Unknown Knowns
The unknown knowns are the fourth quadrant of a 2×2 epistemic taxonomy popularized by US Secretary of Defense Donald Rumsfeld in 2002 — the quadrant Rumsfeld himself left out. Where Rumsfeld named three categories of (un)knowledge, later writers (Slavoj Žižek most famously) supplied the missing fourth: things we don’t know that we know. The concept sits at the intersection of epistemology, psychoanalysis, and organizational theory, and has no single agreed definition — which is itself part of its interest.
Rumsfeld’s Taxonomy (2002)
At a US Department of Defense news briefing on February 12, 2002, Rumsfeld responded to a question about the lack of evidence linking Iraq to the supply of weapons of mass destruction to terrorist groups:
“Reports that say that something hasn’t happened are always interesting to me, because as we know, there are known knowns; there are things we know we know. We also know there are known unknowns; that is to say we know there are some things we do not know. But there are also unknown unknowns — the ones we don’t know we don’t know.”
The remark was widely mocked at the time (it won a 2003 “Foot in Mouth” award from the Plain English Campaign), but philosophers, risk analysts, and decision theorists have since defended it as a genuinely useful classification. The taxonomy structures thinking about risk: known unknowns are risks you are aware of (a cancelled flight), while unknown unknowns are risks so unexpected they were never considered. The framing echoes an older tradition — the terminology was in use in U.S. defense procurement by the late 1960s (a 1968 Aerospace Industries Association study contrasts “known unknowns” with “unanticipated unknowns”; contractors used the shorthand “unk-unks”) and was standard vocabulary inside NASA. Rumsfeld’s own 2011 memoir Known and Unknown traces it to a phrase he heard from NASA administrator William Graham; similar formulations appear in Ernest Hemingway’s Death in the Afternoon.
The conspicuous gap: a 2×2 of knowing vs. not-knowing on both axes has four cells, and Rumsfeld named only three.
Žižek’s Fourth Quadrant: The Disavowed
Psychoanalytic philosopher Slavoj Žižek supplied the most famous filling of the fourth cell in a 2004 essay on the Abu Ghraib scandal (“What Rumsfeld Doesn’t Know That He Knows About Abu Ghraib,” In These Times, May 2004). For Žižek, the unknown knowns are:
“…the disavowed beliefs, suppositions and obscene practices we pretend not to know about, even though they form the background of our public values.”
On this reading — drawing on Freud’s unconscious and Lacan’s “knowledge which does not know itself” — the unknown known is not merely unnoticed but actively repressed: knowledge that would be too destructive to the ego or the status quo to acknowledge. Abu Ghraib, Žižek argued, was not an “unknown unknown” (an unforeseeable threat) but an unknown known — practices everyone in the chain of command effectively knew about and pretended not to.
Žižek’s version is rhetorically powerful but arguably mismatched with the logical structure of Rumsfeld’s quote: it describes refusal to acknowledge rather than genuinely not knowing that one knows. Critics have noted it reduces, in practice, to “things everyone knows but no one says in polite company.”
Alternative Readings of the Fourth Quadrant
Several other disciplines supply their own fourth quadrant, and the term now does multiple jobs:
- Tacit knowledge (Polanyi, 1966). In The Tacit Dimension, Michael Polanyi founded the study of tacit knowledge on the dictum “we can know more than we can tell.” We recognize a face among a million without being able to say how. Tacit knowledge is knowledge we possess and use but cannot articulate or access propositionally — a literal sense in which we “don’t know what we know.” This is Polanyi’s paradox: the skills and intuitions that are hardest to verbalize are precisely the ones hardest to formalize, teach, or automate.
- The Johari window (Luft & Ingham, 1955). The interpersonal-awareness model built by Joseph Luft and Harrington Ingham crosses “known to self” with “known to others,” yielding four panes: open (known to self and others), blind spot (known to others, not self), façade/hidden (known to self, not others), and unknown (known to neither). Its blind spot — what others know about us that we don’t — is a structural cousin of the unknown known, with the axes transposed: Rumsfeld’s grid is epistemic (knowledge × awareness of knowledge), the Johari window is social (self × others).
- Latent knowledge in organizations and culture. A more operational reading: unknown knowns are knowledge that exists somewhere — in an organization’s silos, in a culture’s archive, in the implications of existing data — but is inaccessible at the point of decision because of politics, deference to authority, disciplinary boundaries, or the simple fact that no one thought to ask. On this reading, many scientific “discoveries” are really recombinations of previously separate knowns, and surfacing unknown knowns is the low-hanging fruit of progress. Writer Venkatesh Rao (“Unknown Knowns,” 2024, building on his essay “A Camera, Not an Engine”) maps this quadrant onto generative AI: large models trained on humanity’s recorded output are engines for filling in the latent space of what is already collectively known — turning unknown knowns into known knowns — without expanding the actual frontier. He also connects the concept to Fintan O’Toole’s “unknown known” of collective national identity (Ireland’s “genius for knowing and not knowing at the same time”), where it names repressed collective trauma maintained as useful fiction.
- Errol Morris’s version. In his 2013 documentary The Unknown Known (built around interviews with Rumsfeld himself), Morris initially defines unknown knowns as “the things you think you know, that it turns out you did not” — false certainties — before shifting toward “things that you know, that you don’t know you know.” The false-certainty reading inverts the quadrant’s direction but captures a real epistemic failure mode (see the Dunning–Kruger adjacent territory).
Why the Quadrant Matters
The enduring appeal of the unknown known is that each discipline’s version identifies a distinct failure of access to existing knowledge:
| Reading | What blocks access |
|---|---|
| Žižek / psychoanalysis | Repression — acknowledgment threatens identity or power |
| Polanyi / tacit knowledge | Inarticulability — knowledge is embodied, not propositional |
| Organizational latent knowledge | Structure — silos, hierarchy, unasked questions |
| Rao / AI-era epistemics | Latency — knowledge implied by data but never synthesized |
| Morris / false certainty | Overconfidence — the “known” was never actually true |
A practical consequence: unknown knowns are among the most vulnerable forms of knowledge. Tacit knowledge dies with its holder; siloed knowledge is lost in reorganizations; disavowed knowledge is expunged rather than archived. Their loss is felt acutely even when unnameable — a cultural-scale version of the tip-of-the-tongue state.
Cross-Domain Connections
- post-normal-times: Post-normal science is precisely the regime where the unknown-unknowns cell dominates decision-making (“facts uncertain, stakes high”); the unknown-knowns literature asks what we already know but can’t use.
- vuca: VUCA’s “ambiguity” component overlaps with unknown knowns — multiple interpretations of knowledge already in hand.
- filter-bubble: Algorithmic curation manufactures unknown knowns at scale: knowledge that exists in the corpus but is never surfaced to the person who could use it.
- scientific-idea-diffusion-decline: Empirical evidence that cross-field idea diffusion is slowing — the organizational-latent-knowledge reading measured in citation data.
- technological-singularity: Rao’s “camera, not an engine” thesis is an anti-accelerationist claim: AI fills in the map behind the frontier rather than expanding it, making true exploration (turning unknown unknowns into known unknowns) more scarce and valuable.
Sources
- wikipedia-there-are-unknown-unknowns
- 2002
- 2004
- wikipedia-polanyi-paradox
- wikipedia-johari-window
- 2024