Explanation
Constraint Breeds Clarity
Why forbidding words makes prose sharper
Constraint is a paradoxical engine of clarity. Forbid a single verb, fix an exact grade level, or restrict the dictionary to a thousand approved words, and the writer must find fresh, specific, verifiable language. The freedom to write anything is often the freedom to write nothing in particular.
E-Prime bans every form of to be, which quietly outlaws the passive-voice and the vague assertion of identity; the writer must name an agent and choose a real verb. Simplified Technical English and other controlled natural languages fix the vocabulary so a machine—or a reviewer—can check correctness. Readability metrics turn “make it simpler” into a measurable target.
The shared principle: a checkable constraint is one a language model follows reliably and a human can audit. Vague instructions produce vague compliance. Hard rules produce measurable behavior.
Example
-
Before
The results were surprising, and it was clear the model was better than the baseline.
-
After (E-Prime)
The results surprised us; the model clearly outperformed the baseline.