Who Decides Where We Go? Spatial Algorithms and the Indian City to 2046

A growing share of urban trips now begins with a question typed into a phone. Navigation apps choose routes, and recommendation engines decide which cafés, temples, markets and viewpoints appear first. This article examines how far those systems may be shaping movement in cities, a question long treated as a matter of street geometry and human purpose. It draws on space syntax, the new science of cities, critical work on data-driven urbanism and studies of routing and recommendation. It argues that algorithms should inform spatial decisions without determining human choices, and it sets out five dimensions that Indian planning could use when thinking about movement to 2046.

Introduction: a new author of movement

Urban design has traditionally asked how the form of streets and spaces shapes the way people move. A newer question sits beside it: as digital platforms such as Google Maps, Apple Maps and Waze, together with the recommendation engines built into travel and food apps, become routine, will cities continue to design movement, or will algorithms increasingly design it for them?

The question matters because movement in cities is rarely only a matter of shortest paths. People walk to a shrine because of faith, return to a market out of habit, travel for livelihood, and seek out a lakefront or a mall for leisure or aspiration. An algorithm that optimises for time or for ratings sees only part of that picture. This article first summarises what research says about spatial configuration and movement, then looks at the evidence on routing apps and on algorithmic visibility, and finally proposes a five-part frame for planning. The frame is the author's proposition, not an established finding, and the article states its limits before concluding.

What the research says: configuration, data and the city as a system

The best-established body of work linking urban form to movement is space syntax. Hillier and Hanson (1984) argued that the configuration of a street network, how each space is connected to all the others, structures patterns of co-presence and encounter. Later space syntax research built on this to relate integration in the network to pedestrian and vehicular flows. The important point for this article is that, in this tradition, the street network itself is a powerful, if not sole, generator of movement.

Batty (2013) extended the view of cities as systems of flows and networks, proposing a "new science of cities" in which interactions, not only locations, are the core object of study. This framing makes digital platforms relevant: if movement is a set of flows, any system that systematically redirects flows becomes part of the city's structure.

Critical scholars have been more cautious about what data-driven systems can know. Kitchin (2014) examined the rise of real-time, big-data urbanism and raised concerns about technocratic governance, corporatisation of urban data and the narrowing of urban problems to what can be measured. Kwan (2016) went further, arguing that algorithms used to produce geographic knowledge introduce their own uncertainties, so that "algorithmic geographies" are not neutral reflections of the city but partly products of the methods used to compute them. Read together, these sources suggest that a city increasingly seen through algorithms may be a city partly reshaped by them.

Figure 1. Spatial algorithms of the city: the human layer, the algorithmic layer and five dimensions for planning to 2046. Source: author's own diagram.

Figure 1 sets out the argument of this article in one view. The reader should notice the two layers: on one side, the human reasons for moving through a city; on the other, the digital systems that increasingly mediate those reasons. The five dimensions at the base are the frame developed in the later sections.

Routing apps: when optimisation meets the street

The clearest evidence of algorithms reshaping movement comes from routing. Cabannes et al. (2019) noted that the impact of routing apps on road traffic was still uncertain, and proposed a measure, the average marginal regret, to estimate how far observed traffic departs from the equilibrium that would arise if every driver chose the best route. Their work treats app-guided drivers as a population whose collective choices can be modelled, and it points to a shift: route choice is no longer a dispersed set of individual judgements, but is partly coordinated by a small number of platforms.

Media and community reports in several countries have described routing apps sending through-traffic into residential streets to avoid congestion on arterials. This article has not verified a systematic study of that effect for Indian cities, so it should be treated as a plausible concern rather than an established pattern here. In design terms, however, the concern is clear. A route optimised for travel time does not register that a lane is a children's play space in the evening, that a street is a procession route during a festival, or that a market street depends on slow, browsing movement. Optimisation can be efficient for the individual trip while overlooking what a street means to those who live on it.

Algorithmic visibility: which places get found

The second channel is visibility. Recommender systems research has documented "popularity bias", the tendency of recommendation algorithms to over-recommend items that are already popular and under-represent the long tail. Abdollahpouri et al. (2017) studied this in learning-to-rank recommendation and proposed methods to control it. Their work concerns recommender systems in general rather than places, so its transfer to urban destinations is an inference, but the mechanism is easy to imagine: a well-reviewed café rises in rankings, attracts more visitors and reviews, and rises further, while an older neighbourhood eatery with few online reviews recedes.

Recent urban research begins to test this directly. In a preprint analysing more than 106,000 places in Amsterdam, Shaker and Uitermark (2026) report that Google Maps produces uneven platform geographies, in which consumption-oriented places and neighbourhoods stand out while others are pushed to the background. As a preprint, its findings have not yet completed peer review, and Amsterdam differs greatly from Indian cities. Still, it suggests that the question "can algorithmic visibility decide which destinations become popular?" is researchable, not merely rhetorical.

A third possibility follows from the first two: repeated recommendations may gradually shape everyday choices. If the same handful of routes and places is suggested day after day, habits may form around them. This article treats that as a hypothesis worth studying, not as a demonstrated effect.

Five dimensions for planning to 2046

This article proposes that planning for movement to 2046 should hold five dimensions together. They are offered as a working frame, not as a validated model.

• Human behaviour: the purposes behind trips (faith, habit, livelihood, aspiration and leisure), which no routing metric fully captures.
• Cultural geography: the meanings attached to particular streets, ghats, bazaars and public spaces, which may matter more than their efficiency.
• Spatial configuration: the structure of the street network, which space syntax research links to movement (Hillier & Hanson, 1984).
• Digital intelligence: the routing and recommendation systems that now mediate many trips, understood as active and imperfect producers of geographic knowledge (Kwan, 2016).
• Urban ecosystems: the wider interdependence of flows, land uses and services that Batty (2013) describes, in which a change to one flow alters others.

The central proposition is that algorithms should inform spatial decisions, not determine human choices. The future city, on this argument, must be technologically intelligent and spatially efficient, but also culturally rooted and human-centred.

Implications for Indian planning practice

Three situations illustrate why the frame may be useful in India. First, pilgrim and market streets, such as the lanes approaching major temples or the old bazaars of historic cores, carry movement driven by faith and trade. Route optimisation that pushes vehicles through or around them, without regard to their role, could erode what makes them work. Second, where navigation apps send traffic into residential lanes to bypass congested arterials, local bodies may need ways to monitor and respond, for example through traffic calming, clear street hierarchies and engagement with platforms on how local street classifications are represented. Third, tourism and heritage planning increasingly competes with online rankings. A heritage precinct that is poorly represented on mapping platforms may receive fewer visitors than its significance warrants, while a few highly ranked sites become crowded.

In practice, this suggests that mobility plans and heritage management plans could include a short review of how key streets and places appear on the main digital platforms, alongside conventional surveys. Indian evidence on these effects is limited, so such reviews would also help build the local evidence base this article lacks.

An earlier article on this blog looked at how plot-level rules shape the Indian street; this piece adds a digital layer to the same question of who authors the street.

Limitations

This is a conceptual article, not an empirical study. The routing and recommendation research cited comes largely from the United States and Europe, and from general recommender systems rather than urban places, so its transfer to Indian cities is untested. One key source is a preprint. The article has not drawn on Indian traffic or platform data, and the five dimensions have not been tested against any case. Claims about residential rat-running and about habit formation through recommendation are presented as concerns for research, not as findings.

Conclusion

Movement in cities has always had more than one author: the street network, the purposes people bring, and the institutions that manage both. Digital platforms are now a further author, and an influential one. The research reviewed here suggests that routing and recommendation systems can redistribute flows and visibility, though the size of these effects in Indian cities is unknown. This article has argued that planning should neither ignore these systems nor defer to them. Treating human behaviour, cultural geography, spatial configuration, digital intelligence and urban ecosystems together offers one way to keep algorithms in an advisory role, and to keep the choice of where to go with people.

References

Abdollahpouri, H., Burke, R., & Mobasher, B. (2017). Controlling popularity bias in learning-to-rank recommendation. In Proceedings of the Eleventh ACM Conference on Recommender Systems (pp. 42–46). ACM. https://doi.org/10.1145/3109859.3109912

Batty, M. (2013). The new science of cities. MIT Press. https://doi.org/10.7551/mitpress/9399.001.0001

Cabannes, T., Sangiovanni, M., Keimer, A., & Bayen, A. M. (2019). Regrets in routing networks. ACM Transactions on Spatial Algorithms and Systems, 5(2), 1–19. https://doi.org/10.1145/3325916

Hillier, B., & Hanson, J. (1984). The social logic of space. Cambridge University Press. https://doi.org/10.1017/CBO9780511597237

Kitchin, R. (2014). The real-time city? Big data and smart urbanism. GeoJournal, 79(1), 1–14. https://doi.org/10.1007/s10708-013-9516-8

Kwan, M.-P. (2016). Algorithmic geographies: Big data, algorithmic uncertainty, and the production of geographic knowledge. Annals of the American Association of Geographers, 106(2), 274–282. https://doi.org/10.1080/00045608.2015.1117937

Shaker, R. A., & Uitermark, J. (2026). Seeing Amsterdam through Google Maps: Uneven visibility and crowd-based mapping [Preprint]. SocArXiv. https://doi.org/10.31235/osf.io/fd97a_v2