Those with a vested interest in desktop Linux adoption are trying to sell something that still has critical challenges overcome first. Desktop Linux is ready for everyday use, but there are areas that must to be addressed first before seeking any sort of massive mainstream adoption of this in comparison to OS X or Windows. This is a fact backed by bug reports that leave many existing users without any recourse other than to deal with the issues at hand in frustration.
Heres a list of the areas that I believe are in desperate need of attention. As you read this, bear in mind that Im a long time desktop Linux user, not another biased Windows-using pundit.
1. Preinstalled desktop Linux. With very few exceptions, most users will find their only support options entail calling someone or if they know to take this action, seeking help from a local Linux user group. And for the more technology savvy, user forums.
Regardless of what people might like you to believe, the fact remains that there are two very opposite ends of the desktop Linux spectrum, neither of which is suitable for every day use. Two great examples of opposite ends of the spectrum are Ubuntu and Debian.
For the Debian example, I envision a typical PC vendor taking that distribution, then installing it onto a PC to be sold to the public. The buyer will then end up with a stable, ready to use PC. The flip side to this is that Debian's development means that many cutting edge advancements being seen on other Linux distributions or with OS X/Windows will leave this common user feeling totally left out of the technology improvements.
Yes, that user is free to sit down for weeks and learn how to become a Debian expert, thus becoming comfortable enough to dip into the non-stable Debian repos to test out unstable updates, but this is simply not going to happen with most people.
Then you have the Ubuntu mindset: Fast, reckless and always full of bugs that are marked as 'non-critical'. Users who ends up with a PC loaded with this Linux distribution will find themselves being dropped into the very latest (broadly speaking) that the Linux development world has to offer. Every so often, they will run a system update that breaks one setting or another. Ubuntu is a fantastic example of progress, but with the exception of their last long term support release, the distribution is terribly reckless and full of regressions in the code.
Open Source Apps for Small Biz: Desktop to Backup
GNU/Linux Desktop: The Case Against Running Windows Apps
How the GNU/Linux Community Ranks Distros
In Defense and Praise of Debian
Now it should be noted that there are other examples out there using totally unrelated, non-Debian-based distros that do not share many of these problems. Unfortunately, most of these distributions are entirely too hands-on for most users. And some of them, such as Fedora, have bugs of their own when battling uphill with the RPM-based software installation.
Needed solution: For any sort of preinstalled Linux distribution to be taken seriously outside of geek circles, there simply must be a clearer balance between progressive releases and not diving into beta code regression. At this stage in the game, there are just too many unknowns and "work-a-rounds" that keep driving people back to their original operating systems. Worse, even if this suddenly stopped, the prejudice of past experience has already colored the views of this user.
2. Working Wireless options 802.11a/b/g/n/whatever. Two things to consider here. The first is that a preinstalled Linux notebook nearly always solves this issue, when not plagued with regressive code thanks to sloppy updates. And second, this is only an issue for the most part when considering notebook computers. A desktop machine rarely is in a position where finding a dependable wireless card is an issue. The typical NIC works out of the box 99 percent of the time with zero user interaction.
Whats unfortunate is how nearly every single wireless device vendor in the world has bent over backwards to ignore desktop Linux as a worthwhile platform to provide a working wireless driver for. They cite every excuse from "lack of consistency from distro to distro" to the lack of actual physical user numbers to justify putting work into creating a special Linux-only module for casual use.
Regarding user numbers, much of this is hot air, considering there are entirely more desktop Linux users than being reported. Unlike the proprietary operating systems, it is difficult to track something that is freely distributed on mirror servers throughout the world, as there is no discernible point of sale to grab statistical data from.
Needed solution: I think Linux development needs to center around existing companies that do support open source drivers from the beginning. That and make sure that the support focuses on wireless devices that can easily be plugged into Windows notebooks that now have Linux installed as well.
I watch in amazement as developers and users alike spend days trying to get hack-n-wish solutions like NDISWrapper working with Windows-based wireless drivers. By simply putting a level of importance on the chipsets that support Linux out of the box, the end user is then forced to purchase a compatible wireless device and save everyone a lot of headaches down the road. These devices do exist, we just need to see a central repositories of vendors selling them. And no, hopelessly outdated options like this are not working.