Re: Durability of ISO/IEC 10646-1:2000

From: Markus Kuhn (
Date: Fri Mar 03 2000 - 04:11:47 EST

Rick McGowan wrote on 2000-03-02 23:22 UTC:
> Markus ventured:
> > "Complete ISO Coded Character Set Handbook CD-ROM" [...]
> > for the benefit of the History of Computing PhD student in the
> > year 2280 working on a dissertation about how it all started
> Well, if they wanted to do ANYTHING seriously useful for the PhD student in
> 2280 working on a dissertation about this era, the best thing they could do
> would be to print all that stuff in hard copy on acid-free paper and store
> the bundle at the British Museum in an atmosphere controlled case... By
> 2280, all copies of the CD-ROM will be completely useless and unreadable (if
> they haven't actually disintegrated from within).

On the contrary.

Various CD-ROM manufacturers have kept samples of their first products
in accelerated-aging climate chambers under salty moist high temperature
since the late 1970s and they claim that they are still all readable.
Manufacturers are now confident that if properly stored under not too
horrible conditions, CD-ROMs can easily last 100-200 years, probably
significantly longer if stored a bit carefully (dark, dry, cold, low
vibration, perhaps even oxygen-free) [1]. A few manufacturers discovered
in the late 1980s that their surface printing ink migrates into the disk
and corrodes the aluminium, but this has long ago been fixed by changing
the ink types. Note that this is for CD-ROM (the mass produced ones,
where bits are stored as robust mechanical deformations in an aluminium
foil embedded in a high-quality polymer), not for CD-WO (the ones that
come out of your CD-writer, where bits are stored as colour changes in
a chemical layer instead). For CD-WOs, few experts are at this time
willing to claim more than 30 years lifetime.

The CD-ROM, PDF, and ISO 9660 standards are well documented and so
widely implemented today that it really should be no problem to still
read these artifacts in a couple of hundred years. Don't expect any
executable binary software to be of long-term use. If you have space
left on the CD, fill it with the self-documentation of the medium (i.e,
the CD-ROM, ISO 9660 and PDF 1.3 standards). The only significant data
reliability risk of CD-ROM medium are surface scratches, and those could
be polished away anytime later if someone really wants to get to the
data. So when you prepare CD-ROMs, be aware that these are time capsules
that could become fully readable archeological artifacts one day. Who
knows, may be, the Phaistos Disk was meant to be an early predecessor of
the forthcoming ISO 10646 disk (it would surely make for a neat cover

For comparison: Magnetic media (floppy disks, tapes, etc.) hardly can be
trusted for more than 5-10 years under an office atmosphere (may be
15-20 years in the dry storage room of e.g. a TV station archive). For
lack of evidence, computer archeologists are already debating whether
MS-DOS 1.0 ever existed or was merely a myth, created to make MS-DOS
2.11 look more mature on the market.

When it comes to the long-term preservation of important documents and
specifications, I far prefer to spread a few ten-thousand CD-ROMs over
the surface of the earth as opposed to relying on centralized controlled
storage. The British Museum or the ISO central secretariat burns down or
gets hit by a meteor several times per millenium on avergage after all.
ISO should therefore make it part of their mission, to somehow get every
standard onto at least 10000 sold CD-ROMs, possibly best by bundling
many standards on one inexpensive disk.


  [1] B. Steinbrink: Gibt es ein Mindesthaltbarkeitsdatum für CDs?
       c't Magazin für Computertechnik, Verlag Heinz Heise, Hannover,
       Germany, April 1994, ISSN 0724-8679, p. 62-66. [in German]


Markus G. Kuhn, Computer Laboratory, University of Cambridge, UK
Email: mkuhn at,  WWW: <>

This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:59 EDT