What This Concept Is Saying
Imagine a courtroom scene. A lawyer jumps to their feet, points dramatically at the opposing counsel, and shouts, "Ladies and gentlemen, you see the flaw!" The jury nods along, convinced that something has been exposed. But when the transcript is later read aloud, no one — including the lawyer themselves — can specify what the flaw actually was. The gesture was perfect. The rhetoric was flawless. The "flaw" was not there.
This is what classical Buddhist logic (因明, *hetuvidyā* / pramāṇa) calls 似能破 (*dūṣaṇābhāsa*) — literally "the appearance of a refutation," or in more traditional phrasing, an 邪破 ("deviant refutation"). The classical entry puts it with admirable compactness: it is a refuting utterance that *looks* like it is doing its job but fails to correctly display any fault in the opponent's thesis. The judgement of whether a refutation succeeds has nothing to do with how sharp, confident, or popular the rebuttal sounds; it depends entirely on whether the alleged fault is actually there.
To see why this matters, we need to understand what a "real" refutation — 能破 (*dūṣaṇa*, genuine refutation) — is supposed to do. The opponent has put forward a structured argument: a thesis (宗, *pratijñā*), a reason (因, *hetu*), and an example (喻, *drṣṭānta*) — together a 能立 (*sādhana*, that which establishes). The debater on the other side is supposed to point at the specific point where this structure is missing a required condition. If they can do so accurately, the establishment collapses and the refutation succeeds. If they cannot — if they are simply performing refutation — then they have only produced a *似* (resemblance, *ābhāsa*) of one.
Three classical texts in our lineage treat this single concept from three angles, and the depth of the entry lies in holding all three together rather than collapsing them.
**The first angle — from 《因明入正理論》(*Nyāyapraveśa*, "Introduction to Valid Cognition")** — is the catalogue angle. It lists six specific places where a real refutation is supposed to bite, and therefore six places where a fake one typically bites wrong:
1. 能立缺減 — calling a complete probandum deficient when it is not. 2. 立宗過 — claiming a fault in the thesis when none is present. 3. 不成因 — declaring an established reason to be "unestablished." 4. 不定因 — declaring a determinative reason to be "indeterminate." 5. 相違因 — declaring a non-contradictory reason to be contradictory. 6. 喻過 — finding a fault in an example that has no fault.
In each case the *form* of the accusation is the same form a real refutation would take — the very vocabulary of formal logic is being deployed — but the *content* is empty. The text closes the case bluntly: "以不能顯他宗過失，彼無過故" — "Because [the refuter] cannot display a fault in the opponent's position, the opponent's position being faultless." The fault lies not in the rebuttal's volume but in its referent.
**The second angle — from 《正理滴論》(*Nyāyabindu*, "Drop of Logic," translated from the Tibetan)** — zooms in on the *criteria* a refutation must actually meet to count as genuine. In classical Buddhist logic, a valid reason (因) must possess three characteristics (三相, *trairūpya*): it must be present in similar instances (同品定有, *anvaya*), absent in dissimilar ones (異品遍無, *vyāpti*), and stated of the subject of the thesis. These come in two flavours: 共同性相 (characteristics shared by all valid reasons) and 個別性相 (characteristics specific to the case at hand). A real refutation is one that genuinely shows one of these characteristics is missing. A 似能破 is one that *names* a missing characteristic but hasn't actually shown it. The text is careful to draw a line between a real 不成 ("unestablished") fault and mere 疑惑 (doubt, *saṃśaya*): if the refuter is only uncertain whether the reason covers dissimilar instances, that is not yet a genuine fault of "indeterminate reason" — it is an unsettled question, not an established rebuttal. The concept of 似能破, on this reading, includes the subtle case where a refuter mistakes their own unresolved doubt for a structural failure in the opponent's argument.
**The third angle — from 《入中論善顯密意疏》(*Madhyamakāvatāra-ṭīkā*, Commentary Revealing the Intent of "Introduction to the Middle Way") — is the most philosophically charged. It introduces the principle of 自語同犯** (*svavacanavyādhikaraṇa* — "the same fault rebounds on one's own words"): a refutation is only a refutation if it does not, in the act of refuting, commit the very same fault it alleges against the opponent. If your argument only ends up dismantling your own position, you have not refuted anyone — you have produced a 似能破. The text's vivid examples: a magnet that has not "joined" the iron still only attracts the iron it can attract; an eye that has not "joined" a colour still only sees the colour it can see. To argue that "because an unjoined cause can produce an unjoined effect, therefore any unjoined cause can produce any unjoined effect" is to commit exactly the kind of over-generalisation one is accusing the opponent of. Worse: if such a refutation is taken as the model of how to argue, the result is "無理而謗一切法" — "without reason slandering all dharmas." This is the deepest danger of 似能破: not merely failing in one argument, but establishing a *method* of failure that, if followed, would undermine the legitimacy of reasoning altogether.
The text then sharpens this with a procedural point: 「誰定有宗乃有過，我無此宗故無失」 — "The fault only applies to one who holds the thesis; I do not hold that thesis, so the fault is not mine." Refutations presuppose a thesis on the part of the one being refuted. If the would-be refuter attacks a position that no one in the debate actually holds — perhaps a straw man, perhaps a position they themselves once held and abandoned — the accusation has no target. The third text therefore pairs 似能破 with the figure of the 破法人 — "one who only refutes, never establishes" — someone who has no positive thesis and so cannot legitimately be refuted, but who also cannot legitimately refute, because refutation requires a position to refute *from*.
Holding all three together, we can say: 似能破 is any refuting utterance that fails on at least one of three tests: it names a fault that is not there (Nyāyapraveśa), it invokes a logical criterion it has not actually shown to be unmet (Nyāyabindu), or it commits in its own wording the very fault it alleges against the opponent (Madhyamakāvatāra-bhāṣya).
Walking It Through in Daily Life
The most useful way to test this concept is not to look for it in classical debate manuals but to look for it in the conversations we are already having — and the conversations we are watching others have online.
Scene 1: The work meeting. A colleague presents a proposal. The structure is clear: "We should shift the product line toward mid-market users because (i) our existing enterprise clients are saturated, (ii) three competitors have already opened that lane, and (iii) it leverages our existing engineering capacity." The manager cuts in: "This whole thing is just wrong." No specification of which part. No identification of which of the three reasons is unsupported. No mention of any of the six classical fault categories. The pronouncement *looks* like a refutation. It is the verbal shape of one. The colleague, the team, and the manager all leave the room with the sense that something has been critiqued. Nothing has been. This is a 似能破 of the Nyāyapraveśa type: a refuter naming a category of fault without there being one. The remedy, in the classical terms, would be: state *which* of the three reasons is "not established" (不成因), *which* of the claims in the thesis is "faulty" (宗過), or *where* the example used fails (喻過). Until that is done, the refutation is only performative.
Scene 2: The internet debate. Two people disagree about, say, a policy question. A responds with a long paragraph. B replies, not by addressing any specific claim A made, but by saying "you've just committed a logical fallacy." B does not name the fallacy. Does not show how A's argument exhibits it. Does not even confirm that the structural pattern B has in mind is actually in A's text. B has produced a *gesture* of refutation. In the Nyāyabindu frame, this fails the test of *demonstrating* a missing characteristic of A's reasoning — the refuter has merely *announced* that one is missing. In the Madhyamakāvatāra frame, this often also fails the self-rebound test: B's own dismissal, if examined, would be subject to the very same formal complaint (an unsupported assertion) that B has levelled at A. The conversation then spirals because both parties now believe refutation has occurred when only its *resemblance* has.
Scene 3: The social-media pile-on. Someone posts an opinion. The replies are filled with identical-pattern objections — "you clearly haven't thought about X," "this is so obviously wrong about Y," "any reasonable person can see Z" — applied to hundreds of different posts regardless of what those posts actually argued. Each reply *names* a fault category (logical oversight, factual error, normative inconsistency) without ever checking whether the fault is present in the particular post under consideration. The classical diagnosis is exact: this is 能立缺減 style — faulting a complete probandum as deficient, treating *categories of possible fault* as if their mere invocability were evidence of actual fault. The deeper problem, in the third text's terms, is that the pile-on is also a form of 破法人 behaviour — there is rarely a positive thesis being defended by the pile-on, only a pattern of rejection being performed. The Madhyamakāvatāra warning becomes very concrete here: a society that takes this method of refutation as its model eventually trains itself into "無理而謗一切法" — rejecting every position without ever establishing why.
Scene 4: The intimate argument. A couple are arguing. One says: "You never listen to me." The other responds: "That's just not true, you're being unfair." Each is *performing* refutation. The first names a pattern of behaviour and labels it as a fault. The second *denies* the pattern and labels the accusation as a fault. But neither has done the Nyāyapraveśa work: what *specifically* would count as listening? In which *particular* moment was it absent? What would be the 喻過 — the counter-example showing the claim is overstated? In the Madhyamakāvatāra frame, we can also ask: does the second person's rebuttal itself exhibit the same over-generalisation it accuses the first of? ("You never listen" is an absolute claim; "I always listen" is its mirror absolute claim — and the absolute structure is the same fault.) When two parties in an argument are both committing 似能破 simultaneously, the refutation is symmetric, and no actual ground has been shifted.
Scene 5: The AI assistant and the question of "merely plausible" rebuttal. A useful contemporary analogy, with caveats. Large language models, trained on the surface patterns of disagreement, can produce responses that *look* like refutations — they marshal counter-claims, deploy logical-sounding vocabulary, structure their replies in the rhythm of a rebuttal. But the patterns are pattern-matched, not criterion-checked. They can name a fallacy category without verifying the category applies. They can produce a counter-example that contradicts only a misreading of the original claim. This is structurally identical to the 似能破 problem the classical texts were diagnosing — the *form* of refutation detached from the *criterion* of refutation. (It is worth being cautious here: this is an analogy about *form*, not a claim that contemporary AI systems instantiate Buddhist categories. The underlying mechanism is different. But the diagnostic shape is illuminating: when a system's training optimises for the *appearance* of successful refutation, the system will excel at producing 似能破 at scale.)
A common mistake when people first hear this concept is to map it onto "ad hominem" or "straw man" — both of which are specific informal fallacies. The classical category is wider. 似能破 covers *any* case where the formal structure of a fault-claim is in place but the fault itself is not actually demonstrated, whether or not the surface rhetoric is abusive or caricatured. Equally, the popular notion that a 似能破 is "just a bad argument" is too loose: the classical texts are precise about *which* features must be missing for the designation to apply.
Why Contemporary People Need This
The reason this term deserves more attention than its forbidding Sanskrit-Tibetan-Chinese surface suggests is that we live in what can only be called a saturated refutation environment. Refutations are everywhere; refutations that actually *do* their job are rarer. And the gap between the two has consequences that extend well beyond philosophy seminars.
Consider three domains where this matters.
In public reasoning. Democratic conversation depends on the assumption that objections carry weight because they identify real faults in real arguments. When the public sphere is dominated by refutations whose structural appearance has decoupled from their substantive content — accusations of fault that never specify what the fault is — the medium itself degrades. People learn that the way to participate in a debate is to perform refutation, not to actually refute. The classical diagnosis is exact, and the remedy is specific: name the fault; show the fault; check that the fault does not rebound on the accuser. The training in these three moves is exactly the training that 似能破, as a concept, provides.
In our inner lives. Much of what passes for self-critique is also 似能破. We tell ourselves "I'm such a failure," "this whole project is worthless," "no one really likes me" — these are refuting utterances aimed at ourselves or our situation. They have the *form* of refutation (they name a fault), but they have not done the work of identifying where the fault actually is, in what specific instance, with what counter-evidence. The Nyāyabindu distinction is useful here: there is a difference between *genuine* recognition that a position is unsupported (不成) and mere diffuse 疑惑 — a mood of doubt that hasn't yet located its object. Confusing the two is a particularly human form of 似能破 against oneself.
In the design of systems that argue. As institutions of all kinds incorporate algorithmic mediation of disagreement — recommendation engines optimising for engagement, AI assistants producing argument-like outputs, automated moderation flagging arguments as "fallacious" without showing why — the question of whether the system is optimising for *real* refutation or for *appearance* of refutation becomes structurally important. The classical texts give us the vocabulary for asking the question precisely: is the system's output actually demonstrating the missing criterion (the genuine 三相 analysis), or is it performing the *gesture* of demonstration? This is one of the rare places where a two-thousand-year-old technical term from Buddhist logic has direct diagnostic purchase on a contemporary design problem.
The concept also matters because it is one of the few places where Buddhist philosophy directly addresses the *ethics of disagreement*. 似能破 is not just an error in reasoning; it is, in the Madhyamakāvatāra framing, a form of potential *slander* ("謗") — an attack that, if taken as a method, "slanders all dharmas without reason." This moves the concept out of the logic classroom and into the territory of how we owe each other accuracy when we disagree.
Common Misreadings and Clarifications
Misreading 1: "似能破 just means a weak or unpersuasive refutation." No. The classical texts are explicit that the test is not strength, sharpness, or persuasiveness, but *accuracy*. A quiet, polite, well-mannered refutation that points to a fault which is not actually present is still a 似能破. A loud, rude refutation that points to a fault which *is* actually present is still a 能破. The error is in the referent, not in the tone.
Misreading 2: "If both sides disagree, at least one must be committing 似能破." Not necessarily. Two parties can disagree in good faith; each can present a positive thesis, and the disagreement can be a real disagreement about a fault whose presence is genuinely contested. 似能破 requires a *false claim of fault* — saying there is a flaw when there isn't. Mere disagreement, where both parties have reasons and the question is open, is not yet 似能破 on either side. (It may, of course, *become* 似能破 if one party claims certainty about a fault they cannot demonstrate.)
Misreading 3: "似能破 is the same as ad hominem or straw man." It is a *parent category* that includes those informal fallacies and many others. Ad hominem is one specific pattern (attacking the person rather than the argument); straw man is another (attacking a distorted version of the argument). 似能破 covers these and additionally covers cases where a person attacks the *actual* argument but misidentifies what is wrong with it — for example, calling a thesis "internally inconsistent" when the inconsistency is not there, or declaring a reason "contradictory" when the contradiction has not been shown. The classical category is broader because it is tied to a specific inventory of possible faults (不成因, 不定因, 相違因, 喻過, etc.) rather than to surface patterns of rhetoric.
Misreading 4: "If my opponent's argument really has a fault, my refutation is automatically 能破." Almost, but not quite. The third text's 自語同犯 principle reminds us that a refutation can identify a real fault in the opponent's argument while still committing the same (or a structurally similar) fault in its own wording. "You have over-generalised!" is a real diagnosis if the opponent has in fact over-generalised — but if the refutation itself is then over-generalised in applying the diagnosis (e.g., extending it to all the opponent's arguments on every topic), it commits the very fault it identifies. This is the Madhyamakāvatāra insight: a refutation must be self-consistent *as well as* accurate.
Misreading 5: "似能破 is a problem only for the refuter; the opponent is fine." Not quite. The Nyāyapraveśa framing implies that when a 似能破 is launched at a genuinely faultless position, the opponent is in fact protected — the refutation fails on its own terms. But the Madhyamakāvatāra warning is broader: when a 似能破 style of argument is taken up as a *method* by a community, the cumulative effect is corrosive even to those who never personally commit it. A public sphere in which the *pattern* of unsupported refutation is normalised trains everyone — including those whose own positions are sound — to live inside an environment where actual reasoning is no longer recognised. The damage is to the *medium* of argument, not just to the immediate exchange.
Misreading 6: "The remedy for 似能破 is to be more polite." Politeness is unrelated to the diagnosis. The remedy is structural: when you suspect a fault in an argument, do the work of (i) identifying *which* of the classical categories the alleged fault belongs to, (ii) showing *why* the specific claim, reason, or example in front of you exhibits that fault, (iii) checking that your own wording does not commit the same or an analogous fault, and (iv) ensuring your refutation is actually aimed at a position someone in the exchange holds. The classical texts are, in this sense, not gentle in tone but exacting in procedure. They expect refuters to do work.
A final note on how this concept sits in the wider system of Buddhist thought. 似能破 belongs to a family of terms — 能立, 能破, 似能立, 似能破, 所破, 自語同犯, 餘能破 — that together describe the *discipline* of how a bodhisattva or a serious practitioner is supposed to handle disagreement. In the Madhyamakāvatāra tradition, this discipline is not optional: the entire project of establishing the Middle Way over the philosophical extremes depends on refutations that actually refute. A practitioner who has only learned the *appearance* of refutation is, in a precise sense, not yet equipped to communicate the Middle Way to others, however eloquently they may phrase their objections. The classical demand is that we be able to point to the specific missing condition, in the specific argument, with wording that does not itself exhibit the fault, against a position someone actually holds. It is a high bar. The category of 似能破 exists precisely to remind us how often we fail to clear it — and to give us a name for the failure that is more precise than "being wrong." It is the failure of *looking* like we are right when, on the criterion that matters, we have not yet begun to show it.