You can say something clearly and still not be understood. Understood, in the sense that matters in a decision making forum: whether the reasoning behind the recommendation is accurately read.

The recommendation is clear. The words are plain. But the reasoning that produced it remains opaque to the people who need to weigh it against all the other options.

This is not a communication failure in an ordinary sense. The surface behaviours worked. What didn’t work is something the word “clarity” doesn’t reach.

We need to start with the mechanism.

Decision-makers, especially in senior forums, cannot directly evaluate complex advisory reasoning. Not because they lack intelligence or interest, but because full reconstruction of complicated reasoning is not possible under the conditions of a real executive meeting. Time pressure is structural. Multiple inputs are competing simultaneously. The opportunity for extended interpretation is limited.

So decision-makers don’t try to reconstruct the reasoning. Instead, they build an estimate of the quality of that reasoning from whatever perceptions already exist and any signals that are available in the moment, and they allocate weight on the basis of that estimate. The reasoning is not what provides the weight allocated in the decision, their quality estimate does. That process is inferential compression. It is not a failure of the people involved but part of an adaptive response to the limitations under which they are working.

The question is what are the signals that inform the weight estimate and how accurately do they reflect the reasoning behind the recommendation?

That is what legibility is about.

Legibility is the degree to which an observer can trace the structure of the integrative reasoning from what is visible.

The observer doesn’t need the full reasoning laid out in sequence. They need enough of the architecture to form an accurate attribution of its quality. The question legibility answers is: does what is visible give them that?

When legibility is low, the signals available to inferential compression are surface cues. Tone. Apparent confidence. Fluency. Credentials. These are real signals. They are also weakly correlated with reasoning quality, and the attribution that follows from them is correspondingly inaccurate.

When legibility is higher, what is visible includes elements of the reasoning structure itself: the constraints that were held, the assumptions that bound the recommendation, the trade-offs that were adjudicated. Compression operating on those signals produces attribution that is closer to accurate.

Legibility is what determines how accurately that compression operates.

Now the part that matters most for how this concept is used, and where most of the confusion lives.

Legibility is not clarity. Clarity is a surface property: the quality of the language, the organisation of the argument, the economy of expression. A clear recommendation is one you can follow without effort. A legible recommendation is one whose underlying structure can be traced by an observer. These are related, but they are not the same and improving one does not automatically improve the other. A perfectly clear two-sentence summary can leave the reasoning’s architecture entirely invisible.

Legibility is not confidence. A professional who expresses a position with authority and precision has clarity and composure. Whether the structure behind the position is visible to the observer is a separate question. Confident delivery doesn’t put the reasoning into the signal. It produces a different kind of surface cue.

Legibility is not brevity. Concise recommendations are easier to receive. They are not necessarily easier for the decision maker to trace the reasoning. In fact, extreme conciseness often trades depth of thinking for accessibility, which is sometimes the right call, but it is a trade off.

Legibility is not communication skill. This is the most important distinction. Communication coaching improves how information is conveyed. It is content-agnostic: a better communicator conveys anything more effectively. Legibility concerns whether what is conveyed allows accurate tracing of the underlying reasoning. These are different questions. If the reasoning structure isn’t in the signal, better delivery doesn’t put it there. Improving how you say something does not change what inferential compression has to work with.

What legibility actually requires is a different kind of output: one that makes the architecture of the reasoning available, not just the conclusion it produced.

A recommendation that names the constraints it worked within, the assumptions it made, and the trade-offs it resolved gives the compression mechanism something real to work with. The observer doesn’t need access to the full integration. They need enough of the architecture to attribute reasoning quality accurately rather than inferring it from surface signals.

Legibility doesn’t require articulating the full integration process. That can’t be done, even by the person who did it. What it requires is making the structure visible: the constraints engaged, the assumptions made, the shape of the trade-off resolved.

One further thing legibility is not. It is not the construction of visible architecture where the underlying reasoning doesn’t support it. That looks similar from the outside. It is a different practice.