great 🧵 discussing a network topology-aware simulation strategy for modeling epidemics, proposing a explanation for the unusual periodicity we've observed
while not quite agent-based, this simulation strategy accounts salient nuances of real-world behavior other approaches dont
---
RT @phl43
Many have noted the cyclical nature of the pandemic, with waves that come and go seemingly at random, but the reasons for this remain mysterious. In th…
https://twitter.com/phl43/status/1462806796020686850
important nuance required regarding the limits of this approach and the many assumptions made; it's more useful as a counterargument for assumptions made in many current models, and direction for future exploration
he's not the first to make these points but this is well written
agent simulations weren't too popular historically for a variety of reasons (computational challenges, claims of overengineering, etc) but many real-world systems are sufficiently complex that assumptions of global homogeneity simply don't hold
v glad to see attitudes changing
generally I'm majorly in favor of agent-based models for simulating any domain where network topology is salient (ie where internode interactions are heterogenous and cliquey)
even probabilistic variants tend to better model real-world dynamics than purely analytic approaches