Antibiotics for a chest infection: correct, necessary, the right thing to hand somebody. Antibiotics for a person with no infection: not a smaller good, and not neutral. A harm. Same molecule, same bottle, same dose, and whether it's medicine depends entirely on who's holding it.
Nobody finds that confusing about drugs. The Chan master Pai Chang says the same thing about statements, and it's much harder to take:
All verbal teachings are just remedies to cure illness. Because illnesses are not the same, therefore the remedies are not the same. That is why it is sometimes said that there is Buddha, and sometimes it is said there is no Buddha. True words are those that cure illness, and after the illness is cured, all of them are untrue statements. True words are false words insofar as they produce opinions; false words are true words insofar as they stop people's delusions. Because illness is unreal, there are only unreal remedies to cure them.1
The medicine comparison isn't imported. It's made explicitly a couple of pages later: doctors say one thing to one patient and another thing to a different patient with the same illness, for it all depends upon how to best motivate them to get well.2 Same illness, different words, and the difference isn't dishonesty.
There are three separate claims in that passage and they escalate. The first is that truth is therapeutic. True words are those that cure illness. The test for a teaching isn't whether it matches how things are — it's whether it works on the person receiving it. And the negative form is just as flat: if the words can't cure the sickness, they're false words.
The second is that truth expires. After the illness is cured, all of them are untrue statements. Not no longer needed — untrue. A sentence that was true while it was working becomes false once the work is done, because there's no longer an illness for it to be the remedy of.
The third is the one that bites. True words are false words insofar as they produce opinions. The same sentence, delivered to somebody who converts it into a position rather than a cure, has become a false word — by producing exactly the thing teachings exist to dissolve.
So a teaching can fail by being wrong, and it can fail by being right and held. The second failure has nothing to do with the content, which means no amount of getting the doctrine correct protects you from it.
The other direction is the one people skip, and it licenses something uncomfortable. False words are true words insofar as they stop people's delusions. A statement that isn't the case, said because it will dislodge something, is on this account true.
The worked example is Chao-chou:
When Zen master Chao-chao was asked if a dog has Buddha nature, he replied "No." Of course this isn't true from the ontological sense, but Chao-chao wasn't speaking of this sphere of things. Rather, as a master whose job was to help enlighten others, his response was simply an expedient reply for that particular student at that particular moment to help him awaken.1
The doctrine is unambiguous — every being has Buddha nature — and the answer given was no. And it isn't softened: of course this isn't true. It was the right answer because of what it did to the man who received it.
Which means two registers are running at once, and the passage knows it. There's a register where sentences are true or false about how things are. And there's a register where sentences are true or false about what they do to somebody. The teachers being described work almost entirely in the second. A student reading them in the first will find them contradicting each other constantly — that is why it is sometimes said that there is Buddha, and sometimes it is said there is no Buddha.1
The line that stops all this collapsing into say whatever works is the last one. Because illness is unreal, there are only unreal remedies to cure them. It's the clause that makes the whole thing consistent rather than merely convenient.
Suppose the confusion being treated were a real thing — a substance in a person, a defect in the machinery. Then a remedy would have to be true about that thing, and picking words for their effect would be a lie told for a good reason. But the confusion isn't a substance. It's a wrong reading somebody is holding. And a wrong reading isn't cured by a correct description of it; it's cured by anything that makes it let go.
So the remedies are unreal in the same way and for the same reason the diseases are. Nobody is deceiving anybody. The whole exchange happens at a level where the categories true-about-a-thing and false-about-a-thing were never the relevant ones, because there was no thing.
Which is also the limit on the licence. It doesn't extend to sentences about the world. Chao-chou wasn't lying about dogs; he was doing something to a student. And the general principle keeps one word doing the restraining: for progress in spiritual cultivation you should use whatever virtuous method works.2 Whatever works, and virtuous.
The practical instruction is a sorting job, and the three bins are named:
a spiritual great has a general plan to help society and disseminates a variety of teachings — some of which are higher than others and many which certainly aren't absolute… some are teachings on the ultimate, some are simple guidance mechanisms for society, and some are just responses to the situation of the day.2
Teachings on the ultimate. Guidance for running a society. Answers to a particular afternoon. Three bins, and the trouble is that they all arrive in the same voice, in the same book, from the same person.
The consequence of not sorting is put with the passage's own image:
you can't mix fish eyes and pearls together and say they have the same level of importance, yet that's what people ignorantly do.2
Fish eyes and pearls are the same size, the same colour, and about equally shiny in a bowl.
There's an argument for why no religion's teachings can be absolute, and it's one line. If they were absolute standards, then there would be only one religion in the world, and there isn't.2 With a caution attached, so it isn't licence to be dismissive: nevertheless, we must still be respectful of what they are trying to accomplish.
The examples are deliberately mundane. Sunday in Christianity is no holier than Saturday in Judaism, the Friday of Islam, or the fifteenth day of the lunar month in Chinese Buddhism.2 Four traditions, four days, each certain its own is the anointed one. The mechanism gets named too, and it isn't about religion at all. People attach to a rule just as officials end up following the letter of the law while forgetting the purpose for which the law was legislated in the first place.
The examples of the failure are as ordinary as the ones for the doctrine. Groups that refuse blood transfusions. Groups that refuse doctors. Groups that forbid coffee and tea. And a rule that forbids you, as an example, to flip a light switch during the Sabbath — a solution to a situation that was meant to change when the situation did, carried forward for centuries instead.2
The clearest picture of the whole doctrine in operation is a complaint about a man handing out photographs.
I said, "You are telling these drunkards, liars, and dacoits that they are real bhaktas. They will just go home and carry on their old behaviors."3
The objection is exactly the right one: the statement is false, the recipients are unworthy, and nothing will come of it. The answer isn't a defence of the statement's accuracy.
You don't understand me. If I tell a man he is a great bhakta, I am planting a seed. If a person already has the seed planted and growing, why should I plant another?3
So the words aren't a description that happens to be wrong. They're an operation performed on somebody, and the reason they go to the drunkards rather than to the devotees is that the devotees don't need them. A remedy handed to a person without the illness would be the antibiotic problem in reverse.
Then the mechanism, and then a number:
Some of them will remember what I said of them, and it will make them want to develop this quality in themselves. If ten out of a hundred are inspired in this way, it is a very good thing.3
Ten out of a hundred, and that's called very good. Which is the most honest thing on the page. Ninety of the hundred go home and carry on exactly as the objection predicted, and the practice is judged a success anyway, because the expected yield was never high and was never pretended to be.
The same logic runs the other way into silence. Buddha sometimes refused to answer questions, for he knew that the questioner might take his answer wrongly or pass this information onto others who would misinterpret it. And the thing being avoided is named: always the Buddha feared cutting off a student's wisdom life.2 A true answer, correctly given, can end somebody's capacity to keep looking — so not answering is itself one of the moves.
Take a teaching you hold and ask which of the three bins it goes in. Is this about how things ultimately are, is it a rule for keeping a community working, or was it somebody's answer to a particular situation a long time ago? Most people have never sorted, and the whole of what goes wrong afterwards follows from that. If you can't tell, that's information: it means the thing arrived in a voice that made all three sound identical, which is how they always arrive.
Then check the one you're most attached to against the second failure mode. Not is it right — you've probably already established that to your satisfaction. Ask whether it has turned into an opinion, something you hold and would defend, rather than something doing work on you. On this account that conversion turns a true sentence false without changing a word of it, and the tell is that you'd rather win the argument about it than be finished with it.
Then, when you're the one speaking. Before answering somebody's question about practice, ask what your answer will do to them rather than whether it's accurate. Those come apart more often than people expect. And the passage gives you the harder version of the permission: sometimes the honest move is to say nothing. A correct answer, given to the wrong person at the wrong time, can shut down the searching that would have got them there.
And set your expectations at ten in a hundred. If you're trying to move anybody — a student, a friend, a room — the rate quoted here for a genuinely effective method is ten per cent, described as very good. Which is worth having in advance, because the alternative reading of ninety failures is that the method was wrong.
A student asks a teacher a straightforward question and gets seven different answers.
"How many gods are there, O Yajnavalkya?" He replied: "As many as are mentioned in the hymn to the Vishvedevas, namely, three and three hundred, three and three thousand." "Yes," he said, and asked again: "How many gods are there really, O Yajnavalkya?" "Thirty-three," he said. … "Six," he said. … "Three," he said. … "Two," he said. … "One and a half," he said. … "One," he said.4
Three thousand three hundred and six, then thirty-three, then six, then three, then two, then one and a half, then one. Seven answers to one question, in descending order, each given only because the student said really and asked again.
The reading offered is that every one of them was correct. Yajnavalkya could have stopped at any one of these answers if his student had not pressed him further. The right answer depended on the student's ultimate level of understanding rather than any specific number, and each answer was a test of that level. When the student showed he could take a higher one, he got it.4 Which makes the exchange a working demonstration of the whole doctrine rather than an illustration of it. The teacher isn't withholding the truth and doling it out. He genuinely has seven remedies and is establishing which illness he's dealing with, and the diagnostic instrument is whether the patient comes back.
One and a half is the detail that proves it isn't a countdown to the right number. Nobody thinks there are one and a half gods. It's a rung, placed there because a student who has just let go of two isn't yet ready for one, and something has to be available in between.
And then the passage does the thing to its own example. Even one is wrong: if you say there is "one," that is already too much because "one" implies a one in relation to another, and there isn't another… Emptiness isn't even emptiness because it is empty of emptiness, and thus there's nothing at all.3 So the final answer, the one the whole descent was heading towards, is also a remedy — true while it was curing something, untrue once the cure landed. The staircase has no top step. Which is exactly what after the illness is cured, all of them are untrue statements predicts, applied to the sentence the reader was about to hold onto.
The strongest thing here is that the doctrine gets demonstrated rather than asserted. The Yājñavalkya exchange is a teacher running the procedure in public, with seven graded answers and a stated diagnostic. The one and a half rung is what shows it's a ladder rather than a countdown.4 The Neem Karoli exchange supplies the other half — the objection stated correctly, the answer that doesn't dispute it, and a success rate of ten per cent offered without embarrassment.3
Two tensions stay open. The framework says teachings are remedies with no absolute standing, and it's being delivered inside a six-hundred-page account of graded stages presented as how things actually are. Nothing marks which parts of that account are remedies. And use whatever virtuous method works leans the whole weight of the restraint on one adjective that never gets defined — the licence to say what isn't so is broad, and the only thing bounding it is a word.2
Flags. 🚩 SECONDARY WITHOUT PRIMARY on all four sources: the Pai Chang passage via Thomas Cleary's translation of Nan Huai-Chin's The Story of Chinese Zen, Tuttle 1995, p 132; the Yājñavalkya dialogue via William Stoddart, Outline of Hinduism, 1993, p 26 — a summary volume rather than the Bṛhadāraṇyaka itself; the Neem Karoli material without a locator in the text; and the Chao-chou case characterised rather than cited. ⚠ The Chao-chou reading is [CONTESTED] in the wider literature — the mu answer has several standard interpretations and this one is presented as settled. Everything here is [POPULAR SOURCE].
Open question: the three bins are named and no method is given for sorting anything into them. A teaching on the ultimate and a response to a situation arrive in the same voice, from the same person, in the same book. The passage says so. What would distinguish them for a reader who wasn't there?
Two attitudes to a received teaching run through this material, and they'd have you treating your own tradition very differently. On the first, hold it loosely or not at all. Every dharma is a medicine, no religion's rules are absolute, holy days are interchangeable, and this is also why some people can safely abandon religious or other dogma when they know what they're doing.2 Follow that and the sensible move is to keep what works this year and drop the rest without much ceremony.
On the second, in the same paragraph, hold it carefully. Nevertheless, we must still be respectful of what they are trying to accomplish2 — and the escape clause on abandoning dogma is when they know what they're doing, which is doing a great deal of quiet work. A person who doesn't know what they're doing, abandoning a rule because they've read that rules are expedient, is the failure this whole passage is about, wearing the doctrine as its justification.
Neither is withdrawn, and the second is only ever a clause. Read alongside The Error Is Misuse, Not Content, the same shape recurs — the thing that goes wrong is never the teaching and always what somebody does with it, which makes a doctrine about the non-absoluteness of doctrines exactly as misusable as anything else.
Business — Avoid Self-Labeling. That page's finding is about what happens when a description of yourself hardens into an identity: the label stops being a useful summary and starts being a thing to defend, and defending it costs you the flexibility the description was noticing in the first place. The operator's move is to keep descriptions provisional and disposable — accurate for now, dropped when they stop paying.
Set that beside true words are false words insofar as they produce opinions and the two are describing one mechanism from opposite ends. The Chan version is about statements handed to you by somebody else. The self-labelling version is about statements you make about yourself. In both, the sentence starts as something that does work, and in both it fails in exactly the same way — not by becoming inaccurate, but by becoming something held. And in both, the failure is invisible from inside, because the sentence is still just as true as it was on the day it helped.
What the pairing gives that neither side gives alone is a test that works on either kind, and it's behavioural rather than semantic. Don't ask whether the statement is still true. Ask what happens when somebody contradicts it. A description doing work produces curiosity — you'd want to know what they've seen. A description that's hardened produces defence, and the defending happens before you've considered whether they're right.
Which also explains why the two literatures prescribe the same odd remedy. Neither says find a better description. Both say let it go when its work is done, and both treat the willingness to drop a true thing as the actual skill — harder than acquiring it, and the part almost nobody practises.
One handshake is the section here. A second would restate the same mechanism from a neighbouring page rather than open new ground.
Sharpest implication. A teaching can fail by being wrong and it can fail by being right and held, and the second failure doesn't touch the content, so no amount of getting the doctrine correct protects you from it. Which means the useful question about anything you believe isn't whether it's true. It's whether it's still doing something, and the test is what happens in you when somebody says it isn't so.
The edge left open. Three kinds of teaching are named — the ultimate, the social, the situational — and no way of telling them apart is given. They arrive in the same voice from the same person, which the passage says outright, and the reader is told to sort them anyway. Fish eyes and pearls are the same size in the bowl, and nothing here hands you a way to weigh them.
Generative questions: