Skip Navigation
Analysis

Rescuing History (and Accountability) from Secrecy

Congress should earmark funding for agencies’ own declassification efforts or require agencies to spend a certain percentage of their information security budgets on declassification

September 14, 2018

Cross-posted from Knight First Amend­ment Insti­tute.

Matthew Connelly’s essay on state secrets and archival negli­gence offers a fascin­at­ing new lens on an old prob­lem. The liter­at­ure on national secur­ity secrecy tends to focus on the threat it poses to demo­cratic account­ab­il­ity and parti­cip­a­tion. I, for one, have spent years think­ing and writ­ing about excess­ive govern­ment secrecy, and yet I’ve never considered the issue from the perspect­ive of an arch­iv­ist or histor­ian.

Connelly describes numer­ous chal­lenges to modern record­keep­ing, ranging from the prob­lem of “bit rot” (loss of data result­ing from outmoded soft­ware or hard­ware) to the inad­equacy of keyword searches as a substi­tute for find­ing aids. But one chal­lenge that partic­u­larly concerns him is the govern­ment’s gener­a­tion of a massive amount of clas­si­fied mater­ial coupled with an inef­fect­ive system for declas­si­fy­ing it. I will limit myself to comment­ing on this aspect of his essay, since the rest of it goes beyond my expert­ise.

Connelly aptly conveys the scope of the prob­lem — both how long it has exis­ted and how massive it is. He outlines the skewed incent­ives that lead offi­cials to clas­sify docu­ments unne­ces­sar­ily and in ever-increas­ing numbers. He shows that the pace of declas­si­fic­a­tion has fallen far behind the pace of clas­si­fic­a­tion, creat­ing a grow­ing and poten­tially insur­mount­able back­log. He makes a convin­cing case that our system of histor­ical record­keep­ing is in crit­ical condi­tion, on account of clas­si­fic­a­tion-related prob­lems as well as other causes, and that “[i]f this system collapses, Amer­ica’s commit­ment to learn­ing from its history will become a thing of the past, because the past itself will be impossible to recover.”

Although they are under­stand­ably not Connelly’s focus, it’s worth high­light­ing some of the more commonly cited and imme­di­ate harms of over­clas­si­fic­a­tion. Depriving people of inform­a­tion about the govern­ment’s policies or activ­it­ies impedes their abil­ity to engage in informed debate and to cast informed votes. Over­clas­si­fic­a­tion thus damages the most basic mech­an­isms of demo­cracy. It also under­mines the rule of law, as the govern­ment cannot be held account­able for viol­a­tions that are concealed from the public. And it subverts the consti­tu­tional system of checks and balances, making it more diffi­cult in myriad ways for Congress and the courts to provide mean­ing­ful over­sight.

Connelly proposes several solu­tions, all of which have merit. He leads with the issue of resources, exhort­ing Congress to increase fund­ing for the National Archives and Records Admin­is­tra­tion. Fund­ing is a dry subject, and legal and policy experts usually stick to sexier remed­ies — things like curb­ing exec­ut­ive priv­ilege or beef­ing up judi­cial review. But as Joe Biden has said, “Don’t tell me what you value. Show me your budget, and I’ll tell you what you value.” Connelly is to be commen­ded for under­stand­ing the crit­ical import­ance of fund­ing to this issue. I would expand on his recom­mend­a­tion and urge Congress to earmark fund­ing for agen­cies’ own declas­si­fic­a­tion efforts or require agen­cies to spend a certain percent­age of their inform­a­tion secur­ity budgets on declas­si­fic­a­tion. As Connelly points out, the ratio of declas­si­fic­a­tion spend­ing to clas­si­fic­a­tion spend­ing has decreased dramat­ic­ally in recent years.

Connelly also recom­mends that the exec­ut­ive branch use data-science tech­niques, such as machine learn­ing, to build a “declas­si­fic­a­tion engine” — in lay terms, computer programs that could help identify sens­it­ive inform­a­tion for purposes of speed­ing up declas­si­fic­a­tion review. A CIA pilot project in this area showed prom­ising results, but Congress hasn’t alloc­ated the neces­sary fund­ing to follow up. I agree with Connelly’s recom­mend­a­tion, with one caveat — it is crit­ical that auto­mated tech­no­logy not be used to make decisions about whether to clas­sify inform­a­tion in the first instance (as some have proposed). When “trained” with inform­a­tion that has already been clas­si­fied, computers can identify docu­ments that contain roughly the same inform­a­tion — a conveni­ent aide to declas­si­fic­a­tion. But initial determ­in­a­tions of whether national secur­ity would be harmed by the disclos­ure of inform­a­tion are inher­ently subject­ive and require care­ful thought and judg­ment.

As for “deriv­at­ive” declas­si­fic­a­tion — the prac­tice of clas­si­fy­ing docu­ments on the ground that contain previ­ously clas­si­fied inform­a­tion — auto­ma­tion might seem useful in theory, but it could be disastrous in prac­tice. Computer programs would occa­sion­ally make mistakes, espe­cially in the begin­ning. Given that offi­cials always err on the side of clas­si­fic­a­tion, they would likely correct a computer program’s false negat­ives, but not its false posit­ives. The “machine learn­ing” func­tion would then intern­al­ize, perpetu­ate, and magnify the human tend­ency toward over­clas­si­fic­a­tion.

I would add two other recom­mend­a­tions in the area of declas­si­fic­a­tion. Connelly mentions so-called “auto­matic,” system­atic, and discre­tion­ary declas­si­fic­a­tion, but he leaves out the fourth way in which records may be declas­si­fied: mandat­ory declas­si­fic­a­tion review (MDR). Under MDR, members of the public can submit requests to agen­cies to declas­sify partic­u­lar docu­ments. They can appeal deni­als, first within the agency and then to the Inter­agency Secur­ity Clas­si­fic­a­tion Appeals Panel. The rate of declas­si­fic­a­tion under MDR is orders of magnitude higher than under the Free­dom of Inform­a­tion Act: More than 90 percent of reques­ted docu­ments are declas­si­fied either in whole or in part. But MDR is under­fun­ded, under­staffed, and notori­ously slow. Again, Congress should dedic­ate more fund­ing in this area. In addi­tion, agen­cies should create an exped­ited review track — similar to the exped­ited review track that exists in FOIA — when the reques­ted records address a matter of signi­fic­ant public interest.

To the four exist­ing declas­si­fic­a­tion mech­an­isms, I would add a fifth. All clas­si­fied docu­ments must be marked with a declas­si­fic­a­tion date. In 2017, most docu­ments were declas­si­fied for peri­ods of 10 years or less. Yet incred­ibly, there is no regu­lar system for perform­ing declas­si­fic­a­tion reviews before “auto­matic” declas­si­fic­a­tion kicks in at 25 years. Unless someone happens to request a docu­ment through FOIA or MDR, it is likely to remain clas­si­fied even though its declas­si­fic­a­tion date came and went years ago. Going forward, clas­si­fied docu­ments should be elec­tron­ic­ally tagged to gener­ate a prompt when the docu­ment reaches its declas­si­fic­a­tion date, trig­ger­ing a require­ment to review. This would not only make more inform­a­tion avail­able to the public sooner but also reduce the burden on the auto­matic review process that is triggered at the 25-year mark.*

Even with these reforms, though, declas­si­fic­a­tion will never be able to keep up unless we reduce the amount of clas­si­fied inform­a­tion pour­ing into the system. Connelly alludes to this and notes the need to adopt “a more rational, risk-manage­ment approach to protect­ing sens­it­ive inform­a­tion.” He is correct, but the exec­ut­ive branch needs more than an atti­tude adjust­ment — it needs narrower and more specific criteria for clas­si­fic­a­tion. The exec­ut­ive order that currently governs clas­si­fic­a­tion contains no defin­i­tion of “national secur­ity” or any examples of harm that would justify clas­si­fic­a­tion. Moreover, the categor­ies of clas­si­fi­able inform­a­tion listed in the order are far too broad. For instance, they include “intel­li­gence sources or meth­ods” writ large, even though intel­li­gence agen­cies often rely on open sources and many of their meth­ods are well known. The categor­ies also include “foreign rela­tions or foreign activ­it­ies of the United States,” which encom­passes much of what we read in the news­pa­per every day. While it’s true that offi­cials need a fair amount of discre­tion in making assess­ments of national secur­ity harm, that discre­tion should not be entirely unfettered. I would like to see a White House­–led commis­sion of senior agency offi­cials charged with tight­en­ing the criteria for clas­si­fic­a­tion and provid­ing a defin­i­tion of “damage to the national secur­ity” that sets an appro­pri­ately high bar.

The new criteria also should expand on the categor­ies of inform­a­tion that may not be clas­si­fied. In recent years, the govern­ment has advanced a danger­ous new argu­ment that inform­a­tion may be clas­si­fied if our enemies could use it as anti-U.S. “propa­ganda.” The govern­ment has used this argu­ment to shield photos and videos of Guantá­namo detain­ees, for instance. Of course, the worse the U.S. govern­ment’s conduct, the more likely our enemies could use it to gener­ate anti-U.S. senti­ment. Indeed, the govern­ment’s theory provides a conveni­ent end-run around the current prohib­i­tion on clas­si­fy­ing inform­a­tion to “conceal viol­a­tions of law”: The govern­ment can simply argue that the purpose of clas­si­fic­a­tion was to deny our enemies a propa­ganda oppor­tun­ity. The result­ing abil­ity to clas­sify the govern­ment’s worst abuses endangers core account­ab­il­ity and rule-of-law prin­ciples. New clas­si­fic­a­tion criteria should make clear that concerns about propa­ganda are not a legit­im­ate basis for clas­si­fic­a­tion.

They should also rein in the prac­tice of secret law. Increas­ingly, the exec­ut­ive branch is clas­si­fy­ing rules and legal inter­pret­a­tions that set bind­ing stand­ards for govern­ment conduct. This secret body of law includes not only Office of Legal Coun­sel opin­ions — perhaps the best-known source of secret law — but also pres­id­en­tial direct­ives, intel­li­gence agen­cies’ rules and regu­la­tions, and unpub­lished inter­na­tional agree­ments that have the force of treat­ies. As I explained in a 2016 Bren­nan Center report, clas­si­fy­ing legal author­it­ies raises seri­ous consti­tu­tional ques­tions and leads to distinct demo­cratic harms. Exec­ut­ive branch offi­cials should not have the option of clas­si­fy­ing pure legal analysis or legal stand­ards.

Whatever the applic­able limits on clas­si­fic­a­tion, offi­cials will continue to exceed them until the incent­ives change. In his conclu­sion, Connelly suggests that in the longer run, it might become neces­sary for agen­cies to treat the wrong­ful with­hold­ing of inform­a­tion as seri­ously as they treat unau­thor­ized disclos­ures. I agree, except that I would­n’t wait. Agen­cies should act now to imple­ment systems for hold­ing offi­cials account­able for over­clas­si­fic­a­tion. Clas­si­fi­ers should be required to docu­ment their reas­on­ing (the National Geospa­tial-Intel­li­gence Agency already does this, accord­ing to a recent report of the govern­ment office that over­sees clas­si­fic­a­tion policy). Agen­cies should then conduct peri­odic spot audits, review­ing clas­si­fic­a­tion decisions and the support­ing docu­ment­a­tion. Offi­cials found to engage in inten­tional, negli­gent, or routine over­clas­si­fic­a­tion should be subject to mandat­ory penal­ties. On the flip side, offi­cials should be gran­ted “safe harbor” for good-faith decisions not to clas­sify inform­a­tion. Agen­cies might also consider giving small cash awards to employ­ees who bring success­ful chal­lenges to clas­si­fic­a­tion decisions.**

Finally, the entire clas­si­fic­a­tion system rests on the premise that shield­ing inform­a­tion is an effect­ive method of protect­ing national secur­ity. Given the prac­tical diffi­culties with secur­ing data in the digital era, it might be time to rethink this premise. Indeed, experts have already begun to ques­tion the util­ity of secrecy. In a 2005 memo, Defense Secret­ary Donald Rums­feld concluded that “[t]he United States Govern­ment is incap­able of keep­ing a secret. If one accepts that, and I do, that means that the U.S. Govern­ment will have to craft policies that reflect that real­ity.” In 2011, a distin­guished group of national secur­ity offi­cials and experts convened at a work­shop to discuss how the United States might revamp its national secur­ity strategies for a world where data hacks and “insider threats” argu­ably make secrecy unsus­tain­able, if not impossible. Revers­ing the secrecy-obsessed mind­set that has permeated the national secur­ity state since its incep­tion is easier said than done. But it might prove crucial to our secur­ity. We would never place our faith in, say, a missile defense system that performed as poorly as the secrecy system does today.

Of course, there is zero chance of the Trump admin­is­tra­tion taking up any of these propos­als and little chance that a future one will pursue more than incre­mental change. Exec­ut­ive branch offi­cials are far too commit­ted to the secrecy system and unlikely to embrace limits on their own author­ity. Accord­ingly, my final recom­mend­a­tion picks up on a point that Connelly touches on in his conclu­sion: Congress must end its decades of abdic­a­tion in this area. There are consti­tu­tional dimen­sions to the pres­id­ent’s power to clas­sify inform­a­tion, but it does not follow that Congress is power­less. It has acted boldly in the past. FOIA, for instance, author­izes judges to over­turn pres­id­en­tial clas­si­fic­a­tion decisions, although they almost never do so. Many of the reforms sketched above could be mandated or incentiv­ized by Congress. The next Congress, or a future one, should flex its consti­tu­tional muscle and exert some control over the runaway clas­si­fic­a­tion regime. That would be a win for demo­cracy, the rule of law, and — as Connelly has shown — history.

* I proposed this meas­ure, along with others discussed here, in a short white paper soli­cited by the Public Interest Declas­si­fic­a­tion Board, a pres­id­en­tial advis­ory group.

** These propos­als are explained in more detail in another Bren­nan Center report.

(Image: Shut­ter­stock)