The Crossing Button That Is Not Connected To Anything

Press the button at a pedestrian crossing and something reassuring happens. A light comes on, and eventually the traffic stops. Whether the press caused the stopping is a separate question.

The signal was already running a fixed cycle

Most urban crossings sit inside a network of signals timed together, so that traffic moves through several junctions in one coordinated wave. The cycle length is set centrally.

A pedestrian phase is built into that cycle at a fixed position. It arrives when it arrives, and an early press cannot move it without breaking the coordination further down the road.

At busy times the crossing therefore runs the same sequence whether anybody touches it or not. The button is wired in and powered up, and the cycle takes no notice.

Off-peak the same button does something

Away from the peak, the corridor is released from coordination and the controller reverts to demand mode, calling phases only when a vehicle detector or a waiting person asks for one.

The button is genuinely live then. The identical piece of hardware alternates between mattering and not mattering according to the time of day, without indicating which mode it is currently in.

This is why people disagree so confidently about whether the buttons work. Both camps are describing something they actually experienced, recorded on different roads at different hours.

The little light is answering another question

The illuminated confirmation is not a statement that the request has been granted. It reports that the press was registered, which is true regardless of what the controller intends to do next.

Engineers keep that feedback because an unresponsive button invites repeated jabbing, and repeated jabbing wears out the switch while doing nothing whatsoever for the mood of the person jabbing.

The result is a display that answers a question nobody asked, while staying completely silent on the one everybody standing at the kerb is actually asking.

Keeping people on the kerb is the real target

A crossing succeeds when people wait for the signal instead of judging a gap themselves. A sense of having acted keeps someone standing still for another twenty seconds.

Designers therefore care less about whether a press shortens the wait than about whether the ritual of pressing holds a pedestrian in place until the phase arrives.

The button is doing a job, and doing it fairly well. It is simply not the job written on the front of it.

The hardware outlives the logic behind it

Signal controllers are replaced on very long cycles, while retiming happens far more often. A crossing installed as demand-actuated can quietly become fixed-time when the corridor around it is coordinated.

Removing the now-decorative button would cost money, generate confusion and produce complaints from people who remember using it. Leaving it in place costs nothing at all.

What survives is a control that once did exactly what it claimed, now does so only sometimes, and offers the user no way of telling the difference.