Jerome Fine wrote:
long I can expect any files to still be readable? Also, how long
is it likely
that the magneto optical drive will last. I suspect
that the
drives are likely
This is (I hope) the key. 5.25" floppies don't
appear on mainstream PCs anymore. 8" floppies
are still available (but there were IIRC more
varieties than there were with 5.25"). 3.5" floppies
are too small to last too much longer (although
so far many of the attempted replacements
- like the LS120 - are backwards compatible).
I'll try not to mention 3" floppies (plus all the
others I've never come across).
I expect that CD drives will go the same way over
the next five to ten years. Finding one twenty
years from now will be somewht tricky.
"What sort of strategy is recommended to always be
sure that old
files (stored
on old media which can only be read with old drives) be
rescued/copied to newer
media before that is no longer possible?"
Firstly you have to get them into a portable
digital form. For many things that just means
copying them off the old media and archiving
onto current media (i.e. CD or MO or
whatever you happen to like). For some stuff
(boot disks, copy-protected disks etc.) you
have to archive in some kind of image format
that preserves the original characteristics.
(ISTRC reading a paper about some US uni
doing exactly this and inventing an archival
format ... but I've lost the reference).
Once you have made the initial digital backup
your task is much simpler. It would take me
a lot of time to archive all of my floppy
disks (especially if I try to verify that I can
accurately recreate the original media from the
archive copy). But once I've done that, I can
probably fit all of them onto a handful of CDs
(I'm guessing that I can get 500 floppies
at least onto a CD and I doubt that I have
even a few thousand ... so call it
five CDs to be safe). That might take months,
perhaps years of copying and verifying
(cataloguing might be nice too !).
Once it is done I could run off a second
set in a few hours. And a third, and a fourth
...
We'll skip DVD ... not enough of a leap forward
and too hamstrung by Hollywood fo rthe prices
to drop quickly enough for my liking.
Five to ten years from now, when recordable
C3D is commonly available, I can repeat the
process for all my CDs (which by then may
number a few thousand). At 125GB per C3D
I should be able to get at least 175 CDs per
C3D, so that's maybe 10 C3Ds to archive
absolutely every piece of digital data I
will have then.
Again, a second backup for safety will
(hopefully) be quite quick.
At each stage, I can verify digital backup
copies against each other fairly easily
(I'm assuming that data rates will go
up as capacity goes up ... otherwise
I see a bottleneck looming!)
It all depends on how careful I am making the
initial archival copy.
Obviously, every user would like to skip as many
in-between steps
as possible.
But since safety is perhaps more important, where
should the
compromise be made?
Anything I care about now, I burn twice
for myself. I always check the CD-Rs
individually against the original source.
"What standard needs to be used to determine what
files may be of
interest in
50 years, 500 years or even 5000 years?"
That's easy. You keep everything. Even the most
expensive CD-R media is incredibly cheap
compared to the prices three years ago.
Obviously if you have terabytes of data this
may not be so easy. But if you have terabytes
of data it may still be easier to archive
everything (which is likely to be a fairly
mechanical process) rather than spend time
sifting through deciding what to keep.
It is fine to be discussing the technical details of
how to save
files, but if everything
needs to be saved, that creates many difficulties. For
example, if
some software
is being developed, it's rare that non-distribution
files ever see
the light of day, much
less that they get saved beyond the next distribution.
In most
cases, only the final
source and executable files get saved and it is
probably rare for
OBJ and MAP files
to be saved. What is the best way to develop criteria
that can
determine which
files will be of interest to someone looking to
understand how
"programs" were
written 5000 years ago? Better still would be to
attempt to
determine the questions
that will be asked 5000 years from now. Looking back,
I would
suggest that the
first footstep on the Moon back in 1969 may not have
been as
important as the
technology that was needed to make it all possible.
I'm actually at the beginning of archiving some
stuff at work right now. Copying the source directories
(and the version control stuff) to CD in a PC-readable
format is fairly straightforward. Same for the
final listings, obj, exe etc. (although much
of that could be recreated). The bit I think will
be difficult for a historian to understand in
a hundred years is how to build the product
once you have re-created the environment.
Archiving people's knowledge is not currently
on the agenda ... it takes too much time.
Antonio