Goodharts Law

I just finished the chapter in Messy about incentives. ‘Don’t make a measure a target’ was the big takeaway. Isn’t there a ‘law’ that describes that principle?

kylecoberly [1:33 PM] Goodhart's law

kim [1:34 PM] I can see how organizations default into making them targets.

kim [1:35 PM] It’s so much easier than taking several measures (edited)

kylecoberly [1:35 PM] it's the thing lead/lag measures were designed to fix

kim [1:35 PM] oh I didn’t make that connection

kylecoberly [1:36 PM] A lag is a target, a lead is a measure it's why you're allowed to change your lead measure (and sometimes even encouraged to)

another example is how I used NPS vs. how other people did. GMs would get bonused and shit off of that, and they were incentivized to juice those the easiest way to do that is do things like selectively encourage people to do them I once saw someone crowing about some super good NPS score ...that had like 3 respondents I tried to get every single student in every class to fill it out who cares if the score sucks, don't you just want to know? but if you tie performance shit to that, you care very much if the score sucks, often at the cost of caring about what the score represents Elyse read a really good book about this called "The Tyranny of Metrics"

kim [1:41 PM] I think I’d like that book. I’ll have to check it out