Small-World Networks — high clustering and short path lengths coexist. Rewiring just a tiny fraction (even 1%) of local links into long-range shortcuts collapses average path length while clustering stays intact (Watts & Strogatz). Non-trivial: a few shortcuts do most of the "shrinking" — links aren't equal; reachability ≠ navigability (the famous six-degree chains were ~6 steps, but people couldn't see the path); the same shortcuts that spread ideas also spread contagion and risk. To make something diffuse, invest in the few cross-cluster bridges, not in all links equally.
AI Prompts
中文提示词
我想让 [想法/产品/影响力] 在 [某个网络或群体] 中扩散。请用小世界网络分析：
① 这个网络里，谁是同时连着多个圈子的"长程捷径"节点？
② 我现在的努力是均摊在所有连接上，还是集中在少数桥接节点上？
③ 给出 2 个具体动作：建立或激活哪条跨群体捷径，能最大化缩短扩散路径。
English Prompt
I want [idea/product/influence] to spread through [a network or community]. Analyze with small-world networks:
1. Who are the "long-range shortcut" nodes connecting otherwise-separate clusters?
2. Am I spreading effort evenly across all links, or concentrating on the few bridges?
3. Give 2 concrete moves: which cross-cluster shortcut to build or activate to maximally shorten diffusion paths.
弱连接的力量 · The Strength of Weak Ties
"弱连接，而非密友，是新信息与新机会的最大来源。" — Mark Granovetter, 1973
非平凡点：① 弱连接的价值不在"关系强度"，而在桥接性——它跨越了"结构洞"，连通两个本无往来的群体。强连接通常嵌在同一个稠密小团里，信息在内部反复回荡。② 站在结构洞两侧的人能做信息套利：把 A 圈的知识带到 B 圈。创新常发生在边界，而非中心。③ 功能不可混淆：弱连接提供"信息"，强连接提供"支持与信任"——维护一大批强连接成本高且回报递减，而战略性地激活一批低成本弱连接，反而能极大扩张你的信息半径。
① AI 时代的认知套利：你的弱连接网络（关注的不同领域的人、不同社区）决定了你能接触到多少"非冗余信号"。若信息源全在同一个同温层，再多也是冗余。主动维护跨领域弱连接 = 给认知装上多个独立传感器。② 重大机会的规律：无论是自己的职业转折还是孩子的成长资源，往往来自"朋友的朋友"，而非核心圈。把全部精力投在巩固核心圈，等于亲手关上了通往新机会的那扇门。
English Summary
The Strength of Weak Ties — close friends (strong ties) share highly redundant information; you mostly already know what they know. Acquaintances (weak ties) bridge into clusters you can't otherwise reach, delivering non-redundant information (Granovetter). The value lies not in tie strength but in bridging structural holes; innovation and opportunity arise at boundaries, not centers. Functions differ: weak ties give information, strong ties give support and trust. Strategically activating low-cost weak ties widens your information radius far more than over-investing in a few strong ones.
AI Prompts
中文提示词
我想为 [目标：找机会 / 获取新信息 / 扩大影响] 优化我的关系网络。请用弱连接理论分析：
① 我目前的信息主要来自强连接（同温层）还是弱连接（跨圈）？冗余度有多高？
② 找出 3 个我可以低成本激活的弱连接，它们能桥接到哪些我现在触不到的圈子？
③ 帮我区分：哪些需求该找强连接（支持/信任），哪些该找弱连接（新信息）？
English Prompt
I want to optimize my network for [goal: opportunity / new information / influence]. Analyze using weak-tie theory:
1. Does my information come mostly from strong ties (echo chamber) or weak ties (cross-cluster)? How redundant is it?
2. Identify 3 low-cost weak ties to reactivate — which currently-unreachable clusters do they bridge to?
3. Distinguish which needs call for strong ties (support/trust) vs weak ties (novel information).
① 分布式系统与安全：无标度拓扑解释了为什么系统能扛住随机宕机，却可能被一次针对核心节点的攻击瘫痪——容灾必须给 hub 做重点冗余，而非均匀冗余。② 个人杠杆（"AI 超级个体"）：你的影响力增长同样服从偏好依附——早期获得的关注会吸引更多关注，冷启动最难，但越过临界点后增长自我加速。策略：集中资源突破第一个 hub 级节点（一个关键平台、一个关键人），而非均匀撒网。
English Summary
Scale-Free Networks & Preferential Attachment — real networks (web, citations, social, protein) follow a power-law degree distribution: a few hubs hold most connections, most nodes have few. Mechanism: new nodes preferentially attach to already-popular ones ("rich get richer," Barabási & Albert). Key points: this emerges from growth + preferential attachment, not design; it's simultaneously robust to random failure yet fragile to targeted attacks on hubs; there's no "typical node," so averages mislead (no characteristic scale). To protect a system, harden its hubs; to influence it fast, target hubs; but don't bet everything on hubs — they're the fragile points too.
AI Prompts
中文提示词
我在分析 / 设计 [一个网络：系统 / 社区 / 影响力布局]。请用无标度网络分析：
① 这个网络里的 hub（超级节点）是谁 / 是什么？它们承载了多大比例的连接？
② 鲁棒性审计：哪些 hub 是单点故障？遭遇针对性攻击会怎样？该给哪些 hub 加冗余？
③ 若我想用最小投入最大化影响，该集中撬动哪个 hub，而不是均匀撒资源？
English Prompt
I'm analyzing/designing [a network: system / community / influence map]. Analyze using scale-free networks:
1. Who/what are the hubs? What fraction of all connections do they carry?
2. Robustness audit: which hubs are single points of failure, what happens under targeted attack, which need redundancy?
3. If I want maximum impact for minimum investment, which hub should I concentrate on rather than spreading effort evenly?
① AI 工作流：若你只用一个模型、一种 prompt 风格、一个信息源来训练自己的判断，你是在给认知亲手砌回音室。主动引入异质视角（不同模型、专门唱反调的 prompt、对立观点）= 给思考装上"反共识传感器"。② 深度学习者的陷阱：你越深耕一个领域，圈子越同质，越容易把"圈内共识"误当"客观真理"。佛学讲"所知障"——知道得越多，反被已知遮蔽。定期跨出同温层，是认知健康的必要维护。
English Summary
Homophily & Echo Chambers — people preferentially connect with similar others (age, views, background); at the network level this self-segregates into internally homogeneous, mutually isolated clusters. Combined with recommender feedback, it produces echo chambers and filter bubbles. Key points: polarization emerges without anyone intending it — mild individual preference for similarity plus information flowing along ties suffices (isomorphic to Schelling's segregation model); echo chambers feel high signal-to-noise but are actually high in redundancy (the same signal echoing); recommender systems optimize engagement, which rewards in-group content, accelerating the narrowing. The antidote is deliberate counter-homophily — auditing your information network for genuine diversity, not comfort.
AI Prompts
中文提示词
我想检查 [我的信息环境 / 团队 / 某个社区] 是否陷入了回音室。请用同质相吸分析：
① 我的主要信息源在 [立场 / 背景 / 视角] 上的多样性如何？哪里高度同质？
② 哪些信号看似"很多人认同"，其实只是同一来源的冗余回响？
③ 给出 3 个具体的"反同质"动作：接触哪些异质来源能最有效打破我的过滤气泡？
English Prompt
I want to check whether [my information diet / team / a community] has become an echo chamber. Analyze using homophily:
1. How diverse are my main information sources in [stance/background/perspective]? Where is it highly homogeneous?
2. Which signals that look "widely endorsed" are actually redundant echoes of a single source?
3. Give 3 concrete counter-homophily moves: which heterogeneous sources would most effectively pop my filter bubble?