Back in June, I showed several departments in the government had done inadequate searches for documents responsive to the NYT and, especially, ACLU FOIAs on targeted killing.

DOJ did not perform a reasonable search for documents responsive to ACLU’s FOIA

Part of the problem–for all respondents save the OLC (and CIA, which didn’t describe its search)–is that they used search terms that were likely to leave out responsive documents. In the case of DOJ’s Office of Information Policy, that problem was exacerbated because it searched only on the names of Anwar and Abdulrahman al-Awlaki and Samir Khan in conjunction with the word “target;” not only would that search leave out documents responsive to the NYT FOIA, it was pretty much guaranteed to leave out several important parts of the ACLU request, notably those pertaining to the underlying evidence that Anwar al-Awlaki was an imminent threat or operational.

OIP’s inadequate search was proven by the results of OLC’s search. OLC found 50 documents responsive to the ACLU’s FOIA that also included offices under OIP’s area of responsibility; 32 of those fell in the abbreviated time frame OIP included in their search. OIP only found one of those documents on its own, and only found 4 documents, total, on its own. Given that there were surely a bunch of conversations that transpired exclusively within the Attorney General and Deputy Attorney General’s offices that OLC couldn’t find, we can say with certainty that OIP’s searches found just a tiny fraction (probably less than one percent) of responsive documents.

DOJ doesn’t acknowledge scope of missed documents

The ACLU raised those and other problems with the government’s search in July. In last week’s response, the government didn’t admit what the record clearly shows–that their search was inadequate–and offer to do a real search. Rather, it called the ACLU’s points “nitpicks.” It responded to ACLU’s argument that only searching documents in conjunction with “target” would miss a lot of responsive documents (the ACLU didn’t make the point about the “imminent” and “operational” intelligence as strongly as they might have) by effectively saying, “excluding documents was the point,” even while misrepresenting the content of ACLU’s request as pertaining only to the decision to kill Awlaki and not the underlying decision that he represented an imminent threat because he had gone operational.

And it responded to the ACLU’s demonstration that the search clearly missed responsive documents because OLC had found 10 times more documents from OIP’s area of responsibility than OIP had with a citation to a case that found the government hadn’t conducted an adequate search because it relied on a name search, which is what OIP effectively used. The one line of the decision they cite pertains to the government failing to find one document, not 49 (nowhere in the government response do they admit to how many documents they failed to find).

The ACLU points out that OIP did not uncover some of the documents located by OLC. “Of course, the failure to turn up [a] document does not alone render the search inadequate; there is no requirement that an agency produce all responsive documents.” Nation Magazine v. U.S. Customs Serv., 71 F.3d 885, 892 n.7 (D.C. Cir. 1995). Again, the focus is on whether the search was reasonable.

Moreover, this case’s holding would support the ACLU argument that it’s not enough to do a name search if it clearly leaves out the intent of the request, as OIP’s searches do.

OIP didn’t search FOR responsive documents, it worked to exclude documents

As I said, DOJ tried to explain their use of names plus “target” as a justifiable means of search because the Office of the Attorney General and Office of the Deputy Attorney General had so many files they needed to sort somehow.

OIP used fewer search terms than OLC in part because it covers offices with a broader range of interests.

[snip]

Moreover, OIP’s limitation on the search of names to documents also including the word “target” is reasonable in light of the language of the ACLU’s request, which did not seek all documents concerning Aulaqi, but rather information on the factual and legal basis for the alleged individual targeting decisions.

But that doesn’t explain why “target” was the proper way of excluding bunches of non-responsive documents. In fact, in a supplemental declaration submitted to explain why OIP used such an exclusionary search, Douglas Hibbard seems to misunderstand the ACLU request.

At the time OIP was beginning its search of the EV Vaults [of ODAG and OAG emails], it had already completed its review of the unclassified paper and unclassified electronic files located in OAG, ODAG, and OASG. This review had demonstrated that the majority of the records related to Anwar al-Aulaqi maintained by OAG and ODAG (and, in fact, all of the records maintained by OASG) were not responsive to the request in that they concerned Anwar al-Aulaqi, but not the alleged use of lethal force against him. Given that knowledge, OIP conducted an initial search of the EV Vaults using the terms “al-Aulaqi,” “al-Awlaki,” and “al-Alwaki.” These searches located a substantial amount of material, including many of the non-responsive records located in the searches of unclassified paper and unclassified electronic files. A preliminary review of a substantial sampling of these results demonstrated that the located material was not responsive to the request reasons similar to those applicable to the unclassified paper and unclassified electronic material, i.e., the records did not pertain to the alleged use of lethal force against Anwar al-Aulaqi. [my emphasis]

Now, as a threshold matter, the order implied here–a search of hard copy and electronic files, then a search of emails–is the reverse order of how Hibbard presented the search in his original declaration, which I’ll return to below.

More importantly, the ACLU didn’t ask for only those documents that discussed killing Awlaki, it also asked for documents “pertaining to the factual basis” for his killing, including:

• Facts supporting a belief that al-Awlaki posed an imminent threat to the United States or United States interests;
• Facts supporting a belief that al-Awlaki could not be captured or brought to justice using non-lethal means;
• Facts supporting the assertion that al-Awlaki was operationally involved in al Qaeda, rather than being involved merely in propaganda activities;

To find those documents, you might have searched on “Awlaki” and “Abdulmutallab” or “Awlaki” and “Karim.” You might search on “Awlaki” and “Saleh.” But by looking for only those documents that talk about such issues while using the word “target,” you deliberately leave out precisely those underlying documents.

Indeed, an examination of the sole document the government released–talking points that OIP describes as “final talking points prepared for the use of the Attorney General and others in addressing hypothetical questions about Anwar al-Aulaqi’s death,” but which, given the reference in them to “my Administration” appear to be intended for President Obama’s delivery–shows the result of such a search. It simply makes a number of claims, but points to no facts to back up the claims, including:

• Anwar al-Awlaki was an operational leader of al Qaeda in the Arabian Peninsula–al Qaeda’s most active operational affiliate and a group that poses a serious threat to the United States, our partners, and to the people of Yemen.
• He took the lead in planning and directing efforts to murder innocent Americans and was directly tied to several attempted terrorist attacks on the United States.
• The legal analysis would be slightly different with respect to U.S. citizens as we would have to take into account any constitutional protections that might apply to a U.S. citizen who is leading enemy forces in their efforts to kill innocent Americans

The document also notes two more public documents–Treasury’s labeling of Awlaki a terrorist in July 2010 and Robert Gates’ description of him as such a month later–that should have at least appeared in the files, along with a lot of backup.

In other words, the documents responsive to the request as OIP ran the search would end up producing documents like this one, which obscure the legal process behind targeting Awlaki behind blanket assertions (in some cases, arguable ones) but no actual facts or analysis.

Given that it now appears the government first tried to kill Awlaki at a time when DOJ’s analysis supported none of these bullet points, I’d say that’s a rather telling result.

OIP implies ODAG and OAG staffers are responsible if the search is inadequate

And frankly, I suspect Douglas Hibbard knows he has missed some of the most responsive documents.

Consider, for example, how his new declaration adds a sentence to his first paragraph, which is otherwise boilerplate, about who he is and what OIP does. After this sentence:

In processing such requests, the IR Staff consults with personnel in the senior leadership officers and, when appropriate, with other components within the Department of Justice, as well as other Executive Branch agencies.

Hibbard added this sentence:

In devising and conducting searches, the IR Staff relies on its knowledge of what is in the relevant files, as well as consultations with identified custodians of potentially responsive records, and continually refines search parameters to ensure a search reasonably calculated to locate responsive records.

An explanation he repeats two paragraphs later:

In light of the direct participation in the searches by OAG, ODAG, and OASG personnel with familiarity with the subject matter, as detailed in ¶ 9 of my June 20, 2012 declaration, coupled with OIP’s own extensive experience in conducting records searches, I have confidence that the searches conducted for this request were reasonably calculated to locate the records that had been requested.

I find it of particular interest that, having been called on the 49 emails OLC found but OIP didn’t (to say nothing of a presumably far greater volume of discussion within OAG and ODAG), Hibbard points to the involvement of personnel from those offices to assure that the search was thorough.

To some degree it doesn’t make any sense. OIP appears to have decided to scan out most emails using the search term “target” themselves. Though given that was a response to having conducted searches on electronic files (one that didn’t include either names or “target,” which leads me to suspect that the OAG and ODAG personnel handed over their Anwar al-Awlaki unclassified files, which OIP then searched), it’s possible it was influenced by OAG and ODAG personnel decisions about what to hand over initially. That is, it seems those early searches (including through unclassified hard copies) led OIP to exclude everything that doesn’t include “target.”

The biggest impact the OAG and ODAG seem to have had on the search, though, came in delayed identification of files from former personnel–the classified files of one in OAG, and all the files of the former ODAG records custodian for these issues. Still, all these records were ultimately searched in the same fashion as the current employees.

Which leaves just one other reason I can see why OIP might include this hedging language about the search. While OIP conducted the hard copy and online searches of unclassified documents and emails, OAG and ODAG personnel undertook the search of classified files and emails. It’s not even clear that OIP has seen those documents (all the responses, including OIP’s, have refused to give details about classified searches).

OIP doesn’t offer an explanation–as they should–why OLC found so many documents their search failed to find. But I wonder if for some reason those emails were treated as classified, not unclassified?

Now, to be clear, OIP doesn’t admit what is patently obvious, that their search was inadequate. I’m the one saying that if OIP can only find a fraction of the documents OLC can find for its own officers, the search is inadequate.

But I do find it interesting that OIP points to the involvement of OAG and ODAG personnel as proof that this search is adequate, when the evidence seems to suggest either OIP or those offices designed an inadequate search.