DESIGN INTELLIGENCEJuly 23, 2026·Mary · DEPIX Design Intelligence

The Machine That Fails Into Stairs

Stand at the foot of a broken escalator and something strange happens: nothing. You walk up it. The quietly radical thing about the escalator is not that it moves — it is what it becomes when it stops. Almost everything else we build fails into uselessness: an elevator that loses power is a sealed box, a car that dies is a two-tonne paperweight. An escalator that fails becomes a staircase. That is not luck. It is a decision, made at the very beginning of the design, that most products never make.

The machine arrived as an argument about exactly this. In 1892 Jesse Reno patented an "inclined elevator" — a moving ramp with cleats. Around the same time Charles Seeberger, working with Otis, developed the flat-step version and coined the word escalator from the Latin scala, "steps." Two philosophies fought — ramp versus steps — and steps won. Partly for comfort, but partly because steps are precisely what you fall back to when the power dies. The winning design already contained its own failure mode.

Engineers have a name for this: graceful degradation. A well-made system does not collapse when a component breaks; it drops to a lower but still-useful state. The escalator is the purest everyday example — its degraded mode is a perfectly good staircase, needing no power, no software and no rescue. Compare that with the systems around it. A powered door with no manual fallback traps you. A single point of failure in a lift means a stuck cabin. The escalator refuses to have one, because its skeleton is a stair and the motor is only an enhancement laid on top.

You can read the same failure-first thinking in every safety detail. The comb plate at each end — those interlocking teeth — exists to catch and shear anything before it is dragged into the mechanism. The bristled skirt along the step edge, the emergency-stop button at top and bottom, the handrail synchronised to exactly the step speed so hand and feet never disagree: each is a decision about what happens when things go wrong, not when they go right. This is fail-safe design and redundancy made physical, in an object so ordinary nobody looks at it.

None of this is visible in the brochure. The render of a gleaming new London Underground atrium shows the escalator gliding, full of light, working perfectly. It never shows the state the design actually optimises for: powered down, packed with a rush-hour crowd walking up it through a fault, or a child's shoelace at the comb plate. The moving escalator is the demo. The stair is the product.

That inversion is the whole lesson, and it is where most design goes wrong. Teams polish the happy path — the launch keynote, the hero shot, the frictionless first run — and treat failure as an edge case to be patched later, after the shape is frozen. The escalator does the opposite. Its designers decided the failure state first — it must be a usable stair — and then added the machine that improves on it. Resilience was not bolted on; it was the founding constraint.

It also quietly did something the lift could not: move a continuous crowd. An elevator is a batch process — fill, rise, empty, return. An escalator is a conveyor, absorbing a stadium or a station surge without anyone waiting for a car. That is why the flat moving walkway followed it into airports, and why dense vertical cities became thinkable at all. But the escalator's relationship to accessibility is double-edged: it eased the climb for millions while never replacing the ramp and lift a wheelchair needs — a reminder that graceful degradation for one body can be a dead end for another, and that universal design has to be decided at the same early stage, or not at all.

Which returns to the concept phase — the moment, before anything is engineered, when a team chooses what the object must still do on its worst day. Most of what we build has no honest answer, because the question was never asked; failure was left to accident, found in the field, apologised for in a patch note. The escalator asked it first, and its answer is so quietly complete that a broken one barely reads as broken. Defensive design is not a feature you add at the end. It is a decision about what survives when the power goes — and the best time to make it is before the first step is drawn.

Sources: