Eric Snowdeal is linking to an article entitled The Digital Dark Age, which argues that most of the data on computers today will be lost, resulting in a gap in the historical record. I doubt things are that serious (and most of what's on computers deserves to be lost anyway), but the archiving issue is an interesting one. Once again transparency and openness are part of the answer: your data is much more likely to survive if it's an open file-format like HTML as opposed to MS Word or PowerPoint.
One side point of the article is worth more thought: impermanence by design in our software and most other creative products. The folks at the Long Now Foundation like to talk about long-term thinking (they're building a clock that will go bong once every 10,000 years!), but we don't need to go that far to stretch our time horizons. The folks at Microsoft scoff at Unix because it's a 30-year-old operating system. Um, perhaps that's a good thing? It's withstood the test of time. It's not the latest fashion, as Doc Searls would say. Why not design things with the long view in mind?
I've been listening to the music of Duke Ellington this evening. In the Duke Ellington centennial broadcast on WKCR, Phil Schaap said something that stuck with me: that 500 years from now people will probably have forgotten about most of the figures of jazz music, but they'll still be listening to the music of Duke Ellington. Heraclitus said you can't step in the same river twice (actually he said "Upon those that step into the same rivers, different and different waters flow"). Yet there are, if you will, rocks and boulders that stand fast in the midst of the rushing waters of time. We call them classics, and they are built to endure. Certainly few of us will create anything that will become a classic. But it's something to aspire to.