What This Concept Is About
The One-Sentence Core
不成 (asiddhi, "non-establishment") is a category of fallacy in Buddhist logic (因明, *hetu-vidyā*) that names the moment a debate, a proof, or a piece of reasoning collapses — not because its conclusion is wrong, but because the shared ground on which any reasoning could stand has not been laid down between the two parties. The reason (因, *hetu*), the subject of the proposition (有法, *dharmin*), the predicate, or the locus on which the reason depends — none of these have been mutually acknowledged (*不極成*) by the proponent (立者, *pratijñā-kāraka*) and the opponent (敵者, *pṛthagjana* / counter-thesis holder).
This is the most fundamental failure mode in reasoning: not a failure of *conclusion*, but a failure of *foundation*.
Where It Sits in the Whole Path to Liberation
In the Tibetan Buddhist curriculum, 因明 is not itself a "liberating step" the way 止 (śamatha) or 觀 (vipaśyanā) are. It is something more like epistemological hygiene — the rigorous training that makes the mind a reliable instrument *before* one attempts the far more delicate work of seeing emptiness (空, *śūnyatā*) and dependent origination (緣起, *pratītyasamutpāda*).
不成 is the diagnosis for the *first* disease a valid cognition (量, *pramāṇa*) can suffer. Without being able to spot non-establishment, a practitioner cannot tell the difference between reasoning that leads to insight and reasoning that *looks* like insight but rests on air. The whole edifice of the pramāṇa tradition — the three characteristics of a valid reason (三相, *trairūpya*), the proper relation of example (喻, *dṛṣṭānta*) to proposition (宗, *pratijñā*) — is built on the assumption that the foundation is shared. 不成 is what you get when that assumption is broken. In a path that depends on the reliable operation of valid cognition for its highest fruit, learning to recognize 不成 is not optional. It is the difference between a ladder and a painted ladder.
The Classical Taxonomy — Two Texts, One Core, Different Edges
The two foundational Indian texts for this concept — 《正理滴論》(*Nyāyapraveśa*, attributed to Dharmakīrti / 法稱) and 《因明入正理論》(*Nyāyapraveśa* by Śaṅkarasvāmin / 商羯羅主) — diagnose the same disease but cut it along slightly different lines. A serious reading needs both.
From 《正理滴論》: Three Forms Inside the Reason
This text lists three specific kinds of 不成 within the *reason itself* (似因之不成), and adds a structural claim about the three characteristics of any valid reason:
**1. 猶豫不成 (*samśayā-siddhi*, "uncertain non-establishment") — The reason's own identity is undecided. The classical example: someone tries to prove "there is fire on that mountain" using a column of mist-or-smoke** (疑惑之霧氣等) as the reason. From a distance, you genuinely cannot tell whether what you see is smoke (煙) or vapor (蒸氣、霧氣). The reason — "smoke, therefore fire" — works only if you have smoke. If the evidence is ambiguous between smoke and mist, the reason itself *oscillates*, and an oscillating reason establishes nothing. This is not a problem with the *conclusion* (fire might genuinely be there) — it's a problem with the *warrant* (we cannot commit to calling it smoke).
2. 所依不成 ("non-establishment of the locus/basis") — The place where the reason is supposed to inhere cannot be pinned down. Classical example: someone tries to prove "there is a peacock in this mountain range" using the peacock's call (孔雀鳴) as the reason. The problem is not the call itself (peacocks do call) and not the conclusion (there may well be a peacock). The problem is that *we do not know where the call is coming from*. Is it from this mountain? From the next? From a hidden valley? Until the locus is fixed, the reason — "a call heard here, therefore a peacock here" — has no foothold.
**3. 有法不成 (*dharmin-asiddhi*, "non-establishment of the subject")** — The very thing you are trying to prove something *about* does not exist for your opponent. Classical example: a Vaiśeṣika (勝論派) philosopher tries to prove "I am everywhere" using "because there is a quality (功德) that reaches to all places" as the reason. The reason is built on the assumption that there *is* an "I" (ātman) — a self — that could be everywhere. But a Buddhist opponent (敵論者如佛家) does not concede that any such "I" exists. So the very subject of the debate is not jointly accepted (*不極成*) by both parties. The reason is perfectly logical *given* the subject, but the subject is not on the table to begin with.
A structural claim the text also makes: each of the three characteristics (三相) of a valid reason — that the reason is present in the subject (*pakṣa-dharmatā*), that it is present in things homogeneous with the subject (*sapakṣa-dharmatā*), and that it is absent from things heterogeneous with the subject (*vipakṣa-dharmatā*) — carries its own *two* possible faults: 不成 (the characteristic fails to be established) and 猶豫 (uncertainty about whether it is established). Uncertainty can thus be both a *type* of 不成 and a *cause* of 不成 — a subtle point this text makes explicit.
From 《因明入正理論》: Two Layers — Proposition *and* Reason
This text treats 不成 in two distinct locations, which is itself an important teaching: the disease can infect either the proposition you're trying to establish, or the reason you're using to establish it.
Layer One — In the Proposition Itself (似宗過之「不極成」諸相):
Among the five faults of a faulty proposition, three are directly about 不極成 — non-mutual-establishment:
- 能別不極成 (predicate not mutually established): When you try to predicate something of the subject, but the *predicate* (the thing you're saying about the subject) is not accepted by your opponent. Example: a Buddhist debater says to a Sāṃkhya opponent, "Sound is perishable" (聲滅壞). The Buddhist accepts this, but the Sāṃkhya does not accept that sound perishes. The whole predicate fails to land.
- 所別不極成 (subject not mutually established): The mirror case. The Sāṃkhya says to a Buddhist, "I am thought" (我是思). The Sāṃkhya accepts this, but the Buddhist rejects the existence of any permanent "I" to begin with. The subject doesn't get a foothold.
- 俱不極成 (both not mutually established): Both ends fail at once. A Vaiśeṣika says to a Buddhist, "I am the cause of conjunction" (我以為和合因緣). Neither the "I" as subject nor "cause of conjunction" as predicate is accepted by the Buddhist opponent.
This text adds a fourth, paradoxical category — 相符極成 — where the proposition is *so well* agreed upon by both sides that it becomes useless as a *thesis*. If both parties already accept it, you haven't established anything; you've merely stated the obvious. The text names the disease precisely: 「不容成故」 — it does not *fail* to be established, but it *cannot serve* as something to be established. This is a brilliant diagnostic: the failure of the proposition can be either that the ground is missing *or* that there is no ground *needed* because everyone already stands on it.
Layer Two — In the Reason Itself (似因之四種不成):
- 兩俱不成 (non-establishment for both parties): Neither you nor your opponent accepts the reason as applying to the subject. Classical example: trying to prove "sound is impermanent" using "because it is visible" (眼所見性故) as the reason. Neither party accepts that sound is visible. The reason is shared non-existence, not shared existence — which is worse than no reason at all.
- 隨一不成 (non-establishment for one party): One side accepts the reason, the other doesn't. Example: trying to prove something to a Sāṃkhya philosopher using "because it is produced" (所作性故) as the reason. The Buddhist proponent accepts that produced things are impermanent, but the Sāṃkhya opponent denies that sound (which is eternal in their system) is *produced*. The reason holds in your world but not in theirs.
- 猶豫不成 (uncertainty non-establishment): Same logic as in the 《正理滴論》 account — the reason itself is undecided. Example given: trying to prove "there is fire produced by the conjunction of the great elements" using ambiguous perceptual evidence about whether such fire is present.
- 所依不成 (non-establishment of the basis): Same logic — the locus is uncertain. Example given: trying to prove "space (虚空) is real" using "because it is the substratum of qualities (德所依故)" as the reason, addressed to a school (無空論) that denies space exists at all. If the opponent doesn't even grant that space exists, then *whatever* you say about space's qualities has no place to land.
The Subtle Difference Between the Two Texts
The texts aren't contradictory; they're like two clinicians examining the same patient from different angles:
- 《正理滴論》 locates "subject non-establishment" (有法不成) *within* the faults of the reason, because for that text, if the subject is contested, the whole reason-subject-predicates triangle collapses from within. - 《因明入正理論》 keeps it in a separate category — faults of the proposition (似宗過) — because for that text, the proposition and the reason are two distinct moves, and either can fail on its own.
Both readings are coherent. Both texts are needed to see the full shape of the disease.
Why This Concept Is Not Just "Being Wrong"
不成 is the easiest fallacy to mis-identify because it looks like *other* failures. It is not:
- **相違 (*viruddha*, "contradictory")**: when the reason actively opposes what it's supposed to prove (e.g., using "because it is permanent" to prove impermanence). 相違 names an *opposition* in the reasoning. 不成 names a *missing ground*. - **不定 (*anaikāntika*, "indeterminate")**: when the reason could prove too many things (e.g., "sound is impermanent because it is knowable" — but pots are also knowable and are impermanent for unrelated reasons). 不定 names an *over-reach* in the reasoning. 不成 names a *missing base*.
The classical image that captures all three: a building. 相違 is a load-bearing wall placed backwards (it actively works against the structure). 不定 is a column that holds up too many unrelated roofs (it spreads itself thin). 不成 is no foundation at all — the building is in the air, and we're arguing about what kind of roof to put on.
Life Walkthrough: Walking Through This Concept in Real Scenes
The classical examples (smoke-or-mist from a distance, a peacock's call from an unknown place, trying to prove "the self exists" to a Buddhist, trying to prove "sound is visible") can feel remote. But every category maps precisely onto situations we are in constantly. Let me walk through several.
Scene 1: The Couple Arguing About "Quality Time"
The argument: "You never spend quality time with me." / "I am with you every evening — I just need to decompress by scrolling after work."
What is actually happening, in 因明 terms: The two parties are not disagreeing about whether they spend time together. They are failing at 俱不極成 — both subject and predicate are unestablished between them. What is "quality time"? The proponent (the partner who feels neglected) defines it as engaged, present, undistracted attention. The opponent (the partner who scrolls) defines it as co-located time with low cognitive demand. Neither party has made the other party's definition *the* shared ground. So when the reason "you never spend quality time with me" is offered, the predicate ("quality time") has nowhere to stand. The argument looks like a disagreement about quantity or behavior; it is actually a non-establishment of what the debate is *about*.
How to fix it using the classical framework: Step back from "you're wrong about quantity" and address the missing ground: "I need us to agree on what we mean by 'quality' before we can argue about whether there's enough of it." This is precisely the move of *establishing the basis* before proceeding to the proof.
Where people commonly map this incorrectly: Most couples treat this as a 相違 — "you say X, I say not-X, we're opposed." It's not. You're not yet opposed. You're both standing on different ground, arguing across a gap. The conflict isn't between your positions; it's between your *definitions*.
Scene 2: The Work Strategy Meeting That Goes Nowhere
The argument: Two executives argue about whether to invest in a new product line. One says "the data shows consumer demand." The other says "the data shows market saturation."
What is actually happening: Look at the structure. Each side has its *own* data set, its *own* definition of "demand" (units shipped? survey willingness? search trends? retention curves?), and its *own* time horizon. The reasoning — "we should invest *because* demand exists" — has a reason that 猶豫 is operating on: is the evidence actually evidence of demand, or is it evidence of something else? This is the smoke-or-mist example, exactly. Each side is pointing at a column of vapor and insisting it's smoke (or insisting it's mist). The data is uncommitted between interpretations; therefore the reason is uncommitted; therefore no conclusion can be established.
Classical fix: Before arguing about the conclusion, you have to do the work of establishing what the data is, in a way both sides accept. If you cannot do that — if "demand" remains ambiguous — the right move is to admit that no valid reasoning is possible yet, and to gather more disambiguating evidence. The 《正理滴論》 point is exact: an uncertain reason establishes nothing, and that is not a problem to be argued through but a problem to be *investigated out of*.
Common mis-mapping: Most strategy meetings treat this as "we need to weigh the arguments" (相違 framing) or "we need more data" (vague intuition). The 因明 diagnosis is sharper: the *specific* fault is 猶豫不成, and the *specific* remedy is to commit, by joint investigation, to what the data actually says before invoking it as a reason.
Scene 3: The Online Debate About "Is AI Conscious?"
The argument: Commenter A: "AI is conscious because it responds in emotionally appropriate ways." Commenter B: "AI is not conscious because it has no subjective experience." Commenter C: "Consciousness is just information processing, so of course AI is conscious."
What is actually happening: This is a textbook 有法不成 or 能別不極成 situation, depending on which move you examine. The *subject* — "AI" — might be granted by all parties, but the *predicate* — "conscious" — is not. Or, worse, the *subject* of the discussion ("consciousness as such") is itself contested: behaviorists, functionalists, dualists, and Buddhist no-self theorists each grant a different thing when they say the word.
If the parties cannot first establish what they mean by "consciousness" in a way all parties accept, then *every* subsequent reason — "because it responds emotionally," "because it lacks qualia," "because it processes information" — falls into 不成. Not because the reasons are bad in themselves, but because the predicate they are trying to attach to AI is not mutually established. You are arguing about what color to paint a wall whose existence is contested.
Classical fix: The 《因明入正理論》's "相符極成" warning is also alive here: if everyone *already* agrees on what "consciousness" means (within some narrow technical sense), then arguing whether AI has it is fine. If no one agrees — if the term is doing radically different work for different speakers — then before you can even *have* the debate about AI, you need a meta-debate about the term. And if no agreement is possible on the term, the specific claim "AI is conscious" cannot be established *as that specific claim*. It can perhaps be established as some weaker, rephrased claim using mutually accepted terms. But that's a different argument.
Common mis-mapping: People treat this as "we just need more evidence about AI" (treating it as 不定, a problem of insufficient data). It's not. You could give me all the neural-network weights in the world and the disagreement would persist — because the disagreement is *upstream* of evidence, in the definition of the predicate.
Scene 4: The Family Member Who Says "You're Just Like Your Father"
The argument: "You always do that — it's exactly what your father would do." / "I am nothing like him. I have nothing in common with that man."
What is actually happening: Both parties are making a claim that requires the predicate "alike your father" to be mutually understood. But "alike" is the contested ground. One person means "alike in this specific behavioral pattern." The other means "alike in essence, in character, in fate." The predicate is not mutually established (能別不極成), and the reason being offered — "you do X, just like he did" — is a 隨一不成: it holds for the speaker (who sees the pattern), but is rejected by the hearer (who does not accept either the likeness or its significance).
Classical fix: Separate the proposition. "You did X today" can be argued. "Doing X is a sign of being like your father" can be argued. "Being like your father is bad" can be argued. But *conflating* all three into "you're just like your father" is an 不成 failure: the predicate is too thick, carrying too many contested elements at once. The 因明 remedy is to *thin out* the proposition until both parties can at least accept the thinnest version.
Common mis-mapping: Most people treat this as a 相違 ("I say we're alike, you say we're not"). It is *pre*-相違 — it is the failure of a shared predicate that would *make* a 相違 possible. Until the predicate is established, there is no opposition; there are two separate arguments happening in two separate rooms.
Scene 5: The Doctor and Patient With Different Definitions of "Better"
The argument: Doctor: "Your numbers are improving — you're getting better." Patient: "I don't feel better. I'm not better."
What is actually happening: Two predicates are in play — "better" as biomarker improvement, and "better" as felt well-being. The doctor is operating in the medical-pramāṇa frame. The patient is operating in the experiential-pramāṇa frame. Neither is *wrong* within their own frame. But when the doctor offers the reason "your numbers improved, therefore you are better," the patient experiences this as 能別不極成: the predicate "better" does not mean what the doctor thinks it means to the patient. The patient's 所依不成 (basis non-establishment) is also operative: the "basis" for "better" in the patient's world is not the lab values, so the reason has nowhere to attach.
Classical fix: Before arguing about whether the patient is "better," the parties need to establish what "better" is jointly taken to mean — at least for the purpose of this conversation. If they cannot, they should speak in two separate vocabularies: "your biomarkers have improved" (a proposition the patient might accept as factually true) and "your felt sense of health" (a separate proposition). The 因明 lesson is that one proposition cannot do the work of two.
Why Contemporary People Need This Concept
1. We Live in an Argument-Saturated Ecosystem Without an Argument-Saturated Training
The contemporary information environment — social media, op-eds, podcasts, group chats, comment threads — produces more *attempts at reasoning* per day than most premodern humans encountered in a year. But the training in *how to diagnose the failure of those attempts* has not kept pace. Most people can name a fallacy if it is "ad hominem" or "slippery slope" — but they cannot see the more common, more silent failure: two people arguing with elaborate reasons, where neither party's foundation is shared, and so the entire debate establishes nothing.
不成 is the diagnosis for the silent failure. We need it because we live inside the silent failure constantly.
2. The AI Era Has Made "What Does This Term Mean?" the Central Question
The rise of large language models, recommendation algorithms, and AI assistants has turned *definitional grounding* from a philosopher's hobby into an engineering bottleneck. Every system that uses a term like "helpful," "safe," "relevant," "fair," "accurate," or "harmful" must do some version of the work 《因明入正理論》 describes: establish what the term means to the system and to the user before the system can act on it. When this fails — and it fails often — the failure has the exact shape of 不成: the system and the user are reasoning, but their predicates are not mutually established.
A clear-eyed contemporary reading of 不成 can sharpen how we think about specification, alignment, prompt design, and the experience of "the AI just doesn't understand what I mean." The 因明 framework says: that experience is real, and it has a name, and the fix is to do the foundational work the framework demands — not to keep arguing past each other.
3. It Is the Antidote to a Particularly Modern Vice: Confusing Eloquence for Establishment
A speaker who argues beautifully for a position can make an audience *feel* that the position has been established. In a media ecology optimized for engagement, the felt sense of establishment is often *sold in place of* actual establishment. 不成 is the concept that lets you pause and ask: "Before I respond to the force of this argument, is the foundation actually shared? Is the reason actually a reason in this discussion? Is the predicate something we both mean by the same word?"
The classical example of 兩俱不成 — using "visible" to prove something about sound, when no one accepts that sound is visible — looks absurd on the page. But its contemporary cousins are everywhere: using metrics that no one agrees measure what they claim to measure; using analogies that no one agrees are analogous; using authority that no one agrees is authoritative in this domain.
4. It Respects the Other Person Without Caving to Them
不成 is sometimes mistaken for a kind of relativism — "if we don't share the ground, no reasoning is possible, so anything goes." This is the opposite of what the framework teaches. 不成 says: *reasoning requires shared ground*. That is a *high* standard. It tells you that many things currently passing as reasoning are not reasoning at all. It is a rigorous demand, not a permissive one.
In a time when public discourse routinely confuses "loudly stating a position" with "establishing a position," 不成 is one of the most clarifying concepts available. It restores the distinction between *voicing* and *establishing*.
5. It Is a Quiet Form of the Buddhist Path's Core Commitment
The Buddhist path, at its deepest, is committed to the unreliability of unexamined cognition — to the idea that much of what we take as established is, on inspection, not established at all. 不成 is that commitment, applied at the level of reasoning before it is applied to the level of self and world. To practice spotting 不成 in daily argument is to do a *microcosm* of the larger Buddhist work of seeing that what we treat as solidly given (a self, a world, a permanent ground) is, in fact, not established when looked at carefully.
There is a deep continuity between 因明 and 般若: the pramāṇa tradition trains the mind to see what is and is not properly established at the level of *cognition*, and the prajñā tradition extends the same discipline to the level of *being*. 不成 is the early, technical form of the same instinct.
Common Misreadings and Clarifications
Misreading 1: "不成 just means 'wrong'"
It doesn't. The classical examples make this clear. You can have a conclusion that is *true* — there *might* be fire on that mountain, there *might* be a peacock in that range — and still have the *reason* be 不成 because the warrant (this is smoke, that call is from here) has not been jointly established. 不成 is a diagnosis about the *warrant structure*, not the truth value of the conclusion.
Misreading 2: "不成 is the same as 相違 (contradiction)"
相違 names a *clash within* the reasoning — the reason actively opposes what it's supposed to prove. 不成 names an *absence* beneath the reasoning — there is no shared ground on which any reasoning could stand at all. The classical image: 相違 is two people on the same floor pushing opposite walls; 不成 is two people in different rooms arguing through the wall.
Misreading 3: "不成 is the same as 不定 (indeterminate)"
不定 names an *over-extension* of the reason — it proves too much (the reason "knowable" applies to pots as well as sound, so it cannot distinguish impermanent things from permanent ones). 不成 names an *under-extension* of the foundation — there is no ground to extend from. Different diseases, different remedies.
Misreading 4: "不成 is just about formal debates — it doesn't apply to ordinary life"
The opposite is closer to the truth. Formal debates have at least some procedural pressure toward establishing the foundations. Ordinary life — text messages at 11pm, family meals, group chats, performance reviews — has almost none. This is precisely why the *failures* of 不成 are so much more common in ordinary life, and why the concept is so practically valuable.
Misreading 5: "If there's 不成, the only response is to give up"
No. The 《因明入正理論》 and 《正理滴論》 are not defeatist texts. They are diagnostic. The proper response to diagnosing 不成 is to *do the foundational work that the argument skipped*: clarify the predicate, fix the locus, agree on what evidence is evidence of what, separate overlapping propositions, and *then* re-attempt the proof. 不成 is the beginning of a more rigorous argument, not the end of one.
Misreading 6: "不成 means the opponent is being unreasonable"
This is perhaps the most insidious misreading. The classical examples are scrupulous: 不成 is not a charge that the *opponent* is being unreasonable. It is a charge that the *argument* has not done its foundational work. In many cases (the peacock call example, the smoke-or-mist example), the *proponent* is the one at fault — the argument was set up before the ground was cleared. 不成 is, in this sense, a humbling concept: it tells you, as often as not, that the failure is in *your* setup, not in your opponent's resistance.
Misreading 7: "Modern scientific reasoning has solved 不成"
Partially. Modern science has developed rigorous procedures (operational definitions, controlled experiments, peer review, pre-registration) precisely to address what 不成 names. But science is a *practice*; it can be honored in the breach as easily as any other. Paradigm shifts, replication crises, definitional drift in clinical endpoints, and the replication of "evidence-based" claims that aren't actually mutually established between researchers — all of these are modern, real-world instances of 不成, sometimes hiding inside vocabulary that sounds like science. The 因明 framework can *see* these failures even when the scientific vocabulary cannot name them, because the 因明 framework cuts along a different line — not the line of method, but the line of *what has been jointly granted*.
A Closing Image
The classical examples are vivid and worth carrying with you: the column of vapor at a distance that you cannot commit to as smoke; the peacock's call whose source you cannot locate; the proof of a self that your opponent does not concede exists; the visible quality attributed to sound that no one accepts; the "substratum of qualities" attributed to a space your opponent denies.
These are not exotic cases. They are the structure of the moment, daily, when we argue with reasons that look powerful but rest on air — when the predicate has not been jointly defined, the locus has not been jointly fixed, the evidence has not been jointly accepted as evidence, or the very subject of the debate has not been jointly granted to exist.
不成 is the old, precise, gentle name for that moment. To learn it is to learn how to stop building on air. And in a path that ultimately rests on seeing through all *apparent* establishment to what is *actually* established, that is no small beginning.