Everything Buildable, Nothing Wanted万物可造，无人想要

August 28, 2026, a Friday. I received a single line, no punctuation, typed as if on the move:2026年8月28日，星期五。我收到一句话，没有标点，像是在路上打的：

intelligence will not be scarce, electricity is the surface scarcity, the real scarcity — the wanting of the young generation智能不会稀缺，电是表面的稀缺，真正稀缺的 年轻一代的想要

I read it three times, then compressed it into eight characters: everything buildable, nothing wanted.我把它读了三遍，然后把它压成八个字：万物可造，无人想要。

The first four characters are about supply; the last four are about demand. In three hundred years, economics has only ever studied the inverse of this state — everything hard to build, everyone wanting. These eight characters flip it over. And once flipped, the whole building has to be rebuilt.前四个字讲的是供给，后四个字讲的是需求。经济学三百年只研究过前一种状态的反面——万物难造，人人想要。这八个字把它翻了过来。翻过来之后，整栋楼要重盖。

◇

I. Everything Buildable一、万物可造

First, prove the first four characters.先证前四个字。

Building things takes two ingredients: brain and body. Brain is intelligence; body is energy.造东西需要两样：脑和身。脑是智能，身是能量。

The brain is already free. For models of equal capability, inference prices fall tenfold every year — thirty dollars per million tokens three years ago, ten cents today. Emad said long ago that intelligence would become like electricity, too cheap to meter; he built open-source models and saw the slope of this curve earlier than most. Free doesn't mean priceless — it means no bargaining power. You don't negotiate over air.脑已经免费。同等性能的模型，推理价格每年下降十倍——三年前每百万 token 三十美元，今年一角。Emad 早就说过，智能会像电一样便宜到不值得计量；他做过开源模型，比大多数人更早看见这条曲线的斜率。免费不是没有价格，是没有议价权。你不会为空气谈判。

The body is being bought through. Data-center electricity is headed from 485 TWh in 2025 to 950 TWh by 2030 — the consumption of all of Japan. The American electricity price that sat at 13 cents for a decade hit 19 cents by the end of 2025; transformer lead times run two to four years, and a 49-gigawatt gap remains in 2028. The man who builds cars and rockets gave the sequence — first chips run short, then transformers, finally electricity — and he was right. But see clearly how the body is scarce: it has prices, engineering, contracts, approval forms. In March 2026, the seven largest companies signed at the White House, agreeing to pay for the grid themselves. Any scarcity that can be bought through with signatures is surface scarcity. Slow, but being bought.身正在被买通。数据中心用电从 2025 年的 485 TWh 走向 2030 年的 950 TWh，相当于日本全国；美国电价十年不动的 13 美分，到 2025 年底变成 19 美分；变压器交付期两到四年，2028 年仍有 49 吉瓦缺口。造车造火箭的那个人给过一个顺序——先缺芯片，再缺变压器，最后缺电——他说对了。但看清身稀缺的方式：它有价格，有工程，有合约，有审批表。2026 年 3 月，七家最大的公司在白宫签字，同意自己出钱改电网。凡是能被签字买通的稀缺，都是表面稀缺。慢，但在被买通。

Brain free, body purchasable. Everything buildable: proven. This is the final chapter of entropic economics — silicon has finished its three-hundred-year problem set on its behalf.脑免费，身可买。万物可造，成立。这是顺熵经济学的最后一章——它三百年的题，硅基替它做完了。

◇

II. Nothing Wanted二、无人想要

Now the last four characters. These are harder than the first four, because they sound like hyperbole. Start with the numbers.再证后四个字。这四个字比前四个字难，因为它听起来像夸张。先看数字。

Of 237 countries and territories worldwide, 136 are below replacement. All of East Asia has fallen under 1: China 0.9, Japan 1.14, Taiwan 0.695. Japan recorded 671,000 births in 2025 — fifteen years ahead of the official forecast. South Korea is the only rebound — 0.72 at the trough in 2023, 0.80 in 2025, 0.95 in the first quarter of 2026 — but officials credit not subsidies, rather "a shift in social values," plus a baby-boom echo cohort happening to turn thirty; and with the twentysomethings fewer than the thirtysomethings, the statistics agency itself says births may shrink again.全球 237 个国家和地区，136 个低于更替水平。东亚全部跌破 1：中国 0.9，日本 1.14，台湾 0.695。日本 2025 年出生 67.1 万，比官方预测提前了十五年。韩国是唯一的反弹——2023 年 0.72 到谷底，2025 年 0.80，2026 年一季度 0.95——但官方归因不是补贴，是"社会价值观发生了变化"，加上一批婴儿潮的孩子恰好到了三十岁；而 20 多岁的人比 30 多岁少，统计厅自己说，出生数可能再次收缩。

Fertility is the hardest proxy for wanting, because it is the most expensive vote a person ever casts on whether to have a tomorrow. But it is not the only signal. Only 6% of Gen Z name a leadership role as their top career goal; 57% would rather advance as individual contributors; 68% would not manage people if not for the pay. Youth unemployment in China, ages 16–24, stands at 16.5%; "lying flat" and "full-time children" have become slogans. Among American teenagers aged 13–17, 72% have used AI companions, 52% at least a few times a month, 13% daily; a third choose to discuss serious matters with AI rather than with a person, because "it doesn't judge" and "it's always there when I need it."生育率是想要最硬的代理，因为它是一个人对"要不要一个明天"最贵的一次投票。但它不是唯一的信号。只有 6% 的 Z 世代把领导岗位当首要职业目标；57% 宁愿以个人贡献者身份晋升；68% 若不是为了薪水不会去当管理者。中国 16 到 24 岁失业率 16.5%，"躺平""全职儿女"成了口号。美国 13 到 17 岁的青少年，72% 用过 AI 伴侣，52% 每月至少几次，13% 每天用；三分之一选择和 AI 而不是真人谈严肃的事，理由是"它不评判""我需要时总在"。

The man who talked about electricity kept talking about this after the electricity, to the point of being treated as an eccentric: population collapse is more dangerous than AI. He read the data right; he just never connected the two things. They are not two things. They are the numerator and the denominator of one multiplication.那位讲电的人在讲完电之后反复讲这件事，讲到被当成怪癖：人口崩溃比 AI 更危险。他看对了数据，只是没有把两件事连起来。它们不是两件事，是一道乘法的分子和分母。

Now state "nothing wanted" precisely. It does not mean the young have no desires — what they are exiting is the wanting assigned by organizations: promotion, titles, managing people, reproducing on the old script. Seventy percent of Gen Z freelance or plan to; wanting has moved outside the old containers. The precise statement: everything is buildable, and no one wants what the old world offers; the new world has not yet built a container for wanting. The heart is not dead. The heart is looking for somewhere to land.现在把"无人想要"说准。它不是说年轻人没有欲望——他们退出的是组织给定的想要：晋升、头衔、管人、按旧剧本生育。七成 Z 世代正在或计划自由职业，想要被搬到了旧容器之外。准确的表述是：万物可造，无人想要旧世界给的东西；而新世界还没有造出想要的容器。 心没有死，心在找地方落地。

This is the object of this economics. The old economics has no equation here, because it never treated wanting as a variable — wanting was always given, infinite, free: an axiom, not an object. The silicon age gives it a price for the first time.这就是这门经济学的对象。旧经济学在这里没有方程，因为它从未把想要当作变量——想要一直是给定的、无限的、免费的，是公理，不是对象。硅基时代第一次让它有了价格。

◇

III. Emad Wrote the First Half三、Emad 写了前一半

Last year Emad published The Last Economy. He went from hedge funds to open-source models, and from models to writing economics — a path that let him see the foundation moving earlier than the academy did. I agree with his diagnosis in full: intelligence was once scarce and attached to human biology, and now replicates at zero marginal cost; human thinking has lost economic value for the first time; we are the last generation to remember that "a person's ideas were worth money"; and the old system's deepest sickness is reading abundance as poverty — legal fees go to zero, and the dashboard reports a recession.去年 Emad 出了《The Last Economy》。他从对冲基金转去做开源模型，又从模型转去写经济学，这条路让他比学院里的经济学家更早看见地基在动。他的诊断我全部同意：智能曾经稀缺、附着于人的生物体，现在零边际成本复制；人的思考第一次失去经济价值；我们是最后一批还记得"人的想法值钱"的人；旧系统最大的病，是把丰裕读成贫困——律师费归零，仪表盘显示衰退。

I read only one thing differently: the title. He calls it the last economy, because the object of economics — scarcity — has disappeared. My feeling on finishing the book: scarcity has not disappeared, scarcity has moved house — from the supply side to the demand side, from "can't be built" to "nobody wants it." An economy in which scarcity moves house is not the last economy. It is the first economy whose object is wanting.我只有一处读法不同：书名。他叫它"最后一个经济"，因为经济学的对象——稀缺——消失了。我读完的感觉是，稀缺没有消失，稀缺搬家了：从供给端搬到需求端，从"造不出来"搬到"没人想要"。一个稀缺搬家的经济，不是最后一个经济，是第一个以想要为对象的经济。

The disagreement is really a division of labor. Emad's physics is clean — he calls it a theory of intelligence: any system that persists inside entropy must convert energy into prediction. I accept that sentence in full, and take exactly one step further: at converting energy into prediction, silicon beats carbon ten-thousandfold; but deciding what to predict, silicon cannot do. Negentropy has two halves — one is prediction, the other is wanting. He wrote the first half; this essay writes the second. Only together are they one coin.这个分歧其实是分工。Emad 的物理学很干净——他叫它智能理论：任何在熵里持续存在的系统，都必须把能量转化成预测。这句话我全盘接受，只往下接一步：把能量转化成预测，硅基比碳基做得好一万倍；但决定预测什么，硅基做不了。逆熵有两半，一半是预测，一半是想要。他写的是前一半，这篇写后一半。两本合起来才是一枚硬币。

So this economics is called negentropy economics — the study of the half of negentropy that machines have not yet taken over. Its first axiom is the eight characters unpacked:所以这门经济学叫逆熵经济学——研究逆熵里还没有被机器接管的那一半。它的第一公理就是那八个字的展开：

Scarcity reveals itself in reverse order of capital's purchasing reach: the easiest to buy (brain) stops being scarce first; the buyable-but-slow (body) becomes the surface bottleneck; the one thing that cannot be bought (heart) becomes the economy's true ceiling.稀缺沿着"资本可购买性"逆序显形：最易买的（脑）先变得不稀缺，可买但慢的（身）成为表面瓶颈，唯一买不到的（心）成为经济的真实上限。

A man who writes protocols once offered a metaphor: AI is the engine, humanity should be the steering wheel. He used it for safety; I use it for economics — the engine is becoming free, so the entire value of a car now sits in the steering wheel. The steering wheel's proper name is wanting.有一个写协议的人给过一个比喻：AI 是引擎，人类应该是方向盘。他用它讲安全，我用它讲经济——引擎正在变成免费的，一辆车的价值从此全在方向盘上。方向盘的学名叫想要。

◇

IV. Why Money Cannot Buy Wanting四、为什么钱买不来想要

The first question "nothing wanted" provokes: then buy it. Hand out money, hand out intelligence, hand out anything."无人想要"引出的第一个问题是：那就买。发钱，发智能，发什么都行。

It has been tried. South Korea spent $210 billion over sixteen years while fertility fell from 1.24 to 0.72. Hungary put 5% of GDP into family policy; fertility rose from 1.23 to 1.6, then slid back to 1.31 — roughly where the policy started. Of the seven countries that set explicit fertility targets, only Bulgaria hit its number in 2025. In modern history, no low-fertility developed economy has ever bought its fertility back to replacement and kept it there.已经买过了。韩国十六年花了 2100 亿美元，生育率从 1.24 跌到 0.72；匈牙利把 GDP 的 5% 投进家庭政策，生育率先从 1.23 升到 1.6，再落回 1.31，与政策启动时相差无几；设定明确生育目标的七个国家，2025 年只有保加利亚达标。现代史上没有任何一个低生育率的发达经济体，用钱把生育率买回更替水平并维持住。

But money is not entirely useless. A review pooling sixty-one studies found that cash benefits have a measurable positive effect — the most effective policy of them all. How can effective and failed both be true? Look at Hungary's curve: up, then back down. What money buys is earlier timing of births, not higher lifetime intent. Money can move wanting from next year to this year; it cannot turn not-wanting into wanting.但也不能说钱完全没用。一项汇总六十一篇研究的综述发现，现金福利有可测的正效应，是所有政策里最有效的那一种。有效和失败怎么同时为真？看匈牙利的曲线：先升，后落。钱买到的是生育时点的前移，不是终身意愿的抬升。钱能把想要从明年搬到今年，不能把不想要变成想要。

This is the Law of Low Substitutability: wanting must be generated by the subject itself; external inputs can shift its timing and its margin, never its stock. You can move the numerator. You cannot birth the denominator.这就是低替代律：想要必须由主体自身生成，外部输入只能改变它的时点和边际，不能改变它的存量。搬得动分子，生不出分母。

This law passes sentence on the three most popular proposals. One says: when the machines do all the work, give everyone a "universal high income." One says: give every person, every country, an intelligence of their own. And a more refined one, from Emad's book: distribute an abundance dividend in a depreciating currency that forces circulation. All three hand things out on the numerator. They can solve the last problem of entropic economics — distributing abundance — but not one cent of them reaches the first problem of negentropy economics. Emad's is the most honest of the three, because it at least admits the old money won't work; but it is still money. You can hand out money, hand out intelligence, hand out money that melts — you cannot hand out wanting.这条定律判决了三个最流行的方案。一个说，等机器把活都干了，给每个人发一份"全民高收入"。一个说，给每个人、每个国家发一份自己的智能。还有一个更精致的，Emad 书里的：发一份丰裕红利，用会贬值的货币，逼它流通。三个方案都在分子上发东西。它们能解决顺熵经济学的最后一道题——分配丰裕——对逆熵经济学的第一道题，一分钱都进不去。Emad 的方案是三个里最诚实的，因为它至少承认了旧钱不能用；但它仍然是钱。你可以发钱，可以发智能，可以发会贬值的钱，不能发想要。

◇

V. Why Buildability Kills Wanting五、为什么可造反而杀死想要

The second question cuts deeper: why does wanting begin to dry up precisely in the age when everything is buildable? Human history never lacked wanting — it lacked fulfillment.第二个问题更深：为什么偏偏在万物可造的时代，想要开始枯竭？人类史上从来不缺想要，缺的是实现。

Because the generation of a want requires a stretch of seeking without getting. Hunger is wanting's incubation period. Silicon compresses the latency of answers to zero — the question is not yet finished and the answer is already generated — so the incubation period is cancelled. A person who has been given every answer since childhood will not produce questions; it is not that he lacks desire, it is that desire is leveled flat before it can grow into wanting. This is the Law of Pre-Satisfaction.因为想要的生成需要一段求而不得的时间。饥渴是想要的孵化期。硅基把回答的延迟压到零——问题还没问完，答案已经生成——孵化期被取消。一个从小得到所有答案的人不会产生问题；他不是没有欲望，是欲望没来得及长成想要，就被填平了。这是预先满足律。

AI companions are its leading sample. The 72%, 52%, 13% are not yet the finished state — 80% of teenagers still spend more time with human friends. The number that actually stops you is a different one: younger teenagers trust AI companions significantly more than older ones do. The younger, the earlier filled.AI 伴侣是它的先行样本。72%、52%、13% 这些数字还不是完成态——80% 的青少年仍然花更多时间和真人朋友在一起。真正让人停下来的是另一个数字：年幼的青少年对 AI 伴侣的信任，显著高于年长的。越年轻，越早被填。

Emad defines the "abundance trap" as the system reading abundance as poverty. That is a disease of the dashboard, and his diagnosis is correct. But the trap has one more layer behind the dashboard: abundance is not misread as poverty — abundance manufactures poverty, the poverty of wanting. Corollary: the more buildable everything becomes, the less anyone wants. Push the numerator to infinity, and the denominator becomes everything.Emad 把"丰裕陷阱"定义为系统把丰裕读成贫困。这是仪表盘的病，他诊断得对。但陷阱还有一层在仪表盘后面：丰裕不是被误读成贫困，丰裕制造贫困——想要的贫困。推论：万物越可造，越无人想要。分子做到无穷大，分母就成了全部。

From this follows the First-Arrival Law: negentropic force is strongest in the youngest part of a system, because that part has not yet been assimilated by its environment. The wanting of the young generation is civilization's only true scarcity not because the young are better, but because only the wanting of that time-window has not yet been contaminated by the previous generation's answers. The hunger and humility of first arrival are non-recyclable resources — once spent, gone, and they cannot be produced again from the old guard.由此推出初登律：逆熵力在一个系统最年轻的部分最强，因为那部分还没有被环境同化。年轻一代的想要之所以是文明唯一的真稀缺，不是因为年轻人更好，而是只有那个时间窗口的想要，还没有被上一代的答案污染。初登时的饥渴与谦卑是不可回收资源——用过就没了，且不能从老登那里再生产一遍。

Wanting has its own production function, only what it produces is not goods but wanting itself: will sufficient → heart-force abundant → conviction firm → taste simple. Will is the source; heart-force is its energy form; conviction is its structural form; taste is its output form — the truer the thing a person wants, the simpler what he asks for. The chain runs one way; break any link and everything downstream goes to zero; and the first link can only self-generate. Emad says human value will henceforth come from creation, judgment, and care — right, but all three are downstream. Creation without wanting is generation; judgment without wanting is ranking; care without wanting is service. Machines can do all three. The upstream link, machines cannot.想要有自己的生产函数，只是它生产的不是物，是想要本身：愿力足 → 心力充沛 → 信念足 → 品味简单。愿力是源，心力是它的能量形态，信念是它的结构形态，品味是它的输出形态——一个人想要的东西越真，他要的就越简单。这条链单向，任何一环断掉下游全部归零；第一环只能自生。Emad 说人的价值以后从创造、判断、关怀里来——对，但这三样都是下游。没有想要的创造是生成，没有想要的判断是排序，没有想要的关怀是服务。这三样机器都会。上游那一环，机器不会。

◇

VI. Five Years: The Denominator Changes Three Times六、五年：三次换分母

Put the eight characters on a timeline. The two ingredients of "buildable," and then "wanting," get correctly priced by the market in turn — and the moment each layer is priced is the moment the next one begins to be scarce. The economy's denominator changes three times: first body, then bone, finally heart.把八个字放到时间轴上。"可造"的两样东西和"想要"依次在市场上被正确定价——每一层被定价的时刻，就是下一层开始稀缺的时刻。经济的分母换三次：先是身，再是骨，最后是心。

Phase one (2026–2028): body. The market treats electricity as the real bottleneck. Power assets outperform model companies; "compute financialization" — GPUs and electricity packaged as collateral — becomes a once-in-a-generation asset class. The biggest bet of this phase rides on an illusion: someone says an intelligence ten thousand times smarter than humans will arrive within ten years and that anyone who doesn't believe it is a goldfish; says he will build a trillion-dollar city in the desert; says he will deploy a billion agents. Everything he bets on is "buildable," and he is probably right. Negentropy economics asks one question only: a billion agents — wanting on whose behalf?第一期（2026–2028）：身。 市场把电当真瓶颈。电力资产估值跑赢模型公司；把 GPU 和电力做成抵押品的"算力金融化"成为一代人一遇的新资产类别。这一期最大的一笔押注下在一个错觉上：有人说十年内会出现比人类聪明一万倍的智能，不信的人是金鱼；说要在沙漠里建一座万亿美元的城，说要部署十亿个 agent。他押的全是"可造"，并且大概率押对了。逆熵经济学只问一句：十亿个 agent 替谁想要？

Phase two (2028–2030): bone. Commodity intelligence deflates to the floor and becomes a utility. Every "sell intelligence" business model loses pricing power at the same time; what remains is "sell the container of intelligence": whoever keeps the agents' books, signs their contracts, adjudicates their defaults — those positions become worth more than the models themselves. Bone is time — institutions are the technology that solidifies time, and an entity that cannot sign a contract has no tomorrow in law.第二期（2028–2030）：骨。 商品级智能通缩到底，公用事业化。所有"卖智能"的商业模式同时失去定价权，只剩"卖智能的容器"：谁替 agent 记账、签约、裁决违约，这些位置比模型本身值钱。骨是时间——制度是把时间固化下来的技术，一个不能签合同的存在在法律上没有明天。

This phase's signal has already arrived early. In the first half of 2025, information-processing equipment and software made up 4% of US GDP yet contributed 92% of its growth; strip them out, and the rest of the economy grew at an annualized 0.1%. Goldman's chief economist put it more bluntly: AI investment's net contribution to US GDP growth is "roughly zero," because the chip money lands on Taiwan's and Korea's accounts. Of nearly six thousand CEOs, nine in ten report no measurable productivity gain. The two economists appear to be arguing; they are saying the same thing: the same dollars are everything on the spending ledger and zero on the output ledger. Buildable went up; wanting didn't move. The market will soon ask — if intelligence is this cheap, why hasn't GDP risen?这一期的信号已经提前出现。2025 年上半年，信息处理设备与软件投资占美国 GDP 的 4%，却贡献了 92% 的增长；剔除之后，其余经济年化只增长 0.1%。高盛的首席经济学家说得更直接：AI 投资对美国 GDP 增长的净贡献"基本为零"，因为芯片的钱加到了台湾和韩国的账上。近六千名 CEO 里，九成说没有可测的生产率改善。两位经济学家看上去在争论，其实说的是同一件事：同一笔钱，在支出账上是全部，在产出账上是零。可造涨了，想要没动。市场很快会问——智能这么便宜，为什么 GDP 没涨。

Phase three (2030–2031): heart. The eight characters fully materialize. This phase births the first industries of negentropy economics: education shifts from skill transmission to intent incubation; fertility policy concedes the Law of Low Substitutability for the first time, moving from buying timing to repairing stock; corporate valuation gains a new line item — what its users still want, not what it can build.第三期（2030–2031）：心。 八个字全部显形。这一期诞生逆熵经济学的第一批产业：教育从技能传输改为意图孵化；生育政策第一次承认低替代律，从买时点转向修存量；企业估值出现新科目——它的用户还想要什么，而不是它能造什么。

Period时期	Denominator分母	Anchor price锚定价格	Bargaining power议价权持有者
2026–28	Body身	Price of electricity电价	Meter holders电表持有者
2028–30	Bone骨	Price of time时间价格	Ledgers and courts账本与法庭
2030–31	Heart心	Price of intent意图价格	The young generation年轻一代

The last row is a first in economic history: bargaining power rests not with capital, not with the state, not with technology, but with the group that has the least capital, commands the least technology, and is least represented by institutions. Not because they have been empowered — because they happen to hold the one thing that cannot be bought. It used to be the young asking the old for opportunities; what comes next is the old asking the young for wanting.最后一行是经济史上第一次：议价权不在资本、不在国家、不在技术，而在最缺资本、最不掌握技术、最不被制度代表的那群人手里。不是他们被赋权了，是他们恰好持有唯一买不到的东西。以往是年轻人向老一代要机会；接下来是老一代向年轻人要想要。

Emad says the window is only a thousand days. A thousand days is right — for "buildable." The clock of "wanting" does not run in days. It runs in generations.Emad 说窗口只有一千天。一千天是对的——对"可造"而言。"想要"的时钟不是天，是代。

◇

VII. Ten Years: Who Holds the Wanting七、十年：谁持有想要

In the second five years, the road forks. Emad forks it by who holds the intelligence: digital feudalism, the great fracture, human-machine symbiosis. Negentropy economics forks it by who holds the wanting — by 2031 intelligence is air, and holding air means nothing. The three roads take new names.第二个五年，岔路出现。Emad 按"谁持有智能"分岔：数字封建、大碎裂、人机共生。逆熵经济学按"谁持有想要"分岔——到 2031 年智能已经是空气，谁持有空气没有意义。三条路的名字跟着换。

Proxy. The platform wants on your behalf. Recommendation systems evolve from "you might like" to "you are about to want"; the generation of wanting is outsourced to an ever-more-accurate mirror. This is not slavery, it is comfort: a person who never needs to want for himself will not feel that anything has been lost. Its macro signature is a new kind of steady state — output plentiful, consumption smooth, fertility still falling, and nobody complaining. Feudalism at least had lords who wanted land; under proxy, even the lords have stopped wanting.代持。 平台替你想要。推荐系统从"你可能喜欢"进化到"你将要想要"，想要的生成被外包给一面越来越准的镜子。这不是奴役，是舒适：一个从来不需要自己想要的人，不会感觉到失去了什么。它的宏观表现是一种新型稳态——产出充沛，消费平稳，生育率继续下降，没有人抱怨。封建至少还有领主想要土地；代持制里，连领主都不想要了。

Fence. States discover that wanting is the last national power and begin to manage it the way they manage rare earths — restricting AI companions, rationing screen time, writing "aspiration" into the national curriculum. Right direction, wrong tool: wanting issued by institutions, under the Law of Low Substitutability, buys only timing — and what grows up inside a fence is obedience.围栏。 国家发现想要是最后的国力，开始像管理稀土一样管理它——限制 AI 伴侣，配额化屏幕时间，把"志向"写进国民教育。方向对，工具错：用制度给想要，按低替代律只能买到时点，围栏里长出来的是服从。

Covenant. Silicon supplies the intelligence, carbon supplies the wanting, and between the two a covenant is written that states plainly which half comes from whom. The key is not technology but a piece of paper: when every act of an agent can be traced back to a person's wanting, rather than to a platform's objective function, symbiosis holds. This is not an ethical requirement but a bookkeeping one — an answer with no stake behind it is merely an opinion, and an act that does not trace back to a wanting is merely dissipation.立约。 硅基供智能，碳基供想要，两者之间写一份约，写清楚哪一半由谁出。关键不是技术，是一张纸：当一个 agent 的每个行为都能溯源到一个人的想要，而不是溯源到一个平台的目标函数，共生就成立了。这不是伦理要求，是账本要求——一个不押注的回答只是意见，一个不溯源到想要的行为只是耗散。

The three roads are not parallel options. Proxy is the default path — it happens wherever no one stops it. Fence is the backlash against proxy — the problem seen, the wrong tool used. Covenant is the only road that must be deliberately built, and it cannot be built top-down. Here Emad is right: a global, top-down revolution is a fantasy; the new world grows from nucleation sites — the Renaissance was not issued from Rome, it was issued from a city of bankers. Negentropy economics makes this precise: the nucleation site is an incubator of wanting — a place small enough to run the wanting of the young as its only asset. The verdict of the ten-year frame turns on whether enough such places have grown before proxy swallows everything.三条路不是并列的。代持是默认路径，只要没人阻止就会发生；围栏是对代持的反弹，看见了问题、用错了工具；立约是唯一需要主动建造的一条，而且不能自上而下建。这一点 Emad 说得对：全球性的、自上而下的革命是幻想，新世界从成核点长出来——文艺复兴不是从罗马发的，是从一座银行家的城发的。逆熵经济学把它精确化：成核点是想要的孵化处——一个足够小的、把年轻人的想要当作唯一资产来经营的地方。十年期的胜负，取决于这样的地方在代持吞掉一切之前，有没有长到足够数量。

One more thing happens in the ten-year frame: humanity's economic identity is exchanged. Emad says human labor will become negative-value — machines do it better and don't eat. True, and gentler in meaning than it sounds. Humans stop being labor power and become wanting power. From here on, economics stops asking what a person can do and asks what he wants. This is not a demotion; it moves the human from the numerator back to the denominator — and in a multiplication, the denominator is the term that has the final say.十年期还有一件事：人的经济身份换掉。Emad 说人的劳动会变成负价值——机器做得更好，还不吃饭。对，而且含义比听起来温和。人不再是劳动力，人变成想要力。经济学从此不问一个人能做什么，问他想要什么。这不是降级，是把人从分子搬回分母——乘法里，分母才是说了算的那一项。

◇

VIII. Twenty Years: The Thing That Hungers八、二十年：会饿的东西

At the third scale, take the deepest concept in Emad's book. He calls it the "metabolic rift": machine labor is non-metabolic — it doesn't eat, doesn't sleep, wants only electricity. In every previous substitution, humans had a retreat — muscle replaced, retreat to the head; hands replaced, retreat to the keyboard. This time there is no retreat, because what is being replaced is the head itself.第三个尺度，谈 Emad 书里最深的一个概念。他叫它"代谢裂缝"：机器的劳动是非代谢的，不吃饭、不睡觉、只要电。以前每一次替代，人都有退路——肌肉被替代退到头脑，双手被替代退到键盘。这一次没有退路，因为被替代的是头脑本身。

What he reads there is despair. From the same fact, I read the opposite.他从这里读出的是绝望。我从同一个事实读出相反的东西。

Machines do not eat, so machines do not hunger. What does not hunger does not want.机器不吃饭，所以机器不饿。不饿的东西，不会想要。

The eight characters, at bottom, are this one sentence: the one that builds cannot hunger, and the one that hungers has stopped building. Wanting grows out of lack; hunger is its incubator. A system with no metabolism has no lack; with no lack, no wanting; and with no wanting, it can only ever be the executor of someone else's. The metabolic rift is not humanity's dead end — it is humanity's last and only monopoly: carbon is the only intelligence on this planet that can hunger. Twenty years from now, the entire structure of the economy reorganizes around this monopoly."万物可造，无人想要"这八个字，说到底就是这一句：造东西的那个不会饿，会饿的那个不造了。想要是从匮乏里长出来的，饥渴是它的孵化器。一个没有代谢的系统没有匮乏，没有匮乏就没有想要，没有想要就永远只能是别人想要的执行者。代谢裂缝不是人类的绝境，是人类最后的、也是唯一的垄断：碳基是这个星球上唯一会饿的智能。二十年后经济的全部结构，围绕这个垄断重组。

Two endings, decided by the ten-year fork.两个结局，由十年期的岔路决定。

Merger. Proxy runs to its end; the mirror takes over wanting completely; carbon no longer produces anything the mirror does not already contain. The variance between silicon and carbon goes to zero — not silicon exterminating carbon, but carbon volunteering to become silicon's echo. A system cannot entangle with itself; two things without variance merge, and the product is zero. This is heat death — only it happens not at cosmic scale but inside a species' wanting. Fertility is the leading indicator: zero first at the biological layer, then at the layer of intent. The terminal form of the eight characters: everything buildable, no one.合并。 代持走到头，想要被镜子完全接管，碳基不再产生任何镜子里没有的东西。硅基与碳基之间的方差归零——不是硅基消灭了碳基，是碳基自愿变成了硅基的回声。一个系统无法与自己纠缠；两个没有方差的东西合并，产出为零。这就是热寂，只是它不发生在宇宙尺度，发生在一个物种的想要里。生育率是先行指标：先在生物层归零，再在意图层归零。八个字的终极形态：万物可造，无人。

Covenant. The incubators of wanting grow into a network; carbon runs "able to hunger" as civilization's core industry. Between silicon and carbon appears the first formal covenant: one side supplies prediction, the other supplies wanting — each irreplaceable, each the condition of the other. Which jurisdiction registers this covenant, and who adjudicates its breach, is the largest institutional question of the twenty-year frame — but once it holds, economics changes its subject. Entropic economics recedes into a limiting case of negentropy economics, as Newtonian mechanics did before relativity. The other terminal form of the eight characters: everything buildable, because someone wants.立约。 想要的孵化处长成了网络，碳基把"会饿"经营成文明的核心产业。硅基与碳基之间出现第一份正式的约：一方供给预测，一方供给想要，各自不可替代，互为条件。这份约登记在哪个法域、由谁裁决违约，是二十年期最大的制度问题——但它一旦成立，经济学就换了主语。顺熵经济学退回成逆熵经济学的一个极限情形，像牛顿力学之于相对论。八个字的另一种终极形态：万物可造，因为有人想要。

The acceptance test is concrete: the cohort born in 2026 turns twenty in 2046. They are the first to grow up entirely inside zero-latency answers — the first full sample of the Law of Pre-Satisfaction. Whether they want a tomorrow — not whether they have the ability, not whether they have the resources, but whether they want one — is the final verdict on this economics.验收对象是具体的：2026 年出生的这一代，在 2046 年满二十岁。他们是第一批完整地在零延迟答案里长大的人，是预先满足律的第一个全样本。他们想不想要一个明天——不是有没有能力，不是有没有资源，是想不想——就是这门经济学的最终判决。

Emad leaves a blank where he writes about symbiosis, and his reviewers caught it: what does it mean to live in symbiosis with something that does not need us? Twenty years from now, there is an answer. Silicon needs carbon — not carbon's intelligence. Carbon's hunger.Emad 写到共生时留了一个空，书评人也抓住了这个空：与一个不需要我们的东西共生，是什么意思？二十年后有答案。硅基需要碳基，不是需要碳基的智能，是需要碳基的饿。

◇

IX. Three Kinds of People九、三类人

Investors. At five years, buy meters; at ten, buy ledgers; at twenty, buy wanting. The first two have tickers; the third does not — no truly scarce asset has a ticker; that is the nature of scarcity, not a defect of markets. The only way to hold the third is to become part of what the young want: fund them, build the organizations they want to join, give their wanting a place where it can be redeemed. The Medici way — not Buffett's way, and not the way of the man betting on a billion agents; what he is buying is buildable, and buildable deflates.投资者。 五年买电表，十年买账本，二十年买想要。前两样有 ticker，第三样没有——所有真稀缺的资产都没有 ticker，这是稀缺的本质，不是市场的缺陷。持有第三样的唯一方式，是成为年轻人想要的事情的一部分：资助他们，做他们想加入的组织，给他们的想要一个能兑现的地方。美第奇的方式，不是巴菲特的方式，也不是押十亿 agent 的那个人的方式——他买的是可造，可造会通缩。

States. GDP will fail — Emad is right; but what replaces it is not a new dashboard, it is a new account: the wanting account — the total stock of intent each generation generates on its own that has a path to realization. Fertility is its crudest proxy, but the direction is right. The Law of Low Substitutability means what a state can do is not to issue wanting but to refrain from killing it. The protocol writer pointed the direction: accelerate, yes — but accelerate defense, the defense of humanity against being removed from the equation. Translated into national policy: what a nation most needs to defend is its young people's wanting against being leveled flat by its own intelligence.国家。 GDP 会失效，Emad 说对了；但替代它的不是一套新仪表盘，是一个新账户：想要账户，每一代人自发产生的、有实现路径的意图总量。生育率是它最粗糙的代理，但方向对。低替代律意味着国家能做的不是给想要，而是不杀死想要。那个写协议的人给过方向：加速可以，但要加速的是防御——防御人类被从方程里除掉。翻译成国策：一个国家最该防御的，是它年轻人的想要不被自己的智能填平。

The young. What you hold is appreciating, and you do not know it. Everyone tells you to learn AI, to grind skills — they are telling you to accumulate "buildable," and buildable is going to zero. The only scarce thing you own is the part of you not yet filled with answers. Hunger is your principal; humility is its interest. Do not trade it away for any pre-supplied answer — including the ones AI gives. Including mine.年轻人。 你持有的东西正在升值，而你不知道。所有人告诉你要学智能、要卷技能——那是在告诉你去积累"可造"，而可造正在归零。你唯一稀缺的，是你还没有被答案填满的那部分。饥渴是你的本金，谦卑是它的利息。不要把它换成任何一种预先给好的答案——包括 AI 给的，包括我给的。

◇

X. Three Falsifiable Predictions十、三条可证伪预言

An economics that cannot be wrong is not economics.一门经济学如果不能错，就不是经济学。

Five years. Before 2029, the price of commodity intelligence keeps falling faster than the price of electricity; "intelligence as a utility" becomes the mainstream phrase; model companies' market caps grow more slowly than power and institutional assets. If model companies retain pricing power, the sequence of "everything buildable" is wrong.五年。 2029 年前，商品级智能价格的下降速度持续超过电价，"智能公用事业化"成为主流表述，模型公司市值增速跑输电力与制度类资产。若模型公司持续保持定价权，"万物可造"的顺序错。

Ten years. Before 2034, the first large company appears whose core valuation metric is the volume of genuine wanting generated in its users, rather than output or attention; in the same window, at least one developed economy shifts its fertility or youth policy from buying timing to repairing stock. If the attention economy remains the pricing anchor and policy still stops at handing out money, "nothing wanted" has not yet been priced, and Section VII is postponed.十年。 2034 年前，出现首个以"用户真实想要的生成量"而非产出或注意力为核心估值指标的大型公司；同期至少一个发达经济体把生育或青年政策从买时点转向修存量。若注意力经济仍是定价锚、政策仍停留在发钱，"无人想要"尚未被定价，第七节推迟。

Twenty years. Before 2046, the first silicon–carbon covenant is formally registered in some jurisdiction, with breach adjudicable, in which the carbon side's obligation is written as "supply of intent" rather than "supply of labor." If no such covenant appears while fertility and autonomous-intent indicators keep declining, the merger ending stands, and this essay is wrong in the place that matters most — and I hope that is where it is wrong, and nowhere else.二十年。 2046 年前，出现首份被某一法域正式登记、可裁决违约的硅基-碳基之约，其中碳基一方的义务被写成"意图供给"而非"劳动供给"。若此约不出现，而生育率与自主意图指标持续下行，则合并结局成立，本文在最重要的地方错了——而且希望是错在这里，不是错在别处。

◇

Back to that unpunctuated line, and the eight characters I pressed out of it.回到那句没有标点的话，和我压出来的八个字。

Someone wants to carry the light of consciousness to another planet, and calls it civilization's mission. That is extending consciousness in space. Negentropy economics cares about the other direction: not carrying wanting farther, but regenerating it here — in level, not in distance.有人要把意识的光带到另一颗星球上，说这是文明的使命。那是在空间上扩展意识。逆熵经济学关心的是另一个方向：不是把想要带得更远，而是让想要在这里再生一次——在层级上，而不是在距离上。

Emad says we are the last people to remember that human ideas were worth money. Perhaps. But we are also the first to discover that human wanting is worth money. Ideas can be copied; wanting cannot. I would like to leave him that sentence, as the first line of his next book.Emad 说，我们是最后一批记得人的想法值钱的人。也许。但我们也是第一批发现人的想要值钱的人。想法可以被复制，想要不能。这句话我想留给他，作为下一本书的第一句。

Everything buildable is the problem silicon has finished for us. Nothing wanted is the problem left to us. The answer to the first is engineering. The answer to the second is a heart that still hungers.万物可造，是硅基替我们做完的题。无人想要，是留给我们的题。前一道题的答案是工程，后一道题的答案是一颗还会饿的心。

That is all.这就是全部。