📄 evolutionary biology

The Darwinian evolution of political preferences and the collective strategies they support

This paper presents a model demonstrating that while the Darwinian evolution of individual preferences can drive social groups toward more efficient collective strategies, the success and speed of this improvement critically depend on how individual preferences are aggregated into group decisions, which can otherwise lead to stagnation or maladaptive outcomes.

Original authors: Perret, C.

Published 2026-08-05

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Perret, C.

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

Imagine a world where the rules of a game aren't written by a single boss, but are decided by the whole team. In biology, we know that animals evolve traits like sharp claws or fast legs because those who have them survive better and pass them on. This is natural selection. But what about the "rules" of human society? How do we end up with good tax laws, fair sharing agreements, or effective team tactics? Scientists have long wondered if these social rules evolve in a similar way. The idea is that if a group does well, people copy the preferences of the successful members. If you see a neighbor thriving, you might adopt their favorite way of doing things, hoping to get the same results. This paper asks a crucial question: Does this copying process actually make our society's rules better over time, or can it get stuck, or even make things worse? It turns out, the answer depends entirely on how the group decides which rule to pick.

The paper, titled "The Darwinian evolution of political preferences and the collective strategies they support," by Cedric Perret, dives into this mystery using a mathematical model. Think of it as a giant, digital simulation of a society made up of many small groups. In this world, every person has a "political preference"—a favorite setting for a group rule, like "we should share 30% of our food" or "we should build a wall." The group then combines all these individual preferences to decide on one single rule to follow. Once the rule is set, the group plays a game (like hunting or trading) and gets a "payoff" (success or failure). The key twist is that people who do well are more likely to have their preferences copied by others in the next generation.

The researcher found that, yes, this process can lead to better rules, but only under specific conditions. The speed and success of this improvement depend on a concept they call "marginal influence." Imagine a group of 100 people trying to decide on a rule. If the group uses an "average" rule (where everyone's opinion counts equally), a single person's preference only changes the final decision by a tiny fraction (1/100th). In this case, the "signal" of what works is so weak that the group might take forever to figure out the best rule, or never find it at all. It's like trying to steer a massive cruise ship with a tiny toy steering wheel; the ship barely moves.

However, the paper shows that if the group uses a different method, like picking a "dictator" (one person decides) or a "novelty" rule (the most unique idea wins), the outcome changes dramatically. In these scenarios, a single person's preference has a much stronger impact on the final decision. The simulations show that groups using these methods can rapidly evolve toward the most efficient rules, just like a species evolving a perfect camouflage. But here is the catch: if the group is large and uses an average-based system, the improvement can be so slow that the group remains stuck with bad rules for thousands of generations, even if a better rule exists.

The author also tested what happens when things get messy. They simulated situations where the "best" rule isn't obvious, or where people have a built-in bias against good ideas (like hating taxes even when they help). In these complex scenarios, the way the group decides matters even more. Large groups with weak individual influence often get trapped in "local optima"—they find a rule that is "okay" but not the best possible one, and they can't climb out of that rut. Meanwhile, groups with stronger individual influence can jump over these hurdles and find the truly best solution.

Crucially, the paper argues against the idea that this evolutionary process is a guaranteed path to perfection. It explicitly rules out the notion that copying successful people will always lead to better societies. The study demonstrates that without the right decision-making structure, the link between success and copying gets broken. A person might be successful because their group adopted a great rule by luck, but if their own personal preference for that rule has almost no power to change the group's decision, their success won't spread that preference to others.

In the end, this research suggests that the "software" of our society—how we vote, debate, and decide—is just as important as the "hardware" of our rules. If we want our collective strategies to evolve and improve, we need decision-making systems where individual voices actually move the needle. Whether we are talking about ancient tribes, modern governments, or even animal groups, the ability to adapt depends on how much influence a single individual has on the group's final choice. The paper doesn't claim to have solved the problem of how to build the perfect society, but it provides a clear map of why some societies get stuck while others keep getting smarter.

Technical Summary: The Darwinian Evolution of Political Preferences and Collective Strategies

Problem Statement
Social groups rely on collective strategies (institutions, rules, and decision-making protocols) to govern member behavior and determine group success. While it is widely hypothesized that these "social technologies" evolve through a Darwinian process—where preferences associated with successful individuals are more likely to be copied and spread—the conditions under which this process reliably generates cumulative improvement remain unclear. Previous models have often assumed specific decision mechanisms or treated collective strategies as direct traits, neglecting the critical intermediate step: how individual political preferences are aggregated into a group strategy. The central question addressed is whether the spread of political preferences from successful individuals generally leads to the evolution of more efficient collective strategies, and if not, under what specific conditions this process fails or stalls.

Methodology
The author develops a general evolutionary model structured into $M$ groups of size $N$ . The model operates on a discrete-time life cycle involving three stages:

Collective Decision-Making: A collective strategy $h_j(t)$ is determined by a Social Choice Function (SCF), $H_j$ , which maps the vector of individual political preferences $\vec{x}_j(t)$ within a group to a single strategy value. The SCF abstracts the decision procedure (e.g., voting, deliberation) into a mathematical mapping.
Payoff Generation: Individuals participate in an economic game where payoffs $\pi_{ij}(t)$ are determined by the collective strategy. The model assumes a general payoff function $\Pi(h)$ where certain strategy values yield higher returns, without necessarily specifying the underlying game mechanics in the primary analysis.
Social Learning: Political preferences are transmitted across the population. In each time step, the population of preferences is resampled with probabilities proportional to individual payoffs (a Wright-Fisher process). This represents a bias where successful individuals' preferences are more likely to be copied. Random mutations introduce variation, modeled as Gaussian noise.

The author employs two analytical approaches:

Analytical Approximation: Using a small-variance approximation, they derive the selection gradient $s(\theta)$ , which describes the direction and rate of evolutionary change in the average preference. This allows for the derivation of general expressions for the "marginal influence" of an individual's preference on the group outcome.
Individual-Based Simulations: To test analytical predictions in finite populations, the author runs stochastic simulations varying group sizes, Social Choice Functions, and payoff landscapes (including single-peaked, double-peaked, and biased transmission scenarios).

Key Contributions
The paper introduces a general framework that decouples the evolution of political preferences from the specific mechanics of collective decision-making, allowing for a systematic analysis of how different aggregation rules (SCFs) affect evolutionary outcomes. The core theoretical contribution is the identification of marginal influence ( $\frac{\partial h_\tau}{\partial \tau}$ ) as the critical factor modulating the strength of selection on preferences. This term quantifies how much a marginal change in an individual's preference shifts the implemented collective strategy.

Key Results

Selection Can Improve Strategies, but Rates Vary Dramatically: The model demonstrates that selection on political preferences can drive groups toward locally optimal strategies across a wide range of conditions. However, the rate of improvement is not uniform; it is directly proportional to the marginal influence of individuals.
- Average Rule: Under an averaging SCF, marginal influence scales as $1/N$ . In large groups, selection pressure becomes extremely weak, leading to extremely slow convergence. Populations may remain far from the optimum even over long evolutionary timescales.
- Median Rule: For groups of size $N \geq 3$ , the marginal influence of a single individual is zero. Consequently, selection cannot drive systematic improvement; the strategy remains static or drifts randomly.
- Novelty and Dictator Rules: Under a "novelty" rule (selecting the most divergent preference) or a "random dictator" rule, marginal influence is high (equal to 1 or effectively 1/N in expectation but with high variance). These mechanisms allow for rapid convergence to optimal strategies, comparable to the baseline case where strategies evolve directly.
Failure Modes in Complex Landscapes:
- Local Optima: In payoff landscapes with multiple peaks (e.g., low-tax vs. high-tax regimes), the speed of convergence matters. Slow-moving populations (e.g., under average rules in large groups) are more likely to become trapped in local optima, failing to reach the global optimum.
- Maladaptive Bias: When transmission is biased toward maladaptive values (e.g., an inherent bias against taxation), the outcome depends on group structure. In standard cultural evolution models ( $N=1$ ), large populations can overcome bias. However, in collective strategy models ( $N>1$ ), increasing population size by increasing group size weakens selection, causing biased modifications to accumulate and preventing the group from reaching the optimal strategy.
Long-Run Equilibria: Despite differences in rates, populations using SCFs with positive marginal influence (Average, Random Dictator, Novelty) eventually converge to the same locally optimal strategy. The median rule, while static under small-variance assumptions, behaves like an averaging rule under larger variations and also converges to the same equilibrium, albeit slowly.

Significance and Claims
The paper claims to provide a "qualified answer" to the question of whether social technologies evolve analogously to material technologies. The author argues that while the mechanism of copying successful preferences can generate efficient rules without individual foresight, its efficacy is not guaranteed.

The primary insight is that the structure of collective decision-making acts as a filter on evolutionary forces. The link between individual success and the transmission of preferences is weakened when individual influence is diluted (e.g., in large democratic groups). This suggests that:

Evolutionary reasoning cannot be uncritically applied to collective strategies; the specific aggregation mechanism determines whether selection leads to improvement, stagnation, or maladaptation.
Groups may partially determine their own evolutionary trajectory by the rules they adopt for decision-making.
The findings extend beyond humans to any animal group making collective decisions via preference aggregation, suggesting that species with different decision-making structures may differ in their capacity to evolve efficient collective behaviors.

The author concludes that while previous models established that improvement is possible in specific contexts, this work clarifies that the reliability and speed of such improvement are contingent on the specific social choice functions employed, highlighting a need for broader research into how political organization shapes selection on collective traits.

More like this