This is the kind of stuff I find really interesting. In 1999 the New York Times Magazine ran a contest for the best idea for a time capsule that could store all the existing content published so far by said magazine and have it be preserved for at least a century. Jaron Lanier (who's no crackpot) proposed that the data be digitized and then stored in the DNA of cockroaches. Oblivio has summarized the process excellently. (That man knows how to write.) Lanier's original proposal is no longer online, but you can still read the archived version, appropriately enough, through archive.org.
I'm sold! This is a great idea. The concept has been around for years in science fiction (mostly notably in a 1993 episode of Star Trek TNG, but I didn't know it was actually feasible in the near future or that in could be done safely. When Lanier proposed the idea in 1999, the technology was still out of reach, but he correctly foresaw that it soon would be. Today the only thing that is lacking is a sequencing of the cockroach genome. However, it would be feasible and affordable to store information in the human genome if we were so inclined.
The New York Times Magazine would not be at the top of my list of things to preserve for posterity, but there are plenty of other applications. Since we're now living in what will come to be known as the Digital Dark Ages, this may turn out to be an effective way of preserving some of the more essential information. Also, if civilization comes to an end through nuclear war or unstoppable nanobots or some such thing, it would be nice to have an additional way of of sending the information to our descendants. We may also want to poke around a bit in the various genomes to see if anyone (Atlanteans, aliens, time travelers, God, etc.) has sent us any cool messages.
DNA is a neat storage medium, because each bit can hold four values, as opposed to only two in binary. So if I wanted to encode "tvindy" in binary, it would look like this:
011101000111011001101001011011100110010001111001
Whereas a possible DNA encoding would look like this:
TCTATCTGTGGTTGCGTGTATCGT
I used the following encoding scheme:
00 = adenine-thymine = A
01 = thymine-adenine = T
10 = guanine-cytosine = G
11 = cytosine-guanine = C
Anyway, the point is that the binary sequence is 48 bits long, and the corresponding DNA sequence is only 24 bits long. Cool huh?
Oh, and the cockroach picture is the work of David Kha.