Algos, attribution, and allowances

One of our psychological tendencies (neé biases) is the fundamental attribution error. Also known as: that driver didn’t signal. We have the disposition to note that when others misstep it is their fault, but when we error it’s due to the very specific, unique, unavoidable nuance of the situation.

Conditions matter and we tend to underrate them.

Hannah Fry spoke with Shane Parrish about the role of algorithms in our lives. Though algorithms are impartial in their functionality, they are not in their design. The training set (or set of experiences by the programmer) affects the outcome. Sometimes this means we get the wrong proxies because we use what’s easy to discover, digest, or divide.

“You can’t just build an algorithm, put in on the shelf and decide whether it’s good or bad in isolation. You have to think about how the algorithm actually integrates with the world you’re embedding it in.” Hannah Fry

Conditions matter. For a natural experiment, researchers looked at county-level obesity rates and military transfers. Personnel assigned to counties with higher obesity rates were more likely to be obese. The longer they were assigned in a place, the more they trended toward the average weight.

The researchers supposed that part-of-the-reason was because of the built and natural environment. The number of gyms and parks, the access to healthy food, and the walkability of a region were all associated with a place being more or less healthy. It’s not so much ‘you are the average of the five people you spend the most time with‘ but ‘you are the average of the five places you spend the most time‘

Place and space make some things easier. It’s not impossible to in Ohio winters, but do watch the ice.

The latest pay-what-you-want pdf is now online and covers a handful of ideas from Tyler Cowen. Get it here.