Feedback delay disrupts perceived causality
Aliases: input latency · causality · responsiveness · temporal binding
What it is
Feedback latency and causal attribution describes how a long or unstable wait between an operation and its perceptible consequence makes players unable to attribute that consequence to their input. The problem is not only high average latency: the same operation sometimes being fast and sometimes slow, or feedback arriving after other events, also weakens certainty that “I did this, therefore it happened.”
Why it happens
People use temporal proximity to infer causality. A semantically congruent change that follows input quickly is readily bound into one event. With longer delay, the player has issued new commands and the environment may have changed, creating competing causes. Jitter is especially harmful because no stable predictive model can form, encouraging repeated input merely to probe whether the system heard.
Where it stops holding
Some outcomes legitimately take time: projectile travel, turn resolution, network confirmation, and complex animation. They can delay the final consequence without delaying every acknowledgement. Nor is it enough to shorten client effects while ignoring adjudication mismatch; when local prediction and authority often conflict, premature feedback becomes a false promise.
Applying it
- Measure input sampling, simulation, rendering, audio, network round-trip, and display delay separately. Prioritize unstable queuing and blocking rather than reporting only average frame rate.
- When outcome must wait, immediately acknowledge accepted or predicted input, then correct disagreement with authority clearly and with low penalty.
- Verification: use high-frame-rate capture with input markers to measure end-to-end timing distributions. Combine this with repeat-input rate, errors, and player judgments of whether an action took effect.