Brewster Kahle, Victoria Stodden, and Richard Cox are on a panel, chaired by the National Archive’s Director of Litigation Jason Baron. The conference is being put on by Princeton’s Center Internet for Technology Policy.
NOTE: Live-blogging. Getting things wrong. Missing points. Omitting key information. Introducing artificial choppiness. Over-emphasizing small matters. Paraphrasing badly. Not running a spellpchecker. Mangling other people’s ideas and words. You are warned, people.
Brewster goes first. He’s going to talk about “public policy in the age of digital reproduction.” “We are in a jam,” he says, because of how we have viewed our world as our tech has change. Brewster founded the Internet Archive, a non-profit library. The aim is to make freely accessible everything ever published, from the Sumerian texts on. “Everyone everywhere ought to have access to it” — that’s a challenge worthy of our generation, he says.
He says the time is ripe for this. The Internet is becoming ubiquitous. If there aren’t laptops, there are Internet cafes. And there a mobiles. Plus, storage is getting cheaper and smaller. You can record “100 channel years” of HD TV in a petabyte for about $200,000, and store it in a small cabinet. For about $1,200, you could store all of the text in the Library of Congress. Google’s copy of the WWW is about a petabyte. The WayBack machine uses 3 petabytes, and has about 150 billion pages. It’s used by 1.5M/day. A small organization, like the Internet Archive, can take this task on.
This archive is dynamic, he says. The average Web page has 15 links. The average Web page changes every 100 days.
There are downsides to the archive. E.g., the WayBack Machine gets used to enable lawsuits. We don’t want people to pull out of the public sphere. “Get archived, go to jail,” is not a useful headline. Brewster says that they once got an FBI letter asking for info, which they successfully fought (via the EFF). The Archive gets lots of lawyer letters. They get about 50 requests per week to have material taken out of the Archive. Rarely do people ask for other people’s stuff to be taken down. Once, the Scientologists wanted some copyright-infringing material taken down from someone else’s archived site; the Archive finally agreed to this. The Archive held a conference and came up with Oakland Archive Policy for issues such as these.
Brewster points out that John Postel’s taxonomy is sticking: .com, .org, .gov, .edu, .mil … Perhaps we need separate policies for each of these, he says. And how do we take policy ideas and make them effective? E.g., if you put up a robots.txt exclusion, you will nevertheless get spidered by lots of people.
“We can build the Library of Alexandria,” he concludes, “but it might be problematic.”
Q: I’ve heard people say they don’t need to archive their sites because you will.
A: Please archive your own. More copies make us safe.
Q: What do you think about the Right to Oblivion movement that says that some types of content we want to self-destruct on some schedule, e.g. Facebook.
A: I have no idea. It’s really tough. Personal info is so damn useful. I wish we could keep our computers from being used against us in court; if we defined the 5th amendment so that who “we” are included our computers…
Richard Cox says if you gold, you know about info overload. It used to be that you had one choice of golf ball, Top-Flite. Now they have twenty varieties.
Archives are full of stories waiting to be told, he says. “When I think about Big Data…most archivists would think we’re talking about being science, corporate world, and government.” Most archivists work in small cultural, public institutions. Richard is going to talk about the shifting role of archivists.
As early as the 1940s, archivists were talking about machine-readable records. The debates and experiments have been going on for many decades. One early approach was to declare that electronic records were not archives, because the archives couldn’t deal with them. (Archivists and records managers have always been at odds, he says, because RM is about retention schedules, i.e., deleting records.) Over time, archivists came up to speed. By 2000, some were dealing with electronic records. In 2010, many do, but many do not. There is a continuing debate. Archivists have spent too long debating among themselves when they need to be talking with others. But, “archivists tend not to be outgoing folks.” (Archivists have had issues with the National Archives because their methods don’t “scale down.”)
There are many projects these days. E.g., we now have citizen archivists who maintain their own archives and who may contribute to public archives. Who are today’s archivists? Archival educators are redefining the role. Richard believes archives will continue, but the profession may not. He recommends reading the Clair report [I couldn’t get the name or the spelling, and can’t find it on Google :( ] on audio-visual archives. “I read it and I wept.” It says that we need people who understand the analog systems so that they can be preserved, but there’s no funding.
Victoria Stodden’s talk gloomy title is “The Coming Dark Ages in Scientific Knowledge.”
She begins by pointing to the pervasive use of computers and computational methods in the sciences, and even in the humanities and law schools. E.g., Northwestern is looking at the word counts in Shakespearean works. It’s changing the type of scientific analysis we’re doing. We can do very complicated simulations that give us a new way of understanding our world. E.g., we do simulations of math proofs, quite different from the traditional deductive processes.
This means what we’re doing as scientists is being stored in script, codes, data, etc. But science only is science when it’s communicated. If the data and scripts are not shared, the results are not reproducible. We need to act as scientists to make sure that this data etc. are shared. How do we communicate results based on enormous data sets? We have to give access to those data sets. And what happens when those data sets change (corrected or updated)? What happens to results based on the earlier sets? We need to preserve the prior versions of the data. How do we version it? How do we share it? How do we share it? E.g., There’s an experiment at NSF: All proposals have to include a data management plan. The funders and journals have a strong role to play here.
Sharing scientific knowledge is harder than it sounds, but is vital. E.g., a recent study showed that a cancer therapy will be particular effective based on individual genomes. But, it was extremely hard to trace back the data and code used to get this answer. Victoria notes that peer reviewers do not check the data and algorithms.
Why a dark age? Because “without reproducibility, knowledge cannot be recreated or understood.” we need ways and processes of sharing. Without this, we only have scientists making proclamations.
She gives some recommendations: (1) Assessment of the expense of data/code archiving. (2) Enforcement of funding agency guidelines. (3) Publication requirements. (4) Standards for scientific tools. (5) Versioning as a scientific principal. (6) Licensing to realign scientific intellectual property with longstanding scientific norms (Reproducible Research Standard). [verbatim from her slide] Victoria stresses the need to get past the hurdles copyright puts in the way.
Q: Are you a pessimist?
A: I’m an optimist. The scientific community is aware of these issues and is addressing them.
Q: Do we need an IRS for the peer review process?
A: Even just the possibility that someone could look at your code and data is enough to make scientists very aware of what they’re doing. I don’t advocate code checking as part of peer review because it takes too long. Instead, throw your paper out into the public while it’s still being reviewed and let other scientists have at it.
Q: [rick] Every age has lost more info than it has preserved. This is not a new problem. Every archivist from the beginning of time has had to cope with this.
Jason Baron of the National Archives (who is not speaking officially) points to the volume of data the National Archives (NARA) has to deal with. E.g., in 2001 32 million emails were transferred to NARA; in 2009, 250+ million archives were. He predicts there will be a billion presidential emails by 2017 held at NARA. The first lawsuit over email was filed in 1989 (email=PROFS). Right now, the official policy of 300 govt agencies is to print email out for archiving. We can no longer deal with the info flow with manual processes. Processing of printed pages occurs when there’s a lawsuit or a a FOIA request. Jason is pushing on the value of search as a way of encouraging systematic intake of digital records. He dreams of search algorithms that retrieve all relevant materials. There are clustering algorithms emerging within law that hold hope. He also wants to retrieve docs other than via key words. Visual analytics can help.
There are three languages we need: Legal, Records Management, and IT. How do we make the old ways work in the new? We need both new filtering techniques, but also traditional notions of appraisal. “The neutral archivist may serve as an unbiased resource for the filtering of information in an increasingly partisan (untrustworthy) world” [from the slide].