humancentric

The worn track through the grass is not vandalism.

It is the shortest path asserting itself. It is also the only free, continuous, unfakeable record of how people actually move, and we mostly respond to it by re-turfing.

Move either end and watch the track appear.

The grey line is what was built. The orange one is what people walk. The gap between them is the whole subject.

Evidence that arrives without being asked for.

Every other form of pedestrian data costs money and involves asking people questions they answer badly. A desire line costs nothing, records every user, runs continuously, and cannot be gamed. It is the strongest available signal that a route was designed around something other than where people are going.

Singapore used to lean into this. The habit of paving over worn grass routes, rather than fencing them, shows up repeatedly in accounts of older estates. Newer estates appear less inclined to. Meanwhile there are documented cases of the opposite response: a long-standing desire path near Newton MRT was reportedly closed off by cordoning the entire field rather than paving the line people had already chosen.

The framing has started to circulate locally. Serene Koh's essay Desire lines in Jom opens on exactly this idea and applies it to jaywalking, ending on the right question: not whether people are behaving badly, but "whether we've been designing for them."

It also works. One resident used OneService to suggest paving a route across a field so wheelchair users did not have to detour. MSO acknowledged it the same day and the path was paved about four months later. That is a documented, reproducible route from a citizen report to a built change, and it costs a photograph and five minutes.

Last year this stopped being about convenience.

149road deaths in 2025, a ten-year highSPF
27elderly pedestrians killed, up from 11 in 2024SPF
83.3%of fatalities involving pedestrians who crossed away from a crossing were elderlySPF
84–93%of over-65s walk slower than the speed crossings are timed forAsher et al.

The phrase is the Police's own, not ours. The Annual Road Traffic Situation 2025 describes "a persistent upward trend over the last few years" and calls it worrying.

An elderly person crossing where there is no crossing is a desire line with a body attached. The people least able to walk two hundred metres to a signalised crossing and back are dying at the highest rate doing exactly that. Singapore already concedes that older pedestrians need design accommodation — that is what Green Man Plus is for. The accommodation exists for crossing time, and only for those who register and remember to tap. It does not exist for crossing placement at all.

And age is not a modifier here, it is structural. US crash data shows a 70-year-old struck at 40 km/h faces the same death risk as a 30-year-old struck at 56 km/h. Sixteen kilometres an hour of survivable impact, lost to age alone.

What this does not show. The data establishes that elderly people are dying while crossing away from crossings. It does not establish that better crossing provision would have prevented those specific deaths. Older pedestrians also walk slower, see and hear less well, and survive impacts less often. The defensible claim is narrower: the official response to a doubling of elderly pedestrian deaths is enforcement and education aimed at pedestrian behaviour, and the design question is not being asked with the same energy. A systematic review screening 69,000 records found standalone pedestrian education had no positive effect on injuries, while road-environment change was favourable in 19 of 27 observations.
Singapore already proved its own case. Across 15 Silver Zones — S-curve realignment, narrowed carriageways, extended green time, refuge islands — annual accidents involving elderly pedestrians fell from 14 to 4. That is LTA's programme and LTA's number, not a foreign import. It is currently sited reactively, after the crashes. Proposal 02 is that it should be sited before them.

Crossings interrupt the pedestrian, not the driver.

The same logic that produces detours across grass produces them at junctions. At most Singapore side roads the footpath stops at a kerb, drops to carriageway level, and resumes on the other side. The pedestrian gives way, changes level twice, and is exposed to turning traffic. The alternative, standard in the Netherlands and increasingly in the UK, is the continuous footpath: the pavement runs straight across the side road at its own level, and the vehicle ramps up onto it. Priority is expressed by geometry rather than by a sign.

It costs roughly what a normal junction costs. It just allocates the interruption to the party that is sitting down.

Where Friendly Streets sits. This is the one area where LTA's programme is genuinely delivering: barrier-free and kerbless crossings, raised signalised crossings, wider medians, longer green-man times, lower speed limits. Nine neighbourhoods completed, all 24 towns targeted by 2030, a billion dollars committed. Credit where it is due.

The gap is that the 2023 announcement listed "widened footpaths" among the features, and the completion reports describe crossings, humps and green paint. Which leads to the single sharpest question anyone could put to an MP: how many metres of footpath has Friendly Streets actually widened, and where?

Read the ground before re-turfing it.

Before a worn track is repaired, it should be recorded and assessed against the paved route it bypasses. Where the detour ratio exceeds a stated threshold, the default response should be to pave the line rather than restore the grass. This costs nothing to adopt, because the maintenance visit is already happening.

And where a footpath meets a minor side road, the continuous footpath should be the default treatment rather than the exception.

Photograph one properly

Sources

Status of this page. The desire-line and continuous-footpath material here is less thoroughly sourced than the footpath width page, and the OneService and Newton MRT accounts come from public forum posts rather than official records. Treat them as leads worth confirming rather than as established facts, and if you can improve either, please do.