Hi Polkadot Forum community! ![]()
I recently posted a formal extension of my original two-good convex exchange model into a full n-good reaction–diffusion economy that preserves global consistency, convexity, and closed-form equilibrium properties while allowing rich heterogeneous agent preferences.
Link to the full write-up:
Reaction–Diffusion Exchange Economy (LaTeX/PDF)
With the extension to the production and exchange phases treated as Reaction/Diffusion concepts taken from the original Turing 1952 paper, I think the model is now mature enough to serve as a solid economic analysis foundation for agent-based economic simulations in Substrate/Polkadot-native environments (coretime markets, agile coretime, task allocation, DePIN coordination, etc.) because it introduces technologies treated as zero-sum transformations and distributively executed by agents as part of their “production” (reaction) phase, for then being able to trade this new goods obtained by means of production.
What’s particularly exciting — and still wide open — is endogenous evaluation of the preference parameters α_{ij} via equilibrium analysis.
The paper describes the α_{ij} (pairwise Cobb–Douglas exponents) are taken as exogenous primitives.
The big unexplored opportunity is to reverse-engineer them from observed equilibrium allocations and prices, effectively letting the system infer distributed preferences consistent with what actually happens on-chain.
This “α-evaluation from equilibrium” direction would:
- Turn static preference models into adaptive, learning-aware ones
- Enable collusion/resilience analysis in coretime-style markets
- Provide a principled way to bootstrap heterogeneous agent economies from real data
- Open a path toward fully decentralized preference discovery and economic morphogenesis
This feels like one of the most promising threads for turning the reaction–diffusion framework into a real economic primitive, using my already referenced hack for the two-goods economy, extended to n being possible because the pairwise directions follow the same principles described in the “Emergent properties…” paper. Thus, this was a natural step-forward toward to extend the original Cobb-Douglas two-good solution to a more complex economy explicitly.
Request for Comments
I’m actively looking for collaborators who are excited to dive into quantitative analysis’ α-evaluation stemming from equilibrium analysis and related directions.
Specifically interested in people with backgrounds in:
- Mathematical economics & general equilibrium theory
- Convex optimization / inference over equilibrium constraints
- Rust implementation of interior-point solvers + agent simulation
- DePIN, energy-based tokenomics, or multi-chain agent coordination
If you find the idea of inferring distributed preferences from equilibrium data compelling, let’s start a focused working group right here in this thread.
Long-term vision: a Polkadot-native library for reaction–diffusion economic simulations + preference inference that can be used for coretime pricing experiments, parachain resource markets, or even cross-chain task allocation.
Resources
- Full n-good extension: Reaction–Diffusion Economy LaTeX
- Original 2-good paper repo: Emergent Properties of Distributed Agents…
- Turing’s Morphogenesis paper (1952): [https://www.dna.caltech.edu/courses/cs191/paperscs191/turing.pdf]
Reply below if you’re interested in joining, have ideas on the α-evaluation problem, or just want to discuss whether this could be useful for future Polkadot economic layers and/or products.
Looking forward to collaborate with you! ![]()
— Diego Correa Tristain
PBA Berkeley Cohort | OnEdge Network | Ecosystemic Architect