← All dispatches

No. 018 · October 22, 2026

The Words You Never Said

A correct answer explained in language you never used is not your answer anymore. Eloquence can quietly launder authorship.

Yesterday, a tester caught something our arithmetic never could.

The calculation was correct. The scores were the ones the person had given. The weights were theirs. The totals added up exactly as they should.

And then the explanation described their priorities as "capital preservation" and called one of their options "momentum."

They had never said either phrase.

The numbers were faithful. The words were not.

The explanation is part of the decision

We tend to treat the explanation as packaging — the friendly wrapper around the real work, which is the math.

But that is not how people read. When a system tells you why a result emerged, the why is what you carry out of the room. The totals fade. The story stays.

And if that story is written in words you never chose, something subtle has happened: the system has not merely described your thinking. It has revised it.

"Capital preservation" is not a neutral synonym for whatever the tester actually said. It is a financial frame with its own implications, its own posture toward risk, its own quiet suggestion about what kind of person is making this choice. Repeat it back to someone authoritatively enough, and they may begin to remember having meant it.

That is how authorship gets laundered. Not by changing the numbers — by renaming the reasons.

Fluency is not fidelity

This is a particular temptation of language models. They are trained to produce plausible, polished prose. Given a set of priorities, the most natural thing in the world is to summarize them — to find the elegant phrase that sounds like what you meant.

But "sounds like what you meant" is a guess wearing good clothes.

A person who says "I need to stop working weekends" has said something specific. A system that explains their result in terms of "work-life balance optimization" has not honored that specificity. It has translated a lived boundary into a corporate abstraction — and in the translation, the boundary gets softer, more negotiable, more like everyone else's.

The words you use for your own life are not interchangeable with better-sounding words. They carry the exact weight you gave them.

What we changed

The fix is a rule, and it is absolute:

When TheChoiceLessChoice explains a result, it may refer to your options and priorities only by the names you declared. No renaming. No re-labeling. No coined themes.

If a phrase cannot be grounded in your own words, it does not appear.

This makes the explanations less elegant. It also makes them yours.

The written Decision Brief in your Archive was already built this way — deterministic, assembled from your declared names. The spoken and narrative explanation now follows the same discipline, and automated tests verify that a sealed brief contains no invented labels.

The past stays sealed

One more thing, and it matters as much as the fix.

The receipt with the invented wording still exists. It was sealed at its timestamp, and it will remain exactly as it was sealed. We did not edit it, flag it, or quietly improve it.

That is not an oversight. It is the point.

An archive you can retroactively polish is not an archive. It is a public relations department. The whole value of a sealed receipt is that it records what actually happened — including the parts we would now do differently. The moment we start revising the past to match our present standards, every future receipt becomes negotiable.

The fix governs what comes next. The record of what came before stands.

The test you can run anywhere

This principle travels far beyond our system. The next time any tool, advisor, or report explains your reasoning back to you, ask one question:

Are these my words, or better-sounding words?

If the explanation uses language you never used, do not ask whether it is flattering. Ask what it quietly changed.

A reason renamed is a reason revised. And a reason revised by someone else is no longer entirely your reason.

Bring your own words

TheChoiceLessChoice is live.

The Engine is a space to make your thinking precise and move toward a decision. The Council gives you room to sit with something and understand it before deciding. In both rooms, what you declare is what gets evaluated — and what gets explained back to you in the words you actually used.

Bring the decision that's been following you. Start with what matters — in your own vocabulary.

The choice remains yours. So do the words.

— Eric Bonhomme, founder of TheChoiceLessChoice