I've got some big posts based on emails about a) pricing options, and b) pricing houses coming up (hopefully). But as an interlude, here is Goodhart's Law:
"Goodhart's law is the equivalent in the social sciences of the uncertainty principle in physics. Though it has been expressed in a variety of formulations, the essence of the law is that once a social or economic indicator is made a target for the purpose of conducting social or economic policy, then it will lose the information content that would qualify it to play such a role. "
For example, suppose some social statistic is correlated with good things - like low unemployment. Once we declare it to be an important measure, and say that we want to reduce it, we will reduce the correlation. The reason is that we will tend to affect the statistic in the cheapest and simplest ways, which are probably going to be those which artificially inflate the statistic without addressing the problem its supposed to measure. The correlation measured "in the wild" and the correlation once we start targetting this statistic will usually be different.
The analogy to Heisenberg's principle is fascinating...
"Goodhart's law is the equivalent in the social sciences of the uncertainty principle in physics. Though it has been expressed in a variety of formulations, the essence of the law is that once a social or economic indicator is made a target for the purpose of conducting social or economic policy, then it will lose the information content that would qualify it to play such a role. "
For example, suppose some social statistic is correlated with good things - like low unemployment. Once we declare it to be an important measure, and say that we want to reduce it, we will reduce the correlation. The reason is that we will tend to affect the statistic in the cheapest and simplest ways, which are probably going to be those which artificially inflate the statistic without addressing the problem its supposed to measure. The correlation measured "in the wild" and the correlation once we start targetting this statistic will usually be different.
The analogy to Heisenberg's principle is fascinating...
- Music:Karma of Christianity


Comments
I hesitate to take the analogy any farther by mentioning Bell's Theorem.
Statistics vs Cybernetics of my essay Government is the Rule of Black Magic. Nice to know it well-known and has a name.
Of course, Governments keep sticking to the blessed Statistics beyond the point where it is utterly meaningless, and until even their expedients can't improve the blessed Statistics. Then, they reach the Drudge, and all they can do is adjust the definition and expand their action accordingly, everytime with a similar result, with nothing to stop them but their ultimate failure and fall, taking the whole society down with them, that will find a new set of rulers to start a new cycle from this empoverished point.
Question: how do we escape the downward spiral?
Or am I misunderstanding something?
For instance, in my old home town, town leaf-and-branch pickup was privatized under an incentive system in which the private contractor's pay was reduced based on the amount of yard material left on the street on collection day. So the private contractor simply moved collection day earlier in the spring to before the growth season for most of the local plants (except for those parts of town where most of the politicians live -- no point in jeopardizing the contract for futures years!). The old government-provided service had naively collected leaves and branches, not realizing how much more efficient it would be for them to instead not collect leaves and branches; the private contractor turned the system from mere government inefficiency to simple deadweight rent-gathering.
In short, a corrolary to Goodhart: Market-based solutions will make the indicator meaningless much more quickly and efficently.
Or the problem is contributed by several factors, and by eliminating one of the factors just makes the others more influential as a result eliminating any future correlation from the original variable(if it remained steady).
Thanks for posting this btw... quite a fascinating idea I'd never heard before.
-elzr.com
-----------------------------
http://willowlaboratoriesinc.com
http://willowlaboratoriesinc.org