Joshua PollackSurprise ≠ Intel Failure

That’s the “does not equal” sign up there.

I overlooked something important earlier when discussing what the U.S. government knew about North Korea’s nuclear test preparations and when they knew it:

There are two possibilities. Either A) the Obama Administration saw some advantage to keeping mum, and turns out to be awfully good at keeping mum, or B) someone missed something they should not have missed.

There is an option C) as well: the intel collectors saw all the signs, but the higher-ups failed to draw the proper conclusions.

There was a scattering of leaks in the days ahead of the test, possibly from South Korean intelligence. And afterward, we learned that the IC was watching the preparations intently:

The official said that U.S. intelligence agencies monitoring the test facility had witnessed significant activity in the days before the explosion. The United States had positioned an array of high-tech equipment to monitor the test, including Pentagon aircraft equipped to collect atmospheric samples of any nuclear plume.

I believe it. But the Administration took none of the public steps one would expect to happen in advance of a test, not so much to deter the North Koreans as to build international support for a response after the fact. That led some observers to conclude that the timing of the test came as a surprise. Marcus Noland, for example:

“As much as they understood this was going to be an issue, they weren’t ready for a nuclear test in May,” Marcus Noland, an expert on North Korea at the Peterson Institute for International Economics, said of Mr. Obama and his advisers. “They’re in a situation now where they have to contain and manage a crisis.”

As noted previously, there appears to have been a firm and widely held conviction that North Korea would not test again until it had more plutonium in hand. Potential indications of an imminent test may have been discounted on that basis.

One possible result: the U.S. apparently did not inform anyone in Japan that a test was imminent. Whoops.

For just a moment, let’s turn this blog over to the learned Prof. Richard Betts, ca. 1982:

The principal cause of surprise is not the failure of intelligence but the unwillingness of political leaders to believe intelligence or to react to it with sufficient dispatch.

You see, it pays to be mindful of the classics.

X-posted from TW.


  1. MarkoB

    ah, yes, what a philosopher of science would call the “theory ladenness of observation.” Apparently Stalin and Beria didn’t believe the intel coming from Los Alamos during the war for much the same reason even though the reports they were getting were accurate enough (same goes for Operation Barbarossa).

    On North Korea I feel that the arms control community is missing the big picture here namely; we are in danger of slipping towards war. We can debate how big the Nork’s bang was but a war in Korea will be a big bang and big(er) than Iraq and Afghanistan. The Norks proliferate sensitive strategic technology, to be sure, but the PSI could provide a catalyst setting up the escalatory dynamics that might lead to war. The Norks interpret any tightening of the PSI as a cuban missile crisis style quarantine meant to tighten the screws on the regime.

    Things might get pretty ugly pretty quickly. What’s wrong with opening two party high level (i.e. not Bosworth) talks? It’s better than war, surely.

  2. krepon (History)

    there are other possibilities, no? Like US intel saw evidence of test preps, but a public warning not to test, followed by a test, would not have helped matters.

  3. Andy (History)

    There is at least one more possibility. Option D, for example, is that there was strategic warning but not tactical warning. IOW, the difference between knowing that some event is likely in the future vs predicting exactly when that event will happen. For example, we had strategic, but not tactical warning for the 9/11 attacks – we knew AQ was going to conduct an attack fairly soon, we just didn’t know exactly how, when or where.

    For North Korea, it’s likely the IC knew North Korea would conduct a test, and it’s likely the IC knew where the test would occur, but without observable preparatory indicators, predicting the “when” is very difficult. For a country like North Korea, collection opportunities are limited, to say the least. Perhaps most of the preparations were completed some time ago. In that case, there will be few observable indications of an imminent test and those that are seen occur right before the test occurs, which does not leave enough time for leaks to the NYT and WAPO to be published (but more than enough for a “critic” to reach the national leadership).

    Something similar occurred with the Indian tests – little work was required to actually conduct the tests, so there were few indications available to provide tactical warning. The IC screwed up the strategic warning by assessing that the new BJP leadership’s promises to test were just talk. Strategic mis-assessment coupled with little ability to provide tactical warning is a recipe for an intelligence failure.

    We also need to keep in mind that not everything gets leaked to the press. It’s only an intelligence failure when the national leadership is not given warning (and does not act on that warning, if applicable) – it’s not an intelligence failure simply because something wasn’t leaked beforehand.

  4. Bruce Klingner (History)

    Regarding intelligence surprises and failures: Those of us on the outside of the intelligence community (IC) (i’m former CIA who worked Korea) don’t know what the IC saw and reported to policymakers on the NK nuke test (or other issues). Too often there is a tendency for an outsider to conclude the IC was surprised since it didn’t appear in the paper. Similarly, there are instances of an outsider not being “convinced” by the one satellite photo that was released or a few commercial imagery photos. We saw examples of that on this website with the Syrian reactor. On the inside, analysts might have had hundreds of photos over an extended period of time along with corroborating SIGINT and HUMINT.

    Of course, there are instances of: policy failures (when policymakers were warned of intelligence but disregarded the message); the “paradox of the warning analyst” (when an analyst provides sufficient warning so that policymakers take action which prevents the event which they were warned of which can make the warning analyst look like ‘chicken little’); intelligence surprises (sometimes inadvertently caused by policymakers who give away imagery observables or endanger sources and methods during a previous demarche which enabled the opponent to learn to do things differently this time); and outright intelligence failures (yes, intelligence analysts don’t bat 10 for 10).

    My message is simply not to judge the intelligence analysts, IC, or US Government based on the miniscule amount of intelligence data which makes it over the wall to the outside.

  5. J House (History)

    It seems likely the White House was warned and it didn’t leak.If they weren’t, this goes as another string of IC failures this decade.
    Let’s hope not.
    I agree with Mark B.
    Just when we think we know the other’s intentions, we seem to miss every time.
    One can easily imagine a ‘gulf of tonkin’ or ‘Pueblo’ type incident to set this off into a shooting war.

Pin It on Pinterest