The Logic Off Topic list

marc lindahl:

> BTW, the AES (http://www.aes.org) has a working group on archiving and
> restoration.

And we are just about out of my depth and into theirs.  I'm not motivated
enough on the subject to learn the chemistry!
 
>>> Longer lifespan than analog tape....
>> 
>> Hmm.  The problem is that digital degrades so ungracefully.  Shannon's
>> theories allow you to predict how much EC overhead you'll need to insulate
>> the information from the noise of the "noisy channel", but your calculation
>> becomes invalid as the noise in the storage channel increases over time.
> 
> Well, the calculation doesn't become invalid, if you account for degradation
> over time as well.

Nicely said.  If long-term considerations were to be given weight in
research, it ought to be possible to design digital media optimized for
archival.  If you know you'll have spots of a certain size, or if you know
it will degrade at the edges... whatever.  The challenges are primarily
chemical and material; the information theory analysis and EC design is
straightforward.

> if you have multiple copies, the probablity of them all failing
> at the same point decreases exponentially, and since it's digital, as long
> as you have a known good copy of every sample, you can perfectly reconstruct
> the whole.  

I thought a little more about this.  If all the copies are created at the
same time, and they all experience a crash-knee type degradation around the
"expiration" date, then all copies might be completely unrecoverable.

-- Marvin Humphrey
Mastering Engineer and Graphic Designer, emeritus
CD design website - http://marvin.mrtoads.com