I would imagine new adoption to be comparably small, but why stifle it?
IF new user adoption is the real goal, it always made sense to me to ask if you were a new Linux user during a fresh desktop Linux install. From there, a Yes would result in the next question asking what environment you were coming from [Windows $version|OS X $version|whatever]. Based on that answer you could tailor the DE to be as welcoming to the new user as possible. Over time elements of the old experience could be changed, with tips and guidance along the way, so that in $some_arbitrary_time (maybe 6 months) the user would be comfortable in a completely native Linux environment. Meanwhile people who answered No to the initial question could install whatever DE/WM/Whatever they wanted and go on their merry way. Would this be difficult to implement, require a ton of testing, Q/A, and detail work? Absolutely. Can I think of a less jarring way to sanely on-board new users while not impacting current users? I cannot.
âjeremy
To me, the kind of desktop that you describe here would not be appealing to experienced users anyway, but could it rather be used like a stepping stone to other destops, ones more advanced?
Why would this necessarily stifle it?
For myself, someone who has used linux as my primary (and only) desktop for about 6 years, I donât want to have to manually handbomb init files or recompile kernels or any of the other stuff that goes with that. I donât think the system should require an advanced knowledge of linux, but shouldnât deliberately target new users. As @jonobacon and @sil refered to the desktop should really just get out of the way. There is so many different pathways to go down when the target market is a group aligned more closely to the developer making the changes rather than a mytical new user who is going to bring with them all of their friends.
Virtual Desktops are amazing, something which made Linux Desktop a strkingly better option when I started using it. Building on top of those in a consistent standard way sounds exciting. Dell with their xps 13 tried something along these lines but I think we could continue that.
Mobile interactivity, why when my Android phone runs Linux and my computer runs Linux, can it not automagically tell it Iâm listening to bad voltage and Iâm at 15:36, so that when I stop the podcast on my phone, its ready and waiting to resume playback on my work computer.
Lets attack the things that we want as Linux users, if we love our desktops, that will automatically get new linux users just by our enthusiam, When Desktop Linux has real tangible benefits as compared to other Window managers thats when the Linux marketshare could start to increase.
I want a stable (mostly) desktop environment, where the real engineering is done in trying to make the experience more unified and consistenly improve on the core system.
I want less of the âparadigm shiftsâ for my desktop. There was not a whole lot wrong with Gnome 2, but yet it was replaced in its entirety with something that has in no way persuaded me that it will improve my workflow signficantly. small stable steps (possibly in faster increments) would be better IMO.
The Dell XPS 13 is the thing that pushed me on the Mac side. And honestly, I havenât been happier with my computing experience.
A little prelude. I got the xps from work, but it turned out to be a complete turd for many reasons. The hardware that Dell shipped has quality assurance issues. The screen has (I still have the machine) severe pressure points and there is a noticeable buzz coming from somewhere on the inside.
Thatâs not the whole story. The drivers for the wifi, graphics and HDMI are also a total mess, at least in 14.04 and 14.10. I ended tethering the laptop to my phone for most of the time, because the wifi drivers are a train wreck. Then there were the daily lockups because of weird audio and graphics bugs present somewhere in the software stack, etc. Oh, and, I never managed to make the laptop to go to sleep.
I finally got fed up and bought myself a MBP13 with my own money. The thing works surprisingly well. It goes to sleep without any problems, it has excellent battery life and the quality of the hardware is superb. I got pleasantly surprised by the software as well. Very sleek and pretty and awesomely functional. One thing that surprised in a positive way me was the app installation. Simple apps you install once by dragging their icon (itâs an app package actually) to the Applications folder and youâre done. New versions of the app are downloaded and installed automatically.
After using Linux exclusively for many years I never thought Iâd say this but for regular work macs (and expensive windows PCs) blow Linux out of the water. Linux is fine for supercomputers and servers. RedHat also manages to make a lot of money by selling support to the military and other big companies. But, on the consumer market Linux is a complete failure.
Two things about new users:
- if you tell a new user âok, youâre coming from Windows XP, so letâs install XFCE since this is what looks the most similar to Windows XPâ, then how are you gonna tell him, 6 months later, âalright! You had a blast with XFCE, now time to switch to another Linux DE because you can!â? New users wonât want to change their habits. Hell, they already miraculously landed on Linux, and now you want to change everything again?
- A lot of people I know moved from Windows laptops to Mac OSX laptops. For some reason, they jumped from one DE to another, and never really complained about it. If people can do that, how can they not do it from Windows to Linux?
To get back to the discussion about apps, I would say Iâm always undecided about something:
If I was to create an app, should I create it for a very specific platform so I can enjoy all the tools available for it, or should I create it as generic as possible so I can port it on other platforms?
On the one hand, if applications like Firefox, Thunderbird and LibreOffice were only available on one platform, I could have never asked my mum to switch from Windows to Linux. But since these programs were available on Windows, switching to Linux was super easy: my mum found back the softwares she used everyday and was delighted with that. As @jonobacon says, in that case, my mum didnât care about the desktop, but about the apps she was using.
On the other hand, Iâve used a Mac for a few months (before giving up and switching back to a Linux desktop), and I was quite impressed by a few softwares available only for Mac: GarageBand and Pixelmator for instance. Because they were only available for Mac, they were tailored to make the best use out of all the features available on this specific desktop environment.
So thatâs another problem: should you focus only on one platform and DE when creating an app? Or should you think on how to port it, so that more people will be able to use it?
I think the way is this. If youâre an existing user of the Linux desktop and you think that Ubuntu is too tailored towards mainstream users, then donât use it; thatâs fine and good. But stop telling new people to not use it. I have no problem whatsoever with people who say: I think that Ubuntu isnât working for me, because Iâm a power user, or I donât like Unity, or Canonical prefer usefulness over privacy. But when someone says, hey, Iâm thinking of moving to Ubuntu from the Mac or Windows, do not say âdonât use Ubuntu, because Unity sucks! Use Xubuntu or Fedora or Arch! You should become a power user because thatâs whatâs important!â What that does is makes the prospective new recruit just walk away from the whole thing. There are too many dedicated long-term Linux users who tend to have a very black-and-white dogmatic view of things; theyâre using the Linux desktop because what they care about is FSF levels of freedom, or because they hate content silos such as Facebook, or because they can compile their own device drivers, and thatâs a good thing, but imputing those views onto others â forcing those views onto others â is what I think turns off new users. I prefer to think of someone moving from the Mac to Ubuntu as having made a valuable step in the direction of freedom and control, which is something to be celebrated, and in general I would say that other people using Ubuntu feel the same way. In general I would say that someone using a Linux desktop which isnât Ubuntu tends to see a person moving from Mac to Ubuntu as not any kind of victory at all, which is an alarmingly poisonous attitude.
A little copy-pasted from what I wrote on G+:
IMHO we donât really need the DEs. As soon as apps start using their distinct features, they lose portability. And who needs fancy stuff like a 3d desktop anyway? Iâm only running a window manager on all my desktop machines. Only one of them has just recently been added LXQt, just to see how the new version behaves. I removed lots of features, and basically all I have from it now is the bottom panel with a menu and the clock. What else could one need? That would just bloat the beautiful desktop. 
Of course many apps use some DE-specific features. I donât like those. And I think we have to distinguish between different kinds of apps. If you bind an app very closely to a DE, then it becomes more exclusive. I prefer apps that donât pull in half of the GNOME or KDE libs and stuff. It doesnât have to be that way. We could develop a new layer, something like a widget provider. Then we could effectively run any app on any DE.
Yes, refusing to use any desktop features allows you to run any app on any DE. However, it means that there is no consistency between your apps. How does an app change the volume? Show that a download is happening? Offer quick access to its features? Show that there are pending notifications? With your way, every app does that itself, in a different way. Providing this sort of low-level consistency across applications is what a DE is for.
Just wrapping up: you donât need a DE for a notification daemon. That works by itself very well.
You donât need it for audio control either. I just run pulseaudio and control the volumes there. The apps shouldnât control them. Otherwise I would totally lose overview; I have 3 output devices and several apps that produce audio.
Considering downloads: Why would I need another widget for that? I would prefer to have a notification when something finishes, pretty much like on Android. But we already have the notification daemon for that.
But to better understand you: What consistency do you actually mean? E.g. what âfeaturesâ would I need quick access to? I would guess that Iâm looking at things from a totally different point of view. The consistency I want pretty much includes only the theme and common keyboard/mouse controls.
That is the main reason why I moved aways from DEs: They just overdo things. And they do it heavily!
To show notifications when they come in, yes. To show you that notifications arrived and you havenât read them after theyâve left the screen⌠thatâs not what the existing separate notification-daemon does. Sure, you could build a notification-daemon which remembers the notifications and shows them somewhere, but⌠where? In a window on its own? And if you say, Iâll install this new notification daemon, and Iâll install a convenient pulseaudio monitor, and Iâll install these other things, then basically what youâre doing is building your own DE out of bits. Some people like doing that, but would you not agree that others think that itâs nice to not have to?
Thatâs exactly my point. I donât think this is a user preference thing. You describe the process of picking choices of specific sets of software. That is what single users who want it as well as distros do already: Choose a basic kernel config (and maybe patchset), a boot manager, an init system, display server, display manager, frameworks etc. and then build some more apps on top of it, sometimes their own. And then we have the DEs which are also chosen by the distros to be integrated as well - but they do the same job again by providing their own widgets. That is what I think is unnecessary work. We gain strength from sets of packages we can choose from for specific needs, but we loose that strength if they become tightly coupled. I would prefer the DEs to only provide interfaces. So if you want a fixed set of software, choose a distro and you have it. That would be enough!
So would I. And then people donât implement the interfaces correctly. Since you talked about the notification daemon, letâs take that as an example here. The notification protocol has an option which says ânotification daemon, can you handle having action buttons in a notification?â Gnome and KDE notification daemons replied âyes, we doâ, and they were the only notification daemons. So nobody ever bothered to check the actions-supported value. When Ubuntu released notify-osd, which does not support buttons in notifications, and correctly replied ânoâ to the query asking whether they do, a whole bunch of apps broke because they assumed that buttons worked in notifications without checking, and the developers of those apps blamed Ubuntu and notify-osd for breaking their apps.
People donât code to the spec. They code to the implementation. Raymond Chenâs The Old New Thing blog has spent ten years explaining this sort of implicit backward-compatibility constraint that the OS has to deal with. Saying to people âbut we obeyed the specâ makes them say âbut you broke things so we donât care about the specâ. So thereâs little point in encouraging it. Build apps for Ubuntu. Cross-platform highly integrated apps are basically too much work.
1 Like
Good point here and I totally agree. That is why I suggested widget providers instead. 
I guess I misunderstood you. I agree, apps are the end of the means. A switch is a critical part of the current when lighting up a beautiful candelabra. But,should the switch be the primary focus and not the candelabra?
That is miles away from what I was describing. Asking them to wholesale change DEâs after a small amount of time would be more jarring, not less. I was talking about a few small aesthetic changes which would make non-power users coming from other OSâs more comfortable with Linux initially. As an example: Coming from OS X? Weâll put the dock at the bottom. Then, after some time interval or at the next upgrade have an indicator say something along the lines of: âWe notice that you were a previous OS X user and have the dock at the bottom. Most $DE users prefer to the the dock at $preferred_location because $compelling_reason. Would you like to move it?â. I can think of a couple areas where this may help, although Iâm not completely convinced it actually would help.
âjeremy
1 Like
I see the theory here, but Iâm not sure I actually agree with it. Making a Linux desktop superficially like something youâre familiar with may help, but it may also hinder; the dock at the bottom helps with your OSX-trained muscle memory, but that dock doesnât actually work the same way. Since things are different under a Linux desktop from what youâre familiar with, Iâm inclined to make them look different too; I think that having some things look familiar will actually hinder understanding both in the long and short terms. (Few people wanted their tablet to work like Windows 7, for example.) But I freely admit that this is just my opinion and @jeremyâs opinion is just as likely to be right; what this needs is actual user testing, I think. (I wonder whether Canonicalâs user testing covered it. I wish theyâd published the results.)
1 Like
To be fair, Iâm not convinced either, as I mentioned at the end. I think there is merit to just being different right from the get-go and would have liked to see some actual user testing around this as well.
âjeremy
1 Like
I think the form factor helped a lot there. For a lot of people, smartphones and tablets were a completely new device. I remember when I visited the Computex in Taipei in 2006 and saw a tablet PC with Windows XP on it, I was really wondering why would people buy and use that⌠Fast forward to 2014, and a lot of manufacturers are trying to sell laptops that can be used in 3 or 4 different ways depending on how you orient the screen! (and on the software side, itâs very often Windows 8, because I guess Microsoft tried hard to get a desktop that could also be used as a tablet OS).
OK, OK, I digress.
To be fair, I was playing the devilâs advocate here 
Neither am I. In the end, when you get comfy with a system, you either
- keep using it exactly the same way or, if you are a bit adventurous,
- try to dig into its options. Itâs a lengthy process but sometimes it can be very rewarding. How many computer engineers do I still see right clicking a link and select âopen in a new tabâ instead of just middle-clicking the link?
