The Free-Market Algorithm: Self-Organizing Optimization for Open-Ended Complex Systems

Researcher Martin Jaraiz has published a groundbreaking paper introducing the Free-Market Algorithm (FMA), a new metaheuristic optimization technique fundamentally different from established methods like Genetic Algorithms. Instead of a central controller or a pre-defined fitness function, FMA creates a simulated economy where autonomous agents act as 'firms.' These agents discover rules, trade goods, and compete for demand, with successful solutions emerging organically from the distributed market dynamics. The algorithm's core three-layer architecture keeps the universal market mechanism constant, requiring only pluggable, domain-specific behavioral rules to adapt to new problems.

In validation tests, FMA demonstrated remarkable versatility and power. In a prebiotic chemistry simulation, starting from just 900 basic atoms (C, H, O, N), the algorithm autonomously discovered all 12 feasible amino acid formulas, all 5 nucleobases, and key biochemical pathways like the Krebs cycle—all in under 5 minutes on a standard laptop. It generated up to 240 independent synthesis routes per product. In a completely different domain, macroeconomic forecasting, FMA achieved a Mean Absolute Error of 0.42 percentage points for GDP prediction by reading a single input-output table with zero estimated parameters, a performance comparable to professional forecasters and portable across 33 countries.

The paper argues that FMA provides the first explicit, tunable computational mechanism for the 'selection signatures' described by Assembly Theory, linking it to fundamental physics concepts like causal set theory. This suggests the algorithm's event-driven, relational dynamics may reflect a deeper organizational principle in nature itself, positioning Darwinian market dynamics as a potential universal engine for discovery in open-ended complex systems where the solution space is not known in advance.