[ih] archival quality, was Internet-history Digest

[Robert Brockway]

On Sat, 8 May 2021, John Day via Internet-history wrote:

> Of course it is survival bias, that is the point.
>
> A lot of was destroyed, burned, left to rot, throw away, etc. Some old 
> texts that were to be suppressed (or just considered old) were turned 
> into bookbindings of other books. (Yes, old texts have been recovered 
> this way.)

I used to do a presentation on backups and disaster recovery.  I'd start 
by talking about the Medieval period and make the point that backups have 
nothing to do with technology beyond the fact that technology is used to 
facilitate the backups.

Monks in Medieval Europe and the Middle East made backups (physical 
copies) of important data (texts) and made sure it was stored in off-site 
locations (other monastaries).  Thanks to them many ancient texts, many 
predating the medieval period, survive to the modern day.

There are some fun stories such as one text that was completely lost in 
Europe.  Centuries later Europeans discovered it was widely distributed in 
Ethiopia.

And then of course we have archival storage locations such as Nag Hammadi.

> The whole point is that even with all our efforts, some of it will be 
> lost. HP thought they had a very nice safe archive facility for their 
> corporate papers . . . until it was burned to the ground. We have to go 
> to great lengths to ensure that that probability is low.

Those who forget the past are condemned to repeat it.

I've long argued that generalised solutions are best for backups and 
disaster recovery and that backups should be offline, offsite and tested.

The engineering concept of "failure of imagination" still hasn't been 
widely adopted in IT.  It's necessary to build backup solutions that will 
survive the problems we didn't think of.

> We need something that if left alone for a very long time will still be 
> decipherable.

Yes this is quite challenging and involves both understanding the data 
format and then understanding the information contained.

This is also a good time to mention the Clock of the Long Now, a project 
to create a clock that will keep time accurately for 10,000 years.  They 
are having to consider some of the same problems that we would in 
addressing long-term data recoverability.

Cheers,

Rob