Xiaoyang Yu: Publications

16

那是另一道题：论”若当初不这样”恰如”若是3+3”

关于过去的反事实设想，恰如设想发生的不是1+1而是3+3——结果就不是2而是6。 “若是3+3，则得6”——这句话完全正确，谁也驳不倒。而它对”实际发生的是1+1”这件事，一个字也没说。你不是在讲那道算式，你在讲另一道算式。反事实的处境一模一样。“若钥匙没丢，我现在就进屋了”——作为一句蕴含，它也许没错；可它讲的是另一道题。由此说清了它为什么听起来那么有力：因为算得对。那份笃定不是关于世界的，是关于运算的：3+3确实得6，这一点毫无疑问。而我们把”这一步推得对”的确定感，误当成了”这件事本来会那样”的确定感。两种确定感的分别，可以一句话说清：一句蕴含的正确，只保证从那个前提到那个结论这一步没走错；它不保证那个前提是真的。而这里不必说”更不保证它曾经差一点是真的”——“差一点是真的”就是”不是真的”，加上一个语气。那个词没有刻度：多近算差一点？只有两档：是真的，不是真的。中间没有第三档。而更要紧的一步是：即使那个反事实宇宙是可行的，它所对应的也不是1+1的世界。 3+3不需要任何人去证明它”可行”——它可行得毫无疑问：算得出来，结果确定，内部无矛盾。而它照样不对应1+…Read more
关于过去的反事实设想，恰如设想发生的不是1+1而是3+3——结果就不是2而是6。 “若是3+3，则得6”——这句话完全正确，谁也驳不倒。而它对”实际发生的是1+1”这件事，一个字也没说。你不是在讲那道算式，你在讲另一道算式。反事实的处境一模一样。“若钥匙没丢，我现在就进屋了”——作为一句蕴含，它也许没错；可它讲的是另一道题。由此说清了它为什么听起来那么有力：因为算得对。那份笃定不是关于世界的，是关于运算的：3+3确实得6，这一点毫无疑问。而我们把”这一步推得对”的确定感，误当成了”这件事本来会那样”的确定感。两种确定感的分别，可以一句话说清：一句蕴含的正确，只保证从那个前提到那个结论这一步没走错；它不保证那个前提是真的。而这里不必说”更不保证它曾经差一点是真的”——“差一点是真的”就是”不是真的”，加上一个语气。那个词没有刻度：多近算差一点？只有两档：是真的，不是真的。中间没有第三档。而更要紧的一步是：即使那个反事实宇宙是可行的，它所对应的也不是1+1的世界。 3+3不需要任何人去证明它”可行”——它可行得毫无疑问：算得出来，结果确定，内部无矛盾。而它照样不对应1+1。于是”可行”和”对应”是两件事，而我们一向把前者当成后者的证据。可行是那一侧内部的性质；对应是两侧之间的关系。而两侧之间，没有任何关系。这也把”请拿出计算来”那一路问法摆到了它该在的位置：那不是主战场。就算那份计算交出来了，交上来的也是另一道题的答案。这个类比还带出两件事。其一：式子不会自己变，是有人换了一个数。没有人会无缘无故去设想3+3——“若是3+3”这句话之所以被说出来，是有人先把3写在了那儿。其二：结果不回流。算3+3不是白算，你确实得到了一个正确的结果；可那个结果属于3+3那道题，它不流回1+1那儿去。同样，一个反事实推得再周密，周密全部留在那一侧。而最见分晓的是一处对照：没有人为”没能是3+3”而懊恼。同样的形式，同样正确的一步推理——一个不带任何东西，一个能让人半夜坐起来。所以那份不甘，不是从那一步推理里来的：推理两边一模一样。它从别处来。而本文不追那个别处——本文只指出：它不在这一步里。 Abstract A counterfactual about the past is just like supposing that what occurred was not 1+1 but 3+3—whereupon the result is not 2 but 6. “If 3+3, then 6”—entirely correct; no one can refute it. And about the fact that what occurred was 1+1, it says not a word. You are not discussing that expression; you are discussing another one. Counterfactuals are in exactly this position. “If the keys had not been lost, I would be inside by now”—as an implication it may well be correct; but it is about another problem. This explains why such sentences sound so forceful: because the arithmetic is right. That confidence is not about the world but about the operation: 3+3 does make 6, beyond doubt. And we mistake the certainty of “this step is correctly taken” for the certainty of “this is how it would have gone.” The difference between the two certainties can be put in a line: the correctness of an implication guarantees only that the step from that premise to that conclusion was not misdrawn; it does not guarantee that the premise is true. And one need not add “nor that it once nearly was”—“nearly true” just is “not true,” with a tone attached. The word has no scale: how near counts as nearly? There are two settings only: true, and not true. There is no third. More important still: even if that counterfactual universe is feasible, what it corresponds to is not the world of 1+1. No one needs to prove 3+3 “feasible”—it is feasible beyond doubt: computable, determinate, internally consistent. And it still does not correspond to 1+1. So feasibility and correspondence are two different things, and we have always taken the former as evidence for the latter. Feasibility is a property internal to that side; correspondence is a relation between the two sides. And between the two sides there is no relation at all. This also puts the demand “produce the computation” in its place: that is not the main field. Even were the computation delivered, what is delivered is the answer to another problem. The analogy brings out two further things. First: the expression does not change itself; someone changed a number. No one supposes 3+3 for no reason—“if 3+3” gets said because someone first wrote a 3 there. Second: the result does not flow back. Computing 3+3 is not wasted—you do get a correct result; but that result belongs to 3+3, and does not flow back to 1+1. Likewise, however carefully a counterfactual is worked out, the care stays entirely on that side. Most telling is one comparison: no one regrets that it was not 3+3. The same form, the same correctly drawn step—one carries nothing, the other can wake a person at night. So that sting does not come from the step of reasoning: the reasoning is identical on both sides. It comes from elsewhere. And the paper does not pursue that elsewhere—it points out only this: it is not in this step.

Metaphysics and Epistemology Value Theory Science, Logic, and Mathematics
18

两边都没有解释：论”我有意识”证明不了宇宙非模拟

我所体验到的自己的”意识”，说不定证明了宇宙并非模拟？本文的判断是：证明不了。而理由有三层，一层比一层靠里。第一层：控制那一半先掉。我感觉我的意识能控制我的身体的运动。可理论上，我的身体的物理运动完全可以用模拟中的一个零件来解释——即实际并不由我的意识控制。所以”我能动我的手”这件事，一分证据也不提供：它在两边都说得通。于是剩余的就只剩主观角度的感知了。第二层：而这份剩余还能再削一次。我说得出的关于它的一切，仍然是零件在报告。“我看见红”“我觉得难受”“就是有那么一种感觉”——这些都是那一段活动的输出，跟别的输出没有分别。说不出口的时候，它什么也不是；一说出口，就又落回可被零件解释的那一栏。（须说清一处：那份体验实实在在地发生了。取消的不是它，是它作为判据的资格。）第三层：而就算把那份剩余整个留着，那个论证的形式也不成立。要拿意识证明”非模拟”，得有一条判别式：有意识，所以非模拟。而这一步要求一个前提：模拟里没有意识。这个前提从哪儿来？没有样本可数——我们手上只有一个宇宙、一个我，而这个宇宙是不是模拟，正是待定的那件事。所以那个前提，恰恰要用待证的结论来立：…Read more
我所体验到的自己的”意识”，说不定证明了宇宙并非模拟？本文的判断是：证明不了。而理由有三层，一层比一层靠里。第一层：控制那一半先掉。我感觉我的意识能控制我的身体的运动。可理论上，我的身体的物理运动完全可以用模拟中的一个零件来解释——即实际并不由我的意识控制。所以”我能动我的手”这件事，一分证据也不提供：它在两边都说得通。于是剩余的就只剩主观角度的感知了。第二层：而这份剩余还能再削一次。我说得出的关于它的一切，仍然是零件在报告。“我看见红”“我觉得难受”“就是有那么一种感觉”——这些都是那一段活动的输出，跟别的输出没有分别。说不出口的时候，它什么也不是；一说出口，就又落回可被零件解释的那一栏。（须说清一处：那份体验实实在在地发生了。取消的不是它，是它作为判据的资格。）第三层：而就算把那份剩余整个留着，那个论证的形式也不成立。要拿意识证明”非模拟”，得有一条判别式：有意识，所以非模拟。而这一步要求一个前提：模拟里没有意识。这个前提从哪儿来？没有样本可数——我们手上只有一个宇宙、一个我，而这个宇宙是不是模拟，正是待定的那件事。所以那个前提，恰恰要用待证的结论来立：要知道”模拟里没有意识”，得先有一个已知是模拟的世界去看看——而我们连自己这个是不是都不知道。最后一层，也是最省的一层：两边都没有解释。假如宇宙非模拟，我的意识应被如何解释？同样没有解释。若”非模拟”能解释意识而”模拟”不能，那还算一分证据。可两边一样地交不出解释。而一样东西在两个假说下都同样无解，它就不在这两者之间表态——它不偏向任何一边，因为它对两边一样地不合。所以：在没有其他途径证明宇宙是或不是模拟的前提下，我在身份不明的宇宙中有意识，而这并没有任何令人满意的解释——那么我有意识，就不能用来证明宇宙非模拟。须注意结论的形状：本文不主张”意识证明宇宙是模拟”。那要另一套东西，本文一份也交不出。本文只主张那一条被撤销：这一件证物不作数。 Abstract Might the “consciousness” I experience in myself prove the universe is not a simulation? The paper’s answer is no. Three layers of reason, each further in. First: the control half falls at once. I feel my consciousness governs my body’s motion. But in principle my body’s physical motion can be entirely accounted for by a part inside a simulation—that is, not in fact governed by my consciousness. So “I can move my hand” provides no evidence at all: it holds on both accounts. What remains, then, is only subjective perception. Second: this remainder can be pared once more. Everything I can say about it is still a part reporting. “I see red,” “I feel awful,” “there is just something it is like”—all outputs of that stretch of activity, no different from other outputs. Unspoken, it is nothing; spoken, it falls back into what a part can account for. (One thing must be clear: the experience really occurs. What is withdrawn is not the experience but its standing as evidence.) Third: even granting the remainder entire, the argument’s form fails. To prove “not a simulation” from consciousness requires a test: there is consciousness, therefore no simulation. And that step needs a premise: simulations contain no consciousness. Where would that premise come from? There are no samples to count—we have one universe and one self, and whether this universe is a simulation is the very thing at issue. So the premise must be established by the conclusion it is meant to establish: to know that simulations lack consciousness, one must first have a world known to be a simulation and go look—and we do not know even about our own. Last and cheapest: neither side explains it. Suppose the universe is not a simulation—how then is my consciousness to be explained? There is likewise no explanation. If “not a simulation” explained consciousness while “simulation” did not, that would count for something. But neither delivers an explanation. And a thing equally unexplained under two hypotheses does not take a side between them—it favours neither, because it fits neither. Hence: absent any other route to showing whether the universe is or is not a simulation, my having consciousness in a universe of unknown status, with no satisfying explanation available, cannot be used to prove the universe is not a simulation. Note the shape of the conclusion: the paper does not claim that consciousness proves the universe is a simulation. That would need something else, which the paper cannot supply. It claims only that this exhibit is struck: it does not count.

Science, Logic, and Mathematics Value Theory Metaphysics and Epistemology
9

一边必然，一边偶然：论”判不出来”是一条很强的特征

根据本文对世界的描述，如果世界如所描述的那样，那么它完全可以是个模拟。而更值得说的是下一句：没有和模拟的区别，这本身是个很强的特征。先说清”判不出来”的来由：在模拟的内部，是无法判断这是模拟还是非模拟的——而这本来就是模拟内部零件的特征。判别要一次比对，比对要两样东西：这一个，和一个不是这一个的东西。而零件手上永远只有这一个——它接触到的每一样东西，都在里面。没有第二样可比，就没有判别这回事。所以”判不出来”不是能力不足，是这个位置上不存在那个动作。而”本来”两个字最省：这不是那个模拟做得好，也不是我们不够聪明——做得再糙的模拟，里面的零件也一样判不出来，因为它照样只有这一个可摸。它还顺手废掉两条常被搬出来的路。“找漏洞”——漏洞得能被认出来，而认出来这个动作只能在里面做；一个被认出的漏洞，只不过是里面的又一样东西。“往细看总能看出破绽”——细到什么程度都还是在里面看；粒度换多少档，位置一档没换。而要害在于两边并不对称。在模拟那一侧，“判不出来”是必然的——身在其中的零件，位置决定它没有第二样可比。在非模拟那一侧，“判不出来”只是偶然的——一个非模拟的世界完全可以是通…Read more
根据本文对世界的描述，如果世界如所描述的那样，那么它完全可以是个模拟。而更值得说的是下一句：没有和模拟的区别，这本身是个很强的特征。先说清”判不出来”的来由：在模拟的内部，是无法判断这是模拟还是非模拟的——而这本来就是模拟内部零件的特征。判别要一次比对，比对要两样东西：这一个，和一个不是这一个的东西。而零件手上永远只有这一个——它接触到的每一样东西，都在里面。没有第二样可比，就没有判别这回事。所以”判不出来”不是能力不足，是这个位置上不存在那个动作。而”本来”两个字最省：这不是那个模拟做得好，也不是我们不够聪明——做得再糙的模拟，里面的零件也一样判不出来，因为它照样只有这一个可摸。它还顺手废掉两条常被搬出来的路。“找漏洞”——漏洞得能被认出来，而认出来这个动作只能在里面做；一个被认出的漏洞，只不过是里面的又一样东西。“往细看总能看出破绽”——细到什么程度都还是在里面看；粒度换多少档，位置一档没换。而要害在于两边并不对称。在模拟那一侧，“判不出来”是必然的——身在其中的零件，位置决定它没有第二样可比。在非模拟那一侧，“判不出来”只是偶然的——一个非模拟的世界完全可以是通的：有外面，有别的东西，可以走出去看。那儿”判不出来”就不是位置自带的，只是眼下没查到；它本可以判得出来。于是”判不出来”不是零信息：它跟一边必然相容，跟另一边只是可能相容。但要在这儿刹住：它不给出”所以是模拟”。它给出的是——我们观察到的这份处境，在一边是这个位置的定义，在另一边只是一次尚未成功。而由此问法翻转：不是本文要证明世界是模拟；是主张”不是模拟”的那一边，得指出那个只有非模拟才有、而我们至今没找到的东西，并说清它为什么至今没找到。 “我们判不出来”这句话，在那一边是一份欠账；在这一边，什么也不欠。 Abstract On the description of the world set out here, if the world is as described, then it may perfectly well be a simulation. More worth saying is what follows: that there is no distinguishing it from a simulation is itself a strong feature. First, why nothing can be told: inside a simulation one cannot judge whether this is a simulation or not—and this is simply what it is to be a part inside one. Judging requires a comparison, and comparison requires two things: this one, and something that is not this one. A part has only ever this one—everything it touches is inside. With nothing else to compare, there is no act of judging at all. So “cannot tell” is not a deficiency of capacity; it is that this position affords no such act. And “simply” is the economical word: this is not because the simulation is well made, nor because we are not clever enough—however crudely made, its parts equally cannot tell, since they too have only this one to touch. Two familiar routes are thereby closed. “Find a glitch”—a glitch must be recognizable, and recognizing can only be done inside; a recognized glitch is merely one more thing inside. “Look finely enough and a flaw will show”—however finely one looks, one is still looking from inside; change the grain as often as you like, the position does not change once. The crux is that the two sides are not symmetrical. On the simulation side, “cannot tell” is necessary—being inside, a part has by position nothing else to compare with. On the non-simulation side, “cannot tell” is merely contingent—a non-simulated world may perfectly well be open: an outside, other things, somewhere to go and look. There “cannot tell” is not given by position but only by our not yet having found out; it could have been told. So “cannot tell” is not zero information: it is necessarily compatible with one side and only possibly compatible with the other. But halt here: it does not yield “therefore it is a simulation.” It yields this—the situation we observe is, on one side, the definition of the position, and on the other, merely a failure so far. Hence the question reverses: it is not for this paper to prove the world a simulation; it is for whoever maintains it is not to point out the thing only a non-simulation would have, which we have not found—and to say why it has not been found. “We cannot tell” is a debt on that side; on this side it owes nothing.

Metaphysics and Epistemology Science, Logic, and Mathematics Value Theory
11

只借一半：论”宇宙是个模拟”这个比喻，贴切在哪儿，会把人带偏在哪儿

在本文所描述的世界图景中，我与宇宙的关系，相当于宇宙是个模拟。 “相当于”三个字要守住：这是一个比喻，不是一个主张。本文不主张宇宙由谁运行，不问外面有什么，也不提供任何关于起源的说法——那些需要证明，而本文一份也交不出。这个比喻要借的，是那份处境的形状。贴切之处有五：一、演算不询问我。下一帧怎么算，不查我这边有没有意见。人体并不能屏蔽物理定律——物理定律也不是一种可以被某种物质屏蔽的射线；人体本身就是一种普通的物体，只是我们自以为它不普通。二、只有一份存档，没有分支可以回读。说”我本来可以不那样”，要拿两条线来比：实际走的那一条，和”若我不那样会走的那一条”。而后一条从来没有发生过——它不是被锁了、被删了，是根本没有被写过；没被写过的东西，不需要谁来藏。三、我不是玩家，是被算的一部分。手臂里那些粒子，跟别处的粒子同一套规矩；我不在系统外面握着手柄。四、我以为自己在下指令，其实是在读输出。一个想法冒出来，不是我从外面给系统喂了一条指令，是那一段算到了那儿；而后面的动作，是接着算的。五、模拟不为任何人的账目负责。我们把观察到的事情整理成账目——我的包、我的位置、我该…Read more
在本文所描述的世界图景中，我与宇宙的关系，相当于宇宙是个模拟。 “相当于”三个字要守住：这是一个比喻，不是一个主张。本文不主张宇宙由谁运行，不问外面有什么，也不提供任何关于起源的说法——那些需要证明，而本文一份也交不出。这个比喻要借的，是那份处境的形状。贴切之处有五：一、演算不询问我。下一帧怎么算，不查我这边有没有意见。人体并不能屏蔽物理定律——物理定律也不是一种可以被某种物质屏蔽的射线；人体本身就是一种普通的物体，只是我们自以为它不普通。二、只有一份存档，没有分支可以回读。说”我本来可以不那样”，要拿两条线来比：实际走的那一条，和”若我不那样会走的那一条”。而后一条从来没有发生过——它不是被锁了、被删了，是根本没有被写过；没被写过的东西，不需要谁来藏。三、我不是玩家，是被算的一部分。手臂里那些粒子，跟别处的粒子同一套规矩；我不在系统外面握着手柄。四、我以为自己在下指令，其实是在读输出。一个想法冒出来，不是我从外面给系统喂了一条指令，是那一段算到了那儿；而后面的动作，是接着算的。五、模拟不为任何人的账目负责。我们把观察到的事情整理成账目——我的包、我的位置、我该得的那一份——而这些账目在记账之外不具有任何额外的因果力：撕掉房产证，房子一寸不动；撤销协议，已经交付的货位置不变。我记不记账，它照算。而这个比喻自带一批行李，每一件都会把人带偏，且必须逐件卸掉：行李一：外面有个人在跑它。本文不需要那个人，也不问外面。只借”我在里面”这一半，不借”外面有谁”那一半。行李二：本可以跑成别的样子。重开一局、改个种子、换个参数——而整条线正是拆这个的。这个比喻里最诱人的那一样，恰恰是它不能提供的。行李三：可以被暂停、被中断、被退出。没有出口，因为”出口”要一个外面。行李四：它是为了什么而跑的。目的要一个立目的的人，而那又回到行李一。行李五：它可以被识破。找到”渲染的边缘”、发现一个漏洞——而按本文别处那条，“认出它是什么”这个动作本身只能在被算的这一侧做。卸完之后剩下的，恰恰是本文别处已经写过的那些。所以这个比喻的用处，不在于它多告诉了我们什么，而在于它把一批分散的判断收成一个可以一眼看见的形状。它是一根手指，不是一个发现。 Abstract In the picture of the world set out here, my relation to the universe is as if the universe were a simulation. “As if” must be held to: this is a metaphor, not a thesis. The paper claims nothing about who runs the universe, asks nothing about an outside, and offers no account of origins—all of which would require proof, and the paper has none to offer. What the metaphor borrows is the shape of a situation. Five things fit: One: the computation does not consult me. How the next frame is computed does not check whether I have views. A human body cannot shield physical law—law is no ray that any substance could block; the body simply is an ordinary object, which we merely take to be extraordinary. Two: there is a single save file, and no branch to load. To say “I could have done otherwise” requires two lines: the one actually taken, and the one “that would have run had I not.” The second never occurred—not locked away, not deleted, but never written; what was never written needs no one to hide it. Three: I am not the player but part of what is computed. The particles in my arm follow the same rules as any others; I am not outside the system holding a controller. Four: I take myself to be issuing commands while I am reading output. A thought arising is not my feeding an instruction in from outside; it is the computation having reached that point, and what follows is the computation continuing. Five: the simulation answers to no one’s ledger. We organize what we observe into ledgers—my bag, my position, the share due to me—and these possess no causal force outside the bookkeeping: tear up a deed and the house does not move an inch; cancel an agreement and delivered goods stay where they are. Whether I keep books or not, it computes. But the metaphor arrives with baggage, each piece of which misleads and must be unpacked: One: someone outside is running it. The paper needs no such person and asks about no outside. It borrows the “I am inside” half only, not the “someone is outside” half. Two: it could have run otherwise. Restart, reseed, retune—and the whole line of work is precisely a dismantling of this. The most alluring item in the metaphor is the one it cannot supply. Three: it can be paused, interrupted, exited. There is no exit, because an exit requires an outside. Four: it is run for some purpose. A purpose requires someone to set it, which returns us to the first piece. Five: it can be seen through. Find the edge of the rendering, discover a glitch—but by the paper’s own account elsewhere, the act of “recognizing what it is” can only be performed on the computed side. What remains after unpacking is exactly what was written elsewhere. So the use of this metaphor is not that it tells us more, but that it gathers a set of scattered judgments into a shape one can see at a glance. It is a finger, not a discovery.

Science, Logic, and Mathematics Metaphysics and Epistemology Value Theory
9

应验被读成了改变：论科学只是预测

**计划不是工具，而是预测。同样，科学也只是预测，而非改变世界的工具。** **本文只给一条理由，而这条理由可以逐例去验：所谓"改变世界"的那些例子，逐个看都是应验，不是改变。** **先拿最像纯粹因果发起的那一个：命令。** **命令的内容，其实也是预测。**"明早六点集合"——**内容上，跟"明早六点他们会集合"一字不差**。**唯一的差别是：它被发送给了被预测的对象。** **而这一步正好解释了那份"力"从哪儿来的错觉**：预测行星时，行星收不到我的预测；**预测人时，人收得到。** 那句话进了被预测者的耳朵，成了他后来那一串动作的一段前情——**而我们把"预测被送达"读成了"预测在推动"。送达是真的，推动要那条不存在的第二线。** **而"服从"这个词，不是一段位移的性质：它是把下级的位移记在上级名下的方式**——**跟把砖的位移记在搬砖者名下，是同一个动作。** **这一例还有一处当场可验的：命令不达的那些，从来不进分母。** 一道命令下去，多数照办了——记在他名下；有些没照办——**记成"抗命""执行不力"，而不是"你这个预测算错了"**。**天文学家算错了说"我算错了"，…Read more
**计划不是工具，而是预测。同样，科学也只是预测，而非改变世界的工具。** **本文只给一条理由，而这条理由可以逐例去验：所谓"改变世界"的那些例子，逐个看都是应验，不是改变。** **先拿最像纯粹因果发起的那一个：命令。** **命令的内容，其实也是预测。**"明早六点集合"——**内容上，跟"明早六点他们会集合"一字不差**。**唯一的差别是：它被发送给了被预测的对象。** **而这一步正好解释了那份"力"从哪儿来的错觉**：预测行星时，行星收不到我的预测；**预测人时，人收得到。** 那句话进了被预测者的耳朵，成了他后来那一串动作的一段前情——**而我们把"预测被送达"读成了"预测在推动"。送达是真的，推动要那条不存在的第二线。** **而"服从"这个词，不是一段位移的性质：它是把下级的位移记在上级名下的方式**——**跟把砖的位移记在搬砖者名下，是同一个动作。** **这一例还有一处当场可验的：命令不达的那些，从来不进分母。** 一道命令下去，多数照办了——记在他名下；有些没照办——**记成"抗命""执行不力"，而不是"你这个预测算错了"**。**天文学家算错了说"我算错了"，而上级说的是"他们不听话"：同一处误差，一个归模型，一个归对象。而一个把所有失手都归给对象的预测者，命中率永远是百分之百。** **造一台电机**——不是把世界从一条路推到另一条路，**是"这样绕线、通这样的电流，就会这样转"这个预测成真了**。人做的是**把条件凑齐**；而凑条件那一串动作，**本身也在那一整条线上**。 **登月**——不是改变了月球，**是算准了**。 **疫苗**——不是拨转了疫情，**是"这样处理，人群里会是那样"应验了**。 **每一个都是预测成真。而我们把预测成真，读成了改变。** **为什么会这么读？因为预测里出现了一个人的动作。** 行星那边没有我的动作，所以我们只说"算准了"；**电机这边有我的动作，所以我们说"我做到了"。** **多出来的不是改变，是一个可以被记账的位置——而账落到谁身上，谁就成了因。** **而"改变世界"这句话，要的还是那条不存在的第二线**：有科学的这一条，和"若无科学会走的那一条"。**后一条从来没有。** 我们手上只有一整条扣得极紧的线：有过牛顿、有过麦克斯韦、有过发电机、有过这一切——**而扣得紧不是推动。** **最后须说清一处，否则整篇会被读成反智：本文不主张科学没用，也不主张不该做。** 那些事**确实发生了**——人确实绕了线，确实通了电，机器确实转了起来。**而"改变"和"有用"这两个词，要的都是同一条没有的线。** **发生了，是一句可以说的；改变了、有用，是两句要那条线的。** ## Abstract **A plan is not a tool but a prediction. Likewise, science is only prediction, not a tool for changing the world.** **The paper offers one ground, and it can be checked case by case: the alleged instances of "changing the world" are, one by one, fulfilments rather than changes.** **Begin with the case that most resembles pure causal initiation: an order.** **The content of an order is also a prediction.** "Assemble at six tomorrow"—**in content, word for word the same as "they will assemble at six tomorrow."** **The only difference: it is sent to the object predicted.** **And this explains the illusion of force**: a planet cannot receive my prediction; a person can. The sentence enters the ear of the one predicted and becomes part of the antecedents of what he does—**and we read "the prediction was delivered" as "the prediction is pushing." Delivery is real; pushing requires the missing second line.** **And "obedience" is not a property of a displacement: it is the way a subordinate's displacement is filed under a superior's name**—**the same act as filing a brick's displacement under the man who carried it.** **This case also carries a test: orders not obeyed never enter the denominator.** Most are carried out—filed under his name; some are not—**recorded as "insubordination" or "poor execution," never as "your prediction was wrong."** **An astronomer says "I miscalculated"; a superior says "they would not listen": the same error, one assigned to the model, the other to the object. And a predictor who assigns every failure to the object has a hit rate of one hundred percent.** **Building a motor**—not pushing the world from one path onto another, **but the prediction "wind it thus, pass such a current, and it turns thus" coming true**. What people do is **assemble the conditions**; and that assembling **is itself on the same single line**. **Reaching the moon**—not changing the moon, **but calculating correctly**. **A vaccine**—not diverting the epidemic, **but "treat it thus and the population will behave thus" coming true**. **Every case is a prediction fulfilled. And we read fulfilment as change.** **Why so read? Because a human action appears inside the prediction.** With planets there is no action of mine, so we say only "the calculation was right"; **with a motor there is, so we say "I did it."** **What is added is not change but a place where the entry can be filed—and whoever the entry is filed under becomes the cause.** **And "science changed the world" still calls for the missing second line**: the one with science, and the one "that would have run without it." **There is no second.** We hold only one tightly interlocking line—Newton, Maxwell, dynamos, all of it—**and tight interlocking is not pushing.** **One point must be clear, or the paper reads as anti-intellectual: it does not claim science is useless or should not be done.** Those things **did happen**—wire was wound, current passed, the machine turned. **But "change" and "useful" both call for the same missing line.** **"It happened" can be said; "it changed things" and "it was useful" both require that line.**

Metaphysics and Epistemology Science, Logic, and Mathematics Value Theory
11

残留：论关于过去的反事实，是一份失败的预测被原样端了出来

**关于过去某段时间的反事实设想，可以被理解成：发生于这段时间之前的一个预测失败了，而这个反事实就是对该失败所给出的解释的一部分——即那个预测的内容。** **我的钥匙丢了。"钥匙没丢"可以是我过去的一个预测；钥匙丢了，则表明这个预测失败了。** **而"读回"这个词是对的，因为它本来就是。** 我出门时确实预期钥匙会在兜里——**那份预期在丢之前就已经存在了**。丢了之后我说的"要是钥匙没丢"，**内容跟那份预期一字不差。它不是事后新造的东西，它是那份预期被原样端了出来。** **由此解释了一件一直没被解释的事：反事实的内容为什么那么详尽，又那么不受检查。** 因为**它根本不是内容，它是残留**——**而没有人会去检查一份自己出门时随手带着的预期。** **"解释的一部分"这个位置，须说准。** 预测失败之后人要一个交代，**而一份交代由两半组成**：预测的内容是什么，以及它为什么没兑现。我们通常只把后一半叫解释（"兜太浅了""走得太急"）；**可前一半也在交代里，而且它是先说的那一半。**"要是钥匙没丢"**在做的事，是把失败了的那份预测重新念一遍**——**不念出来，就没有一个…Read more
**关于过去某段时间的反事实设想，可以被理解成：发生于这段时间之前的一个预测失败了，而这个反事实就是对该失败所给出的解释的一部分——即那个预测的内容。** **我的钥匙丢了。"钥匙没丢"可以是我过去的一个预测；钥匙丢了，则表明这个预测失败了。** **而"读回"这个词是对的，因为它本来就是。** 我出门时确实预期钥匙会在兜里——**那份预期在丢之前就已经存在了**。丢了之后我说的"要是钥匙没丢"，**内容跟那份预期一字不差。它不是事后新造的东西，它是那份预期被原样端了出来。** **由此解释了一件一直没被解释的事：反事实的内容为什么那么详尽，又那么不受检查。** 因为**它根本不是内容，它是残留**——**而没有人会去检查一份自己出门时随手带着的预期。** **"解释的一部分"这个位置，须说准。** 预测失败之后人要一个交代，**而一份交代由两半组成**：预测的内容是什么，以及它为什么没兑现。我们通常只把后一半叫解释（"兜太浅了""走得太急"）；**可前一半也在交代里，而且它是先说的那一半。**"要是钥匙没丢"**在做的事，是把失败了的那份预测重新念一遍**——**不念出来，就没有一个可以被解释的对象**：你得先说清"本来该是怎样"，才谈得上"为什么不是"。**而这也说清了它为什么带着不甘：它是一份败诉通知的开头。** **一个当场可做的检验：凡是我事前没有预期过的事，事后也生不出反事实。** 昨天下午三点，某个陌生人在另一个城市摔了一跤——**我事前没预期过任何相关的东西，事后也说不出"要是他没摔"。** 不是说不出这句话，**是说出来不带任何东西**。**而"我出门带了钥匙"预期过，所以事后那句话满载。两边严丝合缝：有预期的地方就有反事实，没预期的地方就没有。** **它还给"其余照抄"一个更省的解释。** 那五分钟的路线、到家的时刻、门还锁着——**不是事后被抄来的，也不只是"从来没被删掉"**；**它们是那份预期里本来就有的**：我出门时预期的不只是"钥匙在兜里"，**是一整套**——走那条路、五分钟、掏钥匙、开门、进屋。**那一整套是一次投出去的，所以事后它也一整套地回来。** ## Abstract **A counterfactual about some past stretch of time can be read as this: a prediction made before that stretch failed, and the counterfactual is part of the explanation given for that failure—namely, the content of the prediction.** **My keys are lost. "The keys were not lost" can be a prediction I once made; their being lost shows the prediction failed.** **"Read back" is the right phrase, because that is what it is.** Leaving the house I did expect the keys to be in my pocket—**that expectation existed before the loss**. What I say afterwards, "if only the keys had not been lost," **matches that expectation word for word. It is not something newly made afterwards; it is that expectation brought out unchanged.** **This explains something long unexplained: why the content of a counterfactual is so detailed and so unexamined.** Because **it is not content at all but residue**—**and no one audits an expectation he carried out of the house without thinking.** **"Part of the explanation" must be stated precisely.** After a prediction fails one wants an account, **and an account has two halves**: what the prediction was, and why it did not hold. We usually call only the second half explanation ("the pocket was shallow," "I was in a hurry"); **yet the first half is in the account too, and is spoken first.** "If only the keys had not been lost" **is the failed prediction recited once more**—**and without reciting it there is nothing for the explanation to be about**: one must first state what should have been before one can ask why it was not. **Which also explains its sting: it is the opening line of a notice of loss.** **A test anyone can run: whatever I did not expect beforehand yields no counterfactual afterwards.** Yesterday at three, a stranger in another city stumbled—**I expected nothing about it, and afterwards I cannot say "if only he had not stumbled."** Not that the sentence cannot be uttered, **but that uttering it carries nothing.** **Whereas "I left the house with my keys" was expected, so the sentence afterwards is fully loaded. The two match exactly: where there was expectation there is a counterfactual, and where there was none there is none.** **It also gives a cheaper account of "the rest copied over."** The route, the arrival time, the door still locked—**not copied afterwards, nor merely "never deleted"**; **they were in the expectation from the start**: leaving the house I expected not merely "keys in pocket" but **a whole set**—that street, five minutes, keys out, door open, inside. **The set was cast at once, so it returns at once.**

Value Theory Metaphysics and Epistemology Science, Logic, and Mathematics
13

当作预测来读：论把"我的计划"换成"我对自己的预测"

**计划也可以不被理解成计划，而被理解成仅仅是对自己未来活动的预测——就像预测行星的轨道一样。** **这不是一个形而上学主张，只是换一个读法。而换完之后，"我本来可以不这样"自己就消失了。** **因为"计划"这个词自带执行者，"预测"不带。**"我计划明天九点出门"——句子里有一个下令的人和一件被下令的事；**"我预测我明天九点出门"——只有一个观察者和一件事。** 同一段活动，同样的内容，**一个带命令，一个只带观察。** **而这个读法的好处是：它可验，而计划那个读法不可验。** **预测行星轨道有对表这一步**：算出来、对一对、错了就改模型。**而"我计划"没有这一步**——做成了画钩，做不成就改单子："临时变了""其实不急""改天再说"。**清单当天就被改了，于是永远对得上。而一个不可能出错的预测，不是预测。** **换成预测读法，标准反而提高了。** 从此那几行字要接受同一种对待：**它算得准不准？我的命中率是多少？哪一类我一直算不准？** 而这些问题用计划那个读法一个也问不出来——**计划只有"执行到位没有"，没有"预测准不准"。** **最要紧的一处：天文学家从不说"…Read more
**计划也可以不被理解成计划，而被理解成仅仅是对自己未来活动的预测——就像预测行星的轨道一样。** **这不是一个形而上学主张，只是换一个读法。而换完之后，"我本来可以不这样"自己就消失了。** **因为"计划"这个词自带执行者，"预测"不带。**"我计划明天九点出门"——句子里有一个下令的人和一件被下令的事；**"我预测我明天九点出门"——只有一个观察者和一件事。** 同一段活动，同样的内容，**一个带命令，一个只带观察。** **而这个读法的好处是：它可验，而计划那个读法不可验。** **预测行星轨道有对表这一步**：算出来、对一对、错了就改模型。**而"我计划"没有这一步**——做成了画钩，做不成就改单子："临时变了""其实不急""改天再说"。**清单当天就被改了，于是永远对得上。而一个不可能出错的预测，不是预测。** **换成预测读法，标准反而提高了。** 从此那几行字要接受同一种对待：**它算得准不准？我的命中率是多少？哪一类我一直算不准？** 而这些问题用计划那个读法一个也问不出来——**计划只有"执行到位没有"，没有"预测准不准"。** **最要紧的一处：天文学家从不说"这颗行星本来可以走另一条"。** 他算错了，说的是**"我算错了"**，不是"行星本可以配合我一点"。**而我们对自己：没做到，说的是"我本来可以做到的"。** **同一个位置，一个把误差归给模型，一个把误差归给对象——而那个对象是我自己。** **所以"我本来可以"其实是"我算错了"的一个走了样的版本：把一次预测失准，读成了一次执行不力。** **而"可我确实能影响我自己，行星不听我的"这句反问不成立**：天文学家也影响不了行星，**可他照样算得准；准不准跟能不能影响无关。** 而若你真能影响自己，**那更该算得准才对——可我们对自己的预测，恰恰是最不准的那一类。** **这个读法还顺手拆掉几样东西。** 那份"先有步骤、后见实现"的自信，**在预测读法下要被重新计分**：画钩那一档粗到不能再粗（"去趟银行"——几点去、走哪条路都不影响画钩），**是我先把成功条件定得那么宽，它才那么容易被满足**；**那不是预测被证实，是描述被兑现**。而细到说哪句话那一档，**自信是从"骂人他会生气"借来的——可那一条几乎是同义反复**，不是知识，是定义的一部分。**在预测读法下，这些都成了应当报告的误差项，而不是可以略过的日常。** **最后须说清这个读法给不了什么：它不供应第二条线。** 换成预测，不等于"我算得准就说明我主导了走向"——**应验只给出预测和后来的事对得上，而对得上是两段之间的扣合，不是其中一段对另一段的推动。** ## Abstract **A plan may be read not as a plan but merely as a prediction of one's own future activity—as one predicts a planet's orbit.** **This is no metaphysical claim, only a change of reading. And once made, "I could have done otherwise" disappears of itself.** **For "plan" carries an executor; "prediction" does not.** "I plan to leave at nine tomorrow"—the sentence contains one who orders and something ordered; **"I predict I will leave at nine tomorrow"—only an observer and an event.** Same activity, same content: **one carries a command, the other only an observation.** **And the advantage of the reading is this: it is testable, while the plan reading is not.** **Predicting an orbit includes checking against the world**: compute, compare, revise the model when wrong. **"I plan" has no such step**—done means a tick, undone means the list gets amended: "things came up," "it wasn't urgent," "another time." **The list is amended the same day, so it always reconciles. And a prediction that cannot fail is not a prediction.** **The reading raises the standard.** Those lines must now face the same treatment: **how accurate was it? what is my hit rate? which kinds do I persistently get wrong?** None of which can even be asked under the plan reading—**a plan has only "was it carried out," never "was it accurate."** **The crucial point: an astronomer never says "this planet could have taken another path."** When wrong he says **"I miscalculated,"** not "the planet might have cooperated." **Whereas of ourselves, when we fall short, we say "I could have done it."** **Same position; one assigns the error to the model, the other to the object—and the object is myself.** **So "I could have" is a distorted version of "I miscalculated": a failed prediction read as a failure of execution.** **And the retort "but I can influence myself, while the planet cannot hear me" does not hold**: the astronomer cannot influence the planet either, **yet he predicts accurately; accuracy has nothing to do with influence.** And if you truly could influence yourself, **you ought to predict better—yet our predictions of ourselves are the least accurate we make.** **The reading dismantles a few other things.** The confidence bred by "steps set out, then seen realized" **must be rescored under it**: the tick is granted at the coarsest grain ("go to the bank"—what time, which route, none of it affects the tick), **and I set the success condition that wide before it was so easily met**; **that is not a prediction confirmed but a description discharged.** And at the finest grain—which words to say—**the confidence is borrowed from "insulting makes people angry," which is nearly a tautology**, not knowledge but part of the word's definition. **Under the prediction reading these become error terms to be reported, not daily matters to be passed over.** **Finally, what the reading does not supply: a second line.** Reading plans as predictions does not license "I predicted well, therefore I steered the course"—**fulfilment yields only that prediction and outcome match, and matching is an interlocking of two stretches, not the pushing of one by the other.**

Metaphysics and Epistemology Science, Logic, and Mathematics Value Theory
当作预测来读：论把"我的计划"换成"我对自己的预测"

**计划也可以不被理解成计划，而被理解成仅仅是对自己未来活动的预测——就像预测行星的轨道一样。** **这不是一个形而上学主张，只是换一个读法。而换完之后，"我本来可以不这样"自己就消失了。** **因为"计划"这个词自带执行者，"预测"不带。**"我计划明天九点出门"——句子里有一个下令的人和一件被下令的事；**"我预测我明天九点出门"——只有一个观察者和一件事。** 同一段活动，同样的内容，**一个带命令，一个只带观察。** **而这个读法的好处是：它可验，而计划那个读法不可验。** **预测行星轨道有对表这一步**：算出来、对一对、错了就改模型。**而"我计划"没有这一步**——做成了画钩，做不成就改单子："临时变了""其实不急""改天再说"。**清单当天就被改了，于是永远对得上。而一个不可能出错的预测，不是预测。** **换成预测读法，标准反而提高了。** 从此那几行字要接受同一种对待：**它算得准不准？我的命中率是多少？哪一类我一直算不准？** 而这些问题用计划那个读法一个也问不出来——**计划只有"执行到位没有"，没有"预测准不准"。** **最要紧的一处：天文学家从不说"…Read more
**计划也可以不被理解成计划，而被理解成仅仅是对自己未来活动的预测——就像预测行星的轨道一样。** **这不是一个形而上学主张，只是换一个读法。而换完之后，"我本来可以不这样"自己就消失了。** **因为"计划"这个词自带执行者，"预测"不带。**"我计划明天九点出门"——句子里有一个下令的人和一件被下令的事；**"我预测我明天九点出门"——只有一个观察者和一件事。** 同一段活动，同样的内容，**一个带命令，一个只带观察。** **而这个读法的好处是：它可验，而计划那个读法不可验。** **预测行星轨道有对表这一步**：算出来、对一对、错了就改模型。**而"我计划"没有这一步**——做成了画钩，做不成就改单子："临时变了""其实不急""改天再说"。**清单当天就被改了，于是永远对得上。而一个不可能出错的预测，不是预测。** **换成预测读法，标准反而提高了。** 从此那几行字要接受同一种对待：**它算得准不准？我的命中率是多少？哪一类我一直算不准？** 而这些问题用计划那个读法一个也问不出来——**计划只有"执行到位没有"，没有"预测准不准"。** **最要紧的一处：天文学家从不说"这颗行星本来可以走另一条"。** 他算错了，说的是**"我算错了"**，不是"行星本可以配合我一点"。**而我们对自己：没做到，说的是"我本来可以做到的"。** **同一个位置，一个把误差归给模型，一个把误差归给对象——而那个对象是我自己。** **所以"我本来可以"其实是"我算错了"的一个走了样的版本：把一次预测失准，读成了一次执行不力。** **而"可我确实能影响我自己，行星不听我的"这句反问不成立**：天文学家也影响不了行星，**可他照样算得准；准不准跟能不能影响无关。** 而若你真能影响自己，**那更该算得准才对——可我们对自己的预测，恰恰是最不准的那一类。** **这个读法还顺手拆掉几样东西。** 那份"先有步骤、后见实现"的自信，**在预测读法下要被重新计分**：画钩那一档粗到不能再粗（"去趟银行"——几点去、走哪条路都不影响画钩），**是我先把成功条件定得那么宽，它才那么容易被满足**；**那不是预测被证实，是描述被兑现**。而细到说哪句话那一档，**自信是从"骂人他会生气"借来的——可那一条几乎是同义反复**，不是知识，是定义的一部分。**在预测读法下，这些都成了应当报告的误差项，而不是可以略过的日常。** **最后须说清这个读法给不了什么：它不供应第二条线。** 换成预测，不等于"我算得准就说明我主导了走向"——**应验只给出预测和后来的事对得上，而对得上是两段之间的扣合，不是其中一段对另一段的推动。** ## Abstract **A plan may be read not as a plan but merely as a prediction of one's own future activity—as one predicts a planet's orbit.** **This is no metaphysical claim, only a change of reading. And once made, "I could have done otherwise" disappears of itself.** **For "plan" carries an executor; "prediction" does not.** "I plan to leave at nine tomorrow"—the sentence contains one who orders and something ordered; **"I predict I will leave at nine tomorrow"—only an observer and an event.** Same activity, same content: **one carries a command, the other only an observation.** **And the advantage of the reading is this: it is testable, while the plan reading is not.** **Predicting an orbit includes checking against the world**: compute, compare, revise the model when wrong. **"I plan" has no such step**—done means a tick, undone means the list gets amended: "things came up," "it wasn't urgent," "another time." **The list is amended the same day, so it always reconciles. And a prediction that cannot fail is not a prediction.** **The reading raises the standard.** Those lines must now face the same treatment: **how accurate was it? what is my hit rate? which kinds do I persistently get wrong?** None of which can even be asked under the plan reading—**a plan has only "was it carried out," never "was it accurate."** **The crucial point: an astronomer never says "this planet could have taken another path."** When wrong he says **"I miscalculated,"** not "the planet might have cooperated." **Whereas of ourselves, when we fall short, we say "I could have done it."** **Same position; one assigns the error to the model, the other to the object—and the object is myself.** **So "I could have" is a distorted version of "I miscalculated": a failed prediction read as a failure of execution.** **And the retort "but I can influence myself, while the planet cannot hear me" does not hold**: the astronomer cannot influence the planet either, **yet he predicts accurately; accuracy has nothing to do with influence.** And if you truly could influence yourself, **you ought to predict better—yet our predictions of ourselves are the least accurate we make.** **The reading dismantles a few other things.** The confidence bred by "steps set out, then seen realized" **must be rescored under it**: the tick is granted at the coarsest grain ("go to the bank"—what time, which route, none of it affects the tick), **and I set the success condition that wide before it was so easily met**; **that is not a prediction confirmed but a description discharged.** And at the finest grain—which words to say—**the confidence is borrowed from "insulting makes people angry," which is nearly a tautology**, not knowledge but part of the word's definition. **Under the prediction reading these become error terms to be reported, not daily matters to be passed over.** **Finally, what the reading does not supply: a second line.** Reading plans as predictions does not license "I predicted well, therefore I steered the course"—**fulfilment yields only that prediction and outcome match, and matching is an interlocking of two stretches, not the pushing of one by the other.**

Science, Logic, and Mathematics Value Theory Metaphysics and Epistemology
14

没有"未查"那一栏：论一本两栏的账

**没看出不可能，就被当做了可能。** **而这之所以能成立，是因为那本账上没有"未查"那一栏。** **若有三栏——不可能、可能、未查——那么绝大多数东西该待在第三栏里。** 我们查过的事屈指可数，**没查过的无穷无尽。** 按理，第三栏该是最挤的一栏。 **可它不存在。只有两栏。** **于是凡是没被判进"不可能"的，自动落进"可能"。** **这不是一次推论，是排水：没有别的地方可去。** **而这个默认是单向的。** 没有人反过来做——**"没看出它可能，所以它不可能"**——这句话谁也不会说，说出来当场就荒唐。**同一个"没看出"，只往一边倒。** **为什么只往那一边？因为"可能"不是判出来的，是随货附上的。** 一件事到我这儿的时候，**它已经带着那个标签了**：所与，不是所作。**而"不可能"要判——要有人指出它在哪一步过不去。所以"不可能"是一次动作，"可能"是一个默认。** **一个当场可做的检验：请举出一件你判为"未查"的事。** 举不出来。你能举出的，要么是你认为可能的，要么是你认为不可能的。**而"我不知道"这句话，实际用法是"我不知道它是哪一种"——它已经预…Read more
**没看出不可能，就被当做了可能。** **而这之所以能成立，是因为那本账上没有"未查"那一栏。** **若有三栏——不可能、可能、未查——那么绝大多数东西该待在第三栏里。** 我们查过的事屈指可数，**没查过的无穷无尽。** 按理，第三栏该是最挤的一栏。 **可它不存在。只有两栏。** **于是凡是没被判进"不可能"的，自动落进"可能"。** **这不是一次推论，是排水：没有别的地方可去。** **而这个默认是单向的。** 没有人反过来做——**"没看出它可能，所以它不可能"**——这句话谁也不会说，说出来当场就荒唐。**同一个"没看出"，只往一边倒。** **为什么只往那一边？因为"可能"不是判出来的，是随货附上的。** 一件事到我这儿的时候，**它已经带着那个标签了**：所与，不是所作。**而"不可能"要判——要有人指出它在哪一步过不去。所以"不可能"是一次动作，"可能"是一个默认。** **一个当场可做的检验：请举出一件你判为"未查"的事。** 举不出来。你能举出的，要么是你认为可能的，要么是你认为不可能的。**而"我不知道"这句话，实际用法是"我不知道它是哪一种"——它已经预设了只有两种。** 那不是第三栏，**那是在两栏之间摇摆。** **而这本账的两栏还不对称：**"不可能"那一栏窄得很，**里面只有那几样撞得最响的**——方的圆、同时在两地。**其余全在"可能"那一栏，而它们绝大多数从来没被碰过。** **开一栏"未查"，要付一样我们不愿付的东西：大部分条目要被移进去，而移进去就意味着不能拿它办事。**"我本来可以不那样"——移进"未查"，那份懊恼就没了着落；"这事还有希望"——移进"未查"，那份指望就悬空。**两栏之所以只有两栏，是因为办事只认两栏。** **所以本文不主张该开那一栏。** 只主张：**它现在没有，而账目正按"没有它"的方式在结。** ## Abstract **What is not seen to be impossible is taken to be possible.** **And this holds because the ledger has no column for "unexamined."** **With three columns—impossible, possible, unexamined—most things would belong in the third.** What we have examined is a handful; **what we have not is endless.** By rights the third column should be the crowded one. **It does not exist. There are two.** **So whatever is not assigned to "impossible" falls into "possible."** **Not an inference but drainage: there is nowhere else to go.** **And the default runs one way.** No one reasons in reverse—**"not seen to be possible, therefore impossible"**—nobody says it; said aloud it is absurd on the spot. **The same "not seen" tilts only one way.** **Why that way? Because "possible" is not adjudicated but comes attached.** When a thing arrives, **it arrives already wearing the label**: given, not made. **"Impossible" must be adjudicated—someone must point out where it fails. So "impossible" is an act and "possible" a default.** **A test anyone can run: name one thing you have filed as "unexamined."** You cannot. What you can name is either what you take to be possible or what you take to be impossible. **And "I don't know," in actual use, means "I don't know which of the two"—it already presupposes there are two.** That is not a third column; **it is oscillation between two.** **The two columns are also unequal:** "impossible" is very narrow, **holding only the loudest collisions**—square circles, being in two places at once. **All the rest sits in "possible," and nearly all of it has never been touched.** **Opening an "unexamined" column would cost something we are unwilling to pay: most entries would move there, and moving there means they can no longer do any work.** "I could have done otherwise"—move it, and the regret has nowhere to rest; "there is still hope in this"—move it, and the expectation hangs in air. **There are two columns because business recognizes two.** **So the paper does not urge that the column be opened.** It claims only: **there is none now, and the books are being closed as though there never needed to be.**

Science, Logic, and Mathematics Metaphysics and Epistemology Value Theory
12

你是怎么知道下一个的：论两栏都空，而只有一边在发货

反事实是按记账规则生成的：记一笔账，没有这笔账就自动呈现为一个反事实。所以那个集合从来没有写完，它还在长——我今天买一样东西、接一份工作、认识一个人，今晚就多出一个。而有人主张：这些反事实所设想的世界，在物理上都是可能的。这句话是在为一批还没写出来的条目预先背书。要那句全称，就请证明它们为什么全都可能——可这个集合尚未穷尽，你连要证的对象都还列不完。而他往往会退：不说所有的，只说”至少我正在思考的这一个”。这一退看似省事，其实更贵：全称还能靠一句原则搪塞，单称要的是一份具体的计算——哪一个涨落，经哪条通道，放大到什么程度，怎么落成那个特定结局。他退到了唯一一处能被明确要求交货的地方，然后照样不交。而更深的一层是：他其实没有在做判断，他在执行一条规则。这一点有一个当场可做的实验：我此刻就知道他会怎么答，虽然我和他都还不知道那是关于什么的反事实。明天他丢一样东西、错过一趟车、被谁辜负一次——那个反事实今晚还不存在，因为那笔账还没记；可等它一出现，他一定会说它在物理上可能。而若那是一个判断，我不可能预先知道：一个真正在看对象的人，得等对象出现了才说得出话。那条规则是：凡是我…Read more
反事实是按记账规则生成的：记一笔账，没有这笔账就自动呈现为一个反事实。所以那个集合从来没有写完，它还在长——我今天买一样东西、接一份工作、认识一个人，今晚就多出一个。而有人主张：这些反事实所设想的世界，在物理上都是可能的。这句话是在为一批还没写出来的条目预先背书。要那句全称，就请证明它们为什么全都可能——可这个集合尚未穷尽，你连要证的对象都还列不完。而他往往会退：不说所有的，只说”至少我正在思考的这一个”。这一退看似省事，其实更贵：全称还能靠一句原则搪塞，单称要的是一份具体的计算——哪一个涨落，经哪条通道，放大到什么程度，怎么落成那个特定结局。他退到了唯一一处能被明确要求交货的地方，然后照样不交。而更深的一层是：他其实没有在做判断，他在执行一条规则。这一点有一个当场可做的实验：我此刻就知道他会怎么答，虽然我和他都还不知道那是关于什么的反事实。明天他丢一样东西、错过一趟车、被谁辜负一次——那个反事实今晚还不存在，因为那笔账还没记；可等它一出现，他一定会说它在物理上可能。而若那是一个判断，我不可能预先知道：一个真正在看对象的人，得等对象出现了才说得出话。那条规则是：凡是我记了账的，其取消都在物理上可能。它从没被说出口过；一旦说出口就荒唐得很明显——我记不记账，跟物理容不容得下，怎么会有关系？而**“紫色的钞票”正是他自己的证据**：他不主张它可能，不是因为考察过，是因为它不在他账上，那条规则不发放；他主张”钥匙没丢”可能，也不是因为考察过。同一条规则，一个不发，一个发——发不发跟物理无关，跟账目有关。于是两个集合的重合成了一件待解释的事：一边的字段只有数额、归属、有无、时间、对手方，而这几个是照着发工钱、追责、继承切的——物理没有这几个字段；两套来路完全不同的东西，逐条对上，一条不漏，而且对未来还没生成的那些也预先对上。这不是默认，这是一个巧合，而巧合要有交代。最后须把本文自己的位置说老实：我也没办法断言所有反事实设想都在物理上不可能——因为我同样没有穷尽它们。而就算只针对他正在思考的那一个，我若要证明它在物理上不可能，同样存在困难。所以两栏都是空的：“可能”那一栏他填不上，“不可能”那一栏我也填不上。可不对称还在，而且全部在这里：他在主张，我在不主张。空栏对主张者是欠账，对不主张者不是。两栏都空，而只有一边一直在往外发货。于是问题的性质就变了。它不再是”谁能证出来”——双方都证不出来；剩下的只有一件事：他为什么在未能证明之前，就认为其在物理上可能？而这已经不是一个逻辑问题或物理问题，这是一个心理学话题。这也就把这一整条线的性质定了：它从来不是要在物理那边赢一场，它是要看清我们这边在做什么。 Abstract Counterfactuals are generated by a bookkeeping rule: enter a line, and the absence of that line arises automatically as a counterfactual. So the set was never finished; it is still growing—buy something today, take a job, meet someone, and by tonight there is one more. Yet some maintain that the worlds these counterfactuals envisage are all physically possible. That sentence endorses, in advance, entries not yet written. For the universal claim, prove why they are all possible—but the set is unfinished, and you cannot even finish listing what is to be proved. So he retreats: not all of them, only “at least this one I am now considering.” The retreat looks economical; it is dearer. A universal claim can be fended off with a principle; a singular one requires an actual computation—which fluctuation, through which channel, amplified how far, ending in that particular outcome. He has retreated to the one place where delivery can be demanded, and still does not deliver. The deeper point: he is not making a judgment; he is executing a rule. There is an experiment, and it can be run now: I already know how he will answer, although neither he nor I yet knows what the counterfactual will be about. Tomorrow he will lose something, miss a train, be let down by someone—that counterfactual does not exist tonight, because the entry has not been made; yet the moment it appears, he will say it is physically possible. And if it were a judgment, I could not know in advance: someone genuinely inspecting an object must wait for the object before he can speak. The rule is: whatever I have entered, the cancellation of it is physically possible. It has never been said aloud; said aloud it is plainly absurd—what could my bookkeeping have to do with what physics permits? And “purple banknotes” are his own evidence: he does not maintain their possibility, not because he investigated, but because they are not on his books, so the rule does not issue; and he maintains “the keys were not lost” is possible likewise without investigating. One rule, issuing in one case and not the other—and the issuing tracks the ledger, not the physics. Hence the coincidence of the two sets requires an account: one side’s fields are only amount, owner, presence, date and counterparty—cut to fit wages, liability, inheritance; physics has no such fields—and yet the two match entry for entry without exception, and in advance even for those not yet generated. This is no default; it is a coincidence, and a coincidence calls for an account. Finally the paper’s own position must be stated honestly: I cannot assert that all counterfactual scenarios are physically impossible—for I have not exhausted them either. And even for the single one he is now considering, proving it physically impossible presents the same difficulty. So both columns are blank: he cannot fill “possible,” and I cannot fill “impossible.” But the asymmetry remains, and it is the whole of the matter: he asserts, I do not. A blank column is a debt for the claimant; for the non-claimant it is not. Both columns blank—and only one side keeps shipping goods. The character of the question thus changes. It is no longer “who can prove it”—neither can; what remains is only this: why does he take it to be physically possible before it has been proved? And that is no longer a question of logic or physics. It is a question for psychology. Which also settles what this whole line of work has been: never an attempt to win something on the side of physics, but an attempt to see clearly what we are doing on ours.

Metaphysics and Epistemology Science, Logic, and Mathematics Value Theory
12

你是怎么知道下一个的：论”反事实在物理上都可能”是一条未说出口的规则

反事实是按记账规则生成的：记一笔账，没有这笔账就自动呈现为一个反事实。所以那个集合从来没有写完，它还在长。我今天买一样东西、接一份工作、认识一个人——今晚就多出一个新的反事实。而有人主张：这些反事实所设想的世界，在物理上都是可能的。这句话是在为一批还没写出来的条目预先背书。两难由此摆出：要这句全称，就请证明它们为什么全都可能——可这个集合尚未穷尽，你连要证的对象都还列不完；退到逐一，那就一个一个来——而最省的那一个（“钥匙没丢”）到今天也没有人算过。一个都没交，却要为无穷多个、且还在增加的条目一次性担保。而本文要指出的更深一层是：他其实没有在做判断，他在执行一条规则。这一点有一个当场可做的实验：我此刻就知道他会怎么答，虽然我和他都还不知道那是关于什么的反事实。明天他丢一样东西、错过一趟车、被谁辜负一次——那个反事实今晚还不存在，因为那笔账还没记；可等它一出现，他一定会说它在物理上可能。而”预先知道一个人对尚不存在之物的判断”，这件事本身就说明了那个判断的来处：它不来自那样东西，它来自他判断的规则。一个真正在看对象的人，得等对象出现了才说得出话。那条规则是：凡是我…Read more
反事实是按记账规则生成的：记一笔账，没有这笔账就自动呈现为一个反事实。所以那个集合从来没有写完，它还在长。我今天买一样东西、接一份工作、认识一个人——今晚就多出一个新的反事实。而有人主张：这些反事实所设想的世界，在物理上都是可能的。这句话是在为一批还没写出来的条目预先背书。两难由此摆出：要这句全称，就请证明它们为什么全都可能——可这个集合尚未穷尽，你连要证的对象都还列不完；退到逐一，那就一个一个来——而最省的那一个（“钥匙没丢”）到今天也没有人算过。一个都没交，却要为无穷多个、且还在增加的条目一次性担保。而本文要指出的更深一层是：他其实没有在做判断，他在执行一条规则。这一点有一个当场可做的实验：我此刻就知道他会怎么答，虽然我和他都还不知道那是关于什么的反事实。明天他丢一样东西、错过一趟车、被谁辜负一次——那个反事实今晚还不存在，因为那笔账还没记；可等它一出现，他一定会说它在物理上可能。而”预先知道一个人对尚不存在之物的判断”，这件事本身就说明了那个判断的来处：它不来自那样东西，它来自他判断的规则。一个真正在看对象的人，得等对象出现了才说得出话。那条规则是：凡是我记了账的，其取消都在物理上可能。它从来没被说出口过；而一旦说出口，它荒唐得很明显——我记不记账，跟物理容不容得下，怎么会有关系？ “紫色的钞票”正是他自己的证据。他不主张紫钞票在物理上可能——不是因为他考察过，是因为它不在他账上，那条规则不发放；而他主张”钥匙没丢”可能，也不是因为他考察过。同一条规则，一个不发，一个发——发不发跟物理无关，跟账目有关。由此，两个集合的重合成了一件待解释的事。一边按记账规则生成，字段只有数额、归属、有无、时间、对手方——而这几个字段是照着我们要办的事切的：发工钱、追责、继承。物理没有这几个字段。另一边是物理许可的范围。两套来路完全不同的东西，逐条对上，一条不漏，而且对未来还没生成的那些也预先对上。这不是默认，这是一个巧合，而巧合要有交代。所以要问他的那一句可以极短：你是怎么知道下一个的？ Abstract Counterfactuals are generated by a bookkeeping rule: enter a line, and the absence of that line arises automatically as a counterfactual. So the set was never finished; it is still growing. Buy something today, take a job, meet someone—and by tonight there is one more counterfactual. Yet some maintain that the worlds these counterfactuals envisage are all physically possible. That sentence endorses, in advance, entries not yet written. Hence a dilemma: for the universal claim, prove why they are all possible—but the set is unfinished, and you cannot even finish listing what is to be proved; retreat to case by case, and then let us take them one at a time—yet the cheapest of them (“the keys were not lost”) has never been computed by anyone. Not one delivered, and yet a blanket guarantee for infinitely many entries still accruing. The deeper point: he is not making a judgment; he is executing a rule. There is an experiment for this, and it can be run now: I already know how he will answer, although neither he nor I yet knows what the counterfactual will be about. Tomorrow he will lose something, miss a train, be let down by someone—that counterfactual does not exist tonight, because the entry has not been made; yet the moment it appears, he will say it is physically possible. And to know in advance a person’s judgment on something that does not yet exist is to have located that judgment’s source: it does not come from the object but from his rule. Someone genuinely inspecting an object must wait for the object before he can speak. The rule is: whatever I have entered, the cancellation of it is physically possible. It has never been said aloud; and said aloud it is plainly absurd—what could my bookkeeping have to do with what physics permits? “Purple banknotes” are his own evidence. He does not maintain that purple banknotes are physically possible—not because he investigated, but because they are not on his books, so the rule does not issue; and he maintains “the keys were not lost” is possible likewise without investigating. One rule, issuing in one case and not the other—and the issuing tracks the ledger, not the physics. The coincidence of the two sets thus becomes something requiring explanation. One is generated by a bookkeeping rule whose only fields are amount, owner, presence, date, counterparty—fields cut to fit our business: wages, liability, inheritance. Physics has no such fields. The other is the range of what physics permits. Two sets of wholly different provenance, matching entry for entry without exception, and matching in advance even those not yet generated. This is no default; it is a coincidence, and a coincidence calls for an account. So the question to put to him can be very short: how do you know about the next one?

Science, Logic, and Mathematics Metaphysics and Epistemology Value Theory
12

自动的另一面：论记一笔账，就同时生出一个反事实

我们时刻把观察到的现象整理成账目，以及账目的变化。而当我们记下一笔账时，没有这笔账，就是自动呈现的反事实。 “自动”这两个字是要点。记下”我的包”那一刻，“没有这只包的我”同时就成立了——我不需要另做一件事去生成它，也不是后来想起来才有的。它跟那一笔同时到场，是同一个动作的另一侧。由此，反事实绝对不是对宇宙任何局部向任何方向的任意改动。它是和账目配套的。这一条可以当场验，而且是本文里最实的一处：没有人说过”要是我的包昨天有三个提手”、“要是那笔钱是紫色的”、“要是我住的这栋楼多出十七级台阶”。这些都是对宇宙局部的改动，一样也没被用过。为什么？因为它们不对应任何一笔账。而所有真被用过的，每一个都恰好是某一笔账的取消或改值。再进一步：反事实的自由度，等于账目的字段数。账上有什么字段，反事实就只能往哪几个方向改——数额、归属、有无、时间、对手方。而账上没有的东西，你连改都改不动，因为那儿没有可改的格子。我们以为反事实是想象力的产物、应该无边无际；实际上它被表结构卡死了，窄得很。你可以虚构任何东西，可你没法为”包有三个提手”生出一丝情绪——因为没有一行被动过。而账、期望、反…Read more
我们时刻把观察到的现象整理成账目，以及账目的变化。而当我们记下一笔账时，没有这笔账，就是自动呈现的反事实。 “自动”这两个字是要点。记下”我的包”那一刻，“没有这只包的我”同时就成立了——我不需要另做一件事去生成它，也不是后来想起来才有的。它跟那一笔同时到场，是同一个动作的另一侧。由此，反事实绝对不是对宇宙任何局部向任何方向的任意改动。它是和账目配套的。这一条可以当场验，而且是本文里最实的一处：没有人说过”要是我的包昨天有三个提手”、“要是那笔钱是紫色的”、“要是我住的这栋楼多出十七级台阶”。这些都是对宇宙局部的改动，一样也没被用过。为什么？因为它们不对应任何一笔账。而所有真被用过的，每一个都恰好是某一笔账的取消或改值。再进一步：反事实的自由度，等于账目的字段数。账上有什么字段，反事实就只能往哪几个方向改——数额、归属、有无、时间、对手方。而账上没有的东西，你连改都改不动，因为那儿没有可改的格子。我们以为反事实是想象力的产物、应该无边无际；实际上它被表结构卡死了，窄得很。你可以虚构任何东西，可你没法为”包有三个提手”生出一丝情绪——因为没有一行被动过。而账、期望、反事实是同一个动作的三个面。记一笔账，就是投出一个期望；而”没有这笔账”，就是那个期望的反面同时成立。所以事前叫期望，事后叫反事实——同一样东西的两个时态。这也解释了”其余那一片”为什么如此详尽、又从不被检查：它不是事后补的，是事前就已经投出去的；那五分钟的路线、到家的时刻、门还锁着，我在丢钥匙之前就已经这么预期了。而期望也是从来不被审的。最后：对于账目的变化的抵抗，则对应最强烈的情感。这解释了静态说法解释不了的事——情绪的分布在时间上，不在水平上。一个长期贫穷的人和一个昨天刚亏一笔的人，处境可以相同，而后者的情绪强得多。情绪在变动那一刻，不在水平上。它也解释了适应：过些日子情绪淡了，不是处境改善了，是那一行已经被改写了，没有可抵抗的变动了。而日常语言早有这个词——“接受”。接受就是把那一行划掉。说一个人”还没接受”，说的正是他还在抵抗那次更新。还有一个不对称，正好是一次检验：账目往上变时没有抵抗。中彩票、被表扬、意外的好事——当场就更新，那一行抬上去就不下来。所以喜的半衰期短，失的半衰期长——不是人性偏悲观，是一边有抵抗，一边没有。（须按前文的分寸读：抵抗不是情绪的原因，抵抗就是情绪的另一面。） Abstract We are continually organizing observed phenomena into ledgers, and into changes in ledgers. And when we enter a line, the absence of that line is the counterfactual that arises automatically. “Automatically” is the point. The moment “my bag” is entered, “me without this bag” stands simultaneously—I need do nothing further to generate it, nor does it wait to be thought of later. It arrives with the entry; it is the other side of one act. Hence a counterfactual is emphatically not an arbitrary alteration of any part of the universe in any direction. It comes paired with a ledger. This can be checked at once, and is the most concrete claim here: No one has ever said “if only my bag had had three handles yesterday,” “if only that money had been purple,” “if only my building had seventeen more steps.” All are alterations of parts of the universe, and none has ever been used. Why? Because none corresponds to any entry. And every counterfactual actually used is precisely the cancellation or revaluation of some entry. Further: the degrees of freedom of counterfactuals equal the fields of the ledger. Whatever fields the books have, those are the only directions of alteration—amount, owner, presence, date, counterparty. What the books lack cannot even be altered, since there is no cell to alter. We suppose counterfactuals are products of imagination and therefore boundless; in fact they are pinned by the table’s structure, and are narrow. You may invent anything in fiction, yet you cannot raise a flicker of feeling over “the bag with three handles”—because no line was touched. Ledger, expectation and counterfactual are three faces of one act. To enter a line is to cast an expectation; and “without this line” is the reverse of that expectation, standing at the same moment. So before the fact it is called expectation, after the fact a counterfactual—one thing in two tenses. This also explains why “all the rest” is so detailed and never inspected: it was not supplied afterwards but cast in advance; the route, the arrival time, the door still locked—I expected all of it before the keys were lost. And expectations, too, are never audited. Finally: resistance to changes in the ledger corresponds to the most intense feeling. This explains what a static account cannot—the distribution of feeling in time rather than in level. A long-poor man and a man who lost a sum yesterday may be in identical circumstances, yet the latter feels far more. Feeling is at the moment of change, not at the level. It also explains adaptation: feeling fades in a few days, not because circumstances improved but because the line has been rewritten, and there is no change left to resist. Ordinary language already has the word—“acceptance.” To accept is to strike the line out. To say someone “has not accepted it yet” is to say he is still resisting that update. And an asymmetry that serves as a test: upward changes meet no resistance. A lottery win, praise, unexpected good fortune—the books update on the spot, and the raised line does not come down. Hence joy has a short half-life and loss a long one—not because human nature is gloomy, but because one side is resisted and the other is not. (To be read with the earlier calibration: resistance is not the cause of feeling; resistance is feeling’s other face.)

Science, Logic, and Mathematics Value Theory
8

改一个字段：论我们真正在用的反事实，内容就是账目

**我们日常真正在用的反事实，几乎都落在自己的账上。** 去数一个人一天里实际说出口的"要不是……""早知道……""本来可以……"——**看有几条落在他自己的账外。** 接近零。 **"假如恐龙没灭绝会怎样"——它不是一个无账的反事实，它是一个没被用的句式。** 我们一天说几十次"要不是"，一次也轮不到它；它是课堂上的、闲聊里的、举例子用的，**不是活着的时候用的东西**。**而它一旦真被用起来，账就出来了**：那还有没有人类？我还在不在？能不能去看？危不危险？——**接的每一句都在算一个利，而算利就是记账。** **这也解释了闲聊型反事实为什么淡而易散。** 恐龙、秦朝、唐朝，说完就完，睡得着；而"要不是那天我多嘴那一句"，能翻来覆去好几年。**差别不在题目大小，在有没有一行是我的。** **而构建反事实的动作，简单到只有几种改账：改一个数额**（那笔钱没被吞）、**改一个归属**（包没丢）、**改一笔进出**（他当初帮了我一把）、**改一个期初余额**（要是我生在有钱人家）。**没有一个需要重造一个世界。** **由此可以答一个一直悬着的问题：那座看起来完整的世界，我们是怎么轻易造…Read more
**我们日常真正在用的反事实，几乎都落在自己的账上。** 去数一个人一天里实际说出口的"要不是……""早知道……""本来可以……"——**看有几条落在他自己的账外。** 接近零。 **"假如恐龙没灭绝会怎样"——它不是一个无账的反事实，它是一个没被用的句式。** 我们一天说几十次"要不是"，一次也轮不到它；它是课堂上的、闲聊里的、举例子用的，**不是活着的时候用的东西**。**而它一旦真被用起来，账就出来了**：那还有没有人类？我还在不在？能不能去看？危不危险？——**接的每一句都在算一个利，而算利就是记账。** **这也解释了闲聊型反事实为什么淡而易散。** 恐龙、秦朝、唐朝，说完就完，睡得着；而"要不是那天我多嘴那一句"，能翻来覆去好几年。**差别不在题目大小，在有没有一行是我的。** **而构建反事实的动作，简单到只有几种改账：改一个数额**（那笔钱没被吞）、**改一个归属**（包没丢）、**改一笔进出**（他当初帮了我一把）、**改一个期初余额**（要是我生在有钱人家）。**没有一个需要重造一个世界。** **由此可以答一个一直悬着的问题：那座看起来完整的世界，我们是怎么轻易造出来的？——我们根本没造。我们只改了一个字段，而账上其余条目一行没动，它们本来就在册。** **而账目的适用范围，比我们承认的大得多：日常里我们煞有介事辩论的，说到底还是个账——不论是家庭琐事、单位人情世故，还是国家大事、世界形势。**"煞有介事"这四个字是要点：**我们不觉得自己在算账**，我们以为在争对错、争道理；**而争到最后停在哪儿，是账上那一行说了算。** 一个当场可做的检验：**任何一场辩论，一直追问"那又怎样"，最后停在的一定是一句带账的话**——谁该得而没得到，谁不该得而得了。**不会停在一句纯粹的事实上，因为纯事实撑不住"那又怎样"。** **而尺度越大，账越隐蔽，姿态越高。** 家里那笔谁都知道是在算账，吵起来还有点不好意思；**而世界形势那一场，账躲在"正义""秩序""历史"这些词后面，于是可以理直气壮地吵。姿态的高低跟账的隐蔽程度成正比——不是账不见了，是它穿得更体面了。** **最后一句最容易挨骂：情感往往对金钱、物质利益最敏感；而关于人际关系的情感，则是虚拟了金钱、物质往来的亏欠、恩情等。** 这跟通常的讲法正好反着：一般以为钱是抽象的、后起的，人情才是本源的。**可语言站在反面：没有一个人情的词不是从物质那儿借来的**——亏欠、还、值不值、代价、补偿、抵——**而反过来一个也找不到。借的方向是单向的，而借的方向就是虚拟的方向。** **这也解释了那份敏感：金钱那儿的账是硬的**——有数目、有日期、有凭据、有第三方可查、有对账日。**账越硬，读数越准，情绪就越即时。** 而人情那本是软账：欠多少没有单位，还没还清没有凭据——**所以它含糊、拖得久、翻来覆去。软账不是没有账，是没有对账日。** ## Abstract **The counterfactuals we actually use in daily life almost all fall on our own books.** Count the "if only," "had I known," "I could have" that someone actually utters in a day—**and see how many fall outside his own books.** Close to none. **"What if the dinosaurs had not died out"—this is not a ledger-free counterfactual but an unused sentence form.** We say "if only" dozens of times a day and it never comes up; it belongs to classrooms, idle talk, illustrations—**not to living.** **And the moment it is really used, a ledger appears**: would there be humans? would I be here? could one go and see? would it be dangerous?—**every continuation computes an interest, and computing interest is bookkeeping.** **This also explains why idle counterfactuals are thin and quick to disperse.** Dinosaurs, the Qin, the Tang: said and done, and one sleeps; whereas "if only I hadn't said that one thing" can run for years. **The difference is not the size of the topic but whether some line is mine.** **And constructing a counterfactual is a matter of a few simple ledger edits: change a figure** (the money was not embezzled), **change an owner** (the bag was not lost), **change one transaction** (he had helped me back then), **change an opening balance** (had I been born rich). **None requires rebuilding a world.** **Which answers a question long left open: how do we build that seemingly complete world so easily?—We do not build it. We change one field, and every other entry stands untouched; they were already on the books.** **And the reach of bookkeeping is far wider than we admit: what we debate with such solemnity in daily life is, in the end, a ledger—household trifles, workplace obligations, affairs of state, the state of the world.** The word "solemnity" is the point: **we do not take ourselves to be doing sums**; we take ourselves to be arguing right and wrong. **Yet where the argument comes to rest is decided by a line in the books.** A test anyone can run: **press "and so?" through any debate, and it will come to rest on a sentence carrying a ledger**—who was owed and not paid, who took what was not theirs. **It will not rest on a plain fact, because a plain fact cannot bear "and so?"** **And the larger the scale, the better hidden the ledger and the loftier the manner.** Everyone knows the household sum is a sum, and there is some embarrassment in it; **whereas in the affairs of the world the ledger hides behind "justice," "order," "history," and one may quarrel with a clear conscience. Loftiness of manner tracks the concealment of the ledger—the ledger has not gone, it is better dressed.** **The last claim is the one that will draw fire: feeling is most acute where money and material interest are concerned; and the feelings of personal relations are a simulation of material exchange—indebtedness, kindness owed, and the rest.** This runs against the usual account, on which money is abstract and late while human feeling is the original. **But the language stands on the other side: not one word of human feeling fails to be borrowed from the material**—owing, repaying, worth, cost, compensation, offsetting—**while not one runs the other way. The borrowing is one-directional, and the direction of borrowing is the direction of simulation.** **This also explains the acuteness: the ledger of money is hard**—figures, dates, receipts, third parties, a day of reckoning. **The harder the ledger, the sharper the reading, the more immediate the feeling.** The ledger of human feeling is soft: no unit for how much is owed, no receipt for settlement—**hence it is vague, drawn out, gone over and over. A soft ledger is not the absence of accounts but the absence of a reckoning day.**

Metaphysics and Epistemology Science, Logic, and Mathematics Value Theory
13

沟通情感就是沟通账目：论”哪一侧是实”这个问题不成立

我们沟通情感时，其实是在沟通账目。沟通账目时，也就是在沟通情感。两条来路各说一遍。其一，我们无法理解不带账目的情感——那是无法在人之间传递的。“我难受。”——难受什么？一答，那一行就在句子里了：他辜负了我，我该得的没得到，我欠了人家的。而最生的那一份呢？一声呻吟确实传得出去，可它被听见的那一刻，立刻被记到了一个人名下（他出事了）。连最生的那一份，出门时也穿上了账目。其二，我们无法理解不带情感的账目——那与任何人都无关。一串没有人在意的数字，跟蒙古草原上一颗石子的登记簿一样：没有人为它高兴或着急，它就只是墨，不是账。一本无人在意的账，连持有者都没有。而有一处现成的证据，摊在语言里没人捡：情感词汇本来就是账目词汇。亏欠、欠他一个人情、还、值不值、代价、补偿、赔罪、心里过不去。说感情的时候，我们用的全是记账的字。这一条不必论证，翻开自己昨天说过的话就看得见。一个最难的反例是音乐：听一段曲子觉得悲伤，账在哪儿？——而它恰好证成上面那条，只要看准说的是”理解”和”传递”，不是”有没有”。那份感受也许在，可它说不出来；一说，就得借账目：“像失去了什么”“像很久以前的一个下午”。…Read more
我们沟通情感时，其实是在沟通账目。沟通账目时，也就是在沟通情感。两条来路各说一遍。其一，我们无法理解不带账目的情感——那是无法在人之间传递的。“我难受。”——难受什么？一答，那一行就在句子里了：他辜负了我，我该得的没得到，我欠了人家的。而最生的那一份呢？一声呻吟确实传得出去，可它被听见的那一刻，立刻被记到了一个人名下（他出事了）。连最生的那一份，出门时也穿上了账目。其二，我们无法理解不带情感的账目——那与任何人都无关。一串没有人在意的数字，跟蒙古草原上一颗石子的登记簿一样：没有人为它高兴或着急，它就只是墨，不是账。一本无人在意的账，连持有者都没有。而有一处现成的证据，摊在语言里没人捡：情感词汇本来就是账目词汇。亏欠、欠他一个人情、还、值不值、代价、补偿、赔罪、心里过不去。说感情的时候，我们用的全是记账的字。这一条不必论证，翻开自己昨天说过的话就看得见。一个最难的反例是音乐：听一段曲子觉得悲伤，账在哪儿？——而它恰好证成上面那条，只要看准说的是”理解”和”传递”，不是”有没有”。那份感受也许在，可它说不出来；一说，就得借账目：“像失去了什么”“像很久以前的一个下午”。说不出来，正因为那一侧没有账目可用。至于哪一侧是实，本文的落点是：这个问题不成立。若认为账目是实，那么关于情感的语言则是对账目的隐喻。若认为情感是实，那么关于账目的语言则是为了把情感公共化、外部化、可表达化。两种读法都站得住，谁也驳不倒谁。而每一种都会把另一侧降格——可两边都能降对方，就说明谁也不是真的被降的那一个：降格不是一件事实，是选了一种读法之后的副产品。再往前一步：问”哪一侧是实”，已经预设了那儿有两样东西。而”没有一边在先”说的正是没有两样。所以这不是一道难题，是一道问不出来的问题——那条界线不承重，它不是那处所自带的，是一对被派下来的标签。 Abstract When we communicate feelings we are communicating ledgers; when we communicate ledgers we are communicating feelings. Two approaches. First, we cannot understand a feeling without a ledger—such a thing cannot pass between people. “I feel awful.”—About what? Answer, and the line is already in the sentence: he let me down, I did not get what was due, I am in someone’s debt. And the rawest case? A groan does carry, but the instant it is heard it is filed under a name (something has happened to him). Even the rawest sort puts on a ledger before going out. Second, we cannot understand a ledger without feeling—such a thing concerns no one. A column of figures nobody minds is like a register for a pebble on the Mongolian steppe: with no one glad or anxious over it, it is only ink, not accounts. A ledger nobody minds does not even have a holder. And there is evidence lying in the language, unpicked: the vocabulary of feeling is the vocabulary of accounts. Owing, indebted, repay, worth it, cost, compensation, making amends, unable to square it with oneself. When we speak of feelings we use the words of bookkeeping. No argument needed; look at what you said yesterday. The hardest counterexample is music: a passage sounds sorrowful—where is the ledger?—and it in fact confirms the above, provided one keeps to “understand” and “transmit” rather than “exist.” The feeling may be there, but it cannot be said; and to say it one must borrow a ledger: “like losing something,” “like an afternoon long ago.” It cannot be said precisely because on that side there is no ledger to use. As to which side is real, the paper’s conclusion is: the question does not arise. If one takes the ledger as real, then the language of feeling is a metaphor for ledgers. If one takes feeling as real, then the language of ledgers exists to make feeling public, external, expressible. Both readings hold, and neither refutes the other. Each demotes the other side—but that both can demote the other shows that neither is really the demoted one: demotion is not a fact but a by-product of having chosen a reading. One step further: to ask which side is real already presupposes two things. And “neither comes first” says precisely that there are not two. So this is not a hard problem but an unaskable one—that boundary bears no load; it is not native to the place but a pair of labels handed down.

Metaphysics and Epistemology Science, Logic, and Mathematics Value Theory
12

没有一边在先：论账目与情感互相解释

账目，就是情感的解释。或者说，情感和账目互相解释。这一句要按”互相”来读，因为它取消了一个很顺手、却不成立的说法。那个说法是：账目在先，情感是它的读数。账上那一行定得高，于是不甘；实际超过了那一行，于是感恩。仿佛账是一块表，情感是指针。不成立。因为我并不能先独立读出那本账，再拿它来解释不甘——我知道那一行定得高，正是因为我不甘。反过来也一样：我说得出这份不甘是冲着什么来的，靠的正是那一行。没有一边在先。所以这不是一个解释，是一次双面描述。同一样东西，从记账那一侧看，叫”我应得的那一份”；从感受那一侧看，叫”不甘”。中间没有因果，因为中间没有两样东西。而这一步会顺手废掉一条看起来很像证据的东西：“有账的反事实带情绪，没账的不带”——这不是一次验证，是同语反复。它们当然重合，因为它们是同一样东西的两面。拿它当证据，是用定义冒充观察。那还剩什么？剩下的比原来那条硬：两侧的可见度不一样。感受那一侧是私有的、无从核的。你说你不甘，我没法查。而记账那一侧是共有的、露在外面的。失物招领的登记簿、保险单、遗产法的条文、“我的”“你的”这两个词的用法、别人替你记的那些——全都…Read more
账目，就是情感的解释。或者说，情感和账目互相解释。这一句要按”互相”来读，因为它取消了一个很顺手、却不成立的说法。那个说法是：账目在先，情感是它的读数。账上那一行定得高，于是不甘；实际超过了那一行，于是感恩。仿佛账是一块表，情感是指针。不成立。因为我并不能先独立读出那本账，再拿它来解释不甘——我知道那一行定得高，正是因为我不甘。反过来也一样：我说得出这份不甘是冲着什么来的，靠的正是那一行。没有一边在先。所以这不是一个解释，是一次双面描述。同一样东西，从记账那一侧看，叫”我应得的那一份”；从感受那一侧看，叫”不甘”。中间没有因果，因为中间没有两样东西。而这一步会顺手废掉一条看起来很像证据的东西：“有账的反事实带情绪，没账的不带”——这不是一次验证，是同语反复。它们当然重合，因为它们是同一样东西的两面。拿它当证据，是用定义冒充观察。那还剩什么？剩下的比原来那条硬：两侧的可见度不一样。感受那一侧是私有的、无从核的。你说你不甘，我没法查。而记账那一侧是共有的、露在外面的。失物招领的登记簿、保险单、遗产法的条文、“我的”“你的”这两个词的用法、别人替你记的那些——全都在明处。于是有力气的不是”两者重合”，而是：那份感受的内容，是从看得见的那一侧领来的。不甘不甘的是一个数额；而那个数额的定法，写在共有的那一侧：跟着比较对象走，跟着邻居改，跟着”人家孩子上的什么学校”改。你查不了自己的心，可你查得了那本共有的账——而两者说的是同一件事。因此结案条件也要换。不该是”举出一个带情绪而指不出账目的反事实”（那验不了，因为两边同源）；该是：若有人指出一份情感，其内容在任何共有的记法里都找不到对应——本文就败。 Abstract The ledger is the explanation of the feeling. Or rather: feeling and ledger explain each other. Read the word “mutually,” for it cancels a convenient but untenable account. That account runs: the ledger comes first, and feeling is its reading. The line is set high, hence resentment; the actual exceeds the line, hence gratitude. As though the ledger were a dial and feeling the needle. It does not hold. For I cannot first read the ledger independently and then explain the resentment by it—I know the line is set high precisely because I resent. And conversely: what lets me say what this resentment is directed at is that very line. Neither side comes first. So this is not an explanation but a two-sided description. One and the same thing: seen from the bookkeeping side it is called “the share due to me”; seen from the feeling side it is called “resentment.” There is no causation between them, because there are not two things between which it could run. And this discards something that looked like evidence: “counterfactuals with a ledger carry feeling; those without do not”—that is not a test but a tautology. Of course they coincide: they are two faces of one thing. To offer it as evidence is to pass off a definition as an observation. What remains is harder than what is lost: the two sides differ in visibility. The feeling side is private and uncheckable. You say you resent; I cannot verify it. The bookkeeping side is public and exposed. Lost-property registers, insurance policies, probate statutes, the usage of the words “mine” and “yours,” the records others keep of you—all out in the open. So the force lies not in “they coincide” but in this: the content of the feeling is drawn from the visible side. Resentment resents a quantity; and how that quantity is set is written on the public side: it tracks the comparison, moves with the neighbours, moves with what school other people’s children attend. You cannot audit your heart, but you can audit the public books—and the two say the same thing. The closure condition must therefore change. Not “exhibit a counterfactual carrying feeling but pointing to no ledger” (unverifiable, since the two are one); rather: should anyone exhibit a feeling whose content finds no counterpart in any public way of keeping accounts, this paper fails.

Metaphysics and Epistemology Science, Logic, and Mathematics Value Theory
10

没被划掉的那一行：论反事实不是造出来的

**"关于"谁，就会被记到谁的账上。** **而构建反事实时，只需要简单地按照账上的内容来翻转就行。** **我的包记在我的账上。它现在在别人手上——保持账目不变，"我的包还在我的掌控下"就成了一个显然的反事实。** **请注意方向：不是反事实把现实翻转了，是现实对账目发生了偏离，而账目没动。** 那个反事实不是被造出来的，**它是没被划掉的那一行。** **这也就解释了一件一直没被解释的事：为什么反事实的其余部分从不被检查。** 说"要不是丢了钥匙，现在我就进屋了"，那五分钟的路线、到家的时刻、门还锁着——**这些不是被抄来的，是从来没有被删掉**：账上"我的钥匙"这一条本来就带着全套内容。**不需要构建，全都还在册。而没有人会去审一条常设条目。** **它还给出一个可验的推论：反事实的分布，跟账目的分布重合。** 我们不为蒙古草原上一颗石子造反事实——**它不在任何人的账上**；而月球也没有账，所以"月球本来可以绕火星"说不出口，"你本来可以走另一条路"却说得理直气壮。**不是句子的差别，是有没有那一行的差别。** **最难的一类是"要是我生在有钱人家"——那儿什么也没丢过。可丢的…Read more
**"关于"谁，就会被记到谁的账上。** **而构建反事实时，只需要简单地按照账上的内容来翻转就行。** **我的包记在我的账上。它现在在别人手上——保持账目不变，"我的包还在我的掌控下"就成了一个显然的反事实。** **请注意方向：不是反事实把现实翻转了，是现实对账目发生了偏离，而账目没动。** 那个反事实不是被造出来的，**它是没被划掉的那一行。** **这也就解释了一件一直没被解释的事：为什么反事实的其余部分从不被检查。** 说"要不是丢了钥匙，现在我就进屋了"，那五分钟的路线、到家的时刻、门还锁着——**这些不是被抄来的，是从来没有被删掉**：账上"我的钥匙"这一条本来就带着全套内容。**不需要构建，全都还在册。而没有人会去审一条常设条目。** **它还给出一个可验的推论：反事实的分布，跟账目的分布重合。** 我们不为蒙古草原上一颗石子造反事实——**它不在任何人的账上**；而月球也没有账，所以"月球本来可以绕火星"说不出口，"你本来可以走另一条路"却说得理直气壮。**不是句子的差别，是有没有那一行的差别。** **最难的一类是"要是我生在有钱人家"——那儿什么也没丢过。可丢的是账上的幸运。** 而这一例比丢包更干净，**因为它从来没兑现过，一天也没有**：账和事从未重合，可它照样丢得动、照样让人不甘。**所以反事实只需要账上那一行，不需要那一行兑现过。** **而"幸运"是最隐蔽的一条，因为它没有入账日期。**"我的包"有买来那天，"我的工作"有入职那天——**"我的份"是什么时候记上去的？** 没有那一天。**它不是账上的一笔，是开账时就在的底数。** **它的默认值定在哪儿，可以当场验：跟着比较对象走。** 村里人人如此的时候，那句话不出现；一进城，天天见着别人的日子，那一行立刻抬高——**而实际处境一分没变。若那份幸运是世界那边的一个量，它不该随邻居改。** **而跟着来的那批情绪，都是这一栏的读数**：不甘、委屈、嫉妒是**账上有而实际没到**；知足、感恩是**实际超过了账上那一行**。**都不是关于事情的，是关于差额的。** **最后一条既是证据也是边界：账的有无，跟情绪的有无重合。**"要是恐龙没灭绝"没有账，所以它不带情绪，只是好奇；"要是我生在有钱人家"有账，所以它带着不甘。**若有人举出一个带情绪、却指不出任何账目的反事实，本文就败。** ## Abstract **Whomever a thing is "about," it is filed on that person's books.** **And constructing a counterfactual requires only flipping what the books already say.** **My bag is on my books. It is now in someone else's hands—keep the books unchanged, and "my bag is still in my keeping" becomes an obvious counterfactual.** **Note the direction: it is not that the counterfactual inverted reality, but that reality departed from the books while the books stood still.** The counterfactual was not built; **it is the line that was never struck out.** **This explains something long unexplained: why the rest of a counterfactual is never inspected.** In "if I hadn't lost my keys I'd be inside by now," the route, the arrival time, the door still locked—**these were not copied over; they were never deleted**: the entry "my keys" comes with the whole set. **Nothing needs building; it is all still on the books. And no one audits a standing entry.** **It also yields a testable consequence: the distribution of counterfactuals coincides with the distribution of ledgers.** We build no counterfactual for a pebble on the Mongolian steppe—**it is on nobody's books**; and the moon has no books, so "the moon could have orbited Mars" cannot be said, while "you could have taken the other road" is said with perfect confidence. **Not a difference between sentences, but between having that line and not.** **The hardest case is "if I had been born rich"—nothing was ever lost there. But what is lost is the luck on the books.** And this case is cleaner than the bag, **because it never once fell due**: books and world never coincided, yet it can still be lost, and still stings. **So a counterfactual needs only the line on the books; the line need never have been honoured.** **And "luck" is the best-hidden entry, because it has no date of entry.** "My bag" has the day it was bought, "my job" the day I started—**when was "my share" written in?** There is no such day. **It is not an entry in the books but the baseline present when the books were opened.** **Where its default is set can be checked on the spot: it tracks the comparison.** In a village where everyone lives so, the sentence does not arise; move to the city, see others' lives daily, and the line rises at once—**while the actual circumstances have not changed by a hair. Were that luck a quantity on the world's side, it would not move with the neighbours.** **And the feelings that follow are readings of this column**: resentment, grievance and envy are **the books exceeding the actual**; contentment and gratitude are **the actual exceeding the books**. **None is about the events; each is about the shortfall.** **A last point, both evidence and boundary: the presence of a ledger coincides with the presence of feeling.** "If the dinosaurs hadn't died out" has no ledger, and carries no feeling—only curiosity; "if I had been born rich" has one, and carries resentment. **Should anyone exhibit a counterfactual that carries feeling yet points to no ledger whatever, this paper fails.**

Science, Logic, and Mathematics Value Theory Metaphysics and Epistemology
13

所与：论那份”关于”要等到它不出现时才认得出来

“这个事是关于我的。” 而”关于”这种感觉，是被给予的，是所与。这一点要先跟一个说法分开。若说那支箭是”我画上去的”，听起来像是我这边有一个动作、一个画的人。没有这样一个动作。那份”关于”到场时已经在那儿了，跟画面的清晰度一样，不由我加工。所与，不是所作。而由此得出本文真正要说的那一句：当我不再感觉到这事是关于我的，我才能认知到这里的”关于”并不在世界那边。不是先认知，然后那份感觉消失；是那份感觉先不出现，认知才有位置。理由很直接：只要那份”关于”还在，箭就在世界那边。你可以承认全部论证，可以把每一步都点头认下来——而下一件事到你面前时，它照样是关于你的。论证进不去那个位置，因为那个位置上不是一个判断。而当它不出现时，那儿不是”我看见了箭是假的”，是那儿没有箭了。事情照旧发生：话被说了，事被做了，包丢了。只是它们不再关于谁。而这时你才第一次知道，先前那样东西在哪儿——因为你先前从来没有看见过它，你只是从它那儿看出去。这也就说明了本文这一类文字的限度：论证不移除所与。读完全部这些篇，那份”关于”照样每天到场。能移除它的不是道理。而那份缺席是可遇的，不是可求的。 …Read more
“这个事是关于我的。” 而”关于”这种感觉，是被给予的，是所与。这一点要先跟一个说法分开。若说那支箭是”我画上去的”，听起来像是我这边有一个动作、一个画的人。没有这样一个动作。那份”关于”到场时已经在那儿了，跟画面的清晰度一样，不由我加工。所与，不是所作。而由此得出本文真正要说的那一句：当我不再感觉到这事是关于我的，我才能认知到这里的”关于”并不在世界那边。不是先认知，然后那份感觉消失；是那份感觉先不出现，认知才有位置。理由很直接：只要那份”关于”还在，箭就在世界那边。你可以承认全部论证，可以把每一步都点头认下来——而下一件事到你面前时，它照样是关于你的。论证进不去那个位置，因为那个位置上不是一个判断。而当它不出现时，那儿不是”我看见了箭是假的”，是那儿没有箭了。事情照旧发生：话被说了，事被做了，包丢了。只是它们不再关于谁。而这时你才第一次知道，先前那样东西在哪儿——因为你先前从来没有看见过它，你只是从它那儿看出去。这也就说明了本文这一类文字的限度：论证不移除所与。读完全部这些篇，那份”关于”照样每天到场。能移除它的不是道理。而那份缺席是可遇的，不是可求的。它通常出现在大病之后、极度疲惫时、久坐之后、丧亲之后——都不是想通了，都是那台机器歇了一下。本文不给方法，也不主张它能被安排；本文只说：它一旦出现过，那一栏就露出来了，而这些文字能做的，是在那之后帮你认出刚才那是什么。指是指，月是月。指可弃，月不可弃。 Abstract “This is about me.” And the feel of “about” is given—it is the given. This must first be distinguished from another phrasing. To say the arrow was “drawn by me” suggests an act on my side and someone drawing it. There is no such act. The “aboutness” is already there when it arrives, like the vividness of an image, not something I work up. Given, not made. Hence the paper’s real claim: only when I no longer feel that the thing is about me can I recognize that this “about” was never on the world’s side. It is not that recognition comes first and the feeling then departs; the feeling must first fail to arrive, and only then is there room for recognition. The reason is direct: so long as the aboutness is present, the arrow is on the world’s side. You may grant the whole argument, nod at every step—and the next thing that comes will still be about you. Argument cannot reach that position, because what occupies it is not a judgment. And when it fails to arrive, what is there is not “I saw the arrow was false” but that there is no arrow. Things still happen: words are said, deeds done, a bag lost. They are simply not about anyone. And only then do you first learn where that thing used to be—because you never saw it before; you only ever looked out from it. Which sets the limit of writings such as this: argument does not remove the given. Read all of these papers, and the aboutness will arrive tomorrow as usual. What removes it is not reasoning. And that absence is met, not sought. It tends to come after grave illness, in extreme fatigue, after long sitting, after a death—none of these is having thought it through; each is the machine resting for a moment. The paper gives no method and does not claim the absence can be arranged; it says only this: once it has occurred, that column stands exposed—and what these writings can do is help you afterwards to recognize what that was. The finger is the finger, the moon is the moon. The finger may be discarded, the moon may not.

Science, Logic, and Mathematics Value Theory Metaphysics and Epistemology
12

世界是我的：论从一只包到整个世界，是同一条账目

**新闻在说他，路人在议论他，车牌是发给他的信号，别人针对我，世界在对我说话。** **改写一遍：新闻是我的，路人是我的，车牌是我的，别人是我的，世界是我的。** **内容一个字没变。** 可前一组听起来像在描述世界，后一组一说出口就荒谬。**而它们是同一笔账：它们都被记在了我的账目里，就像我的包一样。** **反过来改写更有意思："包是我的" → "包是关于我的"。** 这句听着别扭——**而它别扭，只因为我们习惯了前一种说法。** 两种说法是同一个动作，**而我们只对其中一种起疑。** **于是"关于我的"和"我的"合成了一样东西：记在我账上。** **而这条账目是连续的，没有一处天然的断口：** 我的包、我的车、我的房子、我的工作、我的名声、我的学生、我的家人、我的城市、我的时代——**而"在我的世界里"这句话，我们说得毫不脸红。最大的那一行，日常语言里本来就有。** **所以问题不是"该在哪一行停下"。** 那个动作本身**没有提供任何停下的地方**：你停在哪儿，都已经在做同一件事了。哪几行看着正常、哪几行看着有病，是**习惯给的，不是账目给的**。 **而最后一行——"世界是…Read more
**新闻在说他，路人在议论他，车牌是发给他的信号，别人针对我，世界在对我说话。** **改写一遍：新闻是我的，路人是我的，车牌是我的，别人是我的，世界是我的。** **内容一个字没变。** 可前一组听起来像在描述世界，后一组一说出口就荒谬。**而它们是同一笔账：它们都被记在了我的账目里，就像我的包一样。** **反过来改写更有意思："包是我的" → "包是关于我的"。** 这句听着别扭——**而它别扭，只因为我们习惯了前一种说法。** 两种说法是同一个动作，**而我们只对其中一种起疑。** **于是"关于我的"和"我的"合成了一样东西：记在我账上。** **而这条账目是连续的，没有一处天然的断口：** 我的包、我的车、我的房子、我的工作、我的名声、我的学生、我的家人、我的城市、我的时代——**而"在我的世界里"这句话，我们说得毫不脸红。最大的那一行，日常语言里本来就有。** **所以问题不是"该在哪一行停下"。** 那个动作本身**没有提供任何停下的地方**：你停在哪儿，都已经在做同一件事了。哪几行看着正常、哪几行看着有病，是**习惯给的，不是账目给的**。 **而最后一行——"世界是我的"——比看上去更硬。** 它不是一个夸张的说法，也不是一次僭越。**它只是一句关于收件的实话：我所知道的一切，都是通过"到我这儿"这个方式到的。** 光到我眼里，声到我耳里，一切消息经由某条路径落到这一列上。**没有一样东西是从别处进来的——从别处进来的，我根本不知道它发生过。** **所以不是我把世界记进了账，是我这本账只收得到世界的一面。** **而这一行不是账上最后一行，是第一行。** 别的行——我的包、我的家人、我的城市——都写在它下面；**它先在，其余才写得下去。** 这也是它从不被察觉的原因：**它不是账上的一笔，它是这本账的封面。** ## Abstract **The news is speaking of him, strangers are discussing him, the licence plate was a signal for him, people are targeting me, the world is speaking to me.** **Rewrite them: the news is mine, the strangers are mine, the plate is mine, other people are mine, the world is mine.** **Not a word of content has changed.** Yet the first set sounds like a description of the world, while the second is absurd the moment it is uttered. **And they are one ledger: all of them have been entered in my books, just like my bag.** **The reverse rewriting is more telling: "the bag is mine" → "the bag is about me."** That grates—**and it grates only because we are used to the other phrasing.** The two phrasings are one act, **and we suspect only one of them.** **So "about me" and "mine" turn out to be the same thing: entered in my ledger.** **And this ledger is continuous, with no natural break:** my bag, my car, my house, my job, my reputation, my students, my family, my city, my era—**and we say "in my world" without a blush. The largest line of all is already in ordinary speech.** **So the question is not where one ought to stop.** The act itself **offers nowhere to stop**: wherever you halt, you are already doing the same thing. Which lines look normal and which look ill is **given by habit, not by the ledger.** **And the last line—"the world is mine"—is harder than it looks.** It is not hyperbole, nor a usurpation. **It is a plain remark about receipt: everything I know arrived by way of arriving at me.** Light reaches my eyes, sound my ears, every message falls by some path onto this one column. **Nothing enters from elsewhere—what enters from elsewhere, I never learn happened at all.** **So it is not that I entered the world into my ledger; it is that my ledger receives only one side of the world.** **And this line is not the last in the book but the first.** The other lines—my bag, my family, my city—are written beneath it; **it is there first, and only then can the rest be written.** Which is why it goes unnoticed: **it is not an entry in the ledger; it is the ledger's cover.**

Science, Logic, and Mathematics Metaphysics and Epistemology Value Theory
9

我的包：论一次常设的关系妄想

新闻在说他，路人在议论他，那块车牌是发给他的信号——我们把这叫做关系妄想。而”这只包是我的”，要求的是同一样东西。两句话都声称有一支箭，从那件事、那样东西出发，指向我。请把包上那支箭指出来。布料、拉链、金属扣、缝线——一处也没有写着”张三的”。箭是我画的。画在哪儿？画在账上。我的包是我的，这仅仅是因为我把它记到了我的账上。而最能看清这一点的时刻，是它丢了的时候。包到了另一个人手上，我碰不到它，它现在的每一次移动都跟别人发生关系——一切物理往来都断了，而”我的”一分未减。我照样说”我的包丢了”，三个月后想起来还是”我的包”。为什么？因为我还是把它记在我账上呢。那一行还在，我没划掉。平时账和事重合，看不出来；丢的那一刻，两者错开，那支箭就露在明处了。它从来都和我无关，只是有时和我的距离近，有时距离远而已。而”近”和”远”我又是怎么说的？距离近被我称为”来”，距离远被我称为”离开”——这两个概念都是以我自己为中心的。“它离开了我”里，它动，我不动，我是那个原点；可那段距离的变化不归任何一方。同一只包从五个人身边挪开，五个人都能说”它离开了我”——五个原点；若原点在世界那…Read more
新闻在说他，路人在议论他，那块车牌是发给他的信号——我们把这叫做关系妄想。而”这只包是我的”，要求的是同一样东西。两句话都声称有一支箭，从那件事、那样东西出发，指向我。请把包上那支箭指出来。布料、拉链、金属扣、缝线——一处也没有写着”张三的”。箭是我画的。画在哪儿？画在账上。我的包是我的，这仅仅是因为我把它记到了我的账上。而最能看清这一点的时刻，是它丢了的时候。包到了另一个人手上，我碰不到它，它现在的每一次移动都跟别人发生关系——一切物理往来都断了，而”我的”一分未减。我照样说”我的包丢了”，三个月后想起来还是”我的包”。为什么？因为我还是把它记在我账上呢。那一行还在，我没划掉。平时账和事重合，看不出来；丢的那一刻，两者错开，那支箭就露在明处了。它从来都和我无关，只是有时和我的距离近，有时距离远而已。而”近”和”远”我又是怎么说的？距离近被我称为”来”，距离远被我称为”离开”——这两个概念都是以我自己为中心的。“它离开了我”里，它动，我不动，我是那个原点；可那段距离的变化不归任何一方。同一只包从五个人身边挪开，五个人都能说”它离开了我”——五个原点；若原点在世界那边，它该只有一个。而这正是关系妄想的语法形态：世界在动，而我是那个被指向的中心。差别只在于，一处被诊断，另一处从没被当成一个说法。还有一处更难察觉：临床上那个人是把一件事读成关于他的，而”我的包”是把一样东西长期挂在自己名下——后者不是一次误读，是一个常设的原点。它不发作，因为它从不停止。实际上应该是包和我的聚和散。这里我俩都不是中心。而再往下一层，两边也都不是一个东西：“包”是一列，“我”也是一列，都是把一片发生按某个名字归在一起的归档；所以”包不曾属于我”说的是一列不曾拥有另一列。包也从不曾属于我，我也从不曾拥有包。只有那一片发生里，两段轨迹靠近过，又分开了。 Abstract The news is speaking of him, strangers are discussing him, that licence plate was a signal for him—we call this a delusion of reference. And “this bag is mine” demands the very same thing. Both sentences claim an arrow running from the event, or the object, to me. Point out the arrow on the bag. Fabric, zip, metal buckle, stitching—nowhere does it say “Zhang’s.” I drew the arrow. Drew it where? In my ledger. My bag is mine only because I entered it in my ledger. And the moment this shows clearest is when it is lost. The bag passes to someone else, I cannot touch it, its every movement now involves another person—every physical dealing is severed, and “mine” has not diminished by a hair. I still say “my bag was lost”; three months later it is still “my bag.” Why? Because I still have it entered in my ledger. The line is still there; I have not struck it out. Ordinarily ledger and world coincide and nothing shows; at the moment of loss the two come apart, and the arrow stands exposed. It was never connected to me at all; it was only sometimes near me and sometimes far. And how do I speak of near and far? Nearness I call “coming,” distance I call “leaving”—and both notions take me as the centre. In “it left me,” it moves, I do not, I am the origin; yet the change in that distance belongs to neither party. The same bag draws away from five people, and each can say “it left me”—five origins; were the origin on the world’s side, there would be one. And this is precisely the grammatical form of a delusion of reference: the world moves, and I am the one it points at. The only difference is that one case is diagnosed and the other was never taken for a claim at all. One thing is harder still to notice: the clinical case reads one event as being about him, whereas “my bag” keeps an object standing in my name. The latter is not a single misreading but a permanent origin. It does not flare up, because it never stops. In fact it should be a gathering and a parting of the bag and me. Here neither of us is the centre. And one layer further, neither side is a thing either: “the bag” is a column and “I” am a column, both filings that group a stretch of happening under a name; so “the bag was never mine” says that one column never owned another. The bag never belonged to me, and I never owned the bag. There is only, within that stretch of happening, two trajectories that drew near and then apart.

Science, Logic, and Mathematics Value Theory Metaphysics and Epistemology
11

没有一件事是"关于"你的：论最基础的那一份关系妄想

**"这件事是关于我的。"** 有时候这句话被叫做**关系妄想**：新闻在说他，路人在议论他，那块车牌是发给他的信号。我们把这叫做病。 **而本文要说的是：存在某些"关于我"的事情——这一份，是所有人都无法避免的、最基础的关系妄想。其实没有一件事是"关于"你的。** **"关于"这个词声称有一支箭**，从那件事指向我。**请把那支箭指出来。** 那天办公室里发生了什么？空气在振动，光在传播，一些人的肌肉在收缩，一张纸从一只手到了另一只手。**这些都实实在在。可这里面，没有一样携带一个"关于你"。** 没有一支箭从那件事出发，飞到你这儿。**箭是你添的。** **一个当场可做的检验：同一件事，对不同的人"关于"不同的东西。** 一场裁员：对被裁的人，它关于"我的能力"；对人事，它关于"这个季度的成本"；对他孩子，它关于"我们要搬家了"；对楼下卖饭的，它关于"下个月生意会淡"。**同一片发生，四支箭，指向四处。若"关于"是那片发生自带的，它该只关于一样东西。** **而这一份为什么无法避免？** 因为我们所有的记录都是按"我"归档的：我记住的、我在意的、我留下的，无一例外都挂在这一列上。*…Read more
**"这件事是关于我的。"** 有时候这句话被叫做**关系妄想**：新闻在说他，路人在议论他，那块车牌是发给他的信号。我们把这叫做病。 **而本文要说的是：存在某些"关于我"的事情——这一份，是所有人都无法避免的、最基础的关系妄想。其实没有一件事是"关于"你的。** **"关于"这个词声称有一支箭**，从那件事指向我。**请把那支箭指出来。** 那天办公室里发生了什么？空气在振动，光在传播，一些人的肌肉在收缩，一张纸从一只手到了另一只手。**这些都实实在在。可这里面，没有一样携带一个"关于你"。** 没有一支箭从那件事出发，飞到你这儿。**箭是你添的。** **一个当场可做的检验：同一件事，对不同的人"关于"不同的东西。** 一场裁员：对被裁的人，它关于"我的能力"；对人事，它关于"这个季度的成本"；对他孩子，它关于"我们要搬家了"；对楼下卖饭的，它关于"下个月生意会淡"。**同一片发生，四支箭，指向四处。若"关于"是那片发生自带的，它该只关于一样东西。** **而这一份为什么无法避免？** 因为我们所有的记录都是按"我"归档的：我记住的、我在意的、我留下的，无一例外都挂在这一列上。**归档的人看不见自己在归档**——于是那支箭看起来像是从事情那边射来的，而不是从我这边画出去的。 **分寸三条。** 一、**不否认那些发生**：他确实被裁了，工资确实停了，那些都是实实在在的一串。取消的只是那一支箭。二、**不谈谁病谁不病**：临床上那件事在这里只是引子，用来引入正常人身上的那一份；两边如何分辨，是另一个题目，本文不落子。三、**不是说"你不重要"**：重要与否是另一本账上的事，本文一个字也没碰。 ## Abstract **"This is about me."** Sometimes we call that a **delusion of reference**: the news is speaking of him, strangers are discussing him, that licence plate was a signal meant for him. We call it an illness. **This paper says: that there exist certain things "about me"—this is the most basic delusion of reference, and no one avoids it. In fact nothing is "about" you.** **The word "about" claims there is an arrow** running from the event to me. **Point out the arrow.** What happened in that office? Air vibrated, light travelled, some muscles contracted, a sheet of paper passed from one hand to another. **All real. And not one of these carries an "about you."** No arrow left the event and flew to you. **You supplied the arrow.** **A test anyone can run: one event is "about" different things for different people.** A round of layoffs: for the man dismissed it is about "my competence"; for HR, about "this quarter's costs"; for his child, about "we're moving"; for the lunch place downstairs, about "next month will be slow." **One stretch of happening, four arrows, four directions. Were "aboutness" native to the happening, it would be about one thing only.** **And why is this one unavoidable?** Because every record we keep is filed under "I": what I remember, what I mind, what I leave behind—all of it hangs on that column. **The filer cannot see himself filing**—so the arrow looks as though it were shot from the event's side, rather than drawn from mine. **Three calibrations.** First, **the happenings are not denied**: he really was dismissed, the salary really stopped, and all of that is a real series. Only the arrow is withdrawn. Second, **the paper does not adjudicate who is ill and who is not**: the clinical case serves here only as an entry point to the ordinary one; how the two are to be told apart is another topic, on which the paper makes no move. Third, **this is not "you do not matter"**: mattering is an entry in another ledger, which the paper does not touch.

Science, Logic, and Mathematics
11

连"本来可以"也是送来的：论想法不是我做出来的

我们习惯把心里的东西分成两堆：一堆是外面送进来的——看见的、听见的；另一堆是我自己做的——想法、判断、盘算。本文主张第二堆也是送进来的，那一堆是空的。想法到场的时候，已经是成品：你抓不住自己造它的那一步，你能看见的永远是已经成型的东西，"这是我想出来的"是事后加的一句话，加在一件已经完成的事上。最硬的一处检验是"本来可以怎样"那种画面：它看起来只能由我生产，因为世界里根本没有那一支，没有任何对应物可供接收；可它照样自己浮上来，而且常在我最不想要它的时候浮上来——夜里那句"当时要是没答应就好了"，不是我决定去想的。若连这个都不是我做的，"我在制造"就没有留下任何据点。同一类里还有一样：听人说话时到场的"他在压我"。把一段对话从头到尾用物理写全，也不会在任何一处遇到"他被压过去了"——不是漏写，是根本没有这件事可写；可它照样直接到场。于是桌子、脸、"本来可以"、"他在压我"排成同一列，全是送到我这里的成品。那么"我在想"这个感觉从哪儿来？从改得动来。凡是能动的东西，就不像别人给的，像我自己搭的，因为我们默认能动它的人就是做它的人。而改也不是我在改：回头看那次"我把它放下了"，实际发生的是又…Read more
我们习惯把心里的东西分成两堆：一堆是外面送进来的——看见的、听见的；另一堆是我自己做的——想法、判断、盘算。本文主张第二堆也是送进来的，那一堆是空的。想法到场的时候，已经是成品：你抓不住自己造它的那一步，你能看见的永远是已经成型的东西，"这是我想出来的"是事后加的一句话，加在一件已经完成的事上。最硬的一处检验是"本来可以怎样"那种画面：它看起来只能由我生产，因为世界里根本没有那一支，没有任何对应物可供接收；可它照样自己浮上来，而且常在我最不想要它的时候浮上来——夜里那句"当时要是没答应就好了"，不是我决定去想的。若连这个都不是我做的，"我在制造"就没有留下任何据点。同一类里还有一样：听人说话时到场的"他在压我"。把一段对话从头到尾用物理写全，也不会在任何一处遇到"他被压过去了"——不是漏写，是根本没有这件事可写；可它照样直接到场。于是桌子、脸、"本来可以"、"他在压我"排成同一列，全是送到我这里的成品。那么"我在想"这个感觉从哪儿来？从改得动来。凡是能动的东西，就不像别人给的，像我自己搭的，因为我们默认能动它的人就是做它的人。而改也不是我在改：回头看那次"我把它放下了"，实际发生的是又有一批东西被送来——一个新角度自己冒出来，一句解释起了作用，一次重看松开了它；我把这串前后相继的到场，讲成了"我改了它"。作者身份因此是事后追加的，而且追加了两遍：造它的时候，改它的时候。这样一来，"谁在想、谁不在想"这一档整个空了，两边都不在想；剩下的差别只有一条——有的松，有的死。这一条是实的：讲道理有时有用有时没用，正是因为道理只能落在松的那一层，落在死的那一层上就成了一个完全正确、却完全没用的说法。此处守一处分寸：松和死不是两类人，是程度之别；我这边也有死的地方，别人那边也未必样样死。最后，这些东西都送到一个"我"这儿，而这个"我"同样不是我做的，它也是一并送来的——于是最下面一层是：有一个我，且这个我处在某个位置上；判断可以改，这一层不能靠改判断来撤销，出口在哪里本文不论定。回看本文自己：说"一切都是送来的"很容易滑成一句无所不包因而不切开任何东西的话，本文要防的正是这个——取消的是作者身份，不是差别。而我说我这一层改得动，凭什么？证据只能是它确实被改过；没被试过的弹性，和真的弹性，看上去一模一样。 ## Abstract We habitually sort what is in the mind into two piles: one delivered from outside—what is seen, what is heard—and one made by me: thoughts, judgments, calculations. This paper argues that the second pile is also delivered, and that as a separate pile it is empty. A thought is finished on arrival: you cannot catch yourself in the act of making it; what you can observe is always something already formed, and "I thought this up" is a sentence added afterwards to a deed already done. The hardest test is the "it could have been otherwise" scene. It looks as though only I could produce it, since the world contains no such branch and offers nothing corresponding to be received; yet it too floats up on its own, and usually when I least want it—the line at midnight, "if only I had not agreed," is not something I decided to think. If even this is not made by me, "I produce" is left without a foothold. One more item belongs to the same class: the "he is putting me down" that arrives while someone is speaking. Write out the physics of a conversation from beginning to end, and nowhere will you meet "he was put down"—not an omission; there is simply no such thing there to write. Yet it arrives directly all the same. Table, face, could-have-been, being-put-down thus line up together, all finished goods delivered to me. Where, then, does the feeling of "I am thinking" come from? From revisability. Whatever can be moved does not look given but self-assembled, since we assume by default that whoever can move a thing is the one who made it. And revising is not done by me either: look back at the time "I let it go," and what actually happened was that a further batch was delivered—a new angle surfaced by itself, an explanation landed, a rereading loosened it; I narrate this succession of arrivals as "I changed it." Authorship is therefore added afterwards, and added twice: at the making, and at the revising. With that, the register of "who is reasoning and who is not" empties out—neither is—and one difference remains: some things are loose and some are stuck. That difference is real. Reasoning sometimes works and sometimes does not precisely because reasoning can land only on the loose layer; aimed at the stuck layer it becomes a claim entirely correct and entirely useless. One proportion must be kept: loose and stuck are not two kinds of people but a difference of degree; I too have places that are stuck, and not everything on another's side is. Finally, all of this is delivered to an "I," and that "I" is likewise not made by me; it arrives along with the rest. The bottom layer is thus: there is an I, and this I occupies a position. Judgments can be corrected; this layer cannot be undone by correcting judgments, and where the way out lies the paper does not determine. Turning back on itself: saying "everything is delivered" slides easily into a remark so all-encompassing that it cuts nothing, and this is what the paper must guard against—what is cancelled is authorship, not the difference. And on what ground do I say my own layer can be revised? The only evidence is that it has in fact been revised; untested elasticity and real elasticity look exactly alike.

Science, Logic, and Mathematics Metaphysics and Epistemology Value Theory
13

"他在压我"不是想出来的：论改不动的，才是直接给的

有人说了一句话，随后到场的常常有两样东西：他说的那件事，和"这话是冲着我来的、我被压了一头"。第二样通常被当成一个结论——先听懂内容，再掂量语气用词，最后推出一个关于我们两人的判断。本文主张它不是结论。那个关系是直接看见的：还没等我想，它已经在那儿了，跟听见声音的高低、看见一张脸的神色一样，是眼睛耳朵直接交上来的东西，不是拿到东西之后再做的第二步。由此有四层意思。其一，把一段对话从头到尾用物理描述写全，也不会在任何一处遇到"他被压过去了"——不是漏写，是根本没有这件事可写。这个说法干净、当场可验，而它一点用也没有：那个关系不是想出来的，所以说服他"这里没有道理可想"，也就动不了它。于是有本篇的枢纽——一个说法可以完全正确，而完全没用。其二，世界里没有它，我却直接看见它，这两件事不矛盾：前一句说的是世界里有没有这件事，后一句说的是我这边看见了什么，本来就不是同一件事。而且凡是这类东西都这样运作——我看见一张桌子，从来没有先看见一大堆基本粒子；我看见一张脸，从来没有先看见明暗再推断出是脸。其三，也是核心：他那里的关系是直接给的，而且他改不动；我这里的关系也是直接给的，只不过我改得动。正是这…Read more
有人说了一句话，随后到场的常常有两样东西：他说的那件事，和"这话是冲着我来的、我被压了一头"。第二样通常被当成一个结论——先听懂内容，再掂量语气用词，最后推出一个关于我们两人的判断。本文主张它不是结论。那个关系是直接看见的：还没等我想，它已经在那儿了，跟听见声音的高低、看见一张脸的神色一样，是眼睛耳朵直接交上来的东西，不是拿到东西之后再做的第二步。由此有四层意思。其一，把一段对话从头到尾用物理描述写全，也不会在任何一处遇到"他被压过去了"——不是漏写，是根本没有这件事可写。这个说法干净、当场可验，而它一点用也没有：那个关系不是想出来的，所以说服他"这里没有道理可想"，也就动不了它。于是有本篇的枢纽——一个说法可以完全正确，而完全没用。其二，世界里没有它，我却直接看见它，这两件事不矛盾：前一句说的是世界里有没有这件事，后一句说的是我这边看见了什么，本来就不是同一件事。而且凡是这类东西都这样运作——我看见一张桌子，从来没有先看见一大堆基本粒子；我看见一张脸，从来没有先看见明暗再推断出是脸。其三，也是核心：他那里的关系是直接给的，而且他改不动；我这里的关系也是直接给的，只不过我改得动。正是这份改得动，让它在我这里不像是给的，倒像是我自己搭起来的——因为我们默认，能动它的人就是做它的人。所以真正的分界线不是"谁在想、谁不在想"（两边都不在想），而是改不改得动。这样一来，谁更贴近实情就颠倒过来了：他把直接给的当直接给的，是对的；我把直接给的当自己搭的，是错的。而且我改得动这件事，本身也是给我的，不是我修来的本事。承认这一点，"我讲道理、他不讲道理"就不再是一句关于品德的话，它只是说：我这一层有个可以改的口子，他那一层没有。讲道理在这里不是德行，是硬件的弹性。这也顺带说明了为什么第一层那个说法没用——道理只能作用在改得动的那一层。此处须守一处分寸：这不是两类人，是程度之别；我这边也有改不动的地方，他那边也未必样样改不动。其四，这种关系要被看见，得有一个"我"在场当参照点：地位是谁的地位，被压的是谁。而这个"我"同样不是想出来的，它也是直接给的。于是难受的根不在把关系判断错了，而在更下面一层——有一个我，且这个我处在某个位置上；判断可以改，这层设定不能靠改判断来撤销，出口在哪里本文不论定。最后回看本文自己：把一个人的知觉整个说成一台机器，这在关系那一层上是最重的一记，若被他听到，只会作为关系被接收，一个字的内容都进不去。而我凭什么说我这一层改得动？证据只能是它确实被改过；没被试过的弹性，和真的弹性，看上去一模一样。 ## Abstract Someone speaks, and two things typically arrive: the matter he is speaking about, and "this is aimed at me; I have been put down." The second is usually taken for a conclusion—first the content is understood, then the tone and wording are weighed, and finally a judgment about the two of us is drawn. This paper argues it is no conclusion. The relation is seen directly: before I think at all, it is already there, like hearing whether a sound is high or low, or seeing the look on a face—something the eyes and ears hand over directly, not a second step taken afterwards. Four points follow. (1) Write out the physics of a conversation from beginning to end, and nowhere will you meet "he was put down"—not an omission; there is simply no such thing there to write. The claim is clean and checkable on the spot, and it is of no use whatever: that relation was not arrived at by reasoning, so convincing someone that there is no reasoning to be had cannot move it. Hence the pivot: a claim can be entirely correct and entirely useless. (2) That the world does not contain it while I see it directly is no contradiction: the first says what the world contains, the second says what I see, and these were never the same question. Everything of this kind works so—I see a table, never first a swarm of elementary particles; I see a face, never first light and shade from which a face is inferred. (3) The core: in him the relation is directly given and he cannot change it; in me it is equally directly given, only I can. It is precisely this changeability that makes it look, on my side, not given but self-assembled—since we assume by default that whoever can move a thing is the one who made it. So the real dividing line is not who is reasoning and who is not (neither is), but what can and cannot be revised. This reverses who is closer to the truth: he takes what is given as given, and is right; I take what is given as something I built, and am wrong. Moreover, that I can revise is itself given to me, not a capacity I earned. Granting this, "I am reasonable and he is not" ceases to be a remark about character; it says only that my layer has a port that can be edited and his does not. Reasonableness here is not a virtue but an elasticity of hardware. This also explains why the claim in (1) is useless: reasoning can act only on the layer that can be revised. One proportion must be kept: these are not two kinds of people but a difference of degree; I too have places that will not move, and not everything on his side is immovable. (4) For such a relation to be seen at all, an "I" must be present as the reference point: whose standing, who was put down. That "I" is likewise not inferred; it too is directly given. The root of the distress therefore lies not in a mistaken judgment about the relation but one layer below—that there is an I, and that this I occupies a position. Judgments can be corrected; this settlement cannot be undone by correcting judgments, and where the way out lies the paper does not determine. Finally, turning back on itself: to describe a person's perception as an entire machine is the heaviest possible move at the relational layer; were he to hear it, it would be received as relation, and not one word of its content would get in. And on what ground do I say my own layer can be revised? The only evidence is that it has in fact been revised; untested elasticity and real elasticity look exactly alike.

Metaphysics and Epistemology Value Theory Science, Logic, and Mathematics
17

这不是一样吗：论"本来可以"这句话为什么只对人说

**"你本来可以走另一条路的。"** 这句话说得理直气壮。那么： **"这列1号线地铁今天早上本来可以走2号线的。"** **"月球昨天晚上本来可以绕着火星公转的。"** **这不是一样吗？** **本文只提这一问，并把举证责任放回它该在的地方：不一样的话，请指出哪里不一样。** 预计会听到两句回应，本文都不接受，理由如下。 **其一，"月球那句太夸张了。"** ——**请指出夸张在哪里。** 是尺寸？**月球大，难道大小会影响本质吗？** 是距离，是质量，还是年代久远？**请把门槛说出来：多大以上、多远以上，就不能说"本来可以"了？** 说不出这个门槛，"夸张"就只是一句不适感，而不适感不是论据。**而本文不会因为这一句就换成一个温和些的例子——那是替对方把活儿干了。** **其二，"人当然不一样。"** ——这不是回答，这是把立场复述了一遍。**请指出人哪里不一样，并且指出那一处不是一列记账。** 这里的"一列记账"须先说明白：我们对人有一整套记录办法——身份证号、工号、病历号、学号、纳税人编号、朋友圈上那个名字——**每一样都是把观察到的行为或事件按一个名字归在一起，这就是一列…Read more
**"你本来可以走另一条路的。"** 这句话说得理直气壮。那么： **"这列1号线地铁今天早上本来可以走2号线的。"** **"月球昨天晚上本来可以绕着火星公转的。"** **这不是一样吗？** **本文只提这一问，并把举证责任放回它该在的地方：不一样的话，请指出哪里不一样。** 预计会听到两句回应，本文都不接受，理由如下。 **其一，"月球那句太夸张了。"** ——**请指出夸张在哪里。** 是尺寸？**月球大，难道大小会影响本质吗？** 是距离，是质量，还是年代久远？**请把门槛说出来：多大以上、多远以上，就不能说"本来可以"了？** 说不出这个门槛，"夸张"就只是一句不适感，而不适感不是论据。**而本文不会因为这一句就换成一个温和些的例子——那是替对方把活儿干了。** **其二，"人当然不一样。"** ——这不是回答，这是把立场复述了一遍。**请指出人哪里不一样，并且指出那一处不是一列记账。** 这里的"一列记账"须先说明白：我们对人有一整套记录办法——身份证号、工号、病历号、学号、纳税人编号、朋友圈上那个名字——**每一样都是把观察到的行为或事件按一个名字归在一起，这就是一列**。这些东西确实是人有而月球没有的：月球没有身份证，地铁没有工号。**可它们不在月球那边，也不在人那边，它们在我们这边**——是我们建的、我们维护的、我们为了办事而建的。所以若答"人不一样，因为人有名字、有档案、能被追责"，**那答的是我们的账本，不是人。** 他当然可以主张"被记账这件事本身就是那处差别"——**可那就等于承认：差别在我们这边，不在他那边。** 说"人会选择"——那正是要问的事，不是答案；说"人有意图"——那就把意图摆出来，指出它在哪儿；说"人自己走"——那个"自己"在哪儿？他的肌肉收缩、他的脑内状态，也都在那一整片发生里。 **而地铁这个例子不是用来退让的，是用来加码的。** 地铁有调度、有时刻表、有一屋子人的意图卷在里面——**它跟人最像**。可我们照样不对它说"本来可以"；出了事我们说的是"调度出错了"——**句子里立刻出现一个人**。所以他连"因为人有意图"这条路也走不通：**意图在地铁那儿一样不缺，缺的只是我们把这件事记在谁名下。** **本文不提供替代解释，也不打算证明三者相同。** 提出"本来可以"的是他，多说一样东西的是他；**在他指出那处差别之前，那一栏是空的，而我们一向把它当成已经填好的。** ## Abstract **"You could have taken the other road."** This is said with perfect confidence. Then: **"Metro Line 1 could have run along Line 2 this morning."** **"The moon could have orbited Mars last night."** **Is it not the same?** **The paper puts only this question, and returns the burden of proof to where it belongs: if it is not the same, point out where.** Two replies are expected, and neither is accepted. **First: "the moon case is absurdly exaggerated."** —**Point out where the exaggeration lies.** In size? **The moon is large; does size affect the essence?** In distance, in mass, in remoteness of date? **State the threshold: above what size, above what distance, may one no longer say "could have"?** Absent a threshold, "exaggerated" is only a feeling of discomfort, and discomfort is not an argument. **Nor will the paper substitute a milder example on hearing this—that would be doing the opponent's work for him.** **Second: "a person is of course different."** —This is not an answer but a restatement of the position. **Point out where a person differs, and show that the difference is not a column of bookkeeping.** "A column of bookkeeping" must first be made plain: we maintain a whole apparatus of records for persons—identity numbers, employee numbers, medical record numbers, student numbers, tax numbers, the name on a feed of posts—**each of which groups observed behaviours or events under one name; that is a column.** These are indeed things a person has and the moon does not: the moon has no identity card, the metro no employee number. **But they lie neither on the moon's side nor on the person's; they lie on ours**—built by us, maintained by us, built for our business. So to answer "a person is different because he has a name, a file, and can be held to account" is **to answer about our books, not about the person.** He may of course claim that being booked is itself the difference—**but that is to concede that the difference is on our side, not on his.** "A person chooses"—that is the very thing in question, not an answer. "A person has intentions"—then produce the intention and say where it is. "A person moves himself"—where is that "self"? His muscle contractions and brain states are likewise within that one stretch of happening. **And the metro is not a concession but a reinforcement.** The metro has dispatchers, timetables, a roomful of intentions bound up in it—**it is the closest thing to a person**. Yet we still do not say "could have" of it; when something goes wrong we say "the dispatcher erred"—**and a person appears in the sentence at once.** So the route "because persons have intentions" is closed to him too: **intentions are not lacking in the metro's case; what is lacking is only whose name we file the matter under.** **The paper offers no alternative explanation and does not undertake to prove the three alike.** It is he who asserts "could have," he who says one thing more; **until he points out the difference, that column is blank—and we have always treated it as filled.**

Science, Logic, and Mathematics Value Theory Metaphysics and Epistemology
9

论换一个尺度看，"第二条轨迹"就得报出它的来源

**尺寸的感知是主观的。** 人体通常看起来不大；可从蚂蚁的视角看，人就像一座山。**保持这个视角。** **时间的感知也是主观的。** 保持另一个视角：把人的移动感知为极其缓慢。 **在这两个视角下，人巨大而缓慢。** 而此时，人的移动似乎需要极大的动力——因巨大而显得笨重，**于是我们不得不考虑动力的问题**。于是那个问题自己冒出来了：**如果巨人不以一条轨迹移动，而以另一条轨迹移动，那么第二条轨迹所需要的、与第一条不同的那份初始动力，其来源为何？** 若答"来源在那颗巨大的大脑里"，问题只是往里挪了一格：**那么在那颗巨大的大脑之中，此来源的来源又是什么？** 巨大的大脑里也只是一片巨大而缓慢的东西在动——**换个尺度看它，它跟外面那座山没有分别。** **而要害在于：当我们不把人及其大脑感知为巨大、缓慢、笨重时，我们会直接想出第二条轨迹，而会忽略其来源、来源的来源。** 因为**我们在习以为常的日常人体尺度下，已经习惯了这种忽略；我们在其他尺度下则没有这种习惯。** 这一条可以当场检验。换任何一个我们不熟悉的尺度，那份忽略就不见了：**一列火车换了轨，没有人说"它本来可以走另…Read more
**尺寸的感知是主观的。** 人体通常看起来不大；可从蚂蚁的视角看，人就像一座山。**保持这个视角。** **时间的感知也是主观的。** 保持另一个视角：把人的移动感知为极其缓慢。 **在这两个视角下，人巨大而缓慢。** 而此时，人的移动似乎需要极大的动力——因巨大而显得笨重，**于是我们不得不考虑动力的问题**。于是那个问题自己冒出来了：**如果巨人不以一条轨迹移动，而以另一条轨迹移动，那么第二条轨迹所需要的、与第一条不同的那份初始动力，其来源为何？** 若答"来源在那颗巨大的大脑里"，问题只是往里挪了一格：**那么在那颗巨大的大脑之中，此来源的来源又是什么？** 巨大的大脑里也只是一片巨大而缓慢的东西在动——**换个尺度看它，它跟外面那座山没有分别。** **而要害在于：当我们不把人及其大脑感知为巨大、缓慢、笨重时，我们会直接想出第二条轨迹，而会忽略其来源、来源的来源。** 因为**我们在习以为常的日常人体尺度下，已经习惯了这种忽略；我们在其他尺度下则没有这种习惯。** 这一条可以当场检验。换任何一个我们不熟悉的尺度，那份忽略就不见了：**一列火车换了轨，没有人说"它本来可以走另一条"，人们会问是谁扳的道岔；一座山峰移了位，人们问板块；一颗卫星改了轨道，人们问点火。** 这些全是宏观、缓慢、笨重的东西，**于是"来源为何"是自动会被问的**。而人体同样是宏观的——**唯一的差别是我们对它太熟了，而"熟悉"不在对象那边。** 这里还须点破一处，免得本文被当成靠夸张吓唬人：**严格讲，问不问跟规模无关。** 一小时挪一米的巨人和一秒挪一米的人，需要交代的东西同样多；**只是前者我们才想得起来要问。换尺度并没有增加什么，它只是把那份习惯换掉。** **而"来源的来源"这一串，不是无穷追溯的诡辩。** 本文不主张"每件事都必须有原因"。本文只说：**主张有第二条轨迹的人，得为那份不同指出一个来源；而答"来自大脑"只是把地址往里挪了一格——因为大脑里同样是一团要被推动的东西。盒子是不透明的，不是空的。** ## Abstract **The perception of size is subjective.** A human body usually looks not large; but from an ant's vantage, a person is like a mountain. **Hold that vantage.** **The perception of time is subjective too.** Hold a second vantage: perceive a person's movement as extremely slow. **Under these two vantages, a person is vast and slow.** And now the movement seems to require enormous motive power—vastness makes it look ponderous, **and so we cannot help asking about motive power**. Whereupon the question arises of itself: **if the giant moves along a second trajectory rather than the first, what is the source of the initial motive power required by that second trajectory, differing from the first?** Answer "the source lies in that vast brain," and the question has merely moved one step inward: **then within that vast brain, what is the source of this source?** Inside the vast brain there is likewise only something vast and slow in motion—**viewed at that scale, no different from the mountain outside.** **The crux: when we do not perceive a person and his brain as vast, slow and ponderous, we produce a second trajectory directly and omit its source, and the source of its source.** Because **at the familiar human scale we are habituated to that omission; at other scales we have no such habit.** This can be tested on the spot. Shift to any scale we are not used to and the omission vanishes: **a train switches tracks and no one says "it could have taken the other"—they ask who threw the points; a peak shifts and they ask about tectonics; a satellite changes orbit and they ask about the burn.** All macroscopic, slow, ponderous—**and "what is the source" gets asked automatically.** A human body is macroscopic too. **The only difference is our familiarity with it, and familiarity is not on the object's side.** One point must be conceded outright, lest the paper look like intimidation by exaggeration: **strictly, whether we ask has nothing to do with size.** A giant moving one metre per hour and a man moving one metre per second require exactly as much accounting; **it is only with the former that we remember to ask. Shifting scale adds nothing; it merely replaces the habit.** **And the chain "source of the source" is not a regress trick.** The paper does not hold that everything must have a cause. It says only: **whoever asserts a second trajectory owes a source for the difference; and answering "the brain" merely moves the address one step inward, since inside the brain there is likewise something that must be moved. The box is opaque, not empty.**

Metaphysics and Epistemology Science, Logic, and Mathematics Value Theory
12

折纸书：论发生本身没有可供理解的意义，而立体是我们打开出来的

**就像一种折纸书，打开之后书页就立了起来，成为立体的场景。** **我们的大脑会自动地把平铺直叙的发生本身——就像一个平面——想象成立体的东西。** 而这个立体的东西不再平铺直叙，它**充满了冲突、不同的路径**、原本可以怎样、险些如何、幸亏没有。 **而大脑似乎只能生活在这个立体的东西中，因为平铺直叙的那个本身，同时也没有任何可供理解的意义。** 这一句要按字面读：**不是我们看不出它的意义，是那里没有意义可看。** 恰如我们看一张风景照片时，只能把风景看作三维的——**因为二维照片上的像素本身没有任何可供理解的意义。** 一个像素是暗一点还是亮一点，它自己不讲任何事；意义是在被立起来之后才有的。 **现实本身、发生本身，就是二维照片上无意义的像素。** **所以立起来不是一个错误，是理解的唯一形式。** 我们不是贪玩才折，是不折就什么也读不到。这一点与"那些冲突和路径不在纸上"并不冲突：**我们只能住在立体里，而立体不在纸上——两句都对。** **而现实与折纸书有一处关键的不同，也是本文的落点：折纸书的立体场景只有一种，是确定的**——折痕是印好的，谁打开都是那座城堡。**而同一…Read more
**就像一种折纸书，打开之后书页就立了起来，成为立体的场景。** **我们的大脑会自动地把平铺直叙的发生本身——就像一个平面——想象成立体的东西。** 而这个立体的东西不再平铺直叙，它**充满了冲突、不同的路径**、原本可以怎样、险些如何、幸亏没有。 **而大脑似乎只能生活在这个立体的东西中，因为平铺直叙的那个本身，同时也没有任何可供理解的意义。** 这一句要按字面读：**不是我们看不出它的意义，是那里没有意义可看。** 恰如我们看一张风景照片时，只能把风景看作三维的——**因为二维照片上的像素本身没有任何可供理解的意义。** 一个像素是暗一点还是亮一点，它自己不讲任何事；意义是在被立起来之后才有的。 **现实本身、发生本身，就是二维照片上无意义的像素。** **所以立起来不是一个错误，是理解的唯一形式。** 我们不是贪玩才折，是不折就什么也读不到。这一点与"那些冲突和路径不在纸上"并不冲突：**我们只能住在立体里，而立体不在纸上——两句都对。** **而现实与折纸书有一处关键的不同，也是本文的落点：折纸书的立体场景只有一种，是确定的**——折痕是印好的，谁打开都是那座城堡。**而同一段平铺直叙的发生，不同的人、甚至同一个人在不同时刻，却可以打开成不同的立体场景。** **这一处不同自带一个检验，谁都能当场验：若那些冲突与路径是纸上本有的，为什么两个人打开同一段，立起来的东西不一样？** 同一件往事，一个人打开成"我当初太软弱"，另一个打开成"那是唯一可行的做法"；同一个人，二十岁打开是一座城堡，五十岁打开是一片废墟。**纸没有变。变的是折法。折痕在读者手里，不在纸上。** **还有一处，这个比喻自带而少被点出：折纸书的立体是有背面的。** 从背后看，那些楼房只是几片纸板，没有厚度。**立体感靠一个视角撑着，换个方位就塌。** 这与"不同的人折出不同的场景"是同一件事的两面：**不是纸上有立体，是看的方位造出立体。** 最后一层，也是最难受的一层：**那张没有被折起来的纸，恰恰是唯一不能被认出来的东西。** 因为"认出它是什么"这个动作本身，就是一次打开。**你一去看它，它就立起来了。** ## Abstract **It is like a pop-up book: open it and the page rises into a three-dimensional scene.** **Our brains automatically take flat, plain happening—like a plane—and imagine it as something three-dimensional.** And that raised thing is no longer flat and plain; it is **full of conflicts, of divergent paths**, of how things might have gone, of near misses and lucky escapes. **And the brain seems able to live only in that raised version, because the flat, plain thing itself has no meaning available to be understood.** Read that literally: **it is not that we fail to see its meaning; there is no meaning there to see.** Just as, looking at a landscape photograph, we can only see the landscape as three-dimensional—**because the pixels of a two-dimensional photograph have in themselves no meaning available to be understood.** Whether one pixel is darker or lighter tells no story; meaning arrives only once the thing is raised. **Reality itself, happening itself, is the meaningless pixels of a two-dimensional photograph.** **So raising it is not an error but the only form understanding takes.** We do not fold out of playfulness; unfolded, there is nothing to read. This does not conflict with "the conflicts and paths are not on the paper": **we can live only in the raised version, and the raised version is not on the paper—both are true.** **And reality differs from a pop-up book in one crucial respect, which is this paper's point: a pop-up book has exactly one scene, fixed**—the creases are printed, and whoever opens it gets that castle. **Whereas the same stretch of flat, plain happening can be opened by different people, or by the same person at different times, into different three-dimensional scenes.** **This difference carries its own test, and anyone can run it: if those conflicts and paths were on the paper, why do two people opening the same stretch raise different things?** One person opens a memory into "I was too weak then," another into "it was the only workable course"; the same person opens a castle at twenty and a ruin at fifty. **The paper has not changed. The folding has. The creases are in the reader's hands, not on the page.** **One more thing the metaphor carries but is rarely noted: a pop-up scene has a back side.** From behind, those buildings are a few pieces of card with no thickness. **The solidity rests on one vantage; move round and it collapses.** This is the other face of "different people fold different scenes": **it is not that the paper holds a solid, but that the vantage makes one.** The last and hardest layer: **the unfolded page is precisely the one thing that cannot be recognized.** For the act of "recognizing what it is" is itself an opening. **Look at it, and it stands up.**

Metaphysics and Epistemology Science, Logic, and Mathematics Value Theory
11

人并不比足球多任何东西：论人存在，当且仅当有那么一列

人存在，当且仅当，我们以人名为表格里的一列，来分割、记录、组织所观察到的行为或事件。最干净的例子是朋友圈：那儿实际有的，是若干条记录，加一列名字。你从没见过他，甚至不知道那些条目是不是同一个人发的。而”这个人存在”在这里的全部内容，就是那一列把这些记录归在了一起。 “当且仅当”有两头。“当”那一头容易接受：归好了档，人就有了。“仅当”那一头才是真正的内容：撤掉那一列，什么也不剩。最自然的反驳是”可身体在那儿啊”——在那儿的是一片发生，而”人”是把那片发生圈起来的那个圈；圈是我们画的，看边界就知道：法律切到死亡（遗产法还往后拖一阵），医学切到脑死亡，朋友圈切到注销；肺里的空气、肠里的细菌、体外的起搏器算不算他，每一处都由那张表要办的事来定。若”人”是世界那边摆着的一样东西，这些边界该是同一道；它们不是。而表格出错时，我们校对的办法永远是拿一张表去校另一张表——身份证、身体、别人的证词——从来没有诉诸过一个表格之外的”人本身”，因为没有那样一样东西可以去诉诸。反过来做一遍，结论更清楚。如果我们以同样的热情去追踪一颗足球的”行为”——开发一款 app，让很多人帮它记录”生活”——它…Read more
人存在，当且仅当，我们以人名为表格里的一列，来分割、记录、组织所观察到的行为或事件。最干净的例子是朋友圈：那儿实际有的，是若干条记录，加一列名字。你从没见过他，甚至不知道那些条目是不是同一个人发的。而”这个人存在”在这里的全部内容，就是那一列把这些记录归在了一起。 “当且仅当”有两头。“当”那一头容易接受：归好了档，人就有了。“仅当”那一头才是真正的内容：撤掉那一列，什么也不剩。最自然的反驳是”可身体在那儿啊”——在那儿的是一片发生，而”人”是把那片发生圈起来的那个圈；圈是我们画的，看边界就知道：法律切到死亡（遗产法还往后拖一阵），医学切到脑死亡，朋友圈切到注销；肺里的空气、肠里的细菌、体外的起搏器算不算他，每一处都由那张表要办的事来定。若”人”是世界那边摆着的一样东西，这些边界该是同一道；它们不是。而表格出错时，我们校对的办法永远是拿一张表去校另一张表——身份证、身体、别人的证词——从来没有诉诸过一个表格之外的”人本身”，因为没有那样一样东西可以去诉诸。反过来做一遍，结论更清楚。如果我们以同样的热情去追踪一颗足球的”行为”——开发一款 app，让很多人帮它记录”生活”——它需要的只有四样：一个持续的标识，若干条记录，一个次序，和愿意往里填的人。四样齐了，它就像人一样存在了，而且会长出一段可以让人高兴或惋惜的一生。它还会从”果”变成”因”：“是这颗球改变了那场比赛”这句话，一旦它有了履历，就不再刺耳——因为**“因”从来就是我们记账时的落点**：事记在谁名下，谁就是因。不是球变了，是表变了。而这不是空想，我们已经零星地这么做了：名琴造就了演奏家，泰坦尼克号撞上了冰山（我们从不说反过来），飓风有名字，船被称作”她”。凡是我们给了名字、给了持续记录的东西，都开始长出生平，也开始被当成因。至于我们为什么不给足球建这张表——不是因为足球不配，是因为我们没有要跟足球办的事。于是：人并不比足球多任何东西。这句话可以当场核：请指出人多出来的那一样，并且指出它不是一列记录。说”人有内在世界”——你手上是他的话、他的表情、他事后的记述，都是记录。说”人自己记自己”——自记只是又一列，而一个彻底失忆的人我们照样说他存在。说”人会疼”——你手上有的是疼的表现：一串行为、一串陈述、一串读数；你从来没有拿到过表现之外的那一样东西。说”人能被判刑”——那是另一张表的资格，而资格是发的，不是找到的。两处分寸要看住。一、这句话的方向不是”人被降到了球那么低”，而是**“我们从来就没在球之上给人找到过任何东西”**——它说的是我们的账，不是人的贵贱。二、“不多任何东西”不等于”没有差别”：差别多得很，可一样也不在对象那边，全在我们的表这边。人比足球多的，全部是我们给的；而我们给的东西，不算人自带的。 Abstract A person exists if and only if we use the personal name as a column in a table, to partition, record and organize observed behaviours or events. The cleanest example is a feed of posts: what is actually there is a number of records plus one column of names. You have never met him, and do not even know whether one person posts them. “This person exists,” here, consists entirely in that column grouping these records together. The biconditional has two halves. The “if” is easy: file it, and the person is there. The “only if” carries the real content: remove the column and nothing remains. The natural objection—“but the body is there”—meets this: what is there is a stretch of happening, while “the person” is the ring drawn around it; and the ring is ours, as the boundaries show. Law runs to death (and probate a while beyond), medicine to brain death, the app to deletion; whether the air in his lungs, the bacteria in his gut, the pacemaker outside his chest count as him is settled each time by the business that table serves. Were “the person” a thing on the world’s side, these boundaries would be one boundary; they are not. And when tables err, we correct them always by checking one table against another—ID card, body, someone’s testimony—never by appealing to a “person himself” outside all tables, because there is no such thing to appeal to. Run it in reverse and the conclusion is plainer. If we tracked a football’s “behaviour” with the same enthusiasm—an app, many people recording its “life”—it would need only four things: a persistent label, a set of records, an order, and people willing to fill it in. With those, it exists as a person does, and grows a life one can be glad or sorry about. It also turns from effect into cause: “it was this ball that changed the match” ceases to grate once it has a biography—because “cause” has always been where we file things: whoever the entry is filed under is the cause. The ball did not change; the table did. Nor is this fanciful; we already do it patchily: a famous violin made a soloist; the Titanic struck an iceberg (never the reverse); hurricanes have names; ships are “she.” Whatever we give a name and a running record begins to grow a biography and to be treated as a cause. And we do not build the ball its table not because the ball is unworthy, but because we have no business to transact with the ball. Hence: a person has nothing more than a football has. This can be checked on the spot: point out what is extra in a person, and show that it is not a column of records. “A person has an inner world”—what you hold is his speech, his expressions, his later accounts: records. “A person records himself”—self-recording is one more column, and a wholly amnesic person is still said to exist. “A person feels pain”—what you hold is the manifestation of pain: behaviour, statements, readings; you have never obtained the thing behind the manifestation. “A person can be sentenced”—that is a standing on another table, and standings are issued, not found. Two calibrations. First, the direction is not “the person is lowered to the ball” but “we have never found anything in a person over and above a ball”—a statement about our books, not about human worth. Second, “nothing more” is not “no difference”: differences abound, but not one of them lies on the object’s side; they all lie on the side of our tables. What a person has over a football is entirely what we gave him—and what we gave does not count as his own.

Metaphysics and Epistemology Value Theory Science, Logic, and Mathematics
13

这一次不是那一次：论意义的错置，与否定如何重装它所否定的东西

本文处理一条压得很紧的链子：我们感觉意义在事件本身，但意义其实指的是事件的后续；刚刚发生的事件尚无后续，因而尚无意义；于是我们转而预测它的后续；而实际被预测的，是被归为”同类”的诸事件的后续。文章先指出这条链子暴露了什么：意义被感到的地址是错的。它被感到在事件身上，实际住在后续那儿。这不是把内容判断错了，而是把位置弄错了；而位置错误从内部无法察觉，因为人只能在自己相信的那个地址上核对内容。其次给出这一错觉何以极其稳固的机制：后续真的到来时，人不会把它记成”借来的那笔兑现了”，只会记成”果然如此”——兑现被体验成印证。于是那笔借款从头到尾没有一刻以借款的样子出现过。而后续没有到来时，人记的是”我当初理解得不够深”——仍然是一句关于那个事件的话。两种结局都被记到同一个地址上，错觉因此在经验里封死，没有出口。文章后半处理收口问题。这条链子最自然的结尾是”一个事件本质上和任何以前事件都不同类”，但此句有两处病。其一，“本质”正是被拆掉的那类词——底下有个东西撑着，事情因此如此；用它收口，等于在末尾把地基垫回去。其二更要紧：否定那层关系，等于把它重新装上。说”不同类”，仍然承认”同类／不同…Read more
本文处理一条压得很紧的链子：我们感觉意义在事件本身，但意义其实指的是事件的后续；刚刚发生的事件尚无后续，因而尚无意义；于是我们转而预测它的后续；而实际被预测的，是被归为”同类”的诸事件的后续。文章先指出这条链子暴露了什么：意义被感到的地址是错的。它被感到在事件身上，实际住在后续那儿。这不是把内容判断错了，而是把位置弄错了；而位置错误从内部无法察觉，因为人只能在自己相信的那个地址上核对内容。其次给出这一错觉何以极其稳固的机制：后续真的到来时，人不会把它记成”借来的那笔兑现了”，只会记成”果然如此”——兑现被体验成印证。于是那笔借款从头到尾没有一刻以借款的样子出现过。而后续没有到来时，人记的是”我当初理解得不够深”——仍然是一句关于那个事件的话。两种结局都被记到同一个地址上，错觉因此在经验里封死，没有出口。文章后半处理收口问题。这条链子最自然的结尾是”一个事件本质上和任何以前事件都不同类”，但此句有两处病。其一，“本质”正是被拆掉的那类词——底下有个东西撑着，事情因此如此；用它收口，等于在末尾把地基垫回去。其二更要紧：否定那层关系，等于把它重新装上。说”不同类”，仍然承认”同类／不同类”这条轴，只是读数取负；而前面得到的不是负读数，是那儿没有刻度。拆掉一台秤，不等于称出了零。代之以一句更弱也更硬的话：这一次不是那一次。它不主张两者不同类，而是根本不需要”同类”这个词就能说完，因此不欠任何东西。链子的末环应写成”每一次都自成一件，从不曾属于任何一批”——不必替世界宣布类不存在，只需不去替它凑。文末指出：本系列一切结果的力量都来自同一处不对称——预测需要那层关系成立，发生不需要；故”无法被证成”已足够，“已被证伪”既不必要也不可得。 Abstract This paper takes up a tightly compressed chain of thought: we feel that meaning lies in the event itself, yet meaning refers to the event’s sequel; what has just occurred has no sequel yet, and so no meaning yet; we therefore try to predict its sequel; and what we actually predict is the sequel of those events sorted as being of one kind. The paper first identifies what the chain exposes: the address at which meaning is felt is wrong. It is felt in the event and in fact resides in the sequel. This is not a misjudgement of content but a misplacement of location — and errors of location cannot be detected from within, since one can only check contents at the address one believes in. It then gives the mechanism by which the illusion holds so firmly. When the sequel does arrive, it is not recorded as “the borrowed sum has been repaid” but as “just as I thought”: fulfilment is experienced as confirmation. The loan therefore never once appears in the guise of a loan. And when the sequel fails to arrive, what gets recorded is “I did not understand it deeply enough at the time” — still a statement about the event. Both outcomes are posted to the same address, and the illusion is thereby sealed within experience, with no exit. The second half treats the problem of closing the chain. Its most natural ending — that an event is essentially not of the same kind as any previous event — is defective twice over. First, “essentially” is precisely the sort of word under dismantling: something underneath, propping things up. Second and more seriously, to negate the relation is to reinstall it. To say “not of the same kind” is still to grant the axis of same and different, merely taking a negative reading; but what the argument yielded was not a negative reading, it was the absence of a scale. Dismantling a scale is not weighing zero. In its place: a weaker and harder sentence — this one is not that one. It does not assert that they differ in kind; it can be said through without the word “kind” at all, and so owes nothing. The final link should read: each occurrence stands alone and never belonged to any batch. There is no need to declare on the world’s behalf that classes do not exist; it is enough to refrain from assembling them for it. All results in this line of work draw their force from one asymmetry: prediction requires the relation to hold, occurrence requires nothing. “Cannot be established” therefore suffices, and “has been refuted” is neither needed nor available.

Value Theory Science, Logic, and Mathematics Metaphysics and Epistemology
23

借来的后续：论刚刚发生之事的意义

常有人说：我们其实不知道刚刚发生的事情意味着什么。本文认为这个说法方向对，落点却错——它把意义留在那里当作一个藏着的事实，只是我们没看见。按本文的路子应当说：不是不知道，是没有那个东西可知。意义是一笔借来的后续。所谓”这件事意味着什么”，其内容并不是这件事，而是别的事后来跟着什么——是从被归为同类的诸事件的后续里调过来的。而”后续”恰恰是当下唯一没有的东西：刚刚发生的这一件，后面还什么都没有。所以当下的意义不是难以获得，而是结构上不可能在那儿——它所要求的那样东西按定义尚未存在。此刻感到的那份意义，是一段记忆穿着当下的衣服。更进一步，这笔贷款开在一个凑不齐的户头上：调取后续要先有”同类”，而同类不是在事件里找到的，是描述的输出。于是有双重亏空——后续不是这件事的，而供出后续的那个类也不在世界上。这个图景顺手解释两件平常事。误解通常被说成”我理解的和它真正的意思对不上”；按本文，误解只是借来的后续没有兑现，是一次违约——事件那边什么都没变，也没有一个”真正的意思”在那里等着被对上。意义随年月改变通常被说成”我们后来看得更清楚了”；可那件事早就完了，不会再变，变的是卷宗里又积了新的后…Read more
常有人说：我们其实不知道刚刚发生的事情意味着什么。本文认为这个说法方向对，落点却错——它把意义留在那里当作一个藏着的事实，只是我们没看见。按本文的路子应当说：不是不知道，是没有那个东西可知。意义是一笔借来的后续。所谓”这件事意味着什么”，其内容并不是这件事，而是别的事后来跟着什么——是从被归为同类的诸事件的后续里调过来的。而”后续”恰恰是当下唯一没有的东西：刚刚发生的这一件，后面还什么都没有。所以当下的意义不是难以获得，而是结构上不可能在那儿——它所要求的那样东西按定义尚未存在。此刻感到的那份意义，是一段记忆穿着当下的衣服。更进一步，这笔贷款开在一个凑不齐的户头上：调取后续要先有”同类”，而同类不是在事件里找到的，是描述的输出。于是有双重亏空——后续不是这件事的，而供出后续的那个类也不在世界上。这个图景顺手解释两件平常事。误解通常被说成”我理解的和它真正的意思对不上”；按本文，误解只是借来的后续没有兑现，是一次违约——事件那边什么都没变，也没有一个”真正的意思”在那里等着被对上。意义随年月改变通常被说成”我们后来看得更清楚了”；可那件事早就完了，不会再变，变的是卷宗里又积了新的后续，于是那笔贷款被重新定价。“看得更清楚”这个说法，把重新定价说成了看清。文章另指出：预测与意义是同一个动作的两个方向，都是跨着一个并不存在的类做转账。最后设两道防线：拿掉意义并不因此装上虚无——虚无同样是一份判决，要用同一套器械；以及本文不妨碍人照常认出意义、照常行动，那是规范性的问题，此路无出口，尚待研究。 Abstract It is often said that we do not really know what has just happened means. This paper holds that the remark points the right way but lands wrong: it leaves meaning standing there as a hidden fact that we have merely failed to see. On the account developed here the correct statement is: it is not that we do not know, but that there is no such thing to be known. Meaning is a borrowed sequel. The content of “what this event means” is not this event at all but what other events were followed by — drawn from the aftermaths of those occurrences sorted as being of one kind. And a sequel is precisely what the present lacks: nothing has yet followed what has just occurred. Present meaning is therefore not hard to come by; it is structurally absent, since what it requires does not yet exist by definition. The meaning felt at this moment is a memory wearing the clothes of the present. Further, the loan is drawn on an account that was never assembled. To draw on those aftermaths one needs a class of same-kind events, and sameness is not found in events but is the output of a description. The shortfall is thus twofold: the sequel is not this event’s, and the class supplying it is not in the world. The picture accounts for two ordinary phenomena. Misunderstanding is usually described as a mismatch between one’s reading and the true meaning; on this account it is simply a borrowed sequel that failed to materialise — a default. Nothing changed on the event’s side, and no true meaning was ever waiting to be matched. The drift of meaning over years is usually described as our having later seen more clearly; but the event is long over and does not change. What changed is that the dossier accumulated further sequels, so the loan was repriced. “Seeing more clearly” describes a repricing as an act of clarification. The paper also notes that prediction and meaning are one operation run in opposite directions, both transferring across a class that is not there. Two guards close the argument: removing meaning does not thereby install meaninglessness, which is itself a verdict requiring the same apparatus; and nothing here prevents anyone from recognising meanings and acting as before — that is a normative question, and this road has no exit onto it.

Metaphysics and Epistemology Science, Logic, and Mathematics Value Theory
16

同一性不是被找到的，是被供给的：论统计之前的那一步

统计要能进行，先得有一批”同类事件”。本文追问的是这一步：这些事件的同一性从何而来。逐条去找，找不到。同一台机器连续做六次同样的动作，六次的时刻不同、温度不同、部件的磨损不同、周围的空气不同；凡是具体的，全都不同。剩下相同的，只有描述所保留的那一部分——而描述保留什么，在它被造出来的时候就已经定了。所以”同类事件”的准确意思是：过同一台描述、出同一个读数的那些事件。同一性不是在事件里找到的，是那台描述的输出。顺着决定论看，反读甚至更自然：若每一次都被完全决定，则每一次都只有一个结局可言，六次就是六件各自被定死、互不相干的事实。它们彼此之间的差别，与任意两桩历史事件之间的差别并无不同。“实例”这层关系要求先有一个虚拟的东西，再让实际发生去分有它；可实际发生并不分有任何东西，它就是发生。由此得出两点。其一，统计之所以能把诸事件互换，靠的正是把决定了每一次的那些细节删掉；描述越粗，塌进同一格的越多，“概率”涌出得越大方。所以那个数不是从事件里读出来的，是从被删掉的量里长出来的——它是被丢弃之物的度量。描述做到极细，每件事都独一无二，一切频次皆为一，概率归零。其二，分母根本凑不齐：n 要…Read more
统计要能进行，先得有一批”同类事件”。本文追问的是这一步：这些事件的同一性从何而来。逐条去找，找不到。同一台机器连续做六次同样的动作，六次的时刻不同、温度不同、部件的磨损不同、周围的空气不同；凡是具体的，全都不同。剩下相同的，只有描述所保留的那一部分——而描述保留什么，在它被造出来的时候就已经定了。所以”同类事件”的准确意思是：过同一台描述、出同一个读数的那些事件。同一性不是在事件里找到的，是那台描述的输出。顺着决定论看，反读甚至更自然：若每一次都被完全决定，则每一次都只有一个结局可言，六次就是六件各自被定死、互不相干的事实。它们彼此之间的差别，与任意两桩历史事件之间的差别并无不同。“实例”这层关系要求先有一个虚拟的东西，再让实际发生去分有它；可实际发生并不分有任何东西，它就是发生。由此得出两点。其一，统计之所以能把诸事件互换，靠的正是把决定了每一次的那些细节删掉；描述越粗，塌进同一格的越多，“概率”涌出得越大方。所以那个数不是从事件里读出来的，是从被删掉的量里长出来的——它是被丢弃之物的度量。描述做到极细，每件事都独一无二，一切频次皆为一，概率归零。其二，分母根本凑不齐：n 要求先有 n 个同类的可数之物，同一性没有着落，n 就没有着落；不是除法算错了，是除法之前那一步没完成。本文只主张同一性无法被证明，不主张”诸事件确实各不相同”——反读同样立不住。但这已足够，因为不对称在于：统计需要同一性成立，另一侧不需要。文章最后区分本论题与参照类问题：后者问”该选哪一类”，仍承认类之存在，是选择困难；前者问”实例这层关系在世界上有没有”，是位置空缺。至于统计在实践中照常奏效一事，本文不予处理，尚待进一步研究。 Abstract Before any statistics can begin, a batch of “events of the same kind” must be in hand. This paper interrogates that step: where does their sameness come from? Looked for item by item, it is not there. Let one machine perform the same action six times: the moments differ, the temperatures differ, the wear on the parts differs, the surrounding air differs. Everything concrete differs. What remains the same is only what the description retains — and what a description retains was settled when the description was built. “Events of the same kind” therefore means precisely: those events that pass through one description and yield one reading. Sameness is not a finding about events; it is the output of that description. Read along determinist lines, the reverse reading is if anything more natural. If each occurrence is fully determined, each has only one outcome to speak of, and six occurrences are six separately fixed, mutually unrelated facts. The difference between any two of them is no different in kind from the difference between any two historical events. The relation “instance of” requires a virtual item to exist first, in which actual occurrences then partake; but actual occurrence partakes of nothing — it simply occurs. Two consequences follow. First, statistics can substitute its events for one another only by deleting the very details that determined each one. The coarser the description, the more collapses into a single cell, and the more generously “probability” wells up. The number is thus not read off the events; it grows out of the quantity deleted — it is a measure of what was discarded. Push the description to full resolution and every event is unique, every count is one, and probability vanishes. Second, the denominator cannot be assembled at all: n requires n countable things of one kind to be given first; with sameness unfooted, n is unfooted. The division is not miscalculated — the step before the division was never completed. The paper claims only that sameness cannot be established, not that the events are in fact all different; the reverse reading stands no better. This suffices, because of an asymmetry: statistics needs sameness to hold, while the other side needs nothing. Finally, the thesis is distinguished from the reference-class problem, which asks which class to use and thereby still grants that classes exist — a difficulty of choice, where this is a vacancy of position. Whether and how statistical practice continues to work is not addressed here and remains open.

Science, Logic, and Mathematics Metaphysics and Epistemology Value Theory
14

拼接：论对手从不说出那句要害的话，而桥是我们自己搭的

对手往往是分开论证、分开表述的。而这些表述之间，表面上构成了某种关系——如果我不深究的话。他手上通常是这样两句：一、“世界有量子随机过程，因此非决定性。” 二、“大脑内的微观随机涨落可能被放大，导致作出与原本似乎被决定的决定所不同的决定。” 两句都站得住。第一句是物理学的话；第二句带着”可能”和”似乎”，谨慎得挑不出毛病。可他要的结论是第三句：“所以我本来可以不那么决定，所以我的懊恼是有道理的。” 而这第三句，谁也没说过。第一句只讲了世界，第二句只讲了”可能被放大”；从”可能被放大”到”我本来可以”，中间隔着一整段，而这一段没有一个字。它坐在两句之间的空档里——没人主张它，所以没人为它负责。你去问第一句，答的是量子力学；去问第二句，答的是神经科学；去问那个结论，两边都说”我可没这么讲”。这就是分开表述最难对付的地方：每一句单看都无懈可击，而那句真正要害的话，他一句也没说。所以这不是一条论证，是一次拼接——而拼接的活儿是读者干的。桥搭在哪儿？搭在一个字上。“决定”在第一句里是”被前一刻定死”，在第二句里是”我做出的那个决定”。同一个字，两回事。字面接上了，就以为意思也…Read more
对手往往是分开论证、分开表述的。而这些表述之间，表面上构成了某种关系——如果我不深究的话。他手上通常是这样两句：一、“世界有量子随机过程，因此非决定性。” 二、“大脑内的微观随机涨落可能被放大，导致作出与原本似乎被决定的决定所不同的决定。” 两句都站得住。第一句是物理学的话；第二句带着”可能”和”似乎”，谨慎得挑不出毛病。可他要的结论是第三句：“所以我本来可以不那么决定，所以我的懊恼是有道理的。” 而这第三句，谁也没说过。第一句只讲了世界，第二句只讲了”可能被放大”；从”可能被放大”到”我本来可以”，中间隔着一整段，而这一段没有一个字。它坐在两句之间的空档里——没人主张它，所以没人为它负责。你去问第一句，答的是量子力学；去问第二句，答的是神经科学；去问那个结论，两边都说”我可没这么讲”。这就是分开表述最难对付的地方：每一句单看都无懈可击，而那句真正要害的话，他一句也没说。所以这不是一条论证，是一次拼接——而拼接的活儿是读者干的。桥搭在哪儿？搭在一个字上。“决定”在第一句里是”被前一刻定死”，在第二句里是”我做出的那个决定”。同一个字，两回事。字面接上了，就以为意思也接上了。那怎么办？不去跟那两句吵，而是抓住一个大脑、它的一个决定、它的一次懊恼。如果不抓住这一个，世界图景就会被感知得极度复杂，无从下手——散成”世界总体是不是决定性的”这样一个题目，任何一句话都既不能被证实也不能被证伪。没有主语，就没有可以追问的对象。而抓住之后，每一句话都得有主语：谁的决定？谁的懊恼？那个随机取值，谁能要求它出现？——问到这一句，就没人接了。对手靠分开表述来避免被追问；把话收到一个具体的人身上，等于给每一句按上主语，拼接当场就散了。 Abstract Opponents typically argue and state things separately. And these statements appear to stand in some relation to one another—so long as I do not press. He usually holds two sentences: (1) “The world contains quantum random processes, and is therefore not deterministic.” (2) “Microscopic random fluctuations in the brain may be amplified, producing a decision different from the one that seemed to be determined.” Both hold up. The first is physics; the second, hedged with “may” and “seemed,” is beyond reproach. But the conclusion he wants is a third: “so I could have decided otherwise, and my regret is warranted.” And no one has ever said that third sentence. The first speaks only of the world, the second only of “may be amplified”; between “may be amplified” and “I could have,” a whole stretch is missing, and not one word of it has been written. It sits in the gap between the two—no one asserts it, so no one answers for it. Question the first and you get quantum mechanics; question the second and you get neuroscience; question the conclusion and both say, “I never said that.” This is what makes separate statement so hard to counter: each sentence alone is unimpeachable, and the sentence that matters was never uttered. So it is not an argument but a collage—and the assembling is done by the reader. Where is the bridge built? On one word. “Determine” in the first sentence means “fixed by the preceding instant”; in the second it means “the decision I make.” One word, two things. The faces match, so we assume the meanings do. What to do? Not quarrel with those two sentences, but take hold of one brain, one decision, one episode of chagrin. Without holding onto this one, the picture of the world is perceived as impossibly complex, with nowhere to begin—it disperses into “is the world as a whole deterministic,” where nothing can be confirmed or refuted. With no subject, there is nothing to interrogate. Once held, every sentence must have a subject: whose decision? whose chagrin? And who can require that random value to occur?—at this sentence, no one answers. The opponent avoids interrogation by stating things separately; drawing the matter down onto one particular person supplies every sentence with a subject, and the collage falls apart at once.

Value Theory Metaphysics and Epistemology Science, Logic, and Mathematics

Prev.
1
2
3
4
5
6
7
8
9
…
39
40
Next