The fox knows many things, but the hedgehog knows one big thing

Vulpes macrotis mutica with pupsThe title is a (translated of course) quote from Archilochus, a Greek poet from the 7th century BC. Somewhat randomly, this quote from a long dead Greek poet is very popular in the fields of data science and prediction. Nate Silver’s fivethirtyeight.com site has a stylised fox as logo in reference to this quote, although I don’t think he was the first to start using it. The quote is used to classify people, in particular those making predictions, into two groups: the hedgehogs and the foxes.

A hedgehog is someone who has one big fixed idea, and who then bases their predictions on this idea. Examples are a free market ideologue who thinks free markets will solve all problems, or a statist ideologue who thinks the state is always better. A fox is the opposite, they are flexible and have many ideas to understand the world. They try and view problems from several perspectives before making a prediction, and then learn from their mistakes. If they get new data, they reconsider their prediction.

Hedgehogs make good copy, the reasoning behind their predictions are clear and easy to understand, and are consistent from one prediction to the next. In Philip Tetlock and Dan Gardner’s book Superforecasting, they make the point that the media favours hedgehogs as they make for good TV, and for good newspaper columns.

However, Tetlock’s claim to fame is that he ran systematic studies of people’s ability to make accurate forecasts. He recorded these forecasts and then tested whether they came true, or turned out to be wrong. He found that hedgehogs were often about as good at predicting as the proverbial coin-tossing chimpanzee. But that some foxes were much better. With their more complex and nuanced reasoning, foxes may make poor TV pundits, but excellent predictors of the future.

Superforecasting is an interesting book. It makes interesting points about what it takes to predict as accurately as possible, and why hedgehogs are often asked to make predictions, despite the evidence that their predictions are typically worthless.

As an academic scientist, it has made me worry that in principle our job is to develop models to make accurate predictions. So we should be foxes. But doing research costs money, and getting funded research is easier if you are a hedgehog, as selling one simple idea is a lot easier than a more nuanced position. The pressure on scientists to be hedgehogs may be one factor in the reproducibility problems particularly affecting biomedical and psychology research. If so, encouraging fox-like behaviour in scientists could be one way to reduce the number of studies that cannot be reproduced.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s