Algorithms vs. Gut Response

Perhaps it’s only peripherally related to design but this article about baseball statistics blew my mind. I don’t even like baseball much but this is about how analytical statistics and algorithms are often pitted against gut response and instinct and a “cultivated judgement.” At work, I think of this a lot in terms of the machine learning and data modeling done by my team at Microsoft. At home, I think about how Google and Facebook and Netflix use algorithms to craft the media I’m presented. As a musician, I think a lot about the Spotify backend using implicit matrix factorization to suggest I listen to the Pixies.

So basically, the math and probability nerds creating scoring systems and playing “Moneyball” versus the wise and grizzled old baseball scouts chomping cigars at a minor league game, looking for the next superstar. Or the AI engineers vs. the film connoisseur. Or Spotify’s Python libraries versus the cool DJ who can compose the perfect mix tape, just for you.

What’s amazing about this article is that it argues that the distinction between those two strategies is often a false dichotomy. Even when designing algorithms to sort big data, there’s still opinions and judgement at play about what data and what scoring is actually important. Does it matter how far a batter hits the ball? Does it matter more that they get on base? Does it matter how they play at night versus during the day? Despite the ability tech has to compile vast amounts of data and measure correlations, it’s still a judgement call about what correlations matter most.

For some reason, I find this deeply comforting, that often at the root of the most powerful aggregation programs is a judgement call, and human opinion, an educated guess.