Dead Cells
Keep the task stable and make the input method explicit. If Hem’s Inspect is understood but difficult to trigger, test the operation before changing its wording.
Interview first, watch an uncoached task, or observe and then interview?
Someone says the menu felt slow. Was the delay finding the destination, operating the control or waiting after a press? Compare real screens, then choose what you need to see or ask before changing yours.
A small diamond sits beside Campaign. Battle and the other destinations sit in a quiet column beside the lit serpent. The frame records the marker’s position; its input cause is unverified.
Keep the task stable and make the input method explicit. If Hem’s Inspect is understood but difficult to trigger, test the operation before changing its wording.
For a failed Back or Inspect, retain a minimal local trace of semantic input, focus owner and resulting screen. Reconstruct intent from the person’s words.
Keep the naming, reading and operation tasks separate when someone describes the estate as slow.
Compare what the player can infer. The scale labels describe the production context of each example.
These editorial lists rank how well a particular shipped interface supports the stated research question. We inspected publisher and developer imagery across 22 candidate games, including examples from the 1990 Loom lineage through 2023 releases. We compared the specificity of the task, the strength of the primary source, whether an observable action can distinguish plausible explanations, and transfer across genres. The four lists answer different questions. The three Snake Eyes picks in each list are separate editorial recommendations. In particular, the Dead Cells team reports a player panel for its revision, and Turn 10 describes direct consultation on Blind Driving Assists. Other proposed prompts and observations are our exercises. No reference screenshot records a Snake Eyes participant, and no sample here represents all game players. Award labels describe their organizer’s category rather than endorsing our rank or exercise.
Each entry links to the image and the feature evidence. Award badges name the recognized category. The saved direction records the director’s own choices.
Player-panel feedback plus a task
The developers describe working with an AbleGamers panel and the Steam community. This update screenshot shows a shield option that changes holding into a toggle. A person can understand the purpose of a shield and still find the original gesture costly.
Input trace and repeatable replay
The training HUD shows the received directions and buttons, the attack data and the frame display. The recording menu separately allows a fixed dummy action to be replayed. An observer can compare intent, input delivery and game response.
Interview and observe in the person’s setup
A cockpit frame provides road, traffic, speed and gear visually. Turn 10 separately documents co-development of adjustable audio cues with blind consultant Brandon Cole and other players. A sighted observer studying the screenshot has only part of the interface.
Compare a visual and an operational intervention
An authored choice screen explains that the background matters and requires a hold to confirm. Separate options can adjust font presentation, text size or dialogue speed. A long pause could occur at several distinct steps.
Configuration and equivalent tasks
The official guide illustrates presets, UI text size and captions separately. A screenshot of the preset menu plus the actual settings in use gives a more defensible starting description than calling someone a “beginner” or “advanced player.”
First visit versus resumption
The thermal view supplies a legend and describes the selected sawmill as freezing. Another capture shows a building inspection panel in a normal city view. The two screens support different starting conditions for the same diagnostic task.
Shared workspace versus separate participants
One visible kitchen holds multiple chefs, orders, counters and a timer. It gives the observer a vocabulary for recording a handoff and its outcome, apart from the team’s intentions or conversation.
Placement work and personal judgment
A kitchen with an open drawer and ordinary possessions provides several plausible destinations for a held item. The person may be pursuing a domestic arrangement the observer did not anticipate.
Evidence search versus data entry
A ledger can display an unknown person, an apparent cause and an unknown attacker at the same time. There are two valid reasons to leave a field incomplete: uncertainty about the evidence and uncertainty about operating the record.
Historical input and sensory context
The distaff, written notes and highlighted target make the puzzle accessible to a researcher as a set of separate questions. Can the person identify the target, recognize the phrase and play it with their chosen device?
Loading saved references…