Goodhart's Law
Mental architecture from the other side.
There is a moment, in the life of every measurement, when it stops describing the world and begins to replace it.
The measure is introduced honestly. It exists because some quality that matters - health, output, diligence, prosperity - is difficult to see directly, and the number offers a window onto it. For a time, the window is clear. The number moves when the quality moves. Those who watch the number learn to trust it. And then, gradually, the watchers begin to reward the number itself: to pay for it, promote on it, and allocate by it. From that moment the number acquires a life of its own. The people being measured stop producing the quality and start producing the figure. The window becomes a painting of a window. It still hangs in the same place on the wall, but it no longer looks out onto anything.
The men who understood this best were not economists. They were the citizens of the country that ran the largest measurement experiment in human history - and the joke they told about it deserves to be studied as seriously as any paper in the literature.
In the Soviet Union, the economy was commanded by quota. Gosplan, the state planning committee, set targets for every enterprise in the country: so many tonnes, so many units, so many roubles of output per quarter. The targets were the economy’s nervous system. A factory director’s career, his bonus, his standing with the ministry - all of it hung on the fulfilment of the plan. And so the plan was fulfilled. Always, everywhere, by whatever means the arithmetic permitted.
The means became a national art form. Set a nail factory’s quota in tonnes, and it would pour its steel into a small number of enormous, railway-spike monstrosities - heavy, magnificent, and utterly useless for hanging a picture or joining two boards. Reset the quota in units, and the same factory would turn out an avalanche of tacks: thousands upon thousands of pins so small and brittle they could fasten nothing, but which counted, each one, toward the total. The plan was met in both cases. That was the genius of it. The plan was always met. But the country, decade after decade, could not find a decent nail.
The Soviets knew. That is the detail worth savouring. This was not a pathology visible only to Western economists after the archives opened. The regime’s own satirical magazine, Krokodil, ran a celebrated cartoon on precisely this theme: a factory floor, a proud director, and suspended from a crane a single colossal nail the size of a girder - the month’s entire tonnage quota, fulfilled in one unit. The censors let it run. Everyone understood the joke, because they all lived inside it. The system had asked for a number, and the system had received the number, and somewhere in the space between the two, the actual purpose - nails, for building things - had quietly been forgotten.
Two decades before the archives confirmed the scale of it, a Bank of England economist named Charles Goodhart set down the principle in the dry language of his trade: any observed statistical regularity, he noted, will tend to collapse once pressure is placed upon it for control purposes. He was writing about monetary targets - the money-supply measures that central banks of the 1970s had begun to steer by, only to watch the measures buckle and distort the moment they became instruments of policy. But the observation was universal, and it has carried his name ever since. When a measure becomes a target, it ceases to be a good measure. The window, leaned upon, becomes a painting.
The investor would like to believe this is a story about communism. It is not. It is a story about measurement under incentive, and the modern market has not abolished the machinery of Gosplan. It has privatised it.
Consider what the street actually watches. Earnings per share, reported quarterly. Same-store sales. Subscriber additions. Monthly active users. Book-to-bill, backlog, adjusted margins, and the whole apparatus of guidance met and beaten. Every one of these began life as an honest window - a proxy for some underlying quality that genuinely mattered: the durability of the enterprise, the loyalty of its customers, the productivity of its capital. And every one of them, the moment the market began to pay for it, became a quota. The managers being measured are not Soviet directors, but they are subject to the same law, and they respond with the same art.
The earnings number is met - by pulling forward sales, deferring costs, repurchasing shares in the final weeks of the quarter, adjusting the adjustments. The subscriber number is met - by counting trials, discounting to zero, redefining what a subscriber is. The growth number is met — by acquisition, by channel-stuffing, by the thousand small chemistries available to a chief financial officer with a target and a deadline. Each quarter the plan is fulfilled. The colossal nail is hoisted for the analysts, and the call concludes with congratulations. Whether the enterprise is actually accumulating the qualities the numbers were built to indicate - whether anything is being built that will hold two boards together - is a separate question.
The investor’s error is the Gosplan official’s error: mistaking the fulfilment of the figure for the existence of the thing. He screens on the metrics, ranks on the metrics, and congratulates the companies that hit the metrics, and in doing so he becomes the ministry - the distant authority whose demand for a legible number is precisely the force that corrupts it. The more mechanically he rewards the figure, the more surely he will receive figures rather than substance. He is not a victim of the game. He is its sponsor.
Goodhart’s law cannot be repealed, but it can be respected, and respect takes a particular form. The discipline is to treat every number as a window that may have become a painting - to ask, of each metric, what quality it was originally installed to reveal, and then to go looking for that quality directly, by the unmeasured routes. Read the accounts rather than the summary. Notice what the adjusted figures adjust away. Ask what a company would look like if it were manufacturing its numbers rather than earning them, and whether anything distinguishes the two from where you sit. Distrust most, not least, the enterprise that meets its targets with perfect regularity - nature is not that tidy, and Gosplan’s factories never missed the plan either.
Above all, the investor must keep in his mind the distinction the quota destroys: the difference between the number and the nail. The number is what the system reports. The nail is what the world needed. Whole economies have starved amid fulfilled quotas, and investment portfolios have been assembled out of companies whose every metric was immaculate and whose substance had long since been poured into a single, magnificent, but useless spike.
The quota was met, but the country had no nails.
Look past the figure, always, to the thing the figure was supposed to stand for.
Stay still.
Win slow.
Theodore

Great point about the dangers of focusing on particular measures rather than the big picture. This reminds me of stories about the Malcolm Baldrige Quality Award and the broader Six Sigma quality movement that was so popular 30-35 years ago. The focus on quality became so expensive and narrowed focus so much that too much attention to "quality" backfired on some of the "winners." 1990 award winner Wallace Co. filed for bankruptcy, and Six Sigma advocate GE ultimately imploded.