pith. machine review for the scientific record. sign in

arxiv: 2604.23034 · v1 · submitted 2026-04-24 · 💻 cs.SI

Recognition: unknown

The Decay of Impact with Network Distance in Linear Diffusion Processes

Alexander Murray-Watters, Carter T. Butts, Cheng Wang, Cynthia Lakon, John R. Hipp

Pith reviewed 2026-05-08 08:55 UTC · model grok-4.3

classification 💻 cs.SI
keywords social networkslinear diffusioneigenvector centralityimpact decaynetwork distanceadjacency spectrumsocial influence
0
0 comments X

The pith

Total impact between nodes in linear diffusion decays exponentially with network distance, approximated by eigenvector centrality product and spectral terms.

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

The paper derives an approximate expression for the total influence one node exerts on another in forced linear diffusion models on networks. This expression shows the impact as primarily a product of the nodes' eigenvector centralities times a factor that decreases exponentially with their distance, derived from the eigenvalues of the adjacency matrix. Sympathetic readers would care because models like social influence or status scoring rely on such impacts, yet calculating exact totals from all paths is complex. Numerical checks on school networks confirm the average exponential decay and the utility of the first-order approximation. The result allows simpler estimation of influences in social processes.

Core claim

Here, we provide an approximate solution for the total impact of one node on another as a function of network distance, showing that the total impact is given to first order by a product of eigenvector centrality scores together with an expression in terms of the graph spectrum (eigenvalues of the adjacency matrix) that falls exponentially with distance. We also show how this solution can be refined using higher-order eigenvectors of the adjacency matrix. A numerical study on interpersonal networks drawn from educational settings verifies an average exponential decline in impact strength under the linear diffusion model, and shows that the first-order eigenvector approximation can often be a

What carries the argument

The first-order term from the leading eigenvector of the adjacency matrix paired with the dominant eigenvalue, which together produce the exponential decay in summed path contributions.

If this is right

  • Total impact falls exponentially with distance on average in linear diffusion processes.
  • The product of eigenvector centrality scores serves as a practical proxy for pairwise total impact.
  • Higher-order eigenvectors of the adjacency matrix can be used to refine the first-order approximation.
  • This yields a simple model for estimating influence or status effects in social networks without full path summation.

Where Pith is reading between the lines

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

  • The approximation may enable faster influence calculations on large networks by avoiding repeated full solutions of the diffusion equation.
  • Local interaction models could be justified when distant impacts become negligible due to the exponential term.
  • Similar spectral approximations might extend to directed networks or those with weighted edges if the adjacency structure permits.

Load-bearing premise

That the first-order term involving only the leading eigenvector and dominant spectral contribution provides a good proxy for total impact across graphs and diffusion parameters in social networks.

What would settle it

A network where exact total impacts computed from the diffusion model show no exponential decay with distance or deviate strongly from the predicted product of eigenvector centralities.

Figures

Figures reproduced from arXiv: 2604.23034 by Alexander Murray-Watters, Carter T. Butts, Cheng Wang, Cynthia Lakon, John R. Hipp.

Figure 1
Figure 1. Figure 1: Network diffusion under the FLDM. (A) Impact of node view at source ↗
Figure 2
Figure 2. Figure 2: Mean total impact as a function of distance from ego, by network. Left-hand panel view at source ↗
Figure 3
Figure 3. Figure 3: Level of correlation between the first order approximation and the true geodesic view at source ↗
Figure 4
Figure 4. Figure 4: Level of correlation between the first order approximation and the true geodesic view at source ↗
Figure 5
Figure 5. Figure 5: Correlation for first and second order eigenvectors as decay increases. Note that view at source ↗
read the original abstract

Many processes related to status, power, and influence within social networks have been modeled using forced linear diffusion models; examples include the highly successful Friedkin-Johnsen model of social influence, the status/power scores of Katz and Bonacich, and the widely used network autocorrelation model. While a basic assumption of such models is that the impact of one individual on another through any given path falls exponentially with path length, the total impact of the first individual on the second involves contributions from walks of all lengths; thus, while total impact is expected to decline with network distance, the relationship is not trivial. Here, we provide an approximate solution for the total impact of one node on another as a function of network distance, showing that the total impact is given to first order by a product of eigenvector centrality scores together with an expression in terms of the graph spectrum (eigenvalues of the adjacency matrix) that falls exponentially with distance. We also show how this solution can be refined using higher-order eigenvectors of the adjacency matrix. A numerical study on interpersonal networks drawn from educational settings verifies an average exponential decline in impact strength under the linear diffusion model, and shows that the first-order eigenvector approximation can often be a good proxy for total impact as obtained from the exact solution. This suggests a simple model that can be used to approximate total impact for social influence or status processes in a range of settings.

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

1 major / 2 minor

Summary. The manuscript derives an approximate closed-form expression for the total impact of one node on another under linear diffusion processes on graphs. To first order, this impact is proportional to the product of the nodes' components in the leading eigenvector of the adjacency matrix (i.e., eigenvector centralities) multiplied by a spectral factor involving the dominant eigenvalue that decays exponentially with shortest-path distance. Higher-order corrections using additional eigenvectors are outlined, and numerical experiments on educational interpersonal networks are presented to show that impact declines exponentially with distance on average and that the first-order approximation often serves as a reasonable proxy for the exact impact obtained from matrix inversion.

Significance. If the approximation is accurate with controllable error, the work supplies a simple, interpretable link between pairwise impact, eigenvector centrality, and the graph spectrum, enabling efficient estimation of influence decay without full linear-system solves. This is relevant for models such as Friedkin-Johnsen, Katz-Bonacich status, and network autocorrelation. The numerical verification on real social networks adds practical value, though the result's scope is tied to the tested network class.

major comments (1)
  1. [§3] §3 (Derivation of the first-order approximation): The truncation of the spectral expansion of the resolvent (I − αA)^−1 after the leading term is presented without an explicit remainder bound in terms of |λ2/λ1| and the diffusion parameter α. This is load-bearing for the central claim that the leading term provides a good proxy, because the error can become non-negligible when the spectral gap is small, as noted in the stress-test concern; the manuscript supplies no quantitative condition under which the truncation error remains small uniformly in distance d.
minor comments (2)
  1. [Abstract] The abstract states that the total impact 'is given to first order by a product of eigenvector centrality scores together with an expression in terms of the graph spectrum' but does not display the explicit leading-order formula; inserting the concrete expression would improve immediate clarity.
  2. [§2] Notation for the impact function and the precise definition of network distance (shortest-path versus walk-based) should be introduced consistently in §2 before the derivation begins.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their thorough review and for identifying a key point that can strengthen the rigor of our first-order approximation. We address the major comment below and will revise the manuscript to incorporate an explicit error bound.

read point-by-point responses

  1. Referee: [§3] §3 (Derivation of the first-order approximation): The truncation of the spectral expansion of the resolvent (I − αA)^−1 after the leading term is presented without an explicit remainder bound in terms of |λ2/λ1| and the diffusion parameter α. This is load-bearing for the central claim that the leading term provides a good proxy, because the error can become non-negligible when the spectral gap is small, as noted in the stress-test concern; the manuscript supplies no quantitative condition under which the truncation error remains small uniformly in distance d.

    Authors: We agree that an explicit remainder bound is desirable to make the conditions for the approximation fully quantitative. In the revision we will add to §3 a derivation of the truncation error for the resolvent expansion, expressed in terms of the spectral gap ratio |λ2/λ1| and α (under the standing assumption that the graph is connected and α is below the reciprocal of the largest eigenvalue). The bound will be shown to be independent of distance d when the leading eigenvector components are bounded away from zero, and we will discuss the regime in which the error remains small uniformly in d. This addition will also clarify the stress-test scenarios. revision: yes

Circularity Check

0 steps flagged

Spectral truncation approximation derived directly from adjacency matrix eigenvalues without circular reduction

full rationale

The paper's core claim is an approximate closed-form expression for total impact obtained by truncating the spectral expansion of the resolvent (I - αA)^{-1} after the dominant eigenvector term. This is a direct application of standard linear algebra to the forced linear diffusion model and does not reduce to a fitted parameter, self-definition, or self-citation chain. The numerical study on educational networks is described as independent verification of the exponential decay pattern rather than the source of the formula. No load-bearing steps invoke prior author work as an unverified uniqueness theorem or rename an empirical pattern as a derivation.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the spectral decomposition of the adjacency matrix for weighted walks and the premise that per-path impact decays exponentially with length; no free parameters or new entities appear in the abstract.

axioms (1)
  • domain assumption Impact of one individual on another through any given path falls exponentially with path length
    Explicitly identified in the abstract as a basic assumption of the forced linear diffusion models.

pith-pipeline@v0.9.0 · 9911 in / 1078 out tokens · 74349 ms · 2026-05-08T08:55:10.656708+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

35 extracted references

  1. [1]

    Anselin, L. (1988). Spatial Econometrics: Methods and Models . Kluwer, Norwell, MA

    1988

  2. [2]

    Ayala, G. X. and Team, S. D. P. R. C. (2011). Effects of a promotor-based intervention to promote physical activity: Familias sanas y activas. American Journal of Public Health , 101(12):2261--2268

    2011

  3. [3]

    Bonacich, P. (1972). Factoring and weighting approaches to clique identification. Journal of Mathematical Sociology , 2:113--120

    1972

  4. [4]

    Bonacich, P. (1987). Power and centrality: A family of measures. American Journal of Sociology , 92:1170--1182

    1987

  5. [5]

    and Page, L

    Brin, S. and Page, L. (1998). The anatomy of a large-scale hypertextual web search engine. Computer Networks and ISDN Systems , 30(1-7):107--117

    1998

  6. [6]

    Butts, C. T. (2008a). network: a package for managing relational data in R . Journal of Statistical Software , 24(2)

  7. [7]

    Butts, C. T. (2008b). Social network analysis with sna. Journal of Statistical Software , 24(6)

  8. [8]

    Butts, C. T. (2025). A perturbative solution to the linear influence/network autocorrelation model under network dynamics. Journal of Mathematical Sociology , 49(3):192--210

    2025

  9. [9]

    Cliff, A. D. and Ord, J. K. (1973). Spatial Autocorrelation . Pion, London

    1973

  10. [10]

    Doreian, P. (1989). Two regimes of network autocorrelation. In Kochen, M., editor, The Small World , pages 280--295. Ablex, Norwood

    1989

  11. [11]

    Eng, E. (1993). The save our sisters project. a social network strategy for reaching rural black women. Cancer , 72(S3):1071--1077

    1993

  12. [12]

    Freeman, L. C. (1979). Centrality in social networks: Conceptual clarification. Social Networks , 1(3):223--258

    1979

  13. [13]

    Freidkin, N. (1998). A Structural Theory of Social Influence . Cambridge University Press, Cambridge

    1998

  14. [14]

    Friedkin, N. (1991). Theoretical foundations for centrality measures. American Journal of Sociology , 96(6):1478--1504

    1991

  15. [15]

    and Cook, K

    Friedkin, N. and Cook, K. S. (1991). Peer group influence. Sociological Methods and Research , 19:122--143

    1991

  16. [16]

    and Johnsen, E

    Friedkin, N. and Johnsen, E. C. (1990). Social influence and opinions. Journal of Mathematical Sociology , 15:193--206

    1990

  17. [17]

    Friedkin, N. E. and Bullo, F. (2017). How truth wins in opinion dynamics along issue sequences. Proceedings of the National Academy of Sciences , 114(43):11380--11385

    2017

  18. [18]

    E., Proskurnikov, A

    Friedkin, N. E., Proskurnikov, A. V., Tempo, R., and Parsegov, S. E. (2016). Network science on belief system dynamics under logic constraints. Science , 354(6310):321--326

    2016

  19. [19]

    S., Hunter, D

    Handcock, M. S., Hunter, D. R., Butts, C. T., Goodreau, S. M., and Morris, M. (2008). statnet : Software tools for the representation, visualization, analysis and simulation of network data. Journal of Statistical Software , 24(1):1--11

    2008

  20. [20]

    Katz, L. (1953). A new status index derived from sociometric analysis. Psychometrika , 18(1):39--43

    1953

  21. [21]

    A., Peeters, L., Richter, S., Tenfelde-Podehl, D., and Zlotowski, O

    Kosch \" u tzki, D., Lehmann, K. A., Peeters, L., Richter, S., Tenfelde-Podehl, D., and Zlotowski, O. (2005). Centrality indices. In Brandes, U. and Erlebach, T., editors, Network Analysis: Methodological Foundations , chapter 3, pages 16--61. Springer-Verlag, Berlin

    2005

  22. [22]

    M., Pechmann, C., Wang, C., Pan, L., Delucchi, K., and Prochaska, J

    Lakon, C. M., Pechmann, C., Wang, C., Pan, L., Delucchi, K., and Prochaska, J. J. (2016). Mapping engagement in twitter-based support networks for adult smoking cessation. American journal of public health , 106(8):1374--1380

    2016

  23. [23]

    M., Wang, C., Hipp, J

    Lakon, C. M., Wang, C., Hipp, J. R., and Butts, C. T. (2025). Simulating social network-based interventions for adolescent cigarette smoking. Social Science & Medicine , 380:118196

    2025

  24. [24]

    Latan \' e , B. (1981). The psychology of social impact. American Psychologist , 36(4):343--356

    1981

  25. [25]

    Lee, N. H. (1969). The Search for an Abortionist . University of Chicago Press, Chicago, IL

    1969

  26. [26]

    buddy systems

    May, S. and West, R. (2000). Do social support interventions (“buddy systems”) aid smoking cessation? a review. Tobacco control , 9(4):415--422

    2000

  27. [27]

    Minkler, M. (1986). Building support networks from social isolation. Generations: Journal of the American Society on Aging , 10(4):46--49

    1986

  28. [28]

    Moody, J. (2001). Peer influence groups: Identifying dense clusters in large networks. Social Networks , 23:261--283

    2001

  29. [29]

    L., Williams, C

    Perry, C. L., Williams, C. L., Komro, K. A., Veblen-Mortenson, S., Stigler, M. H., Munson, K. A., Farbakhsh, K., Jones, R. M., and Forster, J. L. (2002). Project northland: Long-term outcomes of community action to reduce adolescent alcohol use. Health education research , 17(1):117--132

    2002

  30. [30]

    R: A Language and Environment for Statistical Computing

    R Core Team (2026). R: A Language and Environment for Statistical Computing . R Foundation for Statistical Computing, Vienna, Austria

    2026

  31. [31]

    Salancik, G. R. (1986). An index of subgroup influence in dependency networks. Administrative Science Quarterly , 31(2):194--211

    1986

  32. [32]

    Valente, T. W. (2012). Network interventions. science , 337(6090):49--53

    2012

  33. [33]

    T., Hipp, J

    Wang, C., Butts, C. T., Hipp, J. R., and Lakon, C. M. (2026). Peer influence decay and behavioral diffusion in adolescent networks: a simulation approach. Science , forthcoming

    2026

  34. [34]

    R., Butts, C

    Wang, C., Hipp, J. R., Butts, C. T., and Lakon, C. M. (2022). Insight into selecting adolescents for drinking intervention programs: a simulation based on stochastic actor--oriented models. Prevention Science , 23(1):48--58

    2022

  35. [35]

    and Faust, K

    Wasserman, S. and Faust, K. (1994). Social Network Analysis: Methods and Applications . Cambridge University Press, Cambridge

    1994