0
\$\begingroup\$

How might one store data for the far future? (100 years and up, possibly centuries)

This is entirely hypothetical but would it be possible for semiconductor based data storage to have similar longevity compared to, for example, indents milled into a slab of metal?

Common NAND flash have rated data retention of somewhere around 20 years and I'm sure environmental factors like temperature and humidity would play a role as to the integrity of the silicon itself.

Assuming pretty reasonable physical conditions, what are some possible solutions to achieve much longer retention? I've heard periodic conditioning of the storage device by rewriting data may increase retention by a good amount. Would, say, an FRAM chip with memory controller that is powered on every month or so while kept in a favorable environment possibly achieve data retention of 100 years and beyond?

\$\endgroup\$
9
  • 1
    \$\begingroup\$ a self correcting system such as DNA can store data for millions of years ... there was also an article about storing data in glass for millions of years \$\endgroup\$ – jsotola Nov 11 '20 at 6:01
  • 3
    \$\begingroup\$ @jsotola Yeah, but it needs to feed and you never know when it will end up building a civilization and nuking itself. \$\endgroup\$ – DKNguyen Nov 11 '20 at 6:03
  • 1
    \$\begingroup\$ @DKNguyen lol ... here is the glass article ... en.m.wikipedia.org/wiki/5D_optical_data_storage ... and the DNA article ... en.m.wikipedia.org/wiki/DNA_digital_data_storage \$\endgroup\$ – jsotola Nov 11 '20 at 6:09
  • 1
    \$\begingroup\$ i5.walmartimages.com/asr/… \$\endgroup\$ – jsotola Nov 11 '20 at 6:13
  • 1
    \$\begingroup\$ You could carve it into rock. \$\endgroup\$ – David Nov 11 '20 at 12:19
3
\$\begingroup\$

If you plan to store something that long, why are you using a rewritable medium at all? Rewriting = reversible = inherently won't last as long. At least go for permanently fusible ROM to store the data which gives you a fighting chance.

And just powering on FRAM (or flash, or any other rewritable medium for that matter) won't do anything. You'd have to power it up, reread and rewrite it all so all that stuff has to survive too, plus power source.

Now, a plain, undoped, silicon wafer, which is just a giant crystal, with something etched into might do just as well as a properly chosen metal alloy with something engraved into it. But to rely on the semiconductor properties? No. Higher complexity = more to go wrong. I believe I read that dopants drift which is a huge Achilles' heel for semiconductors.

But if you do go the semiconductor route, the only thing that even remotely has a fighting chance are permanently fusible links (which in the end is not too different from just permanently etching some gold traces into a silicon depending on how you go about it...for the data storage part at least...not the mechanism to read it out).

Something like a pressed CD seems like it would last a long time...or even better...the glass master that was used to make it. And like permanently fusible links, it only addresses the data storage part, not the data readout mechanism.

\$\endgroup\$
1
  • 1
    \$\begingroup\$ I think I read an article once about whisker growth in Si fuse gaps... \$\endgroup\$ – Wouter van Ooijen Nov 11 '20 at 6:56
1
\$\begingroup\$

Stone. Proven since the Egyptians.

Since most of the datas throughout the ages have been lost to fire, one would develop mediums with high temperature materials with solid mechanical strength.

Cast iron is a very stable material. Less primitive ones are high-temperature ceramics, like those used in aerospace. With these materials one should imagine a micro engraving technology to print binary data on them. For text the analog form used since the Sumerian tablets may be more efficient and easier to read with low tech devices than binary. But only binary can stores reliably non-text data like pictures and sounds.

For a more familiar, DVD-like support, ceramic glass with carbon or metal inserts could be a good compromise between ease of use, cost and working temperature (+- 800C, 1400F).

5 dimensional supports are not high temperature and they will melt in the first fire they will be tossed in by the first inquisitor, just like paper. I don't believe the claim of

thermal stability up to 1,000°C

touted for example here. It may be true for the substrate. Not for the reactive material. If it can be modified by an ultra fast laser, it's likely that it can be also modified by an exposure to fire.

For high density storage, that's nonetheless an excellent progress compared to other mediums.

\$\endgroup\$
0
\$\begingroup\$

How long any storage device lasts obviously depends on how badly it is treated. That can never be guaranteed over a 100 year period. Consequently, there can be no fire-and-forget technology.

You get to a point surprisingly quickly where stable societies, conditioned vaults protected against lightning, flooding, fire and NBC attack, and especially that old bogey of every IT manager, the backup regime, become more significant than the theoretical lifetime of the medium itself. More data gets unintentionally lost through reorganizations and disasters than by natural decay.

Isaac Asimov, who wrote much popular science as well as SF, pointed out that information evaporates. It doesn't matter what medium you store it on, a combination of natural and man-made factors will eventually destroy it (One may compare this with the Buddhist stricture that nothing is permanent, all things must pass). His fundamental solution was to replicate it faster than it evaporates.

I have found this solution to be exceedingly unpopular with IT managers and their bean-counting overseers, who live by fire-and-forget budgeting - which is why it so often comes back down the line to bite them in the backside.

\$\endgroup\$

Not the answer you're looking for? Browse other questions tagged or ask your own question.