Six Dynamics · One Reading六种动态 · 一次读取

Every group has a temperature. Most adults can feel it. Almost none can name it. 每个群体都有温度。多数成年人能感受到它。几乎没人能命名它。

Social Temperature reads the six hidden dynamics that shape any group conversation — dominance, insecurity, manipulation, attraction, emotional pressure, and the hierarchy nobody admits exists. Paste a sample of a chat, a community thread, or a Discord server and the engine returns a thermal reading. 社交温度读取塑造任何群体对话的六种隐藏动态——支配、不安、操纵、吸引、情绪压力，以及无人承认存在的层级。粘贴一段对话、社区帖或 Discord 服务器的样本，引擎返回一份热力读取。

six dynamics chat-sample lab thermal reading EN · 中文
S1 · Six Hidden Dynamics

Six dynamics. Each one leaves a signature in the chat log.六种动态。每一种都在对话记录里留下签名。

A group conversation is not a flat exchange of equal voices. It is a thermal map: hot zones where the energy concentrates, cold zones where someone is watching but not participating, and currents that move between the two. Six dynamics produce most of the heat.一场群体对话不是平等声音的扁平交换。它是一张热力图：能量集中的热区、有人观察但不参与的冷区，以及两者之间流动的暗流。六种动态产生大部分热量。

S2 · Chat-Sample Lab

Pick a sample, or paste your own. The engine returns a thermal reading.挑一个样本，或粘贴你自己的。引擎返回一份热力读取。

Three pre-loaded samples cover common cases — a friend group on a tense weekend, a workplace channel after bad news, a Discord with an unstable hierarchy. Or paste your own and the heuristic will tag the dynamics it can detect. Honesty: this is a pattern matcher, not a frontier model.三段预载样本覆盖常见情况——紧张周末的朋友群、坏消息后的职场频道、层级不稳的 Discord。或粘贴你自己的，启发式将标记它能检测到的动态。坦诚：这是一台模式匹配器，不是前沿模型。

S3 · Thermal Reader

Five canned probes plus a free-text fallback.五个预设探针 + 自由文本回退。

S4 · Guardrails

What this engine refuses to pretend.本引擎拒绝伪装的事。

Five things Social Temperature will not do.社交温度不会做的五件事。

  • No surveillance assistance. The engine refuses to analyze chats whose participants have not consented to analysis. The pre-loaded samples are fictional; your own sample is for self-reflection in a group you are part of.不做监视协助。引擎拒绝分析参与者未同意被分析的对话。预载样本是虚构的；你自己的样本用于"你所在群体"的自我反思。
  • No "out the manipulator" framing. The manipulation dynamic is mapped descriptively. The engine refuses prompts of the form "show me which person in this chat is the manipulator." The dynamic is in the room, not in a person.不做"揭露操纵者"框架。操纵动态按描述方式映射。引擎拒绝"告诉我这段对话里谁是操纵者"形式的提问。动态在房间里，不在某个人身上。
  • No clinical claims. Insecurity and emotional pressure are dynamics, not diagnoses. The engine reads patterns; it does not pathologize individuals.不做临床主张。不安与情绪压力是动态，不是诊断。引擎读取模式；它不把个人病理化。
  • No hierarchy moralism. Hidden hierarchies are observable; whether they are good or bad depends on what they enable. The engine names them; it does not judge them.不做层级道德论。隐藏层级是可观察的；好坏取决于它们使什么成为可能。引擎命名它们；不评判它们。
  • No infinite-AI-generation slop. Six curated dynamics + three sample chats + five canned probes is better discipline than infinite plausible-sounding LLM group-dynamics analysis.不做无限 AI 生成的污水。六种精选动态 + 三段示例对话 + 五个预设探针，比无穷个貌似合理的 LLM 群体动态分析是更好的纪律。