Most, if not all, the studies of CD permanence have come out of the
industry, where "archival" may be defined as 5 years or less -- likely
the explanation for the claim that the current generation of disks meet
archival requirements.  With the exception of a gold-coated glass version
(Centuiry Disk) that is relatively expensive and seldom used, there seems
to be consensus that the life expectancy of CDs ranges from 5-20 years
under "average" conditions.  I can't point to a published research study,
but the National Media Laboratory (POB 33015, St. Paul, MN 55133) is
engaged in testing of related materials.
 
The problem with CD as a long-term format is not so much the
vulnerabilities of the physical object (although there are some --
moisture and surface sensitivity to pressure, as from a writing
instrument appear to be among the worst) as the speed of technology
change.  If an institution commits to the medium, it must also commit to
frequent hardware upgrades.  I've seen at least one estimate of a 3-5 year
cycle of technology change through the beginning of the 21st century.
The institution must also commit itself to a systematic program of
collections management including sampling, refreshment (copying within a
format generation), and migration (copying from one generation to the
next).  If the institution is generating its own CDs, it must also
maintain extensive records of programming, electronic architecture, and
disk contents to permit long-term retrieval of disk contents.
 
NEDCC has an excellent free handout on these consideration written by Paul
Conway, head of Yale's preservation programs.  If you'd like a copy, send
your physical mail address to me with a request.
 
_________________________________
 
Karen Motylewski                                508-470-1010
Northeast Document Conservation Center          508-475-6021 fax
100 Brickstone Square                           <[log in to unmask]>
Andover, MA 01810                               Use KM in subject field