0
0
A school wants better reading scores. So it starts rewarding teachers whose classes score high on the reading test. A year later the scores are up. But the children are not reading more books. They have just got very good at the test.
A clinic wants shorter waiting times. So it sets a target: every patient seen within a set number of minutes. The numbers improve. Then someone notices patients are being “seen” by a quick hello at the door, and then waiting just as long as before.
You have probably lived a version of this yourself. At work, at school, maybe on your phone, chasing a step count or a number of likes. There is a name for it. It is called Goodhart’s law, and the short version most people know goes like this: when a measure becomes a target, it ceases to be a good measure.
Here is the strange part. The man the law is named after never wrote that sentence.
What Is Goodhart’s Law?
Goodhart’s law says that a number is useful for telling you how things are going, right up until people are rewarded or punished for that number. Then they start working on the number instead of the thing it was supposed to show. The number goes up. The real thing does not. And the number stops telling you the truth.
Think of a thermometer. It tells you whether a room is warm. Now imagine you get paid every time the thermometer reads high. The easiest way to get paid is not to heat the room. It is to hold a match under the thermometer. The reading looks great. You are still cold.
That is the whole idea. A measure is a window. A target turns it into a mirror, and people start polishing the mirror.
Where Goodhart’s Law Really Came From
Charles Goodhart is a British economist who worked as an adviser at the Bank of England. In 1975 he wrote a paper called “Problems of Monetary Management: The U.K. Experience,” published by the Reserve Bank of Australia. It was about money supply, interest rates and the hard job of steering a national economy.
In that paper he made a dry, technical point. His own words were: “Any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes.”
In plain words: if two things have always moved together, and you try to push one of them to control the other, they will stop moving together. Goodhart was talking about money statistics, not schools or hospitals or social media.
So where did the famous sentence come from? It came more than twenty years later, from a completely different field. In 1997 the anthropologist Marilyn Strathern wrote a paper about the way British universities were being audited and ranked. In it she wrote the line almost everyone now quotes: “When a measure becomes a target, it ceases to be a good measure.” And even she was building on the wording of another scholar, Keith Hoskin, who had made a similar point a year earlier.
Notice that the real line says “ceases to be,” not “stops being.” It is a small thing, but if you want to quote it, now you can quote it correctly.
There is one more twist. An American social scientist named Donald Campbell had described the same problem even earlier, with his own version, now called Campbell’s law. The more a number is used to make social decisions, he warned, the more pressure there will be to bend it, and the more it will bend the very thing it was meant to watch.
So the law named after Goodhart was not first said by Goodhart, is not quoted in his words, and was noticed by others around the same time. In a way that is fitting. The idea was true in so many places at once that several people saw it independently.
Goodhart’s Law Examples You Will Recognise
Once you know the pattern, you start seeing it everywhere.
- School tests. Tests are meant to show learning. Reward schools for scores, and class time shifts toward practising the test.
- Call centres. Average call length is meant to show efficiency. Make it a target, and some calls get rushed or passed along before the problem is solved, so the same customer calls back.
- Social media. Likes were meant to show that people valued a post. Once likes became the goal, posts started being made for likes.
- Fitness trackers. Steps are meant to show how active you are. Some people have been caught shaking their phones or wearing trackers on a pet to hit the goal.
- Artificial intelligence. AI systems are scored on test sets called benchmarks. When the benchmark becomes the prize, systems can get very good at the benchmark without getting much better at the real-world task. This is one reason the term has been searched so much lately.
The Rats of Hanoi
You may have heard the “cobra effect” story: colonial officials in India paid a bounty for dead cobras, so people started breeding cobras. It is a great story. It is also very hard to prove. The economist who made the phrase popular, Horst Siebert, told it in a 2001 book, and historians have struggled to find records that it actually happened.
But there is a real, well-documented case, and it is even stranger. In 1902, French officials in Hanoi, in what is now Vietnam, were afraid of plague. They paid a small bounty for every rat killed. To prove a kill, you brought in the rat’s tail.
Soon, officials began seeing rats running around the city with no tails. People were catching rats, cutting off the tails, and letting the rats go to breed more rats, and more tails. The historian Michael Vann has written about the whole episode in detail.
The tail was supposed to stand for a dead rat. Once the tail itself was paid for, it stood for nothing at all.
A good place to start: The Beginner’s Guide to Feeling God’s Presence Every Day (Free). A short video guide for noticing what matters in an ordinary day, the things no dashboard will ever count. Get the free guide. Free.
Why It Happens: People Are Not Thermometers
It is easy to read all this and think the answer is “people cheat.” Some do. But most of the time Goodhart’s law happens to honest people doing what they were told.
The teacher who drills the test is trying to keep a job. The nurse who rushes the hello is trying to meet a target someone else set. They are not villains. They are responding to the signal they were given. And the signal said: the number is what counts.
That points to the real problem. Numbers are a short summary of something much bigger. A reading score is a small slice of a child who can read. A step count is a small slice of a healthy body. The summary always leaves things out. When the summary becomes the goal, the things it left out stop getting attention, because nobody is counting them.
And that can feel personal. Many people searching for Goodhart’s law are not economists. They are workers who feel reduced to a number on a screen: calls per hour, items per shift, sales per week. They know they are more than that number. They just do not have the words for why the number feels so wrong.
If that is you, and work has started to feel like that, you might find I Hate My Job: How to Tell What You Actually Hate useful. Sometimes what you hate is not the work. It is the scoreboard.
So Should We Stop Measuring?
No. And this is where most articles about Goodhart’s law go wrong.
Throwing away all measures does not fix anything. Without numbers, you cannot tell if a clinic is getting slower, or if a child is falling behind. A world with no measures is not a fair world. It is a blind one.
The fix is to put the number back in its proper place: a tool you look through, not a prize you chase. Some ways people do that:
- Measure more than one thing. One number is easy to game. Several numbers that pull in different directions are much harder.
- Keep asking what the number is for. Say it out loud: “We count this because we care about that.” Then check “that” directly, by talking to people, now and then.
- Do not hang everything on the score. The bigger the reward tied to a number, the harder people will push on it.
- Look at the work, not just the report. The rats with no tails were only noticed because someone actually walked the streets.
The Oldest Goodhart Story on Record
The clearest example of Goodhart’s law might be about two thousand years older than Goodhart.
Jesus once spoke to some very careful religious people. They gave a tenth of what they had, as their law asked, and they were exact about it. They measured out a tenth of the herbs in their kitchen gardens: the mint, the dill, the cumin. Tiny amounts. Easy to count, easy to show.
But Jesus pointed out that they had let the heavier things slip: fairness, mercy, and faithfulness. The parts you cannot weigh on a scale. The things the giving was meant to grow in them in the first place.
Here is the detail that is easy to miss. He did not tell them to stop giving. He did not throw out the measure. He said, in effect, you should have done the counting, and not left the bigger things undone. Keep the measure. Just remember what it was for. That is Goodhart’s law, and its cure, in one breath. (If you want the full story of that moment, we unpacked it in What Did Jesus Actually Say About Tithing?)
Maybe that is the oldest wisdom on the subject: that God was never asking for a spice rack counted to the last leaf. The count was only ever meant to point at a heart.
Actions to Take
- Name your number. Write down one number that rules your week: a sales figure, a weight, a follower count, a grade. Next to it, write the real thing it is supposed to show.
- Check the real thing directly. This week, look at the real thing without the number. Ask a customer how it went. Notice how your body feels after a walk. Read your child a page and listen.
- Add one thing nobody counts. Do one thing today that will never show up on any scoreboard: a kind word, a careful fix, a thank-you. Notice how it feels to do something that is not for the number.
Discussion Question
Where do you think Goodhart’s law does the most damage today: in schools, at work, in healthcare, on social media, or in AI? Tell us in the comments. We’d love to hear which one you think matters most, and why.
Share This
- “When a measure becomes a target, it ceases to be a good measure.” Fun fact: Goodhart never wrote that line. An anthropologist did, 22 years later.
- In 1902 Hanoi paid a bounty for rat tails. Soon there were rats running around with no tails. Goodhart’s law in one picture.
- The fix for Goodhart’s law isn’t to stop measuring. It’s to remember what the number was for.
“A measure is a window. A target turns it into a mirror, and people start polishing the mirror.”
Questions About Goodhart’s Law
What is Goodhart’s law in simple terms?
Goodhart’s law says that when people are rewarded or judged by a number, they start working to raise the number instead of improving the thing the number was meant to show. The number goes up, but it no longer tells the truth. The popular wording is “When a measure becomes a target, it ceases to be a good measure.”
Who said “when a measure becomes a target, it ceases to be a good measure”?
The anthropologist Marilyn Strathern wrote the well-known sentence “When a measure becomes a target, it ceases to be a good measure” in a 1997 paper about auditing in British universities, building on wording by the scholar Keith Hoskin. The idea is named after economist Charles Goodhart, whose original 1975 wording was “Any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes.”
What is an example of Goodhart’s law?
A well-documented example of Goodhart’s law happened in Hanoi in 1902, when French colonial officials paid a bounty for each rat tail handed in to fight the plague. People began cutting the tails off live rats and releasing them to breed, so the tail count rose while the rat problem did not improve. Other common examples include teaching to the test and posting content only to collect likes.
What is the difference between Goodhart’s law and Campbell’s law?
Goodhart’s law and Campbell’s law describe nearly the same problem. Goodhart’s law, from economist Charles Goodhart’s 1975 paper, came out of monetary policy and says statistical relationships break down when used for control. Campbell’s law, from social scientist Donald Campbell, says that the more a numerical indicator is used for social decisions, the more it will be corrupted and the more it will distort the activity it was meant to monitor.
How do you avoid Goodhart’s law?
Goodhart’s law can be softened, not escaped, by using several measures instead of one, keeping rewards tied to any single number modest, regularly checking the real goal directly rather than only the report, and reminding everyone what each number is meant to show. Measuring is still useful; the problem starts when a single number becomes the goal itself.
