How to Hold a World of Tweets
The U.S. Library of Congress is blazing a trail in determining how to store an ever-expanding trove of information that never had physical form.
After the Library of Congress announced last week that it had acquired the rights to Twitter’s entire archive for preservation as a forever testament to how Americans lived and communicated in the early 21st century, the Idea Lobby called up to ask if we could come by and see, well, where they were going to put it. Would tweets be housed amid Thomas Jefferson’s 6,487 tomes? Or on the shelf beneath the Federalist Papers?
“There isn’t really much to see,” conceded Beth Dulabahn, the library’s director of integration management. Sure, there’s a server that physically houses the LoC’s digital collection, although it looks pretty much like any other server.
But, on both a practical and philosophical level, micro-blogging’s arrival inside the nation’s oldest federal cultural institution raises a complex question, one Dulabahn and her colleague Martha Anderson happily accepted our inquiry to discuss.
How exactly do you preserve, archive and house for posterity things that don’t physically exist?
The library has been mulling this question since 2000, when Congress created the National Digital Information Infrastructure and Preservation Program, which Anderson directs.
The library first began working in digital (and in other non-paper media like home movies) in the ’90s. At the time, it was trying to digitize physical objects unique to the LoC, like the Federalist Papers, to make them more accessible to people who weren’t old enough to get into the reading rooms or who lived too far away from Washington to visit. The material was then distributed by videodiscs to remote locations like schools.
“There was serious consideration about putting this stuff out on CDs,” Anderson said. “I just kind of cringe when I think about that – by this time, we would have millions of CDs.”
Around 1996, it dawned on the LoC that it could archive and share the artifacts over the then little-used Internet. Then came a dramatic shift in information outside the library: As the Web exploded in popularity, the LoC’s initiative shifted from digitizing physical objects to capturing born-digital objects that never existed in the physical in the first place.
“We’ve been reminded so much the last few years of all the things that didn’t even exist when our program began in 2000,” Anderson said. “There were no Google maps.”
“There was no Google!” Dulabahn added.
“There was no YouTube of course, there was no Facebook, no Twitter, no Flickr,” Anderson said. “All these things that are just commonplace today did not even exist when our program began. And we thought we had challenges then.”
The library spent the better part of a year trying to figure out how to capture video embedded in websites. Then, suddenly, came YouTube, and overnight the staff had to start all over again relearning the new universal format.
The library’s mandate — to chronicle information in the present to be available in perpetuity — bumps up against the very nature of the dynamic Internet. Digital photos today are created as JPEGs, but what happens when that format becomes obsolete, and 10 years from now, your computer can’t read it? If the library simultaneously preserves software such as a current JPEG reader, does it also need to preserve hardware, like a 2010 MacBook Pro?
And even if the library successfully captures everything — navigable links, text, images — on a website, what happens when that site is updated two hours, or two years, later?
“How often do you go out and try to grab them? What’s enough?” Dulabahn asked. “There are a lot of philosophical questions about, well, what are you actually trying to do? Are you trying to get blow-by-blow, or is once a month enough? You can’t make those decisions across the board; it’s not one size fits all.”
The LoC has focused on sites important for public policy, such as those related to elections and government, or major historic events like Sept. 11 or Pope John Paul II‘s death and succession. (Click here to tour WhiteHouse.gov on March 13, 2003, on the eve of the Iraq war.) Given its limited scope, and even with the coordinated efforts of national libraries across the globe, only a fraction of 1 percent of Web history must have been captured by now, right?
“A fraction at what point in time?” Dulabahn countered, reinforcing the impression that any true understanding of the topic might require a grasp of quantum mechanics.
In more measurable terms, it took the LoC two years to accumulate its first terabyte of Web data. In 2005, it began collecting a terabyte a month. Last spring, that became a terabyte a week. This year, Anderson expects to double that rate. The library isn’t necessarily collecting more websites; rather, the same number of sites today yields exponentially more content.
In total, the collection last year surpassed 3 billion individual “objects” like text files or images – “which was kind of horrifying to us,” Anderson said.
Because how do you sort through all that information to make it useful, and unbundle it from the format restrictions of this moment in time? All the typical standards for how to organize books don’t apply.
To a certain extent, the LoC and its numerous partner organizations are just trying to grab as much content as they can before it disappears, and they hope one day someone will know how to use it.
“This is a reversal of a model we’ve had for decades at the library, where basically we select at the front end,” Dulabahn said. “What’s left you catalogue, and you bind, and you move to the shelves. With this kind of voluminous digital content — like Web archives, things like Twitter archives — it’s a completely different model. You have a chance to get it; it’s probably your only chance looking across the spectrum of time. You have a very small window, so you get it and say, ‘Then we’ll figure out over time how to find the good stuff.’”
But just as they’re betting on future innovation in digital archiving — Web crawlers that can tackle Facebook and search functions that can index time — so too will digital information continue to evolve faster than its would-be preservers can keep up.
Twitter has created a revolution in communication that Dulabahn and Anderson could never have predicted 10 years ago. And so it’s possible something so new will arise in the future that it makes all of the LoC’s efforts at digital preservation obsolete.
“But I want to say this word of encouragement,” Anderson said. “Early video games are still alive — Pacman, those early Donkey Kongs — because people have cared enough about them to try to do something to keep them alive.”
People will also be motivated by more than just nostalgia as businesses become further invested in the enormous new marketplace of digital information.
Predicted Dulabahn: “I think we’ll be able to ride on the coattails of that.”