一条公设换掉以后
When a Postulate Is Replaced
不是抽掉一块砖
把理论比作建筑，会自然地说：公设是地基，抽掉一块，整栋楼就塌。这个比喻过于粗糙。理论不是只有“站立”与“倒塌”两种状态；不同结论依赖不同公设，同一句话也可能在另一条路线取得新的担保。
《第一批判》把替换处理成传导。每项主张登记自己的依赖，一条公设撤回时，变化沿清单抵达相关条目。抵达以后，条目失去原来的着落，被送回“待重审”，而不是自动改判为假。
这给修订留下了精度。依赖 P3 的守恒结论要重审，不依赖 P3 的辖域判断不必陪葬；一条结论若能从另一组前提重新推出，可以取得新地址，但必须明说路线已经改变。
历史地址不能抹掉
替换以后，旧结论曾经在旧前提下成立这一事实仍然保留。过去的论文、决策或模型不能被重写成“当时从未有效”，也不能因为历史上有效就继续占用今天的地址。
这一区分对思想史尤其重要。我们可以准确说某个论证在当时采用的分支与知识条件下成立，同时指出今天已经撤回其中一项承担。既不把过去的人判成愚蠢，也不让过去的成功冻结现在。
版本记录于是成为哲学账目的一部分：何时更换公设，哪些条目降档，哪些通过新担保恢复，哪些仍待处理。修订不是抹除痕迹，而是增加可追踪的时间坐标。
新公理候选怎样入场
若有人提出一项比公设更强、应当列为公理的候选，《第一批判》给出三步程序。第一是施行检验：否认它时，否认动作是否不可避免地施行了它？若没有，就不能凭“非常基本”取得公理地位。
第二是还原检验。候选能否改写成已有公理的表达，能否降为条件命题，是否其实是一项新的可替换承担，或者材料不足而应悬置？这一步防止不同措辞把同一内容重复登记，也防止重要直觉越级。
第三是入册。无论结果是公理、公设、假说还是悬置，都要写明理由与依赖。程序开放，但门槛不因开放而降低。
“自己的哲学”落在承担上
一门哲学是不是自己的，不取决于所有句子都由自己发明。语言、问题、概念和论证大多来自漫长传统。若原创性是署名条件，没有人真正拥有哲学。
《第一批判》把“自己的”放在承担上。你可以继承康德的一项区分、采用他人设计的形式工具、甚至接过一整套公设；只要你知道自己承担了什么，愿意公开其成本，并在替换时接住传导，它便以承担的方式成为你的哲学。
反过来，最原创的措辞若只靠权威、潮流或自动执行而无人承担，也还不是自己的。生成来源与署名责任是两栏。
可修订不是轻浮
有人担心，可替换公设会让理论随时改口。其实隐藏的公设更容易随意改口，因为旧依赖没有记录，失败时只需换一种解释。传导制度让每次更换都付出公开成本。
一条公设换掉以后，我们要知道哪些结论暂时失去着落，哪些历史判断仍按旧址保存，哪些新分支需要重建。修订越自由，账目越要严格。
这使哲学不必在僵硬与崩塌之间选择。它可以承认根本承担发生变化，同时保持局部成果、历史记录和后续作业的连续性。理论不是一栋不能换砖的楼，而是一条有地址、有版本、能说明自己怎样改道的路。
Not a Brick Pulled from a Foundation
Architectural metaphors encourage a simple picture: postulates are foundation stones, and removing one collapses the building. The picture is too coarse. Theories have more than two states, standing and fallen. Different conclusions depend on different commitments, and a claim may obtain a new warrant by another route.
The First Critique treats replacement through propagation. Every claim records its dependencies. When a postulate is withdrawn, the change travels along the list to affected entries. Those entries lose their former standing and return for review. They do not automatically become false.
Revision gains precision. Conclusions relying on P3's conservation require review, while a jurisdictional result independent of P3 need not fall with them. A conclusion derivable from another premise set can acquire a new address, but the changed route must be declared.
Historical Addresses Must Remain Visible
After replacement, the fact that an old result stood under its old premises remains part of the record. Earlier papers, decisions, and models should not be rewritten as if they were never valid. Nor may historical validity reserve the same address indefinitely after its conditions have changed.
The distinction is especially important in intellectual history. An argument may have been sound under the branch and knowledge available at the time, while one of its undertakings is no longer accepted. This neither turns previous thinkers into fools nor allows earlier success to freeze the present.
Version history becomes part of philosophical bookkeeping. When was a postulate replaced? Which entries were downgraded? Which regained standing through new warrants? Which remain pending? Revision preserves traces and adds temporal coordinates rather than erasing the road behind it.
Admitting a New Axiom Candidate
If someone proposes a commitment stronger than a postulate and asks that it be treated as an axiom, the First Critique supplies a three-part procedure. First comes the performative test. Does denying the candidate unavoidably enact it? If not, being intuitively fundamental does not confer axiomatic status.
Second comes reduction. Can the candidate be reformulated as the existing axiom, reduced to a conditional statement, classified as a new replaceable undertaking, or suspended because evidence is insufficient? This step prevents new wording from duplicating an old entry and prevents an important intuition from being promoted without examination.
Third comes registration. Whether the result is axiom, postulate, hypothesis, or suspension, the reason and dependencies are recorded. The procedure is open; openness does not lower the threshold.
One's Own Philosophy Lies in Assumption
A philosophy is not one's own because every sentence was invented by its current author. Language, questions, concepts, and arguments come through long traditions. If originality were the condition of ownership, no one would possess a philosophy.
The First Critique locates “one's own” in assumption. A thinker may inherit a Kantian distinction, adopt someone else's formal instrument, or receive an entire set of postulates. If she knows what she assumes, exposes its costs, and receives the consequences when replacement propagates, it becomes her philosophy through responsibility.
Conversely, wholly original phrasing that runs on authority, fashion, or automatic execution without anyone assuming it is not yet one's own. Generative origin and signed responsibility belong in separate columns.
Revisability Is Not Carelessness
Some worry that replaceable postulates permit a theory to change its story whenever convenient. Hidden postulates permit that behavior more easily: dependencies are absent, so failure can be met with an unrecorded reinterpretation. Propagation makes every change publicly costly.
After a postulate is replaced, we should know which conclusions have lost standing, which historical judgments remain stored at earlier addresses, and which new branch must be built. The greater the freedom to revise, the stricter the account must become.
Philosophy then need not choose between rigidity and collapse. It can acknowledge a change in fundamental commitments while preserving local results, historical records, and continuity of work. A theory is not a building whose stones may never move. It is a road with addresses and versions that can explain how it changed direction.