Risk Detection ── Reading What Is Not Written

The hardest part of material review is not spotting a rule violation. It is catching the information that should be there but is missing, and the quiet "meaning between the lines" that an arrangement creates. Risk detection means reading not the letters on the page but the doctor's head after reading it — picturing what was left behind. That is why its passing bar sits higher than any other skill's. It is like a cook tasting a dish and noticing the missing salt: you guess at what is not on the plate, not what is.

Omission and Hinting — The Danger Lives Outside the Words

Knowledge (Part 1) and intelligence (Part 2) both worked on something right in front of you: a rule, a case, the question of how to handle it. Risk detection steps one foot outside that frame. The source defines it as "the ability to sense regulatory and ethical danger hidden in an expression, whether it is clearly written or merely hinted at." The key phrase is "whether — or." You notice not only written violations but also dangers that were never put in words.

The danger splits into four. (1) The words themselves break a rule (explicit). (2) Each sentence is lawful, yet emphasis or arrangement creates a particular impression (implied = meaning between the lines). (3) Estimating, in advance, what the reader ends up taking away. (4) Noticing that information that should be there is "missing" (omission). The last two never surface no matter how many times you trace the page, because they were never written. In a health-checkup analogy, reading the test numbers is (1)(2); a doctor noticing "this test was never even run" is (4).

Rebuilding the Reader's Head

The key to strong detection is shifting where you stand. The reviewer uses "what the other person receives" as the measuring stick, not "what I read." The source spells out the concrete behavior. Say aloud and rebuild "what stays in the doctor's head after finishing this material." Notice that safety information is "missing" next to the efficacy graph. Check whether the graph's axis, arrow, and layout make something with no real difference look as if there is one. Watch whether the overall flow suggests an unapproved use, even when each sentence is lawful.

Every one of these turns the object over and stands on the reader's side. It is not grading the writing; it is seeing, in advance, the change the writing works on someone else's head. In a soccer-defense analogy, it is reading not the ball (the written words) but the open space the opponent is about to run into (the impression left in the reader). A drug company's information can pull a doctor's or patient's reading away from the facts — catching that point ahead of time is this skill's job. Why ahead of time? Because once it is out, the damage cannot be taken back.

The ability to sense regulatory and ethical danger hidden in an expression, whether clearly written or merely hinted at. (Source, 03, Definition)

The Four Types — "All-Theory" as a Single Wing

Lay this skill out on two axes. The across-axis is "how wide you look" (only the written words, or also hints and gaps). The up-axis is "how deeply you think" (just matching one expression at a time, or generalizing the trick of impression management into a type). That makes four boxes, but the true path is only the two ends of the diagonal. Bottom-left (shallow, narrow) is explicit detection (L1): it catches only written violations. Top-right (deep, wide) is typology discovery (L4): catching the between-the-lines meaning and the gaps of the whole piece, and putting a new danger pattern into words.

The trouble is the other two boxes — and the deep-but-narrow one, armchair detection (all theory, no practice), is the worst. The person can recite the trick of impression management as theory. "Manipulating the axis," "steering by emphasis" — they can classify it. Yet on one actual page, they cannot pick up the real hint. Fluent in theory, wrong on the object. People who only sat through training fall here. The opposite, over-detection (wide but shallow), reacts to any expression but cannot sort it into types, so precision is low. Like the boy who cried wolf, they call everything "dangerous" and stop being trusted. Measured on a single straight L1-L4 ruler alone, both get lumped into "intermediate" and disappear. Only by splitting into two axes can you diagnose all-theory and over-detection as opposite "unfinished forms." In a school-test analogy: the kid who can explain the method perfectly but scores nothing on the real exam (all theory), versus the kid who blindly attempts every question but gets few right (over-detection).

A Concrete Case — A No-Difference Graph Highlighted with an Arrow

The source fixes a single scene for judgment: a material that takes the graph of a trial which showed "no difference" on its main endpoint and highlights the crossing point with an arrow. How L1 through L4 each handle that one page is where the steps stand out clearly. The point: the wording says neither "superior" nor "No.1." So anyone who only watches the words walks right past it. In a map analogy, every road is drawn correctly, yet the placement of one arrow makes you think "this way is the shortcut."

L1 watches only the words. Nothing is written, so it lets the page through. L2 reacts to flashy words but cannot reach a wordless hint like this graph. Only at L3 does the reviewer see, with zero offending words, that the whole arrangement "makes a no-difference look superior." L4 does not let that end as one person's call. It adds the very viewpoint — "even real data can become steering, depending on presentation" — as a new check item, so other reviewers can look with the same eye. The leap from L1 to L3 is the switch from reading the words to reading the reader's head. Why is the switch needed? Because searching inside the words alone will never find a danger that lives outside them.

Why the Passing Bar Sits Highest Here

Perception has four skills (knowledge, intelligence, risk detection, intuition). Among them, this one's passing bar is the highest. The reason lies in the nature of omission and hinting. A plain violation, even if you miss it, may still be caught by someone else or a later step — it stays on the page as words. But omission and hinting pass straight through unless you catch them on the spot. What is never written snags no search and no checklist. There is only one net, and it is this reviewer.

So a reviewer who stops at L1 (words only), however busy they look, is near zero on this skill. Watching only the words is dangerous precisely because it is hard to tell apart from "not really looking." To claim a pass you need at least L3 — the level that notices the between-the-lines meaning of the whole arrangement even with no offending words. All-theory detection (deep but narrow) is easily mistaken for L3 because the talk is smooth. But since it cannot pick up the hint on the object, it has not reached the passing bar.

Here, keeping three words apart pays off especially well. "Scale" is the L1-L4 graduation (the ruler). "Level" is where a given reviewer actually reads on that ruler (where they are now). "Divergence" is the gap between self-rating and others' rating. People stuck in all-theory tend to over-report themselves as L4, opening a wide gap between self and others. In a thermometer analogy: the marks are the scale, the "37°C" reading is the level, and the difference between your thermometer and someone else's is the divergence. Calibration (lining up the rulers) works only when, before deployment, reviewers compare "model real examples" — real material samples like this graph — with each other. Only once those models are in place does this ruler become a shared standard across departments.

Risk detection is the hardest to diagnose and has the highest passing bar among the four skills of perception. Stopping reading the words and starting to read the change those words work on someone else's head — the leap from L1 to L3 is a difference of where you stand, not how much ability you hold. All-theory people, being smooth talkers, can easily pretend that leap is done.

So the call is made on the object, not on the talk. Compare reviewers on the one page that highlights a no-difference graph with an arrow, as a model. Not confusing "nothing written, so no problem" with "the arrangement hints at superiority" is where lining up the rulers for this skill begins.