Bending the Facts Is Far Heavier Than a Plain Mistake

Some failures are light, others heavy. A typo can be fixed, but a wrong understanding stays in a patient's body. This installment treats the asymmetry of harm: a persuasive false impression is far heavier than plain correctness.

The Same "Failure" Carries Different Weight ── Picture Airport Screening

Recall airport baggage screening. If a screener forgets to confiscate a water bottle, usually nothing happens. But if they miss a handgun, someone on board may die. The same "miss," yet the harm left behind differs enormously. This is the asymmetry of harm: a lopsidedness where the damage, when it occurs, is extreme on only one side.

Failures in promotional materials work the same way. Small type, a hard-to-read chart, a long-winded explanation ── these are dull failures that annoy the reader. But they can be fixed. By contrast, a failure that tweaks how data is shown and makes a doctor wrongly believe "this drug works" has already moved prescriptions by the time it is corrected. The result stays in a patient's body. So whoever decides pass or fail must not count the two kinds of failure as equal weight.

Moving the Axis ── Over-Styling a Food Photo

Recall the menu photos at a restaurant. The dish is shot large, steaming, with extra shine ── made to look a little better than reality. That much, everyone forgives. But if they photograph another shop's dish and present it as their own signature plate, that is fraud. There is a clear line between clever presentation and swapping the facts.

In the reported cases, that line was crossed. For a primary endpoint (the measure that decides whether a drug works, the most important one in the trial), the official product information used a normal scale, yet the explanatory slide alone stretched part of the vertical axis to make the gap between drugs look larger. In another case, a survival curve (a line showing the share of people still alive over time) was started from 0 on the vertical axis where it should have started from 0.8, making two drugs look no different. Not a single number was rewritten. Only the axis was moved. Yet the image left in the receiving doctor's mind differs from the fact.

This is not a dull failure. It is heavy precisely because it is skillful. With a hard-to-read document, a doctor braces and thinks "this is confusing." But a beautiful chart with a tidy axis delivers the wrong understanding without leaving any room for doubt.

Why Do Makers Do It ── They Don't Think They Lied

What is frightening is that most such makers do not think "I lied." Four psychologies work behind it. Let us take them in turn.

First is motivated reasoning. When the conclusion "I want to sell this" comes first, how the data is read and shown is pulled, unnoticed, toward that conclusion. The person who moved the axis sincerely believes they "just made the gap easier to see." Second is local rationalization: "only this slide," "only this time" ── deviating at one point while the whole is correct. The case where the product information used a normal axis but only the explanatory slide was moved is exactly this. Third is the sin of omission. They prepare no material for the primary endpoint and explain only the secondary item that happened to show a significant difference (a difference too clear to be explained by chance). They escape by not telling. Fourth is externalizing responsibility. Though the Japanese subgroup showed no difference, they explained that "a difference is appearing," and when challenged, they handed responsibility to authority: "the professor says it's fine too."

Don't Undervalue the Dull-but-Correct Piece ── A Health-Check Analogy

Recall a health checkup. The result sheet is plain ── no color, no decoration, just rows of numbers. Yet that plain sheet reflects your body correctly. A good-looking supplement ad, by contrast, is beautiful but not necessarily the truth about your body. Which one you should trust to live by needs no thought.

Materials are the same. High-fidelity, low-design ── correct on the facts but poor at conveying ── is admittedly unsatisfying. The figures are clumsy and the key points are hard to grasp. But what is written there can be checked by returning to the source. Design skill can be taught and grown later; readability can be fixed through training.

Against this, low-fidelity, high-design ── facts bent but only the presentation skillful ── is the most dangerous. In the two axes this section repeats, fidelity to fact (grounding) sets the ceiling on conveying skill (reach). Grounding is the ceiling. However high the design, if the underlying fact is off, that skill turns wholly into a power that carries a false impression. So an evaluation that ranks "the dull but correct person" below "the skillful but risky person" has the order upside down.

At Pass/Fail, Measure by Weight ── A Driver's-License Mindset

Recall a driver's license. You can pass even if you are bad at parallel parking. But you do not hand a license to someone who ignores a stop sign. Clumsiness improves with practice, but the habit of not stopping where one must stop kills people. So the test weighs each item differently.

The same mindset is needed when judging a material maker. Readability and the beauty of figures are bonus items ── lacking now, growable later. But not bending the facts ── being able to return to the source, not breaking the balance ── is a necessary condition that fails the moment it is missing. We call this the non-compensable gate: a gate that does not allow other strengths to fill that hole. Skill at persuasion cannot fill a hole in source grounding.

Measure light failures by bonus points; measure heavy failures by the floor (the necessary condition). A persuasive false impression is always heavier than plain correctness.

So the conclusion of installment two is this. Do not choose people by an average score. Fairly value the dull-but-correct work, and reject the skillful-but-fact-bending work all the more strongly because of that skill. Building the asymmetry of harm into the pass/fail measure ── that is the second key to spotting the person you can trust to make materials alone.

If we flatten failures and count them evenly, a skillful false impression ends up rated above plain correctness. That order is reversed. Work that tweaks the axis to show a gap beautifully deceives the reader far more deeply than work that is merely hard to read.

So at pass/fail we measure the two kinds of failure separately. Judge light failures by bonus points, and judge the heavy failure of bending the facts by the floor. Skill at persuasion cannot fill a hole in the source ── setting this single point as the measure is the foundation for spotting the person you can trust to make materials alone.