pith. sign in

arxiv: 2604.20872 · v1 · submitted 2026-03-27 · ⚛️ physics.soc-ph · cs.CY

Dynamical Model for the Sustainable Development Goals

Alberto Garc\'ia-Rodr\'iguez , Tzipe Govezensky , Julia Tag\"ue\~na , Kimmo K. Kaski , Rafael A. Barrio This is my paper

Pith reviewed 2026-05-14 22:44 UTC · model grok-4.3

classification ⚛️ physics.soc-ph cs.CY
keywords Sustainable Development Goalsdynamical modelmathematical modelinginternational cooperationresource allocationgoal correlationsUN 2030 Agenda
0
0 comments X

The pith

A dynamical model driven by resource distribution, country cooperation, and goal correlations reproduces real-world SDG progress data.

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

The paper builds a mathematical model for how countries advance toward the 17 UN Sustainable Development Goals over time. It draws on three factors extracted from earlier unsupervised analysis of two decades of data: the way each country allocates its resources across goals, the level of cooperation between countries, and the statistical correlations that link different goals. When run forward, the model closely tracks the observed historical trajectories of SDG indicators. This match indicates the model can generate counterfactual scenarios that test how changes in resource use or cooperation might alter future outcomes.

Core claim

The authors show that a system of coupled dynamical equations, parameterized by country-specific resource vectors, inter-country cooperation strengths, and a goal-correlation matrix, reproduces the measured time series of SDG indicators from 2000 to 2022 across nations.

What carries the argument

The coupled dynamical equations that evolve each country's vector of goal indicators under the combined influence of resource allocation, cooperation terms, and inter-goal correlations.

If this is right

  • Changes in how countries allocate resources can be simulated to forecast effects on multiple goals at once.
  • Varying the strength of cooperation between countries produces testable predictions for accelerated or stalled progress.
  • Scenarios that strengthen or weaken specific goal correlations can be explored to identify high-leverage policy targets.
  • The model supplies a quantitative tool for ranking hypothetical actions by their projected contribution to 2030 targets.

Where Pith is reading between the lines

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

  • If the three-factor structure holds, global agreements that raise cooperation on correlated goals would produce compounding gains across several targets simultaneously.
  • The same equations could be used to test resilience: how much extra cooperation would be needed to offset a sudden drop in resources in one region.
  • Extending the model with explicit time delays in cooperation effects might reveal windows during which early joint action yields outsized later returns.

Load-bearing premise

That resource distribution, cooperation levels, and goal correlations are enough to generate the full observed dynamics without other influences such as sudden policy shifts or external crises.

What would settle it

Running the model from 2023 onward and finding that its predicted indicator values diverge measurably from newly released official SDG statistics for multiple countries and goals.

Figures

Figures reproduced from arXiv: 2604.20872 by Alberto Garc\'ia-Rodr\'iguez, Julia Tag\"ue\~na, Kimmo K. Kaski, Rafael A. Barrio, Tzipe Govezensky.

Figure 1
Figure 1. Figure 1: (A) calculation when the neighbor matrix is complete, (B) when neighbor [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: (A) Map showing the grouping of countries in six clusters. (B) Real data shown [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: (A) Results using real data for the clusters shown in Fig. 2 (A) and (B) [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: (A) Comparison between the real data (color stars) and the results of the [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: (A) Calculation made with gk = 10.5, real data (color stars) and the results of the calculation (color lines). (B) Dynamical behaviour of the 17 goals. Dynamical behaviour of the 17 SDG goals. Observe the remarkable behavior of goals 12 and 13, which now are progressing in spite of the anti correlations [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: (A)trajectories of clusters when gi = 0, meaning that there is no correlation between goals, real data (color stars) and the results of the calculation (color lines). (B) Dynamical behaviour of the 17 SDG goals when they are totally uncorrelated. clear that positive correlation between SDGs would eventually lead to success. 5.3 Effect of economic resources It is challenging to advice how a given country sh… view at source ↗
Figure 7
Figure 7. Figure 7: (A) Trajectories of clusters when gi = 1, and all correlations are positive, real data (color stars) and the results of the calculation (color lines). (B) Dynamical behaviour of the 17 SDG goals when all correlations are positive [PITH_FULL_IMAGE:figures/full_fig_p009_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: (A) Trajectories of clusters when the resources are the same for all clusters [PITH_FULL_IMAGE:figures/full_fig_p009_8.png] view at source ↗
read the original abstract

The 2030 Agenda for Sustainable Development of the United Nations outlines 17 goals as global challenges for countries of the world to address in their development. However, the progress of countries towards these goals has been much slower than expected. In a previous study, we analyzed the data over two decades (2000--2022), using unsupervised machine learning techniques. Based on this study, we take into account three main factors to construct a mathematical model to simulate and predict the dynamical behavior of the SDGs. These factors are: (1) the distribution of amount of resources that each country uses to meet the goals, (2) the cooperation between countries, and (3) the correlations between the goals. In this work, we show that the model is capable of reproducing the real data and therefore could be used to simulate hypothetical scenarios that could help to improve actions towards optimal fulfillment of the goals.

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

3 major / 1 minor

Summary. The manuscript constructs a dynamical model for progress on the UN's 17 Sustainable Development Goals using three factors previously identified via unsupervised machine learning on 2000-2022 data: per-country resource distribution, inter-country cooperation, and goal correlations. It claims this model reproduces the observed historical trajectories and can therefore be used to simulate hypothetical policy scenarios for improved SDG outcomes.

Significance. A validated dynamical model linking resource allocation, cooperation, and goal interdependencies could provide a useful tool for scenario analysis in sustainability policy. However, the current version supplies no equations, fitting details, or quantitative reproduction metrics, so the claimed utility cannot yet be assessed.

major comments (3)
  1. [Abstract] Abstract: the central claim that the model 'reproduces the real data' is unsupported; no dynamical equations, loss function, parameter values, RMSE/MAPE values, or out-of-sample tests on the 17 indicators are provided anywhere in the manuscript.
  2. [Model construction] Model construction section: the three factors are extracted from the identical 2000-2022 dataset that the model is then shown to reproduce, creating a circularity risk; without held-out years, independent benchmarks, or parameter-free predictions, success may be an artifact of post-hoc tuning rather than genuine dynamical capture.
  3. [Results] Results section: with only three classes of free parameters for 17 coupled time series, the system is under-determined; the manuscript must report whether parameters were fit to the full interval or to a training subset and must supply explicit error metrics and sensitivity analysis.
minor comments (1)
  1. [Abstract] The abstract would be clearer if it briefly stated the mathematical form of the dynamical system (e.g., differential equations or discrete update rules) and the number of free parameters.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the detailed and constructive report. The comments highlight important gaps in quantitative detail and validation that we will address in a revised manuscript. We respond point-by-point below.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central claim that the model 'reproduces the real data' is unsupported; no dynamical equations, loss function, parameter values, RMSE/MAPE values, or out-of-sample tests on the 17 indicators are provided anywhere in the manuscript.

    Authors: We agree that the abstract's claim requires explicit quantitative support. The revised manuscript will include the full set of dynamical equations, the loss function used for fitting, the fitted parameter values, RMSE and MAPE metrics for the 17 indicators, and results from out-of-sample tests on held-out years. revision: yes

  2. Referee: [Model construction] Model construction section: the three factors are extracted from the identical 2000-2022 dataset that the model is then shown to reproduce, creating a circularity risk; without held-out years, independent benchmarks, or parameter-free predictions, success may be an artifact of post-hoc tuning rather than genuine dynamical capture.

    Authors: The three factors originate from our prior unsupervised machine-learning study on the same data period (explicitly cited). The dynamical model then evolves the indicators forward using these fixed factors as inputs. To address the circularity concern, the revision will add a cross-validation procedure using held-out years and a sensitivity analysis on parameter perturbations. revision: partial

  3. Referee: [Results] Results section: with only three classes of free parameters for 17 coupled time series, the system is under-determined; the manuscript must report whether parameters were fit to the full interval or to a training subset and must supply explicit error metrics and sensitivity analysis.

    Authors: We will clarify that the three parameter classes were fitted to the entire 2000–2022 interval. The revision will report the explicit error metrics (RMSE, MAPE) for each indicator and include a systematic sensitivity analysis showing how variations in the three factor classes affect the reproduced trajectories. revision: yes

Circularity Check

1 steps flagged

Reproduction of SDG trajectories uses factors extracted from the same 2000-2022 dataset via prior unsupervised ML, reducing claimed dynamical capture to a fit

specific steps
  1. fitted input called prediction [Abstract]
    "In a previous study, we analyzed the data over two decades (2000--2022), using unsupervised machine learning techniques. Based on this study, we take into account three main factors to construct a mathematical model to simulate and predict the dynamical behavior of the SDGs. ... In this work, we show that the model is capable of reproducing the real data"

    Factors are extracted from the 2000-2022 dataset by unsupervised ML; the dynamical model is then constructed from those factors and asserted to reproduce the same dataset. The reproduction success is therefore defined by the identical data used to select the factors, with no reported independent test set or external validation.

full rationale

The paper's central claim is that a three-factor dynamical model (resource distribution, inter-country cooperation, goal correlations) reproduces observed SDG progress. These factors originate from the authors' prior unsupervised ML analysis of the identical historical data interval. The model is then shown to reproduce that data. No out-of-sample validation, independent benchmarks, or parameter-free predictions are described in the provided text; the reproduction therefore reduces to a fitted reconstruction whose inputs were chosen from the target dataset. This matches the fitted-input-called-prediction pattern but is only partial circularity because the dynamical equations themselves are not shown to be mathematically identical to the ML step. Self-citation of the prior study is load-bearing for factor selection but does not collapse the entire derivation by definition.

Axiom & Free-Parameter Ledger

3 free parameters · 1 axioms · 0 invented entities

The model rests on the assumption that three high-level factors extracted from prior data analysis are dynamically sufficient; no independent evidence for this sufficiency is supplied in the abstract.

free parameters (3)
  • resource allocation weights
    Country-specific distributions of effort across the 17 goals, required to initialize and drive the dynamics.
  • cooperation strength parameters
    Coefficients governing interaction terms between countries, fitted or chosen to match observed progress.
  • correlation matrix entries
    Pairwise goal interaction strengths derived from or fitted to the same historical dataset.
axioms (1)
  • domain assumption The three factors (resource distribution, cooperation, goal correlations) are sufficient to reproduce observed SDG dynamics.
    Invoked in the abstract as the basis for constructing the model from the prior ML study.

pith-pipeline@v0.9.0 · 5475 in / 1374 out tokens · 37443 ms · 2026-05-14T22:44:54.438995+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

12 extracted references · 12 canonical work pages

  1. [1]

    THE 17 GOALS — Sustainable Development — sdgs.un.org

    Nations U. THE 17 GOALS — Sustainable Development — sdgs.un.org;. https://sdgs.un.org/goals

    work page

  2. [2]

    Sustainable visions: unsupervised machine learning insights on global development goals

    Garc´ ıa-Rodr´ ıguez A, N´ u˜ nez M, P´ erez MR, Govezensky T, Barrio RA, Gershenson C, et al. Sustainable visions: unsupervised machine learning insights on global development goals. PLOS ONE. 2025;20(3):1–20. doi:10.1371/journal.pone.0317412

    work page doi:10.1371/journal.pone.0317412 2025

  3. [3]

    Global Sustainability Performance and Regional Disparities: A Machine Learning Approach Based on the 2025 SDG Index

    C ¸ elik S,¨Ozt¨ urk OF, Akkucuk U, S ¸a¸ smaz M¨U. Global Sustainability Performance and Regional Disparities: A Machine Learning Approach Based on the 2025 SDG Index. Sustainability. 2025;17(16). doi:10.3390/su17167411

    work page doi:10.3390/su17167411 2025

  4. [4]

    The Agenda 2030 Compass Summary

    Hallding K, Suljada T, Axelsson K, ˚Asa Gerger Swartling, Kemp-Benedict E, Laubacher R, et al. The Agenda 2030 Compass Summary. Stockholm Environment Institute (SEI) and Jernkontoret; 2022

    work page 2030

  5. [5]

    Available from: https://sdgintegration.undp.org/isdg-model

    iSDG Model; 2019. Available from: https://sdgintegration.undp.org/isdg-model

    work page 2019

  6. [6]

    System Dynamics Focuses More on Sustainability Than the Sustainable Development Goals; 2022

    Redivo F. System Dynamics Focuses More on Sustainability Than the Sustainable Development Goals; 2022. Available from:https://systemdynamics.org/ system-dynamics-focuses-more-on-sustainability-than-the-sustainable-development-goals/

    work page 2022

  7. [7]

    Progress in modeling dynamic systems for sustainable development

    Selin NE, Giang A, Clark WC. Progress in modeling dynamic systems for sustainable development. Proceedings of the National Academy of Sciences of the United States of America. 2023;120. doi:10.1073/pnas.2216656120

    work page doi:10.1073/pnas.2216656120 2023

  8. [8]

    An index to measure the sustainability of place-based development pathways

    Topf J, Schultz LA, da Silva JMC. An index to measure the sustainability of place-based development pathways. Ecological Economics. 2023;204:107645. doi:https://doi.org/10.1016/j.ecolecon.2022.107645

    work page doi:10.1016/j.ecolecon.2022.107645 2023

  9. [9]

    Forecasting Future Development under the Interactions among Sustainable Development Goals

    Xu Y, Chen Y, Cao M, Chang L, Bai Y, Li Y, et al. Forecasting Future Development under the Interactions among Sustainable Development Goals. Sustainability. 2023;15(22). doi:10.3390/su152215929

    work page doi:10.3390/su152215929 2023

  10. [10]

    National pathways to the Sustainable Development Goals (SDGs): A comparative review of scenario modelling tools

    Allen C, Metternicht G, Wiedmann T. National pathways to the Sustainable Development Goals (SDGs): A comparative review of scenario modelling tools. Environmental Science & Policy. 2016;66:199–207. doi:https://doi.org/10.1016/j.envsci.2016.09.008

    work page doi:10.1016/j.envsci.2016.09.008 2016

  11. [11]

    Sustainable Development Report Dashboards 2024; 2024

    Index S. Sustainable Development Report Dashboards 2024; 2024. Available from:https://dashboards.sdgindex.org/

    work page 2024

  12. [12]

    Available from: https://www.worldometers.info/gdp/gdp-by-country/

    worlddmeter; 2019. Available from: https://www.worldometers.info/gdp/gdp-by-country/. 12

    work page 2019