Two men, two good deeds, two opposite verdicts. And the verdicts are the wrong way round.

The state of Lu had a law. Any citizen who paid to redeem a fellow citizen from slavery in another state could claim the money back from the treasury.

Zigong redeemed someone and refused the reimbursement. Paid it himself, wanted nothing, told nobody it was owed to him.

Confucius criticised him.

Zilu pulled a drowning man out of a river. The man’s family gave him an ox in thanks. He took it.

Confucius praised him.

Sit with that for a second, because your instinct has it inverted, and the inversion is the point.

The Reasoning

Here’s what Confucius said, in substance.

Zigong was wealthy. Refusing reimbursement cost him nothing he’d notice. But he had just established a standard — that the noble thing is to redeem someone and refuse the money.

Now the next citizen of Lu who sees a countryman in slavery has a problem. Claim the reimbursement and look grasping next to Zigong. Or pay from his own pocket, which he cannot afford.

So he does neither, and walks past.

Zigong’s refusal made it more expensive to do the right thing, so fewer people did it, so more people stayed enslaved.

Zilu accepting the ox did the opposite. He demonstrated that saving a life is rewarded — that you can be a decent person and not be out of pocket for it — and everyone watching updated accordingly. More people go into the river.

One man performed a more admirable act and produced a worse world. The other performed a less admirable one and produced a better one.

The Principle

The moral quality of an action is determined by its effects rather than by its appearance, and the relevant effects include those on people observing it. An action may be admirable in the instance and harmful in aggregate, most commonly by establishing a standard that others cannot meet, in which case the standard suppresses the behaviour it appears to model. This produces a systematic error, since the actions that appear most virtuous — refusing compensation, absorbing costs silently, exceeding what is required — are precisely those most likely to set unaffordable standards. The corresponding test is not whether an action is admirable but what it makes more or less likely for those who see it. Actions taken to demonstrate virtue are particularly susceptible, since demonstration is aimed at observers and observers are exactly who the aggregate effect operates on.

Four ways this goes wrong. The first is Zigong’s.

One: Setting an Unaffordable Standard

The Zigong error, and it’s everywhere.

You do more than required, at your own cost, quietly. It’s admirable. And it resets what “normal” looks like for everyone who sees it.

The colleague who works weekends without being asked has established that working weekends is what committed people do. Nobody said so. The standard is now in the room, and the person with childcare who can’t match it is now visibly less committed.

The person who never claims expenses has made claiming expenses look grabby.

The manager who is always available has made ordinary availability look distant.

None of them intended any of this. Every one of them raised the cost of behaving normally.

And this connects directly to something you may be doing. If you’re the person who never says no — who absorbs the extra, quietly, at personal cost — you’re running the Zigong error at scale. You’ve set a standard of availability, and your colleagues are being measured against it, and the honest answer is that some of them can’t afford it. That’s the cost of being useful, and this is the part of it that isn’t about you.

Two: Removing a Signal Others Needed

The second, and it’s subtler.

Sometimes a cost is doing information work. It tells people something true — that this is expensive, that it isn’t sustainable, that resourcing is short.

Absorb it silently and you’ve deleted the signal.

The team that quietly makes an impossible deadline has told the organisation that the deadline was possible. The next one will be tighter. Nobody lied; a data point was suppressed, and decisions upstream are now being made on false information.

This is the mechanism by which heroic effort damages the system it’s rescuing. The rescue is real. The information loss is also real, and it compounds.

Three: Solving the Instance, Preserving the Cause

Third, and it’s the most common in ordinary working life.

You fix the problem. The problem was a symptom. Fixing it removed the pressure that would eventually have forced the underlying thing to be addressed.

Every time you patch the broken process personally, the process survives another quarter. Every time you cover for someone, the gap they leave stays invisible.

You’ve done a genuinely good thing and extended the life of the bad thing.

I’d be careful with this one, because it’s easily misused as a justification for letting things fail out of spite. The distinction: are you declining to fix it because the failure would surface information that leads to a repair, or because you’re annoyed? Those look identical from outside and are very different acts.

Four: Generosity That Creates Dependence

The fourth, and it’s the one people find hardest to hear.

Help given repeatedly in a form that prevents someone developing the capability to handle it themselves is a net cost to them, delivered as kindness.

The colleague you keep rescuing is not learning. The person for whom you always make the introduction never builds the network. The friend whose problems you consistently solve stops solving problems.

There’s an obvious limit here — sometimes people need help and are not in a position to learn anything, and telling them to develop capability is cruelty dressed as principle. This isn’t an argument against help.

It’s an argument for noticing which kind you’re giving, and being honest that repeated rescue frequently serves the rescuer’s sense of themselves as much as it serves anyone.

The Two Tests

Practically, two questions, and they’re both about people other than you.

What does this make more or less likely for the people who see it?

Not “is this admirable.” What behaviour does it license or suppress in observers?

Could a reasonable person in my position afford to match this?

If the answer is no, you’ve set a standard that will suppress rather than model. Which means either you should do it visibly and make the cost visible — “I did this and it cost me a weekend, which is why I don’t think it’s repeatable” — or do it quietly enough that no standard is set.

Zilu, incidentally, passes both. Accepting the ox made saving people more likely, and any reasonable person could match it, because it costs nothing.

The Unsettling Part

Here’s what this does to you if you take it seriously.

The appearance of virtue stops being evidence of it.

You can no longer tell whether you’re doing good by checking how the action looks or how it feels. Both of those instruments have just been shown to point the wrong way in exactly the cases that matter most — the noble refusal, the silent absorption, the uncompensated effort.

And the actions most at risk are the ones you’re proudest of. The times you didn’t take the credit. The times you covered the cost yourself. The times you did more than anyone asked and said nothing about it.

Some of those were genuinely good. Some of them raised the price of decency for everyone watching, and you have no reliable way of telling which without doing the aggregate analysis, every time, on actions you’d rather just feel good about.

That’s the unsettling part, and there’s no version of this that resolves it comfortably.

What This Is Not

It is not an argument for self-interest. Zilu took the ox, but the story isn’t about him getting an ox. It’s about what accepting it made possible for everyone else. The frame throughout is aggregate consequence, not personal advantage.

It is not a licence to stop helping. Three of the four failure modes involve helping in the wrong form, not helping too much. The response is adjusting the form, not withdrawing.

It is not a claim that intentions don’t matter. They matter for what they say about you and for what you’ll do next time. They just aren’t the thing that determines the effect, and the effect is what the world receives.

The Cost

Two prices, and the second is heavy.

First: you lose the ability to be sure you’re doing good.

The comfortable version of ethics is that a good act is one that feels right and looks right. Give that up and every significant action requires an aggregate check, which is effortful, uncertain, and frequently unresolvable.

Second: some of what you’re proudest of gets reclassified.

Run this over your own history and the quiet uncompensated generosity, the standards you held at personal cost, the times you absorbed something rather than raise it — some of that will turn out to have made things harder for people around you.

That’s not a comfortable audit and it doesn’t come with a way to fix the past.

Common Questions

Can doing the right thing produce a bad outcome?
Routinely. An action can be admirable in the instance and harmful in aggregate, most often by establishing a standard others can’t afford to match — which suppresses the behaviour the action appeared to model.

What is the story of Zilu and Zigong?
Two accounts from the state of Lu. Zigong redeemed a citizen from slavery and refused the legal reimbursement, and Confucius criticised him for making it costlier for others to do the same. Zilu accepted an ox for saving a drowning man, and Confucius praised him for demonstrating that saving lives is rewarded.

Why would refusing a reward be wrong?
Because it sets a standard that the next person may not be able to afford, and rather than declining the reward they decline the action. The refusal looks more virtuous and produces fewer good outcomes.

How do I judge whether an action is actually good?
Ask what it makes more or less likely for the people who see it, and whether a reasonable person in your position could afford to match it. If they couldn’t, the action will suppress rather than model.

Does this mean I should stop being generous?
No — three of the four failure modes are about the form of help rather than its quantity. The adjustments are making costs visible rather than absorbing them silently, and noticing when repeated rescue is preventing someone from developing capability.

Isn’t this just a rationalisation for taking things?
It can be misused that way, which is why the tests are framed around effects on other people rather than on you. Zilu’s ox mattered because of what it demonstrated to observers, not because he got an ox.

Where to Start

Ten minutes. One action, examined.

Pick the thing you do that you’re quietly proudest of — the cost you absorb, the standard you hold, the thing you never mention.

Then answer two questions in writing:

Who can see this, and what does it teach them?

Could they afford to match it?

If the honest answer to the second is no, you’ve got a Zigong on your hands, and you have two options rather than one.

You can stop doing it, which is sometimes right and usually feels wrong.

Or — and this is the better option most of the time — you can make the cost visible. Keep doing it, and say what it costs. I got that done, and it took the weekend, which isn’t something we should plan around.

That single addition converts an unaffordable standard into an honest data point. The good act stays. The suppressive effect goes.

It’s a small change and it’s the whole difference between Zigong and Zilu.

I’ve spent twenty-five years working on how ordinary people develop capabilities that look impossible from the outside, and one of the more uncomfortable things I’ve had to accept is that the most admirable-looking people in an organisation are frequently making it harder to work there.

Not through anything they did wrong. Through what everyone else concluded from watching them.