An information system never lies to its CEO. It is narrated to its CEO. Every audit, every dashboard, every internal report is a deposition, not a statement of fact. We apply the unreliable narrator grid to journalism, to courtroom testimony, to memoir, without any trouble at all. We refuse to apply it to the technical report we just commissioned, because we believe the technical report is somehow different. It is not.

Reports do not lie. They are narrated.

When a CEO opens an IT audit, the reading posture is almost always the same. The document is treated as a state of the system, not as the deposition of the person who wrote it. The figures are scanned. The risk matrix is glanced at. The conclusion is internalized. And the entire reading is done in a frame that assumes the author had no stake, no angle, no positioning interest.

That frame is wrong. The author always has a stake. Not because auditors are dishonest. Because auditors are human, embedded in a market, paid by someone, scored by someone, recruited again by someone. Every audit is written knowing who will read it and who will commission the next one.

This is not corruption. It is positioning. The same word, the same finding, the same recommendation can be phrased twenty ways. Each phrasing is a choice. Each choice carries a small directional vector. Twenty small vectors aligned in the same direction become a story. The CEO reads the story and calls it a state of the system.

Wayne Booth gave us the word in 1961. We never applied it to the audit report.

The literary critic Wayne C. Booth defined the unreliable narrator in The Rhetoric of Fiction in 1961. A narrator, Booth wrote, is reliable when he speaks for or acts in accordance with the norms of the work, and unreliable when he does not. The definition was technical, careful, and meant for fiction. It traveled. It traveled to cinema, to non-fiction memoir, to legal testimony analysis, to oral history.

It never traveled to the corporate technical artifact.

Read a critical review of a presidential biography, and you will see the unreliable narrator grid applied paragraph by paragraph. Watch a film studies seminar discuss Bryan Singer and Christopher McQuarrie’s 1995 film The Usual Suspects, and you will see the same grid applied to Verbal Kint’s voiceover from minute one. Open a courtroom transcript, and you will see attorneys train juries to look for the gap between what the witness says and what the witness has reason to say.

Now open an IT audit. Open the quarterly system review. Open the slide deck the integrator presented at the last steering committee. Where is the unreliable narrator grid? Nobody applies it. Not the CEO. Not the CFO. Not the board. The audit is treated as a state, and the state is acted upon.

This asymmetry is not a small thing. It is the central blind spot of executive decision-making on technical matters. A CEO who would never accept a single-source story from a journalist, and who would interrogate any witness in a deal negotiation for hidden interest, accepts the audit report as if it had no narrator at all.

Why technical reports escape the unreliable narrator grid

There are three reasons the grid does not get applied, and all three are worth naming.

The first reason is that technical reports use the visual codes of objectivity. Numbers. Tables. Maturity matrices on a five-point scale. The visual codes mimic the look of measurement, and measurement is culturally coded as neutral. The CEO who would smell positioning in a paragraph of prose loses the reflex when the same positioning is hidden behind a green-yellow-red column.

The second reason is that the CEO is, by self-assessment, not technical. The lack of subject matter expertise creates a reading posture that is structurally deferential. The audit author is the expert. The expert is treated as the source. The narrator-as-source is exactly what Booth told us, in 1961, to never assume in any narrated artifact.

The third reason is more subtle. It is that the alternative is exhausting. To read an audit report as a deposition is to commit to the work of reading between the lines. It is to ask who paid for the work, who commissioned the framing, who chose the questions, who decided which findings would be in the executive summary and which would be in appendix C. Most executive readers do not have the time, the inclination, or the protected attention to do that. They want a state. They are given a story.

The luxury group that did not know which side to grab

I spent some time, a few years ago, inside a global luxury group trying to harmonize a single technical standard across its subsidiaries. The standard was internal. It was supposed to be the same in Paris, in Milan, in Hong Kong, in New York. It was not.

Each subsidiary told me, over several months, a coherent version of how their implementation of the standard worked, why it deviated from the others, and why their deviation was the legitimate one. The CTO of one subsidiary explained that historical sourcing constraints made his version the original. The CTO of another explained that volume scale forced his version to be the reference. A third explained that regulatory pressure in his geography made his deviation a compliance requirement, not a deviation.

Four versions. Four internal coherences. None describing the same thing.

For months, the central architecture team tried to broker consensus. The brokering followed the natural pattern. Workshops. Comparative documents. Side-by-side tables. Joint sessions. The pattern produced something predictable. Each subsidiary, when asked to describe the standard, doubled down on its own version, because each subsidiary had a deposition to defend.

What ended the deadlock was not consensus. It was not a vote. It was not a synthesis.

It was a formal mandate.

A central architecture committee, convened with formal sponsorship from the group’s executive committee, ruled on the standard. Not by aggregating the four versions. By stating, with authority and on the record, what the standard would be from a defined date forward. The committee did not ask the subsidiaries which version they preferred. The committee stated which version would hold.

The result was immediate. The subsidiaries did not love it. Two of them filed structured objections. But the four narratives collapsed at the moment the standard became a mandate. The story each subsidiary had been telling itself, with full internal coherence, lost its narrative authority because a different authority had been formally constituted.

This is the deep mechanism. Narrators hold their narrative as long as nobody else holds anything. A formal mandate, properly sponsored and properly documented, takes the narration away from the narrators. The standard stops being told. It becomes stated.

The CEO who was watching this from a distance learned something he had not learned in twenty years of running the group. The fight was not between four technical positions. The fight was over who held the right to narrate the standard. The committee did not win on technical grounds. It won by becoming the legitimate narrator.

The detail that does not fit

In Bryan Singer’s film, Detective Dean Kujan listens to Verbal Kint’s story for the entire runtime. The story is internally coherent. The witnesses he can check check out. The narrative does not snag. Kujan validates, in his own mind, almost every beat of it.

What breaks the story is not a contradiction in the story. It is a detail in the room.

A coffee cup falls. The detective looks down. He looks at the bulletin board behind his chair, sees the names of cities and people that match too perfectly the names Verbal Kint just produced. The detail that was always in the room, that Kujan had been looking at and not seeing, suddenly becomes the only thing that matters.

The unreliable narrator does not get caught by interrogating the narration. He gets caught by stopping the interrogation and looking at what was always already in the room.

This is the operational lesson for a CEO reading a technical report. The most useful work is not to demand more reports, more audits, more sources, more triangulation. Triangulation gives you three coherent stories. The most useful work is to slow down at the detail that does not fit, the figure that does not align, the silence in the report on a subject that should obviously be there.

The technical artifact almost always contains the detail that does not fit. It is almost never in the executive summary. It is almost always in a footnote, in an appendix, in a sentence that begins with “we did not have access to” or “this dimension was out of scope”. The CEO who learns to read those sentences first learns more in fifteen minutes than the executive summary can give in two hours.

What changes when a CEO becomes Kujan

The transformation I am describing is not technical. It is a reading posture.

A CEO who reads a technical report as a state of the system will, over years, accumulate a series of decisions taken inside narratives constructed by people whose interests were never on the table. The decisions will look defensible. They will be auditable. They will be presentable to a board. They will also, in aggregate, slowly diverge from any state of the system that exists outside the narration.

A CEO who reads a technical report as a deposition stops doing that. Not because the CEO becomes paranoid. Because the CEO becomes a reader of the kind Booth described in 1961. A reader who assumes that every narrator has a position, and who treats the narration as a piece of evidence about the narrator as much as about the subject.

This posture does not require technical expertise. It requires a frame. The same frame the CEO already uses on memoirs, on press releases, on courtroom testimony, on competitive intelligence about other companies. The same frame nobody currently applies to the artifacts produced inside the CEO’s own organization.

Applying the frame inside the organization is what changes the decision quality. Not more audits. Not more dashboards. Not more cross-checking. A different kind of reading.

The moment a CEO becomes Kujan, the room changes. The narrators sense it. The reports start to be written differently because the writers know they are being read as narrators, not as instruments. The dashboards become more honest, because the green-yellow-red is no longer accepted as a measurement and is instead questioned as a framing choice.

This is the work. Not commissioning more. Reading what already exists with the grid that exists, and that we already use on every other narrated artifact we encounter.

Sources

  • Wayne C. Booth, The Rhetoric of Fiction, 1961 (2nd ed. 1983). University of Chicago Press. https://press.uchicago.edu/ucp/books/book/chicago/R/bo5965941.html
  • Wayne C. Booth, The Rhetoric of Fiction, 1961, digital archive (full text, see definition of reliable/unreliable narrator pp. 158-159). https://archive.org/details/rhetoricoffictio00boot
  • Unreliable narrator, encyclopedic article tracing the concept from Booth to its extension in film and television, 2026. https://en.wikipedia.org/wiki/Unreliable_narrator
  • Confirmation bias, encyclopedic article on the cognitive bias of selectively interpreting evidence consistent with prior expectations, 2026. https://en.wikipedia.org/wiki/Confirmation_bias
  • Authority bias, encyclopedic article on the tendency to attribute greater accuracy to authority figures, with reference to Milgram 1963, 2026. https://en.wikipedia.org/wiki/Authority_bias
  • The Usual Suspects, encyclopedic article on the 1995 film directed by Bryan Singer, screenplay by Christopher McQuarrie, 2026. https://en.wikipedia.org/wiki/The_Usual_Suspects