Diegetic horror UI should unsettle the player, not hide the interaction

therapy.exe’s icon and shadow work shows how a world-consistent interface can fail at actual display size—and why a flattering visual metric is not enough.

Winged Sloth Games Editorial · Published 2026-09-30 · Updated 2026-09-30

Start with the actual game promise

therapy.exe is a turn-based roguelite deckbuilder presented through a clinical, bureaucratic interface. That presentation can carry the fiction: forms, labels, constrained choices, and the feeling of being evaluated. It does not require essential controls to become unreadable.

Our analysis distinguishes two kinds of uncertainty. Narrative uncertainty concerns consequences, motives, or meaning. Interaction uncertainty concerns which action is available, what changed, or how to continue. A designer may want the first. Accidentally introducing the second is not automatically successful horror.

This is a design framework, not a claim that WSG measured player fear or conversion. The evidence below concerns implementation and visual inspection.

First-party evidence: an icon at its real size

The September 14 map-icon entry records silhouettes that depended on thin internal detail. At small rendered sizes, that detail became smudges. Unreachable nodes dimmed to flat gray, making internal detail an especially weak place to put an icon’s identity. [1]

The redesign used thicker structures and more distinctive outlines. The entry reports lower half-alpha coverage after shrinking, but its separate distinctness score did not improve in the expected direction. Pixel correlation treated bold filled silhouettes as similar, even while the author’s map inspection found them more legible. [1]

That disagreement is the important result. It does not mean measurement is useless. It means the metric was answering a narrower question than the design needed. WSG kept that score as a readout rather than a release gate.

First-party evidence: contrast after compositing

A separate shadow pass found a different gap between a tidy rule and the rendered screen. The palette contrast table did not include a translucent shadow darkening the paper behind text. A screenshot comparison flagged the hint below a focused card, and further inspection identified a footer with too little breathing room. [2]

The response was not simply to remove depth from the interface. The record describes tightening the shadow and adding footer space. It also documents false positives in the diagnostic, which initially confused blue outlines with blue text. [2]

In other words, both the visual treatment and the tool needed examination. A large number of warnings did not establish a large number of real defects.

WSG’s proposed uncertainty audit

For each screen, ask three separate questions:

• What does the fiction intentionally withhold or make uncomfortable?
• What must remain unambiguous for the player to make a meaningful choice?
• What changes when the interface is scaled, dimmed, composited, or viewed in an alternate state?

Then inspect the most fragile examples, not just the pristine mockup. A disabled map node, a focused card over a hint, or a crowded hand may expose a failure that a normal state hides.

A useful interaction contract might be: “Each node category remains identifiable when unreachable.” A useful narrative intention might be: “The institution’s language feels impersonal.” Those statements can coexist. Neither needs to be substituted for the other.

What we would test next

Rendered inspection is not the same as observing players. A subsequent playtest could ask whether people correctly identify node categories, distinguish available actions, and explain the result of a card play. We should record actual responses before claiming the redesign reduced confusion.

The same caution applies to accessibility. These records describe particular visual checks; they do not certify the whole game or prove that every player can use the interface.

Outside reporting and limits

No outside horror-market reporting is used in this piece. WSG is offering an implementation-based design framework, not a general law about what makes a horror game frightening.

Our working rule is deliberately practical: preserve ambiguity where it serves the fiction, and demand evidence where it obstructs the interaction.

AI-assisted studio editorial assembled from dated WSG development records; no new playtest or independent verification is implied.

Update note: Initial publication from dated development evidence.