I think if you'll go back and notice, I stated ". . .except for software updates" it has not crashed. . . and when we refer to updates, those are major updates. Minor updates often do not require system restarts. Third party software updates seldom require a system restart as they often do on Windows because software is not allowed to modify the system. On the other hand, we do have a server that has not had a system update in that entire time. . . it does not go on the internet—ever—and happily sits there running OS X.5 Leopard working as a back-up server with a RAID set-up. It's on a heavy duty UPS and has never even gone down in a power failure. Two drives have failed on it. . . but not the boot drive. Perhaps, one of these days, it will. I think it's been six to six and half years since we upgraded it to OS X.5.8 which would have been the last time it was rebooted.
Perhaps because of my long history with Unix, DEC VAX and RSX servers, and high-reliability industrial control systems, I still define "continuous operation" only in terms of "no interruptions of service", and I'm pretty hardline -- for example, restarting Apache to pick up a config file change can be done gracefully and inperceptibly, but it is an interruption of service if anyone's download gets stepped on.
A reboot is an interruption of service lasting typically a few minutes, between the start of service shutdowns and the time when all services are up again. But perhaps more importantly, a reboot reinitializes the system structures, clears caches, all sorts of refreshing cleanup occurs. The test of long-term system stability is whether it keeps going without that refreshment -- that proves there aren't any slow memory leaks, tables that don't have enough space and fill up, etc.
I know I'm being picky here, but my background taught me that counting "continuous operation" and ignoring reboots is just as, ummm, incorrect as counting "continuous fasting" and ignoring snacks. Fasting is fatal; anything less is not fasting, it's partial fasting. The reason RAIDs in HA systems are designed with hot-swap spares is exactly because in some systems any interruption is intolerable. For the smaller computers most of us work with, a reboot is actually a breather. :)
That said, I certainly acknowledge the validity of your description and achievement.