What Was the CIA Really Doing with Merlin by 2003?

On June 26, 2003, CIA posted nuclear blueprints written in English on its website, claiming they were Iraqi.

Bloomberg is reporting that the exhibits released in the Jeffrey Sterling case may lead the UN to reassess some of the evidence they’ve been handed about Iran’s alleged nuclear weapons program.

International Atomic Energy Agency inspectors in Vienna will probably review intelligence they received about Iran as a result of the revelations, said the two diplomats who are familiar with the IAEA’s Iran file and asked not to be named because the details are confidential. The CIA passed doctored blueprints for nuclear-weapon components to Iran in February 2000, trial documents have shown.

“This story suggests a possibility that hostile intelligence agencies could decide to plant a ‘smoking gun’ in Iran for the IAEA to find,” said Peter Jenkins, the U.K.’s former envoy to the Vienna-based agency. “That looks like a big problem.”

Importantly, this story comes from two IAEA officials who are familiar with the evidence against Iran, and therefore would know if aspects of the Merlin caper resemble things they’ve been handed by the CIA, almost certainly including the Laptop of Deathlaundered through MEK to the CIA in 2004.

You’ll recall that immediately upon hearing some of the sketchy details of the Merlin caper I thought of the Laptop of Death and a dubious tale, told by Iraqi nuclear scientist Mahdi Obeidi, involving the blueprints posted above. And I’ve only got more questions about the operation given what we learned since that day.

Here are some of those questions.

  • Why did CIA immediately turn to dealing Iraq nuclear blueprints after such a clusterfuck on Merlin’s first operation — and why wasn’t Sterling involved?
  • Why did both Bob S and Merlin tell the FBI in 2006 that Sterling was just a marginal player in the operation?
  • Did the program get more sensitive over time?
  • Why is the government claiming this part of James Risen’s State of War is as sensitive than his exposure of a massive illegal wiretap program?
  • Did the kind of deception involved change?
  • What was CIA intending with its Iran approach in 2003, and what really happened with it?
  • Why explains the weird reception for Jeffrey Sterling’s complaint at the Senate Intelligence Committee?
  • Why was Bill Duhnke the top suspect?

Why did CIA immediately turn to dealing Iraq nuclear blueprints after such a clusterfuck on Merlin’s first operation — and why wasn’t Sterling involved?

As I have laid out, less than a month after Bob S deemed Merlin unable “to follow even the simplest and most explicit direction” (Exhibit 44), he and one other case officer who was apparently not Jeffrey Sterling (though Sterling was still nominally Merlin’s handler) approached Merlin about repeating the operation with another country (Exhibit 45). David Swanson has compellingly shown that that country was almost certainly Iraq. That operation, however, would be “rather more adventurous” than the Iranian op that Merlin had already proven so inadequate to.

I think it possible they bypassed Sterling because his Equal Opportunity complaints had already so soured his relationship with the CIA they had it in for him already. But I do find it interesting that the transition to Stephen Y happened right as they moved onto this “more adventurous” operation (and Stephen Y handled Merlin through this 2003 leak).

Why did both Bob S and Merlin tell the FBI in 2006 that Sterling was just a marginal player in the operation?

That Bob S was bypassing Sterling in April 2000, over a month before Merlin got a new case officer, also raises questions about why he and Merlin, in what seems remarkably similar testimony to the FBI in 2006, started saying that Sterling was not a central player in the operation. Bob S was doing 70% of the thinking on the operation, he reportedly told the FBI in an February 28, 2006 interview, Sterling just 30%. Sterling served only as a “middleman” editing his letters, Merlin told the FBI in an interview within a month after Bob S’. “The details of this operation were a wild forest to Sterling,” Merlin told the FBI in the same interview (though when asked on cross, he said he meant Sterling didn’t understand the technical details).

Why were Bob S and Merlin both so intent in the months after Risen’s book first appeared on insisting that Sterling’s understanding of the operation was incomplete?

Did the program get more sensitive over time?

Everything introduced at the trial treats the Merlin operation as a clandestine information collection operation. Yet a heavily redacted filing submitted in support of having Retired Colonel Pat Lang testify and other details from the trial suggest the operation got moresensitive as it went along. Like the contemporaneous cables, the filing suggests the operation was clandestine. “The [redacted] operation was conducted as a [redacted] clandestine intelligence operation.” But it also makes it clear that the government was trying to argue that this clandestine operation was covert. Note, for example, the discussion of CIA “electing” to notify Congress, obtain approval from the CIA Director, and … something redacted. That suggests the government went through some or all of the motions of the same kind of notice required under a Finding, without it being a formally covert operation. Risen may have been trying to get at this question, too, when he asked Bill Harlow’s counterpart somewhere (this wouldn’t have been at NSC, but it might have been at Sandia Lab), “he knew that President Clinton had approved the plan…but wanted to know if it had been reapproved by President Bush” (Exhibit 106; note, this appears to have been a seeded question, and not one that Sterling would have reason to pitch).

But two things suggest the program got, formally, more sensitive, perhaps even escalating to a covert operation that the US would want to deny. First, there are the two “facts” mentioned in the Lang filing that had not been shared with the defense, even though Lang was purportedly read into all the evidence pertaining to the Sterling defense. Then there’s an odd exchange that happened with Condi Rice. Eric Olshan asked “did everyone at the NSC know about this specific classified information?” (remember, within weeks, Bob S would tell the FBI more than 90 people were briefed into this compartment). Defense attorney Barry Pollack objected that the question was beyond the protective order. But Olshan insisted it wasn’t, and Judge Brinkema judged that “the government is very sensitive to the protective order and I doubt they would go beyond it.” The suggestion was that very few people at NSC were read into the precise details of the program when Condi talked NYT out of publishing in 2003.

All of this leads me to believe that the program had gotten much more sensitive between the time Sterling was booted off the program in 2000 and the time Risen was reporting the story in 2003.

If so, why?

Why is the government claiming this part of James Risen’s State of War is as sensitive than his exposure of a massive illegal wiretap program?

The program would have had to have gotten more sensitive over time, if any of the implications about the relative sensitivity of the chapters of Risen’s book — including the series of witnesses claiming Chapter 9 was the only one they read (though jurisdictional issues explain some of this, given that Risen’s NSA chapter came under MD’s purview) are to be believed.

After all, elsewhere in Risen’s book, he exposed a massive illegal wiretapping program that directly contravened FISA. He exposed a program that — we now know –directly implicated the Attorney General and Vice President in criminal wiretapping.

Yet the CIA and DOJ want us to believe that this program — described in contemporaneous CIA cables as an effort to give Iran a blueprint to find out if they wanted it — was more sensitive than that massive illegal program? (Admittedly this may all stem from the CIA thinking it is the center of the universe.)

Did the kind of deception involved change?

Those questions all make me wonder whether the kind of deception — and the audience — changed, if the project got more sensitive.

This program was established in January 1997 to,

create and sustain operational access to the Iranian nuclear target. [Merlin’s] goals on behalf of [CIA] will be to gain insight into the stage of development of the Iranian nuclear program and to collect [redacted] information on their contacts with foreign nuclear scientists. Asset will also be involved in the ultimate operational objective of delivery and/or design of a piece of nuclear equipment needed by the Iranians. (Exhibit 5)

As far as we can tell, Merlin’s outreach to Iranian scientists never developed substantive responses, much less insight into their alleged nuke program.

By May 1997, the focus had shifted even more to the blueprints (Exhibit 6).

The goal is to plant this substantial piece of deception information on the Iranian nuclear weapons program, sending them down blind alleys, wasting their time and money, and discrediting Russian designs and equipment in their eyes. The terminology and list of parts are sufficiently specific that we stand a good chance of learning whether the Iranians have in fact adopted the design and trying to make it work.

This seems to suggest this operation was, in part, about trying to get the Iranians to adopt a parts list that would require they purchase in the US, which would be far easier for the CIA to track and thereby monitor Iranian progress. (Such a plan also seems similar to the monitoring of things like aluminum tubes the US was doing before the Iraq War, with all the implications of that.)

That was largely the goal as laid out in the December 1998 cable (Exhibit 16) where Bob S laid out the goal to approach Iranian Subject 1, who apparently was in the process of being assigned to serve as Iran’s IAEA Head of Mission in Vienna (see Exhibit 3). That cable is notable, however, for its judgment that,

The major hurdle here is that neither we nor [Merlin] want him to go to Iran, which would almost certainly be their request. But if we have planted the information and strung them along a bit before facing this issue we would be prepared to let the operation end at that point if necessary.

This admits that the point was dealing the blueprints, and the CIA would even have Merlin balk on an offered nuclear deal — which surely would have alerted the Iranians — if the Iranians asked him to travel to Iran.

In a cable (Exhibit 46) planning the replacement of Sterling (and including 3 offices besides Vienna that were working with local liaison services — one of which would surely be Tel Aviv– to track any Iranian response), Bob S reiterated that goal. “The goal of the operation is to waste as much Iranian nuclear weapons expertise and money as possible.”

Curiously, the cable describing the handover from Sterling to Stephen Y (Exhibit 47) also notes that Merlin and his wife would be taken to a [CIA] setting to go over issues Merlin and his wife had covered in their initial debrief, which should cover more systematic Russian bomb construction.

All of which is to say for the period covering Sterling’s involvement, the story remained consistent. The idea of planting the blueprints was about wasting Iranian time (not unlike the StuxNet plan, though via different means).

What was CIA intending with its Iran approach in 2003, and what really happened with it?

The same cannot be said about the CIA’s plan to use Merlin to approach the Iranians again, after 3 years of having gotten absolutely no response to the first outreach (or, implicitly from Merlin’s testimony, from any of the other countries Merlin approached), as captured in a March 11, 2003 cable (Exhibit 103). The cable, apparently coordinated with the Near East and Counter Espionage Divisions of CIA, is titled, “Surveilling the Iranians in City A for a Classified Program No. 1 Approach.” It describes Bob S’ plan to surveil Iranians in City A. And that’s all the explanation, aside from the indication Bob S plans a unilateral approach to the Iranians at the end of the month, pending meeting with Merlin and his handler.

Whatever this operation was, it seems a more haphazard event than even having Merlin drop off a nuclear blueprint wrapped in a newspaper. And in a mostly redacted cable, none of the unredacted discussion describes the intent of dealing off nuclear blueprints to the Iranians for a second time.

Why was CIA satisfied continually throwing out million dollar blueprints without getting a response?

As noted, with Iran, with (presumably) Iraq, and with the other country or countries which Merlin also approached, Merlin got no response.


The government introduced a stipulation (Exhibit 166) at the Sterling trial revealing the CIA had spent at least $1.5 million to develop the blueprint that Merlin wrapped in a newspaper and left in a mailbox. Presumably, there were additional costs for each time Merlin dropped a newspaper wrapped blueprint in a mailbox. We can presume at least one of those — the blueprint dropped on Iraq, which had given up its nuclear program 9 years earlier — was completely wasted, at least if the purpose was to get the target to waste money on a nuclear program.

And yet the CIA considered the Iranian drop, at least, to be a success.


Why explains the weird reception for Jeffrey Sterling’s complaint at the Senate Intelligence Committee?

Particularly given the curious status of the program involving some kind of Congressional notice, there’s some weird stuff about the Senate Intelligence Committee treatment of Sterling’s Merlin complaint.

Jeffrey Sterling went to the Senate Intelligence on March 5, 2003 to raise concerns about the operation given “current events.” He met with Vicky Divoll — a Democratic staffer Mark Zaid had contacted — and Donald Stone — who was in charge of whistleblower issues. Both staffers showed an appropriate amount of skepticism, given Sterling’s ongoing disputes with CIA, and Stone was a bit peeved that Sterling hadn’t first gone to CIA’s Inspector General. But Divoll and Stone differ about what happened next.

Divoll remembers a meeting with her, Stone, and Staff Director Bill Duhnke immediately after the meeting, with Lorenzo Goco (who covered this portfolio) being pulled in. Divoll also remembers Duhnke ordering them to write a memo right away. Note, Divoll got fired for what she claims credibly were political reasons shortly thereafter, but she blinked, a lot, during cross-examination (and she wears glasses, so it’s not a contacts issue).

Stone, however, recalls a meeting involving just him and Bill Duhnke later. Perhaps at that meeting, Bill Duhnke told him there was an investigation into some kind of compromise (CIA referred the leak on April 7 and FBI opened the investigation on April 8), though Stone insisted he didn’t know it involved a leak to the press. Worse, James Risen had tried to contact him on his direct line. And that’s why, Stone said, he wrote the report, to admit that Risen had tried to contact him but that he hadn’t spoken with him. Divoll said Stone wrote the draft and she reviewed it against her notes (though she appears to have an overestimation of her own note-taking skills). And Stone said he got rid of his notes at some point before his FBI interviews.

The thing is, Stone never got around to writing the report until April 25 (Exhibit 101), coincidentally the very same day Risen called the CIA with a completed draft of his story (Exhibit 112). And it seems no one had done any official channel follow-up on the report until someone — presumably Duhnke, though the sender is redacted, sent Goco an email on April 24 (Exhibit 110) asking about his follow-up and, the next day, instructing, “please attempt to schedule the meeting” to follow-up today. It must have been that last minute follow-up — the day before and day that Stone wrote the report — that Stone refers to when he writes,

To follow up, it was decided that at the next opportunity, the [redacted] account monitor would ask a question on the degree to which such plans are modified and the approach to making sure there is no benefit to the target or a buyer of the plans. Such a briefing is to take place in the near future.

That is, the report and the official follow-up was constructed with the FBI’s leak investigation in mind at a point when Risen already had a story done.

Which is why the details Stone provided the FBI, which would have been captured in his notes but which didn’t show up in the report, are so interesting. First, Sterling said that “they did the equivalent of throwing it over a fence,” an admission of how shoddy the pass-off of the blueprints was. Then, that one of CIA’s two assets involved “got cold feet,” an admission that Merlin almost backed out just before the trip to Vienna. And that one asset (it actually sounds like Stone might have meant Human Asset 2, the other Russian,which the records actually support) “recognized the plans were faulty.”

In other words, Sterling told SSCI a number of details that not only correspond with known details of the operation, but which show up in Risen’s book, but Stone (writing after Bill Duhnke told him of the leak investigation) didn’t include those details in the report. Stone didn’t know when he destroyed his notes, but he didn’t have them when he met with the FBI.

Why was Bill Duhnke the top suspect?

Finally, there’s Bill Duhnke, who not only didn’t testify at the trial, but didn’t cooperate with the investigation. While classified witnesses who did not testify also named Vicki Divoll, as someone who had “a vendetta” against the CIA (as a Bob Graham staffer, she would have been tied to acute criticism of CIA for missing 9/11), Bill Harlow and Special Agent Hunt both said they considered Duhnke a top suspect at the beginning of the investigation. Since that point, because SSCI Chair Pat Roberts refused to cooperate, FBI never really could have ruled out Duhnke.

What I don’t understand is why both people considered Duhnke the leading suspect, especially since he only heard of Sterling’s complaints second-hand. Duhnke was a Richard Shelby staffer (and in recent years, has once again rejoined him as a key staffer), but remained at SSCI after Shelby left in the wake of allegations the Senator had leaked details of a wiretap on Osama bin Laden’s satellite phone (which may have been a critique of CIA’s failures prior to 9/11). But as he resumed the top SSCI staff position in 2003, Duhnke staffed Roberts as the Senator showed great deference to the CIA (as well as to Vice President Cheney). And very significantly, if the Merlin operation did get more sensitive between 2000 and 2003, Duhnke would have attended whatever Gang of Four briefings in this period included staffers. For example, Bill Duhnke staffed Roberts at the February 4, 2003 briefing at which Roberts agreed to the destruction of the torture tapes, quashed oversight efforts Graham had put into place, and — according to the CIA but contested by Roberts himself — said he could think of 10 reasons right off not to exercise more oversight over torture. Having been part of Roberts’ hackery for a few months, why did CIA regard Duhnke to be hostile? And why did they think he had enough information about the operation to be able to leak it to Risen?

There were details of the story Risen had early on — that Merlin had been used (rather than might be, as Risen reported in his book) with other countries, that the fire set handoff was part of a “larger program” to sabotage Iran’s nuke program — that didn’t make it into his book but which reflected knowledge that Sterling didn’t appear to have. They would also seem to reflect larger concerns about the program that had to come from someone with more visibility into what the CIA was doing. Did CIA know overseers at SSCI had such concerns?