Slashdot Mirror
Billion Year Storage Media
Thorfinn.au writes "Even though the data density of digital information storage has increased tremendously over the last few decades, the data longevity is limited to only a few decades. If we want to preserve anything about the human race which can outlast the human race itself, we require a data storage medium designed to last for 1 million to 1 billion years. In this paper a medium is investigated consisting of tungsten encapsulated by silicon nitride which, according to elevated temperature tests, will last for well over the suggested time."
But as we know, you can't trust atoms.
They make up everything.
That's nice and all but can we trust our data formats to stay static for that long? Having the data but being unable to open it seems rather useless to me.
Most of our data are totally uninteresting pieces of garbage. Think of it, a future species recovers an archive of present tweets and facebook comments. They will think that we died out because we were egocentric egoistic maniacs who do not care about their future and legacy. Furthermore, they will see it as direct evidence that we preserved nonsense about our pity lives in a super material, while other knowledge was not stored at all. But maybe, they just come up with the idea that the data must be somewhat scrambled, as it makes no sense at all.
Wow, a slashdot article with a straight-up link to the paper. No multi page article with embedded flash ads, no 'science journalism' minced down through a chain of successively dumber news outlets, no PR bullshit. Just the paper.
Submitter, I'm impressed.
The authors describe a medium that will hold information for 1million to 1 billion years, yet they publish their results on PAPER!
Either they don't trust their own material will last as long as good old paper or they expect irrelevance to do its work faster than wear and tear.
Otherwise, they would publish a "tungsten encapsulated by silicon nitride", not a "paper".
... if much of the world's tungsten ore was laced with silicon nitride "contaminants". Alexandria all over again.
Of course you can, it just requires inventing a warp drive or a hyperspace drive or whatever your FTL of choice is...
... a high-speed object collides with it. Because on a billion-year timescale the universe is a shooting gallery and everything is a target.
... there is something of value to preserve. i know, all the knowledge and that. but, honestly, looking at past and present, if i'd want to build a civilization in some billion years, i'd rather start from scratch. don't spoil them. humanity, what a troll!
Maybe its different when bonded to tungsten, but silicon nitride by itself is extremely brittle, almost as brittle as glass.
Modern natural gas furnaces use silicone nitride hot surface igniters (glows red hot and ignites the gas). These igniters will shatter when dropped as little as 1 foot onto concrete.
As in what quantities of data are they talking about? Historically, there is the Bible with related similar tomes like the Torah (essentially the Old Testament with some minor differences), and the Koran. Perhaps something more recent would be the long-term archiving of something like Wikipedia or some other similar comprehensive encyclopedia. In terms of data sizes, think somewhere on the range of about a GiB to TiB.
Linear Script A is still not decoded - though apparently related to Linear Script B (which has been decoded) it is still not translatable.
There are others - http://www.omniglot.com/writing/undeciphered.htm
The problem is not exactly solvable. All translated texts in existence have something related to base the translation on. The Egyptian Hieroglyphs were untranslatable, until the Rosetta stone provided a sequence of texts. Two were already known, which matched in their translations - thus implying that the unknown third was the same text in that language (http://en.wikipedia.org/wiki/Rosetta_Stone).
All fine to have a storage medium that lasts a million years. How about the drive to read it?
My wife did her thesis on the subject of long term data preservation.
http://explorer.cyberstreet.com/CET4970H-Peterson-Thesis.pdf
The entire concept of storing data for a billion years is nothing but ego. It would be akin to our finding a cave with forty-five thousand little paintings of dots, squares and circles - all perfectly preserved. What the hell does it mean? Curious and interesting to speculate on perhaps, but data? Not so much.
We've been historically terrible at deciphering ancient languages without something to help link it to a current language (such as the Rosetta Stone).
All this talk of data formats spanks of a very digital future, which I think we have a very hard time of predicting. The linked article is very binary... the grooves they explain can have "two or more" readable states, and their use of a QR code is interesting since it's an analog representation of an absurdly hard to decipher technology (without a key, as parent indicates should be the first thing). How would we encode data on these things? ASCII encoded English? Aliens would have to decode a language and then translate it. There's got to be something easier.
At least the QR code is ultimately a 2D picture, though. I'd imagine any thorough storage over that period of time will have to start with something extremely basic. Sculptures or 2D visual instructions that clearly lay things out. I think you could probably describe a mathematical encoding mechanism visually, but a language would take some work. The Arecibo message is somewhat famous for being a digital message that is notoriously difficult to interpret, and that's by people who would actually recognize some of the glyphs. The picture attached to the 1970s Pioneer vessels is higher resolution and easier to identify, and the audio/visual nature of the Voyager Golden Record is also interesting. But still the idea that these will be intelligently deciphered by themselves is tiny.
It's impressive that they're building something to last... they're just going to have to spend a lot of time figuring out what to put on it. Should lead to some interesting conversations.
... at least what the article sort of suggests. I have never seen that group being called a bunch of short-term thinkers before.
That and the concept of DNA storage of massive volumes of information sounds particularly epic. It would be incredible to think you could do something like store the contents of Wikipedia inside of the DNA of a redwood tree. The very thought that an organism could be used in such a way to preserve information is by itself something very interesting to consider... and something that is currently at the threshold of being done.
I'm really quite impressed with the level of thinking that this author has gone in terms of really getting into the grips of what it means for long term data storage systems. I'm sure other ideas will surface too, but this really is a pretty serious issue that has some tragic consequences in the past that need to be remembered too. The Aztecs and Mayans had huge libraries of writings in gold codices that the Spanish thought nothing more than to melt down into bullion bars... where even a fairly large percentage of those bars never even made the trip across the Atlantic Ocean (instead are on the bottom of that ocean) to get deposited into the royal treasury. Similar experiences happened with some things in Egyptian pyramids too. Making something durable yet making the information more valuable than the medium that is holding the data is a rather significant challenge.
It worked for Chairface Chippendale.
http://www.worth1000.com/entries/375722/copycat-chairface-chippendale
I haven't thunk up a cool sig yet.
Sure you can... It just takes a mirror in the right place...
"File to fit, pound to insert, paint to match" - Aircraft Maintenance 101
If you were to preserve this for the next species that evolves here to find, where should you store it?
If you make it easy to find and retrieve, then you run the risk of a primitive culture destroying it as heretical once it's decoded. That risk still exists today.
If you hide it, it may never be found.
Monoliths on the moon are the only thing I can think of at the moment.
Sometimes I wonder if we will ever stumble across a one billion year old time capsule from a sentient species that previously lived on this planet. It's safe to say that sufficient time would erase any trace of even an advanced civilization with the exception of anything that was purposely preserved. Yes, that would be cool.
Brought to you by Carl's Junior.
"Actually, a government bureau is the nearest thing to eternal life we'll ever see on this earth."
- Ronald Reagan
Prove anything by multiplying Huge Number times Tiny Number
The human race isn't going to last a million years!
Optimist!
"File to fit, pound to insert, paint to match" - Aircraft Maintenance 101
Excellent thesis and a most delightful dedication!
A few salient points from this thesis, for the Slashdot crowd: ... you must properly catalog things ... they must be properly rendered
- Accumulation: knowing what to keep and what to toss
- Distribution: where/how to keep copies
- Digital stewardship: maintaining objects isn't enough
- Long term access means more than just saving bits
Convolved on this are problems with copyright, fair use, payment for archives, orphaned collections...
Then there's the cost of creating and maintaining a long term digital repository.
Librarians have done a terrific job with our printed archives. Who will become our digital librarians?
Accounts of trade exchanges, wrehouse contents etc. Generally boring unless you are writing an economics paper.
And who would understand it? I have trouble with Shakespeare's iambic pentameter. Do you think the descendants of humans or some alien race will understand a Slashdot archive of "In Soviet Russia" quips?
Have gnu, will travel.
Last year George Church and colleagues published a paper in Science describing data storage using DNA (Church, Gao, and Kosuri. 2012. Next-Generation Digital Information Storage in DNA. Science 337: 1628. doi:10.1126/science.1226355) . While perhaps not lasting billions of years, given that we've been able to read DNA from creatures that existed millenia ago (whose DNA was definitely stored in non-ideal conditions), DNA data storage could potentially preserve data for very long periods of time.
You mean like "A Canticle for Leibowitz" ... joy.
Archaeology demonstrates that survival over long periods of time is quite random and rare, and does not correlate well to the intent of the creator to preserve the creation for long periods of time. There are always unanticipated threats to the existence of these artifacts: war, natural disasters, rot, rust, erosion, language obsolescence, to name a few. The longer the time period, the more likely that some catastrophe will befall any given artifact.
Works that have survived for millennia tend to be items that were copied prolifically. A few of the many copies or items survive the ravages of time, but not because the creators anticipated all of the things that could destroy their work.
For example, none of the original manuscripts of the Bible have been found, as far as we know. But because those manuscripts were copied and translated so often, we have reasonably accurate copies of those original texts.
A million years from now, nothing much will be left of these new storage media. They will only survive if people in the future consider the information important enough to copy it to new media, and translate it into the new formats of the time.
Works especially well if it's something that pisses her off.
We're the only animals in history that knowingly practice birth control so there's hope for us yet I think.
Not really, there are a lot of animals out there that do this. Dolphins have Sex for fun, and a lot of smaller groups have passive systems that make them fertile only when their living conditions can support offspring (which is better than what we have now as humans).
'...if only "Jumping to a Conclusion" was an event in the Olympics.'
Its all very well storing it ON something for a billion years, but thats no use if that something is subducted into the earths crust in that time period of which there is a close to 100% chance it will happen. So storing it on earth is pointless , where else? Space? Nope. Any satellite would suffer orbital decay or be dragged off into the sun or some other body long before a billion years had passed and who the hell could find it even if it didn't?
Its a nice intellectual exercise but ultimately futile. In 1 billion years humanity and in fact all the rest of life on earth will be history because earth by then will be a boiling hell due to the sun heating up. We should just except that and make the most of the time we're here.
Ok, if it's for aliens, then maybe you need tungsten, and spread about a billion of the things around in the vain hope one of them will actually be found against all odds.
But if it's for humans then consider. There are two scenarios:
1) There is a global catastrophe or mass insanity of such proportions that all trillion of the penny-sized server computers of the near future which each have enough storage to store significant percentages of our data as a whole are wiped out, along with all of the electricity infrastructure and the instructions on how to build more electricity infrastructure.
In that case I submit that the 10 of us who temporarily survived the vancouver-island-sized meteor hit have more problems to deal with than resurrecting our facebook profiles and physics e-textbooks. Maybe something which told you which varieties of cockroaches were not poisonous to eat would be handy mind you.
2) There is no disaster of that scale, and a semblance of our current civilization survives somewhere, ergo, we still have a chunk of Internet somewhere, and a substantial chunk of all its content to date.
Notice that in scenario 2, the key issue is that we've finally learned how to back up our data and adequately distribute rendundant copies of it around the world, a process that should pretty much be automatic. Making sure we keep doing that is one of the most important things we can do to preserve digi-culture. It's really the only major thing we have to learn how to do and keep doing to save information for countless generations.
Where are we going and why are we in a handbasket?
Its called fossils.
Wow! Wow! Wow! Really? We have found "tungsten encapsulated by silicon nitride" fossils?! Those must have been the coolest dinosaurs ever!
A million years *does* seem like overkill. OTOH, for many purposes 1,000 years doesn't seem long enough. Say it needs to be long enough for Mickey Mouse to get out of copyright. Estimate 5,000 years, but a bit longer wouldn't hurt.
But the problem is being able to read it even if you have it preserved. I ran into that problem in less than 2 decades with 200 BPI Even parity mag tape. (Granted, the tapes probably weren't good any longer, but there wasn't anyplace accessible where I could even try.) Binary tape would have been a worse problem, however, because I would have needed an IBM 7094 to interpret them. Like row binary was worse than column binary on punched cards. (Ever wonder why punched cards were 72 + 8 columns rather than just 80? In row binary you got two 36 bit words per row per card, and columns 73 through 80 were for sequence numbers, in case you dropped the deck.)
So say we go through a few decades where they laws prohibit copying for much longer than the durability of the storage media (enforced by DRM, of course, as well as by lawyers and police). Civilization doesn't need to collapse in order to cause near total loss of records. Popular interest waxes and wanes, so illegal copying can't be depended upon as a way around this. It may extend the life of some records, but not even as much as three times the durability. (Most things are only popular for at most a year, with VERY rare exceptions.)
Disney used to figure on releasing their movies once every seven years, so figure that long on how long it takes something popular to sink into obscurity. Which means that illegal copying will only extend the lifetime of a work by about seven years. 21 is being extremely generous. That's not very long measured against even current history. And studios, e.g., have a long history of destroying works that they don't feel are worth reissuing. (Not making them public, destroying them.)
So 5,000 years seems like a good goal. That doesn't require tungsten+silicon nitride. But it does require that the material be stored without DRM and without compression. Even then if the mechanism has been forgotten, it will be difficult to read.
Preservation requires multiple copies as well as durability. And it requires durability as well as multiple copies (unless you are saying something like "This is how you metabolize oxygen".).
To me this sounds like a stabilized matrix that has cells that are either filled or left empty. The cells could be quite small, and it could be written by a specially adapted ink-jet printer. But you would want to fill the cells with something that was both stable and contrasting. Possibly a carbon matrix with a titanium white filler. This wouldn't survive high temperatures, but it could be quite cheap to produce, so you could have multiple copies. I'm thinking of something sort of like a high-tech paper that could be read by a microscope. An adaption of the microfiche idea, but using something a lot more stable than film, a lot cheaper than film, and with a binary coding rather than photographic images. I envision it as being "normally" processed by somthing vaguely like a cross between paper tape readers/writers and mag tape readers/writers. It would NOT be a cheap medium to have the ability to process, but it would be cheap to process once you had the equipment. And it would be intended just about solely for archival purposes.
OTOH, there's nothing that says something similar but easier to handle couldn't be common. It could store information quite densely, though probably only around 2^20 bits/square inch. (Note that tha's only 1024 bits/linear inch, which is a lot less than current mag tape. But I'm using area rather than just linear measure. The difficulties would be that this wouldn't be a mass-market product, so cost would be an issue.) Perhaps a bit less. (The dots would be quite small, after all.) I'm envisioning a matrix of carbon nanotubes as the basis, which makes it rather strong. But it also needs to be handleable and manufacturable...which that isn't, yet.
I think we've pushed this "anyone can grow up to be president" thing too far.
Punching a vanadium tape and storing it on Charon would probably work well.
I'm thinking of cells considerably larger than that. That couldn't be written with an adapted ink-jet mechanism. Think more of something about the size of a skin cell to represent a bit.
I think we've pushed this "anyone can grow up to be president" thing too far.
Better find a place that is likely to not be munged by plate tectonic cycles on average a quarter billion years, for the Earth. I would suggest planning the data repository to reside on the Moon or Mars, some place that doesn't change much in a billion years, I have thought about this off and on and realize that a safe thing to do is to put the archive somewhere in space.