Dan Pink has a new thing called the 1-3-20 podcast. The idea is that each week he selects one important non-fiction book, asks its author three key questions about it (What’s the big idea? Why should I care? What should I do?), and does this all in under 20 minutes.
In the first episode he interviews data scientist Seth Stephens-Davidowitz, about his book Everybody Lies. Davidowitz focuses on all the ways in which data can reveal fundamental truths about human behaviour even when people have a tendency to massage or hide those truths (here's a decent review of the book).
There's a nice story he tells of how horse racing analyst Jeff Seder used data and statistical correllation to better predict how good particular horses were going to be on the racecourse and identify high-potential winners (most notably the horse American Pharoah which won the 2015 Triple Crown). Horses were of-course traditionally judged by their pedigree, size or gait. If you'd grown up in the horse industry this would be how you'd assess how good a horse was likely to be, but when Seder looked at the data all these factors turned out to be relatively poor predictors of greatness. Instead, informed by years of data collection, Seder found that whilst other attributes (like size) needed to be within a certain range, there was one particular attribute that was a far better predictor of racetrack success - the size of the horse's left ventricle.
Davidowitz makes the point that we often have in-built biases that new data sources can challenge, thereby revealing previously hidden insights. This is particularly powerful in domains where deeper data analysis is less prevalent or where people from outside of a domain are more likely to look for (what may seem to those inside that domain to be) unlikely sources of causation.
Sometimes in my client work people say to me that they want to focus on in-sector examples. The belief is that these examples will be more relevant, and therefore more applicable. I always challenge this. Apart from the fact that most teams already know what's going on in their own sector (or at least they should do), it's far more useful to look at the most interesting out-of-sector examples and ask yourself what's distinctive about what they're doing and how that thinking might be applied to your own context. It's less comfortable but if you don't challenge your perspective, how are you ever going to break out of the kind of thinking that led to needing to be challenged in the first place?
Image source
