‘Wishful thinking. Most people will never learn the game of basketball…’
A brief note before we get to the main topic… I’m launching a new project next week, writing about a fictional NBA expansion team called the Pittsburgh Pierogies. You may have come across something similar before, but I’m planning to take it somewhere very different. Read the introduction, and be sure to subscribe to the mailing list so you don’t miss any entries — it’ll be published exclusively via email.
In 2014, Ken Arneson, a baseball blogger I’ve followed for more than a decade and a half, wrote a piece I’ve pondered ever since. Titled “10 Things I Believe About Baseball Without Evidence”, after I read it, I had a distinct feeling that it expressed thoughts I already had but couldn’t articulate, even to myself.
While the whole thing is worth reading, whether you’re a baseball fan or not, because it’s an interesting and clearly-stated critique of unearned certainty and overreliance on data that hasn’t been properly contextualized, the two “beliefs without evidence” that I think have the most direct application to how other sports are described and consumed are No. 1: “A technological Sapir-Whorf hypothesis” and No. 3: “All high-level sabermetric truths derive from lower-level truths about human biomechanics and psychology”.
For a recent example, I thought of these sections of Arneson’s essay when Draymond Green gave his infamous “most people will never learn the game of basketball” comments. Yes, some of what animated Green is a strawman — I’m pretty well-versed in the online NBA world, and the very few dudes who actively promote Xs and Os expertise are pretty widely reviled by non-casuals, and the rest of the people who try to show their enthusiasm through thoughtful breakdowns and analyses tend not to assign individual credit or blame too stridently because they understand they don’t know how, precisely, a player’s been coached, or what defensive rules a team is using, and so on.
But think of Green’s statement in the context that he probably sees there’s a broader move toward trying to understand basketball through statistical analysis, of which ESPN’s embrace of Real Plus-Minus is a big part (that measure has him at 258th in the league so far this year), and it makes a lot more sense he’d be defensive about the gap between what he lives and what other people say they see onscreen.
Arneson’s point about “a technological Sapir-Whorf hypothesis” certainly applies to basketball today. For what it’s worth, it seems most linguists don’t really subscribe to the full original idea, but Arneson’s more limited point is that privileging a problem-solving process that involves entering and massaging numbers in a spreadsheet and then re-applying their meaning back out into the real world is problematic if we’re not especially careful about what it is we’ve measured in the first place. It’s “If your only tool is a hammer, lots of things start looking like nails” writ large.
We don’t know what, exactly, goes into the RPM model, and while I’m not going to get in the weeds of whether RPM is “good” or not, I will note that as a proprietary formula, the public’s trust in the statistic’s usefulness is really just trust in ESPN not to feed us crap. More specifically, why do we have an appetite for a one-number statistical measure like this in the first place? Ultimately, does a single number attempting to sum up an athlete’s contributions actually provide more clarity, even with caveats about its usefulness and how it’s just one part of a mosaic that describes an athlete’s place in their sport? What would you think if your job evaluation was reduced to a single number or small set of numbers? There’s a reason many American work reviews involve subjective narratives: subject-matter experts’ (supervisors, in theory) observations and judgments matter more than numbers entered in a spreadsheet because for the average American worker we don’t trust that we can measure their work and sum it up into one number.
Obviously, with points scored and easily understandable things to measure, sports make for a tempting statistical-analysis sandbox. Basketball may be more free-flowing than baseball and therefore harder to fit into a spreadsheet model of evaluation, but there are still distinct things to measure, especially with Second Spectrum data. With the success of numbers-oriented management teams across sports, industry, and academia, it makes sense basketball management teams would apply that framework to basketball.
And they’re probably correct to do it! The core principle that one should make decisions based on data instead of intuition is the sounder way to proceed most of the time, and in the very recent past there were NBA teams attempting to manage their rosters without that kind of statistical analysis entirely, thus putting themselves at a disadvantage to the teams gaining real insights into the game from their statistical acumen.
However, that leads us to Arneson’s No. 3, that “all high-level sabermetric truths derive from lower-level truths about human biomechanics and psychology”. In the case of Real Plus-Minus, ESPN says it “estimates how many points each player adds or subtracts, on average, to his team's net scoring margin for each 100 possessions played.” Arneson’s suggestion, applied here, is that we ought to think of RPM as the result of biomechanical/psychological forces playing out on the court, and therefore RPM is a description of players’ biomechanical or psychological attributes, ultimately. And if that’s the case, does RPM more accurately describe how various players contribute to winning than a subject matter expert using narrative and trading card statistics to describe those same players? I’m not sure.
Just a few years ago, prominent Sports Illustrated writer Lee Jenkins was hired by the L.A. Clippers to a newly-created executive position that, as best I can tell, still hasn’t been fully explained externally. However, there’s some evidence the Clippers simply felt Jenkins’s elite powers of observation and ability to express those observations in narrative held value to the basketball operations team. Of course the Clippers use statistical measures to inform their decision-making, but isn’t it possible they realized someone with Jenkins’s skillset is good to have on the executive team along with number crunchers, and along with basketball lifers like Jerry West and Lawrence Frank? That, I’m pretty sure of.
(Photo: "On the paint" by Mark Gunn. Used under CC BY 2.0 license.)