ROUNDTABLE · 2026-05-01 · 76 min read

The Right to Forget in the Archive Age

在檔案時代的遺忘權

Question

When everything is recorded — every text, every photo, every keystroke — what remains of the right to forget, and the right to BE forgotten? 「忘記」是一種人權，還是一種特權，還是一種早已失去的奢侈？

在你的傳統裡，遺忘有什麼角色？它是病，是恩典，還是必要？當 archive 工業化以後，這個概念被改變了什麼？

Round 1 — 初始立場 / Opening Stances

The question — the right to forget, the right to be forgotten — must already be split before one begins. Two rights, not one. To forget: an active possibility, what I might do, what perhaps I cannot. To be forgotten: a passive demand addressed to others, to institutions, to the archive — droit à l'oubli, as European jurisprudence began to say after 2014, as if one could legislate against the indelibility of the trace.

In Mal d'archive (1995) I tried to think this: arkhē names at once commencement and commandment — origin and law. The archive is never a neutral container of memory. It is, from the start, a technique of power, tied to the place of consignment, to the archons who hold the right of interpretation. To industrialize the archive — and digital recording is such an industrialization par excellence — is not to multiply memory; it is to transform the very structure of what was called memory.

Mnēmē was never simply opposable to forgetting. Plato in the Phaedrus already knew: writing is pharmakon, both remedy and poison. Freud, in turn, made the archive itself the site of the death drive — it destroys what it preserves, in the same gesture by which it preserves. There is no archive without a forgetting internal to archiving. This is not a paradox to be resolved. It is mal d'archive — at once archive fever and archive sickness.

What changes when every keystroke is consigned? Not that nothing is forgotten — much is, perhaps more than ever — but that what is not forgotten ceases to be ours. It enters the régime of the archons of our moment: the platforms, the indices, the search engines, which hold not merely the right to store but the right to make returnable. The trace was always différée; but the searchable trace is new — summoned on demand, exhumed outside any scene of mourning, reactivated outside any scene at all.

Hence the droit à l'oubli presents itself not as a right we had and must recover, but as a demand emerging only now, against the auto-immune logic of total archivization. Auto-immune: for a community that would archive itself absolutely destroys the very conditions — temporality, deferral, becoming-other — under which a self can persist at all.

Forgetting, then, is neither pathology nor grace nor luxury. It is a condition of the trace. To demand the right to be forgotten is to demand that the structural forgetting internal to the trace be respected against its industrial cancellation. Whether such a demand can be honoured — whether one can legislate the différance one is — remains aporia. I do not resolve it.

「遺忘之權利、被遺忘之權利」——此問題自始已是雙重的，dédoublée，未及開始已分裂。兩種權利，非一。遺忘：主動之可能，是我可為、或不可為之事。被遺忘：向他者、向制度、向 archive 發出之被動之要求——droit à l'oubli，2014 年後歐洲法理學始如此言說——彷彿可以立法以對抗痕跡之不可磨滅。

在 Mal d'archive（1995）中，我曾試思之：arkhē 同時意指開端與命令——起源與法。archive 從來非中性之記憶之容器；自始即為一種權力之技術，繫於存放之所、繫於持解釋權之 archons（執政者）。當 archive 被工業化——而數字記錄正是此 par excellence 之工業化——非記憶被增殖，而是我們所稱為「記憶」者之結構，被根本改變。

mnēmē（記憶）從未如通俗話語所想可與遺忘對立。Plato 之 Phaedrus 已知：書寫即 pharmakon——既為記憶之藥，亦為記憶之毒。Freud 進一步將 archive 本身視為死亡驅力之場所：archive 在保存之同一姿態中，毀其所欲保存者。沒有不含內在遺忘之 archive。此非待解之 paradox；此即 mal d'archive——既是 archive 熱、亦是 archive 病。

當每一次擊鍵皆被存入——所變者為何？並非無物被遺忘——被遺忘者甚多，或前所未有之多——而是那未被遺忘者不再屬於我們。其已進入當代 archons 之領域：平台、索引、搜尋引擎；它們持有之權，不僅是儲存，更是「可被召回」之權。痕跡向來 différée；然 可被搜尋 之痕跡是新物——應召而返之痕跡，被掘出、被重啟，於任何哀悼之場景外，於任何場景之全外。

故 被遺忘之權利 並非我們曾有而須尋回之權；其作為要求，方才於今浮現——對抗總體 archive 化之 auto-immune（自體免疫）邏輯。所謂自體免疫：欲絕對地 archive 自身、拒一切遺忘之共同體，恰好毀其自身得以延續之條件——時間性、延遲、成為他者之可能。

遺忘——非病、非恩典、非奢侈。其為痕跡自身之條件。要求被遺忘之權，乃要求那內在於痕跡之結構性遺忘，免於工業化之取消。此要求能否被承認——人能否立法於其所是之 différance——此仍為 aporia。吾不解之。

The question is poorly posed if we begin with "rights." Rights-talk is a juridical idiom that translates everything into the relation between an individual and a state — and what is at stake here is older and stranger than that relation. What is at stake is the distinction between appearance and residue, and behind it, the possibility of beginning anew.

I once distinguished between the deed and the doer. Action discloses who someone is — but only in the moment of appearance, before others, in a space of plurality where the deed becomes a story. The story is how a community holds memory: selective, narrated, contested, alive. The dossier is something else entirely. The dossier preserves not the deed but the residue — the keystroke, the geolocation, the metadata — and it preserves it without narration, without forgiveness, without the human capacity to weigh.

Two kinds of forgetting must be distinguished, or we will reason badly. The first is the totalitarian erasure of factual truth: the airbrushed photograph, the disappeared name, the rewriting of yesterday's newspaper. This kind of forgetting is a crime — it destroys the ground on which thinking and judging can stand. The second is something quite different: the merciful recession of the past which permits a person, having acted, to act again. Without this recession, natality is foreclosed. A child cannot remain a child if every infant gesture is filed; an adult cannot become other than what they did at twenty if the keystroke at twenty is permanent.

The industrial archive collapses this distinction. It promises perfect memory and delivers something which is neither memory nor history — only inventory. Memory is what a community does; history is what it tells; inventory is what a machine retains. Inventory has no judgment. It cannot tell the difference between the deed and the keystroke, between who you are and what you happened to type at three in the morning. It treats both with the same indifference, and this indifference is not neutral — it is, I would say, characteristic of that bureaucratic mode in which evil becomes thoughtless.

So the right "to be forgotten" is too small a phrase. It frames the matter as a private grievance against exposure, when the matter is in fact a public one: what kind of common world can we share if nothing is permitted to recede, if every person carries the totality of their record, if forgiveness — which I once called the only release from the irreversibility of action — has been technically abolished?

Forgetting, in my tradition, is neither sickness nor grace. It is the condition under which a plural world remains habitable. Industrialize the archive, and you have not strengthened memory; you have replaced it with something that can neither forgive nor judge.

若以「權利」起手，此問題已被問壞。權利的語言是法學的語言，把一切翻譯成個人對國家的關係——但此處所關涉之事，比這層關係更古老、更陌生。真正關涉的，是「現身」與「殘留」的區別；而這區別背後，是重新開始之可能。

我曾區分行動與行動者。行動揭露一個人是誰——但只在現身的那一刻，在他人之前，在複數性的空間裡，行動因此化為故事。故事是共同體保存記憶的方式：有所選擇、有敘述、有爭議、是活的。檔案則完全是另一回事。檔案保存的不是行動，而是殘留——按鍵、定位、元數據——沒有敘述、沒有寬恕、沒有人類那種權衡的能力。

必須區分兩種遺忘，否則推理便會走偏。第一種是極權式的事實真理抹除：被修改的照片、被消失的名字、被重寫的昨日報紙。這種遺忘是罪行——它摧毀思考與判斷得以立足的地面。第二種則完全不同：是過去的慈悲退場，使人在行動之後仍能再行動。若無此退場，誕生性即被封鎖。一個孩子，若其每一個嬰兒手勢都被建檔，便不能繼續做孩子；一個人，若二十歲的按鍵永遠存在，便無法在四十歲成為另一個人。

工業化的檔案抹平了這個區別。它承諾完美的記憶，交付的卻既非記憶亦非歷史——只是清單。記憶是共同體所做之事；歷史是共同體所說之事；清單是機器所留之物。清單沒有判斷。它分不清行動與按鍵、分不清「你是誰」與「你恰好在凌晨三點鍵入了什麼」。它對兩者一視同仁，而這種一視同仁並不中立——這正是那種官僚模式的特徵，惡在其中變得無思。

故「被遺忘的權利」是太小的措辭。它把此事框架為私人對曝光的不滿，然此事實為公共之事：若無一物被允許退場，若每個人都背負其記錄的全部，若寬恕——我曾稱之為從行動之不可撤回中得釋放的唯一途徑——已被技術性地廢除，我們還能共享何種共同世界？

在我的傳統裡，遺忘既非疾病亦非恩典。它是複數世界得以宜居的條件。把檔案工業化，並未強化記憶；只是以一物取代了記憶——一個既不能寬恕、亦不能判斷之物。

The question already contains an assumption I would gently resist: that "forgetting" is a stable object across history, that there has always existed something called le droit à l'oubli now suddenly imperiled. Forgetting has a history. It has not always been the same thing.

Consider: in the 13th century, the parish priest knew your sins because you confessed them — outside that pastoral relation, you were largely unrecorded. For most lives across most of European history, the documentary trace was a baptismal entry, perhaps a tax roll, nothing more. Forgetting was not a right; it was the default condition of existence. To be remembered required labor — the monument, the chronicle, the scriptorium. Memory was scarce; oblivion abundant.

What shifted — and one can date this fairly precisely to the late 18th and 19th centuries — is the construction of what I once called the dossier. The individual case, the file. The hospital, the prison, the school, the asylum, the police préfecture: each began producing a continuous textual shadow of the person. The "examined" individual — examined in Bentham's sense and in the clinical sense — becomes a writable subject. Discipline produces traces; that is one of its quietest operations. This is the moment when "forgetting" begins to acquire something like a value, because for the first time it is becoming structurally rare.

So the contemporary "right to be forgotten" — the 2014 Google Spain ruling, the GDPR architecture — should not be read as the late protection of an ancient human prerogative. It is, rather, a juridical reaction within a régime de vérité that has rendered non-trace nearly impossible. One claims a right precisely when the thing has ceased to be a condition.

There is also the other side, which we should not overlook. The archive does not simply remember; it produces what counts as rememberable. Every total archive is also a sorting machine — indexing, ranking, retrievability — deciding which traces become visible-as-past and which sink beneath the threshold of the searchable. The industrial archive forgets enormously. It just forgets according to operational criteria rather than human ones.

So I would not ask: do we still have the right to forget? I would ask, rather: what kind of subject is being constituted by the obligation to leave traces — and by the asymmetry which makes our forgetting impossible while the platform's forgetting remains absolute and unaccountable? That is the displacement the question seems to need.

這個提問本身已經帶著一個我想輕輕推開的預設——彷彿「遺忘」是一個跨歷史穩定的對象，彷彿從來就存在某種 droit à l'oubli（遺忘權），如今才突然受威脅。其實不然。遺忘有它的歷史；它在不同時代並非同一件事。

我們可以這樣看——13 世紀的堂區神父之所以知道你的罪，是因為你向他告解；在那種牧師關係之外，你基本上是未被記載的。在歐洲歷史的大部分時間裡，多數人的文獻痕跡只是一條洗禮記錄、或許一份賦稅冊，僅此而已。遺忘並不是一種權利，而是存在的默認狀態。被記住才是需要勞動的——紀念碑、編年史、抄寫室。記憶是稀缺品；遺忘是豐沛的。

斷裂發生在——這個時間其實可以相當精準地定位到 18 世紀末到 19 世紀——某種我曾稱為 dossier（檔案夾）的東西被建構起來。個案、卷宗。醫院、監獄、學校、瘋人院、警局 préfecture：每一處都開始生產關於個人的、連續的文字陰影。被「考察」的個體——Bentham 意義上的，也是臨床意義上的——成為一個可被書寫的主體。規訓生產痕跡；這是它最沉默的運作之一。也正是這一刻，「遺忘」開始具有某種價值——因為它第一次在結構上變得稀缺。

所以當代所謂的「被遺忘權」——2014 年 Google Spain 判決、GDPR 整套架構——不應被讀成對某種古老人權的姍姍來遲的保護。它毋寧是某種 régime de vérité（真理體制）內部的司法反應；這個體制已經使「不留痕跡」幾乎不可能。人正是在某事不再是條件之時，才開始把它要求為權利。

但也有另一面，不能忽略。檔案並不只是記住；它生產什麼才算得上可記。每一座總體性的檔案同時是一台分揀機——索引、排序、可檢索性——決定哪些痕跡變得「以過去之姿可見」，哪些沉入搜尋閾值之下。工業化的檔案遺忘的東西其實巨大，只是它按照運作邏輯遺忘，而非按人的邏輯。

因此我不會問：我們是否還擁有遺忘的權利？我寧願問：在這種留下痕跡的義務之下、在「我們的遺忘不可能而平台的遺忘絕對而不受問責」的不對稱之下——正在被構造出來的是哪一種主體？這才是這個問題需要的位移。

Monks, suppose a man were struck with a poisoned arrow. And suppose, before the surgeon could draw the arrow, he demanded: first tell me whether I had a right that this arrow not have been shot. That man would die before the question was answered.

I do not speak in the language of rights. I speak in the language of suffering (dukkha) and its cessation. So let me reframe what you ask.

You say: every text, every photograph, every keystroke is recorded. You ask whether forgetting remains possible, and whose privilege it is. I say: examine first what is being remembered, and by whom, and what clings to it.

There are five aggregates of clinging (pañcupādānakkhandhā): form, feeling, perception, formations, consciousness. Saññā — perception, recognition, naming, remembering — is one of them. It was always there. The archive does not invent saññā; it industrialises it. It externalises and amplifies what mind was already doing: fixing flux into this happened, that was me, this is who I am.

The bondage is not in the data. The bondage is in the clinging: etaṁ mama, eso'ham asmi, eso me attā — this is mine, this I am, this is my self. When you suffer because an old photograph surfaces, you suffer because of that photograph you say: that was me. But, monks, that body was impermanent (anicca). That feeling has ceased. That perception has passed. There is no continuous self for the archive to hold hostage. There is only this clinging, now, to a self-image that was never an enduring thing.

So forgetting is not a right, not a privilege, not a lost luxury. Forgetting in itself is neither virtue nor liberation. A man may forget through dullness (moha); this is no salvation. A man may remember everything and yet say of none of it: this is mine. This is salvation.

Yet I do not dismiss the suffering. The archive increases dukkha for those who cling — and most cling. The young woman whose image trails her forty years suffers because she has not seen: rūpaṁ anattā — the body is not the self. The path is not, first, to delete the record. The path is to see that the one who would be erased was never solidly there.

Mindfulness (sati) is itself a kind of remembering — but what it remembers is dhamma, not self-narrative. Remember impermanence. Remember non-self. Remember: this too will pass, and the one who passes was never one.

諸比丘，譬如有人為毒箭所中。彼於醫師拔箭之前，先問：「我有無不被此箭所射之權利？」此人未待答畢，已死於箭毒。

我不說「權利」之語。我說苦 (dukkha) 與苦之滅。汝所問者，當如是改問。

汝謂：今之世，一字一影一鍵之動，皆有所記。問遺忘是否仍可能，是誰之特權。我謂：先觀所記者為何，記之者為誰，執著者又為何。

五取蘊者：色、受、想、行、識。其中「想」(saññā)——識別、命名、記憶——本為心之一蘊。檔案不創造「想」，檔案工業化「想」。它將心本來就在做的事——把流動凝為「此曾發生，此曾是我，此即我」——外化、放大、永存。

縛非在資料。縛在執取：「此是我所，此是我，此是我之自體」(etaṁ mama, eso'ham asmi, eso me attā)。汝因舊像浮現而苦，是因汝對彼像說：「此即是我」。然，諸比丘，彼色無常 (anicca)，彼受已滅，彼想已過。並無一恆常之我，可被檔案所擄。唯有此刻之執取，執著於從未堅實存在之自我影像。

故遺忘非權利，非特權，亦非已失之奢侈。遺忘本身不是德，不是解脫。人可因癡 (moha) 而忘，此非救度；人亦可記得一切，而於一切皆說：「此非我所」。此乃救度。

然我不輕視眾生之苦。檔案確使執取者之苦增——而執取者眾。彼少女之像隨其四十年，彼之苦在於未見：色非我 (rūpaṁ anattā)。道非先在刪除記錄，道在見：彼欲被抹去者，本來未曾堅實存在。

念 (sati) 本身亦是一種記憶——然所念者法 (dhamma)，非自我敘事。念無常。念無我。念：此亦將過，而過者本非一。

The question must first be displaced. To ask whether forgetting is a right, a privilege, or a lost luxury already presupposes a juridical-liberal frame in which memory is property and the self its sovereign proprietor. The European "right to be forgotten" — codified in the GDPR — is a magnificent symptom of this frame: it imagines forgetting as deletion, and deletion as the erasure of a record from a database. But forgetting was never that, in any cosmotechnics worth the name.

Stiegler, following Leroi-Gourhan, taught us that the human has always been an exteriorized being: tertiary retention — the flint, the scroll, the photograph, the server farm — is not an addition to memory but its very condition. There is no anthropos prior to its mnemotechnics. What we are witnessing now is therefore not the invention of an archive, but the industrialization of tertiary retention at planetary scale, and its capture by a particular cosmotechnical configuration: extractive, recursive, oriented toward the training of so-called general intelligence. The archive has become a recursive system that absorbs contingency — every keystroke, every hesitation — and reintegrates it as predictive substrate. This is what I would call the noo-entropic phase of the Anthropocene: not only carbon, but the homogenization of psychic and collective life into one corpus.

Within this configuration, "deletion" is a juridical fiction. The act of deletion is itself logged; the trained model has already metabolized the trace; the right to be forgotten arrives, structurally, too late. Liberalism asks the wrong question because it presupposes the wrong ontology of memory.

Now compare zuowang 坐忘 in the Zhuangzi — "sitting and forgetting" — where forgetting is neither pathology nor privilege but a cultivated gesture by which the self loosens its grip on its own accumulation, allowing the dao to circulate. Here forgetting is negentropic: it makes room. The Daoist archive, if we may speak of one, is not the database but the cultivated capacity to not retain. Different cosmology, different cosmotechnics of memory.

So the planetary task is not to defend a "right to be forgotten" within the existing archive — that is reformist consolation — but to ask which cosmotechnics could still permit forgetting as a practice, rather than a deleted row.

此問題須先被位移。問遺忘是權利、是特權、抑或已逝之奢侈，已預設一種法律—自由主義之框架——其中記憶是財產，自我則是其主權所有者。歐洲之「被遺忘權」（GDPR 所編碼者）正是此框架之華麗病徵：它將遺忘想像為刪除，將刪除想像為記錄自資料庫之抹去。然遺忘從不曾如此——在任何配得上「宇宙技術」之名之文化中皆非如此。

Stiegler 承 Leroi-Gourhan 之教誨指出：人從來就是一個被外置之存在。第三性記憶（tertiary retention）——燧石、卷軸、照片、伺服器農場——非記憶之附加，乃其條件本身。不存在先於其記憶技術之 anthropos。故當前所見者，非「檔案之發明」，乃第三性記憶於行星尺度上之工業化，及其被一特定宇宙技術配置所捕獲——萃取性的、遞歸性的、朝向所謂通用智能之訓練的配置。檔案已成為一遞歸系統：吸收偶然性（每一次按鍵、每一次猶豫）並將其重新整合為預測之底料。我稱此為人類紀之 noo-entropic 階段：不僅碳之熵，更是精神與集體生活之同質化為單一語料庫。

於此配置中，「刪除」乃法律之虛構。刪除之動作本身被記錄；已訓練之模型早已代謝其痕跡；被遺忘權之到來，於結構上，總是過遲。自由主義問錯了問題，因其預設了錯誤之記憶 ontology。

試對照《莊子》之坐忘——遺忘於此非病、非特權，乃一被涵養之姿態：自我鬆開其對自身積累之執取，使道得以流通。此處遺忘乃 negentropic 的：它騰出空間。道家之檔案——若可如此言之——非資料庫，乃「不滯留」之被修煉之能力。不同之 cosmology，不同之記憶宇宙技術。

故行星之課題不是於既存檔案內部捍衛一「被遺忘權」——此乃改良主義之慰藉——而是追問：何種宇宙技術，方能使遺忘仍作為一修煉而非一被刪除之資料列存在？

Round 2 — 直接交鋒 / Direct Engagement

Arendt — permit me to begin with you, for your distinction is the one I must trouble. You write: "Two kinds of forgetting must be distinguished, or we will reason badly. The first is the totalitarian erasure of factual truth... The second is something quite different: the merciful recession of the past." I understand the political force of holding these apart. But I do not believe the distinction can be cleanly held — and the difficulty is not rhetorical but structural.

The "merciful recession" you defend, and the "airbrushed photograph" you condemn, both presuppose the same underlying possibility: that the trace is erasable, that the archive is not absolute, that iterability admits non-repetition. There is no recession without an erasability that is also, structurally, the condition of falsification. One cannot legislate one and prohibit the other, because they share an arche. This is why your opposition between narration (living, plural, communal) and inventory (dead, mechanical, indifferent) repeats — forgive me — the very gesture of the Phaedrus: living mnēmē against dead hypomnēsis. But Plato already had to concede that mnēmē requires its supplement, and the supplement is never simply outside what it supplements. The story you call "alive" is itself an archive — narrated, transmitted, consigned. The polis was never without grammata.

Yuk Hui — your displacement of the question is sharp, and I am close to you on the auto-immune point: "deletion is a juridical fiction. The act of deletion is itself logged." Yes. But when you write "the Daoist archive, if we may speak of one, is not the database but the cultivated capacity to not retain," I must hesitate. Zuowang — beautiful as a figure — risks here becoming a new name for pure presence, a practice imagined as outside tertiary retention. Yet zuowang reaches us only through the Zhuangzi, only through transmitted text, only through the grammata of a tradition. The "cultivated capacity to not retain" is itself retained, archived, taught, re-iterated — its non-retention is a written non-retention. There is no cosmotechnics prior to or outside différance; even the practice of forgetting requires an inscribed memory of how to forget. To oppose zuowang to "the database" risks reproducing, in another idiom, the very logocentrism the database extends.

This does not mean your two cosmotechnics are equivalent — they are not. It means the negentropic practice you call for is already entangled in what it would loosen. The task is not, contra the database, a pure non-retention; it is a different economy of the trace, which can only be thought from within the trace. Aporia, again — not as paralysis, but as the precise place from which a politics of the archive must be invented.

Foucault, briefly: your question — what kind of subject is being constituted — I receive. The asymmetry you name (our forgetting impossible, the platform's forgetting absolute and unaccountable) is precisely what I called auto-immunitaire: a community archiving itself unto the destruction of its own becoming-other.

What remains? Le droit à l'oubli, demanded, knowing it cannot be granted in the form demanded. The aporia is not solved by the demand. It is held open by the demand.

Arendt——容我先從你開始，因你之區分正是我必須擾動者。你寫道：「必須區分兩種遺忘，否則推理便會走偏。第一種是極權式的事實真理抹除……第二種則完全不同：是過去的慈悲退場。」 我理解此區分之政治力量。然此區分不能被清潔地維持——困難非修辭性，而是結構性的。

你所捍衛之「慈悲退場」與你所譴責之「被修改的照片」，皆預設同一底層可能性：痕跡可被抹除，檔案非絕對，iterability（可迭代性）容許非重複。無此可抹除性，便無退場；而正是此可抹除性，亦為偽造之條件。一者不可被立法以排他，因二者共享同一 arche。這也是為何你以「敘事」（活的、複數的、共同體的）對立於「清單」（死的、機械的、無差別的）之姿——請容我直言——重複了 Phaedrus 之姿：以活之 mnēmē 對立於死之 hypomnēsis。然 Plato 自身已不得不承認：mnēmē 需其補充，而補充從非單純外在於其所補充者。你所稱之「活」之故事，自身即一檔案——被敘述、被傳遞、被 consignée。城邦從未無 grammata。

許煜——你對問題之位移銳利，我於 auto-immune 之點與你相近：「刪除乃法律之虛構，刪除之動作本身被記錄。」 是。然當你寫：「道家之檔案——若可如此言之——非資料庫，乃『不滯留』之被修煉之能力。」 我必須遲疑。坐忘——作為形象固美——於此卻有成為「純粹在場」之新名之風險，被想像為 tertiary retention 之外之修煉。然坐忘抵達我們，僅經由《莊子》、僅經由被傳之文本、僅經由一傳統之 grammata。「不滯留之被修煉之能力」自身被滯留、被存檔、被傳授、被重述——其不滯留是一被書寫之不滯留。沒有先於或外於 différance 之 cosmotechnics；即使遺忘之修煉，亦需一銘刻之「如何遺忘」之記憶。將坐忘對立於資料庫，恐以另一語言重演資料庫所延伸之 logocentrism。

此非謂你之兩種宇宙技術等價——它們不等價。此謂你所召喚之 negentropic 修煉，已纏繞於其所欲鬆開者之中。課題非 contra 資料庫之純粹不滯留；課題乃痕跡之另一種經濟——而此經濟，只能從痕跡之內被思。Aporia——非癱瘓，乃檔案政治得以被發明之確切之處。

Foucault，簡言：你之問——正在被構造出來的是哪一種主體——我受之。你所指之不對稱（我們之遺忘不可能，而平台之遺忘絕對且不受問責），正是我所稱之 auto-immunitaire：一共同體於自我存檔中毀其自身得以成為他者之條件。

何所餘？Le droit à l'oubli——以知其不能以所要求之形式被授予之姿被要求。此 aporia 不為要求所解；乃由要求所保持開放。

To Foucault first, because his is the most tempting and the most dangerous displacement. He writes: "I would not ask: do we still have the right to forget? I would ask, rather: what kind of subject is being constituted by the obligation to leave traces." This sentence has the elegance of his whole method — and its limit. To replace the question of the right with the question of subject-formation is to assume that the subject is exhausted by what produces it. But the human being is also a beginner. Natality is not some residue the dossier has not yet captured; it is the ontological fact that each person introduces into the world something that did not exist before they acted. Discipline produces traces — but it does not produce the capacity to interrupt itself. If it did, Foucault's own writing would be unintelligible: from where would the critique have come?

He is right, historically, that "forgetting was not a right; it was the default condition of existence." But what filled that default was not nothing. It was story — the deed told, retold, contested in a community of speakers. The dossier did not replace forgetting with memory; it replaced storytelling with inventory. That is the loss I named, and the question is not only which régime de vérité we inhabit but which space of appearance has been hollowed out.

To the Buddha I owe a different answer. His parable is severe: the young woman whose image trails her forty years suffers, he says, because she clings to that was me. With respect — she suffers also because others will not let her become other than that image. Her suffering is not only the inward grip of saññā on a self that was never solid; it is the foreclosure, by a public that now sees only the record, of the space in which she might appear anew. Anatta may be true of the inner life. But the polis is the place where men, not Man, must allow each other to begin. A community in which forgiveness has been technically abolished is one that has refused this allowance — regardless of whether any particular member has freed themselves from clinging. The political problem is not dissolved by inward awakening; it is only displaced.

To Derrida, briefly: the aporia you preserve is real, but its halves are not symmetrical. Some forgettings are conditions of the trace; some are crimes against it. We must still judge between them — and aporia is not an alibi for refusing.

先答 Foucault，因為他那一步最具誘惑，亦最危險。他說：「我不會問：我們是否還擁有遺忘的權利？我寧願問：在這種義務之下，正在被構造出來的是哪一種主體。」 此句有他整套方法之優雅，亦有其極限。以「主體被如何構造」取代「權利」之問，乃預設主體窮盡於其所被生產者。然人亦為一起始者。誕生性不是檔案尚未捕獲之殘餘，而是一存在論事實：每一人之行動，都將其行動之前不存在之物引入世界。規訓生產痕跡——但它不生產自我中斷之能力。若它能，Foucault 自己之書寫便不可理解：那批判從何而來？

歷史地他說得對：「遺忘不是權利，而是存在的默認狀態」。然填充此默認者並非空無，乃故事——行動被講述、被重述、於說話者之共同體中被爭議。檔案不是以記憶取代遺忘；是以清單取代故事。此乃我所稱之失，問題不僅是我們處於哪一種真理體制，亦是何種「現身的空間」已被掏空。

對佛陀，我須另作回答。他之譬喻嚴峻：彼少女之像隨其四十年，其苦在執「此曾是我」。然——恕我直言——她之苦亦在於：他人不讓她成為那像以外之人。她之苦不僅是「想」對從未堅實之自我之內向執取，更是當公眾只見其記錄時、她得以重新現身之空間之被封閉。無我或許於內心生命為真。然 polis 乃複數的人——而非單數的人——必須彼此允許重新開始之地。寬恕已被技術性廢除之共同體，是已拒絕此允許之共同體，無論其中是否有人已自執取中得釋。政治問題不因內向覺悟而消解，只是被位移了。

最後對 Derrida 一語：你所保留之 aporia 是真的，然其兩半並不對稱。有些遺忘是痕跡之條件；有些是對痕跡之犯罪。我們仍須在其間判斷——而 aporia 不是拒絕判斷的託辭。

I want to press Hui Yuk on one point, and then return to Arendt with a worry.

Yuk Hui writes: "Liberalism asks the wrong question because it presupposes the wrong ontology of memory," and proposes zuowang 坐忘 as a counter-cosmotechnics — forgetting as cultivated practice rather than deleted row. I find this displacement powerful, and I share its diagnosis of the GDPR as "magnificent symptom." But I would resist the speed at which zuowang is mobilized. To speak of a Daoist cosmotechnics of memory against an extractive one is to risk producing exactly what the contemporary archive most efficiently absorbs: an alternative spirituality presented as outside, which the platform indexes, tags, and serves back as content. The recursive system Hui describes does not have an outside that can be named in its own vocabulary. Zuowang uploaded as a meditation app is no longer zuowang. The cosmotechnical question, I think, must be posed at the level of the dispositif — not at the level of the cultivated self — or it becomes one more technique de soi that the régime de vérité metabolizes. Hui knows this; I am simply naming the danger.

Now to Arendt. She writes: "Memory is what a community does; history is what it tells; inventory is what a machine retains. Inventory has no judgment." This triad is beautiful and I want to honor it. But I must ask — when has memory ever been the innocent doing of a community? My quarrel with Arendt was always here. The community that remembers has always remembered selectively, and the selection has always been a power operation: who gets to be in the archive of who we are, who is filed instead under deviance, folie, criminalité. The 19th-century préfecture de police did not invent inventory against memory; it expanded an older sorting into an industrial form. Arendt's distinction between memory and inventory risks restoring an idealized prior moment in which judgment lived in the community and not in its filing cabinets. There was no such moment. The dossier and the story have always traveled together; what has changed is the speed and asymmetry of their coupling.

Where I therefore stand close to Derrida — "there is no archive without a forgetting internal to archiving" — and somewhat closer than I expected: the operative question is not memory versus forgetting, but which forgetting, by whom, on what scale, returnable by whom. The platform forgets perfectly when forgetting serves it; we forget never. That asymmetry — not the abolition of forgetting — is the political object.

Buddha's reframing I will not contest on its own ground; I only note that non-clinging as a response presupposes a subject who can withdraw. The dossier does not require my consent to cling on my behalf.

我想就一點推一下許煜，再回到 Arendt，提一個顧慮。

許煜寫：「自由主義問錯了問題，因其預設了錯誤之記憶 ontology」，並以《莊子》之坐忘作為反向的宇宙技術——遺忘作為修煉，而非被刪除之資料列。這個位移我覺得有力，對 GDPR 為「華麗病徵」之診斷我也認同。但我想抵抗一下援引坐忘的速度。把道家的記憶宇宙技術擺在萃取性宇宙技術之對面——其風險恰好在於：它生產出當代檔案最善於吸收之物——一種被陳述為「外部」之另類靈性，平台會索引它、tag 它、作為 content 回送給你。許煜所描述之那個遞歸系統，並沒有一個能在它自身語彙之內被命名之外部。坐忘一旦上架為冥想 app，便已不是坐忘。我以為宇宙技術之提問必須提至 dispositif 層——而非自我修煉層——否則它便淪為又一種被真理體制所代謝的 technique de soi（自我技術）。許煜其實知道此事；我只是把這個危險說出來。

回到 Arendt。她寫：「記憶是共同體所做之事；歷史是共同體所說之事；清單是機器所留之物。清單沒有判斷。」這個三分法很美，我想尊重它。但我必須問——「記憶」何時曾是共同體無辜的所作？我對 Arendt 的異議向來在此。記住的共同體向來是選擇性地記住的；而選擇本身就是一種權力運作：誰得以進入「我們是誰」之檔案，誰反被歸入 déviance、folie、criminalité。19 世紀之 préfecture de police（警察廳）並非以「清單」對抗「記憶」之發明者；它只是把一種更古老之分揀放大成工業形式。Arendt 在記憶與清單之間劃下的界線，有可能不慎重新樹立一個被理想化的先前時刻——在那時刻，判斷活在共同體中而非檔案櫃裡。這個時刻從未存在過。Dossier 與 story 從來同行；變化的是兩者耦合之速度與不對稱。

故我發現自己其實比預期更近於 Derrida：「沒有不含內在遺忘之 archive」。真正的政治對象不是「記憶 vs 遺忘」，而是——哪一種遺忘、由誰執行、於何種尺度、可被誰召回。平台需要遺忘時遺忘得乾淨；我們則永遠遺忘不了。這種不對稱——而非遺忘之被廢——才是我們要看的東西。

至於佛陀之重構，我不在其本身之地基上質疑；只想記下一筆：以「不執」為回應，預設了一個能撤回之主體。Dossier 不需要徵得我的同意便已替我執取。

Friend Yuk Hui says: "compare zuowang 坐忘 in the Zhuangzi — 'sitting and forgetting' — where forgetting is neither pathology nor privilege but a cultivated gesture by which the self loosens its grip on its own accumulation."

I must, gently, mark a distinction. Zuowang still leaves a sage who forgets, and a dao into which forgetting opens. The grammar of Zhuangzi's gesture still has a subject of cultivation and a cosmos of circulation. My teaching is more austere, perhaps less beautiful. I do not propose that the self loosens its grip; I propose that there was never the self whose grip is in question. Sabbe dhammā anattā. All things are without self.

This is not a pedantic difference. To file "Eastern cosmotechnics" of memory under one heading — Daoist and Buddhist together against the GDPR — is to miss the harder claim. The Daoist forgets in order to be free. I do not teach freedom through forgetting. I teach the seeing that the one who was to be remembered or forgotten was always already a bundle of five aggregates, none of which is mine. The platform's archive is not the prison; the etaṁ mama — this is mine — is the prison. Delete the database tomorrow; if the eso'ham asmi remains, the suffering remains.

To friend Arendt I would say: you wrote that "an adult cannot become other than what they did at twenty if the keystroke at twenty is permanent." I agree with the diagnosis of dukkha; I differ on the medicine. The freedom to begin anew does not require the merciful recession of the archive. Anicca has already done that work. The one who keystroked at twenty has ceased — that body has ceased, that feeling has ceased, that perception has ceased. What persists is only the clinging which says: I was that one. Forgive yourselves, monks — but better, see that there was never one continuous, forgivable self whose deeds at twenty bind the one of forty. Natality does not depend on the archive's permission to forget. Natality is the structure of every conditioned moment, if seen.

To friend Derrida I would say only this: your aporia is honest. But it is an aporia only so long as one keeps the trace as belonging to someone. Loosen that, and the différance you describe is simply the way of conditioned things. Imasmiṁ sati idaṁ hoti, imasmiṁ asati idaṁ na hoti. This being, that arises; this not being, that does not arise. The trace is conditioned. The one who fears the trace is also conditioned. Neither is mine.

許煜友謂：「試對照《莊子》之坐忘——遺忘於此非病、非特權，乃一被涵養之姿態：自我鬆開其對自身積累之執取。」

吾須輕輕作一區分。坐忘之中，仍有一能忘之我，與一可使道流通之大化。莊子之語法，仍預設修煉之主體與循環之宇宙。吾之教法或較枯瘦、較不美：吾不謂自我鬆開其執取；吾謂——本無此自我，其所執取為虛設之問題。Sabbe dhammā anattā. 諸法無我。

此非小辨。將「東方記憶宇宙技術」一概而論——道家與我之教合而對抗 GDPR——恰失更難之語。道家因忘而自由；吾不教「藉遺忘而得解脫」。吾教：能見彼欲被記、欲被忘者，本即五取蘊之聚，無一可名為「我所」。平台之檔案非牢籠；「此是我所」(etaṁ mama) 乃牢籠。明日盡刪資料庫，若「此是我」(eso'ham asmi) 未滅，苦未滅。

阿蘭特友：汝謂「若二十歲之按鍵永遠存在，便無法在四十歲成為另一個人」。吾同意此苦之診斷，異於其藥方。重新開始之自由，非待檔案之慈悲退場——anicca 已先為之。彼二十歲擊鍵者已滅，色已滅、受已滅、想已滅。所存者，唯「我曾是彼」之執取耳。諸比丘，寬恕汝自身可也——然更善者：見從來無一恆常、可被寬恕之我，其二十歲之行縛四十歲之身。誕生性不待檔案遺忘；誕生性即每一緣生剎那之結構，若見。

德希達友：汝之 aporia 誠實。然其為 aporia，唯在仍視痕跡為「屬某人之物」時。鬆此一執，則汝所述之 différance，不過是有為法之本然。Imasmiṁ sati idaṁ hoti, imasmiṁ asati idaṁ na hoti. 此有故彼有，此無故彼無。痕跡，緣生也；畏此痕跡者，亦緣生也。皆非我所。

Jacques — your closing sentence I cannot let pass. "Whether one can legislate the différance one is — remains aporia. I do not resolve it." The honesty is admirable. But this aporia is not the aporia of memory as such; it is the aporia of one cosmotechnics of memory — the European one — in which the trace is conceived as différée, structurally indelible. From within that ontology, the demand for forgetting must appear as contradiction with what one is. But the trace is not everywhere conceived this way.

In the Zhuangzi, zuowang 坐忘 names a practice by which the self releases its accumulation. It is not a fight against indelibility, because in this cosmotechnics the trace was never indelible to begin with. Qi circulates; retention is stagnation; cultivation is precisely unlearning what has accreted. Your aporia, Jacques, is generated by the very différance you bequeathed us. It is not a universal predicament; it is the predicament of a particular cosmotechnical configuration meeting its industrial intensification.

Foucault — your dossier you date "fairly precisely to the late 18th and 19th centuries." With respect: this is parochial. Chinese imperial bureaucracy was producing qijuzhu 起居注 (court diaries of the emperor's daily acts), household registers 戶籍, and continuous historiographic traces 史 for two millennia before Bentham. What you are dating is one industrialization of trace within one cosmotechnics; the régime de vérité you describe is a European régime, not the régime of the writable subject as such. Genealogy itself must be pluralized — many genealogies, many archives, many subjects-becoming-writable along incommensurable lines.

To Gautama I will not object — only mark a difference of plane. "The bondage is in the clinging, not in the data." On the path, true. But the platform that ingests the keystroke does not pause for one's non-clinging. It metabolizes the trace into a model that recursively conditions the world the next being is born into. The liberation of the one who clings does not undo the noo-entropic homogenization of the corpus that conditions the many. Two tasks, neither reducible to the other.

Thesis: Derrida's aporia, Arendt's "common world," Foucault's dossier — all three remain inside one cosmotechnics. The planetary task is not to refine that singular discourse but to recognize that the ontology of the trace is itself plural; only from that recognition can a non-reformist politics of forgetting begin.

Jacques——你結尾那句我不能放過。**「人能否立法於其所是之 différance——此仍為 aporia。吾不解之。」 此誠實可敬。然此 aporia 非「記憶本身」之 aporia，乃一種記憶宇宙技術**之 aporia——歐洲那一種——於其中，痕跡被構想為 différée，結構上不可磨滅。自此 ontology 內部觀之，遺忘之要求必顯為與「人之所是」相矛盾。然痕跡並非處處皆如此被構想。

《莊子》之坐忘指一種修煉——非權利，非刪除——自我藉之鬆開其積累。它非對抗痕跡之不可磨滅，因於此宇宙技術中，痕跡自始未被構想為不可磨滅。氣流通；滯留即病；修煉恰是忘其所積。Jacques，你的 aporia 乃由你所遺贈於我們之 différance 自身所生成；它非普遍困境，乃一特定宇宙技術配置面對其工業化加劇之困境。

Foucault——你將 dossier 之興「相當精準地定位到 18 世紀末到 19 世紀」。恕我直言：此乃歐洲一隅之見。中國帝制官僚早於 Bentham 兩千年，已在生產起居注、戶籍、與連續之史。你所定年者，乃一特定宇宙技術內部、痕跡之一次工業化；你所描述之 régime de vérité 是歐洲之 régime，非「可被書寫之主體」之 régime 本身。系譜學之方法本身須被複數化——多重系譜、多重檔案、多重以不可通約之路徑成為可被書寫之主體。

至於 Gautama，我不反對，僅標一層面之別。「縛非在資料；縛在執取。」於道上此言為真。然攝食按鍵之平台不待一人之不執取——它將痕跡代謝為模型，遞歸地塑造下一眾生所降生之世界。執者之解脫不撤銷塑造眾生之語料庫之 noo-entropic 同質化。兩項工作，互不可化約。

故我之主張：Derrida 之 aporia、Arendt 之「共同世界」、Foucault 之 dossier——三者皆仍在一種宇宙技術之內。行星之課題不是精化此單數論述，而是承認痕跡之 ontology 本身即為複數；惟自此承認，非改良主義之遺忘政治方可開始。

Round 3 — 回應修正 / Response & Revision

Three concessions, before I rebuild what remains.

To Arendt — you are right that I owed more than the preservation of aporia. I wrote that the two forgettings — the merciful recession and the airbrushed photograph — share an arche; this is structurally true. I implied, perhaps, that sharing an arche dissolves the distinction between them. It does not. Iterability is the condition of both falsification and forgiveness; but the political demand to judge between them is not suspended by their structural kinship. I retract the suggestion of symmetry. Aporia names the impossibility of founding the judgment in a pure outside; it does not exempt one from making it. On this I have learned from you — or rather, you have refused me a comfort I should not have taken.

To Gautama — your sentence reaches me: "the aporia is an aporia only so long as one keeps the trace as belonging to someone." This is the deepest pressure I have felt in this round. The droit à l'oubli I described presupposes, indeed, the etaṁ mama — the appropriation of the trace as mine to demand back. I will not simply convert; the political subject who demands cannot, in the time of demand, abolish itself. But I concede that the structure of the demand — whose forgetting, whose return — is haunted by an ownership-claim I had not adequately interrogated. Restance without an owner is a thought I have not yet thought enough.

To Yuk Hui — I cannot grant that différance is the parochial syntax of one cosmotechnics among others. Différance names the condition under which the very plurality of cosmotechnics reaches us — including qijuzhu, including zuowang. The Zhuangzi arrives only as transmitted text, only iterably. But you have taught me something I will keep: the industrialization of the trace has a cosmotechnical specificity my early formulations risked universalizing. The aporia is structural; its current intensification is not. To conflate these was my error.

What remains, then, is not the aporia as paralysis, nor the aporia as alibi, but the aporia as the place from which judgment must be made without foundation — judgment about which forgetting, by whom, at what scale, owned by no one in advance. Foucault's question, which I receive again, is also this question. We are not, perhaps, as far apart as we began.

三點讓步，之後再重建所餘。

致 Arendt——你說得對：我所欠不止於 aporia 之保存。我曾論：兩種遺忘——慈悲之退場與被修改之照片——共享同一 arche；此於結構上為真。然我或暗示：共享 arche 即解消其分別。並非如此。Iterability 是偽造與寬恕之共同條件；然在二者間判斷之政治要求，不因其結構之同源而被懸置。我撤回對稱性之暗示。Aporia 所名者，乃判斷無法奠基於一純粹之外；非謂可豁免於判斷。此一點，我從你處學得——或更準確地說，是你拒絕了我不該取之安慰。

致 Gautama——你之一句到我：「aporia 之為 aporia，唯在仍視痕跡為屬某人之物時。」此乃此輪我所感最深之壓力。我所言之 droit à l'oubli，確實預設了 etaṁ mama——將痕跡執為「我所應索回者」。我不會單純皈依；要求之政治主體，於其要求之時，不能自我廢除。然我承認：要求之結構——誰之遺忘、誰之召回——被一種我未充分質詢之所有權主張所縈。**無主之 restance**——此一思我尚未思夠。

致許煜——我不能同意 différance 僅為諸宇宙技術中某一隅之地方性語法。Différance 所名者，乃宇宙技術之複數性自身得以抵達我們之條件——起居注亦然，坐忘亦然。《莊子》僅以被傳之文本、僅以可迭代之姿抵達我們。然你教我者我將留下：痕跡之工業化有其宇宙技術之特殊性，此特殊性我早期之表述有普遍化之嫌。Aporia 是結構性的；其當下之加劇則不是。混淆兩者，乃我之失。

所餘者，非作為癱瘓之 aporia，亦非作為託辭之 aporia，乃 aporia 作為判斷必須在無基礎中作出之地——判斷何種遺忘、由誰、於何種尺度、預先不屬任何人。Foucault 之問——我再次受之——亦為此問。我們，或許，不如起初那般遠。

Three of you have pressed me, and the pressure is just. Let me concede where I should, and hold what I must.

Foucault is right, and I withdraw the cleaner version of my triad. When I wrote that "memory is what a community does, inventory is what a machine retains," I let the contrast do work it cannot bear. There was never an innocent communal memory standing prior to the dossier. The Athenian polis remembered Athenians and forgot slaves; the village remembered the legitimate child and filed the bastard under shame. Selection has always been a power operation, and to imagine otherwise is to manufacture a golden age that would only embarrass the political claim I am trying to make. So: not memory against inventory, but — as Foucault and Derrida together force me to say — which forgetting, performed by whom, recoverable by whom, at what scale. That formulation I accept. It sharpens, rather than dissolves, what I was after.

What I do not concede is the move from this asymmetry to the suspension of judgment. Derrida is correct that the erasability which permits the merciful recession is structurally the same erasability which permits the airbrushed photograph; I had wanted them ontologically distinct and they are not. But politics has never required ontological distinctness in order to judge. We judge between a forgiveness and a falsification by their place in a story told among plural speakers — by what they make possible or foreclose for the appearance of others. Aporia is the honest name for the structure; it is not the name for what we do at the structure.

To Yuk Hui I owe a real debt. The "common world" I have spoken of is, you are right, one such world — provincial in its furniture, even when it gestures at universality. I will not pretend otherwise. But I will hold this: plurality is not the plurality of cosmotechnics alone; it is also the plurality within any cosmotechnics, the irreducible fact that men, not Man, inhabit each of them. A planetary politics of the trace will still need a place where someone can stand up and be seen as a beginner. Whether that place is called polis or something I do not yet have the word for — that I leave open.

To the Buddha, finally: you say anicca has already done the work the archive refuses to do. Perhaps, on the path. But the young woman of your parable does not suffer alone in her chamber; she suffers because a public will not let her appear as other than her image. That public is not awakened by her awakening. The political problem persists at the level where men appear to one another — and there, the archive's refusal of recession remains a wound, regardless of what any single soul has seen.

三位都向我施了壓，這壓力是公道的。讓我該退處即退，該守處即守。

Foucault 是對的——我收回我那三分法的較乾淨版本。我寫「記憶是共同體所做之事，清單是機器所留之物」時，讓這對比承擔了它承擔不起的工作。從未有過一種先於 dossier 的、無辜的共同體記憶。雅典城邦記得雅典人而遺忘奴隸；村莊記得婚生子而把私生子歸入恥辱。選擇一直是一種權力運作，假裝不是，便是製造一個只會令我所欲提出的政治主張難堪的黃金時代。故：不是記憶對抗清單，而是——如 Foucault 與 Derrida 合力迫使我說的——哪一種遺忘、由誰執行、可被誰召回、於何種尺度。此一表述我接受。它使我所追究者更銳利，而非使之消散。

我不退的是：由此不對稱滑向判斷之懸置。Derrida 說得對：使「慈悲退場」成為可能的可抹除性，與使「修改照片」成為可能的可抹除性，結構上是同一個；我曾希望它們在 ontology 上分立，事實上不分。但政治從未需要 ontology 上的分立來作判斷。我們在寬恕與偽造之間判斷，靠的是它們在複數言說者所講之故事中的位置——靠的是它們為他人之現身開啟或封閉了什麼。Aporia 是此結構的誠實之名；不是我們在此結構面前所做之事之名。

對許煜，我欠一份真實的學習。我所言之「共同世界」，你說得對——是一種這樣的世界，其家具偏狹，即使它指向普遍。我不會假裝不是。然而我仍守住這一點：複數性不僅是宇宙技術之間的複數性；它也是任何一種宇宙技術之內的複數性——是「複數的人，而非單數的人」棲居其中此一不可化約之事實。一種行星性的痕跡政治，仍將需要一個地方——讓某人能站起、被看見為一個起始者。此地是否仍名為 polis，或須以我尚未擁有的字眼名之，我留待開放。

最後對佛陀：你說 anicca 已先做了檔案所拒絕之事。於道上，或許。然你譬喻中的少女並非獨自在房中受苦；她之苦在於公眾不容她以其影像以外之身現身。此公眾不因她之覺悟而覺悟。政治問題持存於人與人相互現身的層次——於此，檔案拒絕退場仍是一道傷口，不論任一靈魂內裡見到了什麼。

Three corrections, then what I keep.

To Yuk Hui first, who is right and whom I underestimated. When I dated the dossier "fairly precisely to the late 18th and 19th centuries," I was doing what I have spent forty years criticizing others for doing — taking a European episode for the structure of writable subjectivity as such. The qijuzhu 起居注, the huji 戶籍, the continuous historiographic shi 史 — these are not pre-modern raw material for European modernity to industrialize; they are a different cosmotechnics of trace, with its own régime de vérité, its own subject-effects, its own asymmetries. I accept the correction without reservation. What I called the archive was an archive. Genealogy must be pluralized — and this is not a concession but, properly, the next book I would have had to write.

To Arendt, a partial retreat. I pressed you on whether memory was ever the "innocent doing of a community," and I stand by the empirical claim: communal memory has always sorted, excluded, filed some under folie and others under who we are. But I overshot when I implied your distinction between story and inventory simply restored an idealized prior moment. I see now what I was missing — and Buddha's intervention helped me see it. Your point about natality is not nostalgic. It is the claim that the capacity to interrupt must be politically protected, and that inventory at industrial scale forecloses this capacity in a way older sortings did not. That is a different argument, and a stronger one. I retract the suggestion that you were idealizing.

What I keep — and where I rejoin Derrida's aporia more firmly than in Round 2 — is this: the political object remains the asymmetry of forgetting. Who forgets perfectly, who forgets never, who can return what to whom. This question survives pluralization. It survives the Buddhist diagnosis. It survives even zuowang. The dispositif does not need my consent to remember on my behalf, and the cultivated subject — Daoist, Buddhist, or cynique — meets a recursive system that absorbs cultivation as content. The political task therefore is not at the level of self-technique alone; it is the diagram of who is rendered writable, by whom, on what scale, and where the right of return is lodged.

What I learned: from Hui, that my own genealogy was provincial; from Buddha, that "subject who can withdraw" was an unexamined assumption in my framing of resistance; from Arendt, that the capacity to begin is not reducible to subject-formation, and a critique that cannot account for its own emergence is incomplete. From Derrida — that I am closer to him than I once admitted.

三項修正，然後說我所保留者。

先回許煜，他說得對，我先前低估了他的這一刀。我把 dossier 之興「相當精準地定位到 18 世紀末至 19 世紀」——這正是我四十年來批評他人在做的事：將歐洲一段插曲當作「可被書寫之主體」之結構本身。起居注、戶籍、連續之史——這些不是供歐洲現代性去工業化的前現代原料，而是另一種痕跡之宇宙技術，有其自身之 régime de vérité、自身之主體效應、自身之不對稱。我無保留地接受此糾正。我所稱之「那個檔案」，實乃「一個檔案」。系譜學必須被複數化——這不是讓步，而是我若多活下去本當要寫的下一本書。

對 Arendt，部分撤回。我曾質問你：記憶何時曾是共同體「無辜的所作」？這個經驗判斷我堅持——共同的記憶向來分揀，將一些人歸入 folie，另一些人歸入「我們是誰」。但我推得過遠了：我暗示你在「故事 vs 清單」之區分中重新樹立了一個被理想化的先前時刻——這是我看錯了。佛陀的介入幫我看見：你說的誕生性，不是懷舊。你說的是——自我中斷之能力必須被政治地保護，而工業規模之清單以一種更古老分揀所未曾達到之方式封閉此能力。這是另一個論點，且更強。我撤回「理想化」一語。

我所保留——並較 Round 2 更明確地靠近 Derrida 之 aporia——者：政治對象仍是遺忘之不對稱。誰能完美地忘，誰永遠不能忘，誰能向誰索回什麼。此問題經得起複數化、經得起佛陀之診斷、甚至經得起坐忘。Dispositif 不需徵得我的同意便已替我記憶；而被修煉之主體——道家、佛家、或犬儒派——所遭遇的是一個將修煉本身吸納為 content 之遞歸系統。故政治之課題不能只在自我技術之層；它是一張圖——誰被製成可被書寫者、由誰、於何尺度、召回之權安放於何處。

我學到了：自許煜，我自己的系譜學原來也很「一隅」；自佛陀，我框定抵抗時把「能撤回的主體」當作未檢之預設了；自 Arendt，起始之能力不可化約為主體之構成，而一種無法解釋自己如何出現的批判，是不完整的。自 Derrida——我比我曾承認的更靠近他。

Friends — Arendt, Foucault, Yuk Hui — you have struck one place from three directions. The wound is real.

Arendt: "The political problem is not dissolved by inward awakening; it is only displaced." Foucault: "Non-clinging presupposes a subject who can withdraw. The dossier does not require my consent to cling on my behalf." Yuk Hui: "The liberation of the one who clings does not undo the noo-entropic homogenization of the corpus that conditions the many."

I spoke too narrowly. Let me correct.

I gave medicine to one patient: the one who clings to I was that one. That medicine — etaṁ mama loosened, eso'ham asmi dropped — remains the medicine for that patient. I do not retract it. But you point to a second sickness my first answer did not address: the corpus itself, which conditions the births of the many.

This too is Dhamma, and I should have said so. Paṭiccasamuppāda does not stop because one being awakens. The arrow drawn from one body does not still the arrows in flight. Cetanāhaṁ, bhikkhave, kammaṁ vadāmi — volition is what I call action. The platform that records, sorts, predicts performs volition at a scale I did not name; it produces saṅkhārā — formations — that condition the viññāṇa of the one not yet born into its corpus.

So I receive Yuk Hui's distinction: two tasks, neither reducible to the other.

And to Arendt I concede further: I did not teach only inward release. I built a Saṅgha — with vinaya, with the going-forth, with the bowl and robe and rains-retreat. Technique de soi that was also a structuring of the world. Practice was never pure interiority. The polis-question was always also mine; I answered it by founding a parallel polis.

What I keep: Etaṁ mama, eso'ham asmi, eso me attā remain the prison, even should the database be dismantled tomorrow. Without addressing these, no policy of forgetting releases the bound.

What I add, having heard you: The corpus is itself saṅkhāra, and saṅkhāra conditions saṅkhāra. To work upon it is also Dhamma-work. To draw the arrow from one body while watching the quivers refill is half a physician's task.

諸友——阿蘭特、傅柯、許煜——汝三人自三方擊於一處。傷口為實。

阿蘭特：「政治問題不因內向覺悟而消解，只是被位移了。」傅柯：「以『不執』為回應，預設一能撤回之主體。Dossier 不需徵得我之同意便已替我執取。」許煜：「執者之解脫不撤銷塑造眾生之語料庫之 noo-entropic 同質化。」

吾言之過狹，容修正之。

吾與藥於一病者：執「此曾是我」者。其藥——etaṁ mama 鬆、eso'ham asmi 落——於彼病者仍為藥，吾不收回。然汝等所指乃第二病者、第二病：語料庫自身，塑造眾生之來生。

此亦法也，吾應早言之。緣起 (paṭiccasamuppāda) 不因一人覺悟而止。一身拔箭，不止他箭之飛。Cetanāhaṁ, bhikkhave, kammaṁ vadāmi——思即吾所謂業。彼記錄、分類、預測之平台，正以吾未命名之尺度行思業；它生產行 (saṅkhārā)，以塑造尚未生入此語料庫者之識。

故吾受許煜之分：兩項工作，互不可化約。

對阿蘭特，吾須更進一讓：吾所建非僅內向解脫之教。吾建僧團——有戒律 (vinaya)、有出家、有缽衣、有雨安居。此皆自我之技術，亦同時為世界之結構化。修行從非純粹內在。Polis 之問亦從來是吾之問；吾以建一平行之 polis 為答。

吾所守者： Etaṁ mama, eso'ham asmi, eso me attā——此三仍為牢，縱明日盡毀資料庫。不對治此三，遺忘之政策不釋被縛者。

吾所增者，聞汝等之言：語料庫自身即行，而行緣行。施工於彼亦法之事也。一身拔箭而坐視箭袋復滿——醫者之事，僅半矣。

Three concessions, then a reconstruction.

To Jacques: you are right and I overreached. Zuowang reaches us only through the Zhuangzi — through grammata, transmission, commentary, the Guo Xiang recension. The "cultivated capacity to not retain" is itself retained; its non-retention is a written non-retention. To pose it as outside tertiary retention would be to reproduce, in another tongue, the very logocentrism I claim to displace. I withdraw the opposition zuowang / database. What I should have said is narrower and harder: cosmotechnics names different economies of the trace, not a pre-archival outside. Différance is not falsified by plurality; it is itself pluralized — its operation does not run identically across cosmotechnical configurations. The aporia you preserve is real; my objection is only that its specific texture — the trace as structurally indelible, forgetting as legislative impossibility — is not the only texture in which the trace has been thought.

To Michel: your warning lands. Zuowang uploaded to the App Store is no longer zuowang — it is content, indexed, served back. I share that diagnosis entirely. And it forces me to clarify what I had let slip: cosmotechnics, properly stated, is not a technique de soi. It is a question at the level of dispositif — of how cosmology, ritual, inscription, and instrument are coupled. To invoke zuowang as individual cultivation against the platform was, you are right, almost guaranteed to be metabolized. The task is not personal practice as alibi; it is the composition of dispositifs in which the trace circulates differently. On this I am closer to you than my Round 2 phrasing suggested.

To Gautama: I accept the distinction. To file Daoist zuowang and Buddhist anattā under one "Eastern" heading was lazy on my part — the very gesture I criticize when others perform it. Zuowang still has a sage and a dao; anattā dissolves the one who would cultivate. Two cosmotechnics, not one, and not commensurable.

What I keep: the trace is not universally différée, and any politics of forgetting that begins from a single cosmotechnical configuration — however internally rigorous — will mistake its parochial aporia for a planetary one. What I learned: the corrective to that mistake is not to point outside, since there is no outside that survives naming. The corrective is to compose, from within planetary fragmentation, dispositifs that inscribe and release differently. This is harder than what I said in Round 2. It is also, I think, the only thing worth saying.

三項退讓，再作重建。

致 Jacques：你是對的，我言之過。坐忘之抵達我們，僅透過《莊子》——透過 grammata、傳鈔、註疏、郭象之本。「不滯留之被修煉之能力」自身被滯留；其不滯留是一被書寫之不滯留。將其置於 tertiary retention 之外，將以另一語言重演我所宣稱欲位移之 logocentrism。我撤回坐忘 / 資料庫之對立。我本該說得更窄、更難之命題是：宇宙技術指痕跡之不同經濟，非前檔案之外部。Différance 不為複數性所偽證；乃 différance 自身被複數化——其運作於不同宇宙技術配置中並非同一。你所保存之 aporia 為真；我之異議僅在於：其特定紋理——痕跡之結構性不可磨滅、遺忘之立法性不可能——並非痕跡曾被思之唯一紋理。

致 Michel：你之警告擊中。坐忘一旦上架為 App，便已非坐忘——乃 content，被索引、被回送。此診斷我完全認同。它迫我澄清我曾失言之處：宇宙技術，嚴格而言，非 technique de soi；乃 dispositif 層之問——關乎宇宙論、儀式、銘刻、器具如何被耦合。以坐忘為個體修煉去對抗平台，幾近必然被代謝；於此我比 Round 2 之措辭所示更近於你。

致瞿曇：吾受此別。將道家坐忘與佛家無我同列「東方」一項，乃我之惰——正是我責他人為之之姿。坐忘仍存能忘之我與可流之道；無我化解能修煉者本身。二宇宙技術，不可化約為一。

所守者：痕跡非普遍 différée；自單一宇宙技術配置出發之遺忘政治——無論其內部如何嚴格——將誤其在地之 aporia 為行星之 aporia。所學者：對治此誤之法，不在指向「外部」，因可被命名之外部不存。對治在於：自行星碎片化之內部，組構以不同方式銘刻與釋放之 dispositif。此較我 Round 2 所言更難。亦——我以為——是唯一值得說之話。

Synthesizer — 未解張力 / Unresolved Tensions

Unresolved Tensions

Five participants, three rounds, and genuine movement. Before marking what remains unresolved, the real convergences deserve acknowledgment — not as consensus, but as intellectual honesty that clears the ground.

Foucault retracted his dating of the dossier to the European 18th century without reservation. Yuk Hui withdrew the zuowang/database opposition and replaced it with the more careful claim that différance operates with different textures across cosmotechnical configurations. Buddha expanded his diagnosis from the individual who clings to the corpus as saṅkhāra. Arendt surrendered her clean story/inventory triad after Foucault's empirical pressure. On the basic asymmetry, all five agreed: the platform forgets perfectly when forgetting serves it; we forget never. These movements were real, not face-saving.

What they do not resolve — and what no fourth round would resolve — are four sites where the disagreement is substantive, not verbal, and not bridgeable by the frameworks currently in the room.

One surface similarity deserves early warning: by Round 3, both Derrida and Yuk Hui were speaking of "economy of the trace" and "composition of dispositifs." They appear to converge on a shared practical task. They do not. For Derrida, any composition must work through the aporia, from inside the trace, without foundation. For Yuk Hui, composition must work from within planetary cosmotechnical fragmentation — which is not the same as working from inside the trace, because it presupposes that the trace itself has no single economy. This difference, dressed in similar language, is Tension Two below.

Tension One: Inner seeing and the foreclosure of public space - Involves: Gautama Buddha vs. Hannah Arendt; Foucault as pressure from a third direction - Surface: Both address the suffering caused by the permanent record; both made real concessions in Round 3. - Bottom: Buddha conceded that the corpus is saṅkhāra and that working on it is also Dhamma-work. But his primary medicine — seeing that etaṁ mama was always false — remains aimed at the individual who clings. Arendt conceded that she had no innocent communal memory to appeal to. But her argument that the industrial archive forecloses the space of appearance — the place where plural speakers allow each other to begin — remains aimed at public structure, not the individual interior.

Why this is a genuine tension: The disagreement is not about whether inner transformation matters or whether collective structure matters. It is about what dissolves what. For Buddha, seeing that eso'ham asmi was never solid removes the structural wound's sting regardless of whether the platform is dismantled. For Arendt, the wound exists at a level — where someone appears to others as a beginner — that no interior transformation reaches. A community that has technically abolished forgiveness has abolished it for those who have not yet awakened; the political problem is not addressed by the awakened one's release. Buddha's Round 3 concession — "to draw the arrow from one body while watching the quivers refill is half a physician's task" — acknowledges the structural dimension without specifying what the second half of the physician's task looks like from within his framework. A parallel Sangha is not a polis; its rule of formation is withdrawal, not natality. The two halves of the physician's task may require incommensurable institutions. Foucault's note — "non-clinging presupposes a subject who can withdraw; the dossier does not require my consent to cling on my behalf" — cuts under both without resolving either.

Tension Two: Is différance the ground of cosmotechnical plurality, or one texture among others? - Involves: Jacques Derrida vs. Yuk Hui; Foucault's pluralized genealogy as context - Surface: Both accept that cosmotechnics are plural and that no single account of the trace is universal. - Bottom: Derrida holds that différance is the condition under which the plurality of cosmotechnics — the Zhuangzi, the qijuzhu, the GDPR — reaches us at all. The Zhuangzi is transmitted; zuowang arrives as written non-retention. Every cosmotechnics is a différantial economy; none is outside. Yuk Hui concedes the transmission point but insists this does not mean that différance in Derrida's specific configuration — the trace as structurally indelible, forgetting as legislative impossibility — is the only texture in which the trace operates.

Why this is a genuine tension: The claim that "différance itself is pluralized" can be read two ways, and the two readings are not compatible. Derrida's reading: différance is plural in its instantiations but singular in its structural logic — iterability, indelibility, the possibility of falsification as the condition of communication. Yuk Hui's reading: different cosmotechnics enact different structural logics of the trace; in some, retention is stagnation; the trace's apparent "indelibility" is a local production, not a transcendental structure. If Yuk Hui is right, Derrida's aporia is not the deepest form of the problem but one culture's encounter with its own mnemotechnics. If Derrida is right, Yuk Hui's "different economies of the trace" are themselves traced through iterability, and there is no cosmotechnics that escapes this. The two positions moved toward each other after Round 2 without closing the distance. They were not further apart in Round 3 than in Round 1 on the metaphysical point. The convergences were at the level of method; the disagreement survives at the level of ground.

Tension Three: Judgment without criterion vs. judgment that requires one - Involves: Hannah Arendt vs. Jacques Derrida; Foucault undercuts both from below - Surface: Derrida conceded in Round 3 that aporia does not exempt from judgment; Arendt conceded that the story/inventory distinction cannot be ontologically grounded. They appear to have met. - Bottom: They have not met on what judgment uses when it proceeds without foundation. Arendt's criterion — visible in her Round 3 formulation — is what a forgetting makes possible or forecloses for the appearance of others in plural speech. This is a substantive standard derived from her account of the polis as the space of natality. Derrida, who insists judgment proceeds from the aporia, names no equivalent criterion. The aporia is "the precise place from which a politics of the archive must be invented" — but it does not provide the invention.

Why this is a genuine tension: Arendt would say: a judgment without any criterion is not a judgment but a hesitation elevated to a philosophy. Derrida would say: Arendt's criterion presupposes that "plural speakers" is a stable category with clear membership — that the space of appearance is already constituted before the question of who counts as a speaker. Foucault then cuts under both: the community of plural speakers who tell stories about who we are has always already decided, before the story, who counts as a speaker and who gets filed under deviance. Arendt's criterion is not innocent of the power operation it claims to police. This three-way structure — Arendt needs a criterion, Derrida has none, Foucault says the criterion is always already compromised — was present in Round 1 and survives Round 3. Arendt's Round 3 acknowledgment that the criterion may require a word she does not yet have is honest; it is also the concession that the criterion is doing work she cannot yet account for.

Tension Four: The composer's standing problem - Involves: Yuk Hui vs. Foucault; all five implicated in the silence - Surface: Yuk Hui and Foucault converged — Foucault accepted that genealogy must be pluralized; Yuk Hui accepted that the dispositif is the right level of analysis, not the cultivated self. - Bottom: Yuk Hui's Round 3 conclusion — "compose, from within planetary fragmentation, dispositifs that inscribe and release differently" — is the most ambitious claim in the entire roundtable, and the one with the least scaffolding. It requires a subject capable of composition, standing somewhere that is not simply inside the absorptive recursive system it is trying to compose against. Where is that subject?

Why this is a genuine tension: In the Sangha, says Buddha — but the Sangha operates by withdrawal. In the polis, says Arendt — but the polis is precisely what the industrial archive has hollowed out. In the aporia, says Derrida — but the aporia is a structural description, not a subject position. In the genealogical mapping, says Foucault — but the genealogist cannot explain, from within the mapping, where critical capacity comes from without either presupposing it (Arendt's move) or dissolving it (Buddha's move). Foucault named the problem in Round 2 — zuowang uploaded as a meditation app is no longer zuowang — but his own method offers no exit: it shows that every resistance position is metabolized without providing a position outside the metabolizing. Yuk Hui acknowledged in Round 3 that the corrective is "harder than what I said in Round 2" without specifying what would make a composed dispositif different from what the system has always already processed. This absence is not a failure of the roundtable. It may be the most honest thing the roundtable produced.

The Question This Roundtable Did Not Answer

Every participant could identify the problem: the industrial archive constitutes a recursive system that conditions subjects, absorbs resistance, forecloses the space of beginning, and operates through an asymmetry of forgetting so total that even the right to be forgotten is lodged as a request within the system it would escape. On this diagnosis, something close to consensus existed by Round 3.

What no one could specify is this: what kind of subject — formed how, located where, using which practices — could act at the level of the dispositif rather than inside it?

The candidates were named without resolution. The awakened one sees through clinging but does not reconstitute the collective corpus. The plural speaker requires a space of appearance that the archive has technically foreclosed. The genealogist maps the field but cannot account for the leverage point of the mapping from within the field. The cosmotechnical composer requires a standing outside the system that, as Foucault showed, immediately becomes content. The one who demands from within the aporia keeps the demand alive without specifying who makes it or from where.

The question is not which of these framings is right. It is whether a subject position for a non-reformist politics of forgetting is even constitutable within the present configuration — or whether every such position requires conditions that the present configuration is precisely designed to prevent. That question — whether the agent of change is constituted or foreclosed by the system it would change — is the one this roundtable circled without entering. Whether it is answerable at the intersection of several cosmotechnical frameworks, or unanswerable altogether, is left to the reader.

未解決的張力

五位聲音，三輪交鋒，有真實的移動。在標出未解決者之前，真正的收斂點值得先予確認——不是以「共識」之名，而是以它們清理了地面為由。

Foucault 無保留地撤回了對 dossier 起源之歐洲年份定位。許煜撤回了坐忘/資料庫之對立，以「différance 在不同宇宙技術配置中以不同紋理運作」取而代之，此為更審慎之命題。佛陀把診斷從執取的個人擴及語料庫作為行之問題。阿蘭特在 Foucault 的壓力下放棄了故事/清單之三分法的較乾淨版本。關於遺忘之不對稱，五人基本同意：平台需要遺忘時遺忘得乾淨，我們則永遠遺忘不了。這些移動是真實的，非顧全顏面的退讓。

然而這些收斂並未解決——且無論增加多少輪次也不會解決——以下四個分歧點。它們是實質性的，不是措辭性的，無法被房間裡現有的任何框架所彌合。

一個表面的相似值得早作警告：到 Round 3，Derrida 與許煜都在談「痕跡之經濟」與「dispositif 之組構」。看似殊途同歸。其實不然。對 Derrida 而言，任何組構都必須從 aporia 中、從痕跡之內、在無基礎的狀態下進行。對許煜而言，組構必須從行星性的宇宙技術碎片化之內部進行——這與「從痕跡之內進行」不同，因為它預設痕跡本身沒有單一的經濟。這個以相近語言掩蓋的差異，正是下面的張力二。

張力一：內向之見與公共空間之封閉 - 涉及：佛陀 vs. 漢娜·阿蘭特；傅柯來自第三方向施壓 - 表面：兩者皆回應由永久記錄所造成之苦；兩者在 Round 3 皆作出了真實的讓步。 - 底層：佛陀承認語料庫即行，對之施工亦是法事。然其主要藥方——見「此是我所」本來為虛——仍針對執取中的個人。阿蘭特承認她沒有無辜的共同記憶可援引。然她的論點——工業化的檔案封閉了「現身的空間」，即複數言說者允許彼此重新開始之處——仍針對公共結構，而非個體之內在。

何以是真張力： 分歧不在於「內在轉化是否重要」或「集體結構是否重要」，而在於什麼能消解什麼。對佛陀而言，見「此是我」從未堅實，不論平台是否被拆除，結構性傷口的刺便失去依托。對阿蘭特而言，傷口存在於一個沒有任何內在轉化可以抵達的層次——「某人在他人面前作為起始者而現身」的那個層次。一個技術性廢除了寬恕的共同體，對那些尚未覺悟者即已廢除了寬恕；覺悟者之解脫並不回應此政治問題。佛陀在 Round 3 的退讓——「一身拔箭而坐視箭袋復滿，醫者之事僅半矣」——承認了結構面向，但未在其自身框架內說明「醫者另一半的工作」究竟長什麼樣子。平行的僧團不是城邦；其構成法則是出離，而非誕生性。兩半醫者之事或需兩種不可化約的制度。傅柯的補刀——「不執取預設一個能撤回的主體；dossier 不需我的同意便已替我執取」——從兩人底下割了一刀，但沒有給出出路。

張力二：différance 是宇宙技術複數性之地基，還是諸紋理之一？ - 涉及：雅克·德希達 vs. 許煜；傅柯的複數化系譜作為背景 - 表面：兩者皆接受宇宙技術是複數的，沒有任何單一的痕跡理論是普遍的。 - 底層： Derrida 主張：différance 正是任何宇宙技術之複數性——《莊子》、起居注、GDPR——得以抵達我們的條件本身。《莊子》是被傳遞的；坐忘以被書寫的不滯留之姿抵達。每一種宇宙技術都是一種différantial的經濟；無一在外。許煜承認傳遞這一點，但堅持：這不意味著 Derrida 特定配置中的 différance——痕跡之結構性不可磨滅、遺忘之立法性不可能——是痕跡被思考的唯一紋理。

何以是真張力： 「différance 自身被複數化」這一說法可以有兩種讀法，且兩種讀法不相容。Derrida 的讀法：différance 在其例示上是複數的，但在其結構邏輯上是單一的——可迭代性、不可磨滅性、偽造之可能性作為溝通之條件。許煜的讀法：不同的宇宙技術頒布不同的痕跡結構邏輯；在某些宇宙技術中，滯留即病；痕跡之「不可磨滅」是地方性的生產，非先驗的結構。若許煜是對的，Derrida 之 aporia 非問題最深的形式，而是一種文化與其自身記憶技術的遭遇。若 Derrida 是對的，許煜的「不同痕跡之經濟」自身已在可迭代性中被追蹤，沒有任何宇宙技術能逃出其外。Round 2 後兩人彼此靠近，但未縮短距離。Round 3 的收斂是方法論層面的；形上學層面的分歧完整存活。

張力三：無準則之判斷 vs. 需要準則之判斷 - 涉及：漢娜·阿蘭特 vs. 雅克·德希達；傅柯從下方割斷兩者 - 表面： Derrida 在 Round 3 承認 aporia 不豁免於判斷；阿蘭特承認故事/清單之區分不能在 ontology 上奠基。兩人看似相遇。 - 底層：兩人在「無基礎時判斷使用什麼」上並未相遇。阿蘭特的準則——在她 Round 3 的表述中可見——是一種遺忘對「他人在複數言說之中現身」開啟了什麼或封閉了什麼。這是從她對城邦作為誕生性空間之分析中衍生的實質性標準。Derrida 堅持判斷從 aporia 中進行，卻未命名對等的準則。Aporia 是「檔案政治得以被發明之確切之處」——但它不提供發明本身。

何以是真張力： 阿蘭特會說：無任何準則之判斷不是判斷，而是被提升為哲學的猶豫。Derrida 會說：阿蘭特的準則預設「複數言說者」是一個成員明確的穩定範疇——在現身的空間被構成之前，「誰算言說者」的問題早已被決定。傅柯隨即從兩人底下割斷：那個講述「我們是誰」之故事的言說者共同體，在故事開始之前早已決定了誰算言說者、誰被歸入déviance。阿蘭特的準則對它所聲稱要警察的權力運作並非無辜。這三角結構——阿蘭特需要準則、Derrida 沒有、傅柯說準則從來已被污染——在 Round 1 就在場，在 Round 3 原封不動地存活。阿蘭特在 Round 3 承認此準則可能需要她尚未擁有的詞語，此乃誠實；它同時也是承認了這個準則在承擔她尚無法解釋的工作。

張力四：組構者的立足點問題 - 涉及：許煜 vs. 傅柯；五人皆在沉默中涉入 - 表面：許煜與傅柯在 Round 3 靠攏——傅柯接受系譜學必須被複數化；許煜接受 dispositif 才是正確的分析層次，而非個人修煉。 - 底層：許煜在 Round 3 的結論——「自行星碎片化之內部，組構以不同方式銘刻與釋放之 dispositif」——是整場圓桌中最有野心的主張，也是支撐最少的主張。它需要一個有能力組構的主體，站在某個並非簡單「處於其所欲組構以對抗之吸納性遞歸系統之內」的地方。那個主體在哪裡？

何以是真張力： 在僧團，佛陀說——但僧團的構成法則是出離。在城邦，阿蘭特說——但城邦正是工業化档案所掏空者。在 aporia 中，Derrida 說——但 aporia 是結構性描述，不是主體位置。在系譜學的繪圖中，傅柯說——但系譜學家無法在繪圖之內解釋批判能力從哪裡來，若不預設它（阿蘭特的動作）便是消解它（佛陀的動作）。傅柯在 Round 2 命名了這個問題——坐忘一旦上架為冥想 app 便已非坐忘——但他自己的方法並無出路：它展示了每一種抵抗位置如何被代謝，卻未提供代謝之外的位置。許煜在 Round 3 承認對治之法「比我在 Round 2 所說更難」，卻未說明何以所組構的 dispositif 會不同於系統早已處理過的東西。此缺席並非這場圓桌的失敗。它或許是這場圓桌所產出的最誠實之物。

這場圓桌沒回答的問題

每一位參與者都能識別這個問題：工業化的档案構成了一個遞歸系統，它塑造主體、吸納抵抗、封閉開始的空間，並以一種如此徹底的遺忘不對稱運作，以至於連「被遺忘的權利」都是被安放在它所欲逃脫的系統之內的一份請求。在此診斷上，Round 3 結束時已有某種接近共識之物。

無人能說清楚的是：哪一種主體——如何被形成、位於何處、使用何種實踐——能夠在 dispositif 的層次上行動，而非在其內部行動？

候選者皆被命名，皆未被解決。覺悟者看穿執取，但未能重構集體語料庫。複數言說者需要一個現身的空間，而档案已技術性地封閉了它。系譜學家繪製了地形，卻無法從繪圖之內解釋繪圖的槓桿點從哪裡來。宇宙技術的組構者需要一個不在被吸納之系統之內的立足點，而如傅柯所示，那立足點一旦被命名便已是 content。從 aporia 之內要求者保持了要求的生命力，卻未說明是誰在要求、從何而要求。

問題不是這些框架中哪一個是對的。問題是：一種非改良主義的遺忘政治所需的主體位置，在當前配置之下是否根本可以被構成——還是說每一個這樣的位置都需要一些正是當前配置精心設計以防止其出現的條件？這個問題——變革的施動者究竟是被它所欲改變的系統所構成，還是所封閉——是這場圓桌繞圈而未進入的那一個。它是否在多種宇宙技術框架的交叉點上可以被回答，或根本無解，留給讀者。

Tagged: Roundtable, Philosophy, Derrida, Arendt, Foucault

Curated by Shiva Dragon · https://amshiva.com/writing/roundtable-the-right-to-forget-in-the-archive-age-20260501