From: Adrian Stoness
[M?]iror everything guys make copies and stash
From: Paul Koning
The web can make things perpetual if they are stored
redundantly ...
But anything centralized is just as vulnerable as any centralized copy
ever was, whether from risk of fire or flood, or abandonment.
I've been thinking about this issue for a while (although I tend to have a
long scope, e.g. looking forward to a time when everyone currently on this
list is dead; so I think things like 'failed states' need to be a concern
too), and I think history has a key lesson for us.
I've been reading up on the history of the Greek cities after the
Pelponnesian War, down through the War of the Successors (the Diadochi) after
Alexander the Great died. One book I read said that the only surviving source
for many major periods in this stretch was Diodorus (a Greek historian from
Syracuse in the first century BC); he wrote a history of the world in 40
volumes, only 15 of which survive today complete. The sad thing is that there
_was_ a complete set in the library at Constantinope, as late as 1453 (and we
know what happened then). So it survived the best part of 2K years, and was
then lost; the parts that _did_ survive, did so because there were copies in
other libraries.
So the lesson is clear: we need to _replicate_ stuff, in a geographically and
nationally distributed way.
The mirroring of Bitsavers is _very_ good news. However, even in the class of
stuff that it focuses on, e.g. old manufacturer documentation, some things
don't make it in there, but do exist in other online repositories (e.g.
Manx's collections). So one thing we need to do is come up with something
like Bitsavers, but with more curatorial work-sharing. Al has done an
_incredible_ job, for which we are all deeply in his debt - but it would be
good to come up with some way to help him.
(E.g. I've been adding links to online versions of manuals, in articles on
older DEC stuff I'm doing the CHWiki, and I often find things which aren't in
Bitsavers. But sending Al an email saying 'hey, xxx is {here}, you might want
to upload it' is just putting all the load on him.)
Getting all this stuff into the replicated, mirrored system is a key priority.
And in the case of digital data the added complication
is the loss of
the necessary technology.
Multiple independent copies will of course help with this (very real)
problem. The mirrors will likely be using different hardware, and will turn
it over at different times.
We could definitely use more mirrors, though - and geographically
distributed: it looks like there are current (non-US) ones in the UK, and
in Germany - more would be good. New Zealand? Australia? Maybe Japan and
India?
Individual volunteers aren't really what we need ('when everyone currently on
this list is dead'); it needs to be institutions.
The Long Now Foundation has done some good thinking
about this; some
others have as well.
Jerry Saltzer thought about this, especially the 'generations of hardware',
and 'software formats' (e.g old Word docuents) issues. See:
"Technology, Networks, and the Library of the Year 2000"
http://web.mit.edu/Saltzer/www/publications/inria/inria.pdf
(particularly Section 4.3 "Persistence"), and also:
"Fault-Tolerance in Very Large Archival Systems"
http://web.mit.edu/Saltzer/www/publications/fault-tol/fault-tolerance.pdf
I'd say more of us need to be more paranoid about
mirroring stuff.
Yes. Don't just use a link, copy stuff down to a place _you_ control. (I.e.
not Google Drive. Nothing against Google, but their business might go
somewhere different, like Geocities, etc.) I have a large collection of
down-loaded stuff. Already I've run into cases where stuff has gone offline,
and without my local copy...
Noel