Pith. sign in

REVIEW 3 major objections 6 minor 30 references

The Path is the Goal: a Study on the Nature and Effects of Shortest-Path Stability Under Perturbation of Destination

T0 review · 3 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Small destination shifts can rewrite a route in any city, most strongly near the center.

desk verdict Useful empirical extension with reproducible code, but the headline distance-stability trend is likely a trip-length scaling artifact of the radial OD sampling—deserves peer review with mandatory reanalysis. read the letter →

arxiv 2506.09731 v1 pith:LT5SZZOM submitted 2025-06-11 cs.CY

Giuliano Cornacchia , Mirco Nanni This is my paper · ORCID
classification cs.CY
keywords shortestpathstabilityroadnetworkurbannavigationdestinationperturbationroutevariabilitycitycomparisonstructure
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper asks whether the common experience of navigation apps proposing completely different routes for two nearby destinations is a quirk of one city or a general property of urban road networks. It defines path stability—how much the shortest route to a destination changes when the destination is moved by at most 100 meters—and measures it for 46 cities using destination perturbations and a length-weighted Jaccard similarity between original and rerouted paths. The paper's central claim is that the pattern is universal but heterogeneous: stability increases with distance from the city center and plateaus near total stability in peripheral areas, following $y = 0.26e^{-0.43x} + 0.99$ with $R^2 = 0.98$. At the same time, cities divide into stable and unstable profiles, with unstable cities tending to have grid-like, fine-grained networks and longer detours between a destination and its perturbed neighbors. If the claim holds, route sensitivity to small destination shifts is not noise but a measurable, city-specific property tied to network geometry, with concrete uses in traffic distribution and navigation design.

What carries the argument

The central object is the path-stability score $S^{\Delta}(o,d)$ for an origin-destination pair: the destination is displaced by at most $\Delta = [0,100]$ meters, candidate displaced nodes are sampled in eight sectors of a ring around it, shortest paths are recomputed, and the average length-weighted Jaccard index between the original and displaced routes gives the score. The same score aggregated by radius bands and by city carries the analysis: the exponential fit $y = 0.26e^{-0.43x} + 0.99$ formalizes the center-periphery trend, while the normalized detour ratio $R$ and street-level indicators such as average street length, circuity, and bearing entropy explain cross-city differences.

What would settle it

Recompute the analysis with perturbation radii of 50 m, 200 m, and 500 m, and also with the radius scaled to each city's average street length; if the cross-city stability ranking or the exponential center-periphery curve changes substantially under rescaling, the fixed 100-meter band is the real driver of the result. Alongside that, a single city or district where median stability does not increase from center to periphery would break the claimed universal trend.

Watch

Extended reading notes

Core claim

The core discovery, stated on the paper's own terms, is that shortest-path instability under small destination perturbations is a universal but heterogeneous property of cities. Across all 46 cities studied, median stability increases with radial distance from the city center and plateaus near 1 beyond roughly 10 km; the aggregate curve is $y = 0.26e^{-0.43x} + 0.99$ with $R^2 = 0.98$. Cross-city differences align with network structure: cities with grid-like layouts, short average street segments, low orientation entropy, and large normalized detours between a destination and its displaced neighbors rank as unstable, while stable cities have longer, more varied segments and radial or organic layouts. The normalized detour ratio, comparing the path length from the original destination to a displaced one against the full trip length, is the strongest single correlate of instability at city level, with Pearson correlation $-0.831$.

Load-bearing premise

The load-bearing premise is that a 100-meter displacement of a destination counts as the same 'small change' in every city, even though average street length runs from 67 meters in Tokyo to 213 meters in Ottawa, so a fixed shift crosses very different numbers of decision points in different networks.

Editorial extensions

If this is right

  • Navigation providers can produce per-destination stability maps, flagging places where a user's small destination adjustment will likely produce a completely different route.
  • Traffic managers can use low-stability zones around event venues to disperse incoming vehicles across multiple routes instead of funneling them down one path.
  • The center-periphery curve gives a default expectation: beyond roughly 10 km from the city center, small destination shifts should barely change the recommended shortest path.
  • Road-network metrics—average street length, variation in segment length, circuity, and orientation entropy—can be used to predict whether a city or district will be route-stable before any routing simulation is run.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Because a fixed 100-meter shift crosses more decision points in fine-grained grids, the stability ranking across cities may partly encode network granularity rather than intrinsic route robustness; scaling the perturbation by local street length would test this directly.
  • The mechanism suggests a local, testable prediction: adding short cross-streets or increasing intersection density near a destination should lower its stability score, while removing connections should raise it, so the score could serve as a response variable in network-design experiments.
  • The paper treats stability as a descriptive lens, but the same score could be operationalized in routing: a navigation service could deliberately route through unstable zones during events to spread traffic, or keep to stable corridors when concentration on main arteries is desired.
  • Within-city maps show unstable destinations cluster together; a formal spatial-statistical test of that autocorrelation, beyond visual inspection, would quantify how localized the phenomenon is.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. The paper introduces a measure of 'path stability' for urban shortest paths under small perturbations of the destination, defined as the average weighted Jaccard similarity between the original path and paths to destinations displaced by Δ=[0,100] m. The method is applied to 46 cities using OSM data, with O-D pairs sampled from points on concentric circles centered on each city. The paper reports a universal exponential increase of stability with distance from the city center, cross-city correlations with average street length, circuity, and orientation entropy, a four-cluster typology of city networks, and spatial autocorrelation of unstable destinations. The work is a substantial extension of the authors' earlier workshop paper and ships open-source code.

Significance. If the central distance-stability trend were robust, the paper would provide a useful empirical descriptor of urban road networks with direct implications for navigation-system design and traffic management. The main strengths are the multi-city dataset, a clearly specified and reproducible measurement pipeline, and the public availability of the code. However, the headline trend is confounded by the O-D sampling design (trip length scales mechanically with ring radius), and cross-city comparisons rely on a fixed perturbation radius that is not commensurate across networks with very different granularity. The descriptive value is real, but the universal claims require additional controls and inferential statistics before they can be accepted.

major comments (3)
  1. [§3.3.1 and Fig. 4] The radial sampling pairs only points on the same concentric circle, so the expected shortest-path length p(o,d) increases linearly with ring radius r. For a fixed 100 m destination perturbation, the diverged suffix of the path is roughly constant while the shared prefix grows with p(o,d); hence the weighted Jaccard stability increases toward 1 as r grows for purely metric reasons. The exponential fit in Fig. 4 and the Section 6 claim that 'as distance from the city center increases, route stability grows' are therefore not separated from trip-length scaling. Please stratify by trip length or p(o,d) (e.g., regress stability on both radius and trip length, or compare OD pairs of matched length across rings) and report whether the distance-from-center effect persists.
  2. [§4 and Table 4] A fixed perturbation interval Δ=[0,100] m is applied to all cities, but average street length ranges from 67 m (Tokyo) to 213 m (Ottawa). In fine-grained networks a 100 m destination shift can cross many more decision points than in coarse networks, so the cross-city stability rankings and the correlation with average street length (Pearson=0.476, Fig. 7 left) may reflect this granularity mismatch rather than an intrinsic stability property. Please add a robustness check with a normalized perturbation radius (e.g., proportional to city-specific average street length) or report results for several values of Δ.
  3. [§5 (Figs. 4, 6, 7)] The city-level results are reported as Pearson correlations and an exponential R² without confidence intervals, p-values, or multiple-comparison correction. For n=46 cities, correlations around 0.4 are only marginally significant, and the word 'significant' in Q4/Q5 is not supported by inferential statistics. Please provide bootstrap confidence intervals or p-values for all reported correlations and fit parameters, and consider a mixed-effects model that accounts for city and ring effects.
minor comments (6)
  1. [Introduction] The phrase 'is this study' should be 'in this study'.
  2. [Section 5 (introductory paragraph)] 'realtive' should be 'relative'.
  3. [Table 4] 'Moskow' should be 'Moscow'.
  4. [Fig. 4] The text gives fit parameters a=−0.26, b=−0.43, c=0.99, but the caption's equation is garbled (it appears to omit the negative exponent and the sign of a); please correct the equation for clarity.
  5. [Section 5, Q5] The city-shape classification (Grid/Radial/Organic) is based on visual inspection; please include a reproducibility note or an inter-rater consistency check.
  6. [Section 5, Q5] The entropy-of-bearing computation is cited to [27] but the binning and calculation details are not described; a short description would improve reproducibility.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: path stability is empirically measured, and the only self-citation is not load-bearing; the radial-sampling/trip-length confound in Fig. 4 is a validity threat, not a by-construction reduction.

full rationale

This is an empirical measurement study rather than a derivation from first principles. Path stability S^Delta(o,d) is defined directly as the average weighted Jaccard similarity between shortest paths to a destination and to its perturbed counterparts, and all values are computed from OpenStreetMap data through shortest-path queries; no parameter is fitted to a subset of the data and then renamed as a prediction. The exponential curve in Fig. 4 is a descriptive fit to observed median stabilities, not a predictive model, and Q6 explicitly excludes stability from the clustering attributes. The only relevant self-citation is reference [2], the authors' prior workshop paper introducing a first formulation of path instability; the present paper fully redefines the measure, documents the perturbation and OD-sampling protocol, and releases code, so the self-citation is not load-bearing. A real threat to the causal interpretation of Fig. 4 exists, and the paper itself acknowledges the mechanism in Section 6: "short OD distances in these areas mean that p(d, dx) represents a larger fraction of the original path p(o, d), contributing to lower stability, as demonstrated by the negative relationship between R and stability." Because all OD pairs are chords of the same concentric circle, median trip length scales with radius, and stability is a ratio that naturally rises when the denominator (trip length) grows. This is a confound between 'distance from center' and trip length, but it is not a circular reduction: the stability values are not constructed from the explanatory variables, and radius, street length, entropy, and R are not fitted to produce stability. The finding may be over-interpreted, but no step in the paper's claimed derivation chain is equivalent to its inputs by definition.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The ledger captures the hand-chosen experimental parameters (perturbation radius, sector count, filtering threshold, fit parameters) and the domain assumptions (OSM comparability, latlong.net centers, length-based shortest paths, visual shape classification) on which the empirical claims rest.

free parameters (4)
  • Exponential distance-stability fit parameters a, b, c = a=0.26, b=0.43, c=0.99 (text inconsistently reports a=-0.26, b=-0.43)
    Fitted to city-averaged median stability versus radius in Figure 4; R²=0.98. Descriptive, not predictive.
  • Destination perturbation radius Δ and sector count k = Δ=[0,100] m, k=8
    Hand-chosen in Section 4 to represent small destination changes; not normalized to street-network grain, which confounds cross-city comparisons.
  • Perturbed-destination filtering threshold = p(d,d_x)>3 km excluded (ratio approximately 30, 95th percentile)
    Data-dependent exclusion rule in Section 4; removing long detour destinations tends to increase measured stability.
  • K-means cluster count = k=4 (elbow method)
    Selected by elbow method in Section 5 Q6; affects the resulting stability profiles.
assumptions (5)
  • domain assumption OpenStreetMap data is complete and equally comparable across all 46 cities
    Section 4 (Dataset) builds every network metric and path from OSM; global coverage completeness varies.
  • domain assumption Latlong.net coordinates are accurate city centers for radial OD sampling
    Section 3.3.1 and 4 sample concentric circles around these coordinates; an off-center choice shifts the OD distribution.
  • domain assumption The shortest path by length, computed by OSMnx, is the appropriate navigation route
    Section 4 (Shortest path computation) uses minimum length; real navigators typically optimize time and traffic.
  • domain assumption Grid, Radial, and Organic shape labels assigned by visual inspection are reliable
    Section 5 Q5 classifies cities by visual inspection without inter-rater validation; Table 1 feeds shape-based conclusions.
  • domain assumption The fixed-radius concentric-circle OD sample represents real trip patterns
    Section 3.3.1 samples geometrically uniform points, not actual travel demand, so stability distributions may not match real usage.

how reviews work

0 comments
Cite this review

Pith. "Pith review of The Path is the Goal: a Study on the Nature and Effects of Shortest-Path Stability Under Perturbation of Destination." pith.science (2026). https://pith.science/paper/LT5SZZOM

@misc{pith2026250609731,
  author       = {Pith},
  title        = {Pith review of: The Path is the Goal: a Study on the Nature and Effects of Shortest-Path Stability Under Perturbation of Destination},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LT5SZZOM}},
  note         = {Machine review of arXiv:2506.09731}
}
read the original abstract

This work examines the phenomenon of path variability in urban navigation, where small changes in destination might lead to significantly different suggested routes. Starting from an observation of this variability over the city of Barcelona, we explore whether this is a localized or widespread occurrence and identify factors influencing path variability. We introduce the concept of "path stability", a measure of how robust a suggested route is to minor destination adjustments, define a detailed experimentation process and apply it across multiple cities worldwide. Our analysis shows that path stability is shaped by city-specific factors and trip characteristics, also identifying some common patterns. Results reveal significant heterogeneity in path stability across cities, allowing for categorization into "stable" and "unstable" cities. These findings offer new insights for urban planning and traffic management, highlighting opportunities for optimizing navigation systems to enhance route consistency and urban mobility.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

30 extracted references · 25 canonical work pages

  1. [1]

    In: Proceedings of the 12th ACM SIGSPATIAL International Workshop on Compu- tational Transportation Science

    Cheng, D., Gkountouna, O., Z¨ ufle, A., Pfoser, D., Wenk, C.: Shortest-path diver- sification through network penalization: A washington dc area case study. In: Proceedings of the 12th ACM SIGSPATIAL International Workshop on Compu- tational Transportation Science. IWCTS’19. Association for Computing Machin- ery, New York, NY, USA (2019).https://doi.org/1...

    work page arXiv 2019

  2. [2]

    In: EDBT/ICDT Workshops

    Cornacchia, G., Nanni, M., Grassi, F.: Barcelona effect: Studying the instability of shortest paths in urban settings. In: EDBT/ICDT Workshops. Big Mobility Data Analytics (2024)

    work page 2024

  3. [3]

    Barth´ elemy, M., Flammini, A.: Modeling urban street patterns. Phys. Rev. Lett. 100, 138702 (2008) https://doi.org/10.1103/PhysRevLett.100.138702

    work page doi:10.1103/physrevlett.100.138702 2008

  4. [4]

    Chaos, Solitons & Fractals174(2023) https: //doi.org/10.1016/j.chaos.2023.113770

    Lee, M., Cheon, S., Son, S.-W., Lee, M.J., Lee, S.: Exploring the relationship between the spatial distribution of roads and universal pattern of travel-route efficiency in urban road networks. Chaos, Solitons & Fractals174(2023) https: //doi.org/10.1016/j.chaos.2023.113770

    arXiv 2023

  5. [5]

    Physical Review E73(3), 036125 (2006)

    Crucitti, P., Latora, V., Porta, S.: Centrality measures in spatial networks of 21 urban streets. Physical Review E73(3), 036125 (2006)

    work page 2006

  6. [6]

    Physica A: Statistical Mechanics and its Applications 363(1), 89–95 (2006) https://doi.org/10.1016/j.physa.2006.01.051

    L¨ ammer, S., Gehlsen, B., Helbing, D.: Scaling laws in the spatial structure of urban road networks. Physica A: Statistical Mechanics and its Applications 363(1), 89–95 (2006) https://doi.org/10.1016/j.physa.2006.01.051 . Information and Material Flows in Complex Networks

    doi:10.1016/j.physa.2006.01.051 2006

  7. [7]

    The European Physical Journal B84, 563–577 (2011)

    Chan, S.H., Donner, R.V., L¨ ammer, S.: Urban road networks—spatial networks with universal geometric features? a case study on germany’s largest cities. The European Physical Journal B84, 563–577 (2011)

    work page 2011

  8. [8]

    Kalapala, V., Sanwalani, V., Clauset, A., Moore, C.: Scale invariance in road networks. Phys. Rev. E73, 026130 (2006) https://doi.org/10.1103/PhysRevE.73. 026130

    work page doi:10.1103/physreve.73 2006

Show all 30 references
  1. [9]

    Regional Science and Urban Economics39(6), 732–738 (2009) https: //doi.org/10.1016/j.regsciurbeco.2009.07.003

    Levinson, D., El-Geneidy, A.: The minimum circuity frontier and the journey to work. Regional Science and Urban Economics39(6), 732–738 (2009) https: //doi.org/10.1016/j.regsciurbeco.2009.07.003

    2009 doi

  2. [10]

    Nature communications8(1), 2229 (2017)

    Lee, M., Barbosa, H., Youn, H., Holme, P., Ghoshal, G.: Morphology of travel routes and the organization of cities. Nature communications8(1), 2229 (2017)

    2017

  3. [11]

    Wu, L., Xiao, X., Deng, D., Cong, G., Zhu, A.D., Zhou, S.: Shortest path and distance queries on road networks: An experimental evaluation (2012) https:// doi.org/10.48550/ARXIV.1201.6564

    2012 doi arXiv

  4. [12]

    SUMO Conference Proceedings4, 75–87 (2023) https://doi.org/10.52825/scp.v4i.217

    Cornacchia, G., Nanni, M., Pedreschi, D., Pappalardo, L.: The effects of route ran- domization on urban emissions. SUMO Conference Proceedings4, 75–87 (2023) https://doi.org/10.52825/scp.v4i.217

    2023 doi

  5. [13]

    Roughgarden, T., Tardos, E.: How bad is selfish routing? Journal of the ACM49 (2002) https://doi.org/10.1145/506147.506153

    2002

  6. [14]

    In: Proceedings of the 30th International Conference on Advances in Geographic Information Systems, pp

    Cornacchia, G., B¨ ohm, M., Mauro, G., Nanni, M., Pedreschi, D., Pappalardo, L.: How routing strategies impact urban emissions. In: Proceedings of the 30th International Conference on Advances in Geographic Information Systems, pp. 1–4 (2022)

    2022

  7. [15]

    Fluctuation and Noise Letters, 2450016 (2023)

    Cornacchia, G., Nanni, M., Pedreschi, D., Pappalardo, L.: Navigation services and urban sustainability. Fluctuation and Noise Letters, 2450016 (2023)

    2023

  8. [16]

    https://arxiv

    Cornacchia, G., Nanni, M., Pedreschi, D., Pappalardo, L.: Navigation services amplify concentration of traffic and emissions in our cities (2024). https://arxiv. org/abs/2407.20004

    2024 arXiv

  9. [17]

    IEEE Trans

    Li, L., Cheema, M., Lu, H., Ali, M., Toosi, A.N.: Comparing alternative route planning techniques: A comparative user study on melbourne, dhaka and 22 copenhagen road networks. IEEE Trans. Knowl. Data Eng.34(11), 5552–5557 (2022)

    2022

  10. [18]

    New Journal of Physics16(1), 13012 (2014) https://doi.org/10.1088/ 1367-2630/16/1/013012

    Wang, P., Liu, L., Li, X., Li, G., Gonz´ alez, M.C.: Empirical study of long-range connections in a road network offers new ingredient for navigation optimization models. New Journal of Physics16(1), 13012 (2014) https://doi.org/10.1088/ 1367-2630/16/1/013012

    2014

  11. [19]

    European Jour- nal of Operational Research121(2), 232–246 (2000) https://doi.org/10.1016/ S0377-2217(99)00214-3

    Akg¨ un, V., Erkut, E., Batta, R.: On finding dissimilar paths. European Jour- nal of Operational Research121(2), 232–246 (2000) https://doi.org/10.1016/ S0377-2217(99)00214-3

    2000

  12. [20]

    http://www.camvit.com (2005)

    Ltd., C.V.I.T.: Choice Routing. http://www.camvit.com (2005)

    2005

  13. [21]

    The VLDB Journal29(5), 1023–1047 (2020) https://doi.org/10.1007/s00778-020-00604-x

    Chondrogiannis, T., Bouros, P., Gamper, J., Leser, U., Blumenthal, D.B.: Find- ing k-shortest paths with limited overlap. The VLDB Journal29(5), 1023–1047 (2020) https://doi.org/10.1007/s00778-020-00604-x

    2020 doi

  14. [22]

    SIGSPATIAL ’18, pp

    Chondrogiannis, T., Bouros, P., Gamper, J., Leser, U., Blumenthal, D.B.: Finding k-dissimilar paths with minimum collective length. SIGSPATIAL ’18, pp. 404–

  15. [23]

    In: ACM SIGSPATIAL GIS, pp

    H¨ acker, C., Bouros, P., Chondrogiannis, T., Althaus, E.: Most diverse near- shortest paths. In: ACM SIGSPATIAL GIS, pp. 229–239 (2021)

    2021

  16. [24]

    Campbell, M.E.: Route selection and traffic assignment: A compendium of correspondence relating to a suggested technique. (1950). Highway Research Board

    1950

  17. [25]

    SIGSPATIAL ’23

    Cornacchia, G., Nanni, M., Pappalardo, L.: One-shot traffic assignment with forward-looking penalization. SIGSPATIAL ’23. Association for Computing Machinery, New York, NY, USA (2023). https://doi.org/10.1145/3589132. 3625637 .https://doi.org/10.1145/3589132.3625637

    2023

  18. [26]

    Costa, L.: Further generalizations of the jaccard index

    F. Costa, L.: Further generalizations of the jaccard index. CoRR abs/2110.09619(2021) 2110.09619

    2021 arXiv

  19. [27]

    Applied Network Science4(1), 1–19 (2019)

    Boeing, G.: Urban spatial order: Street network orientation, configuration, and entropy. Applied Network Science4(1), 1–19 (2019)

    2019

  20. [28]

    [Online; accessed 2024-10-29] (2024)

    Miller, A.L., Berlo, J.C., Wolf, B.J., Roberts, J.L.: Designing Cities, Partitioning Land, Imaging Utopia. [Online; accessed 2024-10-29] (2024)

    2024

  21. [29]

    2nd edition, Pearson publisher, USA

    Tan, P.-N., Steinbach, M., Karpatne, A., Kumar, V.: Introduction to Data Mining. 2nd edition, Pearson publisher, USA. (2018) 23

    2018

  22. [407]

    https: //doi.org/10.1145/3274895.3274903

    Association for Computing Machinery, New York, NY, USA (2018). https: //doi.org/10.1145/3274895.3274903

    2018

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.