pith. sign in

arxiv: 2604.17643 · v1 · submitted 2026-04-19 · ⚛️ physics.soc-ph · cs.IT· math.IT

Is segregation encoded in urban form? An entropy-based analysis

Vinicius M. Netto , Caio Cacholas , Camila Carvalho , Edgardo Brigatti This is my paper

Pith reviewed 2026-05-10 04:52 UTC · model grok-4.3

classification ⚛️ physics.soc-ph cs.ITmath.IT
keywords urban formresidential segregationbuilt-form entropySao Pauloincome distributionspatial clusteringentropy analysisurban morphology
0
0 comments X

The pith

Built form encodes residential segregation in non-linear entropy patterns.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper examines whether the physical arrangement of buildings can reflect and shape residential segregation by income. It measures built-form entropy from building footprints and aligns it with income distribution and clustering metrics through a shared spatial grid in Sao Paulo. The results show income levels and local clustering rising at both low and high ends of the entropy range, more sharply on the high-entropy side. This matters because it treats urban structure as an active participant in social separation rather than a passive setting shaped only by economics or policy.

Core claim

We find non-linear relationships between built-form entropy, income, and segregation: income levels and residential clustering increase toward both extremes of the entropy spectrum, with a stronger rise at the high-entropy end. This asymmetry suggests that high-entropy urban forms are associated with distinct spatial processes of segregation, including elite enclaving and incremental development in lower-income settlements, while low-entropy forms reflect more selective occupation shaped by planning and market filtering. Overall, the findings suggest that built form is more than a neutral backdrop, functioning as both affordance and signal of segregation.

What carries the argument

Built-form entropy (BFE), a Shannon entropy measure computed from building footprints and aligned with income data through regular spatial tessellation.

If this is right

  • Income levels and residential clustering rise toward both low and high ends of the BFE spectrum.
  • High-entropy urban forms associate with elite enclaving and incremental development in lower-income settlements.
  • Low-entropy forms reflect selective occupation shaped by planning and market filtering.
  • Built form functions as both affordance and signal of segregation.
  • The association is asymmetric, with stronger effects at the high-entropy end.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same entropy-segregation pattern could be tested in other large cities to check if it is a general feature of urban growth.
  • Urban design policies might target intermediate entropy levels to reduce the conditions that support concentrated income groups.
  • The framework could be extended to track how changes in building form over time correspond to shifts in segregation.
  • It links physical morphology directly to social outcomes, suggesting morphology metrics could complement traditional economic models of inequality.

Load-bearing premise

That built-form entropy regimes associate with the spatial distribution of income groups and their local clustering in non-linear ways.

What would settle it

A re-analysis of Sao Paulo building and income data showing linear relationships or no association between BFE values and Gini or Moran’s I measures of income segregation.

Figures

Figures reproduced from arXiv: 2604.17643 by Caio Cacholas, Camila Carvalho, Edgardo Brigatti, Vinicius M. Netto.

Figure 1
Figure 1. Figure 1: (a) Building footprints, tract-level average per-capita income and the regular 3×3 km grid. (b) Method schematic: overlapping the footprint and income layers to com￾pute built-form entropy and residential-segregation metrics on the grid. (c) Grid-cell detail showing high built-form entropy, tract-level per-capita income, and Moran’s I (spatial au￾tocorrelation). Sources: OpenStreetMap/GeoSampa (building fo… view at source ↗
Figure 2
Figure 2. Figure 2: (a) Building footprint (background) and entropy estimation. The urban fabric exhibits spatial continuity in estimated hc. (b) Hot Spot Analysis of BFE with values com￾puted for the 3×3 km grid cells. Red cells indicate statistically significant clusters of high entropy (hot spots), and blue cells indicate clusters of low entropy (cold spots). entropy exhibits clusters of similar values, with low entropy ac… view at source ↗
Figure 3
Figure 3. Figure 3: Income and residential segregation in São Paulo: (a1) The citywide distribution of per-capita income by census tract shows a remarkable centre–periphery pattern; (a2) per capita income distribution averaged at the grid cell scale; (b1) Moran’s I computed per each 3×3 km grid cell; (b2) results for Moran’s 6 I [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Built-form entropy, income distribution and local residential segregation in São Paulo. Top panels: (a) corrected built-form entropy (hc) per grid cell; (b) average per￾capita income, income inequality (Gini) and residential segregation (Moran’s I) on the same grid. Cells outlined with thicker black lines in (a) and (b) mark areas whose values lie near the estimated threshold. (c) Scatterplot of per-capita… view at source ↗
Figure 5
Figure 5. Figure 5: Selected configurations of built-form entropy, income distribution, and residen￾tial clustering. Each column shows a 3×3 km cell. (a) Cell 11 illustrates high built-form entropy, high Gini, and strong local clustering, where income clusters co-occur with con￾trasting morphologies. (b) Cell 27 represents a limiting case of high built-form entropy with a nearly homogeneous high-income group and low internal … view at source ↗
Figure 6
Figure 6. Figure 6: Selected 3×3 km cells from the high-entropy cluster (cells 17 and 35) and the low-entropy clusters (37 and 66). Rows show: (a) built form distributions; (b) tract-level average per-capita income, and (c) local residential clustering of income groups (Moran’s I). Cell 17 (Vila Sonia - Morumbi) ranks first citywide in BFE, while Morumbi has the highest average per capita income. Cell 35 (Pinheiros–Perdizes) … view at source ↗
Figure 7
Figure 7. Figure 7: Illustration of the method for estimating built-form entropy (BFE) in cells 37 (top, historic CBD) and 35 (bottom, Pinheiros–Perdizes). (a) The algorithm scans built-form maps in 3×3 km cells and counts the frequencies of local configurations across blocks of different sizes. (b) Zoomed-in views of the building-footprint arrangement. (c) Detection process for a selected configuration with n = 16. Insets ou… view at source ↗
read the original abstract

The footprints of residential segregation have long been documented, yet the role of urban form as both medium and manifestation of segregation remains under-specified. We investigate whether the configuration of the built fabric may encode residential segregation in its spatial structure, hypothesising that built-form entropy (BFE) regimes are associated with the spatial distribution of income groups and their local clustering in non-linear ways. We examine this by quantifying BFE through a Shannon-based measure computed from building footprints, characterising income-based distributions using the Gini index and Moran's I, and placing both on a common spatial footing through a regular tessellation. Applying this framework to Sao Paulo, Latin America's largest city, we find non-linear relationships between BFE, income, and segregation: income levels and residential clustering increase toward both extremes of the entropy spectrum, with a stronger rise at the high-entropy end. This asymmetry suggests that high-entropy urban forms are associated with distinct spatial processes of segregation, including elite enclaving and incremental development in lower-income settlements, while low-entropy forms reflect more selective occupation shaped by planning and market filtering. Overall, the findings suggest that built form is more than a neutral backdrop, functioning as both affordance and signal of segregation.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper claims that built-form entropy (BFE), measured via Shannon entropy on building footprints, encodes residential segregation in its spatial structure. Using a regular tessellation to align BFE with income-based metrics (Gini index for distribution and Moran's I for clustering) in São Paulo, the authors report non-linear relationships: income levels and residential clustering increase toward both extremes of the BFE spectrum, with a stronger rise at the high-entropy end. This asymmetry is interpreted as evidence that high-entropy forms are associated with elite enclaving and incremental development, while low-entropy forms reflect selective occupation, positioning built form as both affordance and signal of segregation rather than a neutral backdrop.

Significance. If the reported non-linear associations prove robust, the work would be significant for socio-physics and urban studies by providing a quantitative demonstration that physical urban morphology is intertwined with socio-economic segregation processes. The entropy-based framework offers a fresh lens beyond standard indices, with potential implications for how urban form shapes and reflects inequality in rapidly growing cities like São Paulo.

major comments (2)
  1. [Abstract (framework description)] The central claim that BFE regimes are associated with income distribution and clustering in non-linear ways (stronger at high-entropy) is load-bearing on the regular tessellation used to place BFE and income metrics on a common grid. The abstract describes this alignment but reports no sensitivity analysis to cell size. If the non-linearity disappears or reverses at finer or coarser resolutions, the association may be an artifact of the arbitrary spatial aggregation rather than genuine encoding of segregation in urban form. This requires explicit robustness checks to support the hypothesis.
  2. [Abstract (results description)] The abstract states that non-linear relationships were found but supplies no statistical details such as regression coefficients, p-values, error bars, sample sizes per tessellation cell, or robustness to data exclusion rules. Without these, it is difficult to evaluate whether the asymmetry (stronger rise at high-entropy) is statistically reliable or driven by outliers in the São Paulo dataset.
minor comments (1)
  1. [Abstract] Clarify the exact definition and computation of BFE (e.g., how building footprints are categorized into types for the Shannon entropy calculation) to allow replication.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive report. The comments highlight important aspects of robustness and statistical transparency that we will address in the revision. Below we respond point by point to the major comments.

read point-by-point responses

  1. Referee: [Abstract (framework description)] The central claim that BFE regimes are associated with income distribution and clustering in non-linear ways (stronger at high-entropy) is load-bearing on the regular tessellation used to place BFE and income metrics on a common grid. The abstract describes this alignment but reports no sensitivity analysis to cell size. If the non-linearity disappears or reverses at finer or coarser resolutions, the association may be an artifact of the arbitrary spatial aggregation rather than genuine encoding of segregation in urban form. This requires explicit robustness checks to support the hypothesis.

    Authors: We agree that demonstrating invariance to tessellation scale is necessary to rule out aggregation artifacts. The primary analysis uses a 1 km regular grid chosen to balance spatial resolution with data coverage in São Paulo. In the revised manuscript we will add an explicit sensitivity analysis section (or appendix) that recomputes BFE, Gini, and Moran's I at 500 m and 2 km resolutions, reports the corresponding quadratic regression coefficients and p-values, and shows that the non-linear pattern and the stronger high-entropy asymmetry persist across scales. This will directly test the referee's concern. revision: yes

  2. Referee: [Abstract (results description)] The abstract states that non-linear relationships were found but supplies no statistical details such as regression coefficients, p-values, error bars, sample sizes per tessellation cell, or robustness to data exclusion rules. Without these, it is difficult to evaluate whether the asymmetry (stronger rise at high-entropy) is statistically reliable or driven by outliers in the São Paulo dataset.

    Authors: The full manuscript already contains quadratic regression results (including coefficients for the linear and squared BFE terms, associated p-values, R² values, and 95 % confidence intervals) together with cell-level sample sizes and a description of exclusion criteria (cells with fewer than 10 buildings or incomplete income data). To improve accessibility we will revise the abstract to include concise quantitative statements, for example noting the significance of the quadratic term and the direction of the high-entropy effect. We will also add a short paragraph on robustness to alternative exclusion thresholds in the methods if it is not already explicit. revision: yes

Circularity Check

0 steps flagged

No circularity; empirical metrics aligned on independent tessellation

full rationale

The paper defines BFE via Shannon entropy on building footprints, computes Gini and Moran's I separately on income data, and aligns the two via a regular tessellation before reporting observed non-linear associations. No step fits a parameter to the target relationship and then renames the fit as a prediction, no self-citation supplies a uniqueness theorem or ansatz that the central claim rests upon, and the derivation does not reduce any reported association to its own inputs by construction. The analysis remains an open empirical comparison against external spatial data.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

Analysis rests on standard spatial statistics and entropy applied to building footprints; no explicit free parameters or new entities are introduced in the abstract, but several domain assumptions about what the measures capture are required.

axioms (2)
  • domain assumption Shannon entropy computed from building footprints yields a meaningful regime of urban form that can be compared across space.
    Invoked when defining BFE and placing it on a common spatial footing with income measures.
  • domain assumption Gini index and Moran's I on income data adequately represent distribution and local clustering for the purpose of detecting segregation.
    Used without further justification to characterise income-based distributions.

pith-pipeline@v0.9.0 · 5524 in / 1404 out tokens · 41434 ms · 2026-05-10T04:52:48.139305+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

43 extracted references · 43 canonical work pages

  1. [1]

    Burgess, E. W. Residential Segregation in American Cities.The ANNALS of the American Academy of Political and Social Science140, 105–115 (1928). URLhttps://journals.sagepub.com/doi/10. 1177/000271622814000115

    work page 1928

  2. [2]

    Wong, D. W. Formulating a General Spatial Segregation Measure.The Professional Geographer57, 285– 294 (2005). URLhttp://www.tandfonline.com/doi/abs/10.1111/j.0033-0124.2005.00478.x

    work page doi:10.1111/j.0033-0124.2005.00478.x 2005

  3. [3]

    & Rosalino, D

    Netto, V., Krenz, K., Fiszon, M., Peres, O. & Rosalino, D. Decoding segregation: Navigating a century of seg- regation research across disciplines and introducing a bottom-up ontology.arXiv preprint arXiv:2410.08374 (2024)

    work page arXiv 2024

  4. [4]

    Poverty maps of london (1889)

    Booth, C. Poverty maps of london (1889)

    work page

  5. [5]

    Balch, E. G. Hull house maps and papers (1895)

    work page

  6. [6]

    Duncan, O. D. & Duncan, B. A Methodological Analysis of Segregation Indexes.American Sociological Review20, 210 (1955). URLhttp://www.jstor.org/stable/2088328?origin=crossref

    work page arXiv 1955

  7. [7]

    James, D. R. & Taeuber, K. E. Measures of segregation.Sociological methodology15, 1–32 (1985)

    work page 1985

  8. [8]

    & Finizza, A

    Theil, H. & Finizza, A. J. A note on the measurement of racial integration of schools by means of informa- tional concepts (1971)

    work page 1971

  9. [9]

    Reardon, S. F . & O’Sullivan, D. Measures of Spatial Segregation.Sociological Methodology34, 121–162 (2004). URLhttps://journals.sagepub.com/doi/10.1111/j.0081-1750.2004.00150.x

    work page doi:10.1111/j.0081-1750.2004.00150.x 2004

  10. [10]

    On the measure of geographic segregation

    Morrill, R. On the measure of geographic segregation. InGeography research forum, vol. 11, 25–36 (1991)

    work page 1991

  11. [11]

    Wong, D. W. S. Spatial Indices of Segregation.Urban Studies30, 559–572 (1993). URLhttps:// journals.sagepub.com/doi/10.1080/00420989320080551

    work page doi:10.1080/00420989320080551 1993

  12. [12]

    The relationship between physical segregation and social marginalisation in the urban envi- ronment.World Architecture185, 88–96 (2005)

    Vaughan, L. The relationship between physical segregation and social marginalisation in the urban envi- ronment.World Architecture185, 88–96 (2005)

    work page 2005

  13. [13]

    The spatial syntax of urban segregation.Progress in Planning67, 199–294 (2007)

    Vaughan, L. The spatial syntax of urban segregation.Progress in Planning67, 199–294 (2007)

    work page 2007

  14. [14]

    Saboya, R. T. d. & Peres, O. M. Chasm: A configurational measure of socio-spatial residential segregation. Environment and Planning B: Urban Analytics and City Science52, 1464–1481 (2025). 21

    work page 2025

  15. [15]

    M., De Sousa Filho, F

    Brigatti, E., Netto, V. M., De Sousa Filho, F . N. M. & Cacholas, C. Entropy and hierarchical clustering: Characterizing the morphology of the urban fabric in different spatial cultures.Chaos: An Interdisciplinary Journal of Nonlinear Science31, 113138 (2021). URLhttps://pubs.aip.org/cha/article/31/11/ 113138/342078/Entropy-and-hierarchical-clustering-Cha...

    work page 2021

  16. [16]

    M., Brigatti, E

    Netto, V. M., Brigatti, E. & Cacholas, C. From urban form to information: Cellular configurations in different spatial cultures.Environment and Planning B: Urban Analytics and City Science50, 146–161 (2023). URL https://journals.sagepub.com/doi/10.1177/23998083221107382

    work page doi:10.1177/23998083221107382 2023

  17. [17]

    Hillier, B., Penn, A., Hanson, J., Grajewski, T. & Xu, J. Natural movement: or, configuration and attraction in urban pedestrian movement.Environment and Planning B: planning and design20, 29–66 (1993)

    work page 1993

  18. [18]

    & Korver-Glenn, E

    Roberto, E. & Korver-Glenn, E. The spatial structure and local experience of residential segregation.Spatial Demography9, 277–307 (2021)

    work page 2021

  19. [19]

    & Hanson, J.The social logic of space(Cambridge university press, 1984)

    Hillier, B. & Hanson, J.The social logic of space(Cambridge university press, 1984)

    work page 1984

  20. [20]

    Toms, J. D. & Lesperance, M. L. Piecewise regression: a tool for identifying ecological thresholds.Ecology 84, 2034–2041 (2003)

    work page 2034

  21. [21]

    & Owen, D

    Johnston, R., Jones, K., Manley, D. & Owen, D. Macro-scale stability with micro-scale diversity: modelling changing ethnic minority residential segregation–london 2001–2011.Transactions of the Institute of British Geographers41, 389–402 (2016)

    work page 2001

  22. [22]

    Fowler, C

    S. Fowler, C. Segregation as a multiscalar phenomenon and its implications for neighborhood-scale re- search: the case of south seattle 1990–2010.Urban geography37, 1–25 (2016)

    work page 1990

  23. [23]

    Catney, G. The complex geographies of ethnic residential segregation: Using spatial and local measures to explore scale-dependency and spatial relationships.Transactions of the Institute of British Geographers 43, 137–152 (2018)

    work page 2018

  24. [24]

    Alonso, W.Location and land use: Toward a general theory of land rent(Harvard university press, 1964)

    work page 1964

  25. [25]

    K., Thisse, J.-F

    Brueckner, J. K., Thisse, J.-F . & Zenou, Y . Why is central paris rich and downtown detroit poor?: An amenity-based theory.European economic review43, 91–107 (1999)

    work page 1999

  26. [26]

    & Hurst, E

    Guerrieri, V., Hartley, D. & Hurst, E. Endogenous gentrification and housing price dynamics.Journal of Public Economics100, 45–60 (2013)

    work page 2013

  27. [27]

    & Rey, S

    Knaap, E. & Rey, S. Segregated by design? street network topological structure and the measurement of urban segregation.Environment and Planning B: Urban Analytics and City Science51, 1408–1429 (2024)

    work page 2024

  28. [28]

    Schelling, T. C. Dynamic models of segregation.Journal of mathematical sociology1, 143–186 (1971)

    work page 1971

  29. [29]

    The spatial proximity and connectivity method for measuring and analyzing residential segre- gation.Sociological Methodology48, 182–224 (2018)

    Roberto, E. The spatial proximity and connectivity method for measuring and analyzing residential segre- gation.Sociological Methodology48, 182–224 (2018)

    work page 2018

  30. [30]

    Shannon, C. E. A mathematical theory of communication.Bell system technical journal27, 379–423 (1948)

    work page 1948

  31. [31]

    Shannon, C. E. Prediction and entropy of printed english.Bell system technical journal30, 50–64 (1951)

    work page 1951

  32. [32]

    & Grassberger, P

    Schürmann, T. & Grassberger, P . Entropy estimation of symbol sequences.Chaos: An Interdisciplinary Journal of Nonlinear Science6, 414–427 (1996)

    work page 1996

  33. [33]

    Cover, T. M. & Thomas, J. A. Elements of information theory, 1991 john wiley & sons.Inc. Print ISBN 0-471-06259-6 Online ISBN 0-471-20061-1(1991)

    work page 1991

  34. [34]

    universal and accessible entropy estimation using a com- pression algorithm

    Brigatti, E. & de Sousa Filho, F . Comment on “universal and accessible entropy estimation using a com- pression algorithm”.Physical Review Letters129, 029801 (2022)

    work page 2022

  35. [35]

    M.et al.Cities, from information to interaction.Entropy20, 834 (2018)

    Netto, V. M.et al.Cities, from information to interaction.Entropy20, 834 (2018)

    work page 2018

  36. [36]

    & Brigatti, E

    de Sousa Filho, F ., Pereira de Sá, V. & Brigatti, E. Entropy estimation in bidimensional sequences.Physical Review E105, 054116 (2022)

    work page 2022

  37. [37]

    & Ord, J

    Getis, A. & Ord, J. K. The analysis of spatial association by use of distance statistics.Geographical analysis 24, 189–206 (1992)

    work page 1992

  38. [38]

    B.Censo demográfico 2010: Características da população e dos domicílios: Resultados do universo(IBGE, Rio de Janeiro, 2010)

    de Geografia e Estatística, I. B.Censo demográfico 2010: Características da população e dos domicílios: Resultados do universo(IBGE, Rio de Janeiro, 2010)

    work page 2010

  39. [39]

    URLhttps://cran.r-project.org/ package=DescTools

    Signorell, A.DescTools: Tools for Descriptive Statistics(2025). URLhttps://cran.r-project.org/ package=DescTools. R package version 0.99.59.7

    work page 2025

  40. [40]

    Frank, A. I. Using measures of spatial autocorrelation to describe socio-economic and racial residential patterns in us urban areas.Socio-economic applications of geographic information science (Innovations in GIS). London: Taylor & Francis147–162 (2003)

    work page 2003

  41. [41]

    Bivand, R. S. & Wong, D. W. S. Comparing implementations of global and local indicators of spatial asso- ciation.Test27, 716–748 (2018)

    work page 2018

  42. [42]

    & Piras, G.spdep: Spatial Dependence: Weighting Schemes, Statistics and Models (2024)

    Bivand, R., Krainski, E. & Piras, G.spdep: Spatial Dependence: Weighting Schemes, Statistics and Models (2024). URLhttps://cran.r-project.org. R package version 1.3-3

    work page 2024

  43. [43]

    Areal-weighted interpolation

    Prener, C. Areal-weighted interpolation. areal R package vignette (2022). URLhttps://cran. 22 r-project.org/web/packages/areal/vignettes/areal-weighted-interpolation.html. 23

    work page 2022