Gestures with low social acceptability will be actively avoided even if they are technically reliable

What it is

High recognition, low latency, learnable—and still nobody uses it in real life. Users actively avoid motions that make them look strange to bystanders. That is an adoption failure, not a usability failure and not a recognition failure. Everyone can finish in the lab; everyone reaches for the phone on the metro. Technical reliability proves the channel can work; avoidance proves the channel’s social price exceeds the gain of that action.

Why it happens

Choosing an input weighs task gain against several costs. Social cost in public can outweigh the two seconds saved by not taking out a phone. Avoidance looks like: switching to a non-gesture path, shrinking the motion until the recognizer cannot see it, waiting until nobody is around, or uninstalling. In logs those look like “user used the button,” “recognition dropped,” “use clustered at night,” “poor retention,” and get misdiagnosed as something else. Avoidance is contagious: seeing others not do it, one does it less, and field adoption locks low. Fatigue makes people do less because doing hurts; avoidance is deciding not to do it before starting. Interviews can split the two.

Studying it

After measuring recognition and completion in the lab, take the same people to a place with bystanders and see whether they switch to the spare path or refuse the task. On multi-week retention logs, split out “use rate with bystanders present.” Code open reasons for “why didn’t you gesture”: awkward, don’t know how, inaccurate, tired—four classes, kept apart. Reporting only “completable” from the lab will miss avoidance. If almost nobody gestures in the field, do not retrain the model first.

Where it stops holding

In private use with no bystanders, avoidance mostly comes from other costs; this claim does not explain it. A job that requires doing it in public (warehouse scanning, exhibit demo) suppresses avoidance, which does not mean people will still do it after hours. One-shot novelty can outbid awkwardness; long-term avoidance shows up after novelty fades, so a short test underestimates. If the spare path is extremely hard, avoidance looks like “gritting through it” rather than switching, which mimics high acceptance.

Applying it

• Among adoption metrics, split out “still gestures when bystanders are present.” If that line drops, check social price before accuracy.
• Do not put on the public primary path a motion that is most recognizable in the lab and that nobody will perform on site.
• After launch, if button use dwarfs gesture, code reasons and then decide whether to change the vocabulary, change setting strategy, or admit gesture only fits private scenes—do not treat it as a model incident first.

Related

• Same group: C4.32.1 Performing mid-air gestures in public draws bystander attention and creates social pressure · C4.32.2 Large-amplitude, long-duration actions have lower social acceptability than subtle ones · C4.32.3 Social acceptability varies by culture and setting
• Adjacent: C4.28 Necessity of non-gesture paths · C4.05 Three dimensions of gesture evaluation
• Search: avoidance · adoption · social cost