Templates constrain the range of what users will express

What it is

A “competitor comparison” template ships three columns: features, price, reputation. Someone who also wanted lead time and after-sales response sees those three columns and mostly fills those three. Template expression constraint is the scaffold’s mirror: while structure lowers start cost, intents that were never built as slots fall out of view. People are not forbidden to say those things. They stop thinking to say them.

This is not the switch of interaction grammar in which templates turn open input back into a form — the grammar can still allow a free extra sentence, and the slot roster still quietly cuts the range. It is also not the active mislead of a designer’s task assumption missing real work. Even when the task type is right, dimensions still get shaved.

Why it happens

Recognition beats generation. A slot roster is a set of lit categories; unlit categories get less activation in working memory. A classic effect of classifying is that once an object goes in a bin, edge features drop. Column names are bins.

There is also politeness and compliance: official structure is read as recommended practice, even as the only correct one. Editing a template feels like editing an exam paper, socially costlier than adding a sentence to an empty box. The squeeze is therefore cognitive and normative. The model then sees only filled slots; unspoken dimensions are not filled back in. The squeeze is written into the artifact.

Studying it

First, by interview or cards, collect the dimensions people spontaneously mention for the task — a full set uncontaminated by a template. A second group completes the same task with the template. Compare how much of the full set the artifacts cover, and which dimensions appear in think-aloud and then get dropped. Independent variables: coverage of the slot set relative to the full set, whether a free “additional notes” sentence is allowed, whether copy says “you may add columns.” Dependent variables: expression coverage, rate of add-column behavior, items that appear only when you later ask “what else did you want to compare.”

The control matters: empty-box artifacts are often messier and dimensionally wider. Do not read the squeeze as a quality gain because the grid looks tidier.

Where it stops holding

In compliance, finance, and medical records, the squeeze is the point: only allowed fields may be filled. In teaching, a template as a zone-of-proximal-development scaffold is meant to be temporary. If people can add a column in one glance and the addition is actually used, the squeeze becomes optional. Highly practiced people treat the template as a draft and tear it down; the limit mainly hits the moderately skilled who will not edit official structure.

Applying it

• Beside the slot roster, put an “add one” that actually works: the new item enters the same constraint set, rather than falling into a remark the model will ignore.
• Say in a short line that the template is a start, not a ceiling. Avoid exam-paper voice (“please follow the format below”).
• On a schedule, mine empty-box artifacts against the template: high-frequency dimensions the template lacks should become optional slots, not remain a tail.
• Check: dimension lists from template artifacts versus empty-box artifacts on the same task. If dimensions stable in the empty-box group nearly vanish in the template group, and users can name those dimensions afterward, the squeeze has happened. Then look at whether anyone used “add one” — unused equals absent.

Related

• Same group: L2.03.1 Templates lower the cost of getting started · L2.03.3 Templates must be editable, not only fillable · L2.03.4 Templates turn open input back into structured blanks, at the cost of openness · L2.03.5 Placeholder content in templates is kept verbatim; empty slots must be unskippable · L2.03.6 Templates encode the designer’s task assumptions and actively mislead when those do not match real work · L2.03.7 A template that can only be applied whole, with its structure invisible, teaches nothing
• Nearby: L2.01 Openness of Natural-Language Commands and Its Cost · L2.04 Complementarity of Parameter Controls and Natural Language · L2.06 Granularity of Control
• Search terms: template expression constraint · category effect · scaffold as ceiling