Unknown unknowns, 9/11 and a useful anthropology

A caveat:

This post is inspired by the memorials, performed yesterday, for the victims of the terrorist attacks of September 11, 2001. Some readers will say that the appropriate thing to do on this occasion is to honor the memory of people who lost their lives. I respect this view. I believe this might also be an occasion for addressing the causes of terrorist attacks. It turns out, the two issues are connected.

One place anthropologists can make themselves useful is military intelligence. (That most don’t want to, that’s another question.) They can be particularly helpful with what Secretary Rumsfeld calls the “unknown unknowns.”

Unknown unknowns, these are factors we don’t know we don’t know. They are the Achilles heal of any decision making system, military or managerial. We can’t factor what we can’t fathom. We must fear things excluded from the problem set.

Sometimes, perhaps most perilously, the unknown unknowns are the assumptions with which we think. We don’t see these assumptions because, well, we’re assuming them.

That’s where anthropology comes into it: most of what the anthropologists wants to know are the cultural assumptions in the respondent’s head…hence a certainly disciplinary skill in finding these  out.

Invisible assumptions, the most pernicious unknown unknown, are trap doors an enemy may use at their discretion, a place of weakness that will never be defended against because it literally cannot be seen. Invisible assumptions played their part in the terrorist attack of 9/11. Douglas Porch and James Wirtz note that the bin Laden terrorists had discovered an invisible assumption, an unknown unknown. 

[T]he notion of "suicide bombing" is so alien to the American —indeed the Western—outlook, that we find it difficult to fathom the mindset of enemies prepared to conceive of an operation of such horrific proportions, one in which they are prepared to immolate themselves…

The 9/11 attack was hard to defend against because it violated an assumption of the American outlook.  It made combatant self sacrifice a systematic part of the attack.  Wedded as we are to the value of the individual, it was almost impossible to imagine this and other threats. 

Military intelligence cannot see some of the assumptions it entertains.  Consider the report that said some flight students were concentrating on take-off instructions and ignoring landing instructions. 

When the analyst is “crunching” vast bodies of data, each particular finding must respond almost immediately to scrutiny.   (Blink!)  The more it springs from alien assumptions, the more probably it will read as noise.  To capture the significance of the "flight student" report, the analyst would have to perform a kind of intellectual self surgery.  He or she has find and then replace the one assumption that prevents him or her from seeing the significance of the datum at hand.  And again, this has to be done in the real time race through the data that is every analyst’s necessary modus operandi. 

But this of course formidably difficult.  Chances are flight students who appear more interested in take off than landing does not “compute.”  As long as we assume the value of the individual, this datum is odd, counter intuitive, and probably noise.  The moment we make the unknown known (more exactly, the unthinkable thinkable), this piece of intelligence can now “ring a bell.” Now, the defense community is in a position to factor in, and to begin to scan for, the possibility that planes might be used as bombs. 

An American, and extra-American, point of view values the individual extraordinarily.  This was clear yesterday. Every victim of the 9/11 attacks was acknowledged individually and by name. Roughly one in five of them were acknowledged publicly, nationally, and by a member of their family.  This may be taken as a measure of the essential decency of the American approach.  But it is also the very thing that so confirms certain of our assumptions that they become unknown unknown. (Clearly, no  criticism of the memorial is intended. This is a simple  relationship: the more we use certain assumptions, the more we assume them.)

Most of what we know about invisible assumptions tells us that they are almost always impossible to identify and correct on an ad hoc, on the fly, basis. Very, very, very smart people can do this (“Santa Fe Institute smart,” probably), but the rest of us cannot capture and swap out troublesome assumptions to correct the unknown unknown problem, especially not on the fly.

The only systematic way we can deal with this problem (and I guess we can assume the Pentagon is working on this) is to task a team of people and to train them to “think like the enemy.”  It is only by routing out certain assumptions and replacing them with new ones that we can make ourselves the equal of the challenge and escape the real problem of unknown unknowns.  The anthropological task begins be decoding the enemy, discovery the assumptions he makes, building these into the skunk works team, and then recording what the team delivers as new threats that must be defended against. 

I leave for another time the dynamic version of this system, and the more daunting challenge: the system that is capable of finding and mocking up assumptions, as they shift in real time.


Porch, Douglas and James J. Wirtz. 2002. Surprise and Intelligence failure. Strategic Insights. I (7 September) here.

10 thoughts on “Unknown unknowns, 9/11 and a useful anthropology

  1. steve

    Your comment reminds me of what I read about Pizarro’s conquest of the Inca–that the Incas were unable to conceive that anyone would dare to lay his hands on the king, and so did not take sufficient precautions to protect the king’s physical security. Pizarro took him hostage and used that leverage to loot and topple the empire (it helped that the Incas’ subjects weren’t all that enthused with their lot).

    My methodological question is this: How do you know when you’ve learned to “think like the enemy?” By definition, implicit assumptions you have yet to root out are still there, undetected. I suppose you could try to run predictive tests, but given the small samples of enemy action we have to work with, I wouldn’t place a great deal of confidence in them.

  2. Tom Guarriello

    I’ve spent my entire adult life thinking about the issues you describe in this post, Grant. As a graduate student, I studied phenomenological psychology. Phenomenology is the study of experience and the assumptions that are the constituents of that experience. The philosopher, Edmund Husserl, following in the footsteps of his teacher, Franz Brentano, wrote about performing what has come to be called the “phenomenological reduction.” In the reduction, one attempts to place one’s assumptions “in bracket,” holding them in abeyance, and describing an experience in their absence.

    And so, instead of saying, “I walked into a supermarket and headed to the daily case,” one would say, “I entered a large building through a set of doors that opened before me and found myself in a huge space filled with aisle after aisle of shelves stocked full with goods; they stretched far to my left…” and so on. Of course, as this little demonstration of phenomenological description makes evident, one can never ENTIRELY bracket one’s assumptions (what’s a “door,” or an “aisle,” or “goods”?) as long as one uses language. But with practice, as you surely know, empirical description improves.

    And, using the phenomenological method known as “imaginative variation” one can begin to adopt a perspective, what we’re likely today to call a “mindset,” that *resembles* that of “the enemy,” or, more generally, “the other.” But, like the reduction described above, this is always an approximation of the other’s mind. It’s a far cry from the kind of “presuppositionless knowing” that Husserl envisioned when he wrote about conducting a phenomenology *of* consciousness. The existential phenomenologists, like Heidegger, Merleau-Ponty and Sartre, were less interested in studying consciousness as consciousness, and more in studying human experience as lived. Those studies produced great insights into the constituents of human experience and provided the foundation for a qualitative, empirical approach. If you or your readers are interested in further reading on these issues, I’d recommend the book, Psychology as a Human Science, by A. Giorgi.

    I guess that’s a long-winded way of saying that the problem our government is currently grappling with is one that philosophers and psychologists have spent centuries studying.

  3. dilys

    I find both Grant’s and the earlier commentors’ posts instructive. There is also the importance [indispensability?] of what Grant has pointed out earlier: an informant, to trouble-shoot one’s descriptions, conclusions, assigned meanings, and scenarios from “inside the model.”

    I’m also reminded of the creativity-class cliche — take an ordinary object and ask “how many ways can this object be used?” Our inner imagery was entirely identified with airplanes as tools of our safety, strategy, and effectiveness. Our earlier fear, hijacking to Cuba etc., was merely competition between travelers over a destination.

  4. Peter McB.

    As one would expect, the question of surprise events and their prediction has become a hot topic in recent years, with lots of papers and several books (one of which I co-edited) being published. This activity predates 9/11 by a few years; an interesting question is why this interest arose when it did, prior to 9/11. Indeed, this is the second time we in the west have been pre-occupied with the topic of surprise in the last century.

    The first time was in response to Pearl Harbour and the Cold War, when the enemy could launch a surprise attack which would annihilate us all. During the 1950s, western governments sponsored considerable research and even held conferences on the topic of “Surprise”. One, in 1958, included both Soviet and US representatives and sought unsuccessfully to find mutually-agreed ways to eliminate the possibility of surprise nuclear attacks by either side. Economists reading this may know the name of George Shackle, who developed at the same time a model of uncertainty built on the degree of surprise an event would cause a decision-maker.

    An historian of the 20th century may wonder why we were pre-occupied with the issue in the 1950s and again 40 years later, but apparently hardly at all in between. Was it that we became smug, thinking that we’d solved the problem of surprise nuclear attack? There may have been some of that feeling because of the various arms limitation treaties signed between the USA and the USSR. Or, was it that we found nuclear devastation too painful an idea to entertain for long periods, and so put it out of our mind? (Personally, I think this was the reason for the mostly-negative reactions to Herman Kahn’s books on nuclear war.) Whatever the reason, I think there important consequences for our ability, as a society, to ensure that we continue to focus on this issue.

    And, any concerted effort using rigorous methods, such as the ones you describe so well, Grant, to predict these events now, should be looking at the methods and experiences of our nuclear strategists and cold-war-warriors in the 1950s and 1960s to do similar. Different enemy, and different mind, but same task.

  5. Peter McB.

    And related:

    Donald Rumsfeld’s profound statement (12 Feb 2002) about “unknown unknowns” overlooked one category of knowledge: the “unknown knowns”: the things we don’t know that we know. So, for example, lots of information was known about the individual 9/11 terrorists before the attacks, which, if collated, analyzed and used quickly, may have prevented some or all of the attacks. We just did not know that we knew this information.

    Cultural assumptions we forget about are perhaps in this category.

  6. Grant

    Steve, of course, you can’t know, that’s the name of the game, but of course you surely won’t know if you don’t keep looking, a lesson for the military and corporation both. Thanks, Grant

    Tom, wow, splendid, thank you, I knew our field had something in common and now I know what, which is to say, I guess, that we both fall into the cultivated skeptism camp that the West has made pay so richly. This is one of the things the enemy cannot do…not until, as I was laboring to suggest, that actually to learn to swap assumption jamming strategies on the fly. That will be very, very scary. Thanks, Grant

    Dilys, brilliant line about dispute over destinations! Thanks, Grant

    Peter, could we have the name of that, your book, please. That’s interestng to think about, that surprise might be the new condition. I think Drucker says something about this. Or maybe it’s Peter Schwartz of GBN. Thanks, Grant

    I see your point about the assumptions question perhaps being knowns not unknowns. Anthropologically, I think they are the latter. You could interview someone forever and they likely would not report the value in question here. If they did, it would sound fatuous and we would be inclined to dismiss it as a truism. It wouldn’t actually have any value as a corrective against invisible assumptions. Thanks! Grant

  7. Peter McB.

    Thanks, Grant.

    My book is “Chance Discovery”, co-edited with Yukio Ohsawa (Springer, 2003). Here’s the amazon.com page:


    The book is part of a larger effort to find processes for automated discovery of rare events in computer decision support systems, and I think will be of most interest to computer science/artificial intelligence people. Not much marketing or anthropology in it, I’m afraid.

    One of my papers in the book is an attempt (with Simon Parsons) to develop a protocol for computer dialogues which can lead to identification of rare events. There may be connections with the ideas in your post about processes for skunk works.

  8. Grant

    Peter, thanks, looks very interesting indeed; wow, so that’s possible? a protocol that can tell the difference between novelty and noise? wonderful. we should talk. I’ve been working on the something like the same problem, supposing, perhaps incorrectly, that all novelty looks like noise in the first instances, repetition counts not just because it suggests that the thing is formed (and not noise) but because repetition is actually part of the forming process. It is establishing the type of which the novelty can then be a token. Anyhow, I have had a long talk with my accountant and I am ordering your book now. Best, Grant

  9. LK

    4 yrs after 9/11 it’s still hard to process and this post and these comments zero in on why.

    Q: When is a plane not a plane?
    A: When it’s a bomb, owned and operated by the victim, full of unwittingvictims, and pointed straight at thousands of unlucky everyday people,not to mention hundreds of millions of dollars worth of real estate, and even better than that a irrefutable symbol. ceci n’est pas un avion indeed.

  10. Peter McB.

    Grant — not sure that Computer Science is there yet, but we are on the path.

    It is interesting (in the light of your comment) that one reason for CS people to be interested in discovery of rare events was the realization (primarily by Japanese researchers around 1998-9) that standard data mining techniques begin by eliminating noise, yet the really interesting phenomena are usually found in the noise, not in the signal.

Comments are closed.