5 ms·
>Incentivize for the behavior you want. This isn't as straightforward as it seems. It's kind of like that old quote, "When a measure becomes a target, it cease
by imh 10y ago
>Incentivize for the behavior you want.
This isn't as straightforward as it seems. It's kind of like that old quote, "When a measure becomes a target, it ceases to be a good measure." If the behavior you want is there because people are just casually messing around for fake internet points, it will change when you compensate them. It's probably possible, but far from simple.
- erichocean 10y ago[DELETED]
- imh 10y agoYou could say it more formally as, "Many variables correlated to outcomes you want are not causes of those outcomes. When you intervene and change those variables, they cease to be correlated to the outcomes," but it's a bit verbose[0]. I've always read it to be not about measuring things, but about optimizing the wrong thing, because people will game the system. The concept is valid. What's your concern? [0] https://en.wikipedia.org/wiki/Goodhart%27s_law#Expressions https://en.wikipedia.org/wiki/Goodhart%27s_law#Expressions Turns out the original formulation was "Any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes."
- jsemrau 10y agoWell you manage the number. If your target is a high number of internet points your post will aim to maximize that.