Integrative and Divergent

How to try to be less wrong at the organizational level

The common view of the model hockey organization is that of a confident, well-established “hockey man” surrounded by credentialed subordinates all walking in perfect lockstep towards a singular vision. One where every person is comfortably steeped in hockey lore, with an expert intuition guided by tradition and experience.

It’s an intuitive view. And it’s the sort of organization that, if run properly, would be respected and unlikely to embarrass itself.

It’s also a perfect recipe for mediocrity.

Divergent

As noted previously, right and wrong isn’t necessarily a binary, but a gradient (be less wrong).

Not: Wrong / Right

But: Wrong <———————> Right

However, in terms of knowledge markets, we can consider this gradient along two other variables: consensus and non-consensus.

The greatest reward and opportunity lies in being correctly divergent from the consensus - moving along the gradient up and to the right.

But our hypothetical organization mentioned above is mostly optimized to move away from the ruinous upper left corner, and therefore down and to the right - into a field crowded by assumed or obvious truths.

If your org’s implicit mission is to aggressively pursue the mean, then the best that can be achieved is middle-of-the-road outcomes. This is especially true in a bounded environment like the NHL, which has rules in place to create and enforce parity (salary cap, draft lottery).

That is not the explicit mission of NHL teams of course (everyone wants to win!), but the natural incentive and risk structures produce pressures to be not non-consensus wrong. That is, not divergent. Unfortunately, it’s extremely difficult to be non-consensus right if you aren’t willing to risk the humiliation of landing further back on the gradient now and then.

Other perspectives

Challenging the consensus view from within the consensus is easier said than done. Systems are often organized and structured around a given paradigm with all of its embedded assumptions and conventions. That’s why Blockbuster failed to buy Netflix - the big, established incumbent was organized around a specific value chain that obviously worked, or was “true”. Until it wasn’t.

Before the so-called “stats revolution” in hockey, the most common doubt brandished about fancy stats was their origin - why, if these insights are useful, did they come from nobodies and outsiders, and not the highly trained authorities whose job it is to to win Stanely cups?

The appeal to authority is a seductive fallacy, because it is entirely sensible on its face: shouldn’t the recognized pros, the best in the biz, be the ones unearthing the most valuable innovations? But the answer is that it’s easier said than done.

In fact, the more established and proven your organization and leadership group, the more embedded in convention and consensus - and the more likely they are to veer away from divergent thinking. And, of course, incentive structures within these organizations are often based more on avoiding failure/risk than they are about rewarding true innovation.

As a result, “other” and outsider perspectives can be especially valuable when it comes to trying to be non-consensus right. The stats revolution was birthed by a distributed group of individuals across myriad vocations, backgrounds, demographics and experiences - lawyers, engineers, mathematicians, university students, etc. - all creating, challenging, and debating new ideas in an environment relatively free of the pressures endemic to hockey professionals. No risk of career ruination plus an integrated community of diverse perspectives equals a richer milieu for divergent, innovative discoveries.

Integrative

Naturally, it’s not enough to merely generate a cloud of different perspectives. Identifying useful new insights and then integrating them into strategy, tactics, or action is the operational challenge. Especially when those perspectives are seemingly counter to not just the consensus, but to each other.

Roger Martin, author of The Opposable Mind (and many other books on the subject) defines integrative thinking as: 

“…the ability to constructively face the tension of opposing models and, instead of choosing one at the expense of the other, generating a creative resolution of the tension in the form.”

The practice of integrative thinking can help overcome the apparent conflict between a conventional view and a divergent view. This can often be revealed through “binary collapse”. A simple example:

My most recognizable tweet is a simple aphorism. It’s popularity traces to resolving two apparently incompatible views.

Conventionally, blocking shots is highly valued. It’s a behaviour that takes skill, courage, tenacity, and conviction. Blocking shots also has practical, on-ice value - they prevent shots on net, and therefore, potential goals against. Blocking shots is good!

Except new (at the time) quantitative studies consistently revealed that blocked shots was inversely correlated with winning at the team level. So…blocking shots is actually bad?

The resulting binary tension here is understandable, but false - the traditional view that blocked shots are obviously good reacts to the claim that blocked shots are provably bad with understandable derision. The math people are ignoring what’s actually happening on the ice!

But neither perspective is wrong, despite the polarisation. The conflict is resolved when you understand the intervening causal variable - that teams who tend to block a lot of shots also tend to spend a lot of time in their own zone, giving up all sorts of shots on net (including ones they don’t manage to block). As a result, the relationship of shot blocking to winning tends to be inverted, because spending a lot of time defending is usually not conducive to outscoring the opponent.

Nevertheless, that doesn’t mean that blocking shots in isolation is bad, or that the act of blocking shots itself is the primary variable behind losing. Blocking shots is still laudable and you should probably encourage your players to do it. But you should also ensure that your team isn’t doing it a lot.

What I’m reading:

Micah is one of the leaders in hockey modeling and visualization. His updated model might be one of the most important steps forward in hockey analysis.

This is a very long, in-depth piece, but I’ll summarize the importance with this image:

McCurdy’s major break through here is the modeling of the relative impact of the major contributors to a player’s shot impacts. The scale expressed above shows the level impact of each factor, from individual skill and teammates, down to the score state.

Teasing apart impact and causality has long been a major hurdle when it comes to evaluating players. If effective, Micah’s model is a major breakthrough.

A new feature one of the most useful - and free! - advanced stats sites. To sum up:

These sorts of charts will give us a “micro stat” view of individual players in a very specific circumstance. Find out how much your club’s best offensive player spikes expected goals for (XGF%) after offensive zone faceoffs, for instance!

Forget pulling the goalie early. Instead, coaches of losing teams should consider playing four forwards and one defender. Probably earlier than you’d think. Read the whole Twitter thread for the model behind the thinking.

 Breaking down Kaapo Kakko’s historically bad season by Jfresh

I knew Kakko struggled as a rookie this year, but I had no idea it was this bad. There’s some “reverse” survival bias here, in that teenagers who struggle this much early on almost always sent down by the club. For whatever reason, though, the Rangers opted not to do that.

This article leverages some work from Dominic Galamini Jr and looks at how pre-shot puck (and goalie) movement can impact shooting percentage.

Tool recommendation: Meco App

If you’re like me and have a lot of individual newsletter subscriptions, this app helps you consolidate and track them all on your phone (rather than combing through your inbox).