A road is useful because things can move through it.
A crossing is useful because movement has to stop.
That sounds like a contradiction until you stand at a busy Singapore junction and watch what is really happening.
Cars want the same space.
Buses want the same space.
Cyclists may approach the same space.
Pedestrians want the same space.
A child wants to cross it.
An older person wants to cross it more slowly.
A person with visual impairment may need an audible cue.
The junction cannot grant everybody the same physical space at the same instant.
So it grants them time.
A signalised crossing is a machine for turning conflict in space into sequence in time.
Quick Read
Singapore works partly because many competing movements are not left to negotiate road space from scratch. Signals, phases, push buttons, clearance intervals and accessibility extensions allocate time so incompatible movements do not have to occupy the same conflict point simultaneously.
LTA’s current Green Link Determining system, or GLIDE, coordinates traffic signals across corridors and adjusts green time using real-time demand information from road sensors and pedestrian requests. Pedestrian crossings form part of that network. LTA explains that the green-man period includes a steady phase for starting the crossing and a flashing phase intended to let people who have already started finish crossing.
Green Man+ adds another layer. Elderly pedestrians and persons with disabilities can tap an eligible concession card at equipped crossings to obtain between 3 and 13 additional seconds, depending on the width of the crossing. LTA says more than 1,000 crossings already have Green Man+, with another 1,500 planned by the end of 2027.
The deeper causal chain is:
conflicting journeys → detect demand → allocate phases → give one movement right-of-way → provide clearance time → switch safely → adapt to different users → observe network effects → retune.
This article does not claim that pedestrian crossings explain Singapore’s success by themselves, or that every crossing is perfectly safe, convenient or well timed. It isolates one overlooked mechanism: shared road space becomes more usable when conflict is sequenced instead of improvised.
Wait, What? A Crossing Is Mostly About Time?
Look at the asphalt.
The painted stripes do not physically stop a car.
The green walking figure does not physically carry a pedestrian across.
The signal changes behaviour by assigning a temporary right.
For these seconds, pedestrians may enter.
For those seconds, vehicles may proceed.
Then the right changes hands.
A crossing therefore resembles a tiny constitution.
It defines who may occupy a contested space, under what signal, and for how long.
Conflict Is the Real Starting Point
Two paths can intersect harmlessly if nobody arrives at the same time.
The problem appears when demand overlaps.
A moving vehicle and a walking person can both have reasonable destinations.
Their interests become incompatible at the exact crossing point.
One cannot simply say:
everyone go.
So the junction must solve a scheduling problem.
Road engineering often looks like concrete and electronics from the outside.
Underneath, much of it is conflict scheduling.
The Signal Converts Right-of-Way into a Shared Language
Imagine every junction without agreed signals.
A driver gestures.
A pedestrian guesses.
Another driver thinks the gesture was for them.
A cyclist enters from another direction.
Every encounter becomes a fresh negotiation.
Signals reduce negotiation cost because the red, amber, green and pedestrian indications have standardised meanings.
The system is useful only because road users share the grammar.
A crossing works when thousands of strangers agree to obey the same temporary allocation of space without needing to speak to one another.
The Push Button Is a Request, Not a Remote Control
Press.
Nothing happens immediately.
Press harder.
Still nothing.
Some pedestrians assume the button is broken.
But LTA explains that the push button registers pedestrian demand so the pedestrian phase can be inserted into the signal sequence. The request may not be served instantly because the junction is balancing wider traffic coordination and safety.
The button therefore means:
a pedestrian is waiting; include this demand in the next safe feasible phase.
It does not mean:
stop every vehicle immediately because I pressed now.
Demand-Responsive Does Not Mean Instantaneous
This distinction appears everywhere in public systems.
A hospital responds to demand.
It cannot give every patient immediate access to the same scanner.
A bus network responds to travel demand.
It cannot put a bus under every passenger at the moment the passenger arrives.
A traffic system responds to pedestrian demand.
It still has to complete safe phases and preserve coordination across the network.
Responsiveness means demand changes the system’s decisions.
It does not mean every request becomes the only request.
GLIDE Thinks Beyond One Junction
A junction can optimise itself selfishly.
Give maximum green time to whichever queue happens to be largest locally.
But roads form corridors.
A decision at one junction changes arrivals at the next.
LTA’s GLIDE system coordinates signals across multiple junctions and uses real-time traffic demand to manage broader network flow. In a March 2026 reply, LTA explicitly noted that local junction timing has to be balanced against corridor-wide coordination and that drivers may sometimes see red even when no obvious cross traffic is visible.
Why?
Because the system is not optimising your windscreen.
It is optimising a network containing many windscreens and many pedestrians.
Local Inefficiency Can Protect Network Efficiency
You stop at an apparently empty red light.
It feels stupid.
Sometimes there is a valid reason that is not visible from the driver’s position.
A pedestrian phase may have been activated.
A clearance buffer may be completing.
An adjacent junction may need coordinated timing.
The broader lesson is not “trust every red light blindly.”
It is this:
in a network, what looks locally wasteful can sometimes be globally necessary.
The Green Man Has Two Jobs
LTA distinguishes the steady green man from the flashing green man.
The steady phase lets waiting pedestrians begin crossing.
The flashing phase is primarily clearance time for people already in the crossing to finish.
This means “green” is not one undifferentiated permission.
It has states.
Steady green: begin.
Flashing green: finish if already crossing; do not newly enter.
Red: do not enter.
The phase transition carries meaning.
Clearance Time Is the Space Between One Right and the Next
Suppose the pedestrian phase ends and vehicle green starts at the exact same instant.
A slow walker is still finishing.
A turning vehicle begins.
The system has transferred permission before the previous user cleared the conflict zone.
This is why clearance intervals matter.
LTA’s 2026 explanation of GLIDE specifically notes that clearance buffers before the next phase are integral to junction safety.
The general systems principle is larger than roads:
when responsibility or right-of-way changes hands, the outgoing state needs enough time to become safely complete before the incoming state begins.
Green Man+ Changes the Default Without Rebuilding the Junction
Design one pedestrian phase for an average walking speed.
Now place an eighty-year-old person at the kerb.
The road has not changed.
The user has.
Green Man+ lets eligible elderly pedestrians and persons with disabilities request additional time at equipped crossings by tapping a supported concession card. LTA says the extension ranges from 3 to 13 seconds depending on crossing width.
This is adaptive accessibility.
Instead of assuming every pedestrian moves at the same speed, the crossing can temporarily change its service envelope for a user who needs more time.
Equal Time Is Not Always Equal Access
Give everybody exactly twenty seconds.
Looks equal.
But if one person walks twice as slowly, the same time budget produces a different ability to complete the journey.
Accessibility often requires unequal resource allocation to achieve more equal functional opportunity.
Green Man+ makes that principle visible in seconds.
fairness is not always identical treatment; sometimes fairness is enough time for different bodies to complete the same task safely.
More Than 1,000 Crossings Means Accessibility Becomes Networked
One accessible crossing is helpful.
A network of accessible crossings changes everyday mobility.
LTA currently reports Green Man+ at more than 1,000 pedestrian crossings and plans to expand the feature to another 1,500 by the end of 2027.
The number matters because accessibility has weak value if it appears only occasionally.
A wheelchair-accessible station connected to an inaccessible crossing still has a broken journey.
A slow walker needs enough accessible links across the route, not one heroic node.
Audible Signals Turn Light into Sound
A visual green man assumes the user can see it.
Some signalised crossings therefore provide audible indications to support people with visual impairment. LTA explains that locating tones can help users find the crossing and stronger audio cues can indicate the green phase, with audio volume adapted to ambient noise.
The signal’s meaning has been translated across sensory channels.
Same state.
Different representation.
This is inclusive interface design.
Countdown Timers Expose Remaining Time
A green figure tells you the state.
A countdown tells you the remaining budget.
LTA uses integrated pedestrian countdown timers at busy junctions and schools to show the time left in the pedestrian crossing phase.
This changes decision quality.
“Green” is binary.
“Seven seconds remaining” is more informative.
The user can estimate whether to continue, hurry or wait.
The Crossing Contains a Tiny Model of the Human Body
Signal timing depends partly on assumptions about walking speed.
That means the traffic system contains a model of a pedestrian.
How fast can this person move?
How long is the crossing?
How many pedestrians are expected?
What mix of users is present?
LTA explicitly notes that green-man timing considers crossing length, pedestrian volume and composition.
Engineering has put assumptions about human movement into the clock.
The Model Can Be Wrong for a Real Person
No average walking speed describes everybody.
Injury.
Age.
Wheelchair use.
Children.
Carrying groceries.
Temporary illness.
A system that works for the median user can still fail edge users.
Green Man+, Silver Zones, audible signals, kerbless crossings and other features are ways of correcting the model where ordinary assumptions under-serve important groups.
Two-Stage Crossings Break One Hard Task into Two Easier Ones
Some roads are wide enough that crossing them in one continuous movement is demanding.
Silver Zone designs can include centre refuges or rest points that let pedestrians cross in stages.
This is decomposition.
Instead of one long exposure:
cross half → reach protected refuge → reassess → cross remainder.
The road still carries vehicles.
The pedestrian still reaches the other side.
The difficult task has been partitioned.
Raised Crossings Move Safety from Signalling into Geometry
A sign asks a driver to slow down.
A raised crossing changes the road surface so slowing down becomes physically encouraged.
LTA uses raised zebra crossings in road-safety programmes to increase visibility and reduce vehicle speed.
This is a different control philosophy.
Instruction:
please behave safely.
Geometry:
the environment makes unsafe speed less comfortable and less natural.
Friendly Streets Change the Surrounding Context
A crossing is safer when the road approaching it is already telling drivers:
people matter here.
LTA’s Friendly Streets programme uses features such as traffic calming, kerbless crossings and signal changes intended to improve pedestrian safety and comfort, with plans to bring Friendly Streets to all towns by 2030.
The crossing therefore works best as part of a street environment, not as an isolated patch of paint surrounded by hostile geometry.
Safety and Throughput Are Not Enemies, but They Compete
Give pedestrians very long green phases everywhere.
Pedestrian comfort improves.
Vehicle queues may grow.
Give vehicles nearly all the green time.
Vehicle throughput rises locally.
Walking becomes hostile or unsafe.
The real problem is allocation under competing objectives.
Safety imposes constraints.
Within those constraints, timing can pursue efficiency.
optimisation begins after non-negotiable safety boundaries have been respected, not instead of them.
The Fastest Junction Is Not Necessarily the Best Junction
Measure only vehicle delay and the system may starve pedestrians.
Measure only pedestrian waiting time and the corridor may jam.
Measure only average delay and vulnerable users can disappear inside the average.
A junction is a multi-objective system.
Useful measures include:
- vehicle delay;
- pedestrian wait;
- pedestrian clearance;
- queue length;
- conflict risk;
- accessibility;
- network coordination;
- and reliability under changing demand.
Good operation is not one number.
Turning Vehicles Create a Special Conflict
A pedestrian sees green.
A turning driver may also have a permitted movement but must still yield appropriately to pedestrians where required.
This creates a shared phase rather than fully separated phases.
Shared phases demand attention from humans because signalling alone does not eliminate the conflict.
That is a useful reminder:
systems can reduce conflict without removing human responsibility from every remaining conflict point.
A Crossing Is a Human-Machine Contract
The machine promises:
- recognisable signals;
- defined phases;
- safe sequencing;
- reasonable timing;
- and accessibility support where provided.
The human promises:
- do not enter on red;
- do not begin when the flashing phase indicates insufficient time;
- observe turning traffic;
- do not drive into occupied pedestrian space;
- and do not treat a signal as permission to abandon attention.
If either side fails, safety degrades.
The Crossing and The Queue Solve Different Scarcity Problems
The Queue allocates turn-taking among people seeking a scarce service.
The Crossing allocates turn-taking among incompatible flows seeking the same physical conflict space.
Both use sequence.
But the object of scarcity differs.
Queue:
who gets served next?
Crossing:
which flow may occupy this conflict zone now?
The Crossing and The Alert Solve Different Time Problems
An alert gives advance information so a person can act before risk reaches them.
A crossing allocates a protected or prioritised interval during the risk interaction itself.
Alert:
danger is coming; act.
Crossing:
this is your phase; move while incompatible traffic is controlled.
Primary-School Lens: Why Can’t Everyone Be Green?
Draw a simple four-way junction.
Give every direction green at once.
Ask the child what happens.
They will see conflict immediately.
Then let the child invent phases.
North-south cars.
East-west cars.
Pedestrians.
The child discovers scheduling as a safety tool.
Secondary-School Lens: Build a Signal Plan
Give students vehicle counts and pedestrian counts for four directions.
Ask them to allocate sixty seconds of cycle time.
Then add an elderly pedestrian.
Then add a school dismissal peak.
Then require a clearance buffer.
The “simple traffic light” becomes a multi-constraint optimisation problem.
JC Lens: Network Optimisation, Equity and Externality
At JC level, a crossing becomes a public-policy problem.
Every extra second assigned to one movement has an opportunity cost for another.
But the value of a second is not equal for every user.
An extra five seconds for a fast driver may slightly reduce delay.
An extra five seconds for a frail pedestrian may determine whether the crossing is usable at all.
The policy problem therefore combines:
- network efficiency;
- safety constraints;
- equity;
- accessibility;
- and external costs imposed by delay or risk on others.
Optimisation without value choices is an illusion.
Thought Experiment: Every Junction Negotiated by Eye Contact
Remove every signal.
Keep the roads.
Every pedestrian and driver negotiates priority through eye contact, gesture and courage.
Low-volume streets might cope.
Busy arterial roads become unstable.
The most assertive users gain advantage.
Children, older people and less confident pedestrians may lose access.
The signal’s contribution becomes clearer:
it replaces repeated social bargaining with predictable public rules.
Thought Experiment: Optimise Only Cars
Now keep the signals but tell the algorithm to minimise vehicle delay only.
Pedestrians are an inconvenience variable.
Walking time rises.
People cross illegally because waits feel unreasonable.
Short local trips shift toward cars because walking becomes hostile.
Vehicle demand may eventually rise.
The metric starts damaging the system it was meant to optimise.
Thought Experiment: Give Everyone Unlimited Crossing Time
Now solve the opposite problem.
Every pedestrian phase lasts as long as anybody wants.
No vehicle movement resumes until the crossing has been empty for a long interval.
Pedestrian comfort may improve locally.
Network throughput collapses.
Buses lose reliability.
Emergency travel can be delayed.
The crossing reminds us that public systems often succeed not by maximising one user’s convenience, but by sequencing competing legitimate needs under constraints.
Why Singapore Works Does Not Mean Every Crossing Feels Perfect
Pedestrians may wait too long.
Drivers may see apparently unnecessary reds.
Some users still need more time.
Turning conflicts remain.
Signals can fail.
Human beings can disobey them.
Street geometry can still be intimidating.
The serious claim is narrower:
Singapore uses standardised signals, demand detection, corridor coordination, pedestrian clearance timing and targeted accessibility features to turn recurring road conflict into a managed sequence that can be measured and adjusted.
That is a small but deep piece of urban order.
The Fifteen-Question Crossing Test
- Conflict: Which movements physically intersect?
- Demand: How does the junction know who is waiting?
- Phase: Which movements can safely occur together?
- Sequence: In what order are incompatible movements served?
- Start: How much time lets waiting pedestrians begin?
- Clearance: How much time lets people already crossing finish?
- Accessibility: Who needs more time than the default?
- Representation: Can visual, audible and countdown signals communicate the same state to different users?
- Network: How does this junction affect the next one?
- Priority: Which flows receive more time during peaks, and why?
- Geometry: Can raised crossings, refuges or kerbless design reduce risk physically?
- Human behaviour: Which conflicts still require judgment?
- Failure: What happens if the signals stop working?
- Measurement: Are we measuring safety and accessibility as well as throughput?
- Review: How will timing change when travel patterns change?
Frequently Asked Questions
What is GLIDE?
GLIDE is LTA’s Green Link Determining traffic-signal system. It uses traffic-demand information, including data from road sensors and pedestrian requests, to adjust signal timing and coordinate adjacent junctions across the network.
Why does the green man not appear immediately after I press the button?
The button registers pedestrian demand. The system still has to serve that demand at a safe point in the signal sequence while balancing traffic flow and wider junction coordination.
What does the flashing green man mean?
It is primarily clearance time for pedestrians who have already started crossing. LTA advises pedestrians not to begin crossing once the green man is flashing because there may not be enough time to complete the crossing safely.
What is Green Man+?
Green Man+ lets eligible elderly pedestrians and persons with disabilities request additional crossing time at equipped signalised crossings by tapping a supported concession card. The extension is between 3 and 13 seconds depending on crossing width.
How widespread is Green Man+?
LTA currently states that more than 1,000 pedestrian crossings have Green Man+, with another 1,500 planned by the end of 2027.
Why do some pedestrian signals make sounds?
Audible signals support people with visual impairment by helping them identify the crossing and recognise when the pedestrian phase is active.
What is the main student lesson?
When two legitimate activities cannot occupy the same place at once, one solution is to move the conflict into time: define states, sequence turns, protect the transition and adapt the default for users who need more time.
Sources and Further Reading
- Land Transport Authority — Intelligent Transport Systems.
- LTA — Traffic Data Used to Optimise Network Flow While Prioritising Safety, 12 March 2026.
- LTA — Walking and Road Safety Infrastructure.
- LTA — Friendly Streets.
Final Thought: The Road Is Shared Because Time Is Shared
A pedestrian crossing looks ordinary because its logic has become ordinary.
Wait.
Green.
Walk.
Flash.
Finish.
Red.
Cars move.
The simplicity is manufactured.
Under it sit sensors, phases, timings, clearance rules, accessibility assumptions, corridor coordination and human conventions.
That is why Singapore works, in another quiet way:
when too many journeys want the same piece of road, the city does not ask them to fight for space; it gives them different moments in which the space becomes theirs.