1. Pedestrian demand in London can be estimated from spatial and built-environment indicators

The core finding is that pedestrian demand in London can be estimated using a model built from spatial configuration and built-environment indicators. The paper identifies a set of variables that best explain observed pedestrian density, then uses those variables to construct a predictive model that performs reliably under randomised testing.

2. Network accessibility shapes movement at the level of the whole system, not just the street

The spatial dimension is central. The research uses space syntax measures to evaluate network accessibility, which means the street network is treated as a relational structure rather than a set of isolated segments. This is important because it allows the model to account for how a street’s position within the wider network influences pedestrian activity. A highly connected street is not just locally busy. It is part of a larger pattern of accessibility.

3. Land-use intensity adds a second, necessary layer

Land-use intensity is the other major component. The paper uses volume-area ratios as a way of measuring built-environment intensity, which provides a more nuanced account of land use than simple zoning labels. This allows the model to capture whether a place is genuinely dense in activities and destinations, rather than merely designated as mixed use in planning terms. That distinction matters because pedestrian movement is influenced by what is actually present on the ground, not by planning category alone.

4. The model performs reliably under testing and extends to future scenarios

The model is tested across Greater London using pedestrian density data. The result is not merely descriptive. The study shows that the model can reliably predict pedestrian demand and can be used to estimate future demand under different built-environment scenarios. This is one of the most useful aspects of the paper because it means that walking can be evaluated in a way that is closer to how motorised transport is often assessed.

Taken together, these findings establish that pedestrian demand is not random but patterned by the built environment, that spatial configuration matters because it conditions how land use performs, and that the resulting model can support planning at multiple scales, from street level to larger administrative areas.

A Note on the Data: 

The paper is methodologically strong because it works with high-resolution geographic data surfaces and uses them to build built-environment variables that can be tested against observed pedestrian density. This means the model is not built from broad assumptions, but from spatially detailed data that reflect the structure of London as an actual city.

The data can be understood in three layers:

1. Spatial configuration data – These are the space syntax measures used to evaluate network accessibility. They describe how the street network is structured and how accessible different places are within it. This is the main structural layer of the model.
2. Land-use intensity data – The paper uses volume-area ratios to measure the intensity of built form and land use. This helps move beyond crude land-use categories and gives a better sense of how active or dense a place is in functional terms.
3. Observed pedestrian density data – These data are used to test the model’s performance across Greater London. They ground the analysis in actual movement and make it possible to compare prediction with observed demand.

For London, the data conversation becomes richer when the paper is read beside other city datasets that capture health, comfort, and sensory experience.

1. Healthy Streets is useful because it frames streets in terms of how they support people rather than vehicles. Transport for London’s Healthy Streets approach includes indicators such as air quality, shade, shelter, safety, and things to see and do, which makes it a valuable complement to movement modelling (Transport for London, Healthy Streets).
2. Noise mapping adds another layer of urban reality. London road traffic noise maps show how acoustic exposure varies across the city and how traffic corridors generate strong gradients of sound pressure (NoiseMap, London Road Traffic Noise Map). That matters because noise shapes comfort, route choice, and the willingness of a street user to linger, even if it does not always appear directly in movement models.
3. Air quality mapping is equally important. The London Air Quality Network provides maps and monitoring for pollution across Greater London, including data on pollutants that affect health and street-level experience (London Air Quality Network, Pollution in my area and London Air Quality Network). A street can be highly integrated and still be unpleasant to walk in if exposure levels are high.
4. Smell mapping introduces a more subjective but equally important sensory dimension. Kate McLean’s sensory mapping work shows how smell can be used to chart urban atmospheres shaped by food, traffic, waste, vegetation, and weather (McLean, K., Sensory Maps). Smell is rarely included in formal transport modelling, but it can strongly affect memory, attraction, and avoidance.

Taken together, these datasets suggest that a street can be structurally strong in movement terms while still performing unevenly in health, comfort, and sensory experience. Spatial configuration explains where movement is likely to happen, but these London datasets help begin to explain how that movement is felt.

Street Scores

Street Scores is a playful extension of the analysis rather than a formal method. I invite any pedestrian or street user to choose the variables they would use to judge a street and then assign it a score based on their own perception. The point is not to standardise judgement or produce a single correct result. Instead, it creates a light, participatory way of thinking about how streets are experienced.

A street user might judge a street on comfort, calmness, safety, beauty, vibrancy, cleanliness, or even a more personal or humorous criterion. One person might care most about noise, another about shade, another about how walkable a place feels, and another about whether it has good people watching. The score becomes a conversation starter rather than a scientific metric. Try it here: Street Scores

The value of the idea is that it opens up an informal comparison between modelled pedestrian demand and lived judgement. A user might score a street for calmness, safety, beauty, shade, social energy, or just general vibe. The result is not meant to be a scientific measure. It is a way of surfacing how differently streets can be experienced.

The streets that the model predicts as strong in movement terms may not be the streets that people enjoy most. The score can then be compared informally with the structurally predicted pedestrian movement derived from space syntax. What may emerge is not a neat alignment, but divergence. Some streets will perform very well in movement terms but feel unpleasant, intense, or stressful. Others may attract fewer pedestrians but be perceived as pleasant, calm, or socially attractive.

This divergence is what I hope makes the idea interesting. It may reveal   that urban space can be successful in one register and unsatisfying in another. Street Scores may not try to resolve that contradiction but  treat the mismatch as the point, because that is where design judgment becomes most revealing.