Ok, the list of demands sofar:

-the technique should be available between now and 10 years

-the media must last for at least 50 years without any deterioration of the data

-it must be able to store at least 50 TB

-it must be small enough to store and transport cheaply, but not so small that its hard to handle or easily lost, say the size of a pocket book

-both the data storage part and the interface may not have any moving parts

-it must be cheap or at least much cheaper than filmstock containing the same amount of data

-it must have the ability to change interface so it can be connected to future day computers and software

-it must be able to withstand damage from kinetic energy, water, heat, cold, magnetism, electricity and light to a certain degree.

Any other demands we can think of?

Well as suggested here the added requirements sofar

-inedible labeling (clay tablets anyone? ), directly linked to the media, clearly visible on at least two sides where the action of labeling will make the media write protected

 Nice list! I would add to that permanent, indelible physical labeling.

Larry Rubin:

permanent, indelible physical labeling.

 Good point, Andrew! I can't tell you how many times I get handed critical camera original tapes with absolutely NO labels whatsoever!  To borrow a line I saw here once, "Excuse me while I go jump out a window......AAAAAAAAAAAAAAAAAAAHHHHHHHHHHHH!!!!.......[Splat].

Larry Rubin:

I get handed critical camera original tapes with absolutely NO labels whatsoever!

Any format developed thus far has a limit to the length of time it can retain its data.  Paper, (mylar) film, (mylar) magnetic tape, magnetic discs, optical discs, non-volatile memory - they all have to be stored and handled well to get any long-term storage out of them.  Even when handled and stored under the best conditions, the data will start to be lost from any of them at some point in time.  Of course, there's also the fact that the ability to "play back" the data from any of them disappears or it becomes much more difficult to find the appropriate "player" as new technologies come along and push the old ones off the market.

The only way I see for real long-term storage is to do what we've been doing for at least the past 150 years if not for a few thousand years: copy the data from older formats onto the newer ones as they become available or at least before the data starts to be lost from the older format on which it's stored.

The Library of Congress and NIST (National Institute of Standards and Technology) have been working on this question for some time.  I first learned about this research a few years ago.  Some of the results of that research are available on their respective websites.  Here are some links: http://www.loc.gov/avconservation/preservation/, http://www.digitalpreservation.gov, http://www.nist.gov/public_affairs/techbeat/tb2004_1208.htm#cddvd, http://www.itl.nist.gov/iad/894.05/digitalPhysicalMediaAndDevice.html, http://www.itl.nist.gov/iad/894.05/publications.html

At this time, I think, the best answer to the question of what's the archive medium of the future is "All of them as they come along."

The BBC has adopted Avid-compatible MJPEG, MXF and LTO data tapes for their archive. To achieve this they've created an opensource project, which also encompases a multi-stream media capture tool to capture Avid media in studio production.

Ingex - http://ingex.sourceforge.net/
Whitepaper - http://www.bbc.co.uk/rd/pubs/whp/whp155.shtml

smyers63:

copy the data from older formats onto the newer ones as they become available or at least before the data starts to be lost from the older format on which it's stored.

Exactly. Having multiple copies in multiple formats is the only way to safeguard any information. Larry, why don't you want to mess with your u-matic tape? Cause you only have one! What ever medium you choose, if you only leave one copy, any future user trying to read that copy will be scared to mess with it.

In my opinion backing up will be not so much a medium choice. It will be a mathematical probability calculation based on desired length of conservation, medium type and characteristics, human error, natural events and last but not least cost.

In reality if we want things to last "for ever" the best thing would be the oldest one out there: to enter the information (digitally) in a piece of stone. Not very practical... Oh and don't forget to make a copy on a second stone.

And just because this thread was dredged from the bowels of the forum - Vinyl (even though never going away completely) is making a come-back.

Article 1
Article 2
Article 3

Bring on 3/4" baby and data punch-card!!!!!

Solopost:

...Vinyl (even though never going away completely) is making a come-back.

One look inside my archive will easily prove that.

As luck would have it, I have a lot of (recorded) vinyl and maybe a few hundred (punched) IBM/Hollerith punch cards from the "old" days.  Oh, and I have an unused box of 8" floppy disks if anyone has a need for them. 

I think it would be easier to find a working keypunch machine and blank cards these days than it would be to find a working home phonograph recorder and blank records (and, yes, I have quite a number of home-recorded phonograph records and maybe one or two blank records in my collection, too).

If we got down to brass tacks, I think we'd find "vinyl" is still used in some of our currently-popular storage/archive media - but in a slightly different form.

Here's a very relevant video package on this subject.