This is a difficult problem. About 20 years ago I did my Master's thesis on 
the archival storage of electronic media. My conclusions then were that 
electronic media are great for access and outreach in the short term, but 
not for archival storage. I have seen no changes in the technologies or 
market forces that would cause me to change my conclusion.

The biggest problem is technological obsolescence -- of the hardware to 
read the media, and of the software to translate the media to human-
readable form. New formats replace old formats. Operating systems and their 
associated software become obsolete at an incredible rate.

There is also the problem of the media itself often being fugitive 
(delaminating or deteriorating CD-ROM or DVD-ROM disks, the deterioration 
of magnetic data over time, etc).

Example: How difficult (and expensive) would it be today for you to read a 
VisiCalc spreadsheet saved on an 5.25-inch format floppy disk drive? You 
need a disk drive to read the disk, the data on the disk has to still be 
good, and you need a computer with an operating system that can run a 
compatible version of the Visicalc software in which the spreadsheet was 
created.

The costs of constantly upgrading obsolete technologies is too great to 
rely upon contant upgrading to maintain access to the material. 

Example: How much 1960s-era reel-to-reel videotape has been or will be lost 
to history because there are so few machines that can read it and it is so 
expensive to convert to forms we can read today?

A further consideration is the possibility that social upheaval or collapse 
will elmininate our successor's access to any technology capable of 
translating the data to human-readable form.

My conclusion is that data preservation for archival time periods (hundreds 
of years) requires that the data be in a form that requires little or no 
technological translation to be human-readable. 

* Paper print-outs on acid-free paper
* Sound recordings to grooved phonograph records (a low technology society 
could possibly jury rig a machione to play a grooved record. A machine to 
read magnetic tape? Not likely).
* Audio tapes should be transcribed and printed on acid-free paper.
* Motion pictures (whether originbally shot on video or film) on mylar-
backed silver halide emulsion film. 
* Color motion pictures separated a-la 3-negative Technicolor if color is 
important, or prints using pigment-based color technology such as 
Kodachrome. I think the technology to do this is now obsolete.
* Digital photographs printed with pigment-based inks on archival paper.
* Print out databases, spreadsheets, etc., onto acid-free paper in formats 
and using fonts that can be scanned and OCR'd.

You need a three-phased strategy: 

* Keep and maintain working examples of all machines and software for the 
forms of technology you must be able to access from the original media. 
* Attempt to keep up with technological innovation and have a planned 
process for continually upgrading your electronic data so it is accessible 
with whatever technology is current.
* Convert, wherever possible, the informational content of the electronic 
data to a human-readable form. 

Budgetary concerns will ultimately mean we will lose a huge amount of data.