D4.08.2Redundancy in low-load settingsdesignresearch

In low-load settings, redundancy has limited marginal value and may overload

Aliases: marginal gain · overload · unnecessary redundancy

What it is

In low-load settings, redundancy adds little and can overload. When one channel already delivers reliably, a second channel raises delivery no further while consuming extra attention and possibly causing interference. Redundancy's costs become more visible precisely when conditions are good.

Why it happens

Diminishing returns come from a delivery ceiling: when a single channel is near 100%, redundancy can only act on the small remainder. The cost persists, because every extra channel must be processed—at minimum judged as ignorable—and competes for attention with other information. When many events are all redundant, the costs stack, multiplying what must be handled in a short window and lowering overall performance. This is why making every notification audible, visible, and vibrating in a quiet bright office yields irritability rather than reliability.

Studying it

Compare single- and multichannel schemes under low load: measure delivery, response time, perceived workload, and disturbance ratings. Variables include redundancy level, event frequency, and task load. Include workload and disturbance, since when gain is zero the cost is the whole result. Include a high-load condition to show how the gain changes with circumstances.

Where it stops holding

For critical information, redundancy may be insurance against conditions worsening even when the current setting is benign, and judgment should then rest on the worst case rather than the current one. If the redundant channel carries a very small amount of information—a brief tone—its marginal cost is low and redundancy remains acceptable. Where users explicitly prefer a single channel, forced redundancy degrades experience.

Applying it

  • In benign settings, reserve redundancy for critical information and keep ordinary cues single-channel.
  • Evaluate redundancy on gain and cost together rather than delivery rate alone.
  • Size redundancy for critical events by worst-case rather than current conditions.
  • Verification: under low load, measure delivery and perceived workload for single- and multichannel schemes; no gain with higher workload means redundancy is unnecessary.

Related

  • Within the group: D4.08.1 Redundancy improves delivery in high noise or high load · D4.08.4 Blindly stacking channels raises design and maintenance cost with little return
  • Adjacent: D3.05.3 Haptics are poor at conveying quantity and location · D2.11.4 Once sound is off, other channels must carry what was muted
  • Search terms: marginal gain · overload · redundancy cost

Cards in the same group

Quick Actions

Share

Share this page

ios_share

https://hci.top/en/handbook/D4.08.2