Xiaoyang Yu

我们大脑的思考活动，主要就是电信号在神经元之间的传输。而量子随机性并不会使得这些电信号变得随机。 动作电位是巨量离子的集体流动，一次突触释放的是成千上万个分子。在这个尺度上，微观的涨落被平均掉了。 所以主张”我的决策里有随机成分”的人，连第一个前提都还没有拿到。 但本文不停在这里。下面一路让给他。 第一让：就算某个量子层面的随机事件，确实经由某种放大机制，改变了某个神经元的放电时刻——就算脑子里真有那么一处，微观的一次随机取值被放大成了宏观可分辨的差别。 第二让：就算这一下使得大脑做出了一个它自己不喜欢的决策——就是那个让他事后懊恼的决策。 第三让：再就算，他所喜欢的那个决策，刚好对应于这个随机事件的另一个可能取值——也就是说，那件事若取了另一个值，出来的就正好是他想要的那个决策。 让到这里，全让完了。而大脑的懊恼仍然不成立。 因为大脑并没有办法要求这个随机取值出现。 取值出哪一个，由那件事自己定。一样东西在可能取值之列，不等于它能被点到。 而懊恼责怪的是什么？责怪的是”我没让它出现”。而”没有办法让它出现”，恰恰免了这份责怪。 这里还有一个词被混着用了。“允许”有两个意思：一个是”它没被禁止”，一个是”我能促成它”。没被禁止，和我能促成，中间隔着整个问题。 而我们平时把这两句当成一句用：他说”那个决策是被允许的”，听的人就以为他说的是”他本来能做到”。 懊恼其实要三样东西：那一支得存在；那一支得是我要的那一个；而我得有办法让它出现。 随机性也许给得起第一样；第二样它给不了，因为它吐什么算什么；第三样它连边都不沾——因为”没有决定者”里的那个”没有”，同样包括我。 第三条不成立，前两条给再多也没用。 Abstract The thinking our brains do consists mainly of electrical signals transmitted between neurons. And quantum randomness does not make those signals random. An action potential is the collective flow of an enormous number of ions; a single synaptic release involves thousands of molecules. At that scale microscopic fluctuation is averaged out. So anyone claiming that his decisions contain a random component has not yet secured even the first premise. But the paper does not stop there. What follows is a series of concessions. First concession: grant that some random event at the quantum level really did, through some amplifying mechanism, alter the firing time of some neuron—that somewhere in the brain a microscopic random value was amplified into a macroscopically distinguishable difference. Second concession: grant that this made the brain produce a decision it does not itself like—the very decision it later regrets. Third concession: grant further that the decision it would have liked corresponds to another possible value of that same random event—that had the event taken the other value, out would have come precisely the decision he wanted. With every concession made, the brain’s chagrin still does not hold. Because the brain has no way to require that random value to occur. Which value comes up is up to the event itself. That something is among the possible values does not mean it can be selected. And what does chagrin reproach? It reproaches “I failed to make it occur.” And having no way to make it occur is precisely what dissolves that reproach. One word is being used in two senses. “Permitted” means either “it is not forbidden” or “I can bring it about.” Between not-forbidden and I-can-bring-about lies the whole question. Yet we use the two as one: he says the decision was permitted, and the listener hears that he could have done it. Chagrin in fact requires three things: that the branch exist; that it be the one I want; and that I have some way to make it occur. Randomness may afford the first; it cannot afford the second, since it emits whatever it emits; and it does not come near the third—because the “no” in “no determiner” includes me as well. With the third unmet, no amount of the first two helps.

当我说宏观事物由于量子随机性被平均掉而呈现决定性时，有人提出：量子随机性仍然可以支撑反事实设想的可能性。 本文只说一句：他提出的这种世界图景，并不比决定性演化更对我们的常识友好。 它至少同样糟糕。而”至少同样”这一条，就足够了。 因为这一步不是中立的补充。他不是在添一条物理事实，他是在换一张世界的图。 而换图要有理由，理由得是”更好”。 这个理由从来没有被交代过。 翻遍他说的话，找不到一句论证；找得到的是一份印象——决定论听着叫人不舒服，随机性听着像把余地还回来了。可”听着像”不是”是”。 而他若真要说，能拿出来的那几样，恰恰是他丢掉的： “至少不是被定死的。” 可他换来的是”没有决定者”——而没有决定者，也就没有他。定死不由他，随机也不由他。一样不由他。 “至少留了余地。” 留是留了——可留的余地不归他。骰子的余地是骰子的。 “至少不那么冷冰冰。” 这是感受，不是账。 本文不主张那张图更糟——那要另一把尺子，而立”没更好”用不上。只需要这一条：没更好，他就没有理由换图；而换图，是他整套说法的第一步。 Abstract When I say that macroscopic things display determinacy because quantum randomness is averaged out at that scale, someone replies that quantum randomness can still underwrite the possibility of counterfactual scenarios. This paper says one thing: the picture of the world he proposes is no friendlier to common sense than deterministic evolution. It is at least as bad. And “at least as” suffices. For this move is no neutral addition. He is not adding a physical fact; he is swapping the picture of the world. Swapping requires a reason, and the reason must be that it is better. That reason has never been given. Search his remarks and there is no argument, only an impression—determinism sounds uncomfortable, randomness sounds as though it gives the room back. But “sounds as though” is not “is.” And were he to try, the items he could produce are precisely the ones he has lost: “At least it isn’t fixed.” But what he gets instead is “there is no determiner”—and with no determiner, there is no him either. Fixed is not up to him; random is not up to him. Equally not up to him. “At least room is left.” Room is left—but the room left is not his. The dice’s room belongs to the dice. “At least it isn’t so cold.” That is a feeling, not an entry. The paper does not claim that picture is worse—that would need another measure, and establishing “no better” does not require it. Only this is needed: no better, and he has no reason to swap; and swapping is the first step of his entire case.

“如果不是刚才回家路上把钥匙丢了，那么现在我就可以拿钥匙打开门进门回家了。” 说这句话的人在懊恼。而他懊恼的时候，只关注钥匙丢没丢，没有关注是否在五分钟之后进了家门。 懊恼盯着那一个点。而懊恼所依赖的，恰恰是懊恼没在看的那一整片。“现在”要求那五分钟照样过去；“这扇门”要求他走的是那条路；“我站在这儿”要求他此刻在这儿；“门还锁着”要求这五分钟里没人回来开门。没有这一整片照抄，那句话就没有意思。可这一整片，他一眼也没看过——它不是被看过之后判定为”跟实际一样”，它根本没有进入视野。 于是那个干净的两分就露了馅。“没丢”那一格里能装的东西多得很：没丢但走了另一条街，没丢但遇上熟人聊了十分钟，没丢但半路折回买水。“没丢”不是一个情景，是一大堆情景的名字。而我们往那一格里只放了一张画，还以为那一格本来就只装得下这一张。 这份”唯一感”是自带的，不是推出来的：设想一出现就带着”就是这样”，不是从几张里挑出来的，是根本没觉得有得挑。而它一被质疑就退成”只要钥匙没丢就好，之后干了什么不重要”——一推就倒，说明它从来没有根据；可这丝毫不妨碍它在被质疑之前，是他说那句话时的全部底气。 而这句退让本身也不成立：若那五分钟里他坐车去了别的城市，“现在我就可以开门进屋”就是假的。所以”不重要”的真意是”凡是不妨碍我这句结论的都不重要”——他不是放宽了要求，是把筛子藏了起来。 接下来，救兵来了。 有人搬出量子随机性：既然与”丢了”不同的就只能是”没丢”，那随机性一定能生成这种不同。可他继承了前面那一整套，一样也没检查。 他接受了那个两分——而穷尽的是谓词，不是走法；他也没注意到”其余照抄”这一条——而他要拿随机性去满足的，恰恰就是这一条。 于是那个问题落下来：那五分钟的整个过程，是在钥匙丢／没丢的那一瞬间由随机性随机选择出来的吗？如果是，它如何随机地选出一个与实际基本一样的五分钟？ 改一处而抄其余，需要一个抄写员。而随机性不是抄写员。 它没有”照着那一份来”这个功能，只有”吐一个出来”；一个真正的生成器在那一瞬间分岔之后往下只会越走越散。问题不在概率小，在功能：抄写不在它的功能表上。 于是他其实需要两种互相矛盾的随机性：在钥匙那一点上任性到能把”丢”翻成”没丢”，在其余每一点上听话到把那五分钟原样走完。同一样东西，一处任性，处处听话。 而在这之前还有一关：随机性是发生器，不是应答器，你没法向它点单。 他证明的是”存在别的”，他要的是”存在那一个”；两句共用同一个”可能”，字面没变，所以那一跳看不出来。而钥匙这个例子把这一关变得格外难看：他点的那份单子，九成九是”跟实际一模一样”，只有一处不同——这不是难点单，是这份单子的形状跟随机性正相反。 最后还有一记，比前面都短。就算随机性交得出那一片，交上来的那一份也正好取消了懊恼。 因为懊恼要的不只是”本来可以不这样”，它要的是”本来可以由我不这样”；而随机说的不是”留了个位置给你”，是”没有决定者”，这个”没有”同样包括他。如果量子随机性能够使得钥匙没丢，那么这恰恰证明丢钥匙的人没有一丝责任，因此没有什么可以懊恼的。 钥匙没丢那一支若是靠一次随机涨落走出来的，钥匙丢了这一支也是——两支都不是他做的，为此懊恼等于为一次抛硬币的结果责怪自己。他要的是方向盘，拿到的是骰子；而骰子既不能被责怪，也不能被表扬。 于是救兵的处境很尴尬：它救的是设想的可能性，杀的是懊恼的对象。 我们没在看的那一片，正是随机性交不出来的那一片。而它若真交得出来，那一份也正好把懊恼取消了。 Abstract “If I hadn’t lost my keys on the way home just now, I could open the door and go in right now.” The speaker is aggrieved. And while aggrieved, he attends only to whether the keys were lost—not to whether he got in the door five minutes later. Chagrin fixes on the single point. And what chagrin depends on is precisely the whole stretch it is not looking at. “Right now” requires those five minutes to have passed as they did; “the door” requires that street; “me standing here” requires that he is here; “still locked” requires that no one came home in the meantime. Without that whole stretch copied over, the sentence means nothing—yet he has never looked at it once. It was not inspected and found identical to the actual; it never entered the field of view. The tidy dichotomy thus fails. The “not lost” slot holds a great many things: not lost but a different street, not lost but ten minutes of conversation, not lost but turning back for water. “Not lost” is not one scenario but the name of many. And into that slot we placed exactly one picture—believing the slot could only ever hold that one. This feel of uniqueness comes built in, not inferred: the scenario arrives already wearing “this is how it is”; it was not chosen among several—it never seemed there was a choice. Challenged, it retreats to “so long as the keys weren’t lost; what I did afterwards doesn’t matter”—what falls at one push never had a footing, yet that does not stop it from being, before the challenge, the whole of his confidence. And the retreat fails too: had he taken a bus to another city in those five minutes, “I could go in right now” would be false. So “doesn’t matter” means “whatever does not obstruct my conclusion doesn’t matter”—he has not relaxed his requirement; he has hidden the sieve. Then the rescue arrives. Someone invokes quantum randomness: since the only case different from “lost” is “not lost,” randomness must be able to generate it. But he has inherited the whole apparatus above without inspecting any of it. He accepts the dichotomy—yet what is exhausted is the predicate, not the course; and he has not noticed the demand that the rest be copied—which is exactly what he is asking randomness to supply. So the question falls: was that entire five minutes selected by randomness at the instant of the keys? And if so, how does randomness randomly select a five minutes essentially identical to the actual one? To alter one point and copy the rest requires a copyist. Randomness is not a copyist. It has no function “follow that one,” only “emit one”; a genuine generator, once forked, only diverges further. The problem is not low probability but function: copying is not on its list. So he needs two mutually contradictory randomnesses: at the keys, wilful enough to turn “lost” into “not lost”; everywhere else, obedient enough to walk those five minutes exactly as they went. One and the same thing, wilful at one point and obedient everywhere else. And before all this, another gate: randomness is a generator, not a respondent; you cannot order from it. He has shown that “something else exists”; he needs that “that one exists”—both sentences using one word “possible,” so the leap goes unseen. The keys make it uglier: what he orders is ninety-nine percent “exactly as it actually was,” differing at one point. Not a hard order—an order shaped opposite to randomness. One last point, shorter than the rest. Even if randomness could deliver that stretch, what it delivers cancels the grievance. For chagrin wants not merely “it might have gone otherwise” but “it might have gone otherwise through me”; and randomness says not “a place is left for you” but “there is no determiner”—and that “no” includes him. If quantum randomness could have made the keys not get lost, that is precisely proof that the man who lost them bears not the slightest responsibility, and so has nothing to be aggrieved about. If the not-lost branch came of a random fluctuation, so did the lost branch—neither was his doing, and grieving over it is reproaching oneself for the outcome of a coin toss. He wanted a steering wheel and was handed dice; and dice can be neither blamed nor praised. The rescue is thus in an awkward position: it saves the possibility of the scenario and kills the object of the grievance. The stretch we were not looking at is exactly the stretch randomness cannot deliver. And were it delivered, that delivery would itself cancel the grievance.

有人抬出量子随机性，来说事情本来可以不这样。本文向他提三个问题。 第一问：为什么你只在解释反事实设想的时候用它？ 他计划时不提它，预测时不提它，追责时不提它，许诺时不提它。工程算式里不加，医生开药不加，法官判案不加。只有一个地方他提它：当有人说事情就是这么走的时候。 第二问：它在别处会怎样？日常语言里为什么一点痕迹也没有？ 我们不说”明天早上我是否会带伞，将由我大脑中的量子随机性经某种通道放大后决定”，我们说”明天早上我打算带伞”。这不是一处两处的省略，是一处普遍的空白。举证责任因此在他那边：不是我要论证它不在这儿，是他要论证它在这儿而语言里一点痕迹也没有。 他最顺手的一句退路是”日常语言粗糙，不反映精细结构”。可这句话说的正是：它在这个尺度上不显露。而不显露的东西，撑不起’本来可以不这样’。 他用来解释无痕的理由，恰好取消了他要的效果。所以他得同时守住两头：在他要用的地方，它得大到能撑起”本来可以不这样”；在别处，它得小到千百年一句话也没留下。 第三问：能不能精确地用量子随机性解释一个日常的反事实设想？把计算拿出来。 挑任何一个都行。“要是我当初没搬家。”“要是那天没下雨。”“要是他没接那个电话。“任选一个，做出来：哪一个涨落？在哪一刻？经哪条通道？放大到什么程度？最后怎么落成这个特定的结局？ 要数。 一个也没有。 不是做得不够好，是一次也没有人做过。 他会答”原则上可以，只是太复杂”。可原则上可以，跟一次也没有人做过，是两回事。 他正在用”原则上可以”那一栏的信用，去支付”已经成立”那一栏的账。 而这份作业还有一个他不会喜欢的性质：做不出来，他没有依据；做出来了，他没有结论。 因为一旦真的做出来，那就是一份有确定分布的清单，不再是一句自由的许诺——他要的那种开放性，恰恰在他交作业的那一刻消失。 Abstract Some invoke quantum randomness to say that things might have gone otherwise. This paper puts three questions to them. First: why do you invoke it only when explaining counterfactual scenarios? He does not mention it when planning, predicting, blaming, or promising. It appears in no engineering calculation, no prescription, no judgment. It appears in one place only: when someone says that things simply went the way they went. Second: what does it do elsewhere? Why is there no trace of it in ordinary language? We do not say “whether I take an umbrella tomorrow morning will be decided by quantum randomness in my brain, amplified through some channel”; we say “I plan to take an umbrella tomorrow morning.” This is not an omission here and there but a general blank. The burden therefore lies with him: it is not for me to show that it is absent here, but for him to show that it is present while leaving no trace at all. His readiest retreat is that ordinary language is coarse and does not reflect fine structure. But that is precisely to say it does not show itself at this scale—and what does not show itself cannot support “it might have gone otherwise.” The reason he offers for the absence of traces cancels the effect he wants. So he must hold both ends at once: where he needs it, it must be large enough to support “otherwise”; everywhere else, small enough to have left not one sentence in thousands of years. Third: can you explain one everyday counterfactual precisely in terms of quantum randomness? Produce the computation. Take any one. “If I hadn’t moved.” “If it hadn’t rained that day.” “If he hadn’t taken that call.” Pick one and do it: which fluctuation? at which moment? through which channel? amplified how far? and how does it end in that particular outcome? With numbers. There is not one. Not done badly—never done at all. He will answer that it is possible in principle, merely too complex. But possible in principle and never once performed are two different things. He is drawing on the credit of the “possible in principle” column to pay a bill in the “already established” column. And the assignment has a property he will not like: fail it and he has no ground; pass it and he has no conclusion. For once it is actually done, what he holds is a list with a definite distribution, not an open promise—the very openness he wanted disappears at the moment he hands in the work.

一条判据： 如果没有一条物理链条，从某样东西通到李四的脑子里，那么我就不该把李四话里的这样东西，理解为位于、独立于他脑子之外的什么。它反映的只是他脑子的功能。 理由很简单。一句话是一段空气振动。它能反映什么，看有什么通到了它这儿。 “那面墙是砖砌的”——有链条。 砖反射的光进了他的眼睛，改变了他的脑内状态，最后带动了声带。砖挪个位置，他说的话就会不同。所以这段振动携带了关于砖的信息。 “砖的移动是因为张三”里的”因为”——没有链条。 从工地出发的东西里，光带来了张三的身体在动，也带来了砖在动；没有任何一样携带”因为”。没有一个”因为”坐在那儿发出信号。 那这个”因为”该放在哪儿？有两个安放处。 放在脑外，需要一条从未被出示的链条；放在脑内，不需要任何额外的东西——那台机器就在那儿，而且已经足够交代这段振动了。在这两个之间，该放在不需要额外东西的那一个。 而它反映的是什么？是脑子形成文字的某个模式。 这些词的行为完全可以观察：它们出现在什么位置，跟哪些词搭配，什么时候被说出口。这套行为由那台机器的结构就能交代完，不需要在脑外再放一个对象。 很多让人觉得”似乎存在”的抽象概念都是这样——“因为”“应当”“可能”“本质”“意义”。它们并不存在于论证这些概念的人脑之外，它们只是反映了人脑形成文字的某个模式。 那份”它明明是客观的”的感觉从哪儿来？从模式的一致来。 我们在同一套语言里长大，学的是同一套用法，于是这些词在千万人嘴里的行为高度一致。一致性被读成了对象的稳定性。 这条判据不砍掉科学。电子没有谁看见过，可它有链条：它撞在探测器上，探测器响一声，声波进了耳朵。链条可以很长、很曲折，但每一节都是物理的。“因为”没有第一节——它不是一团有位置、会反射光的东西。 Abstract A criterion: If no physical chain runs from some thing into Li’s head, then I should not understand that thing, as it appears in Li’s speech, as located outside and independent of his head. It reflects only a function of his brain. The reason is simple. A sentence is a vibration of air. What it can reflect depends on what reached it. “That wall is made of brick”—there is a chain. Light off the brick entered his eye, altered his brain state, and finally drove his vocal cords. Move the brick and his words change. So the vibration carries information about the brick. The “because” in “the brick moved because of Zhang”—no chain. Of the things that left the site, light carried Zhang’s body moving and the brick moving; nothing carried a “because.” No “because” sits there emitting a signal. Where, then, should that “because” be placed? There are two places. Placing it outside the head requires a chain never exhibited; placing it inside requires nothing extra—the machine is right there and already accounts for the vibration. Between the two, it belongs where nothing extra is required. And what does it reflect? Some pattern by which the brain forms words. The behaviour of such words is fully observable: where they occur, what they combine with, when they get uttered. All of it is accounted for by the structure of that machine, with no object needed outside the head. Many abstract concepts that feel as if they exist are like this—“because,” “ought,” “possible,” “essence,” “meaning.” They do not exist outside the head of the person arguing for them; they merely reflect some pattern by which the human brain forms words. Where does the feeling of objectivity come from? From the consistency of the pattern. We grow up in one language and learn one set of usages, so these words behave alike in millions of mouths. Consistency is read as the stability of an object. The criterion does not cut down science. No one has seen an electron, yet it has a chain: it strikes a detector, the detector registers, sound reaches an ear. A chain may be long and crooked, but every link is physical. “Because” has no first link—it is not a parcel with a position that reflects light.

李四说：“砖的移动是因为张三。” 现在看这句话本身。 它是一段空气的振动。作为物理过程，它和砖的移动是同一类东西：位置随时间变，本来就会这么发生。 本文不去讨论这段振动是不是”因为”李四——那是另一件事。本文要说的是别的：作为本来就会发生的一段振动，它所能传达、所能反映的，不超出李四大脑内部的具体结构。 理由很直接。看有没有一条物理链条，从那样东西通到李四的脑子里。 “砖在那面墙里”这句话有：砖反射的光进了他的眼睛，改变了他的视网膜，一路改变了他的脑内状态，最后带动了声带。这条链条从砖出发，一直通到那段空气振动。 所以这段振动确实携带了关于砖的信息。 “砖的移动是因为张三”这句话，前半截也有：光携带了张三的身体在动，也携带了砖在动，两样都通到了他的脑子里。可”因为”呢？从工地出发的东西里，没有任何一样携带”因为”。 没有一个”因为”在那儿发出信号，也没有一条链条把它送进任何人的脑子。 所以那个”因为”不是从工地进来的。它是在脑子里被加上去的。而一段被加上去的东西，只能反映加它的那台机器。 这不是说没有哪句话能讲世界。能讲的那些，都有链条；那个”因为”没有。 一群人都这么说，也一样。若一万个人都说”因为张三”，从工地出发的东西里仍然没有一样携带”因为”。一万份相同的读数，读的仍然是一万台相似的机器。 Abstract Li says: “The brick moved because of Zhang.” Now look at the sentence itself. It is a vibration of air. As a physical process, it is the same kind of thing as the brick’s movement: position changing with time, and it was going to happen that way. This paper does not take up whether that vibration is “because of” Li—that is a separate matter. It says something else: as a vibration that was going to happen anyway, what it can convey and reflect does not extend beyond the specific structure inside Li’s head. The reason is direct. Ask whether there is a physical chain running from the thing in question into Li’s head. “The brick is in that wall” has one: light off the brick entered his eye, altered his retina, altered his brain state along the way, and finally drove his vocal cords. That chain runs from the brick all the way to the vibration. So the vibration does carry information about the brick. “The brick moved because of Zhang” has one for part of it: light carried Zhang’s body moving and the brick moving, and both reached his head. But the “because”? Nothing leaving the site carries a “because.” No “because” is there emitting a signal, and no chain delivers one into anybody’s head. So the “because” did not come in from the site. It was added inside the head. And what is added can only reflect the machine that added it. This is not to say no sentence can be about the world. The ones that can all have chains; that “because” has none. The same holds when everyone says it. If ten thousand people say “because of Zhang,” still nothing leaving the site carries a “because.” Ten thousand identical readings are readings of ten thousand similar machines.

这块砖的移动，不是因为张三。 “张三搬了一块砖”——这句话是对物理过程的事实描述，不偏不倚。可”砖的移动是因为张三”，这句里的”因为”不成立。 理由从”因为”的成立条件说起。说”因为张三”，就等于说砖本来不必这么移动。 得有一个”本来”在那儿，供张三去改；没有那个”本来”，“因为”就没有地方安放。 而”本来”有两个意思，两个都走不通。 第一个意思：原本就是那样。 在整个宇宙这个系统里，砖本来就会这么移动。砖是宏观物体，这个尺度上不显示什么随机性；它走的就是这一条。而张三也本来就会这么移动——他不站在系统外面往里推，他自己也是本来就这么走的那一段。一个本来就这么走的东西，改不了另一个本来就这么走的东西。 这个意思下，没有任何东西可供张三去改。 第二个意思：若不如此，则会那样。 这是对手真正在用的那个意思。可这个意思一开门，候选就成了无限多：若那天下雨、若砖是湿的、若地面软一寸、若卡车晚到两小时、若那炉砖火候差一点、若工头改了图纸、若张三没生下来、若引力小一点——每一个都同样支持”砖本来不必这么移动”，而它们彼此之间没有等级。在这个意思下，“因为张三”和”因为那天没下雨”完全同格。 谁要是坚持前者而不肯说后者，那他用的就不是这个意思——他偷偷退回了第一个意思，而第一个意思那边根本没有”本来”。 所以两头都落空： 第一个意思给不出可供改的东西；第二个意思给出的是无差别的无限多，把张三稀释成了其中之一。 那我们为什么偏偏从那无限多里挑中张三？因为挑的标准不在那堆候选里，它在我们这边：我们要找的是能被要求的那一个——能被付钱、被追责、被要求下次小心。湿度不能，引力不能，那炉火候也不能。这解释的是我们的挑法，不是砖的移动。 至于工照记、钱照发、赔照赔——那些也都是实际发生的事，照旧运转。动的不是账，是那个”因为”。 Abstract The brick did not move because of Zhang. “Zhang carried a brick” is a plain factual description of a physical process, and it is impartial. But in “the brick moved because of Zhang,” the “because” does not hold. Start from what “because” requires. To say “because of Zhang” is to say the brick need not have moved this way. There must be some “would have been otherwise” standing ready for Zhang to alter; without it, “because” has nowhere to sit. “Would have been otherwise” has two senses, and neither works. First sense: it was going to be that way all along. In the universe taken as one system, the brick was going to move exactly as it did. A brick is a macroscopic body; at that scale nothing random shows itself; it goes the one way it goes. And Zhang too was going to move exactly as he did—he does not stand outside the system pushing in; he is himself a stretch that was going to run that way. A thing that was going to go that way cannot alter another thing that was going to go that way. On this sense there is nothing available for Zhang to change. Second sense: had things been otherwise, it would have gone otherwise. This is the sense the opponent actually uses. But opening that door makes the candidates infinite: had it rained, had the brick been wet, had the ground been an inch softer, had the truck arrived two hours late, had that kiln-load been underfired, had the foreman revised the drawings, had Zhang never been born, had gravity been weaker—each equally supports “the brick need not have moved this way,” and none ranks above the others. On this sense, “because of Zhang” and “because it did not rain that day” are exactly on a par. Anyone who insists on the first while refusing the second is not using this sense at all—he has quietly retreated to the first, where there is no “otherwise” to be had. So both ends fail: the first sense yields nothing to alter; the second yields an undifferentiated infinity that dilutes Zhang into one item among them. Why, then, do we pick Zhang out of that infinity? Because the criterion is not among the candidates; it is on our side: we want the one who can be held to it—who can be paid, charged, told to be careful next time. Humidity cannot, gravity cannot, the kiln’s firing cannot. This explains our manner of picking, not the brick’s motion. As for wages recorded, money paid, damages compensated—those too are things that happen, and they go on as before. What is withdrawn is not the bill but the “because.”

“这块砖是张三搬来的。” 本文只问：这句话有没有在世界里指出一样多出来的东西。 全局里只有位置随时间变，处处相接。而说”因他而如此”，要求世界里多出一个新的物理关系——一头搭在他身上，一头搭在砖上，且不是那套演化本身已经含有的。若他跟砖的全部往来都已经在那套演化里，这句话就只是把里面的东西重念了一遍。 而”张三搬的”确实在描述实际发生的事，一个字不假——可它所描述的正是那套演化本身，只是按”人”这个单位重新编了索引。而索引不是关系：按”张三”归档，跟按”下午三点”“三号料场”“第七车”归档，是同一类操作；归档改变的是我们怎么找到它，不是它跟什么发生了关系。视角自由，物理不动。所以”因”没有落脚处。 两个检验。手推砖：手确实推了，可它是从跟石头同一个门进来的——同一套规矩所作用的另一团物质；没有第二条通道，而”因”要的正是第二条通道。工钱：一头是一千次弯腰抬起，另一头是走到出纳处、抽屉打开、纸币换手；两段都是实际发生，而中间没有一个额外的力——没有任何东西把钱从出纳手里吸到张三手里。 所有说法里，只有”他影响了概率”真主张了一个新关系。它要拿数来：影响了多少，怎么算的。这份计算一次也没有人做过。 守则：账单一分不动。 工照记，钱照发，赔照赔，人照走到出纳处去——这些也都在那套演化里。取消的只是那个”因为”所声称的那根线。 Abstract “Zhang carried this brick.” The paper asks one thing: does this sentence point to anything additional in the world? The whole contains only position changing with time, everywhere continuous. To say “it is so because of him” requires that the world contain one new physical relation—one end on him, one on the brick, and not already contained in the evolution itself. If everything passing between him and the brick is already in that evolution, the sentence merely recites what is already there. “Zhang carried it” does describe something that happened, not a word of it false—but what it describes is that evolution itself, merely re-indexed by the unit “person.” And an index is not a relation: filing under “Zhang” is the same operation as filing under “3 p.m.,” “yard no. 3,” or “truck no. 7”; filing changes how we retrieve it, not what it stands in relation to. Viewpoint is free; the physics does not move. So “cause” has nowhere to stand. Two tests. The hand pushing the brick: it did push, but it entered by the same door as a stone—another parcel of matter under the same rules. There is no second channel, and a second channel is exactly what “cause” requires. Wages: at one end, a thousand bendings and liftings; at the other, walking to the cashier, a drawer opening, notes passing from hand to hand. Both happened, and between them there is no extra force—nothing draws the money from the cashier’s hand into Zhang’s. Of all the candidates, only “he shifted the probabilities” genuinely asserts a new relation. It requires numbers: how much, computed how. That computation has never once been performed. The rule: not one item on the bill changes. Wages recorded, money paid, damages compensated, people walking to the cashier—these too are within that evolution. What is withdrawn is only the thread that the word “because” claims to be there.

本文先立一条守则，因为不先立，后面每一句都会被读成开脱：本文取消的是”因”作为一个新的物理关系的地位，不取消分配。 工钱照付，保修照担，塌了照赔，砌歪了照返工——一项不少，一项不改。而这不是让步，是本文的一个结论：分配本来就没打算引进什么，它不是一份关于世界的报告。 论证从全局起步，因为这一步几乎没有人会反对：包含所有组件的宇宙，其中任何基本粒子的移动轨迹，与任何人都扯不上关系。 那里只有位置随时间变，处处相接。 接着要把”因”这个词要求的东西说清楚。说”这条线之所以是这条线，有一处是因为他”，这句话要求世界里多出一样东西：一个新的物理关系——一头搭在他身上，一头搭在砖上，而这个关系不是那套演化本身已经含有的。没有这样一样东西，“因”就没有落脚处。 而本文的核心判断是：人类文化那套记账方式，仍然是在描述物理定律的那套演化，只是换了个视角，并不能引进新的物理关系。 “这块砖是张三搬来的”，确实在描述一段实际发生的事：一团物质与另一团物质之间的一串相互作用，位置随时间变。这一点完全真实，一个字不假。可它所描述的，正是那套演化本身，只是按”人”这个单位重新编了索引。 而索引不是关系：把同一段位移按”张三”归档，跟按”下午三点”归档、按”三号料场”归档，是同一类操作；归档改变的是我们怎么找到它，不是它跟什么发生了关系。视角自由，物理不动。 这里有一个诱人的分法必须拦住：有人会说，描述那一部分算你说的，可”谁记工、谁负责、谁赔”那一部分不是描述，是我们之间的一项安排，它在描述之外。这个分法不成立，而拆开它的是工钱。“张三搬了一千块砖，得了一百元”——一头是他弯腰、抬起、走、放下，一千次；另一头是他走到出纳处，出纳打开抽屉数出钞票，纸币从一只手换到另一只手。两段都是实际发生，都在那套演化之内；而两段之间没有一个额外的力，没有任何东西把钱从出纳手里吸到张三手里。 所谓”安排”，同样落地为一串位移，落得一分不差。 所以只有一半：全都是描述，全都是换视角，一个新的物理关系也没有引进。 而”因”要的正是一个新的物理关系——所以它没有落脚处。 至于”张三应得一百元”里那个”应当”该如何安放，是一个语义学问题，本文明确悬置为待定。 必填表由此变得简单：主张”这条线跟某个人有关系”的人，所填的每一项，要么是换视角的描述，要么是安排。说他造成了它——他抬手、弯腰、松手，每一下都是同一套演化的一段。说他在场——描述，且不特殊，空气也在场。说那是他砌的——“砌”这个及物动词把描述与安排捆在一起卖，拆开来，一半是位移，一半是记工。四项里只有一项真有资格引进新关系：他影响了概率。 可这一项要拿数来——而这份计算一次也没有人做过。 本文并处理三条常见的反驳。其一，人体不能屏蔽物理定律：定律不是一种可以被物质屏蔽的射线，它同样”穿过”人体，因为人体本身就是一种普通的物体；手确实推了砖，但它是以与石头、与风完全相同的方式进入的——换视角不会屏蔽定律，被屏蔽的只是另一个视角。其二，就算让一步承认轨迹是随机的，也一分不损失：随机说的不是”留了个位置给你”，是”没有决定者”，而这个”没有”同样包括他；缺口不是空座。其三，图纸：砖的位置与三个月前纸上的线吻合，这份吻合是真的——可图纸那段线与墙那段线都在同一套演化之内，吻合是两处形状相似，不是一个新的关系；而且”对得上”发生在我们比对的时候，不发生在砖飞行的途中。 最后是一条梯度：仙女座里的一个粒子（毫无阻力）、山坡上滚下的石头（几乎没有）、一块被砌进墙里的砖（很难）、我抬起的手臂（最难）。论证一个字都没变，多出来的只有抵抗；而抵抗的强度只跟离人多近有关。 至于”我明明感觉是我做的”——那份感觉实实在在地发生了，可它是关于那份感觉的事实，不是关于那条线的事实。 Abstract A rule must be fixed first, or every sentence below will be read as an excuse: the paper withdraws the standing of “cause” as a new physical relation; it does not withdraw allocation. Wages are paid, warranties held, collapses compensated, crooked courses redone—nothing reduced, nothing altered. This is not a concession but a conclusion: allocation never purported to introduce anything; it is not a report about the world. The argument starts from the whole, a step almost no one contests: in the universe comprising all components, the trajectory of any elementary particle bears no relation to any person. There is only position changing with time, everywhere continuous. Next, what the word “cause” demands must be made explicit. To say “this line is as it is, and one part of why is him” requires that the world contain one thing more: a new physical relation, one end on him and one on the brick, a relation not already contained in the evolution itself. Without such a thing, “cause” has nowhere to stand. The paper’s central claim: the way human culture books things is still a description of the same physical evolution, merely from another viewpoint, and it cannot introduce a new physical relation. “Zhang carried this brick” really does describe something that happened: a train of interactions between one parcel of matter and another, position changing with time. This is entirely true, not a word of it false. But what it describes is the evolution itself, merely re-indexed by the unit “person.” And an index is not a relation: filing the same displacement under “Zhang” is the same sort of operation as filing it under “3 p.m.” or “yard no. 3”; filing changes how we retrieve it, not what it stands in relation to. Viewpoint is free; the physics does not move. A tempting division must be blocked here: one wants to say that the descriptive part may be granted, but that the part about credit, answerability and compensation is not description at all—an arrangement among us, outside description. The division does not hold, and what breaks it is wages. “Zhang carried a thousand bricks and received a hundred yuan”—at one end, bending, lifting, walking, setting down, a thousand times; at the other, walking to the cashier, a drawer opening, notes counted out, paper passing from one hand to another. Both are things that happened, both within the same evolution; and between them there is no extra force—nothing draws the money from the cashier’s hand into Zhang’s. The so-called arrangement likewise lands as a train of displacements, and lands exactly. So there is only one half: all of it is description, all of it a change of viewpoint, and not one new physical relation has been introduced. And “cause” requires precisely a new physical relation—so it has nowhere to stand. As for how the “ought” in “Zhang is owed a hundred yuan” should be placed, that is a question of semantics, and the paper expressly leaves it pending. The checklist becomes simple: every entry a claimant can supply is either a re-viewed description or an arrangement. He caused it—his lifting, bending, releasing are each further stretches of the same evolution. He was present—description, and not special; so was the air. He laid it—the transitive verb sells description and allocation bundled; unbundled, half is displacement and half is credit. Of the four, only one could genuinely introduce a new relation: he shifted the probabilities. But that entry requires numbers—and that computation has never once been performed. Three standard objections are handled. First, a human body cannot shield physical law: law is not a ray any substance could block; it passes through the body, because the body simply is an ordinary object. The hand did push the brick, but it entered exactly as a stone or a gust of wind enters—a change of viewpoint shields no law; what is shielded is only the other viewpoint. Second, even granting that the trajectory is random, nothing is lost: randomness does not say “a place is left for you,” it says “there is no determiner,” and that “no” includes him; a gap is not a vacant seat. Third, the blueprint: the brick’s position matches a line drawn three months earlier, and the match is real—but the blueprint’s line and the wall’s line both lie within the same evolution, so the match is a similarity of two shapes, not a new relation; and the matching happens when we compare, not during the brick’s flight. Finally a gradient: a particle in Andromeda (no resistance), a stone rolling downhill (almost none), a brick set into a wall (hard), the arm I raise (hardest). Not a word of the argument changes; what is added is only resistance, whose strength tracks nothing but nearness to the human. As for “but I feel that I did it”—the feeling really occurred, yet it is a fact about that feeling, not a fact about the line.

本文先立一条守则，因为不先立，后面每一句都会被读成开脱：本文取消的是”功劳”与”过错”作为轨迹上一处关节的地位，不取消分配。 工钱照付，保修照担，塌了照赔，砌歪了照返工——一项不少，一项不改。这些是我们要办的事，而它们本来就不靠”线上有那么一处”来立；它们靠的是谁能被要求。把线那一栏腾空，账单一分不动。 论证从全局起步，因为全局那一步几乎没有人会反对：包含所有组件的宇宙，其中任何基本粒子的移动轨迹，与任何人都扯不上关系。 那里只有位置随时间变，处处相接。而归因所要的不是一条线，是线上的一处关节——一处”这里是因为他”的接口。全局里没有这样的接口：每一段都同样是整条线的一部分，没有哪一段比别的更像原因。把某一段挑出来叫”因”，挑的标准不在线上。 论证到这里就已经完成了。以下每一步都不再增加任何东西，只是把同一句话搬到一个越来越难接受的地方。 第一步，抠出一块砖。 从全局里抠一块砖来跟踪，并没有给它添加任何东西：若全局里没有关节，全局的任何一片里也没有。于是——一块砖的移动轨迹，它会进入哪个房子的哪面墙里，与任何人都扯不上关系。 而这一句立刻变得刺耳，尽管论证一个字都没变。多出来的只有抵抗。 抵抗的来源可以指出来：砖是人干活的地方。 于是有一条梯度：仙女座里的一个粒子（毫无阻力）、山坡上滚下的一块石头（也几乎没有）、一块被砌进墙里的砖（很难）、我抬起的手臂（最难）。这不是论证强度的梯度，是离人远近的梯度。 因此举证责任落在主张例外的一边：你既然认了全局那一句，就得说出砖在哪一点上与别处不同。 跨过这条梯度，本文给出一座桥。若你能在一个没有人的宇宙里理解上面那句话，却在有人的宇宙里卡住，那么请注意：人体并不能屏蔽”物理定律射线”，物理定律也不是一种可以被某种物质屏蔽的射线。 定律同样”穿过”人体——或者说，人体其实就是一种普通的物体，只是我们自以为它不普通。人体既不能屏蔽、也不能扭曲作用于某物体的物理定律，因此该物体就如同置身于一个只有普通物体的宇宙之中。 这里要挡住那条必然的反驳：“可手推了砖啊。“手确实推了，而且确实进入了那条线——但它是以与石头、与风、与水完全相同的方式进入的：作为同一套定律所作用的另一团物质，作为边界条件的一部分。没有第二条通道。 而”关节”所要的，恰恰是第二条通道。此外”人体改变了它的轨迹”这个说法本身也有毛病：说”改变”要拿两条线来比——它实际走的那条，和”若无人体它会走的那条”；而后者从来没有发生过，没有第二条线可供比较。“屏蔽”这个比方的用处正在这里：它把那份直觉的形状暴露了出来——把人当成一块定律的阴影区，定律管别处，人这儿归人自己管。没有人肯认这句话；可我们的记账方式一直照它在记。 由此可以把那份”被屏蔽”的错觉说准。物理定律对于物体的移动，有一种计算方式；人类文化对于物体的移动，有一种记账方式——比如这块砖是哪个人搬过来的。记账方式并不能改变计算方式。 可在普通人的认知里，一想记账那一套，就把计算那一套忘了，仿佛定律的作用被屏蔽了。被屏蔽的从来不是定律，是另一本账；屏蔽发生在注意力上，不在世界上。而注意力之所以总往文化那本去，是因为文化那本有情节——谁搬的、谁该赔、谁做得好——物理那本没有情节，它只是位置随时间变。 这里有一句要说准：物理定律的计算方式也是一种记账方式——但它不是人做的。 人做的是我们写下的那些式子、单位、坐标、近似，它们是描摹，可以被修订；而被修订的是描摹，不是被描摹的那样东西。那套计算本身不由谁来记，是世界自己就那样走；说它是”记账方式”，只为跟文化那本并排摆着看，而两者最要紧的不同正在于此——文化那本有作者，它没有。于是差别落在依赖关系上：物理那本不受人类文化那本的影响，而根子是——一本有作者，改作者的笔就改得动；另一本没有作者，没有笔可改。 检验当场可做——把”这块砖是张三搬的”改成”是李四搬的”，或者干脆不记，砖的轨迹一个数都不变；工地上的账本烧了，砖也不会挪一寸。反过来，若那一套计算换了一条，“搬”这件事、工地、砖、以及记账这件事本身，全都不成立。一本改了，另一本一个数不变；另一本改了，这一本整个塌。 而两本账本不冲突，因为它们记的不是同一栏：文化那本记的是谁能被要求，物理那本记的是位置随时间变；冲突只发生在我们把文化那本读成对世界的报告的时候。 本文并把那份关系做成必填表，逐项都塌。说他造成了它——他抬手、他弯腰、他松手，每一下本身也只是线上的另一段位移，用一段线解释另一段线，没有在任何地方造出关节。说他影响了概率——请拿数来，而这份计算一次也没有人做过。说他在场——空气也在场，脚手架也在场。说那是他砌的——“砌”是及物动词，自带一个主语位置，它多出来的不是物理内容，是那个位置。而就算让一步，承认轨迹不是必然的而是随机的，也一分不损失：随机说的不是”留了个位置给你”，是”没有决定者”，而这个”没有”同样包括他。缺口不是空座。 本文并处理砖这边特有的一条反驳：图纸。 砖落在墙上那个位置，跟三个月前一张纸上的线吻合，而且整栋楼都对得上——这份吻合是真的。可吻合是两段之间的扣合，不是任何一段线上的关节：图纸和墙都在这个世界里，两者能对上一点不假；而”对得上”发生在我们比对的时候，不发生在砖飞行的途中。 梯度的最后一级是**“我抬起手臂”**。手臂里那些粒子的移动，同样与任何人扯不上关系。手臂那一截跟砖是同一类东西——它跟砖一样实在，只是它那条线跟砖那条线一样处处相接，没有一处是关节。至于”我明明感觉是我做的”，那份感觉实实在在地发生了，可它自己不带印章：它是关于那份感觉的事实，不是关于那条线的事实。 最后一句关于文献：并不是从来没有人谈过物，而是没有人把物的线当作它自己的线来读。 物的移动一向被追踪得极细——批号、监管链、谁在几点几分做了什么——可每一条记录都以人为索引，被”工具”这个范畴吸收：物的运动被当成人的行动的延伸。不是没人看，是看的时候它已经挂在某人名下了。 Abstract A rule must be fixed first, or every sentence below will be read as an excuse: the paper withdraws the standing of “credit” and “fault” as a joint in a trajectory; it does not withdraw allocation. Wages are paid, warranties held, collapses compensated, crooked courses redone—nothing reduced, nothing altered. These belong to what we have to do, and they never rested on there being such a point on the line; they rest on who can be held to them. Empty that column on the line, and not one item on the bill changes. The argument starts from the whole, because that step is one almost no one contests: in the universe comprising all components, the trajectory of any elementary particle bears no relation to any person. There is only position changing with time, everywhere continuous. What attribution requires is not a line but a joint on the line—an interface at which “here it is because of him.” The whole contains no such interface: every stretch is equally part of the line, and none is more like a cause than another. To single out one stretch as “the cause” applies a criterion that is not on the line. At that point the argument is finished. Every step below adds nothing; it merely carries the same sentence into places progressively harder to accept. First step: carve out a brick. Tracking a brick out of the whole adds nothing to it: if the whole contains no joint, no region of the whole contains one. Hence—a brick’s trajectory, which house and which wall it will end up in, bears no relation to any person. And this at once grates, although not a word of the argument has changed. What has been added is only resistance. The source of the resistance can be named: a brick is where people work. Hence a gradient: a particle in Andromeda (no resistance at all), a stone rolling down a slope (almost none), a brick set into a wall (hard), the arm I raise (hardest). This is not a gradient of argumentative strength but of nearness to the human. The burden therefore falls on whoever claims an exception: having granted the sentence about the whole, you must say in what respect the brick differs. To cross this gradient the paper supplies a bridge. If you can follow the sentence above in a universe with no people yet stall in a universe with them, note this: a human body cannot shield “physical-law rays,” nor is physical law a ray that any substance could shield. Law passes through the body as well—or rather, the human body simply is an ordinary object, which we merely take to be extraordinary. A body can neither block nor bend the laws acting on some object, and therefore the object stands exactly as it would in a universe containing only ordinary objects. The inevitable objection must be blocked: “but the hand pushed the brick.” It did, and it really entered the line—but it entered exactly as a stone, a gust of wind, or water enters: as another parcel of matter acted on by the same laws, as part of the boundary conditions. There is no second channel. And a second channel is precisely what a joint would require. Furthermore, “the body changed its trajectory” is itself defective: to speak of change one needs two lines—the one it actually took and the one “it would have taken without the body”—and the latter never occurred; there is no second line to compare with. The shielding image earns its keep here: it exposes the shape of the intuition, which treats the human as a shadow region where law governs elsewhere but here matters are our own. No one will assert this sentence; yet our bookkeeping has always been done as if it were true. This lets the illusion of “shielding” be stated exactly. Physical law has a way of computing an object’s motion; human culture has a way of booking it—which person carried this brick. The booking cannot alter the computing. Yet in ordinary cognition, once we think in the booking mode we forget the computing mode, as though the law’s operation had been shielded. What is shielded is never the law but the other ledger; the shielding occurs in attention, not in the world. And attention gravitates to the cultural ledger because that ledger has a plot—who carried it, who must pay, who did well—while the physical ledger has none; it is only position changing with time. One sentence must be stated exactly: the physical way of computing is also a way of booking—but it is not made by people. What people made are the formulas we write, the units, the coordinates, the approximations; these are our depiction and are revisable—yet what gets revised is the depiction, not the thing depicted. The computing itself is booked by no one; the world simply goes that way. Calling it a “way of booking” serves only to set it beside the cultural ledger, and the crucial difference lies exactly there: the cultural ledger has an author; this one has none. The difference is thus one of dependence: the physical ledger is unaffected by the cultural one—and the root of this is that one has an author whose pen can be made to write otherwise, while the other has no author and hence no pen to change. The test can be run on the spot—change “Zhang carried this brick” to “Li carried it,” or record nothing at all, and not one number in the brick’s trajectory changes; burn the site’s ledgers and the brick does not move an inch. Reverse it: alter one line of the computation and “carrying,” the building site, the brick, and the booking itself all fail to hold. Change one ledger and the other loses not a number; change the other and this one collapses entirely. The two do not conflict, because they do not record the same column: the cultural ledger records who can be held to it, the physical one records position changing with time; conflict arises only when the cultural ledger is read as a report about the world. The paper turns the alleged relation into a form, and every entry collapses. He caused it—his lifting, bending, releasing are themselves further stretches of a line; explaining one stretch by another produces no joint anywhere. He shifted the probabilities—produce the numbers; that computation has never once been performed. He was present—so was the air, so was the scaffolding. He laid it—“to lay” is transitive and carries a subject slot; what it adds is not physical content but that slot. And even granting that the trajectory is random rather than necessitated, nothing is lost: randomness does not say “a place is left for you,” it says “there is no determiner,” and that “no” includes him. A gap is not a vacant seat. The paper addresses an objection peculiar to bricks: the blueprint. The brick’s position matches a line drawn on paper three months earlier, and the whole building matches—the correspondence is real. But matching is an interlocking of two segments, not a joint on either line: blueprint and wall are both in this world and really do correspond; yet the corresponding happens when we compare them, not during the brick’s flight. The last rung of the gradient is “I raise my arm.” The particles in the arm are, likewise, related to no person. That stretch of arm is the same kind of thing as the brick—it is as real as the brick—only its line, like the brick’s, is everywhere continuous and contains no joint. As for “but I feel that I did it,” the feeling really occurred, yet it bears no stamp of its own: it is a fact about that feeling, not a fact about the line. A last word on the literature: it is not that no one has discussed objects, but that no one has read an object’s line as its own line. Objects have always been tracked in fine detail—lot numbers, chains of custody, who did what at which minute—yet every record is indexed by a person and absorbed into the category of the tool: the motion of things is treated as an extension of human action. It is not that no one looked; it is that by the time anyone looked, it was already filed under someone’s name.

一件东西不见了。我们说：我把它弄丢了。 本文要问的是这句话里那个”把”字。一件物体的移动轨迹，是一条实实在在的线；而在这条线上的某一处，我们安了一个人，说这条线因他而如此。本文的判断是：这个安放从来没有被交代过。 论证分两路走，而两路同向。 第一路：轨迹是必然的。 想想物理定律的作用就知道，一个物体的移动轨迹本来就是那一条。这时非要把这条轨迹与某个人扯上关系，其实并无必要——那条线本来就是那条，人加不进去。丢失的东西，其轨迹本来就会离开你身体的那条轨迹；这两条线在某处不再重合，而这一过程并不真的”因”任何人。 第二路：就算让一步，承认它是随机的。 有人会抬出量子随机性来证明轨迹非必然。好，姑且承认它非唯一而是随机的——可你想把这条随机轨迹与某个人扯上关系，还应讲出该人与这条随机轨迹有何关系。 随机说的不是”留了一个位置给你”，是”没有决定者”，而这个”没有”同样包括你。对手把一处缺口，读成了一个空座。 于是本文把那份关系做成一张必填表，而逐项都塌：说他造成了它——那就不随机了，他越把关系讲清楚，随机性越是被他自己取消；说他影响了概率——那请给出数，多少、怎么算的，而这份计算一次也没有人做过；说他在场——在场的东西多了，空气也在场，桌子也在场；说那是他的东西——所有权是一份分配，不是轨迹上的关系。 本文并指出，那个安放在语言里早已完成。“丢”这个字本身就是及物的，自带一个主语位置。 同一段线，说”石头滚下了山坡”没人追问谁的因；说”我丢了钥匙”，因已经安好了——多出来的不是物理内容，是一个主语位置。而分离本是对称的：“我丢了它”与”它离开了我”，物理上分毫不差，选哪一句不由物理决定，由轴决定；而轴在没人指定的时候，就落在说话的人身上。 那么这个因究竟是从哪儿来的？本文的诊断是：因的位置由”谁能被要求”决定，不由轨迹决定。 说”因你而丢”，办的是三件实事：谁去找、谁赔、下次谁小心。而这套分配的收件人必须是人——只有人能被要求、被赔付、被改进。反过来一验就清楚：没有收件人的时候，我们就不说因。风把纸吹走，没人问这是谁的因；只要有人在场，同一件事立刻长出一个因来。在场与否改变的不是物理，是有没有人可以收件。 本文并挡住一条常见的反驳：“可我确实把钥匙搁在了口袋沿上，那确实是这段轨迹的一部分。“是的——但它是轨迹的一部分，不是轨迹的原因。 整条线上每一段都同样是它的一部分；把其中一段挑出来叫”因”，挑的标准不在线上。因不是线上的一个点，是我们在线上挑出的一个点。 本文守一条必须先立的守则：本文取消的是”因”作为轨迹上一处接缝的地位，不取消分配。 谁去找、谁赔、下次谁小心，照旧运转，一分不减——只是它属于我们要办的事，不属于世界的报告。举证责任因此落在主张者一边：说这条轨迹跟某个人有关系的人，得讲出那是什么关系；而在讲出来之前，那一栏是空的，我们却一向把它当成已经填好的。 Abstract Something has gone missing. We say: I lost it. This paper asks about the transitive verb in that sentence. The trajectory of a physical object is a real line; and at some point along that line we install a person and say the line is as it is because of him. The paper’s claim is that this installation has never been accounted for. The argument runs along two tracks that point the same way. First track: the trajectory is necessitated. Consider what physical law does: an object’s trajectory is simply the one it is. Insisting on relating that trajectory to some person is unnecessary—the line was that line, and a person cannot be inserted into it. The trajectory of a lost object simply departs from the trajectory of your body; the two lines cease to coincide at some point, and this process is not truly caused by anyone. Second track: concede the point and allow that it is random. Some will invoke quantum randomness to show the trajectory is not necessitated. Very well—grant that it is not unique but random. Yet if you would relate this random trajectory to some person, you still owe an account of what relation that person bears to it. Randomness does not say “a place has been left for you”; it says “there is no determiner,” and that “no” includes you. The opponent has read a gap as a vacant seat. The paper turns the relation into a form that must be filled in, and every entry collapses. He caused it—then it was not random, and the clearer he makes the relation the more thoroughly he cancels the randomness himself. He shifted the probabilities—then produce the numbers: how much, computed how; and that computation has never once been performed. He was present—many things were present; so was the air, so was the table. It was his property—ownership is an allocation, not a relation along the trajectory. The paper further notes that the installation was completed in advance by language. “To lose” is transitive; it carries a subject slot with it. Of the same stretch of line: “the stone rolled down the slope” prompts no question about whose fault it was, while “I lost my keys” has the cause already installed—what has been added is not physical content but a subject position. And the separation is symmetric: “I lost it” and “it departed from me” are physically identical; which sentence is chosen is settled not by physics but by the axis—and where no one fixes the axis, it falls on whoever is speaking. Where, then, does the cause come from? The paper’s diagnosis: the cause is placed by who can be held to it, not by the trajectory. Saying “you lost it” accomplishes three practical things: who searches, who compensates, who is more careful next time. And the addressee of such an allocation must be a person—only a person can be held, charged, or improved. The reverse test makes it plain: where there is no addressee, we speak of no cause. When wind carries off a sheet of paper, no one asks whose fault it was; let a person be present, and the same event immediately grows a cause. What presence changes is not the physics but whether there is anyone to receive the allocation. The paper blocks a common objection: “but I did set the keys on the edge of my pocket, and that really is part of the trajectory.” It is—but it is part of the trajectory, not its cause. Every stretch of the line is equally part of it; to single out one stretch as “the cause” applies a criterion that is not on the line. The cause is not a point along the line but a point we pick out along it. One rule must be fixed first: the paper withdraws the standing of “cause” as a joint in the trajectory; it does not withdraw allocation. Who searches, who compensates, who is careful next time—all continue undiminished, but they belong to what we have to do, not to a report about the world. The burden accordingly rests with the claimant: whoever says the trajectory bears a relation to a person must say what relation that is; and until it is said, the column is blank, while we have always treated it as filled.

本文只处理一句常见的辩护。当有人指出”当初要是没那样”这类设想并没有发生过，通常会得到这样的回应：它虽然没发生，但它在物理上是可能的，并没有违反物理定律。 本文的判断是：这句话是一个需要计算才能支撑的断言，而这个计算从来没有人做过。 先说这句辩护为什么一向不必付账。一件实际发生过的事，在物理上一定是可能的——发生了，所以可能，这条不需要任何证明工作。可能性一直是搭着”实际发生”的便车来的。 而反事实场景里没有实际发生的东西，便车没了，可习惯还在。此处还有两条要点明：“没有禁令”不等于”有许可”——“不违反已知定律”是一句关于我们那套写法的话，不是关于世界的话；以及**“我一下子没看出哪里违反”不等于实际上没违反**——没看出来是一句关于看的人的话。相似同样不传递可能性：实际发生的那件事只给它自己发通行证，不给旁边站着的那些发。 那么要证的是什么？两件：该场景内部未违反物理定律，以及该场景与之前的实际情况的接缝处未违反物理定律。而程序是确定的：先在最细的颗粒度上把该场景定义出来，再从接缝处做拉普拉斯妖式的计算，最后看算出的接缝后场景与你所定义的是否完全吻合。吻合了，才说明没违反。 本文并逐一处理三样常被搬出来当许可证的东西，指出它们全都是任务书。其一，若在这件事所在的尺度上随机性不显现（日常反事实正是如此，微观涨落在这个尺度上被平均掉了），接缝处只有两条路：接缝那一刻的状态与实际的一模一样，那么算出来的就是实际历史，与你的场景对不上；或者不一样，可这一动不由前一刻算出，而不由前一刻算出的改动正就是”违反”的意思。其二，量子随机性并不构成豁免，因为随机性仍然是可以计算的——它给出的不是”什么都行”，是一族有确定分布的后继；程序一步没少，只是第三步从”看那一个吻不吻合”变成”看这一族里有没有一个吻合”。门槛降到这里，也仍然只剩一句：你算了吗？没算你凭什么说没违反？ 其三，混沌放大同样不帮忙：混沌给出的是发散，而发散不指向任何特定终点；发散越厉害，落到”我没搬家”这个指定结局上的活儿越重，不是越轻。何况主张”由某个微小涨落放大而成”的人，还欠着两笔：是哪一个涨落，沿哪条路径。 最后须把主张说准：本文不主张那个计算做不到——理论上输入取得到，思想实验也行。本文主张的是：没人做过。 这一条不依赖任何一套写法，决定论也好、量子随机也好、混沌也好，都得算，而三边都没算。所以那句辩护不是被驳倒了，是还没有被提出来：在计算拿出来之前，那一栏是空的，而我们一向把它当成已经填好的。 Abstract This paper addresses one common defence. When it is pointed out that a scenario such as “if I hadn’t done that” never happened, the reply is usually that although it did not happen, it is physically possible and violates no law of physics. The paper’s claim is that this is an assertion requiring a calculation to support it, and that calculation has never been performed by anyone. First, why the defence has never had to pay. Whatever actually happened is thereby physically possible—it happened, so it was possible, and no proof is called for. Possibility has always ridden along with actuality. A counterfactual scenario contains no actuality; the ride is over, but the habit remains. Two further points: the absence of a prohibition is not the presence of a permission—“violates no known law” is a remark about our notation, not about the world; and “I do not offhand see a violation” is not the absence of one—not seeing is a remark about the observer. Resemblance likewise transmits no possibility: what actually happened issues a pass to itself alone, not to whatever stands nearby. What, then, must be shown? Two things: that nothing within the scenario violates physical law, and that nothing violates it at the seam where the scenario joins the preceding actual course of events. The procedure is definite: define the scenario at the finest granularity; compute forward from the seam in the manner of Laplace’s demon; then check whether the computed post-seam scenario matches exactly the scenario as defined. Only then is no violation shown. The paper takes up three things regularly produced as licences and shows each to be a work order instead. First, where randomness does not manifest at the scale in question—as with everyday counterfactuals, microscopic fluctuation being averaged out at that scale—the seam admits only two routes: the state at the seam is identical to the actual one, in which case the computation yields actual history and fails to match your scenario; or it is not, in which case that alteration was not computed from the preceding instant, and an alteration not so computed is precisely what “violation” means. Second, quantum randomness grants no exemption, because randomness is still calculable—it yields not “anything goes” but a family of successors with a definite distribution. The procedure loses not a single step; only the third changes, from “does that one match?” to “does any member of the family match?” With the bar lowered this far, one question remains: have you calculated? If not, on what basis do you say no law was violated? Third, chaotic amplification does not help either: chaos yields divergence, and divergence points at no particular endpoint. The stronger the divergence, the heavier the work of landing on the one specified outcome, not the lighter. Moreover, whoever claims the outcome was amplified from some minute fluctuation owes two further items: which fluctuation, and along which path. Finally the claim must be stated exactly. The paper does not hold the calculation to be unperformable—inputs are in principle available, and a thought experiment will serve. The paper holds that no one has performed it. This does not depend on any particular notation: deterministic, quantum-stochastic, chaotic—all require the calculation, and on all three sides none was done. The defence has therefore not been refuted; it has not yet been made. Until the calculation is produced that column is blank, and we have always treated it as already filled in.

本文讲我们对待心里那些画面的真实方式。通常的说法是：我们要么信它，要么不信它。本文认为不是这样。 **默认不是"信"，是按真实入账，再减去我们不得不承认的那部分。** 一段"当初要是没那样"的画面出现时，如果没人质疑，它就整个挂在"真实"名下。有人指出"可是实际上你并没有那样啊"，我们才承认一处不同；再追问，再承认一处。承认了的那几处，是折扣；没被追问出来的部分，仍然挂在"真实"名下。 这里有三件事要说清楚。 **第一，折扣的大小，不由这段画面假的程度决定，由我们被逼的程度决定。** 折扣不是我们发现的，是被人抠出来的。没人抠，就一分不减。 **第二，开票的和抵扣的是同一个人。** 要算"这一段跟实际发生的差在哪里"，得拿两样东西来比；可比的两边都是心里的画面。这笔账没有外人来审。 **第三，也是最要紧的一条：虚构不是打了折的真实。** 折扣这条轴上量的是"差多少"，零点是完全属实，往下是有出入、大部分不实、几乎全错。可"几乎全错"仍然是**对一件发生过的事的错误描述**。而"当初我没那样"这段画面，不是对任何一件发生过的事的描述——它讲的那件事根本没有原件。所以它不在这条轴上。**在这条轴上一直减，得到的是"一件真事被说得很离谱"，永远得不到"没有这件事"。** 单位不对。 **第四，也是全篇的落点：不是"剔除掉不真实的部分之后，剩下的是真的"。** 那段画面里，那个城市确实存在，我确实有一副身体——可这些零件的真，是它们各自另有出处，是被借来当布景的；借来的真不算这张单子的真。一张假钞上的纤维是真纤维，油墨是真油墨，可没有"这张钞票有九成购买力"这回事，它的购买力是零。真实性不是画面里各个零件的属性，是整段画面所讲的那件事是不是那样；而它所讲的那件事从来没有过。所以**它的真实性不是"很低"，是零**。前面那些被抠出来的"不同之处"，因此也不是折扣——不是从一个正的真值里减掉的份额，而是布景被一件一件认出来的过程。**零不是减出来的，一开始就是零。** **第五，这不等于说这些画面没有意义或功能。** 它们能让人后悔得睡不着，能让人下决心，能让人第二天改主意——这些作用一分不假，确实发生了。但那是**这段活动**的属性，不是**它所讲的那件事**的属性。混淆出在同一个"真"字上："这话说得真"有时指贴切、打动人，有时指所述属实；前者说作用，后者说真实性。**这两样是两个互不相干的领域。** 一段画面不需要任何真实性就能起全部作用——后悔靠的不是它为真，是它被按真实入账了；反过来，一段完全属实却没人看的记录，作用是零。所以本文的结论要分两半说，免得被听成一半：**那些画面的真实性是零；那些画面的作用是实实在在的。** 这两句不冲突，因为它们说的不是同一件事。 **第六，"不真"也不等于"可能"。** 到这里通常有最后一次退守："就算它不真，它至少是可能的吧？"这一次要拆的不是"可能"这个词，是它**从哪里领的**。你**可以**去证明这些画面在物理上可能；但**在证明完成之前，你不应当认为它们在物理上可能**。而这份举证责任之所以一向没人负，是因为可能性一直是白拿的：**一件实际发生过的事，在物理上一定是可能的**——发生了，所以可能，这条不需要任何证明。**可能性一直搭着"实际发生"的便车。** 这段画面里没有实际发生的东西，便车没了，可习惯还在。此处还有两处要点明：**"没有禁令"不等于"有许可"**——"不违反已知定律"是一句关于我们那套写法的话，不是关于世界的话，我们手上没有禁令，跟世界发过许可证，是两件事；以及**相似不传递可能性**——实际发生的那件事只给它自己发通行证，不给旁边站着的那些发。最后须说清这一条主张什么：不授予"可能"，不等于断定"不可能"；本文只说**在证明拿出来之前，那一栏是空的**，而我们一向把它当成已经填好的。 本文并指出一类特别的画面：关于未来的走向。这一类连折扣都提不出来，因为没有东西可以拿来比——不是暂时缺资料，是明天还没发生。所以它们**按全额挂账，一分不减**。我们对未来那种"这是能实现的"的感觉之所以那么硬，不是因为我们对未来特别有信心，而是因为那些画面从来没有被扣过一分钱。反过来，"当初如果"那一类因为有一件真事可以拿来比，反而能被扣掉几项——而正因为它被打过折，它看起来更像一本真账。**能打折的，比不能打折的更骗人。** 这里要补一处难题。有人会说：明天我确实走了左边那条路，那昨天心里那段画面，难道不算真的吗？ 不算。请把两件事分开：一件是昨天那段画面，另一件是明天实际发生的事。两者相似，可"相似"是我们事后在心里比出来的——比的两边，一边是对那段画面的回忆，一边是对明天的回忆。**应验给的是事后的一致，不是当时的真实性。** 那段画面被放出来的时候，没有任何东西使它为真；它后来进入了一套合得上的关系，而这套关系是别处造出来的，不会回填给它。 于是这本账永远结不平：它里面没有一栏叫"作废"。看破一段画面是编的，不是把折扣一路加到十成，而是把这张单子从"真实"这本账上整个拿掉。**不是多减了一笔，是少了一栏。** ## Abstract This paper concerns how we actually treat the images in our minds. The usual account is that we either believe them or we do not. The paper argues otherwise. **The default is not belief but booking at face value, minus whatever we are forced to concede.** When an image of "if I hadn't done that" arises, and no one challenges it, the whole of it is entered under "real." Only when someone objects, "but in fact you did," do we concede one difference; pressed again, we concede another. What we concede is the discount; what was never pressed out of us remains entered under "real." Three things follow. **First, the size of the discount is set not by how false the image is but by how hard we are pressed.** The discount is not discovered; it is extracted. Unpressed, nothing is deducted. **Second, the one who issues the invoice and the one who allows the deduction are the same person.** To compute how this image differs from what actually happened, two things must be compared—and both sides of the comparison are images in the mind. No outside party audits this ledger. **Third, and most importantly: fiction is not discounted truth.** The scale of discounting measures how much something differs; its zero is complete accuracy, and below it lie minor inaccuracy, largely inaccurate, almost wholly wrong. But "almost wholly wrong" is still **a wrong description of something that happened**. The image "if I hadn't done that" is not a description of anything that happened—the event it speaks of has no original at all. It is not on the scale. **Subtract along this scale forever and you arrive at "a real event grossly misdescribed," never at "there is no such event."** The units do not match. **Fourth, and the paper's main point: it is not that what remains after the false parts are struck out is true.** The image contains a city that really exists, a body I really have—but the truth of these components is their own, borrowed from elsewhere and used as scenery; borrowed truth does not accrue to this invoice. The fibres in a counterfeit note are real fibres and the ink is real ink, yet there is no such thing as the note having ninety percent of its purchasing power: its purchasing power is zero. Truth is not a property of the components but of whether the thing the whole image speaks of is so—and that thing never was. Its truth is therefore **not low but zero**. The "differences" extracted earlier are accordingly not discounts subtracted from some positive truth-value, but the scenery being identified piece by piece. **Zero was not arrived at by subtraction; it was zero from the start.** **Fifth, none of this says the images lack meaning or function.** They can keep a person awake with regret, settle a resolution, change a mind by morning—and these effects are entirely real; they really occur. But they are properties of **this activity**, not of **the thing the image speaks of**. The confusion lies in one word: "true" sometimes means apt, moving, on the mark, and sometimes means accurate to what is. The first speaks of effect, the second of truth. **These are two unrelated domains.** An image needs no truth whatever to produce its full effect—regret works not because the image is true but because it was booked as real; conversely, a wholly accurate record no one reads has zero effect. The conclusion must therefore be stated in two halves: **the truth of these images is zero; the effects of these images are entirely real.** The two do not conflict, because they are not about the same thing. **Sixth, "not true" does not amount to "possible" either.** A last retreat usually follows: "granted it is not true, but surely it is at least possible?" What must be examined is not the word "possible" but **where the entitlement was drawn from**. One **may** try to prove these images physically possible; but **until that proof is complete, one should not regard them as physically possible**. The reason no one has ever shouldered this burden is that possibility has always come free: **whatever actually happened is thereby physically possible**—it happened, so it was possible, and no proof is required. **Possibility has always ridden along with actuality.** In these images there is no actuality; the ride is over, but the habit remains. Two further points: **the absence of a prohibition is not the presence of a permission**—"violates no known law" is a remark about our notation, not about the world, and having no prohibition in hand is not the world having issued a licence; and **resemblance does not transmit possibility**—what actually happened issues a pass to itself alone, not to whatever stands nearby. Finally, what this claim asserts must be stated exactly: withholding "possible" is not asserting "impossible." The paper says only that **until the proof is produced, that column is blank**—whereas we have always treated it as already filled in. The paper further isolates a special class: images of how the future may go. For these no discount can even be proposed, since there is nothing to compare them against—not a temporary lack of records, but that tomorrow has not happened. They are therefore **booked in full, with nothing deducted**. Our sense that a future course is realizable is so firm not because we have special confidence in the future but because those images have never had a penny deducted. Conversely, "if I hadn't" images can have items deducted, since a real event stands beside them for comparison—and precisely because they have been discounted, they look more like a genuine ledger. **What can be discounted deceives more than what cannot.** The ledger therefore never balances: it has no column for "void." To see that an image is invented is not to raise the discount to a hundred percent but to remove the entry from the ledger of the real altogether. **Not one more deduction, but one fewer column.**

本文讲我们对待心里那些画面的真实方式。通常的说法是：我们要么信它，要么不信它。本文认为不是这样。 默认不是”信”，是按真实入账，再减去我们不得不承认的那部分。 一段”当初要是没那样”的画面出现时，如果没人质疑，它就整个挂在”真实”名下。有人指出”可是实际上你并没有那样啊”，我们才承认一处不同；再追问，再承认一处。承认了的那几处，是折扣；没被追问出来的部分，仍然挂在”真实”名下。 这里有三件事要说清楚。 第一，折扣的大小，不由这段画面假的程度决定，由我们被逼的程度决定。 折扣不是我们发现的，是被人抠出来的。没人抠，就一分不减。 第二，开票的和抵扣的是同一个人。 要算”这一段跟实际发生的差在哪里”，得拿两样东西来比；可比的两边都是心里的画面。这笔账没有外人来审。 第三，也是最要紧的一条：虚构不是打了折的真实。 折扣这条轴上量的是”差多少”，零点是完全属实，往下是有出入、大部分不实、几乎全错。可”几乎全错”仍然是对一件发生过的事的错误描述。而”当初我没那样”这段画面，不是对任何一件发生过的事的描述——它讲的那件事根本没有原件。所以它不在这条轴上。在这条轴上一直减，得到的是”一件真事被说得很离谱”，永远得不到”没有这件事”。 单位不对。 第四，也是全篇的落点：不是”剔除掉不真实的部分之后，剩下的是真的”。 那段画面里，那个城市确实存在，我确实有一副身体——可这些零件的真，是它们各自另有出处，是被借来当布景的；借来的真不算这张单子的真。一张假钞上的纤维是真纤维，油墨是真油墨，可没有”这张钞票有九成购买力”这回事，它的购买力是零。真实性不是画面里各个零件的属性，是整段画面所讲的那件事是不是那样；而它所讲的那件事从来没有过。所以它的真实性不是”很低”，是零。前面那些被抠出来的”不同之处”，因此也不是折扣——不是从一个正的真值里减掉的份额，而是布景被一件一件认出来的过程。零不是减出来的，一开始就是零。 第五，这不等于说这些画面没有意义或功能。 它们能让人后悔得睡不着，能让人下决心，能让人第二天改主意——这些作用一分不假，确实发生了。但那是这段活动的属性，不是它所讲的那件事的属性。混淆出在同一个”真”字上：“这话说得真”有时指贴切、打动人，有时指所述属实；前者说作用，后者说真实性。这两样是两个互不相干的领域。 一段画面不需要任何真实性就能起全部作用——后悔靠的不是它为真，是它被按真实入账了；反过来，一段完全属实却没人看的记录，作用是零。所以本文的结论要分两半说，免得被听成一半：那些画面的真实性是零；那些画面的作用是实实在在的。 这两句不冲突，因为它们说的不是同一件事。 本文并指出一类特别的画面：关于未来的走向。这一类连折扣都提不出来，因为没有东西可以拿来比——不是暂时缺资料，是明天还没发生。所以它们按全额挂账，一分不减。我们对未来那种”这是能实现的”的感觉之所以那么硬，不是因为我们对未来特别有信心，而是因为那些画面从来没有被扣过一分钱。反过来，“当初如果”那一类因为有一件真事可以拿来比，反而能被扣掉几项——而正因为它被打过折，它看起来更像一本真账。能打折的，比不能打折的更骗人。 这里要补一处难题。有人会说：明天我确实走了左边那条路，那昨天心里那段画面，难道不算真的吗？ 不算。请把两件事分开：一件是昨天那段画面，另一件是明天实际发生的事。两者相似，可”相似”是我们事后在心里比出来的——比的两边，一边是对那段画面的回忆，一边是对明天的回忆。应验给的是事后的一致，不是当时的真实性。 那段画面被放出来的时候，没有任何东西使它为真；它后来进入了一套合得上的关系，而这套关系是别处造出来的，不会回填给它。 于是这本账永远结不平：它里面没有一栏叫”作废”。看破一段画面是编的，不是把折扣一路加到十成，而是把这张单子从”真实”这本账上整个拿掉。不是多减了一笔，是少了一栏。 Abstract This paper concerns how we actually treat the images in our minds. The usual account is that we either believe them or we do not. The paper argues otherwise. The default is not belief but booking at face value, minus whatever we are forced to concede. When an image of “if I hadn’t done that” arises, and no one challenges it, the whole of it is entered under “real.” Only when someone objects, “but in fact you did,” do we concede one difference; pressed again, we concede another. What we concede is the discount; what was never pressed out of us remains entered under “real.” Three things follow. First, the size of the discount is set not by how false the image is but by how hard we are pressed. The discount is not discovered; it is extracted. Unpressed, nothing is deducted. Second, the one who issues the invoice and the one who allows the deduction are the same person. To compute how this image differs from what actually happened, two things must be compared—and both sides of the comparison are images in the mind. No outside party audits this ledger. Third, and most importantly: fiction is not discounted truth. The scale of discounting measures how much something differs; its zero is complete accuracy, and below it lie minor inaccuracy, largely inaccurate, almost wholly wrong. But “almost wholly wrong” is still a wrong description of something that happened. The image “if I hadn’t done that” is not a description of anything that happened—the event it speaks of has no original at all. It is not on the scale. Subtract along this scale forever and you arrive at “a real event grossly misdescribed,” never at “there is no such event.” The units do not match. Fourth, and the paper’s main point: it is not that what remains after the false parts are struck out is true. The image contains a city that really exists, a body I really have—but the truth of these components is their own, borrowed from elsewhere and used as scenery; borrowed truth does not accrue to this invoice. The fibres in a counterfeit note are real fibres and the ink is real ink, yet there is no such thing as the note having ninety percent of its purchasing power: its purchasing power is zero. Truth is not a property of the components but of whether the thing the whole image speaks of is so—and that thing never was. Its truth is therefore not low but zero. The “differences” extracted earlier are accordingly not discounts subtracted from some positive truth-value, but the scenery being identified piece by piece. Zero was not arrived at by subtraction; it was zero from the start. Fifth, none of this says the images lack meaning or function. They can keep a person awake with regret, settle a resolution, change a mind by morning—and these effects are entirely real; they really occur. But they are properties of this activity, not of the thing the image speaks of. The confusion lies in one word: “true” sometimes means apt, moving, on the mark, and sometimes means accurate to what is. The first speaks of effect, the second of truth. These are two unrelated domains. An image needs no truth whatever to produce its full effect—regret works not because the image is true but because it was booked as real; conversely, a wholly accurate record no one reads has zero effect. The conclusion must therefore be stated in two halves: the truth of these images is zero; the effects of these images are entirely real. The two do not conflict, because they are not about the same thing. The paper further isolates a special class: images of how the future may go. For these no discount can even be proposed, since there is nothing to compare them against—not a temporary lack of records, but that tomorrow has not happened. They are therefore booked in full, with nothing deducted. Our sense that a future course is realizable is so firm not because we have special confidence in the future but because those images have never had a penny deducted. Conversely, “if I hadn’t” images can have items deducted, since a real event stands beside them for comparison—and precisely because they have been discounted, they look more like a genuine ledger. What can be discounted deceives more than what cannot. The ledger therefore never balances: it has no column for “void.” To see that an image is invented is not to raise the discount to a hundred percent but to remove the entry from the ledger of the real altogether. Not one more deduction, but one fewer column.

本文讲我们对待心里那些画面的真实方式。通常的说法是：我们要么信它，要么不信它。本文认为不是这样。 默认不是”信”，是按真实入账，再减去我们不得不承认的那部分。 一段”当初要是没那样”的画面出现时，如果没人质疑，它就整个挂在”真实”名下。有人指出”可是实际上你并没有那样啊”，我们才承认一处不同；再追问，再承认一处。承认了的那几处，是折扣；没被追问出来的部分，仍然挂在”真实”名下。 这里有三件事要说清楚。 第一，折扣的大小，不由这段画面假的程度决定，由我们被逼的程度决定。 折扣不是我们发现的，是被人抠出来的。没人抠，就一分不减。 第二，开票的和抵扣的是同一个人。 要算”这一段跟实际发生的差在哪里”，得拿两样东西来比；可比的两边都是心里的画面。这笔账没有外人来审。 第三，也是最要紧的一条：虚构不是打了折的真实。 折扣这条轴上量的是”差多少”，零点是完全属实，往下是有出入、大部分不实、几乎全错。可”几乎全错”仍然是对一件发生过的事的错误描述。而”当初我没那样”这段画面，不是对任何一件发生过的事的描述——它讲的那件事根本没有原件。所以它不在这条轴上。在这条轴上一直减，得到的是”一件真事被说得很离谱”，永远得不到”没有这件事”。 单位不对。 本文并指出一类特别的画面：关于未来的走向。这一类连折扣都提不出来，因为没有东西可以拿来比——不是暂时缺资料，是明天还没发生。所以它们按全额挂账，一分不减。我们对未来那种”这是能实现的”的感觉之所以那么硬，不是因为我们对未来特别有信心，而是因为那些画面从来没有被扣过一分钱。反过来，“当初如果”那一类因为有一件真事可以拿来比，反而能被扣掉几项——而正因为它被打过折，它看起来更像一本真账。能打折的，比不能打折的更骗人。 于是这本账永远结不平：它里面没有一栏叫”作废”。看破一段画面是编的，不是把折扣一路加到十成，而是把这张单子从”真实”这本账上整个拿掉。不是多减了一笔，是少了一栏。 Abstract This paper concerns how we actually treat the images in our minds. The usual account is that we either believe them or we do not. The paper argues otherwise. The default is not belief but booking at face value, minus whatever we are forced to concede. When an image of “if I hadn’t done that” arises, and no one challenges it, the whole of it is entered under “real.” Only when someone objects, “but in fact you did,” do we concede one difference; pressed again, we concede another. What we concede is the discount; what was never pressed out of us remains entered under “real.” Three things follow. First, the size of the discount is set not by how false the image is but by how hard we are pressed. The discount is not discovered; it is extracted. Unpressed, nothing is deducted. Second, the one who issues the invoice and the one who allows the deduction are the same person. To compute how this image differs from what actually happened, two things must be compared—and both sides of the comparison are images in the mind. No outside party audits this ledger. Third, and most importantly: fiction is not discounted truth. The scale of discounting measures how much something differs; its zero is complete accuracy, and below it lie minor inaccuracy, largely inaccurate, almost wholly wrong. But “almost wholly wrong” is still a wrong description of something that happened. The image “if I hadn’t done that” is not a description of anything that happened—the event it speaks of has no original at all. It is not on the scale. Subtract along this scale forever and you arrive at “a real event grossly misdescribed,” never at “there is no such event.” The units do not match. The paper further isolates a special class: images of how the future may go. For these no discount can even be proposed, since there is nothing to compare them against—not a temporary lack of records, but that tomorrow has not happened. They are therefore booked in full, with nothing deducted. Our sense that a future course is realizable is so firm not because we have special confidence in the future but because those images have never had a penny deducted. Conversely, “if I hadn’t” images can have items deducted, since a real event stands beside them for comparison—and precisely because they have been discounted, they look more like a genuine ledger. What can be discounted deceives more than what cannot. The ledger therefore never balances: it has no column for “void.” To see that an image is invented is not to raise the discount to a hundred percent but to remove the entry from the ledger of the real altogether. Not one more deduction, but one fewer column.

本文承接《故事B》与《三维的眼》，追到一个此前两文都停在其外的环节。《故事B》指出：要理解”甲因祸得福”，须自费建造一个永不可查证的世界；《三维的眼》指出：岔路的Y形是空间中的形状，选项一个也没有。两文都保留了一侧未动——“事实”那一侧，仿佛它是我们脚下的实地。本文要说：那一侧也是视频。 论证如下。关于过去的反事实，是我们能用心灵之眼看到的，就像看视频一样；关于未来的走向，同样是我们能用心灵之眼看到的，不论我们看到了几种走向、几个视频。而关于过去的事实本身，其实也是我们能用心灵之眼看到的，就像看视频一样。三者由同一台放映机放出，同样清晰，同样连贯，同样带着”这是真的”那种笃定；而我们把这些视频信以为真。 本文指出”记忆是取回过去”这一说法不能成立：若被取回的真是过去，那它就得是1947年那一段时空本身，而1947年那一段时空并没有被搬进任何人的头里。头里发生的，是此刻的一段活动，其形式与放映一段虚构无异。过去不在视频里，正如反事实不在视频里。 本文并指出妄想这一例的位置。若有妄想的话，视频内容就包含了妄想；而视频本身不会因此显得不同——它照样清晰、照样连贯、照样带着”这是发生过的”那种笃定。这一点通常被当作病理的特例，本文要说的相反：它是一枚显影剂，显出的是正常情况下的那份笃定，同样不是从视频内部挣来的。一段真的记忆之为真，不因为它里面有什么；它里面并没有一枚”真”的印章。 于是我们从形式上难以识别关于过去的事实与关于过去的反事实的不同，因两者都是视频而已。本文把”难以”推到”不可能”：区分所需的那样东西根本不在视频里。真与假不是视频的性质，是视频与视频之外某物的关系；而那个关系，从放映厅里永远看不到。 本文据此对前两文各作一处必要的修正。对《故事B》：A可查证、B永不可查证，这份不对称仍然成立，但须挪位——它在世界那一侧（1947年那笔钱留下了账目、人证、文件；乙还钱那件事没留下任何东西），不在看的那一侧（两段视频一样清晰、一样笃定）；而日常里我们几乎从不走出放映厅去核对，说”1947年甲把钱交给了乙”时，做的是放一段视频，不是查一本账。地位不对称，消费方式相同。 对《三维的眼》：该文第七节说”往过去的方向看，只看到一条已经发生的唯一的四维轨迹”，这里的”看到”须收回——我们往过去看，看到的也是一段视频，不是那条线；那条线唯一，这一点不来自看，也不需要看来支持。此修正不削弱该文，反而让它更稳：该文说定与开的不对称不在线上而在看上，现在可再进一程——在看这一侧，连不对称也没有：往过去是视频，往未来也是视频；差别只在于关于过去的那些视频，原则上还有个外部的东西可以去碰一碰，而关于未来的与关于反事实的，连碰的对象都没有。 本文守数条须格外郑重划定的界线：这不是”记忆都不可信”的怀疑论，本文不主张我们关于过去的说法多半是错的，只主张真假不由视频内部裁定，查证仍然有效——只是查证发生在放映厅之外；这更不主张妄想与事实等价，恰恰相反，正因为真假是视频之外的关系，视频之内的笃定就不足以作准，而这一条对妄想成立，对我们自己那些从未核对过的确信同样成立；至于记忆、证词、历史该如何在此之后被重新安放，本文不下判决，明确悬置为尚待研究。本文这一篇，自己也是一段视频；它没有让谁走出放映厅，它只是在放映厅的墙上，写了”放映厅”三个字。指可弃，月不可弃。 Abstract This paper continues Story B and The Three-Dimensional Eye, reaching a step at which both earlier papers stopped short. Story B argued that to understand “a blessing in disguise” one must build, at one’s own expense, a permanently unverifiable world; The Three-Dimensional Eye argued that the Y-shape of a fork is a shape in space and that there are no options at all. Both left one side untouched—the side of “fact,” as though it were solid ground beneath us. This paper argues that that side is a video too. The argument runs as follows. Counterfactuals about the past are things we can see with the mind’s eye, just as we watch a video; the courses the future might take are likewise things we can see with the mind’s eye, however many courses, however many videos. And the facts of the past themselves are also things we can see with the mind’s eye, just as we watch a video. All three are played by the same projector, equally vivid, equally coherent, equally attended by the conviction that this is real—and we take these videos to be real. The paper argues that “memory retrieves the past” cannot stand: if what were retrieved really were the past, it would have to be that stretch of spacetime in 1947 itself, and that stretch was never carried into anyone’s head. What occurs in the head is an episode occurring now, formally indistinguishable from the screening of a fiction. The past is not in the video, any more than the counterfactual is. The paper then places the case of delusion. Where delusion is present, the content of the video contains the delusion; yet the video does not thereby look any different—it remains as vivid, as coherent, as attended by the conviction that this happened. This is usually treated as a pathological special case; the paper argues the reverse: it is a developing agent, and what it develops is that in the ordinary case the conviction is not earned from inside the video either. A true memory is not true because of anything in it; it carries no stamp reading “true.” Hence we find it formally difficult to tell facts about the past from counterfactuals about the past, since both are merely videos. The paper presses “difficult” to “impossible”: what the distinction requires is simply not in the video. True and false are not properties of videos but relations between a video and something outside it—and that relation is never visible from inside the screening room. On this basis the paper offers a necessary correction to each earlier paper. To Story B: the asymmetry whereby A can be verified and B never can still holds, but must be relocated—it lies on the side of the world (the money of 1947 left ledgers, witnesses, documents; B’s returning the money left nothing), not on the side of the looking (the two videos are equally vivid, equally confident); and in ordinary life we almost never leave the screening room to check, so that in saying “in 1947 A gave the money to B” we are playing a video, not consulting a ledger. The standing is asymmetrical; the consumption is identical. To The Three-Dimensional Eye: its seventh section says that looking toward the past one sees a single unique four-dimensional trajectory already occurred—here “sees” must be withdrawn. Looking toward the past we too see a video, not the line; the line’s uniqueness does not come from the looking and needs no looking to support it. This correction does not weaken that paper but steadies it: it held that the asymmetry between fixed and open lies not in the line but in the looking; now one may go further—on the side of the looking there is no asymmetry either. Toward the past, a video; toward the future, a video. The only difference is that for videos about the past there is in principle something external one might go and touch, whereas for videos of futures and of counterfactuals there is nothing to touch at all. The paper guards several lines with particular gravity: this is not skepticism about memory—it does not claim that our accounts of the past are mostly wrong, only that truth is not adjudicated from inside the video, and verification remains effective, but takes place outside the screening room; still less does it hold delusion and fact to be equivalent—on the contrary, precisely because truth is a relation outside the video, conviction inside the video is insufficient warrant, and this holds for delusion and equally for those convictions of our own we have never checked; and how memory, testimony, and history ought to be re-situated thereafter, the paper renders no verdict, leaving it expressly still under study. This very paper is itself a video; it has let no one out of the screening room—it has only written the words “screening room” on the wall. The finger may be discarded, the moon may not.

本文承接《故事B》，追问一个更前置的环节：当我们比较故事A与故事B，并发表评论”乙可以把钱还给甲”时，这个”可以”由什么支撑。可供选择的只有两个，而两个都退票：其一，故事B已经真实发生过——那样它就不是”可以”而是”已经”，且《故事B》已表明B原则上永不可查证；其二，世界的物理结构容许B出现——可从来没有人在形式上成功地证明过这一点。我们所拥有的只是一组含自由参数的方程，而多解性住在我们的写法里，不住在世界里：世界只走了一个解，它不曾走过一族解，也不曾在一族解中挑过。方程的一般性本是我们不知道是哪一组值的记号，我们却把它读回世界，当成世界自己的复数。这是一张以形式证明为担保而从未兑付的欠条。 我们之所以从不索要这张欠条，是因为我们不是推出了它，是看见了它。我们感觉看到了，就是一个option：我们看到脚下的道路在前方分成两个岔，看到这两个岔，就感觉这两个岔似乎是两个option，而option给我们一种可实现感。本文指出这里有一次范畴错置：岔路的Y形是一件此刻摆在空间中的东西，不是一份未来的清单；在四维中，属于我的那个Y根本不存在，我只是一条不分岔的曲线，分岔的是路而不是我，而”路分岔了”是一件关于沥青的事实。 本文并对这一洞见作一处必要的推进：不只是另一个option是假的，而是剩下的那一个也不是option。说我的轨迹所经过的那条岔道”真正可实现”，只是重复说它实现了；而”选项”是一个要求复数的词——一份只有一项的清单不是清单，一个没有替代者的选项不是选项。于是两边都塌：假的那个是空间形状的误读，真的那个是同语反复。选项一个也没有。 本文另指出：未走的那一支，在四维中不过是我的轨迹从未经过的一片区域，而这样的区域数量无穷——仙女座星系是一片，三世纪的洛阳是一片。它凭什么被叫做option而仙女座不是？只凭空间上离我近、时间上排我前面。邻近不是可能，近不是能；所谓选项，不是世界中的一类对象，是一种距离关系被读成了模态地位。本文并指出唯一性不因看的方向而改变：往过去看是一条唯一的四维轨迹，往未来看不论看到几个option，都并不代表真实的未来轨迹是这几个；并不是只往过去看才唯一，而往未来看不唯一——往未来看不唯一的，是option而已。定与开的不对称不在线上，在看上；我们把”不知道是哪一条”换算成了”有好几条”，而这一步换算从未获得授权。 本文并指出”看到”是借来的动词，而借来处并没有可看的东西：你现在已经在头脑中看到了前方的道路分岔，虽然你的眼前真正看到的只是屏幕上的这篇论文——那里没有沥青，没有光子，没有对象。这不是一次对象缺席的观看，是根本没有观看；而”看到了什么”这个句式一用上，那个”什么”就凭空取得了地位。甲在把钱交出去之前”看到”乙还钱，1950年甲”看到”自己当上盐场老板——这两幅都没有出现在四维时空中。我们却给这种看图改了名字，叫它分析、推理、考虑各种可能，一改名它就取得了论证的身份。其实我们只是在看，只是在虚构地看，只是在看虚构。 本文最后指出：选项被端上来，是因为主角不能出场——我们无法预测自己的坐标在四维时空中的未来轨迹，因而无法谈论它，只好转而谈论眼前在空间中所看到的，似乎那就是未来；替身上台太久，我们忘了主角从未出场。同一层遮蔽也向后蔓延：我们看到四维中真实的过去轨迹时，也没意识到其唯一真实性，而是立刻读成”诸多可能中实现了的那一个”——这个说法已预设了未实现的其余，而其余从未被出示过。未来被预先复数化，过去被追认复数化，唯一的那条线从未被当成唯一。我们缺少四维的眼；但补上四维的眼也不会看到选项，它只会看到那条线，而”选项”一词到那时将无所指——这才是”看错”的分量：错的不是看漏了几个选项，是压根没有选项可漏。 本文守数条须格外郑重划定的界线：这一诊断不否认选项语言在决策、计划、制度中的实际功用；它不是”未来已被写定因而可以被知道”的预测神话——不可知与不唯一是两回事，本文所主张的正是不可知而唯一；至于审议、责任、后悔该如何在此之后被重新安放，本文不下判决，明确悬置为尚待研究。本文这一篇，自己也是一幅被心眼看见的图；它是指月之指；指可弃，月不可弃。 Abstract This paper takes up a question prior to the one in Story B: when we compare story A with story B and say “B could have returned the money to A,” what supports that “could”? Only two supports are available, and both fail. First, that story B actually occurred—but then it would not be “could” but “did,” and Story B has shown B to be in principle permanently unverifiable. Second, that the physical structure of the world permits B—yet no one has ever formally demonstrated this. What we possess is a set of equations with free parameters, and the plurality of solutions lives in our notation, not in the world: the world took one solution; it never traversed a family of solutions, and never picked among them. The generality of the formalism is a token of our not knowing which values obtain, and we read that generality back into the world as the world’s own plurality. This is an IOU underwritten by a formal proof that was never delivered. We never call in the IOU because we did not infer the possibility—we saw it. What we feel we see, we take to be an option: we see the road ahead divide in two, and seeing the two branches we feel that the two branches are two options; options give us a feel of realizability. Here lies a category mistake: the Y-shape of the fork is a thing standing now in space, not a menu of futures. In four dimensions there is no Y belonging to me; I am a single curve that does not fork. The road forks, I do not—and that the road forks is a fact about asphalt. The paper presses the insight one step further: it is not only that the other option is counterfeit, but that the remaining one is not an option either. To say that the branch my trajectory passes through is “genuinely realizable” is only to repeat that it was realized; and “option” is a word requiring plurality—a list with one item is not a list, and an option with no alternative is not an option. Both sides thus collapse: the false one is a misreading of a spatial shape, the true one a tautology. There are no options at all. The paper further notes that the branch not taken is, in four dimensions, merely a region my trajectory never enters—and such regions are infinite in number: the Andromeda galaxy is one, third-century Luoyang is another. What entitles one of them to be called an option while Andromeda is not? Only spatial nearness and temporal priority. Nearness is not possibility; being near is not being able. An option is not a kind of object in the world but a distance relation misread as a modal status. Uniqueness, moreover, does not change with the direction of looking: toward the past one sees a single unique four-dimensional trajectory; toward the future, however many options one sees, they do not constitute the real future trajectory. It is not that looking backward yields uniqueness while looking forward does not—what is non-unique when we look forward is only the options. The asymmetry between fixed and open lies not in the line but in the looking; we have converted “not knowing which one” into “there being several,” and this conversion was never licensed. The paper also argues that “to see” is a borrowed verb, and that where it is borrowed to, there is nothing to see. You have already seen, in your mind, the road ahead dividing, though what your eyes actually see is this paper on a screen—there is no asphalt there, no photons, no object. This is not a seeing whose object is absent; it is no seeing at all, and once the construction “what did you see?” is applied, the “what” acquires standing out of nothing. Before handing over the money, A “saw” B returning it; on learning of the 1950 policy, A “saw” himself as a salt-field owner. Neither ever appeared in four-dimensional spacetime. Yet we rename this looking-at-pictures analysis, inference, weighing the possibilities—and with the renaming it acquires the standing of an argument. In truth we are merely looking; looking fictively; looking at fiction. Finally: options are brought on stage because the protagonist cannot appear. We cannot predict the future trajectory of our own coordinate in four-dimensional spacetime, so we cannot speak of it, and speak instead of what the eye sees in space, as if that were the future; the stand-in has held the stage so long that we forget the protagonist never came on. The same occlusion spreads backward: seeing the real past trajectory in four dimensions, we likewise fail to register its unique reality, reading it at once as “the one among many possibilities that was realized”—a phrase presupposing an unrealized remainder never exhibited. The future is pluralized in advance, the past retroactively, and the one line is never taken as one. We lack the four-dimensional eye; but the four-dimensional eye would not see options either—it would see the line, and by then the word “option” would have nothing to refer to. That is the weight of the error: not that we overlooked some options, but that there were none to overlook. The paper guards several lines with particular gravity: this diagnosis denies the language of options none of its practical utility in decision, planning, or institutions; it is not the predictive myth that the future, being fixed, is knowable—unknowability and non-uniqueness are two different things, and the position here is precisely that the line is unknowable yet unique; and how deliberation, responsibility, and regret ought to be re-situated thereafter, the paper renders no verdict, leaving it expressly still under study. This very paper is itself a picture seen by the mind’s eye; it is a finger pointing at the moon; the finger may be discarded, the moon may not.

本文追一个从未被正面处理过的环节：**"理解"一句评论，究竟需要我们做什么。** 论证从一个具体的历史情境出发：**1947年，甲把所有积蓄交给朋友乙保管；之后甲准备买下一个盐场，但乙不承认甲的钱在他处保管，吞没了甲的钱，甲没开成盐场。但1950年的政策规定盐场老板等类人群都要枪毙，甲因祸得福。** 本文指出：**当我们思考这个故事（A）时，如果我们自称理解了这个故事，则我们必须想象出一个当了盐场老板而被枪毙的甲，存在于一个与1950年的中国几乎一模一样的世界中。** 这个被想象出来的世界，就是**故事B**——而它的虚构性，对我们几乎是隐形的。**在未被质疑的情况下，我们会感觉这个被枪毙的甲所在的世界就是1950年的真实世界**；若被质疑"真实世界里甲未被枪毙啊"，我们会退一步说，这个真实世界里有一处不同，那就是甲由未被枪毙变成了被枪毙。**可再仔细想想，不同之处还应有一些，比如乙把钱还给了甲（而不是没还）。** 而这条线一旦拉开，就停不下来：乙若还了钱，乙便不是那个会吞没朋友积蓄的乙，甲当初的托付、二人的关系，全都跟着变了；甲若从1947到1950一直是盐场老板，这三年的社会位置、来往的人、家人的处境，同样全部不同——**故事B从来不是"1950年的中国加一处修改"，而是一个从1947年就开始分岔、且在除"甲被枪毙"这一点之外的所有地方完全未被定义的世界。** 本文进而指出这份虚构性何以隐形：**故事B之所以难以被识破，恰恰因为它被刻意造得尽量贴近真实世界**——我们造B时只改动一个点，其余全部沿用A的世界，而正是这份"最大程度的沿用"，让B在感觉上几乎就是那个真实世界；**相似性，本是为了让B可理解、可对照，却同时成了它虚构性的最佳掩护：越像真的，越不像是被造出来的。** 本文另指出A与B在证据地位上的绝对不对称：**故事A的事实部分可被查证**（有账目、人证、政策文件），**而故事B永远无法被查证**——不是因为暂时缺资料，是因为没发生过的事不会留下任何记录，原则上不存在任何证据能支持或反驳它；可我们说出B时的笃定语气，与陈述A时几乎无异，这份**语气上的平权**，是虚构性隐形的另一半原因。本文并对A作一处必要的修正：严格讲，A本身也已是一次筛选（从1947至1950那片无限复杂的实际发生里，只挑出了钱、盐场、政策几条线），因此准确的对照不是"真实世界对虚构故事"，而是**"一个真的故事对一个假的故事"**——A的真，是"所述属实"，不是"未经中介地直通实在"。本文最后点明代价：**"因祸得福"是一次价值涂色，而要给一段真实历史涂上这层色，我们必须先自费建造一座完整的、永不可查证的世界，作为涂色的参照系；而这座建筑物，因为造得太像真的，我们从未把它当成建筑物看待过。** 本文守数条须格外郑重划定的界线：这一诊断不否认评论性语句在生活中的实际功用；不构成"故一切评价都不成立"的裁决；至于因果、责任、评价该如何在此之后被重新安放，本文不下判决，明确悬置为**尚待研究**。本文这一篇，自己也是一个为了理解而被建造出来的故事；它是指月之指；指可弃，月不可弃。 ## Abstract This paper examines a step never directly addressed: **what "understanding" an evaluative statement actually requires of us.** The argument proceeds from a concrete historical situation: **in 1947, A entrusts his entire savings to his friend B for safekeeping; A then prepares to buy a salt field, but B denies holding A's money and embezzles it, so A never opens the salt field. Yet in 1950, policy dictates that salt-field owners and similar classes be executed—A's misfortune proves a blessing in disguise.** The paper argues: **when we consider this story (A), if we claim to have understood it, we must imagine an A who became a salt-field owner and was executed, existing in a world almost exactly like China in 1950.** This imagined world is **story B**—and its fictionality is nearly invisible to us. **Unchallenged, we feel that the world containing this executed A simply is the real world of 1950**; challenged with "but in the real world A was not executed," we retreat one step and say this real world differs in one respect: A goes from not-executed to executed. **Yet on further reflection there must be other differences too—for instance, B returned the money rather than keeping it.** And once this thread is pulled, it cannot be stopped: if B returned the money, B is not the man who would embezzle a friend's savings, so A's original entrustment and the pair's relationship both change; if A were a salt-field owner from 1947 to 1950, his social position, his associates, his family's circumstances across those three years all differ—**story B was never "China in 1950 plus one modification," but a world diverging from 1947 onward and left wholly undefined everywhere except at the single point of A's execution.** The paper then explains why this fictionality stays hidden: **story B evades detection precisely because it is deliberately built to resemble the real world as closely as possible**—we alter one point and inherit everything else from A's world, and this maximal inheritance makes B feel like the real world itself; **the resemblance, intended to make B intelligible and comparable, becomes the best cover for its fictionality: the more real it looks, the less it looks constructed.** The paper further notes an absolute asymmetry in evidential status: **story A's factual portion can be verified** (ledgers, witnesses, policy documents), **whereas story B can never be verified**—not from a temporary lack of records, but because what never happened leaves no records, so that in principle no evidence could support or refute it; yet we assert B with a confidence scarcely distinguishable from our assertion of A, and this **parity of tone** is the other half of why the fiction stays hidden. The paper also offers a necessary correction regarding A: strictly speaking, A is itself already a selection (from the infinitely complex actuality of 1947–1950, only the money, the salt field, and the policy were picked out), so the accurate contrast is not "real world versus fictional story" but **"a true story versus a false one"**—A's truth consists in accuracy of what it states, not in unmediated access to reality. The paper closes on the cost: **"a blessing in disguise" is an act of evaluative coloring, and to apply this color to a stretch of real history we must first build, at our own expense, an entire, permanently unverifiable world to serve as the reference frame for that coloring—and this edifice, built too well to look built, we have never regarded as an edifice at all.** The paper guards several lines with particular gravity: this diagnosis denies evaluative statements none of their practical utility; it constitutes no verdict that all evaluation is void; and how causation, responsibility, and evaluation ought to be re-situated thereafter, the paper renders no verdict, leaving it expressly **still under study**. This very paper is itself a story built in order to be understood; it is a finger pointing at the moon; the finger may be discarded, the moon may not.

本文立一个极简、却足以撼动一整类思考习惯的判断，并把它贯彻到底：**我们对事件动力的分析，是一种解释，解释的对象是不内带任何动力的发生本身；因此不要把对事件动力的分析当真。** 这把刀必须被彻底贯彻——**发生里，甚至没有"相互的作用"；看上去似乎存在的相互作用，是我们感觉、解释出来的。** 我们习惯说"空气被台球推开"，可**空气和台球同时改变了位置，并不等于台球推开了空气：前者是发生，后者是解释。** 同样，**这一判断并非在说"重力不存在"——重力恰恰是我们捏造出来的解释；把重力用时空弯曲替代，同样是解释：我们既看不到重力，也看不到时空弯曲。** 刀还须落到时间上：**时间的存在，是对变化的某种解释——我们其实看不到时间的存在；我们先把时间解释、想象成类似空间维度的东西，再用这种东西来替代时间。** 本文进而把这一诊断，扩展到承载一切的那个坐标框架本身：**我们看到各种物体，然后想象它们位于一个由三条相互正交的线所度量的空间／容器／盒子中，给它们这个盒子里的坐标，然后就神奇地感觉这个盒子似乎真的存在了——这个盒子并不真的存在，而只是一种解释／想象。我们再添加一条正交的线来代表"时间"，然后感觉时间就似乎真的存在了——这个四维的新盒子也并不真的存在。** 关键在于，是**先有盒子才给得出坐标，不是坐标证明了盒子**。而本文由此正面回应一个最强的反驳——"可物理公式管用啊"：**即使套用物理公式时，我们想象这个四维盒子的存在，然后物体在这四维上的坐标恰好符合我们所套用的物理公式，这时我们感觉物理公式和四维盒子同时被证明存在了。并没有！只是我们假设它俩存在后，我们的后续推理、预测，仍能自圆其说。但自圆其说的东西／系统，并不证明其中所有要素的存在性。** 剥净之后，"发生本身"只剩并置的变化，没有作用、没有力、没有作为容器的时空。本文进而处理一个实际需求：若非要谈动力，该安放在哪里？**做事的是 condition，而不是人这样的 agent——不需要看到 agent，只需要看到 condition，穿过 agent 的边界。** 这是改进而非又一次虚构，因为 agent 要求一道封闭的边界（而这边界是我们划出的），condition 则不要求，它可以穿过皮肤与系统壁，少虚构一层。本文守数条须格外郑重划定的界线：这不否认物理学的实用效力与预测能力，所拆的仅是把"规律性"读作"内在动力"、把"自洽好用"读作"要素实存"这两步；**"不要当真"不等于"不要再做"——方法照用，地位下调**；这也不构成"故一切解释都无价值"的虚无主义。至于责任、归因这类实践该如何被重新安放，本文不下判决，明确悬置为**尚待研究**。本文这一篇，自己也是一层叠加在发生之上的解释；它是指月之指；指可弃，月不可弃。 ## Abstract This paper advances a claim of extreme simplicity and carries it through without exception: **our analysis of an event's dynamics is an explanation, and what it explains is occurrence itself, which carries no intrinsic impetus; therefore, do not take analyses of event-dynamics as real.** The blade must be applied thoroughly—**within occurrence there is not even "interaction"; what appears to be interaction is something we feel and explain into being.** We habitually say "the air is pushed aside by the billiard ball," yet **the air and the ball changed position simultaneously; this does not amount to the ball pushing the air: the former is occurrence, the latter explanation.** Likewise, **this is not saying "gravity does not exist"—gravity is precisely an explanation we fabricated; replacing it with spacetime curvature is equally an explanation: we see neither gravity nor spacetime curvature.** The blade must fall on time as well: **the existence of time is a certain explanation of change—we do not actually see time existing; we first explain and imagine time as something akin to a spatial dimension, then substitute that for time.** The paper then extends this diagnosis to the very coordinate framework that carries everything: **we see various objects, then imagine them located within a space/container/box measured by three mutually orthogonal lines, assign them coordinates within this box, and then marvelously feel that this box seems really to exist—yet the box does not really exist; it is merely an explanation, an imagining. We then add a fourth orthogonal line to represent "time," and feel that time too seems really to exist—yet this new four-dimensional box does not really exist either.** The crux is that **the box comes first and makes coordinates possible; coordinates do not prove the box.** The paper thereby confronts the strongest rebuttal—"but the physics works": **even when applying physical formulae, we imagine this four-dimensional box's existence, and when objects' coordinates along these four dimensions happen to satisfy the formulae we apply, we feel that both the formulae and the box have been proven to exist. They have not! It only shows that, having assumed both exist, our subsequent reasoning and predictions remain self-consistent. But a self-consistent thing or system does not prove the existence of all its constituent elements.** Once stripped, "occurrence itself" retains only juxtaposed change—no action, no force, no spacetime as container. The paper then addresses a practical need: if one must speak of dynamics, where to locate it? **What does things is condition, not an agent such as a person—one need not see the agent, only the condition, crossing the agent's boundary.** This is an improvement rather than one more fabrication, because agent demands a closed boundary (drawn by us) while condition demands none, crossing skin and system walls alike, fabricating one layer less. The paper guards several lines with particular gravity: this denies neither physics' practical efficacy nor its predictive power—what it dismantles is solely reading "regularity" as "intrinsic impetus," and reading "self-consistent and useful" as "its elements exist"; **"do not take it as real" does not mean "do not do it"—retain the method, demote its status**; nor does this constitute nihilism about explanation. As to how practices of responsibility and attribution ought to be re-situated, the paper renders no verdict, leaving it expressly **still under study**. This very paper is itself a layer of explanation superimposed upon occurrence; it is a finger pointing at the moon; the finger may be discarded, the moon may not.

本文提出对话的一项隐秘风险：当听到的话让人觉得有道理时，这句话便在那一刻获得与自己的内心独白同等的可靠感，被当作自己的思想去调用与承重；然而它是相对陌生的思想，未经自己的测试史检验，驾驭不了它的人会摔跤——如在不熟悉的地形上行走，无人行骗，误导照旧发生。自己的思想是熟的，别人的思想是生的，但我们不感觉它像生的：认知系统登记的是整合的顺滑度，而非来源与测试史。与人工智能的对话把这一风险推至极端：人工智能没有骗人，它的话语甚至比你自己的话语更像你的——它是减去了摩擦的你的声音——归化手续在抵达的瞬间已经办完；其风险的正确模型不是毒物而是换水：整缸混合水可以让鱼生病，纵然新水无毒，只因换水速率超过了代谢速率。本文进而给出时代诊断：思想的输入速率与消化速率之比已改变数量级，而消化端由生理定速——这是一个前所未有的、思想没有时间熟化的时代；我们由此重返婴儿期，必须像婴儿学习使用四肢那样，靠误差信号去适应大脑里新增的内容。文末给出思想检疫的操作方法，并指明熟化的本体论地位：消化不是自我在吸收思想，消化过程本身在铸造下一个自我；困水期不是等待成熟，困水期就是成熟。 Abstract This paper identifies a hidden risk of dialogue: the moment an utterance feels plausible, it acquires the same felt reliability as one’s own inner monologue and is put to use as one’s own thought. Yet it is a relatively foreign thought, unverified by one’s own history of trials; one may prove unable to handle it and may stumble—as one stumbles on unfamiliar terrain—misdirection without any deceiver. One’s own thoughts are cooked; others’ thoughts are raw; but the raw does not feel raw, because cognition registers smoothness of integration rather than provenance or testing history. Dialogue with artificial intelligence pushes this risk to its extreme: the AI does not deceive; its discourse is even more like you than your own—your voice minus friction—so that naturalization is complete upon arrival. The correct model of the risk is not poison but water change: mixed water can sicken the fish even when the new water is clean, because the rate of change exceeds the rate of metabolism. The paper then diagnoses the era: the ratio of input rate to digestion rate has shifted by orders of magnitude, while digestion is physiologically rate-fixed—an unprecedented era in which thought has no time to cure. We have thereby returned to infancy: we must adapt to the new contents of our brains by error signals, as infants learn the use of their limbs. The paper closes with operational methods of quarantine and with the ontological status of curing: digestion is not a self absorbing thoughts; the process of digestion is what forges the next self. The quarantine period is not a wait for maturity; the quarantine period is maturation itself.

本文是《许可的妄想》的姊妹篇，为其补足前提。前篇论证系统化哲学与临床妄想同构、仅隔一纸文化许可；本文追问哲学是从什么之中提纯而来，以及提纯意味着放弃了什么。本文提出：常人的思想不是一个系统，而是一堆互不咬合、彼此牵制的妄想的杂多；它之所以不崩坏，恰因其不成系统——内部矛盾之多，使任何单一读数都无法统摄全局，诸味相互抵消，故无一味走向极端。系统化哲学则是从这团杂多之中，抓住一条读数、剔除其上全部矛盾配重、令其贯彻到底的产物。此即提纯：原则愈纯，而那些使杂多得以安稳的相互牵制愈被弃尽。由此，纯度成为贯穿三者的唯一变量：临床妄想、系统化哲学、常人思想，依”单一原则统摄全局的纯度”由高至低排列。纯度既是”深刻”的度量，亦是”逼近病理”的度量——二者本是同一根标尺。常人之”清醒”不来自看清，而来自不纯；哲学家为深刻所付之代价，是自愿将自己推向谱系中纯净而危险的一端。 Abstract This paper is the sister to “Licensed Delusion,” supplying its missing premise. The former argues that systematic philosophy and clinical delusion are isomorphic, divided only by a sheet of cultural license; this paper asks out of what philosophy is refined, and what refinement entails relinquishing. It proposes that ordinary thought is not a system but a manifold of mutually incompatible, mutually restraining delusions; it does not collapse precisely because it is not systematic — the abundance of internal contradictions prevents any single reading from subsuming the whole, so the many flavors cancel one another and none proceeds to an extreme. Systematic philosophy is the product of seizing one reading out of this manifold, stripping from it every contradictory counterweight, and carrying it through to the end. This is refinement: the purer the principle, the more thoroughly discarded are the mutual restraints that had kept the manifold stable. Purity thus becomes the single variable running through all three: clinical delusion, systematic philosophy, and ordinary thought arrange from high to low along “the purity with which a single principle subsumes the whole.” Purity is at once the measure of “profundity” and the measure of “proximity to pathology” — the two are one and the same rod. The “lucidity” of ordinary people comes not from seeing clearly but from impurity; the price the philosopher pays for depth is to push himself, willingly, toward the pure and dangerous end of the spectrum.

本文提出：系统化哲学与临床妄想是同一种活动。二者同以单一原则统摄万象，同对反例施行辩护性吸收，其持有者同样终生无法自其体系中抽身。区分二者的，既非结构，亦非真假，而仅是一纸文化许可：哲学落在现代文化的协调均衡之内，遂领得"思想"之名；另一种落在均衡之外，遂领得"病"之名。本文据此论证，"病非病"一语的真义不是为病脱罪，而是揭示——被尊为哲学者与被诊断为妄想者，乃同一种心灵活动，只隔着一层文化的命名。文章依次剖析妄想的三项结构特征如何在哲学中逐一重现，说明"可共享性"何以是唯一的分界线而非健康的标志，并指出既有的两道辩护（体系可被主动修改、哲学家生活未受损）如何本身即许可机制的运作。 Abstract This paper proposes that systematic philosophy and clinical delusion are one and the same activity. Both subsume all phenomena under a single principle, both practice the defensive absorption of counter-evidence, and the holders of both are alike unable, throughout their lives, to step outside their system. What separates the two is neither structure nor truth but a mere cultural license: philosophy falls within the coordination equilibrium of modern culture and thereby earns the name "thought," while the other falls outside that equilibrium and earns the name "illness." On this basis the paper argues that the true import of the phrase "sickness is not sickness" is not to acquit illness but to reveal that what we revere as philosophy and what we diagnose as delusion are one and the same activity of mind, divided only by a layer of cultural naming. The paper shows in turn how the three structural features of delusion each recur within philosophy, why "shareability" is the sole line of division rather than a mark of health, and how the two standard defenses (that a system can be actively revised; that the philosopher's life is unimpaired) are themselves the operation of the licensing mechanism.

本文合并七记洞见，把"自然延续"这一概念，从一个看似客观独立的标准，逐层还原成它真正的身份——**我们对自己预测的一次事后追认**，并进一步揭示，这份预测所依赖的方法，正是我们做一切计划、谋划、乃至后悔与犹豫时，共享的同一套通用机制：**功能卡推演**。论证从一个精确的物理例子起步：台球沿斜坡滚下，是最接近普适物理定律的可靠知识，可我们不会说小球本该在斜坡上静止不动——与之相对，"某人的一句话对另一人的功能"这类描述，远不如物理因果可靠，却被感知成同等可靠，且唯有后者会滑向"本该不同"这一步判语。本文追问这份差异的来源：**斜坡台球之所以不产生自然延续本应不同的感觉，是因为物理状态设定简单，我们难以作出错误的预测**；一旦情形复杂——乌云翻滚——我们便可能预测暴雨，若不下，便觉得现实"偏离"了自然延续；**可其实不下暴雨，只是偏离了错误的预测。所谓"自然延续"，只是我们做出的预测，无论对错，都被感知为自然延续本身**——这一诊断，收在一个归谬上：若真准确预测到自然延续，现实又怎能偏离它？本文进而追问，为什么预测出错时，我们责怪的是一个"外来干扰"，而非自己：**狂风没有给我们线索，我们便把"没有线索因此没有预测到"，错读成了"位于待预测的系统之外"——可没有线索，只说明认知失败，不构成关于世界结构的证据。** 本文由此揭示我们实际预测的机制：**这种所谓的预测，其实只是我们心中根据模块的"功能"而做的推演，比如"翻滚的乌云有产生暴雨的功能"；狂风这个模块，虽具有"吹散乌云"的功能，但此时不在我们的视野之内，因此没有进入我们的推演，于是被感知为似乎位于系统之外，实际只是位于推演之外。恰如在我们的推演中，关公可以战秦琼，因为两者都被想象为具有某些武力功能的模块，因此可以推演其相互作用——其实就是想象这些功能是如何被使用的。** 本文进而把"系统边界"这个概念，逼到彻底的悖论——若系统是整个宇宙，没有狂风的系统在物理上是个悖论；若系统缩小到一个小范围，小系统的自然延续又如何定义？不考虑边界外情况的所谓"预测"，与预测关公战秦琼的胜负，又有何不同？关公战不了秦琼，小系统也并非独立。本文由此指出，这套功能卡推演，**也被我们用来在格斗中思考自己的招式，在与他人的非暴力对抗中思考自己的对策，在旅游时计划明天的活动安排，在团队中规划任务的分配——这是我们做任何计划、谋划的基础，因此它显得如此自然，似乎在直接处理现实本身，而非只是处理自己头脑之中的模块。可是，只有在思考台球滚下斜坡这样的简单配置时，头脑之中的模块才能与现实的物理演变严丝合缝；这也解释了为什么我们对复杂配置的推演总不能与现实吻合——这并不说明现实脱离了自然延续，而是说明我们的推演脱离、误判了自然延续，且在复杂配置下，无论采用多么精妙的思维模块组合，也不可能吻合现实的物理演变，因为我们没有采用拉普拉斯妖式的计算方式。** 本文最后把这套机制，用在两种最贴身的心理体验上：**当我们为过去感到后悔时，我们是在用功能卡构造出了另一种（更优的）可能性；当我们思考未来的走向时，用功能卡构造出一种可能性后，又构造出另一种，这时未来就似乎有两个分岔了——分别指这两种可能性。** 两者共享同一个错误：把推演的次数，错当成了世界本身的分支数。本文守数条须格外郑重划定的界线：这一诊断不否认物理因果的真实可靠性，也不否认功能卡推演在日常计划与沟通中的实践价值；不构成"故一切归因与规划都不成立"的裁决；至于"本该如此"这类判语在责任实践中该如何被重新安放，本文不下判决，明确悬置为**尚待研究**。本文这一篇，自己也是一次此刻用功能卡拼出的推演，一场关公与秦琼的对局；它是指月之指；指可弃，月不可弃。 ## Abstract This paper merges seven insights, progressively reducing "natural continuation" to its true identity—**a retroactive medal we award our own prediction**—while revealing that the method underlying such prediction is exactly the shared, general-purpose mechanism we use in all planning, scheming, regret, and hesitation: **function-card reasoning.** The argument begins from a precise physical example: a billiard ball rolling down a slope is the most reliable knowledge, closest to universal physical law, yet we never say the ball should have remained at rest—by contrast, descriptions of "one utterance's function toward another person" fall far short of this reliability, yet are perceived as equally reliable, and only the latter slides into the verdict "should have been otherwise." The paper asks where this difference originates: **the billiard-ball case produces no such feeling because the physical setup is simple, making an erroneous prediction hard to form**; once matters grow complex—churning clouds—we may predict rain, and if none comes, feel reality has "deviated"; **but the absence of rain only deviates from a mistaken prediction. The so-called "natural continuation" is nothing but the prediction we made, perceived as such whether right or wrong**—landing on a reductio: had we truly predicted accurately, how could reality deviate? The paper then asks why, when prediction fails, we blame an "external disruption" rather than ourselves: **the gale gave us no clue, so we misread "unpredicted for lack of a clue" as "outside the system"—yet lacking a clue only reveals cognitive failure, no evidence about the world's structure.** The paper thereby exposes the actual mechanism of prediction: **this so-called prediction is merely a deliberation we run in our minds based on each module's "function"—"churning clouds have the function of producing rain"; the gale, though possessing the function of "dispersing clouds," was not within our field of view and so never entered our deliberation, hence felt to lie outside the system when it in fact lay only outside the deliberation. This is exactly like our deliberation imagining Guan Yu fighting Qin Qiong—both imagined as modules bearing certain martial functions, their interaction deliberated—which is simply imagining how these functions would be deployed.** The paper pushes "system boundary" to a full paradox—if the system is the universe, a windless system is a physical paradox; if shrunk to a small region, how is its natural continuation even defined? How does a "prediction" ignoring its boundary differ from predicting a duel between Guan Yu and Qin Qiong, who could never have fought? The paper then shows this function-card reasoning **also underlies how we think through our moves in combat, our tactics in non-violent contests, our itinerary while traveling, our allocation of tasks within a team—the very basis of all planning, and precisely because of this ubiquity it feels as though it directly handles reality itself, rather than merely the modules in our own heads. Yet only in simple configurations like the billiard ball does what is in our head align exactly with physical evolution; this explains why our deliberation over complex configurations never quite matches reality—not because reality deviated from its natural continuation, but because our deliberation deviated from, misjudged, that continuation; and in complex configurations, no matter how sophisticated the combination of thinking-modules we employ, it can never match physical evolution, because we have not adopted a Laplace's-demon style of computation.** The paper finally applies this mechanism to two of our most intimate psychological experiences: **when we feel regret over the past, we are using function-cards to construct another (superior) possibility; when we contemplate the future's direction, having constructed one possibility with function-cards, we construct another, and the future then seems to have forked into two—corresponding to these two possibilities.** Both share the same error: mistaking the number of times we have deliberated for the number of branches reality itself possesses. The paper guards several lines with particular gravity: this diagnosis denies neither the genuine reliability of physical causation nor the practical value of function-card reasoning in daily planning and communication; it does not verdict that all attribution and planning are void; and how judgments of "should have been otherwise" ought to be re-situated within practices of responsibility, the paper renders no verdict on, leaving it expressly **still under study**. This very paper is itself a deliberation just now assembled from function-cards, one more match between Guan Yu and Qin Qiong; it is a finger pointing at the moon; the finger may be discarded, the moon may not.

本文追一族彼此嵌套的判断，最终收束到一个统一的诊断：**我们所有讲述宇宙中模块间关系的语言与认知，都是在普适性物理定律发明之前，用来讲述宇宙规律的一种原始方式；而故事与反事实情境的设想，正是这种原始讲述方式结构上必然要求的一个（错误的）前提。** 论证从故事的内部结构起步：**故事自带寓意或中心思想，隐性地指明了相关反事实设想的内容**——奋斗的故事，自然带着一个颓废堕落的反事实作对照；可现实本身，并不内在地对应某一个确定的故事，而是同时可以被叙述成无穷多不同的故事，各自配套着不同的反事实。本文进而指出，当我们想象一个人若遭遇不同、此刻性格便会不同时，我们实际上把此刻的他，感受成了一个和过去不连续的陌生人——**我们没有感到他的恒常不变性，相当于认为过去那个他已经消失了**，同时又感觉"似乎有一条他还是他的路径"；这暴露了一个默认设定：我们认为人保持不变才是"自然""正常"的，而这份设定，使我们无法整合与该人互动的体验，等于把该人认知成了两个人。本文由此追到反事实的根：**设想某反事实情景本可发生，相当于认为实际发生的情景并非过去状态的自然延续**——反事实，表明我们认为世界被偷走了，我们被抛入了一个拙劣的替代品中；而"自然延续"，只有在不被认知为自然延续时，才第一次有了被谈论的必要，这时才有了反事实充当自然延续的候补。可**整个宇宙，本来就是其自身的自然延续**——没有切分之前，"这本该是别的样子"这句话无处安放；反事实只在我们把宇宙切成模块之后，才第一次有了栖身之地，而这一步想象，整个颠倒了自然延续与非自然延续的概念，却又能自圆其说。本文进而给出一个形式化的表述：设两模块在两个时刻的状态为A1、B1及A2、B2，A2+B2就是A1+B1的自然演变；可若只看A1到A2这一段，就总能说B破坏了A的自然演变——**这句话的言外之意，是"宇宙本应是A1+B1'而不是A1+B1"，可宇宙本来就是A1+B1而不是A1+B1'**。这类基于自然语言语义的粗略预测，本质上是对模块B之功能的描述，而功能描述天生带着**颗粒度**：狂风与暴雨这类词，从无精密定义，可我们照样能想象"昨天若是暴雨而非狂风，庄稼会怎样"——这想象不会因模糊而瘫痪，只是**我们容忍了它与真实结果之间一道从未被核算过的鸿沟**，而这份轻松的可想象性，恰恰制造了"实际发生的狂风并非必然"这一错觉。本文最后指出一处被系统性忽略的不一致：谈论B对A的影响时，我们往往不再追溯C对B的影响是如何穿行传递到A的，仿佛B屏蔽了C——**这唯一的原因，是思考这种传递会带来的复杂度；我们的文化刻意忽略了影响的传递，只因思考它太复杂。** 本文由此得出统一诊断：以上种种，都不是零散的认知怪癖，是"用模块间关系讲述宇宙"这一种前科学认知方式的结构性产物——在能够用普适物理定律一次性描述整个宇宙演变之前，把宇宙切成模块、谈论模块间关系，就是我们所有知识的本质；而故事与反事实，正是这种原始讲述方式必然要求、却从未被兑现的前提：一个本不存在的"模块独立自然演变"。本文守数条须格外郑重划定的界线：这一诊断不否认任何一次具体归因的实践价值；不构成"故一切归咎与解释都不成立"的裁决；至于责任究竟该如何在模块化的语言之外被重新安放，本文不下判决，明确悬置为**尚待研究**。本文这一篇，自己也是这场思考此刻切出的一个模块、讲出的一个故事；它是指月之指；指可弃，月不可弃。 ## Abstract This paper traces a nested family of judgments to one unifying diagnosis: **all our language and cognition for narrating relations among modules of the universe are a primitive mode of describing cosmic regularity, predating universal physical law; and stories, together with counterfactual imaginings, are a (false) premise this primitive mode structurally requires.** The argument begins within the story itself: **a story's moral or central theme implicitly specifies its matched counterfactual**—the story of striving naturally carries a contrasting counterfactual of decline; yet reality itself corresponds to no single determinate story, but can be narrated into countless different stories, each carrying its own matched counterfactual. The paper then shows that imagining a person's character would differ had they met different circumstances treats the present self as a stranger discontinuous with the past—**we fail to feel his constancy, tantamount to believing the past self has vanished**, while simultaneously feeling "as if there were a path where he remained himself"; this exposes a default assumption that remaining unchanged is what counts as "natural," and this assumption renders us unable to integrate our experience of interacting with that person, effectively cognizing him as two people. The paper traces this to the root of counterfactuals: **to imagine that some counterfactual scenario could have occurred is to hold that what actually happened was not the natural continuation of the prior state**—the counterfactual expresses a feeling that the world has been stolen, that we have been thrown into a shoddy substitute; and "natural continuation" only needs to be spoken of once it is no longer recognized as such, at which point the counterfactual steps in as its candidate replacement. Yet **the universe as a whole is already its own natural continuation**—prior to any slicing, "it should have been otherwise" has nowhere to attach; the counterfactual finds its foothold only once we have cut the universe into modules, an imagining that inverts the very concepts of natural and unnatural continuation while remaining perfectly self-consistent. The paper offers a formal statement: letting two modules' states at two moments be A1, B1 and A2, B2, A2+B2 just is the natural evolution of A1+B1; yet looking only at A1 to A2, one can always say B disrupted A's natural evolution—**the implication being "the universe should have been A1+B1' rather than A1+B1," when the universe simply was A1+B1, not A1+B1'.** Such rough, natural-language predictions amount to a description of module B's function in relation to A, and functional description inherently carries **granularity**: words like gale and rainstorm have no precise definitions, yet we can still imagine "had yesterday been a rainstorm rather than a gale, how would the crops have fared"—an imagining that does not collapse under vagueness, only that **we tolerate a never-measured gap between the imagined and the real outcome**, and it is precisely this easy imaginability that manufactures the illusion that the actual gale was not necessary. The paper finally identifies a systematically overlooked inconsistency: when discussing B's influence on A, we typically decline to trace how C's influence on B might have propagated through to A, as though B shielded C's influence—**the sole reason being the complexity such tracing would demand; our culture deliberately ignores the transmission of influence simply because thinking it through is too complicated.** The paper concludes that none of the above are scattered cognitive quirks but structural products of "narrating the universe through module relations"—prior to describing the universe's evolution once and for all with universal physical law, cutting the universe into modules and discussing relations among them just is the substance of all our knowledge; and stories together with counterfactuals are exactly the premise this primitive mode necessarily requires, yet never actually delivers: an independent "natural evolution of a module" that was never coherently defined. The paper guards several lines with particular gravity: this diagnosis denies no specific attribution its practical value; it does not verdict that all blame and explanation are void; and how responsibility ought ultimately be re-situated beyond modular language, the paper renders no verdict on, leaving it expressly **still under study**. This very paper is itself a module this thinking has just now cut out, a story just now told; it is a finger pointing at the moon; the finger may be discarded, the moon may not.

本文追一条从”故事自带反事实”出发、最终抵达”宇宙本是其自身唯一自然延续”这一判断的论证。起点是：故事自带寓意或中心思想，隐性地指明了相关反事实设想的内容——奋斗的故事，自然自带一个与之对比的颓废、放弃的反事实情境。可一段现实本身，并不内在地对应某一个确定的故事，而是同时可以被叙述成无穷多个不同的故事，各自配套着不同的反事实。本文进而展示这一机制在人格判断上的具体后果：当我们想象一个人若遭遇不同，此刻的性格便会不同时，我们实际上把此刻的他，看作了一个不同于过去的他的、陌生的人——我们没有感到他的恒常不变性，相当于认为过去那个他已经消失了；同时，我们又感觉似乎有一条”他还是他”的路径存在，仿佛这份改变并非必然。这表明我们认为，人保持不变（或只朝我们欢迎的方向改变）才是”自然”“正常”的；我们因此无法整合自己与该人互动的体验，只能分两段处理——等于把该人认知成了两个人。本文由此追问一个更根本的问题：“自然延续”这个概念，究竟是从哪里获得它的立足点的？ 答案是：反事实设想，表明我们认为世界消失了、被偷走了，我们被抛入了一个拙劣的替代品中——而”哪里不是自然延续”，答案是哪里都可以不是；描述”自然延续是什么”，本身就已经是在讲一个故事了。 本文据此立出全篇最重的判断：一切——整个宇宙——本来就是其自身的自然延续。反事实，只在我们把宇宙切分为模块之后，才第一次有了立足之地——设想模块A的实际发生并非自然延续、而应是另一版本A’，这一想象整个颠倒了自然与非自然的概念，却能自圆其说；而这种颠倒，恰恰是我们理解模块间关系的唯一办法，尽管两模块本来并无关系。把宇宙划分成模块、谈论模块间关系，这就是我们全部知识的本质——在我们能用普适物理定律一次性描述整个宇宙演变之前。本文将这一诊断形式化：设两模块在两个时刻的状态为A1、B1与A2、B2，若二者合起来即是整个宇宙，则A2+B2无可争议地是A1+B1的自然演变；可若只看A1到A2这一段，我们便总能说”B影响或破坏了A的自然演变”——而这句话，言外之意是”宇宙本应是A1+B1’而不是A1+B1”，可宇宙本来就是A1+B1，不是A1+B1’；“作为宇宙局部的A的自然演变”这一提法，从未被定义过，因为A从来不是独立于B而存在的整体，只是A+B这个自然延续的一部分。 本文进而指出，这套”B对A的功能”式的粗略预测，天生带着粗颗粒度：狂风与暴雨之间，没有精密的定义边界；可即便如此，我们仍能想象”昨天若是暴雨而非狂风，庄稼会怎样”，这份想象不会因颗粒度粗、定义模糊而瘫痪——只是这份想象，同样可能与真实的暴雨结果相差甚远，我们只是容忍了这份差异，而从未核算它到底有多宽；正是这份毫不费力的可想象性，制造了”实际发生的狂风并非必然”这一错觉。本文最后指出一处影响追溯的不一致性：设宇宙分三个模块A、B、C，我们评判B对A的影响时，往往不再追溯C对B的影响如何穿行至A，仿佛B屏蔽了这份传递——可这一停止追溯，唯一的原因，是思考这种传递会带来的复杂度；我们的文化刻意忽略了影响的传递，只因继续追溯会使思维变得极度复杂，于是简单地停止了这种思考，而非因为这里真有一个天然的、有原则的断点。 本文守数条须格外郑重划定的界线：这一诊断不否认任何具体故事、具体判断在其适用条件内的真实与有用；不构成停止归责、停止理解他人性格变化的理由；至于反事实与责任究竟该如何最终安放，本文不下判决，明确悬置为尚待研究。本文这一篇，自己也是把宇宙这唯一的自然延续，切成模块、讲成故事的又一次尝试；它是指月之指；指可弃，月不可弃。 Abstract This paper traces an argument from “stories carry built-in counterfactuals” to the culminating judgment that the universe is its own sole natural continuation. It begins: a story’s moral or central theme implicitly specifies its matching counterfactual—a story of striving naturally carries, as its foil, a counterfactual of decline or surrender. Yet a stretch of reality does not intrinsically correspond to one determinate story; it can equally be narrated as countless different stories, each carrying its own matched counterfactual. The paper shows this mechanism’s concrete consequence for judging character: when we imagine that had a person met different circumstances, his present character would differ, we in fact regard the present him as a different, unfamiliar person, not a continuation of his past self—we fail to feel his constancy, tantamount to holding that the past him has vanished; yet we simultaneously feel there existed a path on which he “remained himself,” as though this change were not necessary. This reveals that we hold constancy (or change only in directions we welcome) to be “natural,” “normal”; we thereby fail to integrate our experience of interacting with this person, splitting it into two segments—effectively cognizing him as two people. The paper then presses a deeper question: where does “natural continuation” get its footing at all? The answer: counterfactual imagining expresses a feeling that the world has vanished, been stolen, and we have been thrown into an inferior substitute—and as to where is “not” a natural continuation, the answer is anywhere can be; to describe what “natural continuation” consists in is already to be telling a story. The paper thereby states its heaviest judgment: everything—the whole universe—already is its own natural continuation. The counterfactual only gains a foothold once we carve the universe into modules—positing that module A’s actual occurrence is not its natural continuation, but should instead be some A’, entirely inverts the concepts of natural and unnatural, yet remains self-consistent; and this inversion is precisely the only way we have of understanding the relation between two modules, even though the two modules originally bore no relation at all. Carving the universe into modules and discussing the relations among them is the very essence of all our knowledge—prior to being able to describe the whole universe’s evolution at once via universal physical law. The paper formalizes this: let the two modules’ states at two moments be A1, B1 and A2, B2; if together they constitute the whole universe, then A2+B2 is, beyond dispute, the natural evolution of A1+B1; but looking only at the segment A1 to A2, we can always say “B influenced or disrupted A’s natural evolution”—yet this statement implicitly asserts “the universe should have been A1+B1’ rather than A1+B1,” when the universe simply is A1+B1, not A1+B1’; “A’s natural evolution as a piece of the universe” was never defined, since A was never a whole independent of B, only a part of the natural continuation that is A+B. The paper further notes that this “B’s function relative to A” style of rough prediction is inherently coarse-grained: there is no precise boundary between a gale and a rainstorm; yet even so, we can still imagine “had yesterday been a rainstorm rather than a gale, how would the crops have fared”—an imagining not paralyzed by coarseness or vagueness—only this imagining may likewise diverge greatly from the true outcome of an actual rainstorm, a gap we merely tolerate without ever reckoning its width; it is precisely this effortless imaginability that manufactures the illusion that the actual gale was not necessary. The paper finally points to an inconsistency in how far we trace influence: dividing the universe into modules A, B, C, when we judge B’s influence on A, we typically do not further trace C’s influence on B propagating on to A, as though B shielded that transmission—yet the sole reason for stopping the trace is the complexity that continuing it would bring; our culture deliberately neglects the transmission of influence simply because tracing it further would render thought unmanageably complex, and so thinking is simply halted, not because there is any principled, natural stopping point. The paper guards several lines with particular gravity: this diagnosis denies no specific story or judgment its truth or usefulness within its conditions of applicability; it is no reason to stop attributing responsibility or understanding a person’s change of character; and how the counterfactual and responsibility ought ultimately to be settled, the paper renders no verdict on, leaving it expressly still under study. This very paper is itself one more attempt to carve the universe’s sole natural continuation into modules, into a story; it is a finger pointing at the moon; the finger may be discarded, the moon may not.

本文追一个从未被正面问过的问题：同一段真实的物理过程，比如一只猫捉住一只老鼠，可以被压缩成无穷多个互不相容、却都真实的故事——"猫捉老鼠"，也可以是"老鼠甲的惊险的一天"、"感冒病毒对老鼠甲行为的影响"、"血液中的某白细胞捕食了细菌"、"某水分子中发生的量子力学现象"、"适者生存、优胜劣汰"、"猫的捕食活动在代际是如何传递的"。**到底哪个故事，才是这里真正发生的唯一故事？本文的判断是：并没有所谓唯一的故事，所有这些故事都在真实地发生着，即使它们在中心思想或寓意上彼此矛盾。** 可人类个体，对于自己所体验到的人生所涉及的物理过程，往往只会以一到两个故事来记忆一段物理过程，而不是同时用无穷多个故事来记忆同一段过程。本文称这一现象为**故事化坍缩**：无穷多个原则上同时成立的版本，在被记忆的那一刻，被压缩成了一到两个，其余的，即使同样真实、甚至足以推翻我们最终采纳的判语，也被永久错过。本文进而指出，这份错过的根源，往往不是选择性偏见，而是**信息在当时的条件下根本不可得**：设想两个旁观者，同时看着这场猫捉老鼠——我捕捉到的细节，编成了"黑猫捉到了大老鼠"；在场的另一人，捕捉到了一个我也看见、却因情绪基调不同而未曾编入我的故事的细节（一个老鼠本可逃入的洞），编成了"老鼠再跑快一点就能进洞了，真是遗憾"；而"老鼠因为感冒而跑慢了"这个同样真实的故事，则因为信息完全无法获得，被我们两人共同忽略——**正因为这一版本从未进入候选清单，"老鼠跑得不够努力，因此被捉了"这个中心思想，才得以立得住；唯有全知的上帝，才知道老鼠不是不努力，是感冒了。** 本文由此揭示一种寄生在这份永久信息缺失上的辩护策略：**自由意志论者可以声称，老鼠即使感冒了，凭着自由意志，还是能跑得比它实际跑的快一点——因为没有人能够证明这个选项不存在。** 这套辩护的力量，不在于它论证了什么，而在于它展示了一种永远不会输的策略的形状：它不需要证明老鼠"确实能"跑得更快，只需要指出对手"无法证明它不能"——这是一次彻底的举证责任倒置。而这套策略可以被无限地向前递归：自由意志论者同样可以声称，老鼠昨日本可自由地选择留在洞里避雨、从而不会感冒，因为对手同样无法证明老鼠昨天"必须"出去觅食。**"你无法证明我必须如此"与"我确实能够不如此"，中间隔着举证责任转移这一整层，从未被兑现过**——这是一套永不认输、却也永不断言任何具体内容的辩护。本文最后给出这一切的落点：**我不应把记忆中的故事认作现实的本质，而应将其视作一种标识符、一块牌子，这个牌子后面才是现实，现实与故事无关，故事只是一种可被替换、更改的表象。** 本文守数条须格外郑重划定的界线：这一诊断不否认任何一个具体故事本身的真实性，只否认其唯一性与穷尽性；不构成"故一切归责都不成立"的裁决；至于自由意志与责任究竟该如何最终安放，本文不下判决，明确悬置为**尚待研究**。本文这一篇，自己也是这场思考此刻坍缩出的一个故事、一块可被替换的标识符；它是指月之指；指可弃，月不可弃。 ## Abstract This paper pursues a question rarely asked directly: a single real physical process—a cat catching a mouse—can be compressed into countless mutually incompatible, yet equally true, stories: "the cat caught the mouse," or equally "mouse A's harrowing day," "the cold virus's effect on mouse A's behavior," "a white blood cell preying on bacteria," "a quantum event within some water molecule," "survival of the fittest," "how a cat's predatory behavior is passed down across generations." **Which story is the one true story actually occurring here? The paper's judgment: there is no single true story; all of these are truly occurring, even where they conflict in their central moral.** Yet an individual, remembering an episode of lived experience, collapses it into only one or two stories, rather than holding countless versions simultaneously. The paper names this **story-collapse**: countless versions, in principle equally valid, compressed at the moment of memory into just one or two, the rest—however true, however capable of overturning the judgment eventually adopted—forever missed. The paper argues this loss is typically not selective bias but that **the relevant information was simply unavailable at the time**: imagine two observers watching the same chase—the details I captured become "the black cat caught the big mouse"; the other observer, who caught a detail I too saw but, given a different emotional register, never wove into my story (a hole the mouse might have reached), produces instead "if only the mouse had run a bit faster, it would have made the hole—what a pity"; while "the mouse ran slow because it had a cold" is a story neither of us could tell, since that information was simply never available to either—**and it is precisely because this version never entered the candidate pool that the standing moral, "the mouse wasn't trying hard enough, hence was caught," gets to stand; only an omniscient god would know the mouse was not lazy but ill.** The paper thereby exposes a rebuttal strategy that parasitizes this permanent unavailability: **a free-will theorist may claim that even with a cold, the mouse, by virtue of free will, could still have run a bit faster than it actually did—since no one can prove this option did not exist.** The force of this rebuttal lies not in what it proves but in the shape of an unfalsifiable strategy: it need not prove the mouse "could" run faster, only point out the opponent "cannot prove it could not"—a thorough inversion of the burden of proof. And this strategy can be extended recursively backward: the free-will theorist may equally claim the mouse could have freely chosen, the day before, to stay in its burrow out of the rain and thus never catch cold, since the opponent likewise cannot prove the mouse "had to" forage that day. **Between "you cannot prove I had to" and "I really could have done otherwise" lies the entire burden of a proof-shift that is never actually discharged**—an unfalsifiable defense that never loses, and never asserts anything specific either. The paper concludes: **I should not take the story in my memory to be the essence of reality, but should regard it as a label, a tag—behind this tag lies reality, and reality has nothing to do with the story; the story is merely a replaceable, alterable appearance.** The paper guards several lines with particular gravity: this diagnosis denies no specific story its truth, only its uniqueness and exhaustiveness; it does not verdict that "so no attribution of responsibility holds at all"; and how free will and responsibility ought ultimately to be settled, the paper renders no verdict on, leaving it expressly **still under study**. This very paper is itself a story this thinking has just now collapsed into, a replaceable tag; it is a finger pointing at the moon; the finger may be discarded, the moon may not.

本文合并四条彼此独立发展、却共享同一个根源的论证，构成一部关于**我们对因果与责任的判断，如何被语法的恒常性预设、信息的不可得性、以及记忆的坍缩，系统性地扭曲**的整体论述。第一条线索是**语义创世**：一句原本只用来区分两具躯体的话——"他的想法"——在区分对象退场之后，被想象填满了它留下的语法空位，凭空造出一个躯体之外的主体；这一机制依次推广到动词与代词（"想"与"我"，及"我思故我在"之荒谬化）、空间（色盲房间、反事实、柏拉图洞穴）、内部房间的具体挂牌流程（僵尸悖论的自我拆解）、虚拟房间与建模的实存感、时间中的恒常感（塑料袋方案、看病诊断），以及自由意志辩护的收缩策略——辩护者把自由压缩进越来越窄的选择瞬间，这一收缩动作本身，正是他们承认瞬间之外不自由的行为证据。第二条线索是**认知边界如何被错当本体边界**：我们事后能给自己的行为找出各种原因，事前却不将外在的未知因素纳入计划或担忧，因为未知因素没有具体内容可填入图景，而这份"未填"被悄悄读成了"不存在"；这一空缺，在面对未来时比面对过去时更彻底，因为过去有具体结果倒逼我们承认"总有别的原因"，未来却没有这份倒逼——"唯一主宰"之感由此而生。第三条线索是**影响链的无限延伸与可描述性的门槛**：若认了"甲影响了乙"，这条线便会无限延伸，以致任何人此刻的行为都是所有前人的总和，这正面反证了"源于自己"这一感觉的不可靠；而我们判定一份影响是否存在，唯一的操作性标准，是它能否被讲通——这道门槛远比通常以为的容易跨过，只需一个通用的、无法被证伪没有影响的中间变量，便能弥散地覆盖后续一切；而这一门槛，本身又受着道德兴趣的预先筛选：涂不上道德色彩的因果关系，会被搁置，直到我们发明出涂色的办法。第四条线索，本文在此首次系统展开：**故事化坍缩**。一段真实的物理过程，可以被压缩为无数个互不相容、却都真实的故事版本——"猫捉老鼠"，也可以是"老鼠甲的惊险的一天""感冒病毒对老鼠甲行为的影响""某白细胞捕食了细菌""适者生存"；这些故事，没有一个是唯一正确的那个，它们都在真实地发生着，即使彼此矛盾。可个体记忆一段经历时，只会坍缩出一到两个故事，其余版本，即使同样真实、甚至足以推翻我们采纳的判语，也会永远错过——这份错过，往往不是因为选择性采纳，是因为相关信息在当时的条件下根本不可得。而**自由意志论**恰恰寄生在这份永久的信息不可得上：它可以无限地宣称"本可以更努力""本可以昨天就避免"，因为对手永远无法证明"必须如此"——这是一套永不认输、却也永不断言任何具体内容的辩护策略。本文最后给出统一诊断：我们记忆中的故事，从来不是现实本身，只是挂在现实门口、可被替换的一块标识符；现实，不因换了哪块牌子而有分毫改变。贯穿这四条线索的，是同一种对"话语具有脱离情境之恒常性"的误信，与同一种对"信息不可得即等于不存在"的错读。本文守数条须格外郑重划定的界线：这一切不触及第一人称此刻真实的确定性；不构成停止与他人交流、信任、协作的理由；不构成忽视他人痛苦感受的许可证；至于自由意志与责任究竟该如何最终安放，本文不下判决，明确悬置为**尚待研究**。本文这一篇，自己也是这场思考此刻坍缩出的一个故事、一块可被替换的标识符；它是指月之指；指可弃，月不可弃。 ## Abstract This paper merges four independently developed lines of argument, sharing one root, into a unified account of **how our judgments of causation and responsibility are systematically distorted by grammar's presumption of constancy, by the unavailability of information, and by the collapse of memory.** The first thread is **semantic genesis**: a phrase originally meant only to distinguish two bodies—"his thought"—once the disambiguated party recedes, has its grammatical subject-slot filled by imagination, conjuring a subject beyond the body; this mechanism recurs across verbs and pronouns (Cartesian absurdity), space (the color-blind room, counterfactuals, Plato's cave), the concrete labeling of an "inner room" (the zombie paradox self-undoing), the virtual room and its sense of reality, constancy in time (the plastic-bag solution, the diagnosed illness), and the shrinking defense of free will—defenders compress freedom into an ever-narrower instant, the compression itself being evidence they concede everything outside it is unfree. The second thread is **the cognitive boundary mistaken for the ontological boundary**: we readily find causes in hindsight, yet fail to factor unknown future factors into our plans, because the unknown has no specific content to fill a slot, and this blank is quietly read as nonexistence—more thoroughly so for the future than the past, since the past has an actual outcome forcing some acknowledgment while the future has none, whence the sense of sole mastery. The third thread is **the infinite extension of influence and the threshold of describability**: once one grants that A influenced B, the chain extends without limit, refuting the reliability of "sourced from oneself"; and the sole operative criterion for judging whether an influence exists is whether it can be told coherently—a threshold far easier to clear than supposed, requiring only a generic, unfalsifiable intermediate variable; this threshold is itself pre-filtered by moral interest, uncolorable causation being set aside until a means of coloring is invented. The fourth thread, developed here for the first time systematically, is **story-collapse**. A real physical process can be compressed into countless mutually incompatible yet equally true stories—"the cat caught the mouse," or equally "mouse A's harrowing day," "the cold virus's effect on mouse A's behavior," "a white blood cell preying on bacteria," "survival of the fittest"; none is the one correct story, all are truly occurring even where they conflict. Yet an individual, remembering an episode, collapses it into only one or two stories; the rest, however true, however capable of overturning the judgment we adopted, are forever missed—not usually from selective bias but because the relevant information was simply unavailable at the time. **Free-will apologetics** parasitizes exactly this permanent unavailability: it can endlessly claim "could have tried harder," "could have avoided it yesterday," since the opponent can never prove "had to be so"—an unfalsifiable strategy that never loses, and never asserts anything specific either. The paper concludes with a unifying diagnosis: the story in our memory is never reality itself, only a replaceable tag hung at reality's door; reality is not altered in the least by which tag we happen to hang. Running through all four threads is the same false belief in semantic constancy independent of context, and the same misreading of informational unavailability as nonexistence. The paper guards several lines with particular gravity: none of this touches first-person present certainty; none of it is reason to cease communicating, trusting, or cooperating with others; none of it licenses disregarding another's suffering; and how free will and responsibility ought ultimately to be settled, the paper renders no verdict on, leaving it expressly **still under study**. This very paper is itself a story this thinking has just now collapsed into, a replaceable tag; it is a finger pointing at the moon; the finger may be discarded, the moon may not.

本文追一个统一的机制——**语义创世**——它在代词、空间、时间与自由意志辩护这四个看似无关的领域里，反复地凭空造出一个"躯体之外的主体"。机制的原型极朴素：**"他的想法"这句话，本只是一个区分算子。** 两具躯体同时在场时，这句话只做一件事——把话题从"关于乙的情况"切成"关于甲的情况"；此刻，"想法需不需要有个主语"这个问题从未被激活。可一旦乙的躯体退场，只剩甲一具躯体，我们再度说"他的想法"时，语法留下的主语空位，被**想象**填了上去——这个填充没有任何新证据支撑，只因语法结构"读起来该完整"，而这份闭环带来的顺畅感，被错当成了对应真实的正确感。我们遗忘了这句话原是区分算子、其历史任务已经完成，却将错就错，用新的方式解读旧语义，合理化了新的想象而不自知——这背后，是对"话语具有脱离情境之恒常性"的误信。本文进而展示这同一机制的四次推广。其一，**推广到动词与代词**："我想吃饭"与"他想吃饭"共用一个"想"字，可后者对我而言纯属外在、物理性的报告，前者却是我正经历的体验；这份共用制造了对称的假象，而这假象仅在第三方视角下才真的成立。"我"字亦然：只有在第三方发起区分性提问（"谁想去吃饭？""我！"）时，"我"字才在语义上真正对称；离开这一情境（如笛卡尔式独白），"我"字的所指便发生了隐秘的畸变。本文据此论证：**"我思故我在"这句话，以其原初语义（区分算子）理解，是荒谬的**——躯体的存在肯定先于其思考，第三方从不会有论证笛卡尔是否存在的动力；这句话之所以显得深刻，全赖一次隐秘的置换——把"我"从指着躯体，偷换成指着一个被想象出来的灵魂。其二，**推广到空间**：我们想象色盲者"看到"的是一个内部的、同构的灰度房间，这表明我们把日常所见，也想象成了内部的、认知中的东西；可这想象是次生的，原初语义下，所见之物是外部的、独立于认知的。反事实情境与柏拉图的洞穴隐喻，进行的是同一场语义次生。其三，**推广到"挂牌"的具体流程**：我们想象他人"内部房间"的方式，其实是复制一个和外部房间一模一样的房间，只在门口挂一块"内部"的牌子；再把这人的外部形象克隆一份塞进去，挂上"灵魂"的牌子；再挂上"盲人"的牌子，就得到了哲学僵尸——**若这具僵尸的内部房间从不可被访问，这房间存在的唯一理由也随之消失，这正是僵尸概念自身的悖论。** 而这同一套虚拟房间装置，也被我们用来理解自己的思维（如提前关门的假想情景）：虚拟房间与虚拟房客彼此支撑，显得"活了"，其实两者都不存在；某种恒常感更让我们误以为这场景"一分钟前就已构想出"，可它此刻才第一次被造出。这隐喻的实质，只是大脑里的某种物理情况，隐喻的空间形式，却带来了物理情况本不具备的恒常性。其四，**推广到时间与选择**：一个此刻才被制造或确证出的方案／答案，会给人一种"它一直存在、我只是没瞥一眼"的错觉；"当初有几个选项"这类叙事，是选项被事后安放到那个时间点上的明证——实际发生的，只是一段连续、不可切分的物理过程，被硬生生切成几步，再在每一步里倒推出选项。同一机制，也解释了为什么自由意志的辩护者，总是把自由压缩进一个越来越窄的选择瞬间：这套压缩没有天然的止点，可以反复地对任何"自由区间"施加"内部再分出计算与拍板"的操作；而当"大脑的自由选择不可想象"这一物理事实无法回避时，辩护者便把选择的主体，偷偷换成大脑之外的意识／灵魂——这正是语义创世在自由意志辩护里的重演。**辩护者把防线一路收缩到一个瞬间，这个收缩动作本身，恰恰是他们承认瞬间之外不自由的行为证据。** 本文最后指出，贯穿这四次推广的共同根源，是同一种**恒常感**：它把此刻才被制造、填补、确证的东西，错觉成了早就存在、独立于此刻的东西——这份错觉不可见，因为它就发生在语言与想象最基本的语法零件里。本文守数条须格外郑重划定的界线：这一诊断不触及第一人称此刻真实的确定性；不构成停止与他人交流、信任、协作的理由；不构成忽视他人痛苦感受的许可证；至于自由意志、责任该如何最终安放，本文不下判决，明确悬置为**尚待研究**。本文这一篇，自己也是这场思考此刻生成的一段语言；它是指月之指；指可弃，月不可弃。 ## Abstract This paper traces one unified mechanism—**semantic genesis**—recurring across pronouns, space, time, and the defense of free will, each time conjuring a "subject beyond the body" out of nothing. Its prototype is plain: **"his thought" is originally only a disambiguation operator.** With two bodies present, the phrase does one thing—cutting the topic from "about B" to "about A"; whether "thought" needs a subject is never raised. Once body B recedes, leaving only A, saying "his thought" again lets imagination fill the grammatical subject-slot left behind—no new evidence, only grammar's felt need for completeness, whose closure is mistaken for correctness. We forget the phrase's task was done, and rationalize the new imagining unawares—behind this lies a false belief in semantic constancy independent of context. The paper shows this mechanism recur four times. First, **onto verbs and pronouns**: "I want to eat" and "he wants to eat" share one word, though the latter is, for me, a purely external report while the former is my lived experience; this sharing manufactures a false symmetry, true only from a third-party view. "I" is likewise: only when a third party poses a disambiguating question ("who wants to eat?" "I!") is "I" truly symmetrical; outside this, as in Cartesian soliloquy, "I" undergoes a covert deformation. Hence **"I think, therefore I am," read in its original semantics, is absurd**—the body's existence plainly precedes its thinking, and a third party never needs to argue for Descartes's existence; the sentence seems profound only through a covert substitution, from the body to an imagined soul. Second, **onto space**: we imagine what a color-blind person "sees" as an inner, isomorphic grayscale room, revealing that we've reimagined everyday sight itself as internal and cognitive; but this is derivative—originally, what is seen is external, independent of cognition. Counterfactuals and Plato's cave perform the same semantic derivation. Third, **onto the concrete "labeling" procedure**: we imagine another's "inner room" by copying an outer room and hanging a sign reading "inner" on its door; cloning that person's outer form and placing it inside yields a "soul"; hanging a further sign "blind" yields the philosophical zombie—**if this inner room can never be accessed, its very reason for being posited vanishes, the zombie concept's own paradox.** The same virtual-room device explains how we understand our own thinking (imagining having shut a door earlier): the virtual room and its virtual occupant prop each other up into seeming alive, though neither exists; a sense of constancy further misleads us into feeling the scene was conceived a minute ago, when it was only just now constructed. The metaphor's substance is merely some physical state in the brain; its spatial form lends it a constancy the physical state itself never had. Fourth, **onto time and choice**: a solution or answer only just constructed or confirmed gives the illusion of having always existed, merely unglanced-at; narratives of "having had options at the time" are proof that options are retroactively inserted—what actually occurred was one continuous, uncuttable physical process, artificially sliced into steps, each retrofitted with options. The same mechanism explains why free-will's defenders keep compressing freedom into an ever-narrower instant of choice: this compression has no natural stopping point, and can be reapplied indefinitely to any "free interval"; and when the unimaginability of a free brain becomes unavoidable, defenders quietly swap the choosing subject from brain to a soul beyond it—semantic genesis recurring in free-will apologetics. **That defenders keep shrinking the front line to a single instant is itself behavioral evidence that they concede everything outside it is unfree.** The paper concludes that a single sense of **constancy** underlies all four recurrences—mistaking what is only now manufactured, filled in, or confirmed for something that existed all along, independent of the moment—invisible precisely because it operates in language's and imagination's most basic grammatical parts. The paper guards several lines: this diagnosis does not touch first-person present certainty; it is no reason to cease communicating, trusting, or cooperating with others; it licenses nothing regarding disregard for another's suffering; how free will and responsibility ought ultimately to be settled, the paper renders no verdict on, leaving it expressly **still under study**. This very paper is itself a stretch of language this thinking has just now generated; it is a finger pointing at the moon; the finger may be discarded, the moon may not.

本文重写并推进一条论证，最终追到它自己的语言学根源。判断的起点：**若把他人的想法定义为他的内心独白，这就破解了意识的难问题——我们不再需要假设他人具有内心独白之外的"意识"。** 展开这一定义，须拆开"人有想法"捆着的三层意思——内部的、不可见的、似乎不确定的；不确定性被改写为信息只传给了当事人、没有传给旁人这一后果。本文收紧两处曾经的含混：**没有什么所谓"感觉到确定"这件事，当事人知道自己的内心独白，这句话本身就是确定性的完整说明；内心独白，就是主观感受——和他人看待我的客观视角相比，没有再"如何主观感受到内心独白"这一说。** 主观／客观标注的是同一件事被从两个位置观测的两种称呼，不是两种不同的东西。本文进而拆穿僵尸世界的自相矛盾：若僵尸能用内心独白真的描述环境，按已确立的等同，它便已在感受这个世界；若这份描述不是真的描述，僵尸赖以成立的"功能完全相同"这一前提便未被兑现。本文诊断难问题的真正病灶：**它从未证成它悄悄预设的前提——"他人具有与我的意识等同的东西"；这正是皇帝的新装。** 哲学僵尸，其实是唯我论的极端形式。本文将此接回一条更深的结构——**主体之外的主体，原因之外的原因**：灵魂，只是主体感、原因感、解离感这三份第一人称现象的名词化，恰如愤怒把体验隐喻成实体，这一隐喻仅对我自己成立。本文进而指出，我们据以相信他人有意识的真正根据，不是意识，是**语言**——我用同一套语义系统理解自己的内心独白与他人说的话，这份对称感受，才是相信他人有意识的真正原因。可这块地基同样不稳：**哲学僵尸问题，等价于中文房间问题**——除非认定他人正常的说话活动本身就证明了理解能力，否则没有任何证据表明他人理解语言；一具躯体的输出只是输入与结构的函数，读不出结构之外的任何东西。本文如实呈现由此产生的两个分岔（归纳猜想、彻底的唯我论），指出它们分属**两种互不相容的叙事风格**——物理／结构的叙事与灵魂论现代版的叙事，语法互斥，不可通约；本文此刻用后者向"你们"说话，只证明作者仍在使用这套风格，不证明"你们"具有它所预设的东西。本文最后追到这整套困惑的语言学根源：**"语义创世"。** "他的想法"这句话，原本只是一个**区分算子**——两具躯体同时在场时，它只做一件事：把这句话，从"关于乙的情况"切分成"关于甲的情况"，仅此而已，主语是否指着躯体之外的东西，这个问题此刻从未被激活。可一旦乙的躯体退场、只剩甲一具躯体，我们再度说"他的想法"时，语法留下的那个主语空位，被**想象**填了上去——这个填充，没有任何新证据支撑，只因语法结构"读起来该完整"，而这份闭环带来的顺畅感，被我们错当成了对应真实的正确感。我们遗忘了这句话原是区分算子、其历史任务在乙退场时已经完成，却将错就错，用新的方式解读旧语义，从而合理化了新的想象而不自知——这预设了话语具有脱离具体情境的恒常性，而这一预设，本身就是错的。本文进而指出第二重混淆：与躯体交流以获知**已知事实性质**的信息（如图书馆几点开门，预存于结构中）是合理的活动；可试图与躯体交流以获知意识难问题的答案，是不合理的——那答案根本不是已知事实性质的信息，从不预存于任何结构中，等待被检索。**试图通过与他人的交流活动获得意识难问题的答案，是在不可能的前提条件下进行的、不可能成功的认知活动类型。** 本文守数条须格外郑重划定的界线：不触及第一人称确定性；绝非停止与他人交流、信任、协作的理由，语言的社会实践不因这一诊断而停摆，本文自身能被写出、被期待被理解，正是明证；绝非忽视他人痛苦的许可证。他人是否理解语言、是否具有主观体验，该被如何最终对待，本文不下判决，明确悬置为**尚待研究**。本文这一篇，自己也是用灵魂论现代版叙事、向"你们"说出的一段内心独白；它是指月之指；指可弃，月不可弃。 ## Abstract This paper rewrites and advances an argument to its own linguistic root. Starting judgment: **defining another's thought as their inner monologue dissolves the hard problem of consciousness**—we no longer need to posit a "consciousness" beyond the monologue. Unpacking this requires prying apart the trio bundled into "having thoughts"—internal, invisible, seemingly uncertain; uncertainty is rewritten as the consequence of information reaching one party and not another. The paper tightens two prior ambiguities: **there is no "felt sense of certainty" separable from knowing—that the subject knows their inner monologue just is the complete account of certainty; the inner monologue just is the subjective experience, by comparison with another's objective view of me; there is no further question of how it is subjectively felt.** Subjective/objective name one event from two vantage points, not two kinds of thing. The paper then dismantles the zombie world's self-contradiction, diagnoses the hard problem's true ailment—it never argues for its presupposition that another possesses something equivalent to my consciousness, the emperor's new clothes—and connects this to a deeper structure, **a subject beyond the subject, a cause beyond the cause**: the soul is the nominalization of a sense of agency, causation, and dissociation, valid only for oneself. The paper argues what actually grounds belief in another's consciousness is not consciousness but **language**—the same semantic system handling my monologue and another's speech, a symmetry that grounds the belief. Yet this foundation is equally unstable: **the zombie problem equals the Chinese Room problem**; a body's output is merely a function of input and structure, yielding nothing beyond it. The paper presents, without adjudicating, two resulting forks, showing they belong to **two incompatible narrative styles**—physical-structural and modern soul-theory—grammatically exclusive, incommensurable; this paper's own address to "you" proves only the author's current habit of speech, not that "you" possess what it presupposes. The paper finally traces the whole confusion to its linguistic root: **semantic genesis.** "His thought" was originally only a **disambiguation operator**—with two bodies present, it merely marks "about A, not B"; whether the subject points beyond the body is never activated. Once body B recedes, leaving only A, the grammatical subject-slot left behind gets filled by **imagination**—no new evidence, only grammar's felt need for completeness, whose closure is mistaken for correctness. We forget the phrase's task was already done, and rationalize the new imagining unawares, presupposing semantic constancy independent of context—itself false. A second confusion follows: querying a body for **known-fact information** (library hours, pre-stored in structure) is legitimate; querying it for the hard problem's answer is not—that answer is never pre-stored, awaiting retrieval, in any structure. **Seeking the hard problem's answer through communication with others is an impossible type of cognitive activity, attempted under impossible preconditions.** The paper guards several lines: it does not touch first-person certainty; it is by no means a reason to cease communicating, trusting, or cooperating with others—the social practice of language continues regardless, and this very paper's being written and expected to be understood is proof; and none of this licenses disregarding another's suffering. How another's linguistic understanding and subjective experience ought ultimately be treated, the paper renders no verdict on, leaving it **still under study**. This very paper is itself an externalized inner monologue, addressed to "you" in the modern style of soul-theory; it is a finger pointing at the moon; the finger may be discarded, the moon may not.

本文重写并推进一条论证，把它从"意识"这个靶子，逼向一个更深、也更难被绕开的地基——**语言的理解**。判断的起点：**若把他人的想法定义为他的内心独白，这就破解了意识的难问题——我们不再需要假设他人具有内心独白之外的"意识"，内心独白可被视为一种类似说话的纯物理过程。** 展开这一定义，须先拆开"人有想法"捆着的三层意思——内部的、不可见的、似乎不确定的；内部性被继承，不可见性被还原成未发声的语言活动，不确定性被改写为信息只传给当事人、没传给旁人这一后果。本文进而收紧两处曾经的含混：**没有什么所谓"感觉到确定"这件事，与"知道"分离开来还需被单独解释——当事人知道自己的内心独白，这句话本身就是确定性的完整说明；内心独白，就是主观感受——和他人看待我的客观视角相比，没有再"如何主观感受到内心独白"这一说。** 主观／客观标注的不是两种不同的东西，是同一件事被从两个不同位置观测时的两种称呼。本文进而正面拆穿这条路径历史上最强的反驳——僵尸世界的可设想性：若僵尸能用内心独白真的描述环境，按已确立的等同，它便已在感受这个世界，"感觉起来什么都不是"便自我推翻；若这份描述不是真的描述，僵尸赖以成立的"功能完全相同"这一前提便未被兑现。本文由此诊断难问题的真正病灶：**它问"为什么他人能有意识"，却从未证成它悄悄预设的前提——"他人具有与我的意识等同的东西"；这正是皇帝的新装，先假装皇帝穿了衣服，再讨论这件衣服为何难以被看见。** 哲学僵尸的概念，其实是唯我论的极端形式——"意识"在其中的定义，只是"我的意识"的另一个名字。本文将此接回一条更根本的结构——**主体之外的主体，原因之外的原因**：难问题问的是，大脑／身体这个主体、这个原因之外，为何还要另有一个主体、另有一个原因，可根本没有另外的主体或原因；灵魂，只是主体感、原因感、解离感这三份真实的第一人称现象的名词化，恰如愤怒把体验隐喻成实体，这一隐喻仅对我自己成立。本文进而推进到本文最深的一步：**意识的难问题中的意识之定义，是唯我论式的，而又声称他人也有，这一设定本身就是自相矛盾的；真正该问的第一个问题，是为什么他人（从我的视角）没有意识，答案是视角问题；而我们据以相信他人有意识的真正根据，不是意识，是语言——我的内心独白，和我听到的他人所说的话，对我而言是串在一起的，我用同一套语义系统理解这三段内容，于是我对称地感觉别人似乎也能理解他自己说的话。** 语言，因此比意识更根本。可本文进而指出，这块新地基同样不稳：所有这些语言——我的内心独白与他人嘴上说的话——只是同等地被呈现给我，除非我认定他人（从我视角判断）正常的说话活动本身就证明了他的理解能力，否则这里完全没有证据表明他人理解任何语言；**哲学僵尸问题，等价于中文房间问题**，二者都可以对称地宣称除自己外无人理解任何语言，而这就是唯我论——一具躯体的输出只是输入与结构的函数，我们只能从输出反推结构，不可能从输出读出结构之外的任何新东西，这一结构，就是个中文房间，或哲学僵尸。本文如实呈现由此产生的两个分岔，不代为择定：其一，用同一套语义系统顺畅处理全部语言输入，故猜想他人也用同一套系统理解，是一个更简约、更合适的**猜想**；其二，把"理解"一词分两种情况理解——对自己与对他人——由此认为只有自己才真正理解，是**唯我论**，且这也应被还原到它最彻底的程度：只有自己才理解语言。本文最后指出，这两个分岔之所以看似互不相容、却各自自洽，是因为它们分属**两种互不兼容的叙事风格**：**其一，物理／结构的叙事**——在此风格下，"它"只能指那具躯体，它的理解只能指一种物理过程本身，其结构中并不存在任何可以充当"理解之主体"的东西；**其二，玄幻或灵异风格，即灵魂论的现代版本**——唯有在此风格下，"理解"才预设一个主体、才问得出"它是否真的懂了"。两种叙事对"理解"一词的语法安排互斥，不可通约，不是同一件事的两种描述深度。本文据此指出：**我们此刻正用第二种叙事风格与"你们"探讨这一切，这只说明说话者（作为哲学家）此刻仍在使用这套灵魂论的现代版本，丝毫不能反过来证明"你们"具有这套叙事所预设的那个理解主体——使用一套叙事，从不等于这套叙事所断言的东西为真。** 本文由此发展的推论是：他人的主观感受、乃至他人是否理解语言，其证据基础，皆比我们通常以为的要薄弱得多——不是因为它们已被证伪，是因为支撑这一整套信念的地基，一路追下去，最终落在了一种从未被独立证成、只是被言说习惯挪用的叙事风格上。本文守数条须格外郑重划定的界线：这一论证不触及第一人称的确定性；这一论证不构成停止与他人交流、信任、协作的理由——语言的社会实践，无论这一哲学诊断如何，仍完整地、必要地运转着，本文自身能被写出、被期待被读者理解，正是这一实践性事实的明证；且这一切绝不构成忽视他人报告之痛苦、感受或福祉的许可证。至于他人是否理解语言、是否具有主观体验，这一问题该被如何最终对待，本文不下判决，明确将其悬置为**尚待研究**。本文这一篇，自己也是用灵魂论现代版的叙事风格、向"你们"说出的一段内心独白的外化；它是指月之指；指可弃，月不可弃。 ## Abstract This paper rewrites and advances an argument, driving it from "consciousness" as its target toward a deeper, less avoidable foundation—**the understanding of language.** Its starting judgment: **if we define another's thought as their inner monologue, this dissolves the hard problem of consciousness—we no longer need to posit that another possesses a "consciousness" beyond the inner monologue.** Unpacking this requires prying apart the trio bundled into "a person has thoughts"—internal, invisible, seemingly uncertain—and tightening two prior points of ambiguity: **there is no "felt sense of certainty" separable from knowing itself; the inner monologue just is the subjective experience—by comparison with the objective view another takes of me; there is no further question of how the monologue is subjectively felt.** Subjective/objective mark not two kinds of thing but one event named differently by reference frame. The paper then squarely dismantles the strongest historical rebuttal—zombie conceivability: if a zombie's inner monologue genuinely describes its surroundings, then, by the established identity, it is already feeling that world, undoing "feels like nothing"; if the description is not genuine, the zombie's founding premise of functional identity goes unmet. The paper thereby diagnoses the hard problem's true ailment: **it asks why another can have consciousness while never arguing for what it presupposes—that another possesses something equivalent to my consciousness; this is the emperor's new clothes.** The concept of the philosophical zombie is, in truth, an extreme form of solipsism. The paper connects this to a deeper structure—**a subject beyond the subject, a cause beyond the cause**—and supplies an accounting of "soul" as the nominalization of a sense of agency, a sense of causation, a sense of dissociation, valid only for oneself, exactly as anger reifies a feeling into a solid. The paper then advances to its deepest move: **the definition of consciousness within the hard problem is solipsistic, yet claims another has it too—this setup is itself self-contradictory; the real first question is why another (from my viewpoint) lacks consciousness, and the answer is a matter of vantage point; and what we actually rely on to believe another has consciousness is not consciousness at all, but language—my inner monologue and the speech I hear from another are, for me, interlaced into one stream, understood through a single semantic system, and it is this symmetry that grounds my feeling that another likewise understands his own speech.** Language is thus more fundamental than consciousness. Yet the paper shows this new foundation is equally unstable: all this language—my inner monologue and another's spoken words—is presented to me on equal footing, and there is no evidence whatsoever that another understands any language, unless I take his (from my vantage, apparently normal) speech activity itself as proof of his comprehension. **The problem of the philosophical zombie is equivalent to the Chinese Room problem**—both can symmetrically claim that no one but oneself understands any language, and this just is solipsism; a body's output is merely a function of its input and structure, from which we can only infer structure, never anything beyond it—this structure just is a Chinese Room, or a zombie. The paper faithfully presents, without adjudicating, the two forks this yields: that positing another's comprehension, via the same semantic system handling all linguistic input smoothly, is the more parsimonious, apt guess; or that splitting "understanding" into two senses—for oneself and for another—yields solipsism, radicalized to its fullest extent: only oneself understands language at all. The paper argues these two forks appear incompatible yet each self-consistent because they belong to **two mutually incompatible narrative styles**: **the physical/structural narrative**, in which "it" can only denote the body, its "understanding" can only denote a physical process itself, and its structure contains nothing that could serve as a subject of understanding; and **the fantastical or supernatural style, the modern version of soul-theory**, in which alone "understanding" presupposes a subject, and one can ask whether it truly grasps. These two grammars for the word "understanding" are mutually exclusive, not two depths of description of one thing. The paper thereby concludes: **that we are presently conversing with "you" in the second narrative style shows only that the speaker, as a philosopher, is currently still using this modern soul-theory; it in no way proves that "you" possess the subject this narrative presupposes—using a narrative never entails that what it asserts is true.** The paper thus advances the inference that the evidential basis for another's subjective experience, and for another's linguistic understanding, is considerably weaker than ordinarily supposed—not because either has been disproven, but because the whole edifice, traced to its foundation, rests on a narrative style never independently established, only borrowed by habit of speech. The paper guards several lines with particular gravity: this argument does not touch first-person certainty; it is by no means a reason to cease communicating, trusting, or cooperating with others—the social practice of language continues to run, fully and necessarily, regardless of this philosophical diagnosis, and that this very paper can be written and expected to be understood by a reader is itself proof of that practical fact; and none of this licenses, in the slightest, disregarding another's reported suffering, feeling, or welfare. How the question of another's linguistic comprehension and subjective experience ought ultimately to be settled, the paper renders no verdict on, leaving it expressly suspended as **still under study**. This very paper is itself an externalization of an inner monologue, addressed to "you" in the modern style of soul-theory; it is a finger pointing at the moon; the finger may be discarded, the moon may not.