Were the disks really dead? Did he try the disks in just one reader or several? If he still had the writer that wrote them,
did he try them in it? Sometimes heavily used writers go out of alignment, sometimes so severely that disks they write can only be
read in them. In the early days of CDR only certain brands worked with a given writer and CD ROM drives could be just as
problematic.
Was the data actually there in the first place? I've already hinted at this twice and
it is probably the most important
question of all! Did he do a verify after write, or did he just assume that the data was written because no errors occured during
the write?
[a very dangerous assumption!]
Though "burning" a CDR/DVDR is not the dark art that it used to be, it is neither as simple nor reliable as writing to a hard drive.
A lot of users are just writing to disks without doing a verify read afterward and just assuming that the data has been written
because they didn't get a buffer underrun or similar error. Then when they try to retrieve the data later they find it is gone and
think that it has disappeared, when in reality it was never there in the first place!
BOTTOM LINE::
If your software has an option to verify after write, turn it on! If it doesn't, consider buying a seperate
commercial program such as Nero that does have this option.
These stories are very suspicious! Why would their disks go bad
after only 3 years under archival conditions when mine are still readable after 9 years when occasionally stored under less than
ideal conditions? Something is missing from this equation!
Other considerations
There are actually two types of data failure on optical media: unreadability and complete data loss. Unreadability simply means you
cannot read it through normal means. Complete data loss means the data is truly gone and cannot be recovered.
You might not be able to read the disk by simply putting in the drive and looking at it, but there may be ways of recovering the data
just as there are ways of recovering data from a crashed hard drive. Even if no current method can recover the data, it is possible
that disks unreadable today might be readable in the future through methods nobody has even imagined today.
Media obsolescence
A while back there was a documentary about archiving data in the 21st century. It told a cautionary tale about an old archive of
7 track tapes that couldn't be read because no working 7 track tape drive could be found. Ultimately a supply of spare parts was
discovered and a 9 track drive was fitted with 7 track heads. The point they were trying to make was that even if archival estimates
of 70 to 200 years are correct, there may not be a working reader that can read it.
I have always been an advocate of transcribing data to new technologies as they become available but the idea that there will be a
time in the next several hundred years when no working reader will exist for today's disks is a sick fantasy!
7 track drives were an ephemeral (and illogical!) standard quickly replaced by 9 track drives which are still used today. Even a
small university was unlikely to have had a 7 track drive, it certainly never became a consumer item!
Hundreds of millions of drives capable of reading CDRs are still in existence and even DVD drives have reached 9 figures. There are
people who collect working models of "antique" computers, a trend that is likely to increase. Even working 8 inch floppies are still
available!
Look at this way, though consumer data is a recent development, how about consumer audio? Edison's cylindrical recordings can still be
played and turntables capable of playing 78 RPM records are still being manufactured! Projectors for 16mm, 8mm and super 8 home
movies can still be easily obtained as can replacement parts. Even failed formats such as Beta tapes, "laser" disks and 8 track
tapes can still be played. It is ludicrous to suggest that today's disks will outlive the technology that can read them!
Conclusion
Media obsolescence is not a valid concern but can and should be planned for. While estimates of archival life of 70-200 years were
probably not realistic, estimates of 15 to 30 years are probably correct, giving ample time to transcribe to newer media.
Estimates of less than 10 years have already been disproven by actual experience, and anecdotal accounts of much shorter life can be
attributed to poor practices and/or bad media, both of which can and should be avoided.
Over 9 years of experience with optical media has shown it to much more trustworthy than removable magnetic media such as floppies
and data tapes.
