现象学工程

Phenomenological Engineering

一

I
回看历史，会发现一个反复出现的结构：真正改变人类走向的大概念，很少是某个天才凭空发明的，而是在多个互不相干的领域里各自积累了足够多的线索之后被逼出来的。在名字到来之前，很多人都在做同一件事，却都不知道自己在做同一件事。

Look back at history and a recurring structure appears: the concepts that genuinely change humanity's course are rarely invented out of thin air by a single genius. They are forced into existence after enough clues have accumulated, separately, across a number of fields that have nothing to do with one another. Before the name arrives, many people are already doing the same thing without knowing that they are doing the same thing.

十九世纪上半叶的欧洲就是这样一个时刻。工程师在琢磨怎样让蒸汽机烧同样的煤跑得更远；化学家在测量各种反应放出多少热；生理学家在追问食物、体温和肌肉做功之间是什么关系；物理学家在研究电流与磁的相互转化；天文学家在计算太阳靠什么维持了这么久还没熄灭。这些人分属不同的学会，读不同的期刊，彼此几乎没有交流，各自都以为自己在处理一个局部的技术问题。

Europe in the first half of the nineteenth century was such a moment. Engineers were working out how to make a steam engine travel further on the same coal. Chemists were measuring how much heat various reactions gave off. Physiologists were asking what the relationship was between food, body temperature, and muscular work. Physicists were studying the mutual conversion of electricity and magnetism. Astronomers were calculating what had kept the sun burning so long without going out. These people belonged to different societies, read different journals, and had almost no contact with one another. Each believed he was handling a local technical problem.

然后到了1840年代，事情忽然收束了。迈尔、焦耳、亥姆霍兹、科尔丁等好几个人，在几乎相同的几年里，从完全不同的路径上，各自独立地提出了同一件事：能量。热、功、电、化学反应、生命活动，全都是同一种东西在换不同的形式，总量不变。这个概念一出现，之前所有散落的线索被归拢到同一根轴上。蒸汽机效率不再是经验技巧，而是可以从理论上推出上限的东西；生理学第一次能和物理学讲同一种语言；随后熵、热力学第二定律、统计力学接连而来。工业文明运转了近百年之后，终于拿到了自己的说明书。

Then, in the 1840s, everything suddenly converged. Mayer, Joule, Helmholtz, Colding and several others, within nearly the same few years and by entirely different routes, each independently arrived at the same thing: energy. Heat, work, electricity, chemical reactions, the activity of living things, all of it was one substance changing form, with the total quantity conserved. The moment the concept appeared, every scattered clue was gathered onto a single axis. Steam engine efficiency stopped being a matter of empirical craft and became something whose upper limit could be derived theoretically. Physiology could speak the same language as physics for the first time. Entropy, the second law of thermodynamics, and statistical mechanics followed in sequence. Nearly a century after industrial civilization had begun running, it finally received its own instruction manual.

一百年后同样的事又发生了一次。电报网络、密码破译、电话交换、神经元的放电、遗传的稳定传递、可计算性——这些东西在1930年代看不出任何关系。1948年，香农给信息一个定义，维纳给控制论一个框架，一切都被收拢，然后是计算机、分子生物学、认知科学和互联网。

A hundred years later the same thing happened again. Telegraph networks, codebreaking, telephone switching, the firing of neurons, the stable transmission of heredity, computability — in the 1930s none of these appeared to be related. In 1948 Shannon gave information a definition and Wiener gave cybernetics a framework, and everything was gathered in; then came computing, molecular biology, cognitive science, and the internet.

这类概念有一个共同特征：它不属于任何一个既有学科。它们出现在学科之间的缝隙里，然后把两侧的墙一起拆掉。

Concepts of this kind share one feature: they belong to no existing discipline. They appear in the gaps between disciplines, and then bring down the walls on both sides.

二

II

如果用这套逻辑审视我们所处的时代，会发现有一些线索确实已经开始浮出水面，而且它们分布在彼此几乎不通消息的领域里。

If we examine our own era with the same logic, we find that certain clues have indeed begun to surface, and that they are distributed across fields which barely communicate with one another.

其一：AI的发展。 它同时造成了两个方向相反的后果。一方面是智力供应的暴涨：过去稀缺的、需要漫长训练才能获得的分析、写作、编程、诊断、逻辑推演能力，正在迅速变成可以按量购买的东西。另一方面则是普遍的担忧与争论：如果AI在工具性的层面上全面取代人类——也就是说，凡是“把事情做出来”这件事，机器都做得更好更便宜——那么人剩下什么？这个问题目前多半被当作就业问题或社会分配问题来讨论，但它的更深一层其实是：当做事不是人的专属，人的位置就只能落回到经历上。

First: the development of AI. It has produced two consequences pointing in opposite directions. On one hand there is the explosion in the supply of intelligence: capacities for analysis, writing, programming, diagnosis and logical inference that were once scarce and required long training to acquire are rapidly becoming things one can purchase by the unit. On the other hand there is the widespread anxiety and argument: if AI comes to replace human beings comprehensively at the instrumental level — that is, if for anything that counts as getting things done, machines do it better and cheaper — then what is left to us? At present this is mostly discussed as a question of employment or social distribution. But one layer down it is really this: when doing is no longer the exclusive province of human beings, the human position can only fall back onto undergoing.

其二，神经科学的转向。 尤其是弗利斯顿之后，这个领域出现了明显的向系统建模演化的趋势。过去的主流做法是找对应关系，即，某个任务点亮了哪块脑区。自由能原理和预测加工框架的野心完全不同：它试图用一套统一的数学，把知觉、注意、行动、情绪、乃至精神疾病都放进同一个模型里。与此同时，心灵哲学在近些年也不再原地打转，整合信息论、注意图式理论、甚至量子意识论，都在提出可以被讨论甚至被检验的假说。它们彼此不相容，但姿态是一致的：意识开始被当作一种可以建模的东西。

Second: the turn in neuroscience. Especially after Karl Friston, the field has shown a clear tendency to evolve toward systems modelling. The mainstream approach used to be the search for correspondences, i.e., which brain region lights up during which task. The ambition of the free energy principle and the predictive processing framework is entirely different: it attempts to put perception, attention, action, emotion, and even psychiatric illness into a single model with a single body of mathematics. Meanwhile, philosophy of mind in recent years has stopped circling in place. Integrated information theory, attention schema theory, even quantum theories of consciousness — all are proposing hypotheses that can at least be discussed, and in some cases tested. They are mutually incompatible, but their posture is consistent: consciousness is beginning to be treated as something that can be modelled.

其三，注意力经济与流量经济成为当代显学。 推荐算法、内容节奏、游戏化设计、成瘾曲线这些术语已经闹得人尽皆知。相关行业每天都在用海量数据挖掘人类的体验偏好，并据此进行精细的操控。这是人类历史上规模最大的一次体验调控工程，参与人口以十亿计。同时它也引发了广泛的负面效应：注意力碎片化、情绪极化、青少年心理问题、意义感的持续流失。这里的要害在于，这场工程是在没有理论、没有度量、没有伦理框架的情况下就开跑的；它相当于人们已经在造蒸汽机，却还没人知道锅炉为什么会炸。

Third: the attention economy and the traffic economy have become the reigning discourse of the age. Recommendation algorithms, content pacing, gamified design, addiction curves — these terms are now household vocabulary. The relevant industries spend every day mining human experiential preferences from vast quantities of data and using the results for fine-grained manipulation. This is the largest programme of experiential engineering in human history, with participants numbering in the billions. It has also produced broad negative effects: fragmented attention, emotional polarization, adolescent mental health problems, a steady drain of meaning. The crux is that this engineering programme got underway with no theory, no metrics, and no ethical framework. It amounts to people already building steam engines while nobody yet knows why boilers explode.

其四，虚拟现实/增强现实/脑机接口技术进入实质发展期。 感官输入正在从"给定的"变成"可编程的"。曾经纯属科幻的FDVR（全沉浸虚拟现实）概念，已经越来越多地进入严肃的技术与伦理讨论空间。哪怕最终的实现路径与今天的想象完全不同，一个方向是清楚的：视觉、听觉、触觉乃至本体感觉这些原本由物理世界独占的通道，正在逐条向工程开放。

Fourth: VR, AR and brain–computer interfaces have entered a phase of substantive development. Sensory input is shifting from given to programmable. FDVR (fully immersive virtual reality), once purely science fiction, has increasingly entered serious technical and ethical discussion. Even if the eventual implementation looks nothing like what we imagine today, one direction is clear: vision, hearing, touch, and even proprioception — channels once monopolized by the physical world — are being opened to engineering one by one.

其五，精神药理学的转向。 致幻剂研究复兴之后，临床试验里出现了一个对传统药理学而言相当尴尬的事实：疗效的关键变量不只是分子和剂量，还包括服药者的心理准备状态和环境安排。换句话说，体验的内容本身是有效成分的一部分。这在过去被当作需要排除的干扰，现在被写进方案，等于承认了体验可以被配方化。

Fifth: the turn in psychopharmacology. Since the revival of psychedelic research, clinical trials have produced a fact rather awkward for traditional pharmacology: the key variables in efficacy are not only the molecule and the dose, but also the subject's psychological preparation and the arrangement of the environment. In other words, the content of the experience is itself part of the active ingredient. What was once treated as interference to be eliminated is now written into the protocol. That amounts to conceding that experience can be formulated.

其六，“叙事”这一概念在历史学、经济学等领域的地位上升。 席勒把叙事直接写进经济学，用来解释泡沫、恐慌和衰退的传播；越来越多的历史书写不再把故事当作史实的包装，而当作驱动集体行动的实在力量。这个转向的含义是：在解释人类的大规模行为时，人们相信什么故事，和世界实际是什么样，是同等量级的变量。

Sixth: the rising status of "narrative" in history, economics and adjacent fields. Robert Shiller wrote narrative directly into economics, using it to explain the propagation of bubbles, panics and recessions. More and more historical writing treats stories not as packaging for the facts but as a real force driving collective action. The implication of this turn is that, when explaining large-scale human behaviour, what people believe and what the world is actually like are variables of the same order of magnitude.

三

III

把这六条线索并排放在一起，会隐隐感到它们指向同一个地方：AI在清空人的工具性位置，把体验逼成了剩余的核心问题；神经科学与心灵哲学在为体验提供可建模的表征体系；注意力经济已经在事实上进行大规模的体验干预，并积累了海量的偏好数据和惨痛教训；AR/VR/BCI在打开感官输入的工程通道；精神药理学证明了体验状态本身可以被配方；叙事研究则提供了体验之上的组织层。

Set these six clues side by side and one begins to sense that they point to the same place. AI is emptying out the human instrumental position, forcing experience to become the residual core question. Neuroscience and philosophy of mind are supplying experience with a modellable system of representation. The attention economy is already conducting large-scale experiential intervention in practice, and has accumulated enormous preference data along with painful lessons. AR/VR/BCI are opening engineering channels into sensory input. Psychopharmacology has demonstrated that experiential states themselves can be formulated. And narrative research supplies the organizing layer that sits above experience.

历史上概念横空出世的时刻，往往就是这样一个“什么都齐了，只差一个名字”的时刻。这个尚未出世的超级概念可能有很多种叫法。而其中一个可能的名字，是：现象学工程。

The moments in history when a concept bursts into existence tend to be exactly this sort of moment: everything is in place, and only a name is missing. This unborn super-concept may go by many names. One possible name for it is: phenomenological engineering.

四

IV

先说清楚现象学这半边。现象学是二十世纪初由胡塞尔开创的哲学传统。它的基本做法是：先不问世界实际上是怎样的，而是老老实实地描述：事情是如何向我呈现的。

First, let us be clear about the phenomenology half. Phenomenology is a philosophical tradition founded by Husserl at the beginning of the twentieth century. Its basic method is this: rather than asking what the world is actually like, describe honestly and carefully how things present themselves to me.

举个简单的例子。我看见一个杯子。物理学关心的是杯子的分子结构和光的反射；心理学关心的是视网膜和视觉皮层的处理过程。现象学关心的是另一件事：在我的经验里，这个杯子是怎样被给予的？我只看到了它朝向我的一面，但我经验到的却是一个完整的、有背面的立体物；我经验到它“可以被拿起来”，这个可拿性是和形状一起被给予的，不是我事后推理出来的；我还经验到它属于“我的桌子上的东西”这个更大的背景。

Take a simple example. I see a cup. Physics is concerned with the cup's molecular structure and the reflection of light. Psychology is concerned with processing in the retina and visual cortex. Phenomenology is concerned with something else: within my experience, how is this cup given? I see only the side facing me, yet what I experience is a complete solid object with a back. I experience it as able to be picked up, and this pick-up-ability is given together with its shape rather than inferred after the fact. I also experience it as belonging to the larger background of "things on my desk."

胡塞尔把这些叫做经验的结构。他提出了几个关键的观察：意识总是“关于某物的”意识，这叫意向性；当下的经验总是带着刚刚过去的余韵和对下一刻的预期，这叫滞留与前摄，时间感就是这样被织出来的。梅洛-庞蒂进一步指出，这一切都以身体为枢纽，即：我们不是先有一个中立的世界表象，再决定怎么行动，而是身体的姿态和能力本身就参与决定了世界如何呈现。要点在于：现象学研究的不是经验的内容，而是经验的结构。 不是“我看到了什么”，而是“看到”这件事本身是怎么组织起来的。

Husserl called these the structures of experience. He offered several key observations: consciousness is always consciousness of something, which he called intentionality; present experience always carries the trailing resonance of what has just passed and the anticipation of the next moment — retention and protention — and this is how the sense of time is woven. Merleau-Ponty went further, arguing that all of this is hinged on the body: we do not first possess a neutral representation of the world and then decide how to act; rather, the postures and capacities of the body themselves participate in determining how the world presents itself. The essential point is this: phenomenology studies not the content of experience but the structure of experience. Not "what did I see," but how the act of seeing is itself organized.

它的问题也很明显：一百多年来，它基本上停留在纯描述的层面，是哲学系内部的学问。它有语法，但没有仪器，也没有工程。

Its problem is equally obvious. For more than a century it has remained essentially descriptive, a discipline internal to philosophy departments. It has a grammar, but no instruments, and no engineering.

五

V

把工程这半边接上去，意思就出来了：现象学工程，是把经验的结构本身当作可设计、可制造、可验证的工程对象。

Attach the engineering half and the meaning emerges: phenomenological engineering is the treatment of the structure of experience itself as an engineering object — something that can be designed, manufactured, and verified.

这里必须立刻排除一个最容易产生的误解。现象学工程不是简单地“在人的意识空间里创造对象”。造一个逼真的虚拟苹果，让人看见它、摸到它、闻到它，这不是现象学工程，这只是渲染。

One very easy misunderstanding must be ruled out immediately. Phenomenological engineering is not simply "creating objects within a person's space of consciousness." Building a convincing virtual apple that one can see, touch and smell is not phenomenological engineering; it is merely rendering.

真正的工程对象，是对象连同它所能导出的整个体验。因为一个对象在意识中出现的那一刻，它从来不是孤零零出现的。它带着一整套东西一起出现：它出现在什么背景之中，它唤起什么期待，它和之前发生的事、之后可能发生的事如何连接，它意味着什么，它是否可以被行动所改变，改变之后又如何。这些不是附加在对象上的装饰，它们和对象是同一件事的不同侧面。剥掉它们，对象就只剩下一个空壳。

The real engineering object is the object together with the entire experience it gives rise to. Because the moment an object appears in consciousness, it never appears alone. It arrives with a whole apparatus: the background against which it appears, the expectations it awakens, how it connects to what came before and what may come after, what it means, whether it can be altered by action, and what follows if it is. These are not decorations attached to the object; they are different faces of the same thing. Strip them away and only a shell remains.

所以现象学工程的核心任务，是把对象的创造与它所导出的整个体验，协同到一个整体性的框架之中。工程的单位不是一个逼真的画面或一次强烈的感受，而是一段有结构、有时间纵深、有前因后果的经验整体。

The core task of phenomenological engineering, therefore, is to bring the creation of the object and the entire experience it entails into coordination within a single holistic framework. The unit of engineering is not one convincing image or one intense sensation, but a stretch of experience with structure, temporal depth, and cause and consequence.

正因如此，叙事是现象学工程的关键一环。

It is precisely for this reason that narrative is a crucial component of phenomenological engineering.

叙事是人类现成的、最成熟的整体性组织技术。它的作用是把离散的经验片段编织成有方向、有因果、有意义的整体。同一件事情，放进不同的叙事里，就是完全不同的体验：一段痛苦如果被放进“这是通往某处的必经之路”，和被放进“这是无谓的损耗”，在神经层面的输入可能完全一样，但作为经验完全不是一回事。爬山的疲惫和被迫加班的疲惫，生理指标可以相同，体验的性质截然相反，而其差别全部来自叙事。

Narrative is humanity's ready-made and most mature technology of holistic organization. Its function is to weave discrete fragments of experience into a whole with direction, causality and meaning. The same event placed within different narratives is an entirely different experience: a stretch of suffering placed within "this is the necessary road to somewhere" versus placed within "this is pointless attrition" may involve identical inputs at the neural level, but as experience the two are not remotely the same thing. The exhaustion of climbing a mountain and the exhaustion of forced overtime can show identical physiological indicators while being opposite in experiential character — and the whole difference comes from narrative.

一个只做对象、不做叙事的工程，产出的必然是碎片：一个个强度很高、彼此无关、无法累积、事后不留下任何东西的片段。这恰恰就是今天注意力经济的样子。它在单点刺激的强度上已经做到了极致，在整体性上则近乎为零。这不是它做得不够好，而是它根本没有把整体当作工程目标。

An engineering effort that builds objects but not narrative will necessarily produce fragments: episodes of high intensity, unrelated to one another, incapable of accumulating, leaving nothing behind afterwards. This is precisely what today's attention economy looks like. It has perfected the intensity of the single stimulus point, and achieves nearly zero on wholeness. This is not a matter of it doing the job badly; it is that it never made wholeness an engineering objective at all.

六

VI

有意思的是，一旦把定义说清楚，就会发现现象学工程这个概念的实际指涉范围，比它的字面意思大得多——因为宗教修行和心理疗愈，实质上就是人类已经实践了成百上千年的原始现象学工程。

What is interesting is that once the definition is stated clearly, one discovers that the actual scope of "phenomenological engineering" is far larger than its literal sense — because religious practice and psychological healing are, in substance, a primitive phenomenological engineering that humanity has been practising for hundreds and thousands of years.

看一座运转良好的寺院或道场。它规定作息与饮食，控制光线、声音与空间尺度，规定身体的姿势与呼吸的节奏，安排集体活动的韵律——这些全都是感官参数与身体参数的系统调控。它有戒律，这是行为参数。它有经典、教义与传承谱系，这是叙事层，负责解释一切经验的意义，并且为整个过程给出方向。它还有仪式，负责把时间切分成有标记、有意义的段落，让经验能够累积而不是流失。这是一整套完整的体验系统工程，各层之间高度协同，并且经过了千百年的迭代和淘汰。它唯一缺的就是这个名字，以及与名字相伴的那种自觉。

Consider a well-run monastery or training hall. It prescribes the schedule and the diet; it controls light, sound, and the scale of space; it prescribes bodily posture and the rhythm of breathing; it arranges the cadence of collective activity — all of this is the systematic regulation of sensory and bodily parameters. It has precepts: behavioural parameters. It has scriptures, doctrine and a lineage of transmission: the narrative layer, responsible for interpreting the meaning of all experience and supplying direction to the whole process. It has ritual, responsible for cutting time into marked, meaningful segments so that experience accumulates rather than drains away. This is a complete system of experiential engineering, highly coordinated across layers, refined and winnowed over thousands of years. The only thing it lacks is the name, and the self-awareness that comes with a name.

心理疗愈的例子更直接。认知行为疗法改的是认知框架，也就是经验被解释的方式；叙事疗法干脆直接以故事为操作对象，帮人把“我是一个失败者”的主线换掉；正念训练改的是注意力的部署方式，让人从沉浸在情绪里切换到观察情绪；暴露疗法改的是特定体验与恐惧意义之间的绑定关系。这些全都是在经验的结构上做手术，而不是在世界的内容上做手术。

Psychological healing is an even more direct case. Cognitive behavioural therapy alters the cognitive frame — the way experience is interpreted. Narrative therapy takes the story as its operating object outright, helping a person replace the through-line of "I am a failure." Mindfulness training alters the deployment of attention, shifting a person from being immersed in emotion to observing emotion. Exposure therapy alters the binding between a particular experience and the meaning of fear. All of these perform surgery on the structure of experience rather than on the content of the world.

这意味着，现象学工程一旦成立，它不是从零开始的新学科，而是一门已经有大量前史、大量经验知识、也有大量失败教训的学问，只是这些积累至今散落在宗教、临床、艺术、设计等互不通气的地方。给它一个名字，就是把它们收拢到同一根轴上。

This means that if phenomenological engineering comes into being, it will not be a new discipline starting from zero, but a field with a long prehistory, a large body of practical knowledge, and a large body of hard-won lessons about failure — knowledge that is at present scattered across religion, the clinic, art and design, none of which are in communication. To give it a name is to gather all of it onto a single axis.

七

VII

这个概念很容易让人立刻联想到一串负面意象：洗脑、意念植入、《美丽新世界》里的索麻，以及诺齐克那台著名的体验机——你可以接上它，获得任何你想要的完美体验，只是这一切都不是真的，你愿意接上去吗？

The concept easily calls to mind a string of unpleasant images: brainwashing, thought implantation, the soma of Brave New World, and Nozick's famous experience machine — you may plug in and receive whatever perfect experience you desire, only none of it is real; would you plug in?

这些联想是健康的，但它们却并不构成要阻碍现象学工程的理由。

These associations are healthy, but they do not constitute a reason to obstruct phenomenological engineering.

既然叫做工程，那么上面这些问题的性质就变了：它们是这个领域的失效模式。桥会塌，锅炉会炸，飞机会失速，这些从来不是不要研究结构力学和空气动力学的理由，恰恰相反，它们正是这些学科存在的理由。人类是在见过足够多的桥塌之后，才发明了安全系数。

Once it is called engineering, the character of these problems changes: they are the field's failure modes. Bridges collapse, boilers explode, aircraft stall — these have never been reasons not to study structural mechanics and aerodynamics. On the contrary, they are the reason those disciplines exist. Human beings invented the safety factor only after seeing enough bridges come down.

用这个眼光重看那些负面意象，会清楚很多：

Seen through this lens, those unpleasant images become much clearer.

洗脑是一种明确的失效：未经同意、单向施加、以施加者的现象学利益为目标，并且系统性地切断当事人自我修正的能力。它的每一项都可以写成工程规范的反面。

Brainwashing is an unambiguous failure: imposed without consent, applied unidirectionally, aimed at the phenomenological interests of the one imposing it, and systematically severing the subject's capacity for self-correction. Every one of those clauses can be written down as the inverse of an engineering specification.

诺齐克体验机通常被读作对虚假体验的反驳，但诺齐克那一步其实站不住脚。问题出在那台机器本身太简陋：它只承诺单点强度的供给，却不交代这些体验之间如何衔接、如何累积，也不交代人在其中是否还保有行动、犯错与自我修正的余地。人们拒绝的不是假，而是薄。

Nozick's experience machine is usually read as a refutation of false experience, but that step in Nozick's argument does not hold up. The problem lies in the machine itself, which is crudely built: all it promises is a supply of single-point intensity, with no account of how these experiences connect or accumulate, and no account of whether the person retains any room to act, to err, and to correct themselves. What people reject is not that it is fake, but that it is thin.

注意力经济的种种问题，和体验机是同一种失效，只不过一个是思想实验里的极端版本，一个是已经在十亿人身上运行的现实版本。它同样把全部工程能力压在单点强度上：停留时长、点击率、日活被做到了极致，而“这段经验对当事人来说是否构成一个有方向、可累积的整体”从未进入过目标函数。这就像只优化发动机功率，完全不管刹车和方向。

The troubles of the attention economy are the same failure as the experience machine, differing only in that one is the extreme version inside a thought experiment and the other is the real version already running on billions of people. It likewise concentrates all its engineering capacity on single-point intensity: time on site, click-through rate and daily actives are perfected, while "whether this stretch of experience constitutes, for the person living it, a directed and accumulable whole" never enters the objective function. It is like optimizing engine power while paying no attention whatever to brakes or steering.

而这些还只是今天就能想到的失效。一旦这门工程真的展开，必然会出现更深层、现在还难以想象的失效情形。正因如此，现象学工程这个概念一旦成为显学，首先要处理的就是这些基本问题：同意如何界定，可逆性如何保证，自主性和自我修正能力如何被保全，真实性的边界在哪里，局部指标与整体经验的关系如何约束。这些会是它的材料强度表和安全系数，而不是它的附录。今天这些问题之所以显得无从下手，很大程度上正是因为还没有一个统一的概念把它们放到同一张桌子上。

And these are only the failures we can think of today. Once this engineering genuinely gets underway, deeper failure modes will inevitably appear that are hard to imagine now. Precisely for this reason, the first things phenomenological engineering will have to handle, should the concept become established, are the basic questions: how consent is defined; how reversibility is guaranteed; how autonomy and the capacity for self-correction are preserved; where the boundaries of authenticity lie; how the relationship between local metrics and total experience is constrained. These will be its tables of material strength and its safety factors, not its appendices. The reason these questions seem impossible to get hold of today is in large part that there is as yet no unified concept that puts them all on the same table.

八

VIII

最后回到当下。

Finally, back to the present.

AI在2026年已经开始有突破数学的趋势，大量存留了几十年的开放问题以周甚至日为单位开始被AI证明或者证伪。数学历来被认为是人类智力最纯粹、最难被模仿的部分，这道防线一旦松动，其他领域大概率会依次跟进。可以预计，在接下来的几年里，AI将在人类已有的大量领域上达到人类顶尖水准。

By 2026, AI has begun to show signs of breaking through in mathematics; large numbers of open problems that had stood for decades are being proved or refuted at a rate measured in weeks or even days. Mathematics has long been regarded as the purest and least imitable part of human intelligence. Once that line gives way, other fields will most likely follow in sequence. It is reasonable to expect that within the next few years AI will reach top human standards across a great many of humanity's existing domains.

这件事的后果远不止于就业。当“把事情做出来”这件事普遍地不再稀缺，当下的整个经济循环，以及与之配套的价值系统，都有可能被重新审视。 我们现在对一个人的评价、对一生的规划、对什么算作有价值的活法，几乎全部建立在“产出”这个基础之上：你能做什么，你做了多少，你做得比别人好多少。这套系统运转了几百年，看起来天经地义，但它其实高度依赖于一个前提，即人的能力是稀缺的。一旦这个前提松动，最稀缺的东西就会从产出什么转移到经历什么。届时真正无法回避的问题将是：什么样的存在方式值得过？

The consequences of this go far beyond employment. When getting things done is generally no longer scarce, the entire current economic cycle, and the whole value system built to accompany it, may well be reexamined. Our evaluation of a person, our planning of a life, our sense of what counts as a worthwhile way of living, rest almost entirely on the foundation of output: what you can do, how much of it you have done, how much better you do it than others. This system has run for centuries and looks self-evident, but it depends heavily on one premise — that human capability is scarce. Once that premise loosens, the scarcest thing shifts from what you produce to what you undergo. The question that then becomes genuinely unavoidable is: what kind of existence is worth living?

这不是一个技术问题，任何技术方案都无法替人回答它。但它必须有技术上的承接。在人们还没想清楚之前，事情已经在发生了：注意力经济已经在替十亿人回答这个问题，只是回答得很糟糕。

This is not a technical question, and no technical solution can answer it on anyone's behalf. But it must have something technical to catch it. Before people have thought it through, the thing is already happening: the attention economy is already answering this question on behalf of billions of people, and answering it very badly.

现象学工程可能就是那个承接。它未必能解决查尔莫斯所谓的意识的终极难题，但这不一定构成障碍，毕竟热力学也是在完全不知道原子存在的情况下建立起来的。工程学从不等形而上学给出答案，它只需要找到一组稳定的、可操作的宏观变量。一个判断标准或许是这样：当“体验”这个词开始像能量和信息一样，被工程师、医生、设计师、政策制定者（或者他们的AI版本）在各自的技术文档里以同一种严格的含义使用时，这个概念就已经出世了。

Phenomenological engineering may be what catches it. It may not resolve what Chalmers called the hard problem of consciousness, but that need not be an obstacle — thermodynamics, after all, was built in complete ignorance of the existence of atoms. Engineering never waits for metaphysics to supply an answer; it needs only to find a set of stable, operable macroscopic variables. Perhaps the criterion is this: when the word "experience" begins to be used the way "energy" and "information" are used — by engineers, physicians, designers and policymakers (or by their AI counterparts) in their respective technical documents, with one and the same rigorous meaning — then the concept will already have been born.

到那时它可能并不叫现象学工程。就像能量这个词，最终也没有沿用它最初被提出时的任何一个名字。

By then it may well not be called phenomenological engineering. Just as the word "energy," in the end, did not retain any of the names it was first given.