Around 1200 BCE the interconnected societies of the Late Bronze Age (LBA) Eastern Mediterranean — Mycenaean Greece, the Hittite Empire, Ugarit, Egypt, Cyprus, Canaan and others — collapsed or were transformed within a few decades, and no scholarly consensus on the causes exists. Linkov, Galaitsi, Trump, Pinigina, Rand, Cline and Kitsak (2024) proposed the first explicit connectivity-based model of this collapse: a two-layer interdependent network of ten societies whose stress tests show that no single society's failure could trigger a system-wide cascade, but that the paired failure of the Hittite Empire and Ugarit would disintegrate the entire system — a candidate mechanism by which inter-societal connectivity amplified compounding threats. The open problem is whether such connectivity-based systemic models can be established and validated: whether networks grounded in the archaeological and textual record, with justified propagation dynamics, can be shown — robustly to model-construction choices, and against the observed pattern of collapse, survival, and chronology — to explain the LBA collapse better than non-network alternatives, or instead be exposed as artifacts of arbitrary modeling assumptions.

Background

Between roughly 1700 and 1200 BCE, the Late Bronze Age (LBA) Eastern Mediterranean hosted a cluster of centralized societies — the Mycenaeans of mainland Greece, Minoan Crete, New Kingdom Egypt, the Hittite Empire of Anatolia, Babylonia, Assyria, Cyprus, Canaan, Western Anatolia, and the international port city of Ugarit — linked by an intensive system of trade, diplomacy, and treaties. Textual and archaeological evidence documents a minimum of 175 exchanges among 28 separate polities (Cline 2021, DOI 10.2307/j.ctv15r58dw). Shortly after 1200 BCE many of these societies were destroyed or abandoned, regional trade and literacy contracted, and the succeeding centuries (the 'Greek Dark Ages' or Early Iron Age) were far less connected. Proposed causes — drought and climate change (e.g., a severe multi-year drought coincident with the Hittite collapse around 1198–1196 BCE; Manning et al. 2023, DOI 10.1038/s41586-022-05693-y), earthquake clusters, famine, migration and 'Sea Peoples' raids, internal systems collapse — remain debated; even the near-synchronicity of the destruction horizon is contested (Middleton 2023, DOI 10.15184/aqy.2023.187). A recurrent diagnosis is that although single stressors have been studied in isolation, very few studies examine the dynamic relationships among the contributing parameters (Vidal-Cordasco & Nuevo-López 2021, DOI 10.1016/j.jaa.2020.101239), whose agent-based model — a simulation in which individual computational agents interact under explicit rules — explored socio-ecological interactions in the Aegean crisis.

Against this background, Linkov et al. (2024, DOI 10.1016/j.gloenvcha.2023.102792) introduced a network-science model of the collapse. A multilayer network is a graph whose nodes are connected by edges of several distinct types ('layers'); in an interdependent multilayer network, a node's functioning in one layer depends on its state in another, so that failures can propagate across layers and trigger cascading failure — the progressive collapse of nodes triggered by the failures of their neighbors, as studied for infrastructure since Buldyrev et al. (2010). Their model encodes ten LBA societies as nodes with two interdependent layers: a political layer (diplomatic ties, undirected, 21 links after filtering to the two strongest of four tie-strength levels) and a commercial layer (trade flows, directed, whose 54 strength-filtered directed links are consolidated into 27 undirected links for the analysis, since goods almost always flowed both ways), constructed from the evidence assembled in Cline's 1177 B.C. A node is stipulated stable in the political layer if it retains at least two links to stable nodes, stable in the commercial layer if it retains at least one third of its commercial links, and unstable in one layer if unstable in the other. Exhaustive stress tests — removing every combination of 1, 2, 3, and 4 nodes and iterating to a fixed point — showed the network robust to single-node failures (eight of ten single removals leave it fully intact; removing Mainland Greece or the Hittites costs exactly one further node, about 11% damage), while three paired failures (Hittites+Egypt, Hittites+Ugarit, Egypt+Ugarit) destabilize most or all of the system. Dismissing Hittites+Egypt because Egypt demonstrably survived, the authors propose the simultaneous failure of the Hittite Empire and Ugarit as a historically plausible trigger whose cascade (Western Anatolia, Cyprus, Canaan, then Mainland Greece and Crete, then Egypt, finally Assyria and Babylonia) is qualitatively consistent with who collapsed and who endured. A companion perspective by substantially the same team (Galaitsi, Trump, Cline, Kitsak & Linkov 2025, DOI 10.1111/risa.70019) reiterates the hypothesis that LBA network structure amplified compounding threats into a precipitous systemic collapse.

The model's own text concedes its provisional character: the stability thresholds were chosen experimentally ('the only non-trivial choice is two links per node'), goods-based commercial failure rules were impossible for lack of data, the network is built from an incomplete historical record, and the archaeological order in which societies failed is unknown — so corroboration is qualitative consistency, not a discriminating test. In parallel, process-based modeling of societal collapse in general remains stylized: Schunck, Wiedermann, Heitzig & Donges (2024, DOI 10.3390/e26020098), building on Tainter's theory that societies collapse when the marginal returns of complexity decline, developed a dynamic network model of societal complexity whose authors state that process-based understanding of collapse dynamics is 'still in its infancy' and that transferring stylized findings to empirical contexts remains unresolved. What no study has yet done is establish a connectivity-based model of the LBA collapse and validate it against the archaeological record — or show that the network-cascade hypothesis cannot survive such validation.

Problem Statement

Can connectivity-based systemic models of the Late Bronze Age collapse be established and validated against the archaeological record? Concretely: build one or more interdependent-network models of the LBA Eastern Mediterranean system — nodes the LBA societies, links grounded with explicit evidentiary basis in the archaeological and textual record (political/diplomatic ties, commercial flows) — with node-failure and propagation dynamics that are justified rather than experimentally stipulated; and determine whether the network-cascade mechanism these models embody explains the collapse better than alternatives. Validation would require showing that model-generated collapse scenarios are consistent with the observed record — which societies collapsed versus survived, the approximate destruction/abandonment chronology, and the spatial pattern of the collapse horizon — and that this consistency is discriminating (i.e., superior to appropriate null models, such as randomized or non-interdependent networks, and to at least one plausible non-network single-stressor explanation) and robust to model-construction choices (stability thresholds, link-strength cutoffs, node granularity, weighting of evidence). Refutation would require demonstrating either that the cascade conclusion is an artifact of arbitrary modeling assumptions — e.g., it disappears under defensible alternative constructions, or arises equally for networks that do not resemble the LBA system — or that a required precondition or predicted pattern contradicts the dated archaeological record.

References

  1. Linkov, I., Galaitsi, S.E., Trump, B.D., Pinigina, E., Rand, K., Cline, E.H., Kitsak, M. (2024). 'Are civilizations destined to collapse? Lessons from the Mediterranean Bronze Age.' Global Environmental Change 84:102792. DOI 10.1016/j.gloenvcha.2023.102792. https://doi.org/10.1016/j.gloenvcha.2023.102792
  2. Galaitsi, S.E., Trump, B.D., Cline, E.H., Kitsak, M., Linkov, I. (2025). 'Navigating the precipice: Lessons on collapse from the Late Bronze Age.' Risk Analysis 45(8):2079-2082. DOI 10.1111/risa.70019. https://doi.org/10.1111/risa.70019
  3. Vidal-Cordasco, M., Nuevo-López, A. (2021). 'Resilience and vulnerability to climate change in the Greek Dark Ages.' Journal of Anthropological Archaeology 61:101239. DOI 10.1016/j.jaa.2020.101239. https://doi.org/10.1016/j.jaa.2020.101239
  4. Middleton, G.D. (2023). 'Getting closer to the Late Bronze Age collapse in the Aegean and eastern Mediterranean, c. 1200 BC.' Antiquity 98(397):260-263. DOI 10.15184/aqy.2023.187. https://doi.org/10.15184/aqy.2023.187
  5. Schunck, F., Wiedermann, M., Heitzig, J., Donges, J.F. (2024). 'A Dynamic Network Model of Societal Complexity and Resilience inspired by Tainter's Theory of Collapse.' Entropy 26(2):98. DOI 10.3390/e26020098. https://doi.org/10.3390/e26020098
  6. Merelo-Guervós, J.J. (2026). 'Analyzing Late Antiquity Shifts of Trade Regime in the Iberian Peninsula and Their Causes via Change Point Detection Methods.' Complexities 2(2):12. DOI 10.3390/complexities2020012. https://doi.org/10.3390/complexities2020012
  7. Jehn, F.U., Gajewski, Ł.G., Hedlund, J., Arnscheidt, C.W., Xia, L., Wunderling, N., Denkenberger, D. (2025). 'Food trade disruption after global catastrophes.' Earth System Dynamics 16(5):1585-1603. DOI 10.5194/esd-16-1585-2025. https://doi.org/10.5194/esd-16-1585-2025
  8. Loose, D.C., Marcellin, M.C., Linkov, I., Pavur, G., Kitsak, M., Deegan, M.A., Lambert, J.H. (2025). 'Infrastructure Network Resilience Analysis with Disruptions of System Order.' 2025 11th International Conference on Control, Decision and Information Technologies (CoDIT), pp. 349-354. DOI 10.1109/CoDIT66093.2025.11321594. https://doi.org/10.1109/CoDIT66093.2025.11321594
  9. Morgan, A.W., Paine, A.R., Koç, K., Haldon, J., Hofmeister, E., Cheng, H., Tüysüz, O., Matter, A., Edwards, L.R., Haghipour, N., Hajdas, I., Fleitmann, D. (2025). 'Societal responses to cold-season rainfall variability: a speleothem perspective on Byzantine and Hittite climate interactions in Late Holocene Türkiye and southeast Europe.' Quaternary Science Reviews 359:109365. DOI 10.1016/j.quascirev.2025.109365. https://doi.org/10.1016/j.quascirev.2025.109365
  10. Manning, S.W., Kocik, C., Lorentzen, B., Sparks, J.P. (2023). 'Severe multi-year drought coincident with Hittite collapse around 1198-1196 BC.' Nature 614:719-724. DOI 10.1038/s41586-022-05693-y. https://doi.org/10.1038/s41586-022-05693-y
  11. Cline, E.H. (2021). '1177 B.C.: The Year Civilization Collapsed, Revised Edition.' Princeton University Press. DOI 10.2307/j.ctv15r58dw. https://doi.org/10.2307/j.ctv15r58dw