Analytics, Education And A Tribe Mentality
There are 10 kinds of people in the world. Those who understand binary, and those who don’t.
The communication of information. Any recap, summary or rambling you’ve read previously about this year’s MIT Sloan Sports Analytics Conference most likely began with mention of that enduring theme. In this community of like-minded souls the merit and value of analytics is no longer a point of discussion, and attention has shifted to bringing others out of the darkness and into the fold.
In all the talk of communication this weekend I sensed an incredible undercurrent of condescension, one which was uncomfortable to witness at times. The very first panel Friday morning, was entitled ‘Revenge of the Nerds’, ostensibly an opportunity for the analytic-minded to slap themselves on the back and take a victory lap for seizing control of so many professional sports franchises. That subtly divisive separation persisted all the way through to the Basketball Analytics panel on the last day of the conference where Stan Van Gundy repeatedly referred to attendees as ‘you people’, immediately preceding statements about his positions on basketball strategy, positions that more often than not happened to be shared by the audience he was attempting to distance himself from.
I observed panel discussions on data visualization where the focus was on making things as simple and clear as possible. Those are noble goals regardless of context, but I felt more than once that the implication was – make it simple because your audience can’t handle complex. In ballrooms and quiet corridors I overheard the argument made that analytics needed to be made more digestible, as if the potential audience was incapable of understanding the full scope and breadth of the wisdom.
There is undoubtedly value in simplification. If a team is using analytics to communicate information to a player during a game, that situation presents a natural limit on the amount of explanation and context that can be provided. But I would argue that the goal is to lay the groundwork in situations where the natural limits are less restrictive, so that when those time-sensitive situations do arise, less paring and reshaping needs to be done. Admittedly, in some cases simplification correlates to usefulness, which is ultimately what analytics are about. But in the case of changing a culture of thought, simplification seems like an incredible step backwards.
I approach this whole issue from, what I believe is, a slightly unique perspective. I’m not a journalist, at least not in the professional sense of the word. I would consider myself both a producer and consumer of basketball analytics but I spend my days earning a living as a second grade teacher in a rural public school. While the degree of analytic adoption in professional basketball is of great interest to me, it has little effect on my daily life. However, the struggle of highlighting important ideas and helping them to be internalized by an audience, sometimes violently opposed to participation, is a very real and authentic scenario for me. To simplify or not simplify is not an analytic question to me, it’s an education one.
I wrote about this same idea after last year’s MIT Sloan Conference; how language and concepts were the chief barriers to the dissemination of an analytic mindset. For example, asking someone to understand the value of a usage-efficiency curve is useless without first explaining what usage means. Of course, understanding of usage requires the foundational knowledge of what a possession is. Those pieces of information are necessary to understand the product, a usage-efficiency curve, but understanding its value requires an appreciation of efficiency and the trade-off between production as expressed in points, and its cost as expressed in possessions.
In any hypothetical scenario where the information contained in a specific usage-efficiency curve has an immediate, essential and visceral need, ignoring a thorough explanation may be entirely appropriate. But if the basic rationale behind the usage-efficiency curve is going to have a consistent place in the larger consciousness of an organization, or group of fans, than a thorough explanation is absolutely essential. This same battle is being fought in thousands of venues, large and small, across our country’s education system. The current environment of high-stakes testing and statistical teacher evaluation has changed the paradigm. Results can be had in the short term by repititiously exposing students to answers. But in the end this creates students with breadth of knowledge but no depth, students who can follow mathematical procedures accurately and explicitly, but have no idea what that procedure represents or why it works.
Sharing a sliver of statistically gleaned information with an analytically disinclined coach or general manager may be tolerated if it’s presented in a certain way. In the event that the information is acted upon with a successful result, that process may even be repeated. But no real meaningful work has been done to change the way that person measures and evaluates the things they see in front of them.
When most people think of the Sloan conference I imagine they envision an endless parade of graphs and spreadsheets. To an unfortunate majority, those things represent analytics. But they aren’t analytics, those are statistics and visual representations of statistics. Statistics are not analytics. Statistics are evidence gathered to support or refute an idea. Analytics is the critical thinking that is done using that evidence.
My point is that in all the talk of communicating ideas, the focus seemed to be only on outputs and not communicating the underlying ideas and thought processes. This movement is about disseminating ideas, not just research findings. Results can be simplified for the sake of education, but ideas never should be. The point should be to educate and elevate ideas so that things don’t have to be simplified in the future. Operating under the assumption that those of a different perspective will only respond to carefully summarized results and are incapable of interacting with underlying concepts is just as closed-minded as the common characterization of professional sport’s old guard. The analytic revolution is not a campaign to crush and destroy those who disagree, or begrudgingly force them to tolerate a different approach, it should be a movement which is expansive enough to encourage everyone to ask more questions, dig a little deeper, challenge misconceptions and support positions with evidence.
Of course that process goes in both directions. During the Basketball Analytics panel I referenced above, Van Gundy made an argument that was entirely new to me. He pointed out that his teams never ascribed to the ‘two-for-one’ mentality of rushing a shot with 30-45 seconds left in a quarter to ensure that even if the opponent runs the shot clock down to the end, your team will ultimately have the last possession of the quarter. I’ll point out that the mathematics are incredibly opposed to Van Gundy’s perspective. Each NBA game has a finite number of possessions which are essentially distributed equally – each possession for your team ends with the other team taking possession. The only exception to that equal distribution is the end of a quarter, where the expiring clock dictates that one team may be short changed their fair share. The average possession is worth 1.05 points this season, so blatantly waving off the opportunity to grab an extra one seems incredibly foolish. Even if you account for the fact that a rushed possession may not live up to that 1.05 point average, stealing an extra one still undoubtedly puts you on the positive side of the mathematical ledger. But Van Gundy’s point was that rushing a possession, any possession, undermined the system and culture his team spent the other 47 minutes of the game trying to establish. If your offense is built on always taking good shots, encouraging a player to take a bad one, regardless of the situation, chips away at that foundation. Van Gundy’s argument was that the simple math of two possessions is greater than one ignores too many variables.
For years I’ve accepted the ‘two-for-one’ philosophy as statistical gospel, and I had literally never heard Van Gundy’s argument before. My DNA screams for some sort of data to back that up, but actually measuring the effect he’s talking about seems largely impossible. Still I’m willing to accept that his point has merit. But for a team like the one’s Van Gundy has coached, built on a flexibly rigid system and culture it’s not hard to imagine that the value of maintaining and reinforcing that culture might be worth more than an extra point or two a game. Perhaps not in the simple tabulation of wins over the course of a season, but certainly in the playoffs where system and culture are all that separate you from an equally talented opponent. Despite a lack of mathematical evidence on Van Gundy’s side, he makes a compelling case that this issue is murkier than I thought. A call for open minds, is a call for open minds. Like I said above, analytics is really thinking critically and I can’t argue that Van Gundy hasn’t done that. I’m not done thinking about that topic, but I refuse to dismiss it out of hand just because there isn’t currently data to back it up.
Like education this challenge of communication is not about distilling insight down to a digestible format, it’s about contextualizing it and connecting it to other tangible basketball knowledge. Don’t tell, show. The goal should not be to depress the incredible work that’s being done in order to make it more palatable to resistent consumers, on either side of the debates. We are not done with the real, purposeful work that needs to be put into the education side, making sure fans, players and front office personnel are all enthusiastic and willing participants in the search for meaning in basketball.