Somehow I was reminded of the bit-rot problem by this site about the Archimedes Palimpsest.

The background: Several treatises by Archimedes were copied onto parchment during the heyday of the Byzantine Empire, about one thousand years ago. A couple of centuries later, the parchment was reused and made into a prayer book. But faint images of the original transcription of Archimedes' work, including the only known version of one treatise in the original Greek and the only known versions in any language of two others, remained beneath the prayers.

For the last decade, restoration and imaging work has been underway at the Walters Art Museum in Baltimore. A trove of information on the project, including several videos, has now been released online; you can see it at the project site linked in the first paragraph.

Beyond the inherent coolness of rediscovering an ancient manuscript with a history worthy of Indiana Jones, beyond the importance of Archimedes himself ("Give me a place to stand on, and I will move the Earth") to our understanding of math and science, there's a point to this as up-to-date as the phone call I just got pitching a cloud-based storage system.

That's the point Vint Cerf makes in the very first link in this post -- we have an overwhelming amount of digital data, with more on the way every second, and yet preserving it is an issue. It's not a new problem, as the tale of the Palimpsest makes clear, but it's a pressing one.