Ordinary people in the world only like the outward shape of purity and perfection, and so they don't understand what it really means to be a Buddha or Bodhisattva who might suffer outward lives that seem simply miserable. Actually, a great Buddha in the world won't seem so pure precisely because of the methods he might have to use in order to save people. It's most likely that he'll be a combination of saint and hero, Mafioso and thief, for this is the combination that would make their actions the most effective.1
That isn't a metaphor about inner states. It's about methods — about what somebody who actually gets things done in the world has to be willing to do.
And it comes with the reason attached: "At times acting the part of a Mafia don is the only way to get positive things accomplished in the world, so the format of a Bodhisattva's actions are something that cannot be predicted."1 This is the most dual-use passage in the book, and it's worth handling with the guardrails visible rather than either quoting it approvingly or skipping it.
He does specify what each role contributes, which is more than most versions of this argument manage:
in order to establish the good, a Buddha will lean on others like a Mafioso or redistribute things like a thief, recognizing that everything taken must eventually be repaid. To lead a group of asuras you must act like the asuras, but you must act like an asura-hero.2
Mafioso — lean on others. Pressure. Making it costly for someone to do the wrong thing. Thief — redistribute. Moving resources from where they are to where they are needed, without the owner's consent. Saint and hero — the ends and the willingness to bear the cost.
Two clauses in that sentence are doing restraining work and are easy to skim past. Recognizing that everything taken must eventually be repaid. The karmic accounting isn't suspended. The redistribution generates a debt for the redistributor.
You must act like an asura-hero. Not merely like an asura. The mode is adopted; the aim isn't.
Bodri sees the obvious problem and addresses it in the very next sentence:
Don't take this the wrong way and think it's a license to start engaging in immoral deeds. I'm only pointing out that you must recognize the realities of any situation you enter, and become pragmatic in determining how you should set about to accomplish positive change and transformation.2
And he names the price:
your mind must remain in the transcendental and you must be ready to bear the karmic consequences of whatever you end up doing.2
That guardrail is worth something: the bad karma is incurred, not excused. The Bodhisattva doesn't get the thing every dirty-hands argument reaches for, which is that a good enough end washes the act clean. It stays dirty; the actor pays.
But it's weaker than it looks, and here's why. Every element of it is internal. Recognise the realities. Remain in the transcendental. Be ready to bear the consequences. All three are conditions on the actor's inner state, unobservable from outside, and self-certified.
So there's a strong warning here and no external check. And the population that most needs an external check is exactly the population that will find this passage congenial.
The historical example is the most quotable thing on the page and the least defensible:
Remember the lesson of Gandhi who said he would have used guns to free his people, but the problem was that he just didn't have them. That's why he came up with his strategy of non-violence!2
Two things are packed in there, and they pull in different directions. One is about who can act; the other is about what acting costs.
The attributed statement. Gandhi did write about violence and cowardice — most famously that where the choice is only between cowardice and violence, he would advise violence. That strand runs through his writing and it is genuinely hard.
The causal claim, which is Bodri's own. That's why he came up with his strategy of non-violence — non-violence as a workaround for lacking weapons rather than as a principle. That's not what the attributed remark says, it's unsourced, and it's contradicted by the enormous body of Gandhi's writing treating ahimsa as prior to tactics.
The exclamation mark is doing the argument's work. Handling: recorded as Bodri's reading, [CONTESTED], and not carried as a fact about Gandhi. This is a clear instance of the book's documented pattern — careful about the states themselves, confident and unreliable wherever it borrows a historical figure to make a point.
The point the example is supposed to make survives without it, and it gets stated cleanly one sentence later: "Be idealistic in your dreams and vows, but be wise and pragmatic in how you try to accomplish them."2
The argument extends from bodhisattvas to leaders generally, and here it's on much firmer ground:
When people want their leaders to be stainless of character flaws and defects, this type of expectation always just makes me laugh. People all have impure karma, so it's simply impossible. … if the President of a country wasn't a dealmaker or individual who could deal with power politics, then he'd probably never be able to climb to that position in the first place.
Without the ability to bluff, lie, threaten and bully when necessary, a leader might even be placing the security of his country at risk.3
The selection argument in the middle is the strong one and it's not a moral claim at all: the process that produces leaders selects for the capacities the role requires, so expecting someone who arrived through that process to lack those capacities is expecting a contradiction. And the conclusion he draws about spiritual teachers follows the same logic: "the best spiritual teachers always turn out to be multi-talented, like the great Zen masters like Lin-chi, Yun-men and Kuei-shan. These individuals weren't just great monks, but had the mettle to be generals or kings."3
The chapter's summary makes the whole passage a claim about willingness to be soiled:
the Bodhisattva is willing to wear this type of external pollution for the greater good, at times doing what is difficult to do and incurring bad karma for the benefit of others. This is their demonstration of real wisdom as well as compassion. This is what separates them from the Arhats who always want to remain pure. Being liberated means that you don't hold onto holiness and purity, nor fear the world of interdependent phenomena with its inescapable karma.4
That reframes purity itself as an attachment — holding onto holiness — which is consistent with everything else in this corpus about clinging, and which is the strongest version of the argument available. And it has an iconographic payoff he can't resist: bodhisattvas are painted beautiful "due to the merits they've accumulated because of the troubles they've become involved in on behalf of others," while arhats are "usually shown as being poor with strange faces and dirty clothes" because "all they've done is to stay out of public involvement."5
You're in a situation where the clean option won't work. Something needs to happen for somebody else's sake — a resource moved, a person pressured, a truth softened — and the version of you that keeps its hands clean will simply fail to accomplish it. This isn't rare. It's most of institutional life.
The instruction here is genuinely usable. It's a two-part test, and the order matters.
First: name the act accurately. Not managing the situation, not doing what was necessary. If you're leaning on someone, the word is leaning. If you're shading the truth, say shading the truth. The single most reliable marker of the failure mode is euphemism, because a redescribed act doesn't get accounted for.
Second: accept the bill in advance. The framework's whole guardrail is that the karma is incurred. So before acting, name what it will cost — the relationship, the precedent, the thing you'll have become slightly more comfortable doing. If you can't name a cost, you haven't identified the act.
An act you can name and whose cost you've priced is the passage's actual subject. An act you've renamed and whose cost you've argued away is what the passage will be used for. Then a check for the long run, and it's uncomfortable. This reasoning has a ratchet. Each use makes the next one easier, and the transcendental-mind condition is unobservable, so nothing external stops the slide. The only available brake is frequency. If the situation that requires dirty hands is arising monthly, the framework is no longer describing your circumstances; it's describing your habits.
The passage's furthest example isn't the Mafioso. It's a page earlier, and it goes further than anything in the four roles:
in one of Buddha's past lives he was a ship captain who knew the intentions of a murderous criminal to kill a large number of travelers, and therefore killed the criminal first to save him from creating such great negative karma.6
The story is a canonical Mahāyāna text, and note the justification Bodri gives for it: the killing is framed as being for the murderer's benefit — sparing him the karma of mass murder — rather than for the passengers'. That framing is doing something specific and it should be seen clearly. It converts a lesser-evil calculation into an act of compassion toward the person being killed, which removes the last trace of moral remainder. Nobody is wronged. The killer is helped.
Set that beside the guardrail — you must be ready to bear the karmic consequences — and the two don't fit. If the act is compassion toward its victim, what consequences are being borne? The story's own logic has dissolved the cost that the chapter's guardrail depends on.
Three things to carry. This is the exact point where the argument stops being a description of hard choices and becomes an engine. Every step is individually reasonable — foresight, weighing, acting, accepting the cost — and the destination is a justified killing framed as a favour to the deceased.
The pattern isn't unique to this tradition and that's the point. Every ethical system with a lesser-evil provision generates a version of this, and every one of them eventually produces someone who applies it too readily. The structural question is never whether the reasoning is valid but what stops it running.
And this chapter's stopping mechanism is entirely internal. Attainment, transcendental mind, willingness to bear consequences — three self-assessed conditions. The tradition supplies the reasoning and outsources the restraint to the reasoner's own estimate of their attainment, which is the arrangement most likely to fail in precisely the people most confident of it.
The vault's position: the diagnosis is carried, the reasoning is recorded as the source's, and the ship-captain story isn't offered as ethical instruction. What's genuinely valuable here is narrower and it's the selection argument — that expecting effective actors to be unsoiled is expecting a contradiction — which stands without any of the licensing.
Strongest support: the selection argument is sound and doesn't require any of the theory behind it. Processes that produce effective actors select for the capacities effectiveness requires.3 Second: the purity-as-attachment reframing is consistent with the whole corpus. You don't hold onto holiness and purity is the clinging diagnosis applied to virtue itself.4
Third: the karma is explicitly not waived. The dirt stays dirt and the actor pays, which is more than most dirty-hands arguments concede.2
Tension: the guardrail is entirely internal — three unobservable, self-certified conditions, in a passage licensing bluffing, threatening and worse.2
Tension: the ship-captain framing dissolves the cost the guardrail depends on.6 ⛔ The Gandhi causal claim is unsourced, contested, and contradicted by the bulk of Gandhi's writing. Recorded as the source's reading; not carried.2 [CONTESTED]
Open question. What would a non-internal check on this reasoning look like inside a contemplative tradition? Sanghas have vinaya, confession, and expulsion — none of which appear here, and the passage's whole framing (the format can't be predicted) is designed to defeat external assessment. Not addressed in the range read (pp 647–650).
The Defeated Bodhisattva is the same two pages read from the other side, and the pair is more troubling than either half. That page establishes that a bad outcome isn't evidence against a person; this one establishes that bad behaviour may be the method. Held together: neither the results nor the conduct of a bodhisattva can be assessed from outside. The two pages jointly remove every observable handle, and neither notices that the other has removed the one it left standing.
Against The Arhat's Remainder, this page supplies the polemical version of the same contrast and adds a specific charge: the arhat's purity is "holding onto holiness" — an attachment rather than an achievement. That's the sharpest form of the Mahāyāna critique available, and it's worth noting that it's made in a passage arguing for the legitimacy of threatening people. And against Why Bodhisattvas Are Painted Beautiful, this page gives the iconography a second and less flattering reading. That page has the beauty as accumulated merit made visible. Here the same contrast — beautiful bodhisattvas, ragged arhats — is deployed as an argument for engagement over purity, which makes the iconography partisan rather than descriptive.
Kuṭa-Nīti starts from a naming decision. The author of the Arthaśāstra could have published under his birth name, Viṣṇugupta. He chose Kauṭilya — from kuṭa nīti, crooked policy, the methods that bend or break the dharmic rules. A man who spent decades teaching dharma to kings published the operational manual under a name meaning the one who uses immoral strategies, and the naming is treated as doctrine rather than as biographical accident.7
That's the same territory as this page and it handles it completely differently, which is what makes the pairing worth having. Here the argument is that the crooked methods, used with the right mind for the right ends, are what a Buddha's actions look like — the categories are absorbed into sainthood. Kauṭilya's move is the opposite: he keeps the categories apart and puts the label on himself. The methods stay crooked. The manual stays a manual of crooked policy. Nobody is renamed as a saint.
The insight the pairing produces is a discipline neither source states. The safeguard against dirty-hands reasoning isn't a better justification — it's refusing to relabel the act. Kauṭilya's naming keeps the moral remainder permanently visible, in the author's own name, where it can't be argued away. Bodri's guardrail tries to do the same work by requiring the actor to bear the karmic consequences, but he simultaneously supplies the language — saint, hero, asura-hero, Buddha — that makes the act describable as holy.
A framework that lets you call the leaning-on "skilful means" has already spent the protection the accounting was supposed to provide. The Sanskrit strategist's solution is cruder and considerably harder to game: name the method for what it is, and wear the name.
The moral-agency page opens on the explanatory abyss under blame: at what point in a causal chain does someone become responsible? The categorical judgement — they are responsible, they should be punished — sits on a causal story that doesn't have a clean joint in it anywhere.8
Bring that to the ship captain. He kills a man to prevent a mass murder and to spare the murderer the karma of committing it. Where in that chain is the responsibility? The captain acted. The murderer had not yet acted. The passengers were unharmed because of an act performed on their behalf that none of them authorised.
The correspondence is that both accounts locate a place where responsibility is supposed to attach and find that it won't sit still. The clinical version finds no non-arbitrary joint in the causal chain; this chapter finds that the same act is simultaneously a killing, a rescue, a favour to the victim, and a karmic debt for the actor. What the pairing produces is a reason to be suspicious of the smoothness of the ship-captain story rather than of its conclusion. A hard case should feel hard. The absence of remainder — nobody wronged, the killer helped, the captain compassionate — isn't evidence that the reasoning is sound; it's evidence that the framing has been optimised until every party is accounted for. The moral-agency material suggests that in genuinely hard cases the accounting never closes cleanly, and that a story where it does has been told by somebody with an interest in it closing.
Which gives a practical marker for this whole class of reasoning: be most careful with the cases that feel resolved, and treat the residue of unease as data rather than as squeamishness to be trained out.
Sharpest implication. The tradition's own account is that an effective bodhisattva won't look like a good person, and it supplies no way for anybody outside to tell that figure apart from someone who has simply talked themselves into it. The reasoning is coherent, the guardrails are real, and every one of them is inside the head of the person using them.
Generative questions.
Is there any version of skilful-means ethics with an external check? Kauṭilya's naming discipline is one candidate and it's not a check so much as a refusal to launder. What would an auditable version look like — and would any tradition adopt it, given that it would make its exemplars assessable?
And: does the purity-as-attachment argument prove too much? If holding onto holiness is a defect, is there any conduct standard the argument can't dissolve? The chapter says don't take this the wrong way and offers no principle for taking it the right way.