Safety

Definition

Safety is the condition of being protected from harm, and the human need for that condition — one of the small set of requirements that appear in every documented human society and in every serious account of human motivation.

As a Human Universal

Safety appears across the various lists of human universals in several forms: shelter, in-group protection, conflict mediation, sanctions against violence within the group, food storage, care of infants, and the marking of dangerous places and substances. No society has been recorded that does not organise itself substantially around keeping its members alive.

Its most familiar formulation is Abraham Maslow's, where safety needs — physical security, health, property, employment, predictability — sit immediately above physiological needs and below belonging. Maslow's strict hierarchy has not held up well empirically; the categories have, and the observation that unmet safety needs dominate attention and crowd out everything else is well supported.

Its Two Halves

The concept has an objective and a subjective side, and the gap between them is where most of the interesting behaviour lives.

Actual safety is a matter of what can harm a person and how likely it is. Felt safety is a psychological state, and it responds to different inputs: predictability, legibility, control, familiarity, the presence of trusted others, and the ability to see what is happening. The two frequently diverge — people accept large risks that feel controllable and refuse small ones that do not, a pattern documented at length in the risk perception literature.

For design this matters because felt safety is designable and actual safety is not always visible. A system can be secure and feel alarming; it can be dangerous and feel reassuring. The second is worse.

Safety in Technology

Several distinct meanings run together under the word and are worth separating:

• Physical safety. Devices that can burn, crush, electrocute or fail dangerously. When computation is embedded in doors, vehicles and floors, a software fault becomes a physical one — see Everyware.
• Data security. Protection against unauthorised access. Frequently traded against convenience, and usually silently.
• Psychological safety online. Freedom from harassment, exposure and unwanted contact. This is where the asymmetry of risk becomes most visible: a default that is harmless for one user is dangerous for another, and defaults are set by people in the first category.
• Predictability as safety. A system whose behaviour can be anticipated is safe to explore. One that acts on inference, changes without notice, or cannot be inspected is not — regardless of whether it causes harm.
• Recoverability. Norman's point: safety is largely a matter of whether a mistake can be undone. Irreversible actions taken by tired people are the ordinary shape of catastrophe.

Relevance to Cyborg Anthropology

Safety is the need most often invoked to justify surveillance, and most often eroded by it. A system that watches in the name of protection reduces felt safety for anyone who has reason to be watched — see panopticon and geocloaking — which means "safety" in technology arguments should always be answered with the question whose.

It is also the calm-technology criterion that reads oddly until one thinks about it. "Still work when it fails" is a safety principle: a device that becomes inert when a server goes down has made its user's ordinary life dependent on someone else's uptime. See Calm Tech Institute.

Related Reading

• Human Universals
• Variety
• Belonging
• Calm Tech Institute
• Geocloaking
• Locative Feminism