Making radical changes to an incumbent system is difficult and often thankless, even when the new version is, if not objectively better than the previous, very good. Apple encountered this with the transition to OS X. Anyone who remembers the "Classic" environment can attest to how poorly this transition was perceived by many Mac fans. Microsoft has a real problem on its hands trying to convince users that Windows 8 is better than Windows 7.

Ultimately, GNOME 3 feels like a much better designed environment than the previous version. For those that tried an early release and gave up on it, I'd encourage you to go back and try it again.

It's one thing to go through the pain of a major version transition to make something substantially better, but Gnome 3 hasn't done that. I just read through the release notes for the 3.0, 3.2, and 3.4 releases of Gnome to confirm that I wasn't missing anything, and can find literally nothing that actually makes my life significantly better. When you're making major, breaking changes, you need to be able to justify them. As a user, Gnome 3 has done nothing at all to justify the many hours I spent trying (and failing) to get Gnome 3 to a point where it wasn't actively painful to use. (And reading posts like this gives me no confidence that it's going to get any better.)

For the record, I do think GNOME 3 is "substantially better" than previous versions, but I also understand that this is in no way an objective measure and it's perfectly reasonable for us not to hold the same opinion here.

I suspect that most of the major changes have been on the back end, and that it is a substantially better platform to develop on. Unfortunately, that doesn't affect me at all as a user. I want to be able to run terminals and a web browser and not much else (and I don't think I'm alone in that), and so making Gnome a better environment to develop against is unlikely to make my life any better in the foreseeable future.

In this instance, Ubuntu lost me to mint Linux. I expect a similar backlash against Windows 8, although for slightly different reasons.

Enough is enough. Goodbye gnome 3 and Ubuntu.

I've personally been a huge fan of Gnome 3 and Gnome Shell, and I still think Unity is a terrible, ugly mess. But because many still think that Ubuntu is just using Gnome, they project their distaste for Unity onto Gnome 3 instead of where it belongs...

So you've never used it, yet feel the need to coment on its features anyway? Try it. Some things, like the dynamic vertically stacked desktops, simple and attention-grabbing notification UI, and clean overview mode (and even the autohiding status bar, which I hated initially but have grown to like), are actually really nice. Yet they aren't going to look very impressive if your only "experience" with them is screenshots and release notes.

It's good. It broke a lot of stuff, and so many people will forever hate it. And it still has bugs (my biggest pet peeve is that being in the overview mode breaks all the desktop keyboard shortcuts) and misfeatures (app-based window selection will never work for me, though the traditional alt-tab mode is now a well-supported extension). But it's good.

Try it for a few weeks, learn it, then come back and flame.

I read through the release notes to confirm that I wasn't missing anything important in my own use--which I noted in the part of the quoted sentence you excised. I also said, in the same comment you're replying to, that I spent many hours trying to get Gnome 3 to a point where I found it an acceptable desktop environment, and failed.

The lack of a Start-like menu was initially off-putting, but I've grown to appreciate the lack of that bit of complexity. 99% of the time I run: Chrome, terminal(s), Skype, Pidgin and WebStorm/PyCharm, so I don't need to fiddle with settings or wonder about where in a menu an application resides. When I do need to find an app, the action button and search are perfect.

The only thing I miss in Gnome 3 are the many taskbar plugins that show mem/processor usage, but those will come. Otherwise, Gnome 3 does a great job getting out of my way while I do my work.

https://extensions.gnome.org/extension/120/system-monitor/

I think this is currently a big missed opportunity for Gnome 3. The ability to extend the shell using JS is huge and the project has done a terrible job publicizing it. I was just googling around for it and had fairly hard time finding good, official docs on how to build an extension... If they dialed that in and got the GitHub crowd excited, they'd have a vibrant community in little time.

EDIT: Wow. The one click install from Chrome and auto activation is really nice. I clicked the link, approved the extension install (system dialog) and, BOOM!, pretty charts and graphs!

This includes stuff like reload (Alt-F2 r) and a reasonably nice data inspector (Alt-F2 lg).

They definitely should publicise these more.

EDIT: I'm speaking from my experience with Unity, where you could enable sloppy focus with the gconf-editor, but it caused various kinds of wigging out and wonkiness with Unity (probably why there's no easy way to enable sloppy focus).

I tried gnome 3 for a few minutes once, and got the feeling it would probably have similar problems with focus-follows-mouse, and that's probably why there's no easy way to enable it there either.

One thing the gnome3 guys are doing is putting application menu stuff up in the task bar (window-specific menus will remain in the app window), but when focus-follows-mouse is enabled, the app-menu will remain in the menu-bar instead of up in the panel.

    gsettings set  org.gnome.desktop.wm.preferences focus-mode 'sloppy'

Gnome Shell feels overdesigned. I just can't used to it. It draws too much attention to itself instead of getting out of the way. Classical Win7/Gnome2-like panel + exposé would be much better.

Gnome Shell really doesn't help in that scenario, nor does Unity when I tried it. (And Unity's app menu is obnoxious.) A taskbar has been the best UI solution found so far, although hopefully someone can improve on that.

The people most likely to be seriously inconvenienced by the poor multiple window handling are the ones most likely to be able to moan about it.

Gnome 3.4 doesn't let you use Gnome Shell with the "classic" Panel they provide. More annoying for me is if you use a classic session and try to put a panel on the left then it draws the text vertically which is almost unreadable. Modern screens have far more horizontal space available so why won't it go wider and draw horizontally?

Switching from a current emacs to the 6th terminal window isn't particularly practical for me using keystrokes (ymmv). I also drag and drop items in the taskbar so they are in the order that makes sense for me.

I gave up on extensions because they kept breaking, and of course the moment there is a gnome version change they all stop working.

As I mentioned it would be great to be able to use the panel for the many instance apps and shell for the single instance ones but that isn't an option with Gnome 3.4. With Gnome 3.2 that is how I was doing things, but the panel wasn't part of gnome and instead there were several done as extensions each with its own quirks and breakages.

I also don't use Gnome Shell on my laptop for battery savings instead opting for the panel without 3d effects and not using compositing in the window manager. I want as few CPU/GPU cycles as possible spent on the laptop.

"I used gnome 3 shell for 6 months until I could no longer stand it. Please please, Gnome team, go back to Gnome2 or Mate. The desktop really doesn't need reinventing. You're solving fixed problems!"

Unfortunately, I don't know how well the open source adopts the philosophical concept of sunk costs.

I'm surprised people think this. Do you think we're going to still be using the Windows 95 model in 10 years?

Sure, the relative importance of the desktop will diminish as other forms of input become viable, but we're not there yet. I still do 95% of my work in a desktop environment with multiple windows and everything, just like I did 10 years ago. The difference is that the desktop environments I have access to now are better than the ones I had then. I hope that trend continues.

We've got windows. We've got keyboards. We've got mice. In some cases we've now got multi-touch keypads and touch-sensitive screens. Some screens are very large. Some are very small. Some are in-between.

There have been some improvements: processors are vastly faster, pixel densities are increasing, and rendering has improved. We've got some hacks for displaying fonts more clearly. But those are pretty minor. Compiz effects if you want to get flashy, but those are hardly necessary.

I've seen master developers completely satisfied, and massively proficient, in hacked-up twm environments.

Really: the desktop is a solved problem. Move on.

I've looked at many of the "modern" desktops. And I'm still using the same one I was using 15 years ago: WindowMaker, based on NeXTstep. It's simple, it's fast, it's light, it's stable. It's not visually distracting. It gets out of the way. It supports keyboard shortcuts and navigation. It presents me with persistent pinnable window lists (something no other window manager / desktop environment provides me) and pinnable submenus.

What the desktop developers for GNOME, Unity, Microsoft, et al are forgetting is Anoine de Saint Exupery's dictum: perfection comes not when there is nothing left to add, but nothing left to take away. There's an implied "and what is left is sufficient to the task at hand".

Whose workflows and at what point? Do we think the children born today will be satisfied with GNOME 2 (or GNOME 3 for that matter)? I agree that we're playing fast and loose with the definition of "reinvention" here (and that GNOME 3 is clearly not a reinvention of anything--to be fair, he started it!), but I think there's room for better than incremental improvements in all facets of computing, including the desktop.

For workflows I just think of some complex things I do that may end up being a 30-40 step process with tons of key presses and clicks. It's probably way more complex than it needs to be but I know it so well that I use the complexity to my advantage. So for example I know step 15 is a good spot to do a backup because step 16 might be a destructive process. The granularity this offers is so powerful. If I could replace it with a 10 step process I might lose the granularity. If I replace it with a new 30-40 step process it wouldn't really benefit me.

People enjoy the idea of not having to sync their data because it's all "on the cloud" (eyeball roll) and because they have a 3G modem.

Sooner or later that enthusiasm will fizzle out.

Maybe. We used the command-line model for 25+ years before that.

I sure hope so. It's ideal for laptops and desktop PCs, or at least none of the alternatives are good enough.

The question is, is the desktop metaphor still a good one? I would argue it never was good, just necessary. My question is, why do I have to move my mouse here and there to get things done on a desktop? Why is it so painful to use a laptop trackpad to get here and there with any accuracy? Newer window managers try to mitigate that by removing the need for a mouse altogether. That's a step in the right direction, I think.

Don't take my word for it, this is a falsifiable statement. Run usability studies on both novice and experience users with different desktop models and metaphors.

Ensure that you're not just testing initial impressions, but usability after a week, a month, a quarter, a year.

I've used Windows, Macs, and multiple Linux desktops for years on end. Hands down the environment that lets me get the most done with the minimum of hassle and interference is Linux.

My desktop (used on both traditional PCs and laptops) minimizes mousing actions by maximizing keyboard inputs to perform work and manage my GUI).

WindowMaker.

- Hit F2 to invoke run dialog, type command (I've substituted gmrun for the default WMaker run dialog which lacks history or command completion).

- Your defined hotkey. I have launchers for terminals, edtiors, web, and mail clients -- stuff I need frequently.

- Menus can be keyboard-driven. F12, and arrow-navigate to your desired item.

The problem is that the devices where the traditional WIMP interface make the most sense (desktop & laptop PCs) are on track to become a small niche in the overall computing landscape, thanks to the wild growth in mobile and tablet computing. While desktops and laptops aren't going to disappear anytime soon, they're not where the action is.

So if you're a maker of an operating environment, you can stick with an interface that's highly optimized for desktops and laptops, but that severely reduces the relevance of your work.

It's like choosing to make an operating environment today that only runs on IBM mainframes. Those mainframes exist, they're still out there, and you can still make money selling software for them; but that choice makes your product irrelevant to the vast majority of computer users today. And when you slot yourself into a niche, you can never grow larger than the niche you've slotted yourself into. If you want to be an industry leader, you have to be playing in the same space where the users are; you can't get any further back in the parade than the middle.

All the significant commercial developers of traditional WIMP environments today (which means Microsoft, Apple, and Canonical) are struggling with how to make this transition -- to stay relevant on the desktop while also extending themselves into the tablet/mobile realm. It's a big challenge, and it's not clear at this point that any of them have figured out how to do it yet. But they have to do it to stay viable as mass-market vendors.

I would imagine the pressure to make that transition is even greater on open-source projects like GNOME. If you're a commercial player stuck in a niche, that's not necessarily a bad outcome if the niche is at least profitable. But open-source development is driven more by developer ambition. Developers want to work on hot new projects that lots of people will use, not old stable projects for niche audiences. In a commercial environment you can salve their disappointment at being stuck with the boring old projects somewhat with cash and perks. That's less true in FOSS development. So the talent flees your ecosystem to go work on things that are more fun, and you slowly die by inches.

Do people honestly see long-time LINUX users dumping their laptops / Desktops for Tablets / Smartphones?

I really don't get the pull for "unified" interfaces. It doesn't make sense. I don't do the same things on my mobile devices that I do on desktop / laptops. Not. At. All.

There is no reason to Unify them. The closest thing I could see is similar interfaces, with features useful on both included, but otherwise different.

GNOME/Fedora and Ubuntu are not targeting long-time Linux users. They are targeting people who have never used Linux.

GNOME 3? A new project was started; mutter and gnome-shell.

You say that starting a new project is better than dumping other projects. So I guess you want people not only to start something new, but also keep maintaining their old things? It seems a bit strange to me.

In case of GNOME 3, gnome-panel and metacity are still available.

There's certainly plenty of room for improvement though. I can remember trying Window managers 10 years back, and workspaces - and thinking how great they were. And how they were a refreshing change. Gnome shell sounded really promising in the early days. Challenge existing paradigms by all means.

My powerbook with OSX 10.4 died just before Xmas. And despite it's age it still beats a lot of the current offerings. Raise the bar.

When you've got your own people, tons of power users, ordinary users, and key kernel devs telling you you've gone astray, something's wrong. What we're seeing is fortress mentality and hunkering down. Wrong move.

Ted T'so has commented on his desktop woes at length on G+. I lifted one a reply he made to his own post which I feel gets to the heart of the problem, and what is really necessary for a desktop solution (probably XFCE4 for most people these days). Note that Ted's including both GNOME and KDE in his criticisms:

My concern with KDE is how bloated it's become. I don't want highly integrated applications. My main applications are a terminal, emacs, open office, and the Chrome browser. I do way more with web applications these days. Bloat is bad not just because of the increased disk and ram usage (although I do worry about that); bloat is bad because of the complexity. The more you rely on an Object broker, the more it becomes harder and harder to fix things when they break --- and things do break, with appalling regularity. It's the fact that Xfce and Network-Manager uses D-Bus and PolicyKit which is why I still get asked for my password from time to time even with my localauthority hack. How do I fix it? Who knows?

Unfortunately wicd isn't fully-featured enough, but it has as a huge win the fact that it doesn't use PolicyKit; it just uses a simple Unix group membership check for its access control. Once wicd can handle large numbers of access points with the same wireless network, I'll be removing network manager from my system and using wicd instead, because network manager is just too damn complicated, and relies on too much complex infrastructure which I can't debug because of the !@#@! D-Bus design. KDE is no better in this regard, and in fact it may be worse.

Why don't I use Openbox? It's too simple. For one thing, it doesn't support 3x3 workspaces. At the end of the day, I want a desktop environment that makes the things that I do simple. Which means wireless roaming, the terminal, emacs, open office, and a browser. Anything more than that increases the memory and disk usage, but worse, it means more infrastructure which can break, which means I end up having to debug things like PolicyKit. And that's just not productive time for me. So Xfce seems like the right balance point for now, although it definitely still has some rough edges.

https://plus.google.com/u/0/117091380454742934025/posts/VjbF...

Once again: HN needs a proper blockquote markdown.

I don't really see how you're quoting this as a good example of why GNOME doesn't "get it", while the situation is exactly the same as 2.x.

That said, my situation is much like Ted's. I don't generally use, or find much value in highly and tightly integrated applications. They're misfeatures. Most of my work is in terminal windows, editors, remote access tools, and browsers.

Early in the GNOME and KDE experience, it became very clear that these where complex systems with many moving parts. My distro of choice (Debian) makes using things that need to be updated in lockstep (GNOME's home environment, Red Hat, does this) kind of difficult to sustain. Less so in stable, but very much so in testing/unstable (where I prefer to live).

The systems (KDE and GNOME) keep getting more complex, they keep stripping out more power-user features, they keep getting in my way. Really, I've got little or no need for this.

KDE at least supports PulseAudio better than ESD but it could as easily have supported ESD if it were necessary, especially if there was as much support from ESD devs as was provided by PulseAudio devs (ESD was already in its nadir before KDE 4.0 thanks to broad usage of gstreamer making ESD redundant for most audio needs).

Also, Gstreamer is not the same as PulseAudio.

Especially the latter is easy to verify. Go to their mailinglists and see the level activity, it sure does not look very good. Reading the actual mailinglists gives sometimes a quite disturbing image as well, with a few examples already linked in Mr. Ottes mentioned blog post.

However, note that the current slide in the numbers is something that only happened after February-March 2012, which was when the previous major release was launched. In 2010 another slide lasted from April until October, so I suppose there is a need to worry if contributions don't pick up around Sep-Oct.

That said, the situation for GNOME is certainly different now, given that they lost their biggest distribution channel in Ubuntu doing their own thing.

(Or could it be that there aren't any users?)

Gnome and Unity are both targeted at some sort of hypothetical casual linux user, that doesn't really exist.

Casual users DO exist, but they want iOS, not Linux - or more generally - an appliance, not a general purpose tool.

The truth is, there are people who like Unity and there are people who like Gnome 3. You're just not hearing them because the negative types are yelling over top of them. The people who like the new WMs don't have anything to add to this discussion because the entire discussion is about how those people don't exist. I feel like if I mention that I'm using Unity right now and I like it, I feel like I'll be told "no you're not and no you don't, case closed."

You don't like it. We get it. But claiming that these people don't exist seems a little disingenuous when obviously they do or Ubuntu wouldn't be quite such a popular distro. Unless you're suggesting everyone downloads Ubuntu and then immediately installs a different WM?

The people complaining usually are those who were happy with Gnome2 and just want it the way it was. Guess what, nobody erased Gnome2 from existence, and as far as I know it was in maintenance mode anyway, so nothing has changed for them. Use Mate if you prefer it.

And many (like Linus) probably just want a lightweight WM to use CLI, a browser and maybe another application. Guess what, you already have that, and there are several of them to choose.

But many of us want a complete and integrated desktop environment with lots of eye candy, multiple applications with similar look, complex file search, GUI installation and the whole shebang. We don't mind if it takes extra RAM and CPU because we have it. KDE4, Gnome3 and Unity try to give that, and many people do enjoy it, as I assume the developers of those DE's.

We get it, you don't like it. That's fine! You have choice! Just don't assume your opinion is universal, and don't be a dick about it.

to just take an example, the user-friendly way to update Ubuntu linux is

$ sudo apt-get upgrade

try to do it with the GUI and it's a kafka nightmare. sometimes it just hangs up for hours with the progress bar stuck. Installing software on the command line is just

$ sudo apt-get install X

whereas you can screw around forever with the GUI and never find the package you're looking for

Oh, what's that? You typed apt-get install solitaire? What are you, a fool? What you actually wanted to type was apt-get install xmividev_2_2.0.2.3.1-dev-final-release-source. It's totally more user-friendly than a GUI.

When I worked internal tech support I had users call and complain about Windows 7, asking to go back to DOS. These people stood out more than the 90% of users who enjoyed Windows 7 simply because the complainers were motivated enough to complain and curmudgeonly enough to make such a ridiculous request. The difference is, those users were forced to use software they didn't like. That's hardly the case here.

OS X has a better combination of desktop quality and application support, but if Unity could magically run OS X apps, I don't think I'd ever use or recommend anything else.

A handful of tweaks and extensions have made it great. Use it every day for work.

This is the camp I fall in. It's not that I don't mind the new, improved, eye-candy-focused drive of Unity and Gnome3. But why abandon the products that worked to begin with?

After about a year or so of using KDE I switched to Gnome, and then to Blackbox and then Fluxbox and eventually ION.

For me, now that I am accustomed to the various shortcuts and key combinations, gnome3 is amazingly fast to use compared to gnome2 (and any "start menu" based desktop). I really don't know how I could go back! Whenever I have to use a relatives computer it's an incredibly frustrating and slow affair with my hand instinctively flying up to the top left only to find that nothing happens. First thing I do when sitting at my own desk is throwing the mouse pointer (or hitting the "windows" key) and typing "ter" and pressing enter.. terminal appears and work begins..

Gnome3 is open source and Free Software, those who don't like it will move to the gnome2 fork and I am sure will be very happy. You can't force everyone to use (and develop) the desktop that you want, that's the whole point!

"But no one is using it!": well, I think the gnome3 developers are probably using it daily.. and if noone else in the world was a user, maybe they would still be producing it for their own benefit.. would you still ridicule it in that case?

I always thought GNOME was a desktop system.

Not sure how they plan to take that desktop system and target it as a server platform?

What saved Unity, I think, is firstly that it had time to get farther along the development curve before Canonical released it into the wild as the default (where Fedora 15 was shockingly rough around the edges) and second that Ubuntu has become the default development environment for cooking Android images, drawing a bunch of new users who aren't attached to the old Gnome 2 models.

As long as you can access a terminal quickly, everything else is really window dressing.

Anyway, this is not the future of GNOME. Different projects and all that.