In this blog, I have spent quite a bit of time (here and here) talking about the Linux desktop chicken and the egg problem: there are not enough people for application vendors to justify the port to Linux on the desktop; however, users are not willing to move to desktop Linux until it supports the applications they require. I have also talked about the lack of vendor driver support that would allow users to plug and play with any device they happen to buy at the local electronics store (scanners, printers, digital cameras, MP3 players, etc.) The human tendency to resist change is another factor slowing the growth of the Linux desktop.
In an interview with Michael Dell, he brings up another excellent point specific to OEM sales of Linux on the desktop. Right now, the Linux desktop market is highly fragmented with literally hundreds of distributions that could be selected for installation on a new PC. If Dell or any other OEM picks just a couple to install (Red Hat & Novell), people who wanted a different distribution (Mandriva or Ubuntu) will choose not to buy one of Dell’s Linux desktops. If they let users choose any one of the many Linux desktops, the OEM can end up with a support nightmare trying to provide user support for too many Linux desktop configurations.
In most fragmented markets, as the market matures, several winners tend to emerge to narrow the list of popular choices. Until a few winners emerge, we should expect to see resistance from OEMs like Dell; however, if a few winners do emerge, it will become much easier for OEMs to support Linux on the desktop.