Secrecy News

DARPA Seeks Technology to Support Declassification

The Defense Advanced Research Projects Agency has issued a new solicitation to industry and academia in an attempt “to discover new technologies to support declassification.”  Researchers are invited (pdf) to submit ideas for innovative approaches to declassification that will support the National Declassification Center in achieving its goals.

Can technology actually make a difference in declassification?  It seems clear that it can, at least within certain limits.

One thing that technology cannot do is to render a decision about exactly what should be classified or declassified.  That is a policy question which is dependent on a complex, rapidly changing factual environment (e.g. what related information is already available in the public domain) as well as a largely subjective threat assessment (e.g. what damage might conceivably result from disclosure and what benefits might ensue).  Such a decision does not easily lend itself to a technological formula.

Besides that, the executive order that governs the national security classification system is permissive, not mandatory;  it allows the classification of eligible information, but does not require it.  So any algorithm that dictates the continued classification of a certain category of information is likely to be wrong at least sometimes.

However, the declassification process is composed of several discrete steps, many or all of which could be facilitated by new technologies.  These steps include the collection and assembly of records for review, the circulation of records to reviewers as needed, the actual review and redaction process, and the distribution of the declassified records, among others — each of which might be streamlined and expedited by new technological measures.

So, for example, if it were possible to routinely incorporate the digitization of records into the declassification process, and to make the digitized records available online so that readers would not have to come to the National Archives or to the Presidential Libraries just to view them, that action alone would multiply the utility of the declassification process many times over.

But perhaps the strongest contribution that technology could make involves the future declassification of records that are being classified today.  Classified records that are being created now could be tagged in such a way as to expedite their ultimate declassification.  In fact, the goal should be to eliminate the need for declassification processing altogether, or as far as possible.  Instead, most classified records should literally be self-declassifying.  Their classification controls should expire and be automatically canceled.  In principle, this ought to be readily achievable.

The Public Interest Declassification Board will hold a public session on the potential role of new technology in declassification on Thursday, September 23 at the National Archives.  The agenda is here (pdf).

The new DARPA solicitation was reported in “Darpa Wants You To Build An Anti-Secrecy App” by Spencer Ackerman in Wired Danger Room, September 14.

0 thoughts on “DARPA Seeks Technology to Support Declassification

  1. DARPA wants you to build an anti-secrecy app

    How juvenile. You’re the anti-secrecy app.

    One supposes if the anti-secrecy app could be made for the iPod, iPhone and iPad, we could use touch screens and banish it all while listening to the virtual piano tune they love so much.

    In between all the users farting sunshine and urinating champagne at the iTunes store.

  2. Not sure the assertion that technology cannot make the decision (recommendation to protect human egos) to declassify a document with a high degree of precision.

    By definition a classifier of information will tend to think that the classified information is important notwithstanding the absence of objective criteria; the short half-life of most, if not all data; the inability to even remember why the document was classified; and that classified information is routinely declassified without objective harm (the classifiers will always forecast subjective harm)!

    The benefits of using elapsed time as a very powerful declassifier may far outweigh the burdens of those instances where harm (even grave harm) occurs from relying on time and technology. Don’t be surprised if you’re told that such an algorithm, where it to exist is classified!

    In this regard it should be noted that a group of so called “policy declassifiers” sitting around a table, in some cases decades later will experience false positives and negatives, too. Whether those false positives-negatives are better than time-technology is worth objectively testing.

    An imperfect analogy is our jury system that sends thousands of innocent people to jail and even death with surprising regularity and without so much as a murmur. Now, contrast this with the rhetoric you hear from our national security classification system.

  3. Not sure the assertion that technology cannot make the decision (recommendation to protect human egos) to declassify a document with a high degree of precision.

    I understand the desire of software gadgeteers to think of classification (or, indeed any human activity), as a substance within a virtual rock or landfill, which if it can only be digitally measured and parsed, will furnish a result.

    That doesn’t mean they need even more indulging.

Leave a Reply

Your email address will not be published. Required fields are marked *