作者：陀先润 编辑：李晶 校对：程筱筱 翻译：彭小梅

前几天，一名网友讲述了这样一段经历：他在北京前往毛主席纪念堂参观，安检时身份证连续两次无法通过。执勤人员查看屏幕内容后，询问他最近是否通过 12345 投诉过小米。他承认确有其事，且投诉尚未结案。对方点头示意，随后放行。

这件事的关键，并不在于它是否经过完整核实，而在于它在逻辑上完全自洽，在现实中完全成立。它不需要额外加工，也不依赖任何阴谋论假设。相反，它之所以令人不安，正是因为它看起来过于“正常”。

不少人第一反应是质疑企业权力，怀疑是否某家公司的影响力已经可以左右公共秩序。这种理解方式看似尖锐，实则安全。它把问题压缩为“资本勾结权力”的个案，好像只要换一家企业、换一次投诉对象，就能规避风险。但事实恰恰相反：决定一个人能否“刷得过去”的，从来不是企业，而是一个早已运行多年的国家级数据治理体系。

在今天的中国，个人并不是以权利主体的身份被对待，而是以数据对象的形式被管理。身份、行踪、消费、医疗、婚姻、信访、投诉、表达记录，被持续拆解、整合、关联，最终形成一个可调用、可标记、可联动的“人像”。技术上，这并不新鲜；真正发生变化的是，这套系统已经从后台管理走向前台治理，并且缺乏任何有效的外部约束。

这意味着，投诉不再只是表达不满，维权不再只是寻求救济，反映问题首先是一条数据，其次才可能是一项诉求。当数据进入系统，它的命运就不再由当事人掌控，而由算法、规则、权限和所谓“稳定需要”共同决定。是否反复提交、是否继续追问、是否涉及特定对象，都会被转译为风险特征。这些特征不需要公开，不需要论证，也不需要申诉路径，只需要在某个关键节点弹出一个提示：注意、限制、人工处理。

这正是现代控制最危险的形态：它不靠公开镇压，而靠隐形分类；不宣布罪名，却制造障碍。个人不会收到正式通知，只会发现“刷不过”“进不去”“被多看了一眼”。没有文书，没有解释，更没有纠错机制。当事人甚至不知道自己属于哪一类，也不知道如何退出这套系统。

有人会反驳：可他最终还是被放行了。恰恰在这里，控制的逻辑暴露得最为清楚。真正成熟的治理机器，并不追求每一次都拦下个体，而是要让人意识到，自己随时可能被拦下。今天放行，是因为系统判断“问题不大”；明天限制，也不需要新的理由，只需要“情形不同”。当权利从规则变成裁量，从确定性变成不确定性，社会行为就会自动发生变化。

于是，自我驯化开始出现。人们减少表达，回避公共事务，反复权衡“值不值得留下记录”。不需要命令沉默，沉默会自行生成；不需要禁止投诉，投诉会自然减少。久而久之，制度不再需要高强度维稳，因为社会已经完成了自我维稳。

更具讽刺意味的是，这一切往往以“治理现代化”“数字政府”“便民服务”的名义推进。数据被高度集中，却没有独立审计；权限被层层叠加，却没有责任追溯；裁量被不断下沉，却没有清晰边界。公民被要求实名、配合、守法，却被拒绝查询自己是否被标记、因何被标记、如何纠错。系统可以在关键时刻告诉一线人员某人“因为什么被注意”，却不允许当事人作为权利主体知情。

在这样的结构中，“正常人没事”成为一种极其危险的安慰。因为“正常”并不是法律概念，而是系统判断。它可以随时间、随环境、随政策需要而改变。今天是普通投诉者，明天可能成为反复申诉对象；今天是消费纠纷，明天可能被归入“影响稳定因素”。当规则不透明、标准不公开、救济不存在时，所谓安全感，只建立在“暂时没轮到我”之上。

这件事真正揭示的，不是某一次安检的偶发异常，而是一个社会如何在缺乏边界的情况下，把技术优势转化为控制优势。它清楚地表明：在一个以数据库为神经系统的治理结构中，公民不是被说服的对象，而是被管理的变量；权利不是不可侵犯的底线，而是可以被临时暂停的状态。

当一个社会需要公民“尽量少留下记录”才能自保，当参与公共事务本身需要进行风险评估，这个社会就已经不再是现代意义上的公民社会。它只是一个运转良好的系统，以及一群学会绕开系统的人。

真正令人恐惧的，从来不是暴力本身，而是暴力不再需要露面。当一个国家不必告诉你为什么限制你，只需要让你感受到它随时可以限制你，这种权力就已经摆脱了法律的约束，进入了纯粹的技术统治阶段。

而在这种体制下，权力最忌惮的，从来不是犯罪，而是记忆。因为记忆意味着记录，记录意味着追问，而追问本身，就是对控制的否定。

When the State Machine Begins to Remember You

— You Are Not Being Censored, You Are Being Marked

Author: Tuo Xianrun Editor: Li Jing Proofreader: Cheng Xiaoxiao Translator: Peng Xiaomei

Abstract:The Chinese Communist Party is using advanced digital technology to attempt to mark every Chinese person, in order to identify those who may resist and to impose surveillance and suppression on them, all for the sake of maintaining the dictatorship of this regime.

A few days ago, an internet user described the following experience: while traveling in Beijing to visit the Chairman Mao Memorial Hall, his ID card failed to pass security screening twice in a row. After seeing the contents displayed on the screen, the on-duty staff asked him whether he had recently filed a complaint against Xiaomi through 12345. He admitted that he had indeed done so, and that the complaint had not yet been resolved. The other party nodded, then allowed him to pass.

The key point of this incident does not lie in whether it has been fully verified, but in the fact that it is entirely self-consistent in logic and entirely plausible. It requires no additional embellishment, nor does it rely on any conspiracy theory assumptions. On the contrary, what makes it disturbing is precisely the fact that it appears overly “normal.”

Many people’s first reaction is to question corporate power, suspecting whether the influence of some company has already reached the point where it can affect public order. This way of understanding the issue appears sharp but is in fact safe. It compresses the problem into an isolated case of “capital colluding with power,” as if changing the company or changing the target of the complaint would avoid the risk. But the truth is exactly the opposite: what determines whether a person can “get through the scan” has never been the company, but rather a national-level data governance system that has already been operating for many years.

In China today, the individual is not treated as a subject of rights but is managed as a data object. Identity, movements, consumption, medical care, marriage, petitions, complaints, and records of expression are continuously dismantled, integrated, and correlated, ultimately forming a retrievable, markable, and linkable “human profile.” Technically, this is nothing new; what has truly changed is that this system has already moved from backstage management to frontstage governance and lacks any effective external constraint.

This means that a complaint is no longer merely an expression of dissatisfaction, and rights defense is no longer merely a search for remedy. Reporting a problem is first a piece of data, and only secondarily may it become a claim. Once the data enters the system, its fate is no longer controlled by the person concerned, but is jointly determined by algorithms, rules, permissions, and so-called “stability needs.” Whether someone submits repeatedly, whether someone continues asking questions, and whether a matter involves certain specific targets, all can be translated into risk characteristics. These characteristics do not need to be public, do not need to be argued for, and do not need any path of appeal. They only need to trigger a prompt at some key node: attention, restriction, manual handling.

This is precisely the most dangerous form of modern control: it does not rely on open repression, but on invisible classification; it does not announce charges yet creates obstacles. The individual will not receive any formal notice, but will only discover that something “won’t scan,” “can’t be entered,” or that he has “been looked at one more time.” There is no document, no explanation, and even less any correction mechanism. The person concerned may not even know what category he belongs to, nor how to exit this system.

Someone may object but, in the end, he was still allowed through. It is precisely here that the logic of control is exposed most clearly. A truly mature governance machine does not seek to stop the individual every single time, but rather to make people realize that they may be stopped at any time. Letting him through today is because the system judged that “the issue is not serious”; restricting him tomorrow does not require any new reason, only that “the circumstances are different.” When rights change from rule-based to discretionary, from certainty to uncertainty, social behavior will automatically begin to change.

Thus, self-domestication begins to appear. People reduce expression, avoid public affairs, and repeatedly weigh whether “it is worth leaving a record.” There is no need to order silence, because silence will generate itself; there is no need to forbid complaints, because complaints will naturally decrease. Over time, the system no longer needs high-intensity stability maintenance, because society has already completed self-stability maintenance.

Even more ironic is that all of this is often advanced in the name of “modernized governance,” “digital government,” and “convenient public services.” Data is highly concentrated, yet there is no independent audit; permissions are layered one upon another, yet there is no tracing of responsibility; discretion is constantly pushed downward, yet there are no clear boundaries. Citizens are required to use their real names, cooperate, and obey the law, yet they are denied the right to inquire whether they have been marked, why they have been marked, and how to correct errors. The system can, at a critical moment, inform frontline personnel why a certain person has “drawn attention,” yet it does not allow the person concerned, as a subject of rights, to know.

Within such a structure, “normal people are fine” becomes an extremely dangerous consolation. Because “normal” is not a legal concept, but a system judgment. It can change with time, with environment, and with policy needs. Today one may be an ordinary complainant; tomorrow one may become a repeated-petition subject. Today it is a consumer dispute; tomorrow it may be classified as a “factor affecting stability.” When rules are not transparent, standards are not public, and remedies do not exist, the so-called sense of security rests only on the fact that “it has not reached me yet.”

What this incident truly reveals is not some accidental abnormality in one security inspection, but how a society, in the absence of boundaries, transforms technological advantage into control advantage. It clearly shows that: in a governance structure that uses databases as its nervous system, citizens are not subjects to be persuaded, but variables to be managed; rights are not inviolable bottom lines, but states that can be suspended.

When a society requires its citizens to “leave as few records as possible” to protect themselves, and when participation in public affairs itself requires risk assessment, that society is no longer a modern civil society in the true sense. It is only a well-functioning system, along with a group of people who have learned how to maneuver around the system.

What is truly frightening has never been violence itself, but the point at which violence no longer needs to show its face. When a state no longer needs to tell you why it is restricting you but only needs to make you feel that it can restrict you at any time, that power has already broken free of legal constraint and entered a stage of pure technological rule.

And under such a system, what power fears most has never been crime, but memory. Because memory means records, records mean questioning, and questioning itself is a negation of control.