Madhyamaka Wiki

Madhyamaka Is a Method, Not a Standalone System

17 min read

The claim

Madhyamaka is not a self-standing philosophical system that competes with the Ābhidharmika, Sautrāntika, Yogācāra or non-Buddhist darśanas on their own ground by offering a rival inventory of what exists. It is a definitive (ངེས་དོན་, nītārtha) soteriological method: a procedure of analysis that operates on a conventional truth (ཀུན་རྫོབ་བདེན་པ་, saṃvṛti-satya) the practitioner has already received — from whichever vehicle instructed them — and carries them through to the direct experience of suchness by establishing the two selflessnesses (the selflessness of persons, pudgala-nairātmya, and the selflessness of phenomena, dharma-nairātmya). Its defining feature is that it declines to build a positive conventional ontology of its own, and does so by design, because supplying one is no part of its function.

Four things follow, and together they form the argument:

  1. Against the ontological shoe-horn. Because there is no positive conventional system in the text, reading Nāgārjuna as advancing one — and then auditing it for coherence — is a category error (the move shared by Burton and Wood).
  2. The nihilism impression explained. Because no ontological alternative is offered after the refutation, a reader who expects a system reads the resulting silence as annihilation.
  3. Flexibility. Because the conventional layer is supplied from outside, it is swappable — which is exactly what the Svātantrika sub-schools demonstrate, and is the in-tradition proof that Madhyamaka is a method rather than a system.
  4. The avatāra title. Because the method works out dependent origination (རྟེན་འབྲེལ་, pratītyasamutpāda) and the emptiness of phenomena, its fullest deployment is the bodhisattva path — which is why Candrakīrti called his treatise Entering (avatāra) the Middle Way.

The whole of this works only because of the Two Truths framework: the method is the act of passing from a received conventional register to the ultimate without positing a second reality behind appearances. Strip the framework and the same procedure reads as a bare campaign of denial — i.e. nihilism. The page is therefore the positive, characterising complement to the negative diagnosis of framework-absence-yields-nihilism and the rebuttal of nihilism-charge-refuted: it says what Madhyamaka is such that those two pages come out true.

Evidence

(A) Declining a positive conventional ontology — and why ontological analysis shoe-horns

The Prāsaṅgika does not merely refrain from a conventional ontology by oversight; it refuses one as a matter of method. On the conventional level the Prāsaṅgika negates all six synonyms of intrinsic existence, where the Svātantrika retains them — “existence by virtue of essential nature,” “existence through intrinsic characteristic,” “intrinsic existence” (Prāsaṅgika-Svātantrika, on tsongkhapa-illuminating-intent-1418 Ch 9). What survives is conventional truth held only “in accordance with the world”: MA 6.160 — “in the relative, the existence of the chariot should be accepted in accordance with the world” (Madhyamakāvatāra, the chapter’s lodestar against nihilistic over-reading). The conventional is kept, but with no objective grounding — the precise axis named in the tradition’s own voice as Phya pa Chos kyi seng ge’s twelfth-century Question 5: do conventional truths have philosophical grounding, or only everyday-practice grounding? (westerhoff-candrakirti-2024, via Prāsaṅgika-Svātantrika). The Prāsaṅgika answer is: everyday-practice grounding only.

This is why the ontological-analysis readings shoe-horn. Burton’s regress — prajñaptisat requires an unconstructed basis and an unconstructed constructor (burton-emptiness-appraised-1999 Ch 4) — and Wood’s asat-khyāti “nihilistic idealism,” on which appearances are mere prajñaptisat like the son of a barren woman (wood-nagarjunian-disputations-1994 Ch V), both audit a conventional ontology for what it grounds. But there is no such ontology in the text to audit. The method declines to supply the very thing their analysis presupposes; finding it absent, they read the absence as abhāva. (This is the same diagnostic the wiki applies to the cataphatic-while-apophatic error: the critic demands a positive remainder the method was built not to give.)

A modern philosopher who reads Madhyamaka correctly on exactly this point is Garfield. In garfield-causality-2001 he derives from MMK I and VII a conventionalist regularism: Nāgārjuna rejects causal powers (the hetu “capable of bringing about another by virtue of a power that is part of its nature”) and retains only conditions (pratyaya) — regularities embedded in ever-wider regularities, with no “occult causal powers lurking as the unique and genuine targets of our theoretical activity” (p. 512) and no totality to ground. Applied to ontology this yields a catholic realism that privileges no level as the really-real (“let a thousand entities bloom”), under the Quinean formula “to exist is to exist conventionally, dependently” (p. 511). This is method-not-system stated in the analytic register: Garfield supplies no positive Madhyamaka inventory of his own and grounds the conventional only in the conventions of scientific theory and the dependencies its laws discover — a modern analogue of conventional truth held “in accordance with the world” (MA 6.160). The urge to privilege one level over another, he notes, always derives from a prior claim about where genuine causal power resides (Churchland’s eliminativism, van Fraassen’s anti-realism) — exactly the reificatory expectation of a positive remainder that this page identifies as the engine of the shoe-horn.

His full MMK commentary (garfield-fundamental-wisdom-1995) states the method-not-system reading in its general form. “If the analysis in terms of emptiness is the substantive heart of Mūlamadhyamakakārikā, the method of reductio ad absurdum is the methodological core”: Nāgārjuna “systematically eschews the defense of positive metaphysical doctrines… arguing rather that any such positive thesis is incoherent and that, in the end, our conventions and our conceptual framework can never be justified by demonstrating their correspondence to an independent reality.” The via media on causation (Ch I) is reached “by taking conventions as the foundation of ontology, hence rejecting the very enterprise of a philosophical search for the ontological foundations of convention” — i.e. declining a positive conventional ground by design (the line is stated verbatim a year before FWMW at garfield-dependent-arising-1994 p. 226, where Garfield’s “why start with causation?” hermeneutic makes the emptiness of causation the first instance of the method-not-system reading). And the positionlessness is primary-grounded at MMK 24:18: because everything, “including this very thesis, has only nominal truth,” the self-refutation reductio fails, “one more point at which ladders must be kicked away” (VV 28/29). Garfield is thus a modern witness not only for the regularist reading of causation but for the general structure this page asserts: a reductio method that operates on the received conventional and refuses to terminate in a positive inventory.

(B) Why the move looks like nihilism

The therapeutic procedure ends by removing the reification of intrinsic nature (རང་བཞིན་, svabhāva) and offering nothing positive in its place — no rival inventory, no substitute ground. To a reader trained to expect that a philosophical analysis terminates in a thesis about what exists, the terminus reads as “nothing exists.” The impression is structurally predictable, and Nāgārjuna predicts it: Śūnyatāsaptati 70 — those who do not understand what the Tathāgata explains as “conventionally existent and empty of the sign of true existence are frightened by this teaching” (komito-seventy-stanzas-1987). The fright is the nihilism impression, named in advance.

Put in the terms of nihilism-charge-refuted: what the system-builder registers at this terminus is nihilism₂ — the saṃvṛti-level absence of a grounded conventional ontology, which is exactly this page’s thesis seen from the outside, and which is true. He then misreads it as nihilism₁ — the metaphysical “nothing exists” — by running the absent remainder through the svabhāva-equation (existence = intrinsic existence). The method concedes nihilism₂ and refuses nihilism₁; the nihilism impression is precisely the artefact of conflating the two. This is why the present page is the positive characterisation that makes the rebuttal’s disambiguation true: there is a real, conceded sense in which Madhyamaka “supplies no ontology,” and naming it as the conventional-level reading is what stops it being heard as the ultimate-level denial.

Guardrail (this must be stated precisely, or the point self-defeats). The claim is that Madhyamaka declines to ground conventional truth objectively — not that it has no conventional truth. Nāgārjuna engages saṃvṛti structurally and indispensably: MMK 24:10 (“Unless the transactional realm is accepted as a base, the surpassing sense cannot be pointed out”), which he self-quotes at VV 28 in defence of universal emptiness; Mabja’s “just as one who wishes for water needs a vessel” on the same verse (mabja-ornament-of-reason). The conventional is the indispensable vehicle of the method. Saying instead that “Nāgārjuna does not engage conventional truth at all” is the over-statement Wood needs — it concedes him the very nihilism the registerial Two Truths refuses. The correct formulation throughout is ungrounded conventional truth, not no conventional truth.

(C) The final move on an already-received conventional truth

The method presupposes a prior conventional instruction and completes it. In the Provisional and Definitive (དྲང་དོན་ / ངེས་དོན་, neyārtha / nītārtha) structure, Madhyamaka is the definitive register that supervenes on whatever provisional conventional teaching the student has already received — and the graded-teaching device is stated by Nāgārjuna himself at MMK 18:6–8 (“this is the Buddha’s graded teaching”). The conventional truth on which it operates is not fixed by Madhyamaka: it may be sūtra-derived, Ābhidharmika, Sautrāntika, or otherwise — what matters is that the student has been given a working register of persons, aggregates and karma to then see through. The terminus is the experience of suchness via the two selflessnesses: the selflessness of persons (the chariot analysis, MA 6.150 ff.) and the selflessness of phenomena, the Mahāyāna extension that Mipham glosses as the two no-selves being “of one taste” (mipham-introduction-middle-way-2002). This is the sense in which emptiness is “a soteriological therapy, not an ontological category” (della-santina-madhyamaka-western-1986) — and it is intelligible only against the Two Truths, because the move just is the disciplined passage from the received conventional register to the ultimate (framework-necessity, framework-as-pedagogy).

Mipham makes the terminal character of this move explicit, and in doing so supplies the wiki’s clearest in-tradition statement of Madhyamaka as the method that takes the final step into suchness. On the Madhyamakālaṅkāra he sets out a two-stage pedagogy (shantarakshita-madhyamakalankara, General Introduction): the approximate ultimate (the reasoned, non-implicative negation of true existence) “as a first step… destroys clinging to reality”; then the actual ultimate “halts clinging to non-reality” as well — and that actual ultimate is “free / From constructs and elaborations” (MA v. 70), beyond all four extremes, “the object of neither thought nor word” (MA vv. 71–72). Crucially the actual ultimate is not a further proposition but a state: it is “what noble beings on the Bodhisattva grounds see with the utterly stainless primordial wisdom of meditative equipoise,” and the Prāsaṅgika method “aims to place the mind immediately and directly in the state of freedom from conceptual elaboration.” This is exactly the page’s thesis in Mipham’s voice: the negation is the kindling-stick (cf. Atiśa’s fire-sticks at nihilism-charge-refuted Field 2b), consumed in the act of carrying the practitioner from the received conventional register across to suchness — not a terminus in a new ultimate fact.

(D) Flexibility — the in-tradition proof

If Madhyamaka were a standalone system it would carry its own conventional ontology and could not borrow one. It does borrow one — visibly, and by the tradition’s own classification. The Svātantrikas supply the conventional layer from elsewhere: Bhāviveka takes external objects as real on the conventional level (the Sautrāntika-Svātantrika construal); Śāntarakṣita takes the conventional in Yogācāra terms (he is classed Yogācāra-Svātantrika, Prāsaṅgika-Svātantrika). And the tradition reads this borrowing not as doctrinal rivalry but as pedagogical skilful means: Tsongkhapa presents the Svātantrika as “great skilful means to help guide those who are, for the time being, not capable of easily realising” the subtler view (tsongkhapa-illuminating-intent-1418 Ch 9); Mipham reads the Svātantrika–Prāsaṅgika difference as pedagogical emphasis, not philosophical rank (shantarakshita-madhyamakalankara). That the conventional layer is swappable between Sautrāntika and Yogācāra registers without changing the ultimate analysis is the cleanest evidence that what is constant — the analysis to emptiness — is a method, and what varies — the conventional inventory — is a received input the method does not itself fix.

The sharpest single demonstration comes from the Madhyamakālaṅkāra itself, as read by Mipham. The text’s whole argument is the “neither one nor many” reasoning, and Mipham shows that the very same reasoning can be set out either as a prasaṅga (drawing the unwanted consequence from the opponent’s own assertion that things truly exist) or as an autonomous inference (taking the merely mind-posited subject through “other-elimination”) — citing the Madhyamakāloka that the two methods “refute… equally” and that no division of cases is needed (shantarakshita-madhyamakalankara, “A Prāsaṅgika or a Svātantrika argument?”). The analysis is method-constant; the form (consequence vs. autonomous inference) is a presentational variable calibrated to the disciple. And the conventional input here is overtly received: Śāntarakṣita takes the Cittamātra inventory of mind-only-appearances as the conventional truth (MA v. 91, “All causes and effects / Are consciousness alone”), then runs the constant analysis over it to empty mind itself (v. 92). The conventional register is supplied from the prior turning; the method does not generate it.

The signature across the whole tradition (cross-ref framework-internal-debate-is-productive). The swappability is not an isolated feature of the two Svātantrikas; it is visible as a pattern across the entire Indo-Tibetan debate, and the pattern is itself the strongest in-tradition evidence for method-not-system. The schools diverge sharply at the level of conventional truth — Sautrāntika external realism (Bhāviveka), Yogācāra mind-only (Śāntarakṣita), Tsongkhapa’s pramāṇa-warranted conventional, Gorampa’s snang tsam — while converging at the level of the ultimate: freedom from the four extremes, the non-affirming negation, no svabhāva. This divergence-at-saṃvṛti / convergence-at-paramārtha signature is what a method operating on a received, interchangeable conventional would produce, and what its two rivals would not: a self-standing system would carry its own conventional and so converge there too, while a mere shared fiction would show no principled split between the levels at all. The signature is therefore positive evidence for the thesis, not merely an illustration of it — developed as the bounded-fertility argument at framework-internal-debate-is-productive.

(E) Mahāyāna-directedness and the avatāra title

Because the method works out dependent origination and the emptiness of phenomena (not only the selflessness of the person), its fullest deployment is the bodhisattva path — the practitioner who must sustain a wide and deep realisation of emptiness while remaining in saṃsāra for the welfare of beings, relatively freed from the fetters of the kleśas (the ārya bodhisattva on the bhūmis). This is why Candrakīrti titled his treatise Madhyamakāvatāra, “Entering the Middle Way”: entering the view is entering the bodhisattva path, and the text’s architecture is that claim — the ten bhūmis from the first-ground path of seeing (Ch 1) to the buddha-ground (Ch 11), with the analysis of emptiness placed at the sixth ground, the perfection of wisdom (Madhyamakāvatāra). The MA thus fulfils what Mabja frames as the third step of ascertaining the Two Truths.

Calibration (do not collapse the live disagreement). “Directed especially at the Mahāyāna” must not be over-read as “emptiness is Mahāyāna-exclusive.” The MA 1.8 majority — Tsongkhapa, Gorampa, the Ninth Karmapa — holds that arhats do realise the selflessness of phenomena; only Mipham dissents, and even he grants arhats “some” realisation while reserving complete dharma-nairātmya as “definitionally Mahāyāna” (Madhyamakāvatāra, MA 1.8). The careful claim is therefore about application, not availability: what is Mahāyāna-specific is the breadth and depth of emptiness sustained while remaining on the bhūmis, and the bodhisattva-path framing that the avatāra encodes — not emptiness as such, which the majority extends across the vehicles.

(F) Working hypothesis (Tenpa) — is a “pure Prāsaṅgika” even possible?

Flagged as the wiki author’s working position, not yet an adjudicated wiki claim — to be developed on sp-hierarchy-is-tibetan-construction. If Madhyamaka is a method overlaid on a received conventional ontology, and if the Tibetan schools in practice accept the four tenets as a graded conventional ontology (deploying Sautrāntika/Yogācāra commitments “depending on the situation and the step along the gradual path”), then the operative, discursive Madhyamaka of the textbooks is really Svātantrika scaffolding, and “pure Prāsaṅgika” names not a standpoint one occupies in debate but the equipoise terminus of the method. Mipham supplies three textual supports that make this more than speculation (shantarakshita-madhyamakalankara): (i) the genuine no-assertion ultimate belongs to ārya meditative equipoise — even Āryas in post-meditation “remain within the scope of thought and word, assertion or denial” in order to teach and debate; (ii) an assertion-making “Prāsaṅgika” is, by Mipham’s own criterion, “no different from” a Svātantrika (“inasmuch as certain ‘Prāsaṅgikas’ remain on the level of the approximate ultimate truth, making assertions… there is no distinguishing them from Svātantrikas”); and (iii) Mipham’s own observation that “in Tibet, even the explanation of the Prāsaṅgika view reverts to that of the Svātantrikas.” On this reading the supposed Prāsaṅgika/Svātantrika rank dissolves into a pedagogical difference (Mipham’s settled view, sp-hierarchy-is-tibetan-construction), and being a “Prāsaṅgika” is something one transitions into as one is prepared — fully consummated only in the ārya bodhisattva’s equipoise. Caveat to verify before load-bearing use: the precise claim “a true Prāsaṅgika can only be an ārya bodhisattva” is substantiated in spirit by the equipoise passages (raw pp. 24–25, 113) but is not a verbatim Mipham slogan; state it as the equipoise-terminus reading, not as a direct quotation.

Evidence against / objections

  • “Method vs system” may be a false dichotomy. A critic (Wood above all) can grant that Madhyamaka is therapeutic and still insist the therapy commits its practitioner to propositions — universal niḥsvabhāvatā, dependent origination, the two truths — which together are a minimal system with a definite (nihilist, he says) content. Reply: the commitments are real but are no positive thesis about the ultimate (VV 29, emptiness of emptiness; see nihilism-charge-refuted Field 3) and carry no conventional ontology — which is the specific sense of “not a system” this page asserts. The dichotomy is between method-with-received-conventional-input and system-with-own-conventional-ontology, not between method and any commitment whatever.
  • The Svātantrika “borrowing” may understate a real ontological difference. If Bhāviveka genuinely accepts svalakṣaṇa on the conventional level, the conventional layer is not cleanly “swappable” — it is a substantive commitment that, on Tsongkhapa’s substantive (not merely methodological) reading of the distinction, exposes residual realism (Prāsaṅgika-Svātantrika). So (D) holds most strongly for the Prāsaṅgika and should be stated as: the Prāsaṅgika declines a conventional ontology, while the Svātantrikas show the conventional layer is an input even when they ground it more robustly than the Prāsaṅgika would allow.
  • Object-side Tsongkhapa is the hard case. Tsongkhapa’s “each phenomenon has two natures… found by two distinct cognitive processes” (conventional-truth) is the wiki’s lone object-side voice and the one most exposed to the charge that Madhyamaka does carry a conventional-ontological commitment after all. The page’s “declines a positive conventional ontology” claim is cleanest on the subject-side coalition (Gorampa, Karmapa, Mipham, Dzongsar Khyentse) and should be pressure-tested against object-side Tsongkhapa — the same residual soft point flagged at nihilism-charge-refuted (Wood Step 4).
  • “Received conventional truth” risks relativism. If any conventional system will do as input, is the method indifferent to its conventional starting point? Reply: no — the input is ranked by the provisional/definitive structure for efficacy of removal, not by resemblance to a hidden object (nihilism-charge-refuted Field 2c); the method is calibrated to dismantle whatever reification the received register installed, which is why the same method takes different conventional inputs.

Linked pages

Graph View

Backlinks