Journal Short Circuit's Journal: Linux stability 13

I've got authorized shell access to three Linux computers on the Internet, soon to be four. One of those machines is a production system, one is semi-production (printer and fileserver at a friend's house), and the other two are personal machines. I've got partial admin access on the production machine (I set up the guy's spam filtering.), and regular user accounts on the semi-production and one personal-use machine, and I currently give administrative advice on the second personal-use machine.

Now, Linux is supposed to be pretty stable, right? You're not supposed to have to reboot every week, or have massive data loss resulting from power loss or hardware failures. I've run Debian GNU/Linux on my computers since 1999, and can only recall one time when a hardware failure led to data loss, on an old machine with a 6.4GB hard drive. My current machine has gone through various upgrades since 2000, but my home directory has remained intact through the period, save for a couple tens of megabytes of accidentally rm'd files. (Out of an estimated 100 GB+, that's not bad. And I've got disk space to grow to 310GB, if I switch to a LiveCD with a HD-based homedir. Or an NFS homedir solution and a small drive on the main machine.) I've run parted and qtparted on ext3 and fat filesystems without losing data. I've played with adding and removing partitions without the luxury of a backup. (Where am I going to put 100GB of data, with a total capacity of 310GB spanned between one each of 20GB, 40GB and 250GB drives, and the data spread across all three?)

But I've still only lost a few tens of megabytes of data. And nothing I couldn't reproduce by download or a script.

But two out of three systems I have access to frequently get rebooted, one of them keeps getting its Linux flavor switched, and one of them seems to go through cyclical hardware failure. What gives? Why does my Linux experience have to be so different from most other people I know?