Regarding ease-of-use (number three on my list), a major problem with Linux is installing third-party applications. Programs which are in the repositories of your distro are very easy to install, in my opinion easier than on Windows. Downloading a separate package file (a .deb or .rpm for example) for your distro is also very easy (this however assumes some technical knowledge: the version number of the distro you're using).

But then there is the case where there is no package available for your distro. Maybe there is a package available for some other distro, which could be converted with alien. However, this is cumbersome, slow, and will sometimes even not install the application correctly.

The worst case is however when none of that is available, and just a third-party installer, or source files you have to compile yourself. Last year, I installed the commercial statistical package SPSS on my Ubuntu 7.10. It didn't work. First after a few months, when Hardy came out, it actually did work.

The compile-it-yourself way is however an option that none of non-technical users can ever do. Clearly, a lot could be improved when it comes to supporting third-party applications.
I hope that the LSB can standardize Linux sufficiently for compatibility across distros. But the program installation process itself needs to be improved a lot. The Burgdorf API (also known as LSB Packaging API) seems very promising. Small open-source projects which don't get enough attention to get native packages in all major distros could quite easily (supposed the right tools are developed) make their own install file.

When I wanted to install NES emulator, the ones found in the Ubuntu repositories didn't have all the features I wanted, so I ended upp downloading and compiling Nestopia. If the Burgdorf API had been in widespread use, they could have made an install file using that, which would have been much simpler for me.

The same would have held true for installing the commercial SPSS program. Had standardization been thorougly implemented, it would have worked on the first try to install it.

So, let's just hope that development of the Burgdorf API proceeds well, and that all distributions will adopt it! Then I think there would just be a matter of time before almost every program will be easy to install on Linux, removing one more of the obstacles for a really user-friendly Linux.

In response to comment on my previous post, let me define what mean with 'The year of the Linux desktop'. I would be something like: Linux is quite widespread on the home desktop market. Apple has around 6-7 %, something like that would suffice. My prediction is that this will take maybe four years (meaing sometimes in 2013). That would mean that Linux would be so easy and useable that (compared to for example Windows XP) a normal, non-technical user can use it with spending a similar amount of time on configuration and system management issues, it would need a similar level of knowledge (note that this can be entirely different *kind* of knowledge than other OS:es, but total computer illitterate user could learn them in a similar amount of time), and that most normal tasks could be done with a similar level of functionality, hassle and ease-of-use.

Currently, this is not the case for the majority of non-technical user.

What are the biggest problems with desktop Linux?

A lot of people has been talking about 'The Year of the Linux Desktop', and that this year it will come. But the years has passed, and the The Year of the Linux Desktop never seems to come. Why is that? What are the main problems with Linux that holds it back from becoming a success?

In my experience, the main problems are the following:
1. Hardware support
2. Multimedia support
3. Ease-of-use / Useability
4. Support for running Windows programs (WINE)

Other people may have a different list, or a different priority of the items in the list. This is just according to me. I will post about these items for some time, describe the problems, give an analysis of why the situation is as it is, and give some suggestions for solutions.