This discussion came up on our internal mailing list today. As this turned out too long, and I didn’t want to start a mailing list frame war, I decided to post it here instead. Feel free to comment below, or ignore 
.
Here’s what Apple does well in my mind: Apple does a good job of providing a solution that works out of the box for a limited environment and a very limited supported set of hardware. They invest considerable resources in that and it works quite well, most of the time, for most people, especially non-tech ones that don’t turn screws. It even works well for many tech users (e.g., myself). I personally consider it the best compromise for my productivity. And that is even though I miss the freedom I had during my decade of using Linux as the main OS, and I am annoyed by how they close they platform and much of the corporate behavior they are exhibiting.
Ubuntu desperately tries to emulate that user experience, with way too few resources and for a completely undefined hardware landscape. They fail miserably.
Two developers at tt gave up on 11.04 this summer and switched over to Windows 7 as their desktop OS because of recurring issues with Eclipse crashing, menu bars vanishing, X hanging (yes, we give everyone their OS of choice and let them administer their own desktops). Two different sets of hardware, two different people. For UNIX only stuff, they now work via VNC on my workstation which is still 8.04 because I am a conservative bastard . From my experience, and much of what I have heard, 11.04 is simply more broken than 10.04 and that was more broken than 8.04 (and arguably, that was more broken than 6.06).
My claim is that Ubuntu should stop trying to emulate what Apple is doing, because they cannot possibly do so successfully, given the resources they have and the hardware landscape they have to support. And they should stop making these lonesome, ill-communicated, ill-documented decisions, like, using an untested event framework to reduce boot times so they look good in quick first-glance reviews. Or switching from gnome standards to an unready custom shell that was designed for *Netbooks* of all things. These decisions alienate both their power users and the external OSS developer community.
They should return to embracing the open source community and try to build a working desktop out of community supported OSS components for power users that know what they are doing. We don’t need shiny demos. We need a working system.
Let Apple cater to non-tech people — let’s be serious, they are never going to use Linux on a general-purpose PC anyway. Most of them will be using iOS / Android type devices as their main platforms before too long, because that is all they need.
You are getting the underlying assumption wrong: Ubuntu is no desktop operating system!
The fact that you cannot reliably boot your system, if you use any of these services on an upstarted ubuntu should tell a clear enough story: kerberos-ldap , nis , samba, autofs . All of them have triaged and confirmed bugs open at least since maverik, regarding upstart dependencies only. That means, that Ubuntu is essentially completely useless in in any server-client environment since serveral releases. Note, that all these bugs are unassigned, too, just to get a notion of canonical’s priorities.
I consider Ubuntu is a pretty fancy tech-demo for “how to port the OSS ecosystem to a tablet”, and honestly they are doing nothing wrong there. They only need to find a major tablet vendor to buy them out, before google turns it’s office suite into something usable on mobile devices….
Well two points here, a) the Linux on the Tablet question has been decided. It’s Android. There’s not going to be a second one. It takes incredible market strength to push a mobile ecosystem these days, because of the network effects with content and apps. If Canonical thinks they stand a chance at doing that, with or without a vendor, they are plain delusional.
And b): Where do you see any evidence that Ubuntu is making a sensible move towards Touch devices? Unity was created in the brief moment when Netbooks were hyped as the next relevant thing. It’s a bad interface for mouse-driven devices with small screen-real estate. I don’t see anything that makes sense in a touch environment. Nor, frankly, does the established software ecosystem make sense for it either. Touch is a completely different user experience and requires specifically designed interfaces to work well (The same is true in the other direction, incidentally: Apple reverse-ported the look and feel of some of their touch iOS apps to MacOS in Lion, and they annoy the hell out of me).
To me, this is all still Netbook crap. It seems Shuttleworth is the only one who hasn’t noticed these sorry devices were dead the minute Jobs stepped on the stage with the iPad in his hand.
Why were your developers using a technical release (11.04) which is used to test new features and is not guaranteed to be stable, instead of a LTS (Long Term Support) release 10.04 for doing their development work?
It sounds to me like someone didn’t do their homework (i.e. RTFM) before using Ubuntu (it still is a *NIX same rules apply).
I run a network of 20 point of sale systems, 4 back offices and 2 development machines all running 10.04 and they are solid as a rock. My 3 laptops run 11.10 which again you have to do your homework before using Unity. Unity is all about Keyboard shortcuts, once you learn to use the keyboard shortcuts (just as you had to learn the gestures on the OS-X product) then Unity makes sense and is fast and easy to use.
Glad to hear you like Unity. Many people, some of them called Linus Torvalds or Eric S. Raymond don’t and have switched away from Gnome for that reason. Are you sure they are all just too lazy to ‘do their homework’, as you so charmingly put it?
Also, I am sorry to report that from my experience I cannot second your praise of Lucid’s stability. Have a look at the my most recent post, or at the bug reports referenced by commenter #1 above. All confirmed bugs with serious productivity implications that have been unaddressed for considerable time. Likewise, there exist to this day well-known race-conditions in upstart (unresolved at least since hardy) that break system startup every so often in more complex configurations (resulting, e.g., in a host coming up with no NTP peers, because ntpd gets started before the network is up.)