5D Library › Collective Dynamics, Coupled-Oscillator Synchrony and Distributed Consensus · 中文
Preface: How Many Come Together, and How They Avoid Becoming One
Note
Please read this first. This book is a teaching and research text on mathematical models, statistical physics and network science. It is not management consulting and makes no political claim of any kind; nor is it investment, medical, psychological or legal advice. Its simulations are computed under stated assumptions and describe only the properties of the models; they say nothing about the future of any real organisation, community, platform or country. Unless a source is given, the “teams”, “committees”, “crowds”, “opinions” and data in the book are fictional or synthesised by the programs, and the parameters are illustrative.
Whenever a concept from physics or mathematics (phase transition, synchronisation, spectral gap, order parameter) is applied to people or society, the book labels it explicitly as an analogy or a model and says where it fails. The book never says that a rule “guarantees” anything, is “inevitable”, or is “proved optimal”.
In 2000 Néda, Ravasz, Brechet, Vicsek and Barabási reported in Nature a phenomenon everyone has witnessed and few have thought about: applause in a theatre sometimes turns from a noisy roar into a regular beat, falls apart after a while, and then becomes regular again (Néda, Ravasz, Brechet, et al. 2000). In the same year the new Millennium Bridge over the Thames in London swayed from side to side as crowds walked across it (Dallard et al. 2001). A few years later Strogatz and colleagues explained the bridge with the same kind of mathematics, the theory of synchronisation of coupled oscillators (Strogatz et al. 2005). Flocks of birds, schools of fish, the opinions in a committee and the views on a social network all seem to raise a similar question: many individuals interact with only a few neighbours each, yet the whole group may fall into step, or may split into several pieces.
This book asks four questions. First, what do the three families of models considered here — coupled-oscillator synchronisation, the collective motion of self-propelled particles, and distributed consensus — actually conclude, and under what assumptions? Second, when is “being in step” useful and when is it harmful? The applause and the bridge show that synchrony itself can be a risk. Third, when a group combines many people’s estimates, how does the network structure decide whether the result is “the wisdom of crowds” or “the amplified opinion of a few”? Fourth, can these models be used to design a test that compares centralised and decentralised communication structures when a core member carries a bias?
What this book covers
Chapter 1 discusses the Kuramoto model: the complex order parameter, the self-consistency equation, the derivation and the conditions of the critical coupling , how the shape of the frequency distribution changes the type of the transition, fluctuations at finite size, and the mechanism of synchronised applause. Chapter 2 discusses the synchronisation model of the London Millennium Bridge and the later criticism of it, together with the master stability function for network synchronisation and the Laplacian eigenratio criterion . Chapter 3 discusses the self-propelled particle model of Vicsek and the three-zone model of Couzin: how noise destroys collective order, how the widths of three zones of interaction produce swarms, tori and parallel groups, and “collective memory”. Chapter 4 discusses leadership without signals: how a few informed individuals guide a group, how the proportion needed falls with group size, and how uninformed individuals weaken the control of a strongly opinionated minority. Chapter 5 discusses DeGroot’s distributed consensus: the influence vector, the spectral gap and the speed of convergence, and the criterion of Golub and Jackson for the “wisdom of crowds”. Chapter 6 discusses bounded-confidence models (the Hegselmann–Krause and Deffuant models) as the confidence bound shrinks from consensus to clusters, and the evidence for and limits of the “echo chamber” metaphor. Chapter 7 reads this research beside “the gentleman harmonises but does not conform” from the Analects (“Zilu”), Shi Bo on harmony and sameness in the Discourses of the States (“Discourses of Zheng”), Yanzi on harmony and sameness in the Zuo Tradition (Duke Zhao, year 20) and “music makes for sameness, ritual makes for difference” from the Record of Music in the Book of Rites, and states the boundary between ancient and modern. Chapter 8 offers a “four-step checklist against echo chambers and for calibrating synchrony” and a pre-registered cluster-randomised controlled trial.
The book borders on several other volumes of 5D Library, which it recalls only briefly without repeating their programs: Volume 8 (the default mode network and neural oscillations), which checked for the Lorentzian by simulation and whose result Chapter 1 only cites; Volume 18 (acoustic persuasion and interpersonal physiological synchrony), which discussed the difference between common driving and mutual coupling and read the opening of the Record of Music, recalled in Chapters 1 and 7; Volume 25 (social network topology and complex contagion), which discussed small-world networks, structural holes and information cascades, cited in Chapters 2 and 5; Volume 26 (probabilistic judgement and collective intelligence), which discussed the diversity prediction theorem and correlated errors, recalled in Chapter 5; Volume 33 (scientific methodology), which explained what “90 % power” means, followed in Chapter 8; and Volume 35 (the evolution of cooperation), adjacent to the group decisions of Chapter 4.
Claims corrected after checking
Several statements in the author’s original framework needed, after checking, to be corrected, supplemented or qualified.
First, the name. Kuramoto is the Japanese physicist Kuramoto Yoshiki (蔵本由紀, simplified 藏本由纪). Volumes 8 and 18 of 5D Library transliterated the name as 库拉莫托; to keep both the usual Chinese rendering and consistency with those volumes, the Chinese text of this book writes 藏本(库拉莫托)throughout, while the title keeps the author’s wording 藏本相变.
Second, the critical coupling. is the threshold, obtained by Kuramoto for global coupling, and a frequency distribution that is symmetric about its mean and unimodal (non-increasing on ), at which the incoherent state loses stability and a synchronised state appears; for a smooth unimodal distribution the state above threshold is partially locked (Kuramoto 1975; Strogatz 2000). For the Lorentzian, holds exactly for every ; for a general smooth unimodal distribution there is only a square-root law near threshold, a continuous (second-order) transition when . The framework’s “second-order phase transition” must be qualified: for the uniform distribution jumps from 0 to at , where all oscillators lock at once without passing through partial locking, a first-order transition (Pazó 2005). At finite there is no phase transition in the strict sense; fluctuates at order and the threshold is smeared (program in Chapter 1). Kuramoto’s 1975 paper is a short note in volume 39 of Lecture Notes in Physics; the forerunner of the model is Winfree’s 1967 work (Winfree 1967).
Third, applause. Néda and colleagues concluded that once the audience doubles its clapping period (halving the frequency) the dispersion of frequencies shrinks with it, the critical coupling falls, and only then does synchrony appear; during synchrony the average sound intensity falls, the audience speeds up to be louder, the dispersion grows again and synchrony vanishes (Néda, Ravasz, Brechet, et al. 2000; Néda, Ravasz, Vicsek, et al. 2000). The framework’s “when synchronised, the clapping frequency halves and the noise intensity falls” is factually right, but the direction of cause must be stated: halving the frequency is the precondition for synchrony, not its result. The measured ratio of dispersions is about 2.5, and the Nature note gives no formula for ; the argument is in the longer Physical Review E paper of the same year.
Fourth, the Millennium Bridge. Strogatz and colleagues’ 2005 paper is a brief communication in Nature (Strogatz et al. 2005), and its parameter was obtained by fitting engineering test data. The framework’s “critical crowd of about 165 people” must be qualified: Belykh and colleagues in 2017 used about 165 as the observed value (Belykh et al. 2017), whereas Lenci and Marcheggiani in 2011, quoting the tests of Dallard and colleagues, give about 155 (Lenci and Marcheggiani 2011); secondary accounts differ. The criticism the framework mentions must be supplemented: Belykh and colleagues in 2017 held that large-amplitude wobbling needs pedestrian phase-locking but that the initial small wobble need not be caused by synchrony; Belykh, Bocian, Champneys and colleagues went further in Nature Communications in 2021 and argued that synchrony of footsteps is a consequence of the instability, not its cause (Belykh et al. 2021).
Fifth, the master stability function. The method was introduced by Pecora and Carroll in 1998 (Pecora and Carroll 1998), and the ratio condition for a bounded stable region follows directly from it; using the eigenratio as an index for comparing the “synchronisability” of different networks, and applying it to small-world networks, is the work of Barahona and Pecora in 2002 (Barahona and Pecora 2002). The framework’s attribution of the criterion to 1998 is not wrong, but the 2002 source must be added, and the criterion must be restricted to the case in which the stable region is a bounded interval ; when the stable region is unbounded only matters. The ratio also depends on whether the coupling is normalised by degree: in the program of Chapter 2 a star network can synchronise under normalised coupling, whereas without normalisation its ratio equals .
Sixth, Couzin’s three-zone model. The three-dimensional model of Couzin and colleagues in 2002 yields four collective states: swarm, torus, dynamic parallel and highly parallel (Couzin et al. 2002); the framework lists three, and the book completes the list. The main control parameter is the width of the zone of orientation (together with the width of the zone of attraction ), and the states show hysteresis: under the same rules, the state of the group depends on its history (“collective memory”). The program in this book is a two-dimensional simplification and reproduces only the qualitative conclusions.
Seventh, the informed minority. The conclusion of Couzin and colleagues in 2005 is qualitative: the larger the group, the smaller the proportion of informed individuals needed to guide it (Couzin et al. 2005); the paper gives no universal proportion, and “about 5 %” quotes the share of scout bees in honeybee swarms, not a law. The framework’s “how an undecided majority restrains hijacking by an extreme minority” needs two corrections: first, in the 2011 experiment of Couzin and colleagues the untrained fish added numbered 0, 5 or 10 and were not necessarily a majority; their effect was to return control to the numerical majority (6 against 5) (Couzin et al. 2011); second, the paper’s “strongly opinionated minority” refers to the strength of preference and carries no political meaning; applying it to human politics can only be an analogy. The program in Chapter 4 also shows that in the book’s simple model, in which individuals copy one random other, the uninformed have no effect at all; Hartnett and colleagues found in 2016 that the direction of their effect depends on how nonlinear individual responses are (a relatively linear response makes them reinforce the majority, a highly nonlinear one makes them favour the strongly biased minority instead) (Hartnett et al. 2016). The effect is therefore not a general law; it depends on the specific form of the interaction rules.
Eighth, DeGroot consensus. Convergence requires the network to be strongly connected and aperiodic (DeGroot 1974; Golub and Jackson 2010); the speed is set by the largest-modulus non-unit eigenvalue , and gives only the asymptotic rate. Golub and Jackson’s criterion — a society is wise if and only if — concerns a sequence of networks whose size tends to infinity, with independent signals whose variances are bounded below; they also point out that the speed of convergence and wisdom are unrelated.
Ninth, bounded confidence. “Consensus at ” in the Hegselmann–Krause model holds only for large populations (about 0.2 in Fortunato 2005, 0.19 in Lorenz’s density-based model) (Fortunato 2005; Lorenz 2007); in small populations the threshold is smeared, and the probability of consensus need not be monotone in (Lorenz 2006). The Deffuant model always reaches consensus for , and the number of major clusters is roughly (Deffuant et al. 2000; Weisbuch et al. 2002), often with a few outliers left at the extremes. The framework’s “polarisation” is, in these models, just two clusters, not necessarily at the extremes; “consensus → polarisation → fragmentation” is a broad trend, not three sharply bounded phases. “Echo chamber” is an analogy, and the empirical evidence does not support the simple inference that more exposure to opposing views reduces polarisation (Bail et al. 2018; Nyhan et al. 2023).
Tenth, the classical texts. “The gentleman harmonises but does not conform; the petty man conforms but does not harmonise” (Analects, “Zilu”), Shi Bo’s “harmony truly generates things, sameness does not continue; to balance one thing with another is called harmony” (Discourses of the States, “Discourses of Zheng”) and “music makes for sameness, ritual makes for difference; sameness brings mutual affection, difference brings mutual respect” (Book of Rites, “Record of Music”) were all checked against the received texts. The book adds the sentence that follows in the Record of Music, “when music prevails there is laxity, when ritual prevails there is estrangement”, and uses the passage the framework suggested, Yanzi on harmony and sameness in the Zuo Tradition (Duke Zhao, year 20; the received text reads 若琴瑟之专壹). The “difference” of the Record of Music means the distinction of noble and base ranks, not pluralism in the modern sense, and Chapter 7 states this boundary.
Eleventh, the Chapter 8 trial. The framework’s “core node carrying a bias” is implemented in the trial as a programmed seat disclosed to participants in advance. The program in Chapter 8 shows that a centralised structure does not always increase the error: when the bias of the programmed seat is small (log bias 0.40), the centralised structure also pulls everyone’s estimates together, the fall in variance offsets the bias, and the two arms have almost the same error. The trial must therefore fix the size of the bias (the book takes “showing twice the true value”), and its conclusion holds only for that setting. This agrees with the experiment of Becker and colleagues: the error grew when the central person’s initial estimate lay away from the truth and fell when it lay towards it (Becker et al. 2017).
Each chapter states these points as they arise, and they are collected in the accompanying editorial notes.
How to read the boxes
As in the other books of 5D Library, different kinds of statement are marked separately.
Evidence
Evidence boxes report published research findings, with sources.
Calculation
Calculation boxes give formulas and short Python programs. Every program runs unchanged in a standard Python 3 environment, uses only the standard library and fixed random seeds, and is followed by its actual output. The “groups”, “teams” and “opinions” in the book are synthesised by the programs; the parameters are illustrative and are not estimates for any real system.
Hypothesis · not established
Hypothesis boxes contain the author’s ideas, are marked “not established”, and say what result would show them to be wrong.
Source text
Source boxes quote classical texts, with the author’s plain-language explanation.
Note
Note boxes give reminders and explain choices.
About this edition
This is version 1.0.0, the first complete edition: a preface and eight chapters, in Chinese and English. The book contains 16 programs and 5 vector figures. The text, boxes and code were drafted with an AI writing tool from the author’s title, framework and notes; the author is responsible for reviewing and revising them and for checking every source. Corrections and references the author has missed are welcome through the 5D Library website.
Disclosure
The author uses the works of 5D Library, including this book, as the methodological basis of his community courses, the 5D Ascension Cards (Dimensional Shift Card Decks) and the Yafu AI knowledge base, and therefore has a commercial interest in the subject of this book. The book promotes no course, card deck or service; the Chapter 8 trial is designed to be run and analysed by researchers independent of the author’s business and does not recruit from the author’s community, students or customers; all the code and the study protocol are published openly for anyone to check. The author founded and runs the Yafu Global and Asia Influencer X community businesses and has a direct commercial interest in the subject of this book. The book is not management consulting and makes no political claim of any kind.
References
Fan, Y. (2026). Collective Dynamics, Coupled-Oscillator Synchrony and Distributed Consensus: From the Kuramoto Phase Transition and Vicsek Flocking to DeGroot Spectral Convergence, Read Beside Classical Chinese Thought on Harmony and Conformity [集群动力学、耦合振子同步与分布式共识的数学模型] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.23828063 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).