Historical Context for Today's Surveillance Debates: The 1945 Legal Memo on What Became Operation Shamrock

By Robert Chesney
Wednesday, March 29, 2017, 2:00 PM

Section 702 is coming up for renewal later this year, and it is clear we'll be hearing a lot in that context about the impact of SIGINT collection activities on US person communications. When that topic comes up, inevitably there follows at least a brief reference to the Church Committee's exposure of Operation SHAMROCK and Operation MINARET. In light of all this—or perhaps just because I love archival finds—I'm writing this post to capture some 1940s history that helps us better understand those 1970s revelations—and how if at all they pertain to today's controversies.

The archival find in question isn't mine.  It's Steve Budianksy's.  His recent book Code Warriors: NSA's Codebreakers and the Secret Intelligence War Against the Soviet Union is a must-read volume for those interested in the history of SIGINT in general and the NSA in particular.  It’s not primarily about the legal regulation of SIGINT activities, yet there are many nuggets of legal interest in it. One that caught my eye was a reference to an August 1945 memo from the Army Judge Advocate General (Myron Cramer) to Secretary of War Henry Stimson, titled “Legality of Signal Intelligence Activities.”

As near as I can tell, Cramer’s memo hasn’t been discussed previously in legal literature, and though declassified it’s not posted online in a freely-available way. You can get it through the Digital National Security Archive if you have a Proquest subscription, though, so with the help of the library at UT I was able to get a copy.  I don’t feel free to just the post the pdf, of course, but I can at least provide a detailed review of its contents.

What was the historical context for Cramer's 1945 memo?

There was no NSA in 1945.  America’s SIGINT functions at that time remained split between distinct Army and Navy operations.  For several years, both had been running at full speed in support of the war effort.  Some of these activities actually began in the tense period leading up to the war.  For example, as Steve relates in Code Warriors, the Army in January 1940 asked RCA (the Radio Corporation of America, at the time a giant of the communications industry) to allow the Army to post a soldier to an RCA office in order to make copies on a daily basis of all the international telegraph traffic RCA transmitted into or out of the United States.  Steve notes that no one “at the time seemed overly concerned, and in any case America’s entry into World War II in December 1941 made the question of legality moot for the duration when official wartime censorship began, requiring all cables to be turned over to the government for inspection.” (Code Warriors, at 23) 

But the war would not last forever, and with its end looming (V-E Day came in May 1945, and the atomic bombings of Hiroshima and Nagasaki in early August 1945 portended the end of the war in the Pacific as well) it was foreseeable that legal obstacles might soon become more significant.  This helps explain why, sometime that summer, the Army G-2 (i.e., the assistant chief of staff for intelligence) reached out to Major General Cramer requesting advice across a broad range of collection activities. 

What sort of collection activities did the G-2 have in mind?

Cramer writes that the G-2 asked for an analysis of three sets of issues, which I paraphrase below:

(i) radio and wire intercepts targeting agents of foreign governments or agents of foreign commercial enterprises, taking account of various combinations of geographic location such as foreign-to-foreign, foreign-to-US (or vice-versa), and US-to-US communications);

(ii) receiving copies of communications to or from such foreign persons through voluntary cooperation from telecom companies like RCA;

(iii) disclosing the fruits of (i) and (ii) to other parts of the government.  

Notably, Cramer frames these questions in a way that assumes a targeted approach to collection, focused on communications to or from agents of foreign governments or foreign commercial enterprises.  The memo is not framed as a study of the legality of wholesale collection of all communications crossing a particular channel.

Why engage in collection at all, once the war is over?  

Cramer opens his analysis by addressing this question. After noting the manifest importance of SIGINT during wartime, Cramer writes:

“In time of peace, [such] information … is equally required for the national defense and the effective control of international relations, to avert the danger of surprise attack by an unfriendly nation, and to contribute to the satisfactory adjustment of the complicated differences between nations that lead to war…. The catastrophic consequences inherent in the use of the atomic bomb dramatically emphasize the necessity for our knowing as much as possible of the plans of a possible enemy.” (para.3)

So the memo focuses on the Fourth Amendment, right?

No.  In 1945, the prevailing understanding of the Fourth Amendment remained the narrow doctrine described by the Supreme Court in 1928’s Olmstead decision.  Olmstead posed the question whether government wiretapping violated the Fourth Amendment if the tap occurred outside of a person’s home or private office.  The Court answered in the negative; so long as agents did not make an unauthorized entry to search in a specially-protected physical location like the home, the Fourth Amendment had no bearing at all. 

The modern, more-flexible “reasonable expectation of privacy” conception of the scope of Fourth Amendment protections was still more than two decades away at the time Cramer wrote, and hence it did not seem to him that the activities the G-2 described would be likely to implicate (nevermind violate) the Fourth Amendment.  For this reason, the memo takes up the Fourth Amendment question only after a lengthy analysis of statutory concerns (discussed below), and for the most part treats it as an easy question involving the settled law of Olmstead.

There is one significant caveat, however.  The final paragraph of Cramer’s Fourth Amendment analysis shifts from consideration of radio and wire intercepts towards the voluntary provision by telecom companies of copies of telegrams.  Rather than relying on Olmstead, Cramer instead invokes what we know of today as the “third-party doctrine,” which the Supreme Court would recognize in the 1970s in decisions like US v. Miller and Smith v. Maryland.  The idea behind the third-party doctrine is that one cannot raise Fourth Amendment objections if the government acquires information that one voluntarily put into the hands of a third party (such as a telecom service provider like RCA).  All that said, Cramer takes note of a 1936 D.C. Circuit decisions (Hearst v. Black, 87 F.2d 36) that arose when a Senate committee (investigating allegations of unlawful forms of lobbying by Hearst and other New Deal opponents) worked with the FCC to compel telecom companies to produce copies of all telegrams sent or received in the summer of 1935.  The Court described this as a “dragnet seizure,” and found it unlawful on that basis.  This suggested a rather significant problem for the third-party doctrine argument Cramer had just advanced (and note the problem this sort of analysis poses for the post-9/11 application of the third-party doctrine to the bulk collection of communications metadata).  Cramer responded with a factual distinction: the Hearst decision “does not affect the legality of the proposed activities,” for

“here there is no dragnet seizure, but [only] a restricted examination…of only communications of foreign governments, their agents or representatives, and certain communications of foreign commercial enterprises.” (para. 33) 

The clear implication is that Cramer might have been obliged to take a different view had the question instead been framed as bulk collection of all telegrams.

What issue did Cramer instead think mattered most?

The obstacle that concerned Cramer was the Federal Communications Act of 1934, section 605 of which provided in substance that “no person” may intercept and divulge wire or radio communications on an unauthorized basis.  A similar prohibition had been in place just with respect to radio communications for some time, and in 1934 Congress extended it to the increasingly-important realm of wiretapping.  On its face, it seemed a significant obstacle to the intercept activity Cramer was called upon to address.

Cramer opened his analysis of section 605 by observing that the legislative purpose behind it focused on the actions of private persons spying on one another.  Indeed, the government previously had taken the position in litigation that section 605 should be construed to apply only to such persons, excluding the government.  The Supreme Court had rejected that view in the 1937 decision Nardone v. United States, however, based on a plain-language approach to the statutory phrase “no person.”  Section 605 forbade government agents from testifying at a trial about the contents of an intercept, the Court concluded.

This left open the question that mattered for Cramer: could government agents still conduct intercepts and use the fruits for internal purposes such as intelligence analysis? Not surprisingly, that question had never come up in a litigation setting, so there was no guidance in caselaw.

In concluding that section 605 did not forbid the intelligence-use of intercepts, Cramer first distinguished the two scenarios on what he described, somewhat strangely, as “ethical” grounds.   What he seemed to mean was this: Section 605 reflected a judgment that the law enforcement benefits of the public use of intercept evidence were not worth the costs, but that judgment did not automatically extend to other government interests, particularly those that would not involve public disclosure of the information (such as support to the conduct of foreign relations, preservation of neutrality, or protection of national security). (para. 9)

Cramer then pointed out that it was widely known in Congress that the executive branch conducted intercepts of this kind for intelligence purposes, and that when the subject came up in a prominent way during an investigation of the Pearl Harbor disaster, then-Senator Harry Truman had asserted that “under the law as it now stands the Federal government could have subpoenaed copies of all telegrams sent by Japanese spies…,” and that nothing “in the present law” prevented intercepts in such circumstances. (para. 10)

Cramer next added that Attorneys General Biddle and Jackson both had concluded previously that section 605’s “divulge” language meant that intelligence-focused wiretapping remained permissible. (para. 11)  Of course, one might object on the ground that even without public disclosure, intelligence intercepts surely involved one executive branch employee sharing the fruits with at least one other executive branch employee. Cramer argued, however, for this purpose each executive employee was simply an arm or extension of the President, and thus intra-executive sharing was not “divulging” in the statutory sense. (para. 12)

For his next argument, Cramer came back to the Nardone decision.  The Court had observed in a general way that ambiguous statutory prohibitions should not be read to reach government employees if such construction either “would deprive the sovereign of a recognized or established prerogative,” or “would work an obvious absurdity” (such as enforcing the speed limit on an ambulance in the midst of an emergency). (para. 13)  Cramer concluded that both principles applied here, but did not elaborate either point at that stage.  A few pages later he would return to the issue of sovereign prerogative, but first he digressed to make a distinct argument that flowed from, of all things, the publication in 1931 of Herbert Yardley’s book The American Black Chamber.

That book, famously, had broken new ground by publicly describing SIGINT activities during and after World War One, and it had caused much consternation by those who wanted to keep all details of such activities secret.  This was the backdrop against which Congress in 1933 passed a statute forbidding persons from publishing information about foreign diplomatic codes if the person had the information by virtue of government employment.  According to Cramer, this demonstrated Congressional awareness of, and approval of, SIGINT activities including the sort under his consideration.  Section 605 had been enacted just a year later, with no indication of an intent to change the status quo with respect to intelligence intercepts. (paras. 18-20)   

And this, at last, brought Cramer around to a final set of argument that he expanded upon at great length: that section 605 should in any event be construed to exclude this set of activities in order to avoid a potential constitutional clash with the Article II authority of the president.

What did Cramer say about the President’s constitutional authority?

It’s tempting to say that Cramer simply made an extended Curtiss-Wright argument, and leave it at that.  But if you’ve read this far, you might have an appetite for just a bit more detail than that, so here goes.

Cramer opens by noting that the President (as Chief Executive and Commander-in-Chief) does not require an affirmative grant of statutory authority to engage in foreign intelligence collection activities.  He then asserts that any statute that attempted to take that power away from him would, by extension, “be of doubtful constitutionality.”  (para. 21)

Unpacking this argument, Cramer first provides a standard recitation of relevant Article II authorities enjoyed by the President.  He then focuses in on the President’s broad power over the conduct of American foreign relations, including in passing the (controversial) claim that such power derives in part not from the Constitution of the United States but as a “heritage from the British Crown.” (para. 22)  Naturally, extended citations to Curtiss-Wright follow.  Cramer then contends that the President not only enjoys various powers under this heading, but also has an affirmative duty to exercise those powers “within Constitutional limits, of course, when the national security so demands.” (para. 24)

Against this backrop, Cramer points out a couple of occasions when presidents asserted power “to control communications between the United States and foreign countries.” First, he cited President Grant’s decision, supported by an Attorney General’s opinion, to authorize a French company to “land submarine cables in this country.” (para. 25)  (Cramer admits that a district court later denied the Presidential has such unilateral authority, and that the Second Circuit agreed with that ruling, but argues that all of that was superseded by Curtiss-Wright).  Second, Cramer cites Woodrow Wilson’s 1914 executive order barring US radio stations from carrying messages for non-neutral parties to World War One, and establishing a censorship regime to enforce that rule.  Rather remarkably, Cramer notes that the supporting Attorney General’s opinion undergirding that set of actions emphasized, among other things, that “[s]uch powers…are not likely to be abused”. (para. 26)  At any rate, Cramer concludes that if Wilson could actually censor radio communications, Truman surely can simply obtain copies of or intercept communications involving foreign governments—and he asserts that Congress arguably might be unable to limit that power.  (para. 27) 

What about international law?

Cramer concludes his memo by noting that the collection at issue is to no small extent targeting diplomatic correspondence, and thus a question arises as to whether international law protects that correspondence from espionage.  He concludes that it does not, observing that “apparently all major powers engage in these intelligence activities,” and thus if various international agreements involving the protection of diplomacy had been “intended to prohibit” such espionage, “an express prohibition thereof would have been made.” (para. 37).

What happened next, and does any of it matter today?

Let’s return to Steve Budiansky’s Code Warriors to pick up the thread of the story.  The war ended, but the telegram-copying program remained in place (perhaps in part on the strength of Cramer’s memo, though of course it’s hard to say what practical impact the memo actually had).  RCA, ITT World International, and Western Union all cooperated, but also expressed repeated anxieties about the legality of it all. Steve writes that they “sought repeated assurances from the government that the program was essential to national security and that their companies would not face prosecution for their actions.” (Code Warriors, at 66)  Secretary of Defense Forrestal in 1947 gave a personal assurance to the companies that President Truman “himself had authorized the program,” yet anxieties remained. (Id.)  In 1948, an effort to obtain affirmative statutory authorization apparently came up for discussion, but (in an especially-fine research nugget) Steve reports that one of the attorneys involved in the discussion warned that “such a legislative effort might result in “ill-advised screams from some members of the press that the bill invades freedom of the press and civil rights.” (67)  It seems that was that. 

The telegram-copy program just kept chugging along in the decades that followed, becoming “Project SHAMROCK” under the auspices of the NSA.  In the interim, some of the premises undergirding Cramer’s analysis changed.  First, the Supreme Court in Katz upended the Fourth Amendment jurisprudence of Olmstead by adopting the flexible “reasonable expectation of privacy test” in 1967 (though the question remained whether foreign-intelligence collection constituted some kind of categorical exception even then).  Second, SHAMROCK may have been foreign-intelligence focused as Cramer had assumed, yet the scope of the collection appears all along to have been bulk rather than targeted.  Third, SHAMROCK eventually became entangled with Project MINARET, whereby NSA would search its collections (including SHAMROCK content) for communications to, from, or about people—including US citizens—identified on watchlists generated by the FBI and others. 

All of this was unknown to the public for decades, but it came out in the 1970s during the Church Committee investigations.  It was the NSA’s first public black eye, and ever since it has been customary to point to SHAMROCK and MINARET as original sins, ur-examples of NSA activity unlawfully and unwisely impacting the privacy of Americans.

No small amount of the subsequent developments in the legal architecture governing NSA activities—including FISA and the US person protections rooted in 12333—can best be understood as responses to those revelations, designed to prevent this sort of thing from recurring. 

The episode continues to cast a long shadow over today's debates.  Recent events involving President Trump and Russia have revived attention to the impact of foreign-intelligence collection on U.S. person communication, and they’ve done so just when the looming sunset of Section 702 collection authority ensures Congress must re-engage on this topic.  Inevitably, we’ll hear some about SHAMROCK and MINARET.  We probably won’t hear so much about Myron Cramer, but familiarity with his part of the story does help us understand the larger SHAMROCK/MINARET story better.