, thus not being eligible here.
Cinnamon and Mate are their main desktops. Although forks of old Gnome 2, they are actively supported and in the LM 18 (based on Ubuntu 16.04) have an updated GTK interface that makes them very fresh. Mate can be considered more limited than Cinnamon, especially considering aesthetics, and Linux Cinnamon as possibly one of best GTK systems.
LM Xfce stands as a snappy, simple but complete DE on Mint, to the point that it may be one of the the best overall choice in Linux world, especially if you have a system low on resources.
Mint KDE (using KDE4) in LM 17.2 (ubuntu 14.04) was for me the best Linux experience ever. I thus discovered KDE4 as an outstanding DE, possibly the best DE there is. Plasma 5 will be used in LM 18 (ubuntu 16.04) but was not released yet (June 2016).
Many people take it already as the best Linux, namely for its excellence in design. It is an Ubuntu-based project aiming to create a perfect Linux system with a design inspired by Mac. The project includes numerous in-house applications, especially the DE, Pantheon. Overall, it provides arguably one of the best if not the best Linux GTK desktop (especially if you prefer Mac to Windows design, and considering that Unity is buggy and Mint Cinnamon a bit Windows-style because of the panel). It aims at great stability and a Mac-like rigidity to hazardous tweaking (which in fact is impossible to achieve in Linux, at least by present standards). It is only at the second major release and not only follows long time releases, but they do not necessarily create a new release for each Ubuntu LTS version. Each release is there to stay for a few years at least. The change from Luna to Freya was great in speed and stability but nothing was changed in the general experience, a quality that is possibly one of the rarest and precious in Linux world. (Unity stability of design also aims at that, but with mixed results.) It needs no tweaking, just installing other browsers beside Midori and some supplementary players like VLC, because the default players, and the default browser Midori, are limited. They keep them because they want to develop them into in-house fully integrated programs (Mac-style) and if they are not yet fully developed is because the project itself is in fact in its teens. The system is so good in fact that people forget it is still in versions marked under 1.0 (0.3 at the present Freya). It would be easy to say that the stability comes from Ubuntu, and it is clear that stability is the goal they aimed at when selecting Ubuntu. But creating a new DE to integrate with the Ubuntu base lead to a feat so great that only Mint can compete with.
It's the looks, but also the unified concept, the stability and the continuity. The Plank/Docky panel is one of the coolest things in Linux.
Based on Arch in the sense Ubuntu is based on Debian, fives access to a lot of programs but creates the ready-to-go system that is excellent, with the option of selection between a lot of desktop environments, each with is very active and supportive community (where usually you get rapid help from the very developers; one just built for me the printer drivers I was lacking). My Arch experience was great in Xfce, Fluxbox, KDE. For medium and advanced users, but a way for a beginner to have a taste of the deep. I would recommend it in fact as a permanent Linux choice for most people, and the only reason I don't have a Manjaro permanently on my laptop is that it doesn't work with Grub Configurator, which is a must for me.
OpenSuse - it is a renowned system that provides very good Plasma 5 and other desktops. I see it best for people that are for some reason not prone to use Ubuntu. I find it unpolished sometimes, with a concept possibly oriented more on enterprise integration that on individual use. Also, I have no reason not to take advantage of the Ubuntu-Mint systems myself, but will always keep an eye on that. (I was only able to install it from a DVD.)
Fedora - I haven’t tested the Gnome 3 version and the one I tested seemed to me like it's not making any efforts to satisfy the expectations of the larger public. I’ll have to test again. (Gnome 3 too)
Deepin - A good looking and complete system, based on Debian. Too heavy on my system as it looked close to Gnome 3.
Solus - An interesting Linux build from scratch focusing on end-user simplicity and great looks. It challenges elementaryOS as far as good looks are concerned, but instead of taking ubuntu or other base, they built one of their own, with whole new packages, installation commands and a new DE, Pidgin. Still incomplete in mid-2016: no printing and no keyboard layout switch.
Bodhi Linux - Mentioned from time to time by some people when it used the Enlightment desktop, I found it a flashy bad taste kinky unfinished buggy of a Linux. I was asking myself what's the point of making an ubuntu-based Linux oriented on looks that needs such effort and competence to make it work - which I never succeeded. It's Ubuntu, FGS! - Now I see they gave up Enlightment ... in favor of Mocksha Desktop, which is based on Enlightment, but with a default style that goes to the other extreme: it’s fully gray. But it is easy to go back to the old flashy style. It benefits from the Ubuntu base, possibly helpful for people that want a bit of good looks on older hardware. (But then I would use Fluxbox with Docky/Plank or even LM Xfce 17.2) It smells like a desktop from 15 years ago.
SUMMING IT UP: Which one is THE BEST LINUX?
The main candidates are:
- Linux Mint 17.2 KDE4 is the best Linux I have used or the one that I was most impressed with.
- Linux Mint 18 Cinnamon - But lately I am very impressed by Linux Mint 18 Cinnamon, while the new KDE version using Plasma 5 is not yet here.
- elementary OS. For a long time I was skeptical when it came to a project so focused on Mac-style design and limitation of user's initiative. But then I realized that too many freedom is chaos in Linux and that I was only asking for total freedom because the Linux systems I used were so unfinished. But elementary aims at being a perfect desktop and, on top of that, its stability and coherence (and also success compared to others) make it a really good candidate.
- Xubuntu 16.04. Ubuntu needs a chance, and I will re-try Unity. The last try was with 16.06 LTS: the dash froze a few times which was enough to dismiss it again. But the Plasma 5 version (Xubuntu) was impressive: the unfinished details that were present allover in KDE4, fixed by the Mint edition but not by older Xubuntus, are now absent in Plasma 5, and Xubuntu 16.04 takes advantage of that. I hope Mint 18 KDE will be even better.
Manjaro is great but I think that user intervention is unavoidable.
Systems using Gnome 3 might be great, For that I needs a more discrete fan.
Considering desktops (DE) as such:
Gnome 3 – I have avoided it as it is too heavy on my fan, but I have to try it again. One may never know if it’s the DE or the kernel etc.
Plasma 5 - Remarkably mature already in Kubuntu 16.04. The problems from earlier versions have been rapidly fixed. Avoid updating to the testing version though - if you want the best experience. (I have found Kubuntu 16.04 very lean and light, but I made the mistake of upgrading to Kubuntu latest sources and the fan went mad.) I am looking forward to Linux Mint 18 (ubuntu 16.04 LTS base) with Plasma 5.
Xfce - Light, it is also one of the most complete and configurable desktops in Linux. Although in itself looks a bit un-polished, it is fine in Mint. Maybe also in Manjaro, which just like Xubuntu may need some tweaking.
Fluxbox – tested in Manjaro, it is impressively complete in itself, given the simplicity and lightness. Can be easily completed with Docky and Synapse.