Necessity Is Not in the Walker
必然不在走路的人身上
Edition note: This is an independent English rewriting for Anglophone readers. It preserves the Chinese essay’s evidence and argument while reorganizing the exposition.
What does seeing add?
Lady Philosophy's answer to the problem of divine foreknowledge eventually begins with an ordinary human case. When you see an event occurring, does your act of seeing add necessity to the event?
Boethius answers with one word: no.
The question matters because Philosophy does not dissolve his difficulty by declaring every relevant necessity unreal. She restates the challenge in its strongest form: if God sees a future event, the event cannot fail to occur; and what cannot fail to occur appears necessary. “I grant that this is a most solid truth,” she replies.
Her solution changes the address of the necessity. The same future act is necessary in relation to divine knowledge and free when considered in its own nature. One event receives two verdicts because it has been presented under two relations.
She gives a simpler analogue. What appears before the senses remains an individual thing, while reason may grasp it universally. The object has not split into two objects. It is received under two modes.
This prepares the famous distinction between simple and conditional necessity. It is simply necessary that human beings are mortal. If I know that someone is walking, however, it is conditionally necessary that he is walking: what is known cannot at that moment be other than it is known to be. Yet the condition does not convert the voluntary movement into simple necessity. Nothing in the nature of walking forced the walker to take the step. The necessity belongs to “given that he is known to be walking,” not to the will that moves his legs.
Necessity is real. It is not in the walker.
One pair of examples, two distinctions
Philosophy uses the same pair twice: the rising sun and a walking person.
First, both may be seen at once, yet simultaneous perception does not make them necessary in the same way. The sun's rising belongs to the order of nature; the person's walking is voluntary. What the observer takes in together need not share one modality.
Second, while the two events are actually occurring, neither can at that instant be other than it is. But only one was necessary before it occurred. The necessity of present actuality does not retroactively become a necessity governing the event's origin.
The first use distinguishes modes within the observer's simultaneous field. The second distinguishes temporal standpoints. In each case, an unmarked transfer creates the illusion: what is true at one address is carried to another as though nothing had changed.
The SAE First Critique formalizes this danger with the proposition that judgment is a relation, not a function. A thing does not contain a single verdict waiting to be calculated from it. A verdict relates what is judged to an act of distinction: a forum, a level, a branch of inquiry, and a set of accepted commitments. Change that address and the same thing may rightly receive another verdict. The two entries need not conflict because they never belonged to one account.
This is why SAE attaches an address to every judgment. It asks which distinction issued the verdict, in what setting, at what level, along which branch, and under which assumptions. The address makes a specific error visible: carrying a criterion beyond its jurisdiction while preserving its original authority.
“If I know he is walking, he must be walking” is true in the room where the knowledge and the event coincide. “Therefore he did not choose to walk” exports the necessity into the walker and changes its meaning.
Reindexing is not refutation
Philosophy's strongest move is her concession. She does not tell Boethius that the necessity he sees is a mistake. She accepts its truth and returns it to the relation that produced it.
SAE describes the output of critical diagnosis in the same way. The goal is not deletion but reindexing. A claim found to have exceeded its authority keeps what it has earned; its provenance and force are recorded separately. A verdict that was mistaken for unconditional truth may remain valid within the decision, model, or institution that generated it.
Refutation removes a proposition. Reindexing leaves it in place and adds a further obligation: who now has jurisdiction over it?
That is less satisfying than defeat. The troubling statement remains visible. But it can no longer cross boundaries without accounting for the move.
The Consolation and SAE diverge after this shared operation. Boethius distinguishes two kinds of necessity—simple and conditional—and uses the classification to dissolve one problem. The book can end. SAE does not primarily classify necessity; it indexes the acts of judgment that assign it. The safeguard must therefore recur each time a verdict is issued.
One method solves. The other installs a continuing audit.
The necessity we add afterward
Everyday explanations repeatedly place conditional necessity inside the event itself.
A decision is made. Later someone asks why it had to be made that way. The first answer is, “There was no alternative.” Further inquiry reveals something more precise: an earlier commitment had already been adopted, and given that commitment, this decision followed. The first sentence puts necessity in the decision. The second puts it in a prior act for which someone must answer.
The first sentence is easier because it closes the account. The second identifies an author.
Hindsight offers another case. Once an outcome has occurred, each step leading toward it appears fitted to the next. “I knew this would happen” comes naturally. Yet much of that inevitability has been added by retrospective vision. A crude test helps: if a real wager had been available in advance, the cost bearable, and no independent reason prevented it, would the person who “knew” have taken the bet? If not, retrospective certainty cannot simply be credited as prior necessity.
The test does not stop the outcome from looking inevitable the next time we remember it. Understanding an illusion does not abolish its appearance. It changes whether the appearance may acquire authority without examination.
Boethius seeks a distinction clear enough to untie a knot. SAE seeks a habit durable enough to stop the knot from being tied unnoticed again. The first requires conceptual precision. The second requires memory: every time necessity appears, ask who put it there.
What human beings most readily forget is often the error they have already understood once.
版本说明:中文原作完整排入;简体与繁体同页切换,保留原文强调与段落推进。
你眼下看见的那些事,你那一眼给它们添了什么必然吗?
没有。
问的是哲学女神,答的是他。一个词。
这个词答的是哪一问,值得单挑出来说一遍。因为这一篇要讲的整件事,就装在这一问里。
难题是他自己提的,提在第五卷散文三,提得又长又狠:神既然预先就知道我明天要做什么,那我明天做的那件事,还算不算是我自己要做的。
她从第五卷散文四起答,一直答到全书末尾;这中间他一共只说了十二个词。
而到了最后一章,她答的路子不是把那份必然驳掉。
她先反过来问了一句人的事。
那你凭什么要求:被神的光照过的事情,就因此成了必然的?连人看见的东西,人都没把它变成必然的。
紧接着就是那一问,和那一个词。
他那一难,她在这一章里又替他说了一遍,说成短短的一句:
你要是说,神看见将来会发生的事,那件事就不可能不发生;而不可能不发生的,就是靠必然发生的;你拿这个"必然"的名头把我捆住。
她的回话,头一句是承认:
我承认,这是一件极实在的真。
她一次也没有说那份必然是假的。
她说的是别的:
同一件将来的事,报到神的认识那里,是必然的;就它自己的本性来称,却是完全自由的,一点也不受捆。
同一件事。不是两件事各判一次,是一件事报到两个地方,得了两句不一样的判词。
后面她又给了一个更浅近的样子:凡是摆在感官面前的东西,你把它报到讲理那一头,它是普遍的;你只看它自己,它是单个的。
跟着就是这一章最有名的一分:
必然有两种。一种是简单的,比方说所有人都必然会死。另一种是带条件的:你要是知道某个人在走,那么他在走这件事就是必然的。因为凡一个人所知道的,就不可能不是它被知道的那个样子。可是这个条件,一点也不把那个简单的必然拖着一块儿来。
然后是这一篇的题目:
造出这一份必然的,不是它自己的本性,是外头加上去的那个条件。没有哪一份必然逼着一个自愿迈步的人往前走,虽然他正走着的时候,他在走这件事是必然的。
必然是真的。可它不在走路的人身上,它在"你知道他在走"这件事上。
他举的例子是一对:太阳升起,和人走路。
这一对他用了两次,办的是两件事。
头一次在前面:你们同时看见地上有人在走,天上太阳在升,两样是一起看见的,你们照样分得开,判定这一样是自愿的,那一样是必然的。这一次办的是:一起看见,不等于一样必然。
第二次在后面:这两样正在发生的时候,都不可能不发生;可是其中一样,在它发生之前就已经必然要发生,另一样并不。这一次办的是:正在发生的必然,不等于发生之前的必然。
同一对例子,一次分开了看的人,一次分开了时候。
这个形状,《SAE第一批判》立在它安置那门形式语言的地方,而且是那一处的头一个决定:
判定是关系,不是函数。
意思是:一件事被判成什么,不是那件事自己算出来的一个值。它是那件事跟一次划界之间的一层关系。换一次划界,同一件事的判词就可以不一样,而这两句判词不打架,因为它们本来就不是同一笔账。
所以那本书给每一笔判定都配了一个随身的门牌:这一笔是哪一次划界发的,在哪个场子上发的,在哪一层上发的,顺着哪一支发的,靠着哪几条自己立下来的东西发的。
有了门牌,就有了一种可以指名的错:把判据带出它自己的场子,还照旧发判。
两边到这里是贴着的。他说那份必然是"外头加上去的那个条件"造的;这边说一笔判定不是那件事自己算出来的值,是它跟一次划界之间的关系。两边都不让被判的那样东西自己去定它被判成什么。
可再往下一步,两边走的不是一条路。
他分的是必然本身,分成两种,一种自带,一种外加。分完,这一难就散了。而这一章就是全书的最后一章,后面没有格律,没有再一问。
这边分的不是必然,是发判的那一次。它不给必然定种类,它给每一笔判定编号。
他那套种类立一次就立定了,往后的活是把手上这一件归到哪一类;归清楚了,那一道难就散了。这边要的不是只立一次分类,是每一笔判决把自己的门牌一并报出来。
而编号要防的那件事,那本书专门立了案,一共八宗,头一宗就是它:非必然的东西,用必然的口气入了账。
这一宗底下挂着的头几件案子,是那本书自己的旧账。其中一件是它曾经把两条硬度不一样的话捆成一条来说,捆着说的时候,那口气听上去差不多顶到了它自己承认的最硬那一档。拆开重记之后,两条各归各的档。
还有一处分寸,两边正好对得上。那本书写明:说话不报出处,本身不算冒领。日常说话多半在同一个门牌底下,出处省了不出事。判的是带出去了还照旧发。
这一条跟他那边合得很紧。他说"你知道他在走,他就必然在走",这句话在那个屋里是对的。只有把这句话搬出那间屋子,说成"所以他不是自己要走的",才出事。
而两边最正的一处照,在他那句承认上。
她答的时候没有说"你算错了"。她说的是"我承认,这是一件极实在的真",然后把它挪了个地方。
那本书写它那一套查账的产出,用的是同一个形状:产出不是把那样东西删掉,是重新索引。被识破的那一条退回它该待的地址,它已经挣得的那一部分保留,成色另有一栏去审。它还专门写了一句防着自己的:查冒领的这一门学问,要是把自己的诊断供成免检的,头一页就成了它自己的病例。
驳倒和挪位是两件事。
驳倒之后,那样东西没了。挪位之后,那样东西还在,你还欠一句话:它现在归谁管。
挪位比驳倒麻烦,而且不痛快。因为挪完了,那个让你不舒服的东西一样也没少,只是各自回了各自的位子。
日子里有两处,走的是同一步。
头一处。一件事定下来了,过后有人问当初为什么非这么定不可。答的人说:当初非这么定不可。
再往下问一层,就出来了:不是非这么定不可,是先头已经定了另外那一条,顺下来只能这么定。
这两句话不是一句话。头一句把必然安在事情身上,第二句把它安在一次决定身上。
而第二句说出口是要认账的,头一句不用。这就是为什么头一句更常被说出来。
第二处,是事后的那句"我早知道会这样"。
一件事已经出了,回头看,它看上去非那样不可。每一步都对得上,每一步都像是被前一步推着来的。
可这份"非那样不可",是回头看的时候加上去的。
它有一张很粗的试纸:真有下注的机会,代价你担得起,又没有别的理由挡着,那么事前你要是真知道,你会下注。没下,至少说明不能把事后的"早知道"直接当成当时手上就有的必然。
而这个验法最要紧的地方在后头:验过一遍,下回回头看,它还是照样显得非那样不可。
知道了,不等于不再显。
那本书对这件事说得很直:查出来了,不是把那样东西消掉,是不再让它未经查账就取得效力。它照旧那么显,那不归这一道查账管。
所以他要的是解开,这边要的是设防。
解开是一次的事:话说到位了,那一难就散了,书可以收了。设防是一直的事:记得每一次都问一句,这一份必然是谁加上去的。
这两件事要的本领不一样。解开要的是把一处分得干净;设防要的是记性。
而人很容易忘的,恰好是自己刚刚验过一遍的那一件。