By Josh Hemann
I had dinner recently with a good friend, Will Deaver, who is Managing Director at USA Ultimate, the governing body for the sport of Ultimate. Will recounted a conversation he once had with Joey Gray, a former USA Ultimate colleague (and recent Seattle mayoral candidate). Here is the gist of their conversation: The rise and fall in popularity of various sports can be seen as mirroring larger socio-political trends. Some examples bring this to life:
- Baseball becomes popular in the late 1800s. The dynamic of the game is distinctly different in the infield vs. the outfield. This mix of infield and outfield dynamics mirrors the mix of rural (outfield) and urban (infield) life rapidly evolving at this same time during America's Gilded Age
- Football grows in popularity through the 1930s-1950s. It's highly structured rules, positions, and strategy for moving a dividing line between two sides mirror America's emergence from two world wars as a military, manufacturing and science superpower
- Ultimate grows in popularity in the 1990s. In fact, it is one of the fastest growing sports in the world. Its lack of fixed positions and advocacy of Spirit of The Game are unique in team sports, but mirror the increasingly individual-focused yet highly social trends in our culture
Will's recounting of this conversation was fascinating to me because one, I love the sport of Ultimate and two, I think I have always thought popular sports and their icons act upon the larger culture, not the other way around. I had thought of sports as things in themselves, not as manifestations of the American zeitgeist.
So what does all this have to do with the subject of this blog - analytics? Well, I could not help but to see analogs with 20th century statistical development.
Gosset's foundational work on small sample properties, as well as design of experiments, was firmly in the context of mechanized, industrialized agriculture, with the same juxtaposition of rural and urban cultures that baseball's form and popularity rose from.
While football became increasingly popular during America's increasing industrialization and involvement in two world wars, a battle in statistics was emerging over significance and hypothesis testing. Today, most of what is taught in undergraduate statistics is a (confusing) combination of the Fisher vs. Neyman/Pearson schools of thought, but each was developed with much rigor and yielded defined strategies for making inference.
And much like Ultimate reflects the power and fluidity of networked individuals, the modern statistical era has perhaps reflected the same cultural shifts. For example: the increasing application of Bayesian methods replaces the Frequentist view of probability and inference for subjective, fluid notions of belief; machine learning methods tackle questions using data from networks of actors, often looking for mere correlation, not causation; and of course, there is the Data Scientist, who must be adept in many roles, from statistician to software developer to business analyst.
* Image: http://bravoultimate.org/
