📚 Collectivist Anarchism: Game Theory, Networks, and the Mathematics of Self-Organisation | 集体无政府主义：博弈论、网络与自组织的数学
Collectivist anarchism is a political tradition that seeks to replace the state and private capital with voluntary, decentralised networks of workers' associations and communities. At first glance, this topic belongs to politics or history, not mathematics. However, many of its core claims - about self-organisation, cooperation, resource allocation, and decision-making without a central authority - can be modelled and tested using tools from A-Level Mathematics, especially game theory, networks, decision mathematics, and probability.
集体无政府主义是一种政治传统，主张以自愿的、去中心化的工人协会和社区网络取代国家与私人资本。乍一看，这个主题属于政治或历史，而不是数学。然而，它的许多核心主张——关于自组织、合作、资源配置以及没有中央权威的决策——都可以用 A-Level 数学中的工具加以建模和检验，尤其是博弈论、网络、决策数学和概率。
1. What Is Collectivist Anarchism? | 什么是集体无政府主义？
Collectivist anarchism, associated with Mikhail Bakunin and the anti-authoritarian wing of the First International, argues that the means of production should be owned collectively by voluntary associations of producers. Unlike centralised state socialism, it rejects a transitional state and instead envisions coordination through federations, mutual aid, and direct democracy. The key mathematical challenge is how to achieve order without a top-down coordinator.
集体无政府主义与米哈伊尔·巴枯宁以及第一国际中的反权威派别相关，主张生产资料应由生产者的自愿联合体集体所有。与中央集权的国家社会主义不同，它拒绝过渡性国家，转而设想通过联合会、互助和直接民主进行协调。关键的数学挑战是：如何在没有自上而下协调者的情况下实现秩序。
The tradition has a recognisable set of institutional features, each of which has a mathematical analogue:
这一传统有一套可辨识的制度特征，而每一条特征都有对应的数学类比：
- Collective ownership of production by voluntary associations - modelled as a shared resource with a common pool.
- Rejection of a transitional state - modelled as a system with no central controller.
- Federations of local groups - modelled as a graph of nodes and edges.
- Mutual aid as an everyday practice - modelled as a repeated game strategy.
- Direct democracy and recallable delegates - modelled as a voting or consensus process.
- 生产资料由自愿联合体集体所有——建模为具有公共池性质的共享资源。
- 拒绝过渡性国家——建模为没有中央控制器的系统。
- 地方团体的联合会——建模为节点与边构成的图。
- 作为日常实践的互助——建模为重复博弈中的策略。
- 直接民主与可罢免代表——建模为投票或共识过程。
The central mathematical question is therefore a control question: can a system of self-interested agents, connected only to their neighbours, produce stable, efficient and fair outcomes without a central planner? That is a question about equilibria, connectivity and convergence, and all three are examinable topics.
因此，核心的数学问题本质上是一个控制问题：一个由自利主体组成、彼此只与邻近者相连的系统，能否在没有中央计划者的情况下产生稳定、高效而公平的结果？这是关于均衡、连通性与收敛性的问题，而这三者都是可考查的内容。
2. Why Use Mathematics? | 为什么使用数学？
Mathematics is a natural language for analysing decentralised systems. Prose arguments about freedom and solidarity are hard to test; a model forces every assumption to be stated explicitly, and then allows consequences to be derived rather than asserted. If a claim about self-organisation is true, it should survive being written as a formal system.
数学是分析去中心化系统的天然语言。关于自由与团结的散文式论证难以检验；而模型迫使每一个假设都被明确写出，然后让结论可以被推导出来，而不是被断言出来。如果某个关于自组织的论断是真的，它就应该能经受住被写成形式系统的检验。
A standard way to begin is to define a strategic model as an ordered triple:
一个标准的起点是把策略模型定义为一个有序三元组：
M = (N, S, u)
where N is the set of agents (workers, associations, communes), S is the set of available strategies, and u is the payoff function that assigns a numerical value to each outcome. Once M is fixed, questions about cooperation become questions about the sign and size of u.
其中 N 是主体集合（工人、协会、公社），S 是可选策略集合，u 是把每个结果映射为数值的支付函数。一旦 M 确定，关于合作的问题就变成了关于 u 的符号与大小的问题。
It is useful to separate three scales of analysis. At the micro level we ask what a single rational worker should do. At the meso level we ask how the network of relationships shapes what is possible. At the macro level we ask what aggregate outcomes - prices, output, stability - emerge from millions of individual choices.
把分析分成三个尺度很有用。微观层面问的是：单个理性工人应该做什么。中观层面问的是：关系网络如何塑造了可能性。宏观层面问的是：从千百万个体选择中涌现出什么样的总体结果——价格、产量、稳定性。
A caution is essential from the start. A model is not a proof of a political claim. Mathematics can show that cooperation is stable under stated conditions, or that it collapses under others; it cannot by itself tell us which conditions we ought to want. The value of the mathematics is precision, not persuasion.
从一开始就必须谨慎。模型并不等于政治主张的证明。数学可以表明在某些给定条件下合作是稳定的，也可以表明在另一些条件下合作会崩溃；但它本身无法告诉我们应当希望哪种条件。数学的价值在于精确，而不在于说服。
3. The Prisoner's Dilemma and the Free-Rider Problem | 囚徒困境与搭便车问题
The simplest formal objection to voluntary cooperation is the Prisoner's Dilemma. Two workers must decide whether to contribute to a shared tool library or to take a free ride. Their payoffs can be summarised in a 2 x 2 matrix.
对自愿合作最简单的形式化反驳是囚徒困境。两名工人必须决定是共同出资建立共享工具库，还是搭便车。他们的收益可以用一个 2 x 2 矩阵概括。
| A \ B | B cooperates / B 合作 | B defects / B 背叛 |
| A cooperates / A 合作 | R = 3, R = 3 | S = 0, T = 5 |
| A defects / A 背叛 | T = 5, S = 0 | P = 1, P = 1 |
The four payoffs are conventionally labelled: T is the temptation to defect, R is the reward for mutual cooperation, P is the punishment for mutual defection, and S is the sucker's payoff. The defining inequalities of a Prisoner's Dilemma are:
这四个收益通常这样标记：T 是背叛的诱惑，R 是相互合作的回报，P 是相互背叛的惩罚，S 是当傻瓜的收益。囚徒困境的定义性不等式是：
T > R > P > S and 2R > T + S
he game is a genuine dilemma: both would be better off at (R, R) than at (P, P), yet each has a strictly dominant strategy to defect. The unique Nash equilibrium (D, D) is Pareto inefficient.
取 T = 5、R = 3、P = 1、S = 0，则 5 > 3 > 1 > 0，且 2R = 6 > T + S = 5，所以这确实构成一个困境：双方在 (R, R) 处都比在 (P, P) 处更好，但每一方都有严格占优的背叛策略。唯一的纳什均衡 (D, D) 是帕累托低效的。
This is the free-rider problem in its purest form. If a communal workshop, a shared irrigation channel or a mutual insurance fund is a common-pool resource, then each individual has an incentive to enjoy the benefit while shifting the cost onto others. Any theory of self-organisation must explain how this incentive is neutralised.
这就是最纯粹形式的搭便车问题。如果公共作坊、共享灌溉渠或互助保险基金属于公共池资源，那么每个人都有动机去享受收益而把成本转嫁给他人。任何自组织理论都必须解释这种动机是如何被抵消的。
4. Iterated Games and the Strategy of Reciprocity | 重复博弈与互惠策略
The dilemma changes when the same two agents meet repeatedly. In an iterated game, a defection today can be punished tomorrow, so cooperation can become rational even for purely self-interested agents. If future payoffs are discounted by a factor δ with 0 < δ < 1, then conditional cooperation is sustainable provided δ is large enough.
当同样的两个人反复相遇时，困境就改变了。在重复博弈中，今天的背叛可以在明天被惩罚，因此即便对纯粹自利的主体而言，合作也可能变得理性。如果未来收益以因子 δ（0 < δ < 1）贴现，那么只要 δ 足够大，有条件的合作就是可持续的。
δ ≥ (T − R) ÷ (T − P)
s care about the future at least as much as they care about the present, mutual aid is an equilibrium rather than a naive hope.
用上面的数字：(T − R) = 5 − 3 = 2，(T − P) = 5 − 1 = 4，所以 δ ≥ 0.5。如果主体对未来至少与对当下同等重视，那么互助就是一种均衡，而不是天真的期望。
Robert Axelrod's computer tournaments in the 1980s showed that a very simple strategy, Tit-for-Tat, performed remarkably well. Tit-for-Tat cooperates on the first move, then copies the opponent's previous move. It is nice, retaliatory, forgiving and clear - four properties that correspond exactly to the informal ethics of mutual aid.
罗伯特·阿克塞尔罗德在 1980 年代组织的计算机竞赛表明，一个非常简单的策略"针锋相对"（Tit-for-Tat）表现得格外出色。它在第一步合作，之后复制对手上一步的选择。它友善、会报复、会宽恕、而且清晰——这四个特性恰好对应互助的非正式伦理。
Tit-for-Tat is not unbeatable. Against an unconditional defector it loses, and in noisy environments where moves are misperceived it can spiral into mutual retaliation. Generous variants such as Tit-for-Two-Tats, which forgive a single defection, are more robust in noisy conditions. The mathematical lesson is that reciprocity works, but it needs forgiveness to survive error.
针锋相对并非不可战胜。面对无条件背叛者它会输，而在动作可能被误解的噪声环境中，它可能陷入相互报复的螺旋。像"两报还一报"这样会宽恕单次背叛的宽容变体，在噪声条件下更稳健。数学上的教训是：互惠有效，但需要宽恕才能在错误中存活。
For a collectivist federation, this suggests a design principle: repeated interaction, stable membership and freedom of association are not merely ethical preferences, they are the structural conditions that make cooperation individually rational.
对一个集体主义联合会而言，这提示了一条设计原则：重复互动、成员稳定和结社自由不只是伦理偏好，它们正是使合作在个体层面变得理性的结构性条件。
5. Coordination Games and Nash Equilibrium | 协调博弈与纳什均衡
Not every social problem is a Prisoner's Dilemma. Many are coordination problems, where the worst outcome is not betrayal but failure to align. The classic example is the Stag Hunt, in which two workers may jointly hunt a large animal for a big payoff, or each may individually catch a small animal for a safe payoff.
并非所有社会问题都是囚徒困境。许多问题属于协调问题，其中最糟的结果不是被背叛，而是未能对齐。经典例子是猎鹿博弈：两名工人可以联手猎取大型猎物以获得高收益，也可以各自去抓小型猎物以获得稳妥收益。
| A \ B | B hunts stag / B 猎鹿 | B hunts hare / B 抓兔 |
| A hunts stag / A 猎鹿 | 4, 4 | 0, 3 |
| A hunts hare / A 抓兔 | 3, 0 | 3, 3 |
A Nash equilibrium is a strategy profile in which no agent can improve their payoff by changing strategy alone. Formally, for every player i and every alternative strategy sᵢ:
纳什均衡是指这样一种策略组合：任何主体仅凭单方面改变策略都无法提高自己的收益。形式化地说，对每个参与者 i 和每个备选策略 sᵢ：
uᵢ(sᵢ*, s₋ᵢ*) ≥ uᵢ(sᵢ, s₋ᵢ*)
The Stag Hunt has two pure Nash equilibria: (stag, stag) with payoffs (4, 4), and (hare, hare) with payoffs (3, 3). The cooperative equilibrium gives both players more, but it is also riskier, since a mistaken partner leaves you with 0. This tension between payoff dominance and risk dominance is the mathematical core of the problem of trust.
猎鹿博弈有两个纯策略纳什均衡：(猎鹿, 猎鹿)，收益为 (4, 4)；(抓兔, 抓兔)，收益为 (3, 3)。合作均衡让双方获得更多，但也更具风险，因为如果伙伴判断错误，你将得到 0。收益优势与风险优势之间的这种张力，正是信任问题的数学核心。
= 3 ÷ 7 ≈ 0.429. If coordination is attempted at random with that probability, the chance of successful coordination is q² + (1 − q)² ≈ 0.184 + 0.327 ≈ 0.51 - barely better than a coin toss.
此外还存在一个混合均衡。令 a = 4、b = 3，并令 q 为对手选择猎鹿的概率。当 4q = 3(1 − q) 时，参与者无差异，故 q = 3 ÷ 7 ≈ 0.429。如果以该概率随机尝试协调，则成功协调的概率为 q² + (1 − q)² ≈ 0.184 + 0.327 ≈ 0.51——几乎不比抛硬币好多少。
This result is important for direct democracy. Assemblies, shared calendars, common standards and public commitments act as coordination devices: they raise the probability that others will also choose the cooperative action, moving the system towards the payoff-dominant equilibrium without requiring a central authority.
这个结果对直接民主很重要。集会、共享日程、通用标准和公开承诺都起到协调装置的作用：它们提高了他人也会选择合作行为的概率，从而在没有中央权威的情况下把系统推向收益占优的均衡。
6. Networks: Federations as Graphs | 网络：作为图的联合会
A federation of workers' in V is an association and each edge in E represents a channel of communication, exchange or mutual aid. This single abstraction allows a large number of political questions to be turned into countable ones.
工人协会的联合会很自然地可以建模为图 G = (V, E)，其中 V 中的每个顶点是一个协会，E 中的每条边代表一条沟通、交换或互助的渠道。仅此一个抽象，就把大量政治问题转化为可计数的问题。
The degree of a vertex is the number of edges incident to it, that is, how many partners an association is directly connected with. The adjacency matrix A of the
更多咨询请联系16621398022（同微信）
或截图后用「扫一扫 → 相册」选择本图