This is going to be one of my really weedy posts, but if it’s any consolation, I went crazy while weeding (in the garden) yesterday and accidentally pulled up a huge tomatillo plant that was just about to bear tons of tomatillos. So I’m suffering from having gotten myself lost in the weeds right now.

Back in June, I did a close review of which documents from its index the CIA had described for ACLU in its Vaughn Index (Part One, Part Two) of documents pertaining to events described in the torture tapes. It appeared that the CIA had included fewer documents from May–the period when CIA was fighting with FBI over control of Abu Zubaydah’s interrogation–than it did from later in its document series.

By comparing the total index with the Vaughn index, though, we can get a sense of what the CIA did include. For most of the series of cables reporting to and from the field, the CIA submitted fairly regular cables–every 10, 11, or 12 cables. From June 22, 2002 through August 20, 2002, they appear to have submitted every 10 document, like clockwork (in addition to the handwritten log dated August 4). (It’s impossible to exactly identify a pattern from after that because so many of the cables are the same length, though it is possible that it sticks pretty close to the every tenth cable pattern.)

But things in April–when the FBI and CIA were fighting over control of the interrogation and Abu Zubaydah was reportedly cooperating with the FBI–and May–when the small box was introduced at least two months before OLC approved its use–things are a bit more irregular. In April, for example, the CIA submitted documents 1, 12, 1922, 32, 42 or 43, and 53 (plus the handwritten log, which was document 3); that gives you gaps of 11, 7, 13, 10, 10, 10, 10, and 10 documents (not including the log in the series). In May, CIA submitted documents 64, 65, or 66, 77 or 78, 89 or 90, document 99, 110, 123, 134, 146, document 155 or 156, ad 165; while it’s harder to pin down the gaps, there is necessarily one 13-document gap early in the month, a 9-document gap between May 8 and 11,  another 13-document gap between May 14 and 17, a 12-document gap between May 20 and 24, and one 9-document gap between May 24 and 30. 

In other words, whereas later in the series the CIA just provided every tenth document, for this early period, they cherry-picked what they submitted. [my emphasis]

I asked the ACLU why the gaps were irregular, they asked the CIA, and the CIA has finally answered that question. Here’s their response, with my comments about each response.

First, the apparent deviation from the "every tenth contemporaneous cable" portion of the sample is because (1) three of the entries from the mostly-cable section of the May 18, 2009 list were not cables, so they were skipped over (i.e., not counted) for the one-in-ten sampling; and (2) twenty cables were non-contemporaneous cables, so they were also skipped over for the one-in-ten sample. Second, as you surmised, the non-cable records identified on the May 19 [sic], 2009 list that were excluded from the Vaughn were excluded because they are non-contemporaneous (i.e., derivative).

Third, Vaughn item number 64 corresponds to document 568 on the May 18, 2009 list. Although the May 18, 2009 list refers to the document as an "outline," and the Vaughn index refers to the document as a "memo," both entries refer to the same document.

This explanation does provide a plausible explanation for the known gaps, though it identifies some interesting new communication patterns for key periods of Abu Zubaydah’s torture.

Let me start from the back: with Vaughn document 64. It was pretty clear that this document had to be document 568, as it was the only undated outline of 3 pages in length. But note, for the moment, the description of it:

Waterboard Summary

This document is a three-page memo which summarizes details of waterboard exposures from the destroyted videotapes. [my emphasis]

Next, this document made something clear that wasn’t clear to me before (but should have been)–the CIA succeeded in exempting documents it considered derivative. This goes back to their April 9 letter, in which they said:

The Government will not create Vaughn-like entries for the following categories of documents: documents that do not describe the interrogations but contain the raw intelligence that was collected from the interrogations, and derivative documents that merely summarize information contained within interrogation records.

So long as derivative records can be claimed not to relate to the reasons behind the destruction of the videotapes, nothing in Judge Hellerstein’s April 20 order requires CIA to hand over derivative documents. 

Which explains why CIA excluded all the timelines and outlines described in the index. I guess the CIA is making a (fair) distinction between notes made while actually viewing the videotapes, the waterboard summary above, and summaries of interrogation records themselves.

Bummer. Being deprived of timelines for me is almost as sad an event as ripping out my almost-bearing tomatillo plant.

Which brings us to the last (actually first) explanation the CIA gave ACLU–meant to explain why there are 13 and 11 document gaps in May, unlike the other months. They say, 

(1) three of the entries from the mostly-cable section of the May 18, 2009 list were not cables, so they were skipped over (i.e., not counted) for the one-in-ten sampling; and (2) twenty cables were non-contemporaneous cables, so they were also skipped over for the one-in-ten sample

By "mostly-cable section" of the index, I assume they mean documents 1 through 549 of the index. And, yes, there are three documents that are not cables: documents 3 and 255, which are handwritten interrogation logs included in the Vaughn Index as documents 57 and 58; and document 540, which was a memo sent from HQ to the Field at about the time that people started dying from torture in Afghanistan and close to the time the CIA stopped videotaping interrogations. That leaves 546 cables, of which they claim 20 are derivative, leaving 526 non-derivative cables. Altogether, they have included 53 cables in their Vaughn index, so they have sampled one tenth of those cables.

This appears to be a plausible (or at least convenient) explanation. As the second set of columns in this spreadsheet shows, if you take every tenth cable from the index as ordered, you always have at least a 10-cable break, with a number of 11, 12, and 13-cable breaks.

What that reveals, then, are certain periods during which the field was sending summaries of the torture sessions back to HQ in addition to original interrogation reports:

April 25 to April 28: One "derivative" cable

April 28 to May 1: One to three "derivative" cables

May 1 to May 5: One to three derivative cables (four total derivative cables from April 28 to May 5)

May 5 to May 8: Two derivative cables

May 11 to May 14: One derivative cable

May 14 to May 17: Three derivative cables

May 17 to May 20: One derivative cable

May 20 to May 24: One derivative cable

[break]

May 30 to June 4: Three derivative cables

[break]

June 8 to June 13: One derivative cable

June 13 to June 22: One derivative cable

[long break]

August 24 to August 31: One derivative cable

[long break]

September 22 to September 24: One derivative cable

The CIA was not cherry-picking. Rather, they had defined one set of communication as derivative and thereby shielded it according to the terms of Judge Hellerstein’s order. (Note, my math shows 19 of these derivative cables, though this method is error-prone). 

So the CIA was not–assuming these are derivative cables–withholding these cables improperly. But the communication pattern is interesting on its own right. It suggests that in the early period–particularly during the period when the FBI remained onsite–the Field was sending their normal records, and sending a second set of summary records. Perhaps they were sending updates to Tenet’s office in addition to Counterterrorism. Perhaps they were sending–say–the White House updates (though they do specify HQ). Perhaps the second set of cables were cables intended to be shared with the FBI (I can see why CIA would want to withhold those). Perhaps James Mitchell was sending updates to Bruce Jessen so they could plan more torture.

Whatever the explanation, these derivative cables seem to indicate a parallel set of communications during the early period of Abu Zubaydah’s interrogation.