Open source can be a solution, but it's limited by bitrot; what it runs on, and who or what is capable of interpreting it.
I feel there needs to be something else, haven't quite found out what that might be.
 - http://www.edge.org/response-detail/10113
That's the source of my frustration whenever the open source solution is brought up. It only works when there are hands present to maintain the source.
Which is much more effort than, say, writing a book and leaving it some place to be rediscovered. Only the popular or the intensely niche seem to survive.
But more importantly, I think _why's most recent incarnation is a nod to a perceived solution. It's perhaps not 50 million years, but the persistence of the internet and the protocols surrounding it may give us hope that document stores can still be read in 100 or 200 years. Maybe the code wont run, but we have the logic and the ability to decipher the language. Now, I'm just going to go scribble out a Fortran to Python generator so we don't lose all those handy formula translations :)
EDIT: I also bought a house from a former book collector inheriting a great many old books. Aside from a 1787 edition of Virgil's Aeneid, I can tell you, my ultimate frisbee stat collector from high school has an analog in the book world. I'm staring at Oliver Onion's Poor Man's Tapestry, what a world shaker.